Synthesis, molecular modelling and evaluation of (−)-isopulegol-based 2,4-diaminopyrimidines as promising aurora kinase inhibitors

Abstract

A library of (−)-isopulegol-based 2,4-diaminopyrimidines was prepared from commercially available (−)-isopulegol. Aminodiols, derived from (−)-isopulegol according to literature methods, were added to 5-substituted 2,4-dichloropyrimidines; the resulting products were then subjected to microwave-assisted S_NAr coupling reactions with aniline derivatives to produce 2,4-diaminopyrimidines. All 2,4-diaminopyrimidine adducts were evaluated for their in vitro cytotoxicity against human colon adenocarcinoma cell lines, including Colo205 and Colo320. Among these derivatives, compounds 6a and 7b exhibited significantly greater efficacy against the two cancer cell lines within concentrations around 2.0 μM. Furthermore, these derivatives displayed higher selectivity for cancer cells over normal cells (SI > 44) compared to the positive controls, doxorubicin (SI > 2) and cisplatin (SI = 5). Molecular docking analysis indicated that these compounds (6a and 7b) form interactions with the aurora A kinase receptor both from structural and energetic perspectives. These results suggest that derivatives 6a and 7b have potential for further development as aurora A kinase inhibitors for colorectal cancer treatment.

(−)-Isopulegol-based 2,4-diaminopyrimidines were prepared and evaluated for cytotoxicity against human colon adenocarcinoma cell lines. Molecular docking analysis indicated their interaction with the aurora kinase A receptor.

Introduction

The aurora kinase family, including three kinases designated as aurora A, B, and C, is a subfamily of serine/threonine kinases that have emerged as crucial factors in not only mitosis and cytokinesis, but also human carcinogenesis.^1,2 Among them, aurora kinases A and B have received the most attention to date as anticancer targets.³ Aurora A kinase, which plays a pivotal role in various cellular processes, including mitotic entry, centrosome maturation and spindle formation, is frequently overexpressed in tumour cells^4–7 while aurora B kinase, which plays a role in cell cycle progression, is frequently linked to tumour cell invasion, metastasis and drug resistance.^8–10 Both aurora kinases A and B have been validated as targets for anticancer drugs.^11–25

2,4-Diaminopyrimidines, an important pharmacological core, exhibit potent inhibitory activity against a number of protein kinases (Fig. 1),^26–28 including aurora kinase (PF-03814735),^29,30 polo-like kinase 1 (PLK1) (BI2536 and DAP-81),³¹ focal adhesion kinase (FAK) (such as VS-6062,^32,33 CEP-37440³⁴ and NVP-TAE 226),^35,36 spleen tyrosine kinase (SYK) (R406),³⁷ Rad3-related kinase (ATR) (NU6027),³⁸ cyclin-dependent kinases (CDK) (R547),³⁹ multi-targeted tyrosine kinase (XL228),⁴⁰ and anaplastic lymphoma kinase (ALK) (such as NVP-TAE684⁴¹ and GSK1838705A).⁴² Some of them were approved by the FDA for cancer treatment, including the multi-targeted inhibitor pazopanib (Votrient®, approved 2009),⁴³ ALK inhibitors such as ceritinib (Zykadia®, approved 2014)⁴⁴ and brigatinib (Alunbrig®, approved 2017),⁴⁵ SYK inhibitor fostamatinib (Tavalisse®, approved 2018),⁴⁶ and JAK-2 inhibitor fedratinib (Inrebic®, approved in 2019).⁴⁷ Others remain in clinical trials, including the SYK inhibitor cerdulatinib (phase 2 study),⁴⁸ PLK1 inhibitor volasertib (phase 2 study),⁴⁹ and FAK inhibitor defactinib (phase 1 study).⁵⁰ The discovery of novel and selective inhibitors of protein kinases, including aurora A kinase,^51–58 anaplastic lymphoma kinase (ALK),^59–66 cyclin-dependent kinases (CDK2,^67,68 CDK7,⁶⁹ and CDK9),^70,71 EGFR-targeted tyrosine kinase (EGFR-TK),^72–75 focal adhesion kinase (FAK),^76–79 tropomyosin receptor kinase (TRK),⁸⁰ p21-activated kinases (PAKs),⁸¹ Janus kinase (JAK2⁸² and JAK3),⁸³ glyoxalase I (GLO-1),⁸⁴ mesenchymal epithelial transition factor (c-Met) kinase,^85,86 hematopoietic progenitor kinase 1 (HPK1),^87–90 and casein kinase 1 epsilon (CK1ε)⁹¹ continue to attract the attention of many research groups.

Fig. 1. Some 2,4-diaminopyrimidine-based kinase inhibitors.

Our previous work indicated that N²-(p-trifluoromethyl)aniline-substituted pyrimidines, prepared from (−)-isopulegol-based aminodiols, displayed potent inhibitory effects on the growth of cancer cells.⁹² Docking studies suggested that the pyrimidine core and the amine group of the aminodiol ring form hydrogen bonds with the hinge region of the aurora A kinase domain.⁹² According to the literature, the displacement of the p-trifluoromethyl group on the aniline ring by substituents containing hydrophilic and acid–base interactions, such as carboxylic acid, carboxylic ester, and amide groups may lead to significant aurora A kinase inhibitors.⁵⁵ Therefore, in order to identify novel and highly potent aurora A kinase inhibitors using the pyrimidine moiety, a new series of 2,4-diaminopyrimidines were designed and synthesized.

Results and discussion

The synthetic routes started with the preparation of aminodiols 2a and b as key intermediates. Aminodiols 2a and b were obtained from commercially available (−)-isopulegol 1 by a three-step sequence including epoxidation with m-CPBA followed by ring-opening of the corresponding oxiranes with benzylamine and subsequent hydrogenolysis on 5% Pd/C.⁹³ The (−)-isopulegol-derived 2,4-diaminopyrimidines 4a–18a were prepared as shown in Scheme 1 and Table 1via the coupling of commercially available 5-fluoro-2,4-dichloropyrimidine with building block 2a, followed by microwave-assisted S_NAr coupling reactions with aniline derivatives in moderate to high yields (Scheme 1 and Table 1).^92,94 Intermediate 2b underwent a similar transformation to afford compounds 4b–18b with a 5-fluoro substituent in the pyrimidine scaffold (Scheme 1 and Table 1).

Scheme 1. (−)-Isopulegol-based 5-fluoro-2,4-diaminopyrimidines.

Table 1. (−)-Isopulegol-based 5-fluoro-2,4-diaminopyrimidines.

Entry	Compound	R	Yield (%)
1	4a	4-Trifluorophenyl	58
2	4b		72
3	4c		58
4	5a	4-Carboxyphenyl	60
5	5b		60
6	5c		53
7	6a	4-(Methoxy-carbonyl)phenyl	73
8	6b		76
9	6c		66
10	7a	4-(Ethoxycarbo-nyl)phenyl	85
11	7b		71
12	7c		57
13	8a	4-Acetamidophenyl	73
14	8b		59
15	8c		59
16	9a	4-Hydroxyphenyl	82
17	9b		82
18	9c		65
19	10a	4-Methoxyphenyl	79
20	10b		63
21	10c		71
22	11a	4-Cyanophenyl	40
23	11b		48
24	11c		40
25	12a	4-Nitrophenyl	23
26	12b		15
27	12c		15
28	13a	4-Morpholinophenyl	83
29	13b		55
30	13c		69
31	14a	4-(4-Methylpiperazin-1-yl)phenyl	34
32	14b		34
33	14c		27
34	15a	4-Sulfamoylphenyl	70
35	15b		56
36	15c		42
37	16a	Phenyl	67
38	16b		75
39	16c		50
40	17a	Pyridinyl	53
41	17b		44
42	17c		44
43	18a	1-Methyl-1H-pyrazol-4-yl	93
44	18b		84
45	18c		76

Attempts to directly synthesize 19a and b from 3a and b and p-aminobenzamide via a microwave-assisted heating method were not successful despite the elongated reaction time. Fortunately, this was achieved by hydrolysing their precursor nitriles 11a and b under alkaline conditions.⁹⁵ The analogues 19a and b were obtained in satisfactory yields as final products (Scheme 2).

Scheme 2. Preparation of 2,4-diaminopyrimidine carboxamide.

