Dynamic radical recombination enabling stereodivergent construction of spirocycles with nonadjacent stereocenters

Abstract

Chiral spirocyclic scaffolds have garnered significant attention in drug discovery and chiral ligand development due to their rigid structures and potential bioactive properties, yet stereodivergent synthesis of such systems, particularly those bearing non-adjacent stereocenters, remains a formidable challenge. Herein, we report a dynamic radical recombination (DRR) strategy that enables cobalt-hydride-catalyzed ligand-controlled stereodivergent olefin hydroalkylation, delivering a series of spirocyclic compounds bearing 1,3-non-adjacent stereocenters with up to >99:1 er and >20:1 dr. Density functional theory (DFT) calculations reveal that different ligands induce distinct reaction mechanism, resulting in diastereoselectivity reversal through dynamic radical recombination pathway. Biological evaluations demonstrate that selected newly synthesized spirocyclic products markedly suppress lipopolysaccharide (LPS)-induced neuroinflammation in microglial cells, effectively reducing pro-inflammatory cytokine levels (TNF-α, IL-6, IL-1β) and restoring cells to pre-inflammatory states.

Chiral spirocyclic scaffolds have garnered significant attention in drug discovery and chiral ligand development due to their rigid structures and potential bioactive properties. Herein, the authors report a dynamic radical recombination strategy that enables cobalt-hydride-catalyzed ligand-controlled stereodivergent olefin hydroalkylation, delivering a series of spirocyclic compounds bearing 1,3-non-adjacent stereocenters.

Introduction

Chiral molecules with distinct absolute and relative configurations often exhibit markedly different physiological or pharmacological properties due to their precise interactions with chiral targets¹. Stereodivergent synthesis to access all possible stereoisomers of chiral molecules is of paramount importance for elucidating stereochemical structure–activity relationships². Chiral spirocyclic scaffolds containing multiple stereocenters are widely present in natural products, bioactive molecules, drugs and chiral ligands^3–6. Notably, the incorporation of spirocyclic scaffolds into pharmaceutical candidates can improve pharmacokinetic properties, such as potency, selectivity, protein binding affinities, and metabolic stability^7,8. Consequently, the development of efficient methods for the stereodivergent synthesis of the privileged spirocycles is highly desirable.

However, achieving stereodivergence in the construction of multiple stereocenters through a single catalytic transformation is notoriously difficult, owing to the inherent predisposition of a specific diastereomer, while the other diastereomers are often difficult to obtain directly with satisfactory efficiency^9–11. Since the pioneering work by Carreira in 2013, dual catalysis has become a powerful strategy for stereodivergent synthesis^12–14, and various metal/organo^15–31, metal/metal^32–52, and organo/organo^53–55 dual catalytic systems have been developed. However, such strategies mainly provide the stereodivergent synthesis of adjacent stereocenters in acyclic and cyclic frameworks (Fig. 1a). In contrast, stereodivergent asymmetric construction of much more rigid spirocyclic scaffolds in a single catalytic operation is exceptionally scarce, due to their increased complexity and structural constraints as well as the difficulty of overcoming the stereochemical bias of the substrate in rigid spirocyclic structure formation⁵⁶. To date, only a few examples of stereodivergent asymmetric catalytic synthesis of spirocycles have been reported^57–60. Strategically, these reports typically rely on annulation or cycloaddition reactions, and provide spirocyclic compounds containing adjacent stereocenters (Fig. 1b). To our knowledge, diastereodivergent asymmetric catalytic synthesis of chiral spirocycles bearing non-adjacent stereocenters are rare. In 2007, Trost group disclosed an elegant construction of chiral spirocyclic oxindolic cyclopentanes through Pd-catalyzed asymmetric [3 + 2]-cycloaddition with 3-alkylidene-oxindoline-2-ones⁶¹. However, the scope of substrates was rather limited, and the diastereoselectivity was inadequate. Thus, the development of new catalytic systems (such as non-precious metal catalysts) and synthetic strategies (such as non-annulation reactions) for diastereodivergent asymmetric synthesis of spirocycles bearing non-adjacent stereocenters remains to be developed and challenging.

Fig. 1. Background of diastereodivergent asymmetric catalytic synthesis and our work based on dynamic radical recombination.

a Stereodivergent catalysis for synthesizing different product types or frameworks. b Diastereodivergent asymmetric catalytic synthesis of rigid spirocycles: state-of-the-art. c This work. d Representative biologically active spiro cyclopentanes featuring multiple stereogenic centers.

