Caffeine derivatives as promising multi-target anti-Alzheimer’s agents: Investigating dual inhibition of AChE and BChE

2.2.1 1,3-dimethyl-7-[(pyridin-3-yl)methyl]-1H-purine-2,6-dione (3a)

White solid. Yield: 88%. MS (ESI): m/z = 294.10 [M + Na]⁺, calculated m/z= 271.11. IR (νmax/cm^–1) 3094 (CH), 1695 (C=O), 1625 (C=C), 1563 (C=N). ¹H NMR (CD₃Cl, 500 MHz): δ 8.56 (d, J = 4.5 Hz, 1H, H₂₂), 7.84 (s, 1H, H₂₇ theophylline), 7.70 (td, J₁= 2 Hz, J₂ = 7 Hz, 1H, H₂₄), 7.44 (d, J = 4.5 Hz, 1H, H₂₃), 7.25 (td, J₁= 1 Hz, J₂ = 4.5 Hz, 1H, H₂₁), 5.60 (s, 2H, CH₂, H₂₅, H₂₆), 3.60 (s, 3H, CH₃, H₃₁, H₃₂, H₃₃), 3.40 (s, 3H, CH₃, H₂₈, H₂₉, H₃₀). ¹³C NMR (CDCl₃, 125 MHz) δ: 155.3 (C₄), 154.6 (C₁₆), 151.7 (C₁₄), 149.8 (C₂), 148.8 (C₁₁), 141.8 (C₉), 137.3 (C₆), 123.4 (C₅), 122.8 (C₁), 106.7 (C₁₂), 51.5 (C₈), 29.8 (C₂₀), 28.0 (C₁₉). Elemental analysis (%) calc for C₁₃H₁₃N₅O₂: C 57.56; H 4.83; N 25.82. Found: C 57.51; H 4.86; N 25.78.

2.2.2 1,3-dimethyl-7-[(pyridin-4-yl)methyl]-1H-purine-2,6-dione (3b)

White solid. Yield: 65%. MS (ESI): m/z = 272.11 [M + H]⁺, calculated m/z = 271.11. IR (νmax/cm^–1) 3125 (CH), 1695 (C=O), 1628 (C=C), 1563 (C=N). ¹H NMR (CD3Cl, 500 MHz): δ 8.60 (s, 1H, H₂₂), 8.55 (d, J= 4.5 Hz, 1H, H₂₁), 7.68 (d, J = 8.5 Hz, 1H, H₂₃), 7.60 (s, 1H, H₂₇ theophylline), 7.22 (dd, J1= 4.5 Hz, J2 = 8.5 Hz, 1H, H₂₄), 5.50 (s, 2H, CH₂, H₂₅, H₂₆), 3.55 (s, 3H, CH₃, H₃₁, H₃₂, H₃₃), 3.36 (s, 3H, CH₃, H₂₈, H₂₉, H₃₀). ¹³C NMR (CDCl3, 125 MHz) δ: 155.2 (C₁₆), 151.5 (C₁₄), 150.0 (C₃), 149.2 (C₁), 148.9 (C₁₁), 140.7 (C₉), 135.7 (C₄), 131.2 (C₅), 123.8 (C₆), 106.7 (C₁₂), 47.7 (C₈), 29.8 (C₂₀), 28.0 (C₁₉). Elemental analysis (%) calc for C₁₃H₁₃N₅O₂: C 57.56; H 4.83; N 25.82. Found: C 57.60; H 4.89; N 25.84.

2.2.3 1,3-dimethyl-7-[(pyridin-2-yl)methyl]-1H-purine-2,6-dione (3c)

White solid. Yield: 66%. MS (ESI): m/z = 272.11 [M + H]⁺, calculated m/z = 271.11. IR (νmax/cm^–1) 3109 (CH), 1687 (C=O), 1625 (C=C), 1570 (C=N). ¹H NMR (CD₃Cl, 500 MHz): δ 8.68 (d, J= 4 Hz, 2H, H₂₁, H₂₄), 7.73 (s, 1H, H₂₇ theophylline), 7.23 (d, J = 4 Hz, 2H, H₂₂, H₂₃), 5.60 (s, 2H, CH₂, H₂₅, H₂₆), 3.68 (s, 3H, CH₃, H₃₁, H₃₂, H₃₃), 3.46 (s, 3H, CH₃, H₂₈, H₂₉, H₃₀). ¹³C NMR (CDCl₃, 125 MHz) δ: 155.2 (C₁₆), 154.4 (C₁₄), 151.6 (C₂, C₆), 148.8 (C₁₁), 142.0 (C₄), 139.0 (C₉), 122.4 (C₃, C₅), 106.6 (C₁₂), 50.6 (C₈), 29.9 (C₂₀), 27.9 (C₁₉). Elemental analysis (%) calc for C₁₃H₁₃N₅O₂: C 57.56; H 4.83; N 25.82. Found: C 57.58; H 4.89; N 25.78.