Besides the effects of stereochemistry on the aminoalcohol moiety, the antiproliferative activity also depends on the stereochemical aspects of the hydroxy substituent on the cyclohexyl core.^92,94 Therefore, the synthesis of library 4c–18c was performed using the synthetic routes and protocols shown in Scheme 3. The preparation of building block 2c was achieved from (+)-neoisopulegol 20, obtained from 1 in two steps by Jones oxidation of the hydroxy group, followed by stereospecific reduction of (S)-isopulegone with a stoichiometric amount of l-Selectride into the desired cis diastereoisomer,⁹⁶ according to literature-reported protocols⁹⁷ as mentioned in Scheme 3. The starting material 2c was subsequently reacted with 5-fluoro-2,4-dichloropyrimidine to obtain intermediate 3c in high yield. The displacement of the 4-chloro group on the 2,4-dichloropyrimidine moiety by various anilines provided the final targets 4c–18c with moderate to excellent yields (Scheme 3 and Table 1).⁹² The hydrolysis of nitrile 11c under alkaline conditions led to the formation of carboxamide 19c in moderate yield (Scheme 2).

Scheme 3. (+)-Neoisopulegol-based 5-fluoro-2,4-diaminopyrimidines.

The difference between the halogen atoms at the 5-carbon position on the pyrimidine ring could affect the electron-withdrawing properties improving in vitro antiproliferative activity.⁵⁴ The synthesis of library compounds 22–26 with 5-chloro substituents on the pyrimidine skeleton is described in Scheme 4. The key intermediates 21a–c were synthesized from commercially available 2,4,5-trichloropyrimidine and building blocks 2a–c using literature-reported protocols.⁹² In the same manner, these intermediates were directly reacted via nucleophilic aromatic substitution with aniline derivatives to obtain the final library compounds 22–28 possessing chlorine in the 5-position of the pyrimidine skeleton in moderate yields (Scheme 4, Table 2). Treatment of nitriles 26a–c with powdered potassium hydroxide in tert-butyl alcohol gave benzamides 29a–c (Scheme 2) in good yields.⁹⁵

Scheme 4. (−)-Isopulegol-based 5-chloro-2,4-diaminopyrimidines.

Table 2. (−)-Isopulegol-based 5-chloro-2,4-diaminopyrimidines.

Entry	Compound	R 1	Yield (%)
1	22a	4-Carboxyphenyl	49
2	22b		31
3	22c		38
4	23a	4-(Methoxycarbo-nyl)phenyl	50
5	23b		45
6	23c		45
7	24a	4-(Ethoxycarbo-nyl)phenyl	54
8	24b		43
9	24c		51
10	25a	4-Acetamidophenyl	57
11	25b		56
12	25c		52
13	26a	4-Cyanophenyl	46
14	26b		32
15	26c		43
16	27a	4-Morpholinophenyl	64
17	27b		28
18	27c		36
19	28a	4-(4-Methylpiperazin-1-yl)phenyl	52
20	28b		31
21	28c		27

Cytotoxic effects

Since several 2,4-diaminopyrimidines exhibited significant cytotoxicity on colorectal cancer cell lines,^70,91 the in vitro cytotoxic effects of the prepared isopulegol-based 2,4-diaminopyrimidine analogues were also screened on human colon adenocarcinoma lines, including Colo205 and Colo320 using the MTT assay. Doxorubicin (DOX) and cisplatin (CIS), clinically applied anticancer agents, were used as reference compounds, and the results are summarized in Tables S3 and S2 in the SI.

The results indicated that 2,4-diaminopyrimidine derivatives 6a (SI > 44), 6c (SI > 7), 7b (SI > 77) and 19b (SI > 12) exhibited potent inhibitory properties on doxorubicin-sensitive Colo 205 and doxorubicin-resistant Colo 320 adenocarcinoma cell lines comparable to those of the reference agent, doxorubicin and cisplatin, with no cytotoxicity against fibroblasts. Eight compounds (11b, 14b, 24b, 25b, 25c, 27b, 27c, and 28b) showed considerable cytotoxic action, with IC₅₀ values below 1 μM (SI < 1). The structure activity relationship (SAR) was investigated as below.

Compound 7b displayed a high potency in the range of the most active cytotoxic agents (SI > 77). Replacement of the p-ethyl ester function in 7b with a p-methyl ester, cyano or methylpiperazin-1-yl group as shown in 6b, 11b or 14b, respectively, retained the activity. In contrast, it resulted in over 200-fold toxicity in fibroblasts (SI < 1), emphasizing the importance of the ethyl ester moiety. In addition, a 2–5-fold loss of inhibitory activity was also observed when the ethyl ester group in 7b was replaced with trifluoromethyl (4b), acetamido (8b), methoxy (10b), nitro (12b), morpholino (13b), or even hydrogen (16b). In all cases, these replacements led to much less selectivity between cancer and normal cell lines (SI < 1) than the parent molecule (7b). Compound 19b (SI > 12), where the aniline ring has a carbamoyl substituent in the para position, was 8-fold less active than compound 7b, whereas the introduction of a p-COOH group into the aniline portion (compound 5b) resulted in the total loss of in vitro potency. These observations suggested that the p-COOC₂H₅ moiety on the phenyl ring was important to maintain the inhibitory activity. Furthermore, the activity of 16b (SI < 0.03) was significantly higher than that of 17b and 18b (SI < 1), indicating the importance of the phenyl ring for cytotoxic effects. Next, the effect of the halogen atom in the pyrimidine ring was studied, and the results indicated that the in vitro cytotoxicity of compounds 22b–28b, substituted with a chloride at C-5 of the pyrimidine ring, was more potent against the two cancer cell lines and fibroblasts than that of compounds 4b–19b, in which C-5 in pyrimidine was substituted with fluorine (Table 3). The influence of the stereochemistry of the hydroxy group in the cyclohexyl ring was analysed. The results indicated that in the case of 5-fluoropyrimidine, the hydroxy group with an R configuration (6a and 7a) was found to be more effective than its corresponding isomers (6c and 7c), whereas upon replacement of 5-fluoro with 5-chloro in the pyrimidine system, the (S)-isomers (23c and 24c) were more potent than the (R)-isomers (23a and 24a). From the results of the in vitro cytotoxic assay, it was found that compounds (6b, 7b, 23b, and 24b) with the (R)-OH substituent in the aminoalcohol scaffold showed more potent cytotoxic effects in comparison to the (S)-analogues (6a, 7a, 23a, and 24a).

Table 3. Cytotoxic activity of 2,4-diaminopyrimidines on Colo 205 and Colo 320 colon adenocarcinoma cell lines and the MRC-5 normal lung fibroblast cell line.

Compound	IC50 (μM)
	Colo 205	Colo 320	MRC-5
4b	3.38 ± 0.06	2.09 ± 0.17	1.72 ± 2.35
5b	>100	>100	>100
6a	2.24 ± 0.36	1.47 ± 0.03	>100
6b	1.20 ± 0.24	0.46 ± 0.02	<0.19
6c	7.18 ± 0.68	12.63 ± 0.45	>100
7a	13.75 ± 1.46	12.97 ± 1.96	<0.19
7b	1.03 ± 0.02	1.29 ± 0.24	>100
7c	34.33 ± 1.85	20.54 ± 0.69	>100
8b	4.52 ± 0.4	6.99 ± 0.47	2.46 ± 2.92
9b	7.11 ± 0.13	3.53 ± 0.19	3.62 ± 4.94
10b	1.56 ± 0.34	1.7 ± 0.03	0.95 ± 0.86
11b	0.99 ± 0.04	0.51 ± 0.06	0.52 ± 0.68
12b	2.34 ± 0.24	2.01 ± 0.07	<0.19
13b	1.46 ± 0.13	1.71 ± 0.19	0.79 ± 0.95
14b	0.98 ± 0.04	0.88 ± 0.03	0.51 ± 0.66
15b	14.58 ± 1.47	>100	8.03 ± 9.27
16b	5.78 ± 0.39	5.56 ± 0.17	<0.19
17b	>100	>100	>100
18b	16.3 ± 1.16	11.53 ± 1.29	8.73 ± 10.7
19b	7.91 ± 0.78	6.89 ± 0.23	>100
22b	14.75 ± 0.22	8.29 ± 0.82	>100
23b	1.03 ± 0.04	0.8 ± 0.15	0.53 ± 0.7
24b	0.75 ± 0.04	0.8 ± 0.09	0.39 ± 0.5
25b	0.52 ± 0.02	0.47 ± 0.01	0.27 ± 0.36
26b	1.25 ± 0.1	0.88 ± 0.04	0.67 ± 0.82
27b	0.64 ± 0.05	0.38 ± 0.02	0.35 ± 0.41
28b	0.26 ± 0.02	0.3 ± 0.01	0.14 ± 0.17
DOX	1.85 ± 0.77	3.96 ± 0.66	>8.62
CIS	14.52 ± 0.79	16.46 ± 1.67	85.39 ± 1.78