Recently, metal hydride (MH)-catalyzed reductive hydroalkylation of olefins with alkyl electrophiles has emerged as an efficient strategy in organic synthesis and become an area of intensive interest^62–67. This method typically employs inexpensive earth-abundant transition metals (such as Cu, Ni and Co) as catalysts and utilizes easily available olefins as pro-nucleophiles, rather than the moisture- and air-sensitive organometallic reagents typically required in traditional cross-coupling⁶⁸, providing an attractive mild method for the stereoselective construction of C(sp³)−C(sp³) bonds^69–84. Although there is significant progress in MH-catalyzed enantioselective hydroalkylation, achieving diastereodivergence in MH-catalyzed diastereo- and enantioselective hydroalkylation of olefins is challenging and remains elusive. Mechanistically, in the reported MH-catalyzed asymmetric alkene hydroalkylation, insertion of an alkene into a metal hydride species generates in situ a metal alkyl species. The enantioselectivity is set at this migratory insertion step (enantioselective hydrometallation). Then the metal alkyl species captures with an alkyl radical to form a metal bis(alkyl) intermediate, which undergoes a direct stereospecfic (stereoretentive) reductive elimination to inherently produce a single stereoisomer from a given alkene configuration, thereby precluding the diastereodivergence. To address this inherent challenge and limitation in MH catalysis chemistry, we envisaged that precise dictation of chiral ligand might lead to a new and different mechanistic pathway involving a dynamic radical recombination (DRR) process, in which the metal bis(alkyl) intermediate undergoes a reversible C-M bond cleavage (homolysis) followed by an reversible diastereoselective C-M bond re-formation rather than a direct stereospecfic reductive elimination, providing a potential opportunity for the unexplored diastereodivergent synthesis. Given the challenge of diastereodivergent construction of non-adjacent stereocenters, considering the importance of chiral spirocycles and in combination with our interest in MH-catalyzed hydrofunctionalization chemistry^85,86 and spirocycle chemistry^87,88, we tested our proposed new strategy in the context of stereodivergent synthesis of the privileged spirocycles containing 1,3-non-adjacent stereocenters by exploring the reactivity of spirocyclic olefins toward MH-catalyzed desymmetric diastereodivergent and enantioselective hydroalkylation. Prior to this work, metal-hydride-catalyzed asymmetric olefin hydroalkylation for constructing chiral spirocycles has not been reported. Moreover, diastereo- and enantioselectivity control of rigid spirocyclic olefins remains a challenge, with no successful examples of the use of spirocyclic olefins in MH-catalyzed hydroalkylation.

Herein, we report our successful efforts to implement such a strategy and present a cobalt hydride catalyzed ligand-tuned enantio- and diastereodivergent hydroalkylation of spirocyclic olefins (Fig. 1c). A wide range of spirocyclic olefins, including those derived from indolinones, benzofuranones, 3-isochromanones, pyrrolidones and piperidones, can be utilized within this reaction system, providing a general stereodivergent access to structurally rigid chiral spirocyclic compounds bearing 1,3-non-adjacent stereocenters. This work represents the pioneering example of diastereodivergent and enantioselective construction of non-adjacent stereocenters via MH-catalyzed olefin hydroalkylation and also the diastereodivergent asymmetric synthesis of spirocycles by MH-catalyzed olefin hydroalkylation. Density functional theory (DFT) calculations reveal that the stereochemical control of the spiro quaternary carbon is governed by a cobalt(II) hydride insertion step, while the chiral induction at the tertiary carbon proceeds through distinct mechanisms under L1 and L2 conditions. In the presence of L1, the spirocyclic alkylcobalt(III) intermediate undergoes direct reductive elimination to afford the cis-spirocyclic product. Through ligand-modulated mechanistic bifurcation, L2 triggers a DRR pathway, where configurational inversion of the spirocyclic alkylcobalt(III) intermediate occurs via radical recombination before reductive elimination, selectively yielding the trans-spirocyclic product. Importantly, certain spirocyclic products exhibited excellent anti-inflammatory effects against LPS-induced neuronal inflammation, restoring mouse microglial cells to pre-LPS stimulation level.

