-
Medium-sized rings typically refer to non-aromatic heterocyclic structures composed of 7 to 9 atoms [1−4]. These frameworks are prevalent not only in biologically active natural products, such as laulimalide, salicylihalamide, and manzamine alkaloids, but also serve as versatile scaffolds in drug discovery owing to their unique conformational and electronic properties [5−8]. Medium-sized rings offer diverse functionality and stereochemical complexity compared to the more common small rings including only five- and six-membered and larger macrocycles, endowing them with the ability to engage protein targets with high affinity and selectivity, while retaining sufficient bioavailability to access intracellular compartments [9−12]. Such advantageous properties have spurred a growing interest in the synthesis and biomedical applications of medium-sized rings and macrocycles, both in academic research and pharmaceutical development [13, 14]. In the past five years, a variety of drugs containing middle-ring structures have been approved by the FDA, covering multiple fields such as anti-infection, anti-cancer, and immune regulation [4]. In recent years, extensive attention has been directed toward the study of various isomers within medium-sized heterocycles, particularly in systems where minor structural modifications can significantly impact chemical properties [15−17]. Medium-sized heterocyclic compounds often possess multiple chiral centers or unique conformations, making the selection and control of isomers critical factors influencing pharmacological efficacy and drug properties [18−20].
Among these medium-size rings, LY-411575 (FIG. 1(a)), a γ-secretase inhibitor, has played a pivotal role in Alzheimer’s disease research [21, 22]. It has a 7-membered-ring dibenzolactam, a dibenzo[bI,d]azepin-6-one moiety (1 in FIG. 1(b)). The stereochemistry of 1 exhibits intriguing complexity due to the combination of the C7 chiral center and the axial chirality generated by the sp2-sp2 axis of the biphenyl system. Previous studies have successfully separated the a-R- and a-S-atropisomers of 7-, 8-, and 9-membered dibenzazepinones (compounds 1–3 in FIG. 2(a)) using chiral high-performance liquid chromatography (HPLC), and the absolute configuration of 1–3b-S has been unambiguously determined by single-crystal X-ray diffraction analysis [23]. The thermal stability of these isomers is commonly characterized by their interconversion free energy barrier (ΔG‡) from R- to S-, and existing research has demonstrated pronounced differences in the stability of the 7-, 8-, and 9-membered ring systems. Intriguingly, the 8-membered ring (2a) is found to exhibit unusually the highest configurational stability compared to its 7- and 9-membered analogs. However, the origin of this stability trend remains poorly understood. Since typically only one atropisomer exhibits the desired pharmacological activity, while interconversion (e.g., R to S) often leads to loss of efficacy or even toxic effects, it is essential to design molecules with sufficiently high rotational barriers to preserve a stable conformation.
To elucidate the structural and energetic factors controlling these phenomena, we used density functional theory (DFT) calculations including a solvation model for both geometry optimization and relative free energies calculation for atropisomers and their methylated derivatives. The results were systematically compared with the experimental data on the free-energy barrier to rotation and the stereoselective C7 methylation reactivity. Our analysis reveals how ring expansion affects the stability against racemization, the thermodynamic preferences of atropisomers, and the stereochemical outcomes of methylation reactions, thereby elucidating the inherent conformational properties of each ring system.
-
Referring to the recent work by Houk et al. [24], the combination of Truhlar’s Minnesota hybrid functional M06-2X [25], Grimme’s GD3 dispersion correction [26], and the Pople’s 6-311G basis with double diffusion and polarization (d,p) was adopted for both optimization and frequency analysis. The Truhlar’s solvation model SMD [27] was used to simulate the solvation effect corresponding to the experimentally used solvent toluene. To overcome the inaccuracy of harmonic frequencies for such flexible ring structures, and thereby obtain free energies, we have adopted quasi-rigid-rotor-harmonic-oscillator (QRRHO) model [28] for anharmonic frequency calculations. For locating all transition states (TS), we used a relaxed scan along the dihedral angle C8-C7-C6-C5 first, then extracted the structure with the highest potential energy, and finally used the TS optimization to further relax the structure. For comparison, we also used the popular combination of Becke’s three parameter exchange functional [29, 30] with Lee-Yang-Parr’s correlation functional [31], combined with 6-311++G(d,p) basis, and second-order Møller–Plesset perturbation theory (MP2) [32] combined with Pople’s 6-31G basis with d-polarization functions for heavy atoms (6-31G*) [33]. For B3LYP/6-311++G(d,p) calculation, we used Grimme’s dispersion correction GD3 and solvation model SMD, and QRRHO for correction of anharmonicity of low frequency modes. For MP2/6-31G* calculation, we also used solvation model SMD and QRRHO model. The reaction equilibrium constant was calculated by
$K = {k_1}/{k_2} = \text{e}^{ - \Delta G/RT}$ [34]. The rate coefficients were calculated by transition state theory${k_{{\text{TST}}}} = \dfrac{{{k_{\text{B}}}T}}{h}{\text{e}^{ - \Delta G/RT}}$ [35].All M06-2X DFT calculations were performed using the Gaussian 16, Revision A.01 [36]. All B3LYP DFT calculations were performed using the BDF program [37, 38]. The QRRHO was calculated using Shermo program [39]. All the free energy barrier calculations were performed at the same temperature as the experiments. The molecules were visualized using Chimera [40].