Molecular docking analysis

To elucidate the binding mechanisms and verify the consistency between the structural features and potential inhibitory activity, molecular docking simulations were performed to predict the interactions of compounds 6a, 6c, 7b, 19b, 25b and 28b with the aurora kinase A receptor (PDB: 4DEE). As summarized in Fig. 2 as well as Fig. 1 and Table 2 in the SI, these derivatives occupy the ATP-binding pocket through distinct electrostatic and hydrophobic networks.

Fig. 2. The 3D docking poses of aurora A inhibition (ID: 4DEE) with 2,4-diaminopyrimidines.

Compound 7b establishes a stable binding pose anchored by hydrogen bonds to Asn261 and Glu260, alongside a critical metal–acceptor interaction with Mg501 and a halogen (fluorine) bond to Gly142. This stability is further reinforced by π–anion interactions with Glu260 and a dense van der Waals shell involving Leu139, Val174, Phe144, and Trp277. Notably, the binding of 7b is tightly packed, although it exhibits some unfavourable donor–donor interactions with Lys143 and acceptor–acceptor repulsion with Asp274.

In contrast, compound 6a displays a shifted hydrogen bonding network, engaging Asp274, Lys141, and Lys162, while maintaining the conserved metal–acceptor contact with Mg501. Unlike 7b, compound 6a lacks the specific halogen bond to Gly142 and faces steric challenges, indicated by an unfavourable interaction with Mg502, which may constrain its conformational fit compared to its analogues. Its hydrophobic enclosure is primarily supported by π–alkyl interactions with Lys143 and van der Waals contacts with residues such as Ala273, Leu164, and Val147.

Compound 6c preserves the hydrogen bonds to Lys141 and Lys162 and expands its halogen bonding to Gly142; however, it suffers from significant electrostatic mismatches. Specifically, 6c is destabilized by unfavourable donor–donor and metal–donor repulsions with Lys143 and the Mg ion, respectively, suggesting a less optimal fit despite its broad van der Waals footprint.

Compound 25b appears to integrate key favourable features of the series, forming a stabilising hydrogen bond triad with Asp274, Asn261, and Glu260. Similar to 7b, it utilizes a fluorine-mediated halogen bond with Gly142 and a π–anion interaction with Glu260, effectively bridging the binding motifs observed in the other ligands. Compound 28b, while sharing the halogen bond at Gly142 and hydrogen bonds at Asp274 and Asn261, is penalised by multiple unfavourable interactions, including donor–donor clashes at Lys143 and acceptor–acceptor repulsion at Asp274.

Furthermore, compound 19b exhibits a binding mode that is consistent with the interaction patterns established for the previously analysed compounds, particularly mirroring the stability seen in 25b. It demonstrates a robust interaction profile anchored by a dual hydrogen-bonding network with Lys143 and Lys162. Its orientation is further secured by a metal–acceptor contact with the Mg ion and a halogen bond to Lys141, while a dense hydrophobic shell, comprising π–σ interactions with Leu263 and π–alkyl contacts with Ala273, reinforces the binding.

In summary, the docking results indicate that halogen bonding and metal coordination with Mg ions serve as the key anchoring interactions within the aurora A active sites. These ligands 7b, 19b, 25b and 28b emerge as the most promising candidates due to their ability to maintain these critical interactions effectively, with 28b achieving high affinity (−9.8 kcal mol⁻¹) despite the presence of minor unfavourable electrostatic interactions. In contrast, 6a and 6c show reduced binding performance due to specific structural conflicts: a steric clash with the Mg ion in 6a and cumulative donor–donor or metal–donor repulsions in 6c. Ultimately, these interaction profiles identify 25b as the most balanced derivative, while 28b remains a high-affinity analogue of interest. These results also explain the most effective cytotoxicity of adducts 25b and 28b in the in vitro assay.