Results

Reaction development

We began our investigation with the model reaction between spirocyclic olefin 1a with alkyl iodide 2a to construct spirocyclic oxindolic cyclopentane 3a (Table 1). After careful screening, we have finally determined the best reaction Condition A as follows: CoCl₂ (15 mol%), bisoxazoline ligand L1 (18 mol%), K₃PO₄·H₂O (2.5 equiv), DMMS (3.0 equiv), and MTBE (0.5 mL) as the solvent, the reaction is conducted at 0 °C for 96 h. During the screening process, we found that chiral bisoxazoline ligands exhibited excellent catalytic activity and were able to effectively control enantioselectivity and diastereoselectivity (entry 1, 56% yield, 10:1 dr, and >99:1 er). Bisoxazoline ligands L4 and other cobalt salts also produced similar results with a slight decrease of dr values (entries 5 and 13). Many ether solvents have been tested for the reaction, with isopropyl ether performing well but with a low yield of 30% (entry 15). When using DME as the solvent, excellent enantioselectivity control can be achieved, although the diastereoselectivity control is relatively average (entry 14). The use of many bisphosphine ligands, ferrocene ligands only resulted in alkene isomerization product 4 (entries 10 and 11). Pyridine oxazoline ligands failed to produce the desired product and instead yielded a large number of protonated products (entry 12). Using other silanes or reducing the silane equivalent resulted in a decrease in yield but did not affect enantioselectivity and diastereoselectivity control (entries17 and 18). Interestingly, when using the bisoxazoline ligand L6, we observed a diastereoselective inversion of the product (entry 7). Therefore, we conducted extensive screenings on this class of ligands and ultimately identified L2 as the optimal ligand, achieving the diastereomer with 75% yield, >20:1 dr and 90:10 er by employing DME as the solvent (entry 20). The inability to obtain the target product when using ligand L7 suggests that the reaction is extremely sensitive to steric hindrance of the chiral ligand (entry 8). Using other cobalt salts produced similar results (entry 19). When nickel was used, no target product was detected (entry 16). After carefully screening other experimental parameters, we determined the optimal condition for achieving trans-spirocycles as Condition B: 15 mol% CoCl₂, 18 mol% bisoxazoline ligand L2, 2.5 equiv of K₃PO₄·H₂O, 3.0 equiv of DMMS, and 0.5 mL of DME as the solvent. The reaction was conducted at 0 °C for 96 h (entry 21, 87% yield, >20:1 dr, 90:10 er). (see the Supplementary information for details on other condition screenings).

Table 1.

Summary of selected reaction condition optimization

entry	variants	3a			4
		yield [%]a	erc,d	drd (cis:trans)
1	none	56	>99:1	10:1	ND
2b	L1	65	>99:1	10:1	ND
3	L2	44	90:10	1:8	ND
4	L3	trace	60:40	1:2.5	ND
5	L4	53	>99:1	8.4:1	ND
6	L5	22	80:20	1:2.7	ND
7	L6	64	91:9	1:1.6	ND
8	L7	ND	-	-	ND
9	L8	70	87:13	1:10	ND
10	L9	ND	-	-	30
11	L10	ND	-	-	43
12	L11	ND	-	-	<5
13	CoBr2•DME	64	>99:1	7.5:1	ND
14	DME	60	>99:1	5:1	ND
15	(i-Pr)2O	30	>99:1	10:1	ND
16	NiCl2•DME	ND	-	-	<5
17	(EtO)3SiH	41	>99:1	10:1	ND
18	DMMS (2.0 equiv)	47	>99:1	10:1	ND
19	L2 and CoBr2	72	88:12	<1:20	ND
20	L2 and DME	75	90:10	<1:20	ND
21e	L2 and DME	87	90:10	<1:20	ND

Conditions: 1a (0.10 mmol, 1.0 equiv), 2a (0.2 mmol, 2.0 equiv), cobalt catalyst (10 mol%), ligand (12 mol%), silane (3.0 equiv), base (2.5 equiv), solvent (0.2 M), 0 °C, 48 h.

Reactions were carried out under an argon atmosphere.

MTBE tert-butyl methyl ether, DME 1,2-dimethoxyethane, DMMS methyldimethoxysilane, ND not detected.

^aNMR yield. 1,3,5-trimethoxybenzene was used as an internal standard.

^bCoCl₂ (15 mol%), L1 (18 mol%) and extend the reaction time to 96 h, isolated yields.

^cMajor diastereomer.

^dDetermined by high-performance liquid chromatography (HPLC) analysis.

^eCoCl₂ (15 mol%), L2 (18 mol%) and extend the reaction time to 96 h, isolated yields.

Mechanistic investigations

The mechanism of CoH-catalyzed diastereodivergent asymmetric hydroalkylation of spirocyclic olefin remains elusive, especially the diastereodivergent stereocontrolled process. Thus, a series of control experiments were undertaken to elucidate the reaction mechanism. Under both Condition A and Condition B, the alkylation reaction was inhibited or completely prevented when added different equivalents of TEMPO (2,2,6,6-tetramethyl-1-piperidinyloxy) (the alkyl/TEMPO radical adducts were captured by high-resolution mass spectrometry (HRMS)), suggesting that the activation of alkyl iodide is a radical process (Fig. 2a). Additionally, the spirocyclic olefin/TEMPO radical adducts were also trapped, leading to the hypothesis that the reaction might involve a C-Co bond homolytic cleavage, which was discussed by DFT Calculations. Ph₂SiD₂ was used instead of DMMS to perform the deuterium-labeling experiments, indicating that the hydrogen source originates from silane (Fig. 2b). Subsequently, nonlinear effect experiments were performed (Fig. 2c). The ee values of cis-3a and trans-3a exhibited a linear correlation with the ee values of the respective ligands L1 and L2. These results suggest that under standard Conditions A and Conditions B, the active catalytic species are likely mononuclear cobalt complexes coordinated by a single ligand. Finally, we prepared a complex of cobalt chloride with ligand L1 (structure of the CoCl₂/L1 complex was determined by single-crystal X-ray diffraction) for subsequent ultraviolet-visible measurements on the model reaction mixture. The UV absorption curve of CoCl₂/L1 differed significantly from those of the reaction system at 24 h, 36 h, and 48 h, but the curve of CoI₂/L1was similar to the curve obtained with the addition of sodium iodide and curves of the reaction system at different time, indicating that the active species in the catalytic system may be obtained by replacing the chloride in cobalt chloride with iodide. The concentration of the active catalytic species decreased slowly over time while the yield increased, and a relatively high concentration was still present after 48 h of reaction. Based on the relevant references^70,77,80,85, this indicate that during the insertion of CoH into spirocyclic olefin, the possibility of Co(Ⅱ)H insertion followed by alkyl radical absorption and reductive elimination to obtain the target product is higher than that of Co(Ⅰ)H insertion, although the latter process cannot be completely ruled out (Fig. 2d)^77,80.

Fig. 2. Mechanistic investigations.

a Radical capturing experiments. b Deuterium experiments. c Nonlinear effect study. d Ultraviolet visible measurements.

To gain deeper understanding on the reaction mechanism and probe the origin of the ligand-tuned stereoselectivities, we conducted DFT calculations. Given that 3 d metal complexes possess multiple spin states close in energies, we examined both low-spin and high-spin states in our computations⁸⁹. As shown in Fig. 3a, the high-spin species ⁴CoL1–H is more stable than its low-spin analogue ²CoL1–H by 8.0 kcal/mol. However, the doublet migratory insertion transition state ²TS1 is energetically more favored than the quartet transition state ⁴TS1 by 17.6 kcal/mol, indicating the involvement of a spin crossover. The corresponding minimum energy crossing points (MECP1 and MECP2) during this course were located to be slightly higher than ²CoL1–H and ²INT1 by around 5 kcal/mol, respectively, in terms of electronic energies, suggesting that the two-state reactivity is feasible in this process. The alternative pathway via MHAT were calculated to be significantly higher than migratory insertion, being consistent with our previous findings (Fig. S9)⁸⁵. Following this migratory insertion, the Co(II) intermediate ⁴INT1 captures a propyl radical to form the relatively stable Co(III) intermediate ³INT2 via ³TS2 with electrons on C and Co antiferromagnetically coupled with each other (Fig. S10). Subsequent reductive elimination affords the product cis-3l, being consistent with experimentally observed configuration. Considering the relatively low bond dissociation energy (BDE) of Co(III)–C bonds⁹⁰, we also considered a competing radical pathway from ³INT2. Our calculations revealed that homolytic cleavage of Co–C via ³TS4 is less stable than reductive elimination via ³TS3, with an energy difference of 2.7 kcal/mol. This is attributed to the thermodynamic stability of the archetypal pyramidal configuration of ³INT2, where the alkyl group in the ligand backbone impose little steric hindrance to the spirocyclic alkyl group, evidenced by the nearly vertical (89.1°) N–Co–C bond angle. The energy profiles on low spin state were calculated to be significantly higher than that on high spin state (Figs. S11 and S12), thus ruling out their possibilities. The doublet migratory insertion was established as stereo-determining step according to the energy profiles, and we thus examined all plausible TSs with different configurations (²TS1-1 to ²TS1-3; see Fig. S13 Supplementary Information). ²TS1 was the lowest in energy among them, benefiting from favorable C–H···O electrostatic interactions and C–H π interactions depicted in Fig. 3a.

Fig. 3. DFT studies.

a Calculated energy profiles on cobalt catalyzed cross coupling between propyl radical and substrate 1 g using L1 as ligand; b Calculated energy profiles on cobalt catalyzed cross coupling between propyl radical and substrate 1 g using L2 as ligand; low spin state is presented in red and high spin state in blue; the relative free energies and electronic energies are given in kcal/mol.