2.3.1 1,3-dimethyl-7-[(4-{[2-(3,7-dimethyl-2,6-dioxo-1,2,3,6-tetrahydro-1,3,7-triazin-1-yl)ethyl]amino}pyrimidin-2-yl)methyl]-3,7-dihydro-1H-purine-2,6-dione (3d)

White solid. Yield: 95%. MS (ESI): m/z = 464.17 [M + H]⁺, calculated m/z = 463.17. IR (νmax/cm^–1) 3125 (CH), 1669 (C=O), 1687 (C=C), 1570 (C=N). ¹H NMR (CD₃Cl, 500 MHz): δ 7.75 (t, J= 8 Hz, 1H, H₃₇), 7.73 (s, 2H, H₄₀, H₅₀ theophylline), 7.41 (d, J = 8 Hz, 2H, H₃₅, H₃₆), 5.56 (s, 4H, CH₂, H₃₈, H₃₉, H₄₈, H₄₉), 3.61 (s, 6H, CH₃, H₄₄, H₄₅, H₄₆, H₅₄, H₅₅, H₅₆), 3.36 (s, 6H, CH₃, H₄₁, H₄₂, H₄₃, H₅₁, H₅₂, H₅₃). ¹³C NMR (CDCl₃, 125 MHz) δ: 155.1 (C₂, C₄), 151.6 (C₁₆, C₂₇), 150.5 (C₁₄, C₂₉), 149.0 (C₁₁, C₂₄), 144.3 (C₉, C₂₆), 141.0 (C₆), 121.9 (C₁, C₅), 106.8 (C₁₂, C₂₃), 49.0 (C₈, C₂₁), 29.9 (C₂₀, C₃₄), 28.0 (C₁₉, C₃₃). Elemental analysis (%) calc for C₂₁H₂₁N₉O₄: C, 54.42; H, 4.57; N, 27.20. Found: C 54.41; H 4.58; N 27.20.

2.5.1 In vitro AChE and BChE inhibition assays

The ChEI activities of the synthesized compounds were evaluated against AChE and BChE following the protocols provided with commercially available kits. Stock solutions of the test compounds were prepared in high concentration and serially diluted in the corresponding assay buffer, with galantamine employed as the reference inhibitor. For AChE inhibition, 10 µL of 20X diluted samples was added to each well of a 96-well microplate, followed by 10 µL of enzyme solution and the chromogenic substrate mixture. The absorbance was monitored at 412 nm for 40 min at room temperature (Ellman et al., 1961). BChE inhibition was assessed in black 96-well plates by combining 100 µL of each test solution with 50 µL of a fluorescent reagent containing butyrylthiocholine. After a 20 min incubation, fluorescence intensity was recorded at excitation and emission wavelengths of 390 nm and 510 nm, respectively (Soliman et al., 2023; Gabr & Brogi, 2020). All experiments were conducted in triplicate, and IC₅₀ values were calculated using non-linear regression after background correction. Statistical analyses were performed using GraphPad Prism software (version 9.0.1) (GraphPad Software, San Diego, California, USA). One-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison test was applied to determine statistically significant differences between the tested compounds and the reference standard. A p < 0.05 was considered statistically significant.

2.5.2 Docking study

To evaluate the synthesized compounds as potential ChEI, molecular docking studies were performed against AChE (PDB ID: 4M0E) and BChE (PDB ID: 1P0P) (Ullah et al., 2024). Protein structures were prepared using AutoDockTools, wherein all crystallographic water molecules and co-crystallized ligands were removed, followed by the addition of polar hydrogen atoms and assignment of Gasteiger partial charges. The ligand structures were generated using ChemDraw, then geometry-optimized at the DFT/B3LYP/6-311G(d,p) level prior to docking. Docking simulations were carried out in PyRx (v0.8) (Kondapuram et al., 2021; Chandramohan et al., 2023) employing AutoDock Vina (v1.5.7) (Trott & Olson, 2009), with energy minimization executed under default parameters. The optimal binding conformations were selected based on the lowest binding free energy and key molecular interactions within the active site. Visualization and interaction analyses of ligand-protein complexes were performed using Discovery Studio Visualizer (v21.1.0.20298) (Dassault Systèmes 2021).