-
FIG. 2 shows the structures of 1a, 1b, 2a, and 3a, as well as their optimized TS. Table I presents a detailed comparison of the energetics between the calculated ΔG‡ and experimental values. Considering M06-2X/6-311++G(d,p) results, the enantiomer of 1a (R=H) exhibits stereochemical stability toward racemization, with a calculated ΔG‡ value of 25.18 kcal/mol. Notably, our results indicate that the 8-membered ring 2a is more stable, with a ΔG‡ value of 27.86 kcal/mol, while the 7-membered (1a) (25.18 kcal/mol) and 9-membered (3a) (26.64 kcal/mol) rings are less stable. The B3LYP results show the interconversion free energy barrier of 2a is the highest, which is consistent with the experimental observations. However, compared to the M06-2X results, B3LYP still overestimates the interconversion barrier of 3a. MP2 calculations revealed that 2a has the highest stability, which is consistent with the experimental observations and the M06-2X results. However, the MP2 results significantly underestimate the differences in free energy barriers among 1a, 2a, and 3a. Overall, the M06-2X method remains an ideal choice between accuracy and efficiency. So, in further calculations, we use the protocol of M06-2X/6-311++G(d,p). From the rate calculations based on M06-2X results, the half-life at room temperature for the isomerization of 2a is 8750 h, which is about one year, whereas that of 1a is 94 h (about 4 days). Our M06-2X results support the experimental findings that the 8-membered ring has the strongest stereochemical stability. Tabata et al. qualitatively attributed this stability to the deeper and more rigid cage structure of 2a, but this reason cannot explain why 3a is also instable [23]. Here, we quantitatively describe the cage depth of the 7-membered, 8-membered, and 9-membered rings using the dihedral angle of C8-C7-C6-C5 (D1) and C11-C12-N1-C16 (D2), as shown in Table II. The results show that the 9-membered ring of 3a has the deepest cage structure, with dihedral angles of 86° for D1 and 89° for D2. Despite this, the longer continuous methylene chain makes the 9-membered ring more flexible, requiring only a 24° for D2 to invert from 3a-R to its TS. In contrast, the cage structure of the 8-membered ring is more rigid, requiring a deeper rotation for inversion. It is therefore reasonable that the 8-membered ring has stronger stereochemical stability.
-
Having finished the discussion about the stability issues of these enantiomers, we now try to answer how the substitution affects the stability of the 7- and 8-membered ring diphenylmethane derivatives. The substitution occurs at the C7 position of the 7- and 8-membered ring diphenylmethane 1b and 2a and influence the stereochemistry of the entire molecule. As shown in FIG. 3(a), compounds 1b and 2a are methylated to introduce an asymmetric center at the C7 position. Such methylation is expected to generate two diastereoisomers (17b-R (7R), 17b-S (7S) and 18a-R (7R), 18a-S (7S)). Experimental reports indicate that this methylation exhibits strong selectivity, with only 17b-R (7R) being produced in 99% yield, while 18a-R (7R) is obtained in 79% yield [23]. Table III presents a comparison of our calculated ΔG‡ and equilibrium concentration ratios for 17b and 18a at 80 °C with experimental values. From the perspective of free energy, it is reasonable that the 7R isomer has a lower energy and a higher yield. For 17b, the calculated R:S equilibrium concentration ratio at 80 °C is 2.9:97.1, which agrees well with the experimental data. The calculated interconversion barrier between 17b-R and 17b-S is 27.39 kcal/mol. In contrast, for 18a, the R:S equilibrium constant at 80 °C is 15.5:84.5, with an interconversion barrier of 25.17 kcal/mol. Both the experimental and theoretical results indicate that the rotation free energy barriers of the methylated products are lower for both 1b and 2a. This suggests that although the isomerization occurs along the sp2-sp2 axis of the biphenyl system, the barriers of rotation is significantly influenced by the overall stereochemical stability of the molecule. As proposed in the Ref.[23], the C7-methyl group lowers the free energy barrier by raising the ground state of the pseudoaxial conformer. Our theoretical calculations support this conclusion, as reflected by the lower interconversion potential barriers calculated for the methylated products.