We then focused on the reaction pathway employing L2 as ligand (Fig. 3b). Similarly, the doublet migratory insertion transition state ²TS7 is kinetically more favored than the quartet transition state ⁴TS7 by 12.8 kcal/mol, indicating a spin crossover is involved in the process in which both related MECPs were located to be feasible. After migratory insertion, the Co(II) species ⁴INT7 couples with the propyl radical to generate ³INT8 through ³TS8. This intermediate then undergoes a very facile homolytic Co–C bond cleavage via ³TS9 with a barrier of only 3.9 kcal/mol, generating secondary radical species 1g-rdc. Then 1g-rdc re-approaches the cobalt center with another side, forming Co(III) species ³INT10 via radical recombination transition state ³TS10. The subsequent reductive elimination from ³INT10 delivers the trans-3l product, agreeing with the experimental results. On the other hand, reductive elimination directly from ³INT8 via ³TS12 is energetically less favored than the radical dissociation pathway (higher than ³TS10 by 0.9 kcal/mol according to Curtin-Hammett principle). Carefully comparing Co(III) complexes with two different ligands, we found that while ³INT2 with L1 as ligand is more stable than Co(II) intermediate ⁴INT1, ³INT8 using L2 is prominently unstable compared to those Co(II) complexes. As shown in Fig. 3b, in ³INT8, due to the steric repulsion from the bulky ligand arms, the N–Co–C angle was bent to 124.8°, significantly deviated from pyramidal geometry. This spiro alkyl group further forces the propyl group point towards the ligand, resulting in steric repulsions. The more congested steric environment of L2 destabilizes Co(III) complexes and facilitates the corresponding radical dissociation over direct reductive elimination. In the meanwhile, the existence of radical 1g-rdc was confirmed by radical trapping experiments and HRMS, further supporting the involvement of a homolytic cleavage process when L2 was employed. Overall, the energy profile suggests that when using L2, the chirality of spiro-carbon is governed by the migratory insertion step. Compared with other configurations, ²TS7 was calculated to be more favored due to the stabilizing C–H···O and C–H π interactions (see Fig. S14 Supplementary Information). Since the two Co(III) complexes in different configurations could interconverted to each other via radical dissociation and association, the chirality of secondary carbon was controlled by radical association and reductive elimination processes according to Curtin-Hammett principle. In summary, different ligands induce variations in the stability of intermediates, which alter the reaction mechanism, thereby enabling distinct stereoselective controls.

Substrate scope

With the optimal reaction conditions in hand, we proceeded to investigate the substrate scope for the asymmetric construction of cis-type products (Fig. 4). A wide range of alkyl iodides bearing various functional groups, including alkyl ethers, alkyl silyl ethers, alkyl/aryl esters, underwent smoothly to afford the target spirocyclic products containing 1,3-non-adjacent stereocenters (cis-3a)−(cis-3f) in 62–86% yield with excellent diastereoselectivity (8:1−15:1 dr) and enantioselectivity (all >99:1 er). Notably, even the relatively inert iodomethane could successfully underwent coupling to achieve diastereodivergent asymmetric alkylation with high yields, high enantioselectivity and diastereoselectivity (cis-3h), which is still challenging yet of great significance in alkene hydromethylation^91,92.

Fig. 4. Substrate scope to afford the cis-type products, Condition A.

^aTBAF (1.0 mL, 1.0 M in THF), THF (1.0 mL), rt, 2 h. Bn, benzyl; PMP, 4-MeO-C₆H₄.

Next, we examined the substrate scope of spirooxindole substituents. Alkyls (methyl, ethyl), aryls, and tert-Butoxycarbonyl (Boc) protecting group on the nitrogen atom were well-tolerated in the reaction system, maintaining excellent stereoselective control (cis-3i)−(cis-3l). A variety of spirooxindole substituents bearing electron-withdrawing groups (cis-3m)−(cis-3o), electron-donating groups (cis-3p, cis-3q) were readily accommodated. The difluoro-substituted spirocyclic alkene substrate was also successfully applied to this reaction system, delivering the target difluoro-substituted spirocyclic product in high yield with excellent enantioselectivity and diastereoselectivity (cis-3r, 62% yield, >20:1 dr, >99:1 er). The configuration of the product cis-3o was confirmed by X-ray crystallographic analysis. Compatibility studies with other spirocyclic olefins were also conducted, A wide range of spirocyclic olefins, including those derived from indolinones, benzofuranones, 3-isochromanones, pyrrolidones, can be utilized within this reaction system to construct the corresponding spirocyclic compounds that contain 1,3-non-adjacent stereocenters in 55−82% yield with exceptional enantioselectivity (cis-3s)−(cis-3v), although diastereoselectivities of (cis-3w) were relatively lower.

We tested the substrate scope to afford the trans-type products. As shown in Fig. 5, Simple iodoethane, iodopropane, ether- and ester-functionalized alkyl iodides, as well as spiroindolone endo-cyclic alkenes with nitrogen atoms substituted by methyl or phenyl groups, were all successfully applied to this reaction system, affording the target spirocyclic products in 55−95% yield with excellent diastereoselectivity and good enantioselectivity (up to >20:1 dr and 91:9 er) (trans-3a, trans-3x, trans-3y, trans-3l, trans-3z). The electron-donating and electron-withdrawing groups substituted on the spirooxindole could all smoothly yield the desired products (trans-3aa)−(trans-3ad). Other spirocycle skeletons, such as pyrrolidinone, piperidinone and 7-azaindolone derived spirocyclic olefins underwent asymmetric hydroalkylation to obtain corresponding spirocyclic product with excellent diastereoselectivity and good enantioselectivity (trans-3s, trans-3ae, trans-3af, trans-3w, trans-3ag−trans-3ai). We next apply this protocol for the functionalizations of drugs, with respect to alkyl iodides derived from naproxen (used for the treatment of mild to moderate pain and arthritis), asymmetric hydroalkylation proceeded smoothly to afforded corresponding drug-modified spirocyclic product (trans-3aj) in moderate yield and excellent stereoselectivity.