2.5.3 ADME-T prediction

Computational predictions of the pharmacokinetic properties of the newly synthesized compounds were performed using the SwissADME online server to evaluate absorption, distribution, metabolism, and excretion (ADME) profile (Daina et al., 2017). Additionally, the potential toxicity of these compounds was assessed using the ProTox-II platform (Banerjee et al., 2018).

3.2.1 ¹H and ¹³C NMR spectral analysis

RMN spectroscopy is a powerful analytical tool for elucidating the structural and functional characteristics of molecules, providing detailed insights into their chemical environments. The DFT has proven particularly valuable for predicting NMR spectra accurately, facilitating the correlation between molecular structure and chemical shifts (Zazouli et al., 2022). In this study, the ¹H and ¹³C NMR chemical shifts of the newly synthesized compounds were computed at the B3LYP/6-311G(d,p) level of theory. The optimized geometries with atomic numbering of 3a-d have been illustrated in Supplementary file. Tables S1 and S2 present the experimental and calculated chemical shifts, demonstrating a strong agreement and confirming consistency with the proposed molecular structure (See Supporting Info).

The experimental ¹H NMR chemical shifts for aromatic protons in compounds 3a-d were observed in the range of δ 8.69 to 7.22 ppm in CDCl₃, whereas the corresponding theoretical values ranged from δ 8.81 to 6.36 ppm. The observed consistency across the compounds reflects the influence of the nitrogen atom within the pyridine moiety, particularly in compounds 3a-c, which differ in para, meta, and ortho positioning relative to the theophylline scaffold. In contrast, compound 3d lacks protons adjacent to the pyridine nitrogen, accounting for a notable difference in its proton chemical shifts.

Specifically, for proton H17, the experimental chemical shifts were δ 7.80, 7.61, 7.73, and 7.69 ppm for compounds 3a, 3b, 3c, and 3d, respectively, while the theoretical values were δ 7.31, 6.73, 6.75, and 5.10 ppm. In the aliphatic region, comparable ¹H NMR signals were observed across all derivatives. For instance, three singlet signals appeared at δ 5.60 (for CH₂ protons H25 and H26), δ 3.60 (for CH₃ protons H31, H32, and H33), and δ 3.40 (for CH₃ protons H28, H29, and H30). The corresponding calculated chemical shifts were δ 5.08-5.30 for CH₂ (H25, H26), δ 3.19-3.51 for CH₃ (H31, H32, H33), and δ 3.57-3.55 for CH₃ (H28, H29, H30). In certain cases, significant discrepancies between experimental and theoretical values were observed. Minor discrepancies between experimental and theoretical values are attributed to the free rotation of methyl groups in theophylline moiety and the CH₂ linkers connecting the pyridine and theophylline moieties in the experimental solution, which introduce conformational flexibility not fully captured by theoretical calculations.

For the ¹³C NMR spectra, experimental chemical shifts ranged from δ27.9 to 155.3 ppm in CDCl₃, while theoretical values calculated at B3LYP/6-311G(d,p) spanned δ20.55 to 159.45 ppm. Notably, carbon atoms adjacent to electronegative atoms, such as oxygen (carbonyl) and nitrogen (pyridine), exhibited consistently higher chemical shifts due to the deshielding effect induced by electron-withdrawing. The close agreement between experimental and theoretical ¹³C chemical shifts further supports the validity of the calculated structures.

Aromatic protons exhibited resonance signals between δ7.18 and 7.50 ppm experimentally, and between δ7.39 and 7.92 ppm theoretically. The observed discrepancies are likely due to intermolecular hydrogen bonding interactions, which are not fully accounted for in the theoretical calculations. Modeling the dimeric form of these compounds could further improve the agreement between experimental and theoretical chemical shifts.

3.2.2 Vibrational assignment

FTIR spectroscopy is a widely employed technique for characterizing chemical bonds and functional groups in molecular structures. The vibrational assignments, experimental frequencies, theoretical wavenumbers, and calculated infrared intensities of the synthesized compounds have been summarized in Table S3. The symmetric and asymmetric C-H stretching vibrations typically appear between 3100 and 3000 cm^-1 (Stuart, 2004; Larkin, 2018). In the experimental IR spectra, these bands were observed at 3094, 3125, 3109, and 3125 cm^-1 for compounds 3a-d, respectively, while the corresponding theoretical values were calculated in the range of 3209-3184, 3204-3162, 3199-3179, and 3205-3118 cm^-1.

