Introduction

ACC is an enzyme contributing to the regulation of fatty acid biosynthesis and oxidation and consists of two isoforms, ACC1 and ACC21. It is also an important mediator of carbohydrate, amino acid and lipid metabolism. Abnormalities in metabolism related to the roles of ACC1 and ACC2 lead to excess fatty acids in the body and cause metabolic syndrome (MetS)2.

MetS is a series of cardiovascular risk factors, including abdominal fat, glucose intolerance, triglyceride increment and hypertension, related to an increment in crude fat content in the human body. This leads to a higher chance of stroke, type 2 diabetes and cardiac death. According to the National Health and Nutrition Examination Survey (NHANES), during the past decade, the proportion of men and women aged between 12 and 19 with MetS has increased by more than 50%. In addition, there are approximately more than 200 million adolescents worldwide who have suffered from MetS today2. In supporting the treatment of fatty acid metabolic disorder at the molecular level, the inhibition of ACC activity plays an important role in constraining the synthetic biology reaction rate of fatty acids (ACC1) and promoting their oxidation process (ACC2) in the human body3.

Many studies have demonstrated that ACC inhibition offers potential in controlling fatty acid biosynthesis, leading to the development of new solutions for the prevention and treatment of MetS-related diseases. Studies by Jeffrey and Matthew (2014) reported an ACC inhibitor model performed on yeast4. Accordingly, these inhibitors have diverse and complex structures, including bipiperidylcar-boxamide derivatives, polyketides (soraphen A), aryloxyphenoxypropionate (haloxyfop), cyclohex-anedione (sethoxydim), synthetic acyl-CoA fatty acid chains (TOFA, MEDICA), competitive substrate inhibitors (CABI-CoA), acylsulfonamides and related derivatives. In addition to the above inhibitor groups, many studies by Taisho, Sanofi-Aventis, AstraZeneca, Takeda, Pfizer, Boehringer Ingelheim, Haselkorn, Nimbus and Amgen, which were performed on yeast and mice, resulted in several other nonselective inhibitor structures that can be attached to both the carboxyl transferase (CT) and biotin carboxylase (BC) active sites at the same time.

Inheriting the results from our previous research about "the effective model of human acetyl-CoA carboxylase inhibition by aromatic-structure inhibitors"5, the aim of this research is to reveal the inhibition mechanism of the enzyme by studying the interaction direction, preferential binding sites of aromatic ligands on the two isoforms and conformational changes of enzyme-ligand complexes in the body’s environmental conditions (temperature 37 degrees Celsius and water solvent).

Software and Methods

Identification of binding sites on ACC1 and ACC2

The two isoforms are structurally extracted from the RCSB protein data bank7: 2YL2 (BC), 4ASI (CT) for ACC1 (Figure 1) and 3JRX (BC), 3FF6 (CT) for ACC2 (Figure 2).

To identify binding sites on ACC1, we utilized Discovery Studio Visualizer 2017 R2 client software8 for 32 ligand binding sites (Figure 3). Accordingly, combined with docking and virtual screening of ligands corresponding to 4 different methods by AutoDock9 and AutoDock Vina 10, including simulated annealing, genetic algorithm, Lamarckian genetic algorithm and hybrid-scoring function, another 12 ligand binding sites were detected.

In the case of ACC2, 13 ligand binding sites were detected, including 5 binding sites available from X-ray diffraction data (Figure 4) and 8 binding sites from virtual screening.

Raccoon9 is utilized to run virtual screening of ligand structures by receiving parameters from AutoDock. The docking grid volume is determined by two methods: preselected ligand binding sites identified by Discovery Studio and the grid overlaid with the entire enzyme structure.

Cygwin was utilized to run virtual screening of ligand structures by receiving parameters from Autodock vina, including the enzyme.pdbqt, the ligand.pdbqt and the docking grid volume determined on each preselected ligand binding site identified by Discovery Studio.

Results

QSAR enzyme-inhibition models

The results of our previous research show that in the case of ACC1, the hybrid neural fuzzy inference system (HYFIS) model (Figure 5) of the compounds optimized by the PM3 method indicates that four descriptors, GATS2c, JGI5, ETA_EtaP, and CIC5, correlate closely with logIC (R2 = 84.1%). The higher the ACC1 inhibitory activity is, the higher the number of aromatic rings and the higher the number of atoms belonging to the small period, which have more electronegativity and electron resonance effects, such as O atoms. In the case of ACC2, the HYFIS model (Figure 6) of the compounds optimized by the DFT method indicates that the four descriptors, GATS6s, GATS1e, MATS3s, and GATS3s, correlate closely with logIC (R2 = 90.8%). The higher the ACC2 inhibitory activity is, the more C=O or C=N double bonds, the higher the number of atoms that have the large Sander- son electronegativity (O>N>S) and the small Sanderson electronegativity difference between first order interleaved atoms (the preferable N-X-H structure due to O-X-H > N-X-H).

Correlation between binding energies at preferential binding sites and the experimental log(IC) of ligands on ACC1 and ACC2

For both ACC1 and ACC2, the BE values of ligands are closely correlated with the experimental log(IC), which has statistical significance (< 0.05): R2ACC1 = 96.3% (Figure 7) and R2ACC2 = 78.9% (Figure 8).