Fig. 5. Substrate scope to afford the trans-type products and late-stage drug modification, Condition B.

^aTBAF (1.0 mL, 1.0 M in THF), THF (1.0 mL), rt, 2 h. Bn, benzyl.

Stereodivergent synthesis of all four stereoisomers

After conducting a thorough investigation of the substrate scope, we initiated an exploration into the potential of constructing all stereoisomers through stereodivergent asymmetric hydroalkylation (Fig. 6).

Fig. 6. Stereodivergent synthesis of all four stereoisomers.

The cis-3a/3ak and trans-3a/3ak compounds were synthesized under standard Conditions A and B, respectively, while their enantiomers were obtained using the enantiomeric forms of ligands L1 and L2 under the corresponding standard conditions.

By merely switching the two chiral ligands to their enantiomers, all four stereoisomers of the spirocyclic product 3a containing 1,3-non-adjacent stereocenters could be precisely obtained in 65−87% yield with 90:10 − 99:1 er and 10:1 − > 20:1dr (cis-3a, ent-cis-3a, trans-3a, ent-trans-3a). We have also extended this reaction system to achieve stereodivergent synthesis of the corresponding spirocyclic product 3ak, showcasing the robust stereoselective control of this catalytic system.

Biological investigations

Lipopolysaccharide (LPS), a major component of the cell wall in Gram-negative bacteria, is widely utilized as an inflammatory inducer⁹³. It was used to mimics the activation of microglia and astrocytes, trigger the release of pro-inflammatory cytokines such as TNF-α, IL-6 and IL-1β via TLR4/NF-κB signaling pathways⁹⁴. In organisms, the nervous system plays a pivotal role in regulating cellular physiological functions. However, stimulation by LPS can induce chronic neuroinflammation. In this study, we synthesized a series of spirolactamides and spironolactones, but their anti-inflammatory effects on LPS-induced microglia cells remained unclear. To investigate whether chiral spirocyclic products exert anti-inflammatory effects on cellular inflammation, we first established an anti-inflammation assay based on LPS-induced mouse microglial cell model. Stimulating with 100 ng/mL LPS⁹⁴, mouse microglia was treated with chiral spirocyclic products at a concentration of 10 μM for 24 h. Several chiral spirocyclic compounds significantly attenuated LPS-induced pro-inflammatory cytokine release in mouse microglia. Among the tested compounds, thirteen inhibited IL-6 release by >30% (excluding cis-3m and trans-3ac). Notably, six compounds—trans-3w, cis-3p, cis-3o, cis-3b, trans-3aa, and cis-3f—exhibited potent IL-6 inhibition >80% (inhibition >50% is generally considered highly active). Further validation in cellular and animal models is ongoing. For TNF-α, 33% of compounds demonstrated >30% inhibition, including trans-3ab, cis-3j, cis-3i, trans-3y, and cis-3f. In contrast, most compounds showed negligible inhibition of IL-1β, with only trans-3ab exhibiting >30% inhibition. Crucially, trans-3ab was the sole compound demonstrating >30% inhibition against all three cytokines (IL-6, TNF-α, and IL-1β). Given the complexity of inflammatory signaling pathways, compounds inhibiting multiple key pro-inflammatory cytokines (e.g., IL-6, TNF-α, IL-1β) suggest potential specificity for those key signaling cascades (Fig. 7).

Fig. 7. The tested compounds affect the gene expression of LPS-induced inflammatory factors in the microglia cell model.

a, c, e qPCR analysis of inflammatory gene expression in microglia treated with compounds. a IL-6, b Inhibition of IL-6, cTNF-α, d Inhibition of TNF-α, e IL-1β, and f Inhibition of IL-1β. All data represent mean ± S.E.M. from three different experiments (n = 3).

Discussion

In summary, we have introduced an DRR in metal hydride catalysis chemistry that not only expands classic mechanistic pathways but also overcomes inherent stereochemical challenge, enabling the diastereodivergent synthesis. Based on this strategy, the cobalt-hydride-catalyzed ligand-modulated asymmetric hydroalkylation of spirocyclic olefins has been developed, enabling enantio- and diastereodivergent synthesis of rigid spirocycles bearing non-adjacent stereocenters, which represents a significant challenge in stereodivergent synthesis. Bioactivity evaluations revealed that several new spirocyclic compounds exhibited excellent anti-inflammatory activity by reversing LPS-induced neuronal inflammation and restoring microglial cell homeostasis in murine models. DFT calculations elucidated the origin of diastereoselectivity inversion, identifying ligand L2-controlled DRR as the key mechanistic determinant.