In silico ADME prediction

As summarised in Tables S2–S4 in the SI, the pharmacokinetic properties of derivatives 6a, 6c, 7b, 19b, 25b and 28b were estimated using the ADMETlab 2.0 web tool. All synthesized analogues adhered to Lipinski's Rule of Five with zero violations, exhibiting physicochemical properties conducive to drug-likeness. Furthermore, unlike many early-stage leads, these compounds displayed high gastrointestinal absorption and moderate Caco-2 permeability values ranging from −4.96 to −5.56, suggesting favourable oral bioavailability.

Metabolic stability was assessed against the cytochrome P450 (CYP) superfamily, the primary enzyme system responsible for the oxidation of xenobiotics in the liver and intestines. The data reveal a noticeable divergence in metabolic liability. Compounds 6a, 6c and 7b act as broad-spectrum inhibitors, showing inhibitory activity against all major isoforms, including CYP1A2, CYP2C19, CYP2C9, CYP2D6, and CYP3A4. Conversely, 19b, 25b and 28b demonstrated improved selectivity. Notably, 19b and 25b spared CYP1A2 and CYP2C19, while 28b exhibited the most selective CYP inhibition profile, inhibiting only CYP3A4 while remaining non-inhibitory toward CYP1A2, CYP2C19, CYP2C9, and CYP2D6. In summary, while 6a, 6c and 7b present a higher potential for clinically meaningful drug–drug interactions (DDIs), the refined profiles of 19b, 25b and especially 28b suggest a lower risk of metabolic interference upon co-administration with other therapeutics.

In silico toxicity profiles

Additionally, the safety profiles of the analogues were predicted using the Deep-PK computational tool (Tables S2–S4 in the SI). In terms of acute toxicity, compound 25b suggested the most favourable acute toxicity profile with an LD₅₀ of 2888 mg kg⁻¹, classifying it within GHS category 5, which is low toxicity.⁹⁸ In contrast, compounds 6a (1498 mg kg⁻¹), 6c (1643 mg kg⁻¹), 7b (1547 mg kg⁻¹), and 28b (1213 mg kg⁻¹) fall into GHS category 4.⁹⁸ Compound 19b exhibited the highest acute toxicity risk with an LD₅₀ of 832 mg kg⁻¹.

Regarding organ-specific endpoints, the toxicity estimation for compounds 6a, 6c, 7b and 25b predicted safety concerning carcinogenesis. However, compounds 19b and 28b were flagged for potential carcinogenesis, which may compromise their developability. All compounds in the series were predicted as safe for drug-induced liver injury type I (DILI) but were flagged for potential liver injury type II and micronucleus formation, suggesting a potential risk of hepatotoxicity and genotoxicity that necessitates monitoring in downstream assays.

Mechanism-based toxicity screening revealed that while all compounds are predicted to be safe for the androgen, estrogen, including their LBDs, and thyroid receptors, there is a divergence in glucocorticoid receptor (GR) interactions. Compounds 6a, 6c, 7b and 25b were predicted as GR toxic, suggesting potential GR-mediated off-target effects. Interestingly, compounds 19b and 28b were the sole analogues predicted as GR safe. Finally, all six compounds showed a potential hERG liability, a common signal in kinase inhibitor discovery. Nevertheless, the favourable acute safety profile of 25b marks it as the most promising lead for further toxicological optimization.

Mechanism-linked SAR

To further rationalize the observed structure–activity relationship (SAR) trends, the cytotoxicity data were interpreted in conjunction with molecular docking results within the ATP-binding pocket of aurora kinase A (PDB: 4DEE).

The superior activity and selectivity of compound 7b can be mechanistically explained by its ability to establish a multi-point anchoring network within the binding pocket. Docking analysis revealed that the para-ethyl ester substituent plays a crucial role in mediating hydrogen bonding interactions with key residues such as Asn261 and Glu260, thereby stabilizing the ligand orientation toward the hinge region. In addition, this group enhances complementarity within the hydrophobic pocket formed by residues, including Leu139, Val174, and Phe144. In contrast, replacement of the ester group with more polar functionalities such as –COOH or less optimally oriented substituents such as –CN and –CONH₂ disrupts this interaction network, leading to reduced binding stability through the loss of key hydrogen bonds or increased off-target interactions. These effects are consistent with the experimentally observed loss of selectivity (SI < 1). Therefore, the para-ethyl ester moiety can be regarded as a dual-function pharmacophore that balances hydrogen bonding and hydrophobic interactions to achieve optimal binding and selectivity.