Investigation of enzyme-ligand complexes in the body’s environmental conditions

The binding patterns of the enzyme average conformations, including the enzyme-ligand complexes associated with Pfizer6 having the highest inhibitory activity and Astra5 — Astra1 having the lowest inhibitory activity on ACC1 (Figure 9) and ACC2 (Figure 10), were analyzed by Discovery Studio.

The preferential binding regions on ACC1 and ACC2

The preferential binding sites of ligands reveal the five reaction niches on ACC1 (Figure 11), involving the two preferential binding regions in the order of amino acids 152 to 643 and 1632 to 2100 (Figure 13), and the four reaction niches on ACC2 (Figure 12), involving the two preferential binding regions in the order of amino acids 270 to 704 and 1831 to 2299 (Figure 14).

Discussion

The suppression of ACC activity is crucial in limiting the synthetic biology reaction rate of fatty acids (ACC1) and encouraging their oxidation process (ACC2) in the human body, which supports therapy of fatty acid metabolic disorder at the molecular level. Numerous studies have shown that ACC inhibition has the capacity to regulate fatty acid production, opening up new avenues for the prevention and treatment of MetS-related disorders. Studies of Cho. (2008) revealed the crystal structure of the BC and CT domains of human ACC213. Timothy's research (2014) discovered the domain crystal structure of some microorganism ACCs14. Jeffrey and Matthew (2014) reported an ACC inhibitor model performed on yeast that resulted in several nonselective inhibitor structures that can be attached to both the carboxyl transferase and biotin carboxylase active site at the same time4. In addition, Chen (2019) reviewed and updated some human ACC inhibitors in clinical studies3. All of these studies have focused on studying structures and ACC inhibition models in yeasts and microorganisms or determining the experimental activity of human ACC inhibitors based on clinical studies. Therefore, the inhibition mechanism of human ACC has not yet been elucidated.

The findings of this study indicate that the inhibition mechanism of the human ACC is explained through the preferential binding sites of ligands on the two isoforms and conformational changes of enzyme-ligand complexes in the body’s environmental conditions. Accordingly, ligands prefer to interact with amino acids 152 to 643 and 270 to 704 on the biotin carboxylase domain and amino acids 1632 to 2100 and 1831 to 2299 on the carboxyl transferase domain of the human ACC, described as the five reaction niches (A – E) on ACC1 and the four reaction niches (F – I) on ACC2. These calculation reaction sites are consistent with the research results of Cho. (2008) about the crystal structure of the BC and CT domain of the human ACC2. In addition, there is a close correlation between the experimental log(IC), based on the research of Matthew (2014) and Chen (2019), and the calculated binding energy values of ligands with statistical significance (< 0.05). The conformational changes of enzyme-ligand complexes in the body’s environmental conditions, as the new finding of this study, show that each enzyme-ligand complex has an unfavorable bump and breaks three types of unstable bonds: Pi-Pi bonds between the conjugated Pi system of amino acids on the enzyme and the conjugated Pi system of aromatic rings on the ligand, sulfur-Pi bonds between the conjugated Pi system of amino acids on the enzyme and sulfur on the ligand, and Pi-amide bonds between amide groups of amino acids on the enzyme and the conjugated Pi system of aromatic rings on the ligand. In addition, four new types of bonds are formed: hydrogen bonds, Pi-alkyl bonds between the alkyl group of amino acids on the enzyme and the conjugated Pi system of aromatic rings on the ligand; Pi-cation bonds between the cation of amino acids on the enzyme and the conjugated Pi system of aromatic rings on the ligand; and alkyl-alkyl bonds. Therefore, ligands have fewer stereogenic interactions, and the easier it is to form these types of bonds, the higher their inhibitory activity. However, the mechanism of bond breaking and formation has not been clarified and will be reviewed in our future research.

Conclusions

The results of the study have solved the main objectives, including the interaction direction, preferential binding sites of aromatic ligands and conformational changes of enzyme-ligand complexes in the body’s environmental conditions.

The research reveals the five reaction niches on ACC1 and the four reaction niches on ACC2, including the two preferential binding regions in the order of amino acids: 152 to 643 and 1632 to 2100 for ACC1, 270 to 704 and 1831 to 2299 for ACC2.

The results of molecular dynamic simulation of enzyme-ligand complexes in the body’s environmental conditions show that each complex has an unfavorable bump and breaks three types of unstable bonds, including Pi-Pi, sulfur-Pi and Pi-amide bonds. In addition, each complex forms four new types of bonds: hydrogen, pi-alkyl, pi-cation and alkyl-alkyl bonds. If ligands have fewer stereogenic interactions and are ready to form four types of bonds, including hydrogen, pi-alkyl, pi-cation and alkyl-alkyl bonds, their inhibitory activity will be higher.

Abbreviations

ACC: acetyl-CoA carboxylase; BC: biotin carboxylase; CT: carboxyl transferase; MetS: metabolic syndrome; QSAR: quantitative structure-activity relationship; HYFIS: hybrid neural fuzzy inference system; BE: binding energy; IC: half maximal inhibitory concentration.

Competing interests

The authors declare that they have no competing interests.

Funding

This research was funded by Ho Chi Minh City, University of Science.

Authors’ contributions

CMNT carried out calculations and data analysis. TTB conceived of the study and participated in research coordination. TLX provided theories of the research. XTTL drafted the manuscript and designed and supported the research. The authors read and approve the manuscript.

Acknowledgments

None.