Methods

Co-catalyzed enantioselective synthesis to afford the cis-type spiral products

To an oven-dried 8.0 mL Teflon-screw cap test tube containing a magnetic stir was charged with CoCl₂ (2.0 mg, 15 mol%) and ligand L1 (7.5 mg, 18 mol%) under Ar atmosphere using glove-box techniques. Subsequently, anhydrous MTBE (0.5 mL) was added, and the mixture was stirred for 15 minutes at room temperature. Then, K₃PO₄·H₂O (58.0 mg, 0.25 mmol, 2.5 equiv.), Spirocyclic olefin 1 (0.10 mmol, 1.0 equiv.), alkyl iodide 2 (0.20 mmol, 2.0 equiv.), and (OMe)₂MeSiH (48.0 µL, 0.30 mmol, 3.0 equiv.) were sequentially added. Afterwards, the tube was sealed with airtight electrical tapes and removed from the glove box and stirred at 0 °C for 48–96 h at 650 rpm. After the reaction was completed, the reaction mixture was diluted with saturated NH₄Cl (aq., 1.0 mL) and ethyl acetate (5.0 mL). The aqueous phase was extracted with ethyl acetate (2 × 5.0 mL) and the combined organic phases were concentrated in vacuo. The crude mixture was purified by flash column chromatography on silica gel using a mixture of PE/ethyl acetate as eluent to obtain the desired product cis-3.

Co-catalyzed enantioselective synthesis to afford the trans-type spiral products

To an oven-dried 8.0 mL Teflon-screw cap test tube containing a magnetic stir was charged with CoCl₂ (2.0 mg, 15 mol%) and ligand L2 (8.5 mg, 18 mol%) under Ar atmosphere using glove-box techniques. Subsequently, anhydrous DME (0.5 mL) was added, and the mixture was stirred for 15 minutes at room temperature. Then, K₃PO₄·H₂O (58.0 mg, 0.25 mmol, 2.5 equiv.), Spirocyclic olefin 1 (0.10 mmol, 1.0 equiv.), alkyl iodide 2 (0.20 mmol, 2.0 equiv.), and (OMe)₂MeSiH (48.0 µL, 0.30 mmol, 3.0 equiv.) were sequentially added. Afterwards, the tube was sealed with airtight electrical tapes and removed from the glove box and stirred at 0 °C for 48–96 h at 650 rpm. After the reaction was completed, the reaction mixture was diluted with saturated NH₄Cl (aq., 1.0 mL) and ethyl acetate (5.0 mL). The aqueous phase was extracted with ethyl acetate (2 × 5.0 mL) and the combined organic phases were concentrated in vacuo. The crude mixture was purified by flash column chromatography on silica gel using a mixture of PE/Ethyl acetate as eluent to obtain the desired product trans-3.

Anti-inflammatory Activity Assays

BV2 microglial cells

BV2 cell line (Serial: GNM45) was obtained from National collection of Authenticated Cell Cultures. STR profiling was used as the authentication procedure for BV2 cell line. Upon lipopolysaccharide (LPS) stimulation, BV2 cells exhibit significantly upregulated gene expression of pro-inflammatory mediators, leading to the secretion of inflammatory cytokines such as TNF-α, IL-6, and IL-1β. These cytokines, particularly TNF-α and IL-6, play critical roles in neuroinflammatory processes by activating downstream inflammatory cascades, thereby exacerbating neurological inflammation.

BV2 cell culture

Cells were cultured under standard conditions: 95% air, 5% CO₂, 37 °C, and 70–80% humidity. Complete culture medium consisted of 88% (w/v) DMEM high-glucose medium, 10% (v/v) fetal bovine serum (FBS), 1% (v/v) L-glutamine, and 1% (v/v) penicillin-streptomycin. For cell resuscitation, a cryovial containing 1 mL cell suspension was thawed in a 37 °C water bath, gently mixed, and transferred to a 15 mL centrifuge tube containing 4 mL complete medium. After centrifugation at 800 rpm for 3 min, the supernatant was discarded, and the pellet was resuspended in 2 mL medium. Subsequently, 1 mL cell suspension was transferred to a T25 flask supplemented with 5 mL complete medium and incubated under standard conditions. Subculturing was performed at 80% confluency. The culture supernatant was removed, and cells were washed with 1 mL PBS, followed by incubation with 1 mL digestion solution at 37 °C for 1 min. Upon detachment (observed as cell rounding), digestion was terminated by adding 2 mL complete medium. The cell suspension was centrifuged (800 rpm, 3 min), resuspended in 1 mL fresh medium, and transferred to a new T25 flask with 5 mL complete medium.