Asymmetric and symmetric CH₂ stretching vibrations typically occur between 3000 and 2850 cm^-1 (Abraham et al., 2018). The symmetric CH₂ stretches were experimentally observed at 2938 and 2937 cm^-1 for 3a & 3b and 3c & 3d, respectively, in good agreement with theoretical predictions of 3126, 3127, 3165, and 3095 cm^-1.

The carbonyl (C=O) stretching vibration typically occurs between 1650 and 1800 cm^-1, a region minimally affected by other absorptions, facilitating its identification. In the experimental spectra, the symmetric C=O stretches were detected at 1695-1687 cm^-1 for compounds 3a-d, while theoretical values were calculated as 1669-1669 cm^-1. This minor discrepancy is likely due to strain within the heterocyclic ring system, providing insights into the geometry of the carbonyl-containing ring.

C=C stretching vibrations within aromatic rings generally appear in the range 1675 to 1500 cm^-1. Experimentally, these were observed at 1625 cm⁻¹ for compound 3a, 1628 cm⁻¹ for 3b, and 1687 cm⁻¹ for compounds 3c and 3d, which align closely with the calculated values of 1638, 1639, 1668, and 1669 cm⁻¹ for compounds 3a, 3b, 3c, and 3d, respectively. C=N stretching vibrations, typically occurring between 1700 and 1500 cm⁻¹ (Ermiş et al., 2018), were experimentally measured at 1563 cm⁻¹ for compound 3a, 1577 cm⁻¹ for 3b, and 1570 cm⁻¹ for compounds 3c and 3d, with corresponding theoretical values of 1598, 1599, 1639, and 1609 cm⁻¹ for compounds 3a, 3b, 3c, and 3d, respectively. These results are consistent with symmetric C=N stretching vibrations reported in related studies (Tamer et al., 2014; Evecen et al., 2025).

3.2.3 FMO analysis

FMO analysis provided critical insights into the electronic properties and chemical reactivity of the investigated compounds. As illustrated in Supplementary file, the HOMO and LUMO orbitals of compounds 3a-c are delocalized throughout the molecular framework, whereas in compound 3d, the HOMO is predominantly localized on the caffeine moiety. Notably, compound 3d exhibited the smallest HOMO-LUMO energy gap (ΔE_GAP) and the lowest chemical hardness, indicating higher chemical reactivity and reduced thermodynamic stability. It also demonstrated increased electron transfer capability and polarizability. Additionally, 3d displayed the lowest electrophilicity index (ω = 0.107 eV) and a comparatively high dipole moment, further supporting its enhanced chemical reactivity (Table 1). Collectively, these results suggest that compound 3d is the most electronically active and potentially the most biologically relevant of the series (See supporting information).

3.2.4 Mulliken atomic charges

As depicted in Figs. S1-S28 (Supporting information), the electronic distribution in the synthesized compounds is strongly influenced by the functional groups and heterocyclic ring systems. The carbon atoms C14 and C16 serve as principal electrophilic centers, while the nitrogen atoms in the pyridine moiety represent key nucleophilic sites. Notably, the pronounced negative charge on N3 in compound 3d further indicates its potential for strong intermolecular interactions, which may contribute to the enhanced bioactivity observed for this derivative (See supporting information).

3.2.5 NBO

The NBO analysis revealed that the synthesized compounds exhibit significant intramolecular charge transfer (ICT) and electron delocalization, predominantly mediated through key bonding interactions and lone pairs. These electronic interactions contribute to the overall stabilization of the molecular systems. Among the series, compound 3d displayed the highest stabilization energies, indicating enhanced electronic delocalization and suggesting potentially greater chemical reactivity and bioactivity. The summarized NBO results have been presented in Table S4 and further detailed in the Supporting Information.

3.2.6 MEP

The synthesized compounds 3a-d exhibit pronounced polarity and chemical reactivity, which are critical for effective interactions with biological targets. As illustrated in Supplementary file, regions of high negative electrostatic potential are localized on the carbonyl and amine groups, indicating their potential significance in binding through electrophilic interactions. Conversely, areas of positive electrostatic potential on hydrogen atoms highlight possible sites for nucleophilic attack (See supporting information).

3.3.1 In vitro AChE and BChE inhibitory activities

As summarized in Table 2, the tested compounds consistently demonstrated preferential inhibition of BChE over AChE. Among the series, compound 3d emerged as the most potent BChE inhibitor (IC₅₀ = 7.87 ± 0.04 μM) while exhibiting moderate AChE inhibition (IC₅₀ = 39.15 ± 0.26 μM), indicating the highest selectivity for BChE. Compounds 3a-c displayed comparatively lower inhibitory activity but maintained greater efficacy against BChE than AChE. In contrast, the reference drug galantamine selectively inhibited AChE. Statistical analysis confirmed that compound 3d was significantly more potent against BChE than all other derivatives, whereas both 3d and galantamine were markedly more active against AChE relative to the remaining compounds. This selective inhibition profile highlights the potential of these compounds as therapeutic candidates targeting BChE-mediated pathways, which is particularly relevant in the later stages of Alzheimer’s disease when BChE activity rises as AChE activity declines (See Supporting Information for more detailed information).

3.3.2 AChE and BChE inhibitory activity in silico docking

Molecular docking analysis represents a pivotal tool in drug discovery, providing detailed insights into the interactions between potential drug candidates and their target proteins at the molecular level (Pinzi & Rastelli, 2019). The 2D and 3D interaction profiles of the synthesized compounds within the active sites of AChE and BChE have been illustrated in Fig. 1 and 2, with binding energies and rankings presented in Table 3. Galantamine served as a positive control to benchmark the binding affinities of the tested compounds.

Close

Fig. 1.

The molecular docking 2D and 3D interactions between compounds 3a-d and the active site of 4M0E.

Export to PPT

Close

Fig. 2.

The molecular docking 2D and 3D interactions between compounds 3a-d and the active site of 1P0P.

Export to PPT

Across all compounds, the carbonyl group of the caffeine moiety consistently engaged in hydrogen bonding with key residues, including HIS405, GLU197, and THR120. The amine functionalities on the pyrimidine and imidazole rings acted as principal hydrogen bond donors, most prominently in compound 3d, which formed multiple polar contacts with residues including Arg296, Leu286, and Gly287. Aromatic and heterocyclic rings, including pyridine, imidazole, and purine scaffolds, participated in π-alkyl and π-π stacking interactions, further stabilizing the ligand-protein complexes. Notably, compound 3d exhibited the most diverse and extensive interaction network, combining strong hydrogen bonding and hydrophobic interactions, which likely underpins its superior binding affinity. Therefore, the observed interactions for compound 3d, including multiple hydrogen bonds and hydrophobic contacts, provide meaningful insights into its potential bioactive conformation, although future MD simulations may further refine these predictions.

These findings underscore the importance of combining polar and hydrophobic groups within a molecule to optimize binding efficiency and selectivity toward cholinesterase enzymes (See supporting information).

3.3.3 In silico ADME-T studies

In silico ADME analysis was conducted, representing a crucial component of early drug discovery by providing predictions of compound behavior in an in vivo environment (Özkay et al., 2017). Additionally, toxicity parameters were evaluated, with the combined results summarized in Table 4. All compounds exhibited favorable drug-like properties, including moderate lipophilicity (log P = 0.20-0.57), optimal topological polar surface area (TPSA = 74.47-136.53 Ų), and molecular weights compatible with oral bioavailability. The compounds also complied with Lipinski’s and Veber’s rules and displayed a favorable bioavailability score of 0.55. Importantly, compounds 3a-c were not predicted to be P-gp substrates, suggesting enhanced oral absorption and tissue distribution. Regarding metabolic stability, compounds 3a-c were predicted to inhibit CYP1A2, whereas compound 3d inhibited CYP3A4, indicating potential drug-drug interaction risks. All compounds were predicted to be non-toxic (See Supporting Info for full details).

3.3.4 Gastrointestinal absorption and Brain penetration prediction [BOILED-Egg]

The BOILED-Egg model provides a straightforward computational approach for predicting a compound’s potential of gastrointestinal (GI) absorption and blood-brain barrier (BBB) penetration, based on polarity and lipophilicity. This assessment is particularly relevant for evaluating the oral bioavailability and central nervous system accessibility of candidate therapeutics for Alzheimer’s disease. According to the BOILED-Egg predictions summarized in Supplementary File, compounds 3a-c are located within the white region, indicating high GI absorption and favorable oral bioavailability, thereby supporting their potential as orally administered Alzheimer’s therapeutics. In contrast, compound 3d is located in the grey, suggesting limited GI absorption and poor passive BBB penetration. These results imply that, while 3d may exhibit reduced oral bioavailability and limited brain access through passive diffusion, alternative delivery strategies or structural modifications may be required to optimize its therapeutic efficacy in AD treatment.