The nature of the halogen substituent at the C-5 position of the pyrimidine ring also plays a critical role in modulating ligand–protein interactions. Fluorinated derivatives, exemplified by compound 7b, were found to form halogen bonding interactions with Gly142, which contribute to proper positioning within the ATP-binding site while maintaining a favorable selectivity profile. In contrast, substitution with chlorine, as observed in the 22–28 series, enhances hydrophobic contacts and overall binding affinity, which is reflected by lower IC₅₀ values. However, this increased hydrophobicity also promotes non-specific interactions and reduces discrimination between cancer and normal cells, resulting in diminished selectivity. These findings suggest that fluorine provides a more balanced interaction profile, whereas chlorine favours stronger but less selective binding.

Another important feature observed among active compounds such as 7b, 25b, and 28b is their ability to interact with the Mg²⁺ ion present in the active site, forming metal–acceptor interactions that further stabilize ligand binding. Moreover, hydrogen bonding interactions with hinge region residues, particularly Glu260 and Asn261, mimic the canonical binding mode of ATP-competitive kinase inhibitors. However, less active compounds such as 6a and 6c exhibit steric clashes with the Mg²⁺ ion and unfavourable donor–donor interactions, for instance with Lys143, which likely compromise binding efficiency and account for their reduced biological activity.

The stereochemistry of the hydroxy substituent on the cyclohexyl ring also significantly influences ligand binding. The R configuration, as observed in compound 7b, enables optimal spatial alignment of the aminoalcohol moiety toward polar residues within the binding pocket, facilitating the formation of stable hydrogen bonding networks without introducing steric hindrance. In contrast, the corresponding S-isomers tend to misalign key functional groups and may introduce electrostatic repulsion or steric clashes, thereby reducing binding affinity. This stereochemical dependence highlights the importance of three-dimensional complementarity between the ligand and the protein binding site.

Overall, the biological activity of these 2,4-diaminopyrimidine derivatives is governed by a combination of hinge-binding hydrogen bonds with residues such as Glu260 and Asn261, halogen bonding interactions with Gly142 for positional stabilization, metal coordination with Mg²⁺, and hydrophobic enclosure within the ATP-binding pocket. Among the series, compound 7b achieves the most favorable balance of these interactions, which explains its superior potency and selectivity. In contrast, highly potent but non-selective compounds such as 25b and 28b likely derive their activity from increased binding affinity at the expense of specificity. These findings suggest that fine-tuning the balance between polar anchoring interactions and hydrophobic complementarity within the aurora A kinase binding pocket is critical for achieving both potency and selectivity in this scaffold.

Conclusions

Aurora kinases have been of interest as potential therapeutic targets. In the current investigation, a series of (−)-isopulegol-derived 2,4-diaminopyrimidines that exerted their cytotoxic effects in a panel of colorectal cancer cell lines were designed and synthesized.

All of the compounds were screened for their cytotoxic effects on human colon adenocarcinoma lines. Compounds 6a and 7b exerted outstanding activities against the malignant cells with no action on fibroblasts (SI > 44), indicating considerable cancer selectivity. The structure–activity relationship clearly indicated the importance of stereochemistry on the cyclohexyl ring and aminodiol moiety, as well as the effects of substituents on the pyrimidine scaffold.

The molecular docking studies revealed that the ester group at the para position of the aniline moiety in compounds 6a and 7b played a crucial role in binding with aurora A kinase. These data together suggest the potential of 6a and 7b as promising therapeutic candidates for addressing colorectal cancer based on aurora A kinase inhibition.

Author contributions

T. M. L.: writing – original draft, validation, methodology, and investigation; N. S.: methodology, investigation, formal analysis, and investigation; M. C. N.: investigation; H. N. K. T.: writing – original draft and formal analysis; G. S.: review & editing, formal analysis, and data curation; K. M. T.: methodology and review & editing; Z. S.: supervision, resources, and funding acquisition.

Conflicts of interest

There are no conflicts to declare.

Data availability

Data supporting this article have been included as part of the supplementary information (SI) and available therein.

Supplementary information is available. See DOI: https://doi.org/10.1039/d6md00278a.