Compound treatment in LPS-Induced BV2 inflammation model

Confluent BV2 cells were harvested, centrifuged, and resuspended. After 10-fold dilution and counting, cells were seeded into 6-well plates at 3.2 × 10⁴ cells/mL and cultured for 12–24 h until 50% confluency. The cells were then treated with 100 ng/mL LPS alone or in combination with test compounds. After 24 h, RNA was extracted for reverse transcription and qPCR analysis, or culture supernatants were collected for ELISA.

qPCR analysis of inflammatory cytokine expression

RNA extraction from BV2 cells

After removing the medium, cells were washed with 500 µL PBS and lysed with 500 µL Buffer RL. Lysates were transferred to FastPure gDNA-Filter Columns III, centrifuged at 12,000 rpm for 30 s, and the filtrate was collected. RNA was precipitated by adding 250 µL absolute ethanol to the filtrate. The mixture was loaded onto FastPure RNA Columns III and centrifuged (12,000 rpm, 30 s). Columns were sequentially washed with 700 µL Buffer RW1 and 700 µL Buffer RW2 (with ethanol), followed by a final wash with 500 µL Buffer RW2 (centrifugation at 12,000 rpm, 2 min). After air-drying (12,000 rpm, 1 min), RNA was eluted with 50 µL RNase-free MQ H₂O and stored at −80 °C.

Reverse transcription

Genomic DNA was removed by incubating 1 µg RNA with 4 µL 4× gDNA Wiper Mix and RNase-free MQ H₂O (total 16 µL) at 42 °C for 2 min. Reverse transcription was performed by adding 4 µL 5× HiScript III qRT SuperMix and incubating at 37 °C for 15 min, followed by 85 °C for 5 s. cDNA was stored at 4 °C for immediate use.

Quantitative Real-Time PCR (qPCR)

Reaction mixtures contained 10 µL 2× Taq Pro Universal SYBR qPCR Master Mix, 0.4 µL each of forward and reverse primers (5 µM), 2 µL cDNA, and RNase-free H₂O to 20 µL. Cycling conditions: 95 °C for 30 s (pre-denaturation); 40 cycles of 95 °C for 10 s and 60 °C for 30 s; followed by a melting curve analysis (instrument default settings).

Data analysis

Cycle threshold (Ct) values were normalized to housekeeping genes (ΔCt = Ct_target − Ct_reference). Relative expression levels were calculated using the 2^(−ΔΔCt) method (ΔΔCt = ΔCt_sample − ΔCt_control). Primer sequences are listed in Table 2.

Table 2.

Primer sequence for IL-1β, IL-6 and TNF-α

gene	primer sequence
IL-1β	FORWARD: TCGCAGCAGCACATCAACAAGAG
	REVERSE: AGGTCCACGGGAAAGACACAGG
IL-6	FORWARD: CTTCTTGGGACTGATGCTGGTGAC
	REVERSE: TCTGTTGGGAGTGGTATCCTCTGTG
TNF-α	FORWARD: CGCTCTTCTGTCTACTGAACTTCGG
	REVERSE: GTGGTTTGTGAGTGTGAGGGTCTG

The forward primer sequence for IL-1β is shown as SEQ ID NO.1, and the reverse primer sequence for IL-1β is shown as SEQ ID NO.2. The forward primer sequence for IL-6 is shown as SEQ ID NO.3, and the reverse primer sequence for IL-6 is shown as SEQ ID NO.4. The forward primer sequence for TNF-α is shown as SEQ ID NO.5, and the reverse primer sequence for TNF-α is shown as SEQ ID NO.6.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Z.H.S. conceived and directed the project. Z.S., J.R., K.Z. and Y.L. performed the chemistry experiments. J.X., X.C. and J.L. performed bioactivity experiments. Z.Z. and J.H. performed the DFT studies. Z.S., Z.Z., J.X., Y.O., Y.J. and F.P. analyzed the results and wrote the manuscript.

Peer review

Peer review information

Nature Communications thanks Rambabu Chegondi, Xiuxiu Li, Lisa Roy and Xiaolong Yang for their contribution to the peer review of this work. A peer review file is available.

Data availability

The authors declare that the data supporting the findings of this study are available within the article and the Supplementary Information as well as from the authors upon request. The coordinates of the optimized structure are available from the source data. The X-ray crystallographic coordinates for structure(CoCl₂·L1), (cis-3o) and (trans-3ab) reported in this study have been deposited at the Cambridge Crystallographic Data Centre(CCDC), under deposition numbers CCDC 2328562, CCDC 2328807 and CCDC 2395424. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif. Source data are provided with this paper.

Competing interests

The patent application (patent number: CN202510465238.X) was filed by Z.S., Z.S. and J.L. from Yunnan University and J.X., X.C. and Z.L. from Shanghai University of Traditional Chinese Medicine. The patent application covered through stereodivergent hydroalkylation to access chiral spirocyclic compounds and their application into effects against LPS-induced neuronal inflammation. The remaining authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-64491-y.