-
Overall, the computed free-energy barriers are in close agreement with the experimental values, although a systematic deviation persists for compounds 1−3. The root-mean-square error (RMSE) and mean absolute error (MAE) of our results relative to the experimental values are 1.43 kcal/mol and 1.28 kcal/mol, respectively. The major source of error originates from compound 1−3. The discrepancy between experiment and theory is shown in detail in FIG. 4, where the deviation is quantitatively fitted. The calculations of 7-, 8-, and 9-membered ring dibenzolactams 1−3 systematically overestimate the barrier heights. One likely reason is that the flexible, long carbon chains contain many low-frequency vibrational modes, which causes the zero-point energy and thermal corrections to be underestimated. And the QRRHO can only elevate the deviation within limited accuracy, since we have not used more sophisticated methods to calculate the frequencies, such as vibrational perturbation theory [41]. Another possible source is the intrinsic error between the chosen exchange-correlation functional and the solvation model. In our molecule, the C=O group and sp3 N group are both hydrogen bond acceptors. Electrostatic potential analysis shows that the region near the C=O group in dibenzolactams 1–3 is negatively charged, indicating that it serves as the primary hydrogen bond acceptor. So, the ignorance of explicit water molecules around them would introduce non-neglectable error [42]. Nevertheless, as shown in FIG. 4, our computational results are fitted using linear regression, yielding a slope of 0.90. The coefficient of determination (R2) is 0.97. The calculated relative errors are small, and the ordering of the free energy barrier heights is correct. For the methylated products 17b and 18a, Mulliken charge analysis (in Table S1 in Supplementary materials, SM) reveals that the oxygen atoms bear lower partial charges compared to those in the corresponding compounds 1b and 2a, and the sp3 N atom is buried with both methyl groups. As a result, hydrogen bonding interactions are relatively weaker. In this case, the solvation model effectively describes the system, leading to calculated ΔG‡ that are in great agreement with experimental values.
-
In this study, we investigated the stereochemical stability of 7-, 8-, and 9-membered dibenzazepinone rings using density functional theory (DFT) calculations, providing valuable insights into their thermal stability reflected by free energy barriers (ΔG‡) and the factors influencing their stability. The results confirm that the 8-membered ring exhibits the highest stereochemical stability, attributable to its more rigid cage-like structure. In contrast, despite possessing the deepest cage structure, the 9-membered ring is less stable due to its increased flexibility. For compounds 17b and 18a, the introduction of an asymmetric center via methylation leads to a pronounced enantiomeric stereoselectivity, which is consistent with experimental results. By comparing the computed free energy barriers with experimental observations, we are able to accurately predict the enantiomeric distribution of the methylated products. The calculated results are in good agreement with experimental data, validating the theoretical model employed. These findings contribute to a deeper understanding of the structural and energetic factors governing the stability and reactivity of medium-sized heterocyclic compounds.
Stability of Atropisomers in Medium-Sized Ring: DFT Insights of Dibenzolactam Chirality
- Received Date: 03/06/2025
- Available Online: 27/10/2026
-
Key words:
- Quantum chemistry /
- Reaction mechanism /
- Axis chirality /
- Drug design
Abstract: Medium-sized rings have emerged as critical scaffolds in the field of drug development. Owing to their unique conformational features and the existence of multiple isomeric forms, even subtle structural modifications can give considerable rise to pronounced differences in chemical properties. This study presents a comprehensive theoretical analysis of the free energy barriers to interconversion of R-series and S-series of enantiomers (ΔG‡) governing the atropisomerization and stereochemical stability of 7- (1a), 8- (2a), and 9- (3a) membered ring dibenzolactams using density functional theory (DFT), which complement experimental findings. The computed ΔG‡ values for racemization agree with experimental trends, with a uniform deviation for each compound calculated, revealing enhanced stability for the 8-membered ring derivatives (ΔG‡=27.86 kcal/mol) compared to 7- (ΔG‡=25.18 kcal/mol) and 9-membered (24.39 kcal/mol) analogs, attributed to its rigid cage-like conformation. The enhanced conformational flexibility of 9-membered ring dibenzolactams (3a) leads to a further reduction in their ΔG‡, thereby facilitating a more facile crossing of the transition state. Based on the ΔG‡ for methylated derivatives (17b and 18a), we have successfully reproduced the equilibrium constants reported experimentally, and our computed free-energy barriers exhibited trends consistent with those observed in experiment.
首页
登录
注册


DownLoad: