2-Oxoindoline-based hydroxamic acids: novel HDAC inhibitors with promising anticancer activity

Abstract

Histone deacetylase (HDAC) inhibitors are important epigenetic anticancer agents that regulate gene expression, induce cell cycle arrest, and promote apoptosis. In this study, a novel series of 2-oxoindoline-capped hydroxamic acids was designed, synthesized and evaluated for their capacity to inhibit histone deacetylases and suppress cancer cell proliferation. The screening panel incorporated multiple cancer models spanning different organ systems, including colorectal adenocarcinoma (SW620, HCT116), triple-negative breast cancer (MDA-MB-231), non-small cell lung carcinoma (A549), and prostate cancer (PC-3). Comparative assessment against non-transformed fibroblasts (MRC-5) enabled evaluation of selectivity and tolerability profiles. Several derivatives exhibited potent HDAC inhibition at submicromolar concentrations, with compounds 7c, 10b, and 10c showing stronger activity than the reference inhibitor SAHA. Among them, compound 10c demonstrated broad antiproliferative effects while maintaining relatively low toxicity toward normal cells. Mechanistic investigations revealed that 10c induced S-phase cell cycle arrest and promoted apoptosis in SW620 colorectal cancer cells. Molecular docking studies against multiple HDAC isoforms supported the experimental findings by revealing favorable zinc coordination and key interactions within the catalytic pocket. To further elucidate the binding behavior and dynamic features of the most active derivatives, molecular dynamics simulations were performed for HDAC complexes with 7c, 10b, and 10c. The simulations revealed stable protein–ligand interactions without perturbation of the overall protein structure, while highlighting distinct binding dynamics among the compounds, with 10c exhibiting the highest binding persistence, followed by 7c and 10b. In addition, in silico ADME and toxicity predictions were carried out for compound 10c as a representative highly active derivative, indicating acceptable drug-like properties and a favorable safety profile. Overall, 2-oxoindoline-based hydroxamic acids, particularly those bearing extended alkyl linkers, represent promising scaffolds for further development of HDAC-targeted anticancer agents.

Histone deacetylase (HDAC) inhibitors are important epigenetic anticancer agents that regulate gene expression, induce cell cycle arrest, and promote apoptosis.

1. Introduction

Histone deacetylases (HDACs) are enzymes that remove acetyl groups from lysine residues on histone tails, increasing the positive charge of histones and promoting tighter binding to the negatively charged DNA. This results in chromatin condensation and suppression of gene expression. Conversely, histone acetyltransferases (HATs) add acetyl groups to histones, facilitating chromatin relaxation and transcriptional activation. The balance between HDAC and HAT activity is essential for maintaining epigenetic stability and regulating gene expression.¹ In addition to histones, HDACs, also known as protein deacetylases (PDACs), target numerous non-histone proteins involved in diverse cellular processes. These include transcription factors (e.g., p53, Rb), components of the transcriptional machinery, cytoskeletal proteins (e.g., α-tubulin, actin), and proteins involved in DNA repair, apoptosis, and signal transduction.^2,3 HDAC dysregulation has been implicated in various cancers, where it contributes to the epigenetic silencing of tumor suppressor genes and disruption of cellular homeostasis. As such, HDACs have emerged as promising therapeutic targets. Inhibiting HDACs can restore gene expression, induce cell cycle arrest, and promote apoptosis in cancer cells.⁴ These observations emphasize the pivotal function of histone deacetylases across transcriptional networks, malignant transformation, and cell proliferation regulation. To date, 18 HDAC isoforms have been identified in humans. These enzymes are categorized into four classes based on their sequence homology with yeast HDACs.^5–9

In 2006, the FDA licensed Vorinostat (1, Fig. 1) as the pioneering HDAC inhibitor for cutaneous T-cell lymphoma (CTCL) therapy.¹⁰ Several other HDAC inhibitors, such as chidamide (2, Fig. 1), belinostat (3, Fig. 1), panobinostat (4, Fig. 1), and romidepsin (5, Fig. 1), have gained approval for various cancer.¹¹ HDAC inhibitors typically adopt a well-established three-component structural design: a zinc-binding group (ZBG) for metal coordination, a linker for spatial organization, and a hydrophobic cap (CAP) for surface engagement.^12–14 Metal coordination involves the ZBG establishing contacts with catalytic zinc at the enzyme's catalytic pocket. The linker mediates optimal conformational alignment, while the CAP domain stabilizes the inhibitor through surface residue interactions.^14–16

Fig. 1. Ligand design strategies and zinc-chelating pharmacophores of approved HDAC-targeting cancer therapeutics.

Isatin, a versatile indole-based scaffold, has emerged as a promising cap group in the design of histone deacetylase inhibitors (HDACi) due to its favorable structural and electronic properties. Its rigid, planar aromatic system facilitates π–π stacking interactions with aromatic residues lining the rim of the HDAC active-site channel, thereby enhancing binding affinity and isoform selectivity. Moreover, the isatin core offers both hydrogen bond donors and acceptors, facilitating stabilizing interactions with surface-exposed amino acid residues.^17–19 In our previous studies, a series of isatin-based HDAC inhibitors exhibited potent enzymatic inhibition and promising antiproliferative activities against several human cancer cell lines, including colorectal cancer (SW620), prostate cancer (PC-3), and pancreatic cancer (AsPC-1), highlighting the potential of the 2-oxoindoline scaffold as an effective surface recognition moiety in HDAC inhibitor design.^19–24 Beyond HDAC inhibition, the isatin scaffold has been widely recognized as a privileged structure in medicinal chemistry and has been incorporated into numerous anticancer agents targeting diverse molecular pathways. This scaffold is also found in many compounds with potential anticancer activity targeting various molecular pathways, including dual VEGFR-2/CA,²⁵ Hsp90,²⁶ PI3K,²⁷ EGFR,²⁸ caspase,²⁹ and CDK2.³⁰ (Fig. 2).

Fig. 2. Structures of 2-oxoindoline derivatives as anticancer agents.

In continuation of research aimed at optimizing HDAC inhibitor scaffolds, we synthesized and evaluated a collection of 2-oxoindoline-functionalized hydroxamic acids (Fig. 3). The indoline-derived cap was incorporated with the expectation that it would strengthen electrostatic and hydrogen-bonding contacts at the binding site entrance, thereby enhancing HDAC suppression and anticancer activity. Herein, we present the chemical synthesis, experimental biological screening, and structural prediction of these novel hydroxamic acid compounds.

Fig. 3. Rational design of 2-oxoindoline based hydroxamic acids as novel HDAC inhibitors.

2. Materials and methods

2.1. Chemistry

The starting materials, solvents, and reagents were sourced from Aldrich, Fluka Chemical Corp. (Milwaukee, WI, USA), or Merck without further purification. The reaction mixtures were monitored by thin-layer chromatography (TLC) using Whatman® 250 µm Silica Gel GF Uniplates with different mobile phase solvents and visualized under UV light at 254 or 365 nm. Purification of synthetic products was accomplished via column chromatography on Merck silica gel 60 (240–400 mesh). Melting point determination was performed on a Gallenkamp apparatus (LabMerchant, London, United Kingdom). Spectroscopic characterization included: (1) ¹H and ¹³C NMR analysis acquired at 600 and 150 MHz, respectively, on Bruker AC-500/AC-600 spectrometers using DMSO-d₆ solvent with TMS as the reference standard; (2) high-resolution mass spectrometry (HRMS) conducted on PE Biosystems API 2000 and Mariner instruments with electrospray ionization; and (3) infrared (IR) spectroscopy obtained on a Shimadzu FTIR Affinity-1S instrument using the KBr pellet method.

2.1.1. General procedure for synthesis of compounds 4a–c

As outlined in Scheme 1, a three-step synthetic route was established to obtain the target 2-oxoindoline-based N-hydroxybenzamides (4a–c). In the first step, 5-bromoisatin (1.5 mmol) was dissolved in DMF (5 mL), followed by the addition of K₂CO₃ (6 mmol). The reaction mixture was stirred at room temperature for 1 hour, after which methyl 4-bromomethylbenzoate (1.5 mmol) and a catalytic amount of KI were introduced. The mixture was stirred at room temperature for 6 hours and monitored by TLC (EA/hexane = 1 : 3). Upon completion, the resulting mixture was poured into ice-cold water, and the resulting precipitate was collected by filtration and washed with water to afford the N-alkylated isatin derivative (2).

Scheme 1. Synthesis of hydroxamic acids incorporating 2-oxoindoline scaffold 4a–c, 7a–c and 10a–c.

For the second step, compound 2 (1.0 mmol) was dissolved in DMF (5 mL) with 3 drops of water. Subsequently, boronic acid pinacol esters (1.5 mmol), PdCl₂[P(o-tolyl)₃]₂ (10 mol%), PPh₃ (20 mol%), and K₂CO₃ (4 mmol) were added. The mixture was stirred under a nitrogen atmosphere at 80–100 °C for 7 hours, monitored by TLC. After completion, the reaction was extracted with DCM and water. The combined organic layers were washed with cold water (4 × 30 mL) to remove residual DMF, dried over anhydrous sodium sulfate, evaporated under reduced pressure, and purified by silica gel column chromatography (EA/hexane = 5–10%) to yield intermediates 3a–c.

In the final step, compounds 3a–c were dissolved in methanol and treated with hydroxylamine hydrochloride (10 mmol) at 0–5 °C for 15 minutes. NaOH was then added dropwise to adjust the pH to ∼11. The reaction progress was monitored by TLC (DCM/methanol = 9 : 1). Upon completion, the mixture was diluted with water and acidified to pH ∼7 using HCl. The precipitate was filtered, dried, and purified by silica gel column chromatography (DCM/methanol = 3–6%) to afford the desired N-hydroxybenzamides (4a–c).

2.1.1.1. N-Hydroxy-4-((3-(hydroxyimino)-2-oxo-5-phenylindolin-1-yl)methyl)benzamide (4a)

Yellow solid; yield: 34%; mp: 178.7–179.9 °C; R_f = 0.46 (DCM : MeOH = 9 : 1). IR (KBr, cm⁻¹): 3360 (NH); 3222 (OH); 2972 (CH, arene); 2902 (CH, CH₂); 1717, 1665 (C O); 1615 (C N); 1541 (C C). ¹H NMR (600 MHz, DMSO-d₆, ppm): δ 13.68 (¹H, s, N–OH); 11.17 (¹H, s, CONHOH); 9.00 (¹H, s, NHOH); 8.27 (¹H, d, J = 1.8 Hz, H-4′); 7.73 (2H, d, J = 8.40 Hz, H-2, H-6); 7.65 (¹H, dd, J = 7.80 Hz, J′ = 1.8 Hz, H-6′); 7.57 (2H, d, J = 7.80 Hz, H-3, H-5); 7.45–7.42 (4H, m, H-2″, H-6″, H-3″, H-5″); 7.34 (¹H, t, J = 7.50 Hz, H-4″); 7.06 (¹H, d, J = 8.40 Hz, H-7′); 5.03 (2H, s, CH₂). ¹³C NMR (150 MHz, DMSO-d₆, ppm): δ 163.28, 143.32, 141.97, 139.59, 139.28, 135.19, 132.04, 130.35, 128.94, 127.29, 127.19, 127.09, 126.34, 125.10, 116.01, 109.94, 42.39, 30.62. HRMS (ESI) m/z calculated for C₂₂H₁₇N₃O₄, [M + H]⁺ 388.1292. Found, 388.12753.

2.1.1.2. 4-((5-(1,3-Dimethyl-1H-pyrazol-4-yl)-3-(hydroxyimino)-2-oxoindolin-1-yl)methyl)-N-hydroxybenzamide (4b)

Yellow solid; yield: 29%; mp: 183.4–184.9 °C; R_f = 0.41 (DCM : MeOH = 9 : 1). IR (KBr, cm⁻¹): 3254 (OH); 2972 (CH, arene); 2902 (CH, CH₂); 1701, 1653 (C O); 1616 (C N); 1559 (C C). ¹H NMR (600 MHz, DMSO-d₆, ppm): δ 8.99 (¹H, s, NHOH); 8.06 (¹H, d, J = 1.80 Hz, H-4′); 7.83 (¹H, s, H-5″); 7.72 (2H, d, J = 8.40 Hz, H-2, H-6); 7.42–7.38 (3H, m, H-6′, H-3, H-5); 6.99 (¹H, d, J = 7.80 Hz, H-7′); 4.99 (2H, s, CH₂); 3.76 (3H, s, N–CH₃); 2.23 (3H, s, C–CH₃). 13C NMR (150 MHz, DMSO-d₆, ppm): δ 163.15, 143.61, 143.39, 140.51, 139.33, 130.13, 129.24, 128.34, 127.25, 127.09, 125.39, 118.97, 115.71, 109.67, 42.32, 38.12, 34.39, 12.91. HRMS (ESI) m/z calculated for C₂₁H₁₉N₅O_4, [M + H]⁺ 406.1510. Found, 406.15131.

2.1.1.3. 4-((5-(4-Fluorophenyl)-3-(hydroxyimino)-2-oxoindolin-1-yl)methyl)-N-hydroxybenzamide (4c)

Yellow solid; yield: 41%; mp: 188.3–189.8 °C; R_f = 0.44 (DCM : MeOH = 9 : 1). IR (KBr, cm⁻¹): 3276 (OH); 2972 (CH, arene); 2902 (CH, CH₂); 1704, 1653 (C O); 1616 (C N); 1543 (C C). ¹H NMR (600 MHz, DMSO-d₆, ppm): δ 13.70 (¹H, s, N–OH); 11.17 (¹H, s, CONHOH); 9.01 (¹H, s, NHOH); 8.24 (¹H, d, J = 1.8 Hz, H-4′); 7.73 (2H, d, J = 8.40 Hz, H-2, H-6); 7.64–7.61 (3H, m, H-6′, H-2″, H-6″); 7.43 (2H, d, J = 8.40 Hz, H-3, H-5); 7.26 (2H, t, J = 9.00 Hz, H-3″, H-5″); 7.07 (¹H, d, J = 8.40 Hz, H-7′); 5.04 (2H, s, CH₂). ¹³C NMR (150 MHz, DMSO-d₆, ppm): δ 163.88, 163.25, 162.64, 160.70, 143.27, 141.94, 139.27, 136.10, 134.18, 132.05, 130.34, 128.38, 128.32, 127.30, 127.09, 125.05, 115.99, 115.79, 115.62, 109.94, 42.39. HRMS (ESI) m/z calculated for C₂₂H₁₆FN₃O_4, [M + H]⁺ 406.1198. Found, 406.12009.

2.1.2. General procedure for synthesis of compounds 7a–c

Analogously, the synthesis of 2-oxoindoline-based N-hydroxypropenamides (7a–c) was carried out using methyl (E)-4-bromomethylcinnamate in place of methyl 4-bromomethylbenzoate (Scheme 1).

2.1.2.1. (E)-N-Hydroxy-3-(4-((3-(hydroxyimino)-2-oxo-5-phenylindolin-1-yl)methyl)phenyl)acrylamide (7a)

Yellow solid; yield: 43%; mp: 192.4–193.8 °C; R_f = 0.49 (DCM : MeOH = 9 : 1). IR (KBr, cm⁻¹): 3255 (OH); 2987 (CH, arene); 2902 (CH, CH₂); 1734, 1667 (C O); 1619 (C N); 1540 (C C). 1H NMR (600 MHz, DMSO-d₆, ppm): δ 13.64 (¹H, s, N–OH); 10.74 (¹H, s, CONHOH); 9.01 (¹H, S, NHOH); 8.27 (¹H, d, J = 1.80 Hz, H-4″); 7.67 (1H, dd, J = 7.80 Hz, J′ = 1.80 Hz, H-6″); 7.59 (2H, d, J = 7.80 Hz, H-2′, H-6′); 7.54 (2H, d, J = 8.40 Hz, H-2‴, H-6‴); 7.45 (2H, t, J = 7.50 Hz, H-3‴, H-5‴); 7.40 (2H, d, J = 8.40 Hz, H-3′, H-5′); 7.39 (¹H, d, J = 15.00 Hz, H-3); 7.35 (¹H, t, J = 7.50 Hz, H-4‴); 7.08 (¹H, d, J = 8.40 Hz, H-7″); 6.43 (¹H, d, J = 15.60 Hz, H-2); 5.01 (2H, s, CH₂)^. 13C NMR (150 MHz, DMSO-d₆, ppm): δ 163.25, 143.32, 142.00, 139.58, 137.73, 137.46, 135.15, 134.10, 130.35, 128.93, 127.81, 127.73, 127.18, 126.33, 125.07, 119.12, 115.98, 109.98, 42.41. HRMS (ESI) m/z calculated for C₂₄H₁₉N₃O_4, [M + H]⁺ 414.1448. Found, 414.14316.

2.1.2.2. (E)-3-(4-((5-(1,3-Dimethyl-1H-pyrazol-4-yl)-3-(hydroxyimino)-2-oxoindolin-1-yl)methyl)phenyl)-N-hydroxyacrylamide (7b)

Yellow solid; yield: 36%; mp: 197.2–199.1 °C; R_f = 0.45 (DCM : MeOH = 9 : 1). IR (KBr, cm⁻¹): 3340 (NH), 3082 (OH); 2927 (CH, arene); 1714, 1702 (C O); 1590 (C N); 1559 (C C)^. 1H NMR (600 MHz, DMSO-d₆, ppm): δ 8.07 (¹H, d, J = 1.20 Hz, H-4″); 7.83 (¹H, s, H-5‴); 7.53 (2H, d, J = 7.80 Hz, H-2′, H-6′); 7.42 (¹H, d, J = 16.20 Hz, H-3); 7.40–7.38 (3H, m, H-6″, H-3′, H-5′); 7.00 (1H, d, J = 7.80 Hz, H-7″); 6.43 (1H, d, J = 15.60 Hz, H-2); 4.98 (2H, s, CH₂); 3.78 (3H, s, N–CH₃); 2.24 (3H, s, C–CH₃). 13C NMR (150 MHz, DMSO-d₆, ppm): δ 163.22, 143.67, 143.46, 140.56, 137.59, 134.12, 130.15, 129.29, 128.34, 127.84, 127.79, 125.38, 119.14, 119.05, 115.75, 109.75, 42.40, 38.17, 12.95. HRMS (ESI) m/z calculated for C₂₃H₂₁N₅O_4, [M + H]⁺ 432.1666. Found, 432.16702.

2.1.2.3. (E)-3-(4-((5-(4-Fluorophenyl)-3-(hydroxyimino)-2-oxoindolin-1-yl)methyl)phenyl)-N-hydroxyacrylamide (7c)

Yellow solid; yield: 38%; mp: 195.5–197.3 °C; R_f = 0.48 (DCM : MeOH = 9 : 1). IR (KBr, cm⁻¹): 3247 (OH); 2972 (CH, arene); 2901 (CH, CH2); 1717, 1668 (C O); 1616 (C N); 1508 (C C). 1H NMR (600 MHz, DMSO-d₆, ppm): δ 8.24 (1H, d, J = 1.80 Hz, H-4″); 7.64–7.61 (3H, m, H-6″, H-2′, H-6′); 7.54 (2H, d, J = 8.40 Hz, H-2‴, H-6‴); 7.43 (1H, d, J = 15.60 Hz, H-3); 7.39 (2H, d, J = 7.80 Hz, H-3′, H-5′); 7.26 (2H, t, J = 8.70 Hz, H-3‴, H-5‴); 7.07 (1H, d, J = 8.40 Hz, H-7″); 6.43 (1H, d, J = 15.60 Hz, H-2); 5.01 (2H, s, CH2). 13C NMR (150 MHz, DMSO-d₆, ppm):δ 163.24, 162.62, 160.68, 143.27, 141.94, 137.69, 137.45, 136.10, 134.12, 130.29, 128.36, 128.30, 127.80, 127.72, 124.98, 119.12, 115.97, 115.77, 115.60, 109.96, 42.40 HRMS (ESI) m/z calculated for C₂₄H₁₈FN₃O_4, [M + H]⁺ 432.1354. Found, 432.13583.

2.1.3. General procedure for synthesis of compounds 10a–c

Similarly, N-hydroxyheptanamides (10a–c) were prepared following the same procedure using methyl 7-bromoheptanoate (Scheme 1).

2.1.3.1. N-Hydroxy-7-(3-(hydroxyimino)-2-oxo-5-phenylindolin-1-yl)heptanamide (10a)

Yellow solid; yield: 31%; mp: 177.5–178.8 °C; R_f = 0.55 (DCM : MeOH = 9 : 1). IR (KBr, cm⁻¹): 3221 (OH); 3054 (CH, arene); 2951, 2852 (CH, CH₂); 1706, 1638 (C O); 1615 (C N); 1590 (C C). ¹H NMR (600 MHz, DMSO-d₆, ppm): δ 13.53 (¹H, s, N–OH); 10.31 (¹H, s, CONHOH); 8.62 (¹H, s, NHOH); 8.25 (¹H, d, J = 2.40 Hz, H-4′); 7.72 (¹H, dd, J = 8.40 Hz, J′ = 1.80 Hz, H-6′); 7.61 (2H, dd, J = 8.10 Hz, J′ = 1.20 Hz, H-2″, H-6″); 7.47 (2H, t, J = 7.80 Hz, H-3″, H-5″); 7.35 (¹H, t, J = 7.80 Hz, H-4″); 7.22 (¹H, d, J = 7.80 Hz, H-7′); 3.73 (2H, t, J = 7.20 Hz, H-7a, H-7b); 1.93 (2H, t, J = 7.20 Hz, H-2a, H-2b); 1.62–1.60 (2H, m, H-6a, H-6b); 1.49–1.47 (2H, m, H-3a, H-3b); 1.30–1.29 (4H, m, H-4a, H-4b, H-5a, H-5b). 13C NMR (150 MHz, DMSO-d₆, ppm): δ 169.02, 163.03, 143.44, 142.43, 139.68, 134.77, 130.38, 128.96, 127.14, 126.31, 125.01, 115.86, 109.57, 32.13, 28.16, 26.91, 25.90, 24.94. HRMS (ESI) m/z calculated for C₂₁H₂₃N₃O_4, [M + H]⁺ 382.1761. Found, 382.17462.

2.1.3.2. 7-(5-(1,3-Dimethyl-1H-pyrazol-4-yl)-3-(hydroxyimino)-2-oxoindolin-1-yl)-N-hydroxyheptanamide (10b)

Yellow solid; yield: 27%; mp: 179.5–181.4 °C; R_f = 0.51 (DCM : MeOH = 9 : 1). IR (KBr, cm⁻¹): 3279 (OH); 3043 (CH, arene); 2918, 2846 (CH, CH₂); 1708, 1655 (C O); 1617 (C N); 1589 (C C). ¹H NMR (600 MHz, DMSO-d₆, ppm): δ 13.43 (¹H, s, N–OH); 10.31 (1H, s, CONHOH); 8.62 (¹H, s, NHOH); 8.04 (¹H, d, J = 1.80 Hz, H-4′); 7.85 (¹H, s, H-5″); 7.46 (¹H, dd, J = 8.10 Hz, J’ = 1.80 Hz, H-6′); 7.13 (¹H, d, J = 7.80 Hz, H-7′); 3.78 (3H, s, N–CH₃); 3.69 (2H, t, J = 7.20 Hz, H-7a, H-7b); 2.26 (3H, s, C–CH₃); 1.93 (2H, t, J = 7.20 Hz, H-2a, H-2b); 1.60–1.58 (2H, m, H-6a, H-6b); 1.49–1.46 (2H, m, H-3a, H-3b); 1.29–1.28 (4H, m, H-4a, H-4b, H-5a, H-5b). 13C NMR (150 MHz, DMSO-d₆, ppm): δ 169.01, 162.88, 143.61, 143.54, 141.01, 130.25, 129.22, 127.96, 125.41, 119.08, 115.57, 109.31, 39.08, 38.14, 32.13, 28.15, 26.88, 25.90, 24.93, 12.94. HRMS (ESI) m/z calculated for C₂₀H₂₅N₅O_4, [M + H]⁺ 400.1979. Found, 400.19836.

2.1.3.3. 7-(5-(4-Fluorophenyl)-3-(hydroxyimino)-2-oxoindolin-1-yl)-N-hydroxyheptanamide (10c)

Yellow solid; yield: 33%; mp: 185.5–186.4 °C; R_f = 0.56 (DCM : MeOH = 9 : 1). IR (KBr, cm⁻¹): 3258 (OH); 3056 (CH, arene); 2934, 2855 (CH, CH₂); 1701, 1654 (C O); 1617 (C N); 1540 (C C).¹H NMR (600 MHz, DMSO-d₆, ppm): δ 10.32 (¹H, s, CONHOH); 8.63 (¹H, s, NHOH); 8.22 (¹H, d, J = 1.80 Hz, H-4′); 7.70 (¹H, dd, J = 8.40 Hz, J′ = 1.80 Hz, H-6′); 7.66–7.63 (2H, m, H-2″, H-6″); 7.28 (2H, td, J = 7.80 Hz, J’′ = 1.80 Hz, H-3″, H-5″); 7.21 (¹H, d, J = 8.40 Hz, H-7′); 3.73 (2H, t, J = 7.20 Hz, H-7a, H-7b); 1.94 (2H, t, J = 7.20 Hz, H-2a, H-2b); 1.62–1.60 (2H, m, H-6a, H-6b); 1.49–1.47 (2H, m, H-3a, H-3b); 1.30–1.29 (4H, m, H-4a, H-4b, H-5a, H-5b). 13C NMR (150 MHz, DMSO-d₆, ppm): δ 169.03, 163.00, 162.62, 160.67, 143.39, 142.42, 136.19, 133.77, 130.38, 128.34, 128.27, 124.99, 115.84, 115.79, 115.62, 109.58, 32.13, 28.17, 26.90, 25.90, 24.94. HRMS (ESI) m/z calculated for C₂₁H₂₂FN₃O_4, [M + H]⁺ 400.1667. Found, 400.16733.

2.2. Cytotoxicity assay (MTT method)

To evaluate compound cytotoxicity, we employed a diverse cell panel including colorectal adenocarcinoma (SW620), colorectal carcinoma (HCT116), triple-negative breast carcinoma (MDA-MB-231), lung carcinoma (A549), prostate carcinoma (PC-3), and normal lung fibroblasts (MRC-5). All cell lines were supplied by the Cancer Cell Bank of the Korea Research Institute of Bioscience and Biotechnology (KRIBB), while culture reagents were sourced from GIBCO (Grand Island, NY, USA). Cells maintained in Dulbecco's Modified Eagle Medium (DMEM) were harvested upon reaching confluency, treated with trypsin, and adjusted to a working density of 3 × 10⁴ cells mL⁻¹. For the MTT assay, 180 µL of the cell suspension was dispensed into each well of a 96-well plate (day 0) and incubated for 24 h at 37 °C in a humidified atmosphere containing 5% CO₂. The cells were then exposed to 30 µL of the test compounds for an additional 24 h, followed by the addition of 20 µL MTT solution (5 mg mL⁻¹) and incubation for 4 h at 37 °C. The resulting formazan crystals were dissolved in 100 µL of DMSO, and absorbance was measured at 570 nm to determine cell viability, calculated as (A₅₇₀ treated/A₅₇₀ control) × 100 (ref. 31 and 32) IC₅₀ values for the active compounds were determined using the Probit analysis method,³³ with each value representing the mean of three independent experiments (SD ≤ 10%).

2.3. HDAC enzymes assay

Inhibitory activity against histone deacetylases was determined using a fluorescence-based HDAC assay kit (Abcam, MA, USA) with HeLa nuclear extracts as the enzyme source. Test derivatives (4a–c, 7a–c, 10a–c), vehicle, and SAHA (reference) were incubated with enzyme and fluorometric substrate at 37 °C for 30 min. Developer reagent was then added, and fluorescence was quantified on a VICTOR plate reader (PerkinElmer, MA, USA) at 360/460 nm. IC₅₀ values were derived from background-corrected dose–response curves using GraphPad Prism.

2.4. Cell cycle analysis

SW620 colorectal cancer cells (2 × 10⁵ cells per mL) were supplied by the Cancer Cell Bank of KRIBB and seeded in 6-well culture plates and incubated for 24 h to permit attachment. Compound exposure at specified concentrations continued for 24 h. Cells were then harvested, washed with ice-cold PBS, fixed in 75% ethanol, and labeled with PI/RNase (30 min, room temperature). DNA content distribution was determined using a FACSCalibur flow cytometer (BD Biosciences), and phase frequencies were calculated with CellQuest Pro software.

2.5. Apoptosis assay

Apoptotic responses were assessed via Annexin V-FITC/PI dual staining. SW620 cells (2 × 10⁵ cells per mL) were cultured in 6-well plates for 24 h prior to compound treatment at the indicated concentrations for 24 h. Following harvest and PBS washing, cells were stained with Annexin V-FITC and PI in 1× binding buffer (15 min, dark, ambient temperature). Flow cytometric acquisition on a FACSCalibur instrument (BD Biosciences) enabled quantification of apoptotic populations using CellQuest Pro analysis software.

2.6. Molecular docking studies

Compound structures were generated from ChemDraw Professional version 16.0 and subsequently subject to energy minimization with an RMS gradient of 0.001 kcal mol⁻¹ Å⁻¹ in MOE (version 2025.06, Chemical Computing Group, the software is operated under a valid institutional academic license managed by Chungbuk National University). All ligands, with hydroxamic acid groups converted to their O-deprotonated keto tautomeric form, were assigned charges according to the 94x variant of the Merck Molecular force field (MMFF94x). The X-ray crystallographic structures of HDAC isoforms were retrieved from Protein Data Bank with the IDs of 4LXZ (HDAC2), 4A69 (HDAC3), 5EEI (HDAC6), and 7U6B (HDAC10). These structures were prepared using the QuickPrep tool in MOE, including the addition of hydrogen atoms, protonation, water removal, atom type editing and assignment of AMBER99 charges. Molecular docking employed rigid receptor/flexible ligand methodology using the Triangle Matcher algorithm in MOE. Among the 20 generated poses, those exhibiting favorable Zn²⁺ coordination geometries were selected for detailed analysis. Binding affinity evaluation employed sequential scoring: London ΔG (primary score) followed by GBVI/WSA ΔG (refinement score) for estimation of binding free energies. Visualization of ligand–protein complexes was performed using BIOVIA Discovery Studio 2024 Client.

2.7. Molecular dynamics simulations

All protein–ligand complexes (SAHA, 7c, 10b, and 10c) were subjected to all-atom molecular dynamics (MD) simulations to evaluate the stability of binding modes and characterize ligand–pocket dynamics. The initial complex structures were obtained from the docking poses. The protein was parameterized using the AMBER ff14SB, while ligand parameters were generated using GAFF2/Antechamber. Partial atomic charges for ligands were assigned using AM1-BCC.^34,35 The catalytic Zn²⁺ center and its coordination environment were treated using non-bonded model to maintain a physically meaningful metal–ligand interaction during the simulations. Each complex was solvated in an explicit TIP3P cubic box with a minimum solute-box distance of 2.5 nm ³⁶ Counterions were added to neutralize the system, followed by addition of 0.15 M NaCl to mimic physiological ionic strength. The MD studies were performed by using GROMACS software suite and input files were generated with CHARMM-GUI.^37,38

Energy minimization was performed using the algorithm of steepest descent until the maximum force was below 1000 kJ mol⁻¹ nm⁻¹. The system was then equilibrated in two stages: (i) NVT equilibration for 125 ps with position restraints on heavy atoms, followed by (ii) NPT equilibration for 500 ps, allowing the solvent and ions to relax while gradually releasing restraints. Temperature was maintained at 303.5 K using the thermostat of V-rescale with a coupling time constant.³⁹ Pressure was controlled at 1 bar using the barostat of Parrinello–Rahman with a coupling time constant.⁴⁰ Long-range electrostatics were computed using Particle Mesh Ewald (PME) with a real-space cutoff of 1.2 nm and van der Waals interactions were truncated at 1.0 nm.⁴¹ All bonds involving hydrogen atoms were constrained using LINCS, enabling an integration time step of 2 fs.⁴² Production MD simulations were performed for 500 ns for each complex under periodic boundary conditions in the NPT. Coordinates were saved every 0.1 ns for subsequent analyses. For fair comparison between ligands, the same simulation protocol and analysis settings were applied to all systems.

Protein and ligand stability were evaluated using backbone (or Cα) RMSD and ligand heavy-atom RMSD, respectively. Protein compactness and flexibility were assessed by the radius of gyration (R_g) and RMSF analyses. Ligand solvent exposure and intermolecular interactions were characterized by SASA and hydrogen bond analyses. The persistence of metal coordination was monitored by tracking the distance between the ligand coordinating atom (O1) and Zn²⁺. Ligand pose clustering was performed based on ligand heavy-atom RMSD using a GROMOS-like algorithm with a cutoff of 0.20 nm. Collective ligand–pocket motions were analyzed using principal component analysis (PCA) on combined ligand and pocket coordinates, retaining only bound-state frames. The first three principal components were used to describe dominant binding dynamics, and representative structures were extracted from extreme PC1 states. Trajectory processing and analyses were carried out using GROMACS tools and in-house Python scripts based on MDAnalysis and NumPy, with data visualization performed using Matplotlib.⁴³ Molecular visualization and structural inspection were conducted using VMD and Discovery Studio Visualizer.^44,45

2.8. In silico ADME and toxicity predictions

The simplified molecular-input line-entry system (SMILES) representing the chemical structure of 10c was generated using ChemDraw Professional version 16.0. The line notation was then employed as the input for the SwissADME platform, available online at https://www.swissadme.ch/, to predict the compound's physicochemical and pharmacokinetic profiles.

3. Results and discussions

3.1. Chemistry

A three-step pathway was designed to synthesize the target 2-oxoindoline-based N-hydroxybenzamides (4a–c), as shown in Scheme 1.

First, N-alkyl isatin derivative (2) were synthesized via a substitution reaction between 5-bromoisatin (1, 1 mmol) and methyl 4-(bromomethyl)benzoate in DMF under a basic condition using K₂CO₃, with a catalytic amount of KI. The reaction proceeded efficiently at ambient temperature, affording the desired product in excellent yields (89%). Second, the Suzuki reaction between N-alkyl isatin derivative (2) and aryl boronic acid pinacol ester was carried out at 80–100 °C with dichloro-bis(tri-o-tolylphosphine)palladium(ii), triphenylphosphine, and K₂CO₃ as catalysts in DMF solvent under a nitrogen atmosphere to reduce the formation of black palladium. These intermediates 3a–c were acquired in moderate yields. In the final step, the target isatin-based hydroxamic acids were obtained by the reaction of hydroxylamine hydrochloride with the corresponding esters 3a–c at 0 °C under basic conditions (NaOH), and methanol was determined to be the most suitable solvent. The desired products 4a–c were obtained in acceptable yields ranging from 29% to 41%.

The targeted compounds, isatin-based N-hydroxypropenamides (7a–c), were similarly synthesized to 4a–c, using methyl (E)-3-(4-(bromomethyl)phenyl)acrylate to replace methyl 4-(bromomethyl)benzoate (Scheme 1). Likewise, several isatin-based N-hydroxyheptanamides (10a–c) were obtained following the same procedure as 4a–c (Scheme 1), but with methyl 7-bromoheptanoate substituted for methyl 4-(bromomethyl)benzoate. Compounds 7a–c and 10a–c were obtained in generally acceptable yields, ranging from 27 to 43%.

The Suzuki cross-coupling reactions afforded the desired products (3a–c, 6a–c, 9a–c) in moderate yields. The relatively lower yields can be attributed to the steric and electronic properties of the N-substituted isatin scaffold, which may hinder oxidative addition of the aryl bromide to the palladium catalyst. In addition, partial protodeboronation of the boronic esters, catalyst deactivation by coordination of heteroatoms, and competing side reactions such as debromination or homocoupling may also contribute to the reduced isolated yields. The subsequent oximation/hydroxamic acid formation (4a–c, 7a–c, 10a–c) also proceeded in moderate to low yields. This may result from the relatively low reactivity of the conjugated C-3 carbonyl of the isatin nucleus, the reversible nature of oxime formation, and competing side reactions occurring under the basic NH₂OH conditions. Moreover, the highly polar oxime–hydroxamic acid products were difficult to purify by column chromatography, which likely further decreased the isolated yields.

All the hydroxamic acid compounds synthesized 4a–c, 7a–c and 10a–c were determined the structure by spectroscopic data, including IR, HRMS, ¹H and ¹³C NMR. In the down-field of the ¹H NMR spectra, single peaks corresponding to NH (∼10 ppm) and OH (∼9 ppm) represent the protons of the hydroxamic acid group. Meanwhile, a singlet signal, occasionally absent from the spectrum, at around 13 ppm is characteristic of the oxime ( NOH) proton.

3.2. Bioactivity

Antiproliferative potential was assessed across a panel of six mammalian cell models: colorectal lines (SW620, HCT116), breast carcinoma (MDA-MB-231), pulmonary carcinoma (A549), prostatic carcinoma (PC-3), and normal lung fibroblasts (MRC-5). Reference compounds SAHA and adriamycin (ADR) were included for comparative evaluation. Cytotoxicity data are presented in Table 1.

Table 1. HDAC inhibitory activity and cytotoxicity of the synthesized compounds against selected cancer cell lines.

Cpd	R	HDAC (Hela extract) inhibiton (IC50a, µM)	Cytotoxicity (IC50a, µM)/cell linesb
			SW620	HCT116	MDA-MB-231	A549	PC-3	MRC-5
4a		0.300 ± 0.085	6.97 ± 0.11	2.58 ± 0.07	4.59 ± 0.45	12.84 ± 0.01	3.42 ± 0.57	13.02 ± 0.62
4b		0.149 ± 0.009	239.83 ± 15.94	26.08 ± 5.49	39.83 ± 13.91	127.76 ± 27.47	11.45 ± 3.35	424.25 ± 102.75
4c		6.446 ± 0.411	6.57 ± 0.77	2.66 ± 0.58	3.95 ± 0.48	21.75 ± 12.74	3.94 ± 0.70	15.06 ± 2.76
7a		0.418 ± 0.039	3.53 ± 0.27	0.78 ± 0.04	1.41 ± 0.01	4.00 ± 2.10	2.27 ± 0.31	5.36 ± 3.06
7b		1.003 ± 0.061	Non toxic	8.89 ± 0.22	31.10 ± 8.34	79.91 ± 6.46	4.74 ± 1.28	23.70 ± 8.35
7c		0.078 ± 0.002	5.13 ± 0.39	0.94 ± 0.03	2.10 ± 0.50	4.63 ± 0.97	2.49 ± 0.31	12.21 ± 1.93
10a		0.480 ± 0.023	4.54 ± 0.07	1.59 ± 0.37	0.81 ± 0.05	8.80 ± 2.61	0.91 ± 0.54	12.58 ± 3.93
10b		0.043 ± 0.001	16.65 ± 1.45	4.82 ± 0.45	6.29 ± 0.43	8.16 ± 1.97	9.19 ± 1.05	41.49 ± 16.72
10c		0.014 ± 0.001	3.62 ± 0.31	2.95 ± 0.11	2.06 ± 0.21	15.68 ± 3.60	6.31 ± 0.07	72.45 ± 51.35
SAHAc		0.122 ± 0.003	1.65 ± 0.25	1.18 ± 0.31	3.51 ± 0.65	9.48 ± 0.40	3.93 ± 0.68	8.22 ± 0.44
ADRd		#e	0.35 ± 0.28	0.26 ± 0.14	1.71 ± 0.35	0.26 ± 0.05	0.16 ± 0.05	0.87 ± 0.15

^aThe concentration (µM) of compounds that produces a 50% reduction in enzyme activity or cell growth.

^bCell lines: SW620 (human colorectal adenocarcinoma), HCT116 (human colorectal carcinoma), MDA-MB-231 (human triple-negative breast cancer), A549 (human lung carcinoma), PC-3 (human prostate carcinoma), and MRC-5 (non-malignant human lung fibroblasts).

^cSAHA: suberoylanilide hydroxamic acid, a positive control, this positive control was evaluated in the same assay batch together with the compounds described in [ref. 54].

^dADR: adriamycin, a positive control.

^eNot tested.

The biological evaluation indicated that several synthesized hydroxamic acids exhibited substantial HDAC inhibitory activity together with diverse cytotoxic profiles across the tested cancer cell lines. All nine compounds were capable of inhibiting HDACs extracted from HeLa cells, with IC₅₀ values in the micromolar to submicromolar range. Among them, compounds 7c, 10b, and 10c displayed the most potent activity, with IC₅₀ values of 0.078 ± 0.002 µM, 0.043 ± 0.001 µM, and 0.014 ± 0.001 µM, respectively, all outperforming the reference inhibitor SAHA (IC₅₀ = 0.122 ± 0.003 µM). The enhanced inhibitory effects of these derivatives are likely attributed not only to the favorable positioning of their aromatic substituents, which may strengthen zinc-binding interactions and facilitate deeper accommodation within the HDAC active-site pocket, but also to the compatibility between these substituents and the distinct linker architectures. An appropriate balance of linker length and flexibility appears to support optimal spatial arrangement of the cap group and the hydroxamic acid moiety, thereby improving overall binding affinity. In contrast, derivatives bearing unsubstituted phenyl groups in the benzamide, cinnamamide, and heptanamide series exhibited weaker HDAC inhibition, highlighting the significant role of both electronic and steric factors in modulating binding efficiency. A notable structural trend was observed in the 4-fluorophenyl-substituted analogues: while this modification markedly enhanced activity in the cinnamamide and heptanamide series, it did not provide similar benefits in the benzamide scaffold. As a consequence of these structural influences, compound 4c showed the weakest HDAC inhibition, with an IC₅₀ value of 6.446 ± 0.411 µM, approximately fifty-three times higher than that of SAHA.

In the cytotoxicity assays, several derivatives exhibited notable and, in some cases, selective antiproliferative activity toward the evaluated cancer cell lines. Consistent with their potent HDAC inhibition profiles, compounds 7c, 10b, and 10c also demonstrated strong cytotoxic effects, with IC₅₀ values ranging from 0.94–5.13 µM for 7c, 4.82–16.65 µM for 10b, and 2.06–15.68 µM for 10c across the five cancer cell lines. Interestingly, several derivatives with relatively weak HDAC inhibition still showed considerable antiproliferative activity. For example, compound 4c, despite exhibiting poor HDAC inhibition (IC₅₀ = 6.446 ± 0.411 µM), inhibited the growth of all tested cancer cell lines with IC₅₀ values between 2.66 ± 0.58 µM (HCT116) and 21.75 ± 12.74 µM (A549). Similarly, compound 10a, whose HDAC inhibitory potency was approximately fourfold weaker than SAHA, displayed even stronger cytotoxicity than SAHA against several cell lines, including MDA-MB-231, A549, and PC-3. These findings suggest that certain derivatives may exert cytotoxic effects through additional mechanisms beyond HDAC inhibition or may possess physicochemical properties, such as more favorable polarity or membrane permeability, that enhance their cellular accumulation and biological activity. In contrast, compound 4b presented an opposing trend. Although it inhibited HDAC with potency comparable to SAHA (IC₅₀ = 0.149 ± 0.009 µM vs. 0.122 ± 0.003 µM for SAHA), its cytotoxicity toward all tested cancer cell lines was substantially weaker, with IC₅₀ values ranging from 11.45 ± 3.35 µM to 239.83 ± 15.94 µM. Pronounced differences in cell-line sensitivity were also observed; for instance, compound 7b inhibited PC-3 cells with an IC₅₀ of 4.74 ± 1.28 µM, yet was approximately 17-fold less active against A549 cells and showed no detectable cytotoxicity in SW620 cells. The study also explored the effect of varying linker types on cytotoxicity; however, based on the current dataset, no clear relationship between linker architecture and cellular potency could be established.

An important consideration in anticancer agent development is the discrimination between malignant and non-malignant cells. In this regard, MRC-5, a normal fibroblast line, was included to evaluate off-target cytotoxicity. Most compounds exhibited higher IC₅₀ values against MRC-5 compared with cancer cell lines, indicating an acceptable degree of selectivity. Notably, compound 10c retained relatively low toxicity toward MRC-5 (IC₅₀ = 72.45 ± 51.35 µM) while demonstrating substantial cytotoxicity toward the SW620, HCT116, and MDA-MB-231 cancer cell lines, with IC₅₀ values of 3.62 ± 0.31 µM, 2.95 ± 0.11 µM, and 2.06 ± 0.21 µM, respectively. This favorable differential suggests that 10c possesses promising therapeutic potential with a widened safety margin. Conversely, compounds such as 7a and 7b showed less pronounced selectivity, with comparable IC₅₀ values between malignant and normal cells, suggesting that further structural optimization is required to reduce undesired cytotoxicity.

The structure–activity relationship (SAR) analysis of the synthesized compounds is summarized in Fig. 4, indicating that the inhibitory activity was influenced by both the nature of the aromatic substituents and the linker moiety. The 4-fluorophenyl substituent enhanced potency only in the cinnamamide and heptanamide series, whereas the unsubstituted phenyl derivatives exhibited relatively weak HDAC inhibitory activity but strong antiproliferative effects. In contrast, the pyrazole-containing derivatives showed the weakest cytotoxic activity. Regarding the linker moiety, the six-carbon aliphatic linker generally conferred stronger HDAC inhibitory activity than the aromatic linker, which may be attributed to its greater conformational flexibility.

Fig. 4. SAR of the novel hydroxamic acids incorporating 2-oxoindoline scaffold 4a–c, 7a–c and 10a–c.

Given compound 10c's potent HDAC inhibition and broad-spectrum antiproliferative properties demonstrated across multiple cancer models, further mechanistic studies were undertaken to elucidate its impact on cell-cycle progression and apoptotic pathways in SW620 colorectal carcinoma (Fig. 5).

Fig. 5. Cell cycle phase distribution induced by compound 10c in SW620 colorectal cancer cells (5 × 10⁵ cells). Cells were exposed to 10c (5 µM) or SAHA (5 µM, reference) for 24 h in triplicate experiments. Following PI/RNase staining, DNA content analysis was performed by flow cytometry. Controls: UN (untreated), VH (vehicle: 0.05% DMSO). Results displayed as flow histograms (left panel) and quantified cell-cycle phase data (right panel).

To investigate the cellular effects of the compounds, SW620 cells were treated with 10c and SAHA (5 µM, 24 h) and analyzed by flow cytometry and apoptosis assays. Compound 10c induced a pronounced S-phase arrest (44.66 ± 0.94%), whereas SAHA mainly caused a G₂/M arrest (30.30 ± 0.25%) with a weaker accumulation in the S phase (29.55 ± 0.59%). Annexin V-FITC/PI dual staining further demonstrated that both agents significantly increased early and late apoptosis, with 10c showing pro-apoptotic activity comparable to SAHA (Fig. 6). At the tested concentration, a substantial necrotic fraction was also detected, with necrosis rates of 11.20 ± 0.10% for SAHA and 9.74 ± 0.75% for 10c. Consistent with these findings, microscopic observations revealed that 10c induced characteristic apoptotic morphology, including cell shrinkage and membrane blebbing, similar to those observed in SAHA-treated cells (Fig. 7).

Fig. 6. Annexin V-FITC/PI flow cytometric analysis of apoptotic responses to compound 10c in SW620 cells (5 × 10⁵ cells). Cells received 24 h incubation with 10c (5 µM) or SAHA (5 µM control). After dual staining (Annexin V-FITC/PI in RNase-containing buffer), populations were characterized by flow cytometry. Quadrant definitions: area 1, PI+ (late apoptotic/secondary necrotic); area 2, Annexin V+ (early apoptotic). Controls as above. Dot plots displayed alongside quantified apoptotic percentages.

Fig. 7. Brightfield microscopy of morphological alterations in SW620 cells upon exposure to compound 10c. Cells (2.5 × 10⁵ cells per mL, 2 mL per well) were pre-equilibrated for 24 h prior to 24 h treatment with 10c (5 µM) or SAHA (5 µM). Images captured using a ZEISS Celldiscoverer 7 microscope at magnifications of 20× (Panel A) and 40× (Panel B). Scale bars: 100 µm (A); 50 µm (B).

3.3. Docking studies

Docking simulations undertaken during lead discovery could provide beneficial insights into the ligands' interaction profiles within the enzyme's binding domain, thereby supporting the rationale behind the design of our isatin-based derivatives as novel HDAC inhibitors. To confirm the applied docking method, re-docking simulations were performed on HDAC2 (PDB code: 4LXZ) and HDAC6 (PDB code: 5EEI), with SAHA (suberoylanilide hydroxamic acid) co-crystallized in the structure of these two HDAC isoforms. The notably low values of RMSD (root mean square deviation), specifically 0.225 Å for HDAC2 and 0.652 Å for HDAC6, indicated that the native and re-docked SAHA conformations practically overlapped, as reinforced in Fig. 8. These results, therefore, demonstrated the reliability of the docking procedures.

Fig. 8. Structural overlay of co-crystallized (yellow) and re-docked (green) conformations of SAHA within HDAC2 (A) and HDAC6 (B).

Adhering to the validated protocols, SAHA, serving as the reference control, and three series 4a–c, 7a–c, and 10a–c were consecutively docked into the binding cavities of some HDAC isozymes, namely HDAC2, HDAC3, HDAC6, and HDAC10, with the respective PDB IDs of 4LXZ, 4A69, 5EEI, and 7U6B. Our synthesized derivatives, along with SAHA, were appropriately accommodated in the enzymatic pockets of these four HDAC isoforms, providing concrete evidence of proper binding geometries exhibited among the ligands. Their favorable docking scores and short bond distances between the oxygen atoms of the chelators and the zinc cofactor (the critical element in metal-dependent HDACs, contributing to their lysine deacetylase activity of N-terminal histone tails) were briefly summarized in Table 2, indicating potential potency as HDAC inhibitors. As a means of exemplifying the plausible binding patterns of the ligands, the docking poses of all compounds within HDAC2 are illustrated in Fig. 9.

Docking scores and bond distances of series 4a–c, 7a–c, 10a–c, and SAHA into the catalytic pockets of HDAC2, HDAC3, HDAC6, and HDAC10.

HDAC2					HDAC3
Cpd	E_Score1a	E_Score2a	Distance to Zn2+b		Cpd	E_Score1a	E_Score2a	Distance to Zn2+b
			–OH	O				–OH	O
4a	−15.3083	−19.8244	1.97	2.34	4a	−22.4294	−15.0307	2.05	2.54
4b	−14.5944	−19.8599	1.97	2.35	4b	−14.6338	−15.1421	2.05	2.57
4c	−18.1872	−19.8388	1.97	2.33	4c	−11.0437	−15.0780	2.06	2.54
7a	−12.7655	−20.3266	1.97	2.30	7a	−12.2639	−15.5480	2.23	2.12
7b	−13.6115	−20.5790	1.96	2.33	7b	−11.9439	−15.7327	2.22	2.15
7c	−17.2179	−20.6237	1.92	2.64	7c	−11.3263	−15.3030	2.21	2.16
10a	−10.9332	−21.3818	1.94	2.45	10a	−13.1687	−17.6744	2.04	1.97
10b	−10.6470	−21.0128	1.94	2.42	10b	−10.8050	−17.9185	2.05	2.04
10c	−13.0789	−20.8870	1.94	2.40	10c	−13.1697	−17.8862	2.03	1.99
SAHA	−13.1587	−20.4148	1.94	2.41	SAHA	−15.4134	−20.0710	2.06	2.07

^aThe docking score (kcal mol⁻¹) calculated from the London (with refinement) and GBVI/WSA affinity scoring function from MOE software.

^bDistances (Å) from oxygen atoms ( O and –OH) of hydroxamate group to zinc ion.

HDAC6					HDAC10
Cpd	E_Score1a	E_Score2a	Distance to Zn2+b		Cpd	E_Score1a	E_Score2a	Distance to Zn2+b
			–OH	O				–OH	O
4a	−12.7199	−17.1759	2.01	2.23	4a	−15.4638	−19.0049	1.99	2.27
4b	−12.9773	−17.4105	2.01	2.23	4b	−18.4534	−19.1470	1.99	2.27
4c	−13.0645	−17.2923	2.01	2.26	4c	−16.1583	−19.0475	1.99	2.27
7a	−12.5379	−17.2273	2.02	2.14	7a	−17.4411	−20.3480	1.90	2.29
7b	−12.0600	−17.8052	2.02	2.14	7b	−15.1667	−20.7154	1.90	2.29
7c	−19.7729	−17.3095	2.02	2.14	7c	−14.5957	−20.4082	1.90	2.29
10a	−10.8377	−17.5156	2.00	2.26	10a	−15.2399	−21.2297	1.89	2.25
10b	−9.4171	−17.6367	1.99	2.25	10b	−20.4374	−19.8997	1.91	2.26
10c	−10.2946	−17.4013	1.99	2.25	10c	−13.1907	−20.1232	2.10	1.96
SAHA	−13.0754	−16.8278	2.02	2.17	SAHA	−27.7134	−19.8535	1.87	2.11

Fig. 9. Three-dimensional visualization showing the binding poses of the designed ligands within HDAC2. The ligands were colour-coded in green #00ff00 (4a), red #ff0000 (4b), blue #0000ff (4c), violet #aa55ff (7a), orange #ff5500 (7b), magenta #ff00ff (7c), sky blue #00aaff (10a), cyan #00ffff (10b), indigo gray #55557f (10c), yellow #ffff00 (SAHA) with corresponding RGB codes.

Also contributing to the binding affinity of the ligands were many intramolecular interactions from three typical pharmacophore features of our newly synthesized hydroxamate-based HDAC inhibitors to multiple residues within the HDAC binding pockets. Consistently detected in all isoforms, the hydroxamic acid moieties generally facilitated the formation of three hydrogen bonds towards one tyrosine and two histidine catalytic residues, implying the rational design of our novel compounds capable to reach the bottom of the HDAC pockets and establish bidentate zinc coordination. Astonishingly, the linker regions of the three series shared similar interactions with various residues lining across the tube-like channels of HDACs. The benzamide moiety of series 4a–c displayed π–π stackings with His146, Phe210 (for HDAC2), or Phe144, His135, Phe200 (for HDAC3), or Phe583 (for HDAC6), or His137 (for HDAC10). The cinnamamide scaffold of 7a–c, while maintaining this trend, also engaged in unconventional non-covalent interactions with the side chains of certain aspartate or serine residues. Inspired by the identical structural motif in SAHA that enables it to fit into the narrow hydrophobic tunnels, the long and straight heptanamide fragment of 10a–c adopted mostly comparable interactions. Like SAHA, 10a–c displayed van der Waals contacts with a variety of residues, e.g. His183, His146, Phe155, Gly154 (for HDAC2), or Cys145, Phe144, His172, His135 (for HDAC3), or His574, His614, Phe643, Gly582 (for HDAC6), or His176, Phe146, Trp205 (for HDAC10). While the isatin-based cap group of series 4a–c, coupled with reasonably rigid N-hydroxybenzamide template, participated in analogous contacts when docked into four investigated HDAC isoforms, the same heterocyclic fragment of other synthesized derivatives exhibited fluctuating interactions with residues on the entrance of HDAC2 and HDAC3 cavities. This observation suggested that the similarity in binding behaviour of compounds in a series within HDAC active sites might be influenced, to some extent, by their structural flexibility. The 5-position substituents incorporated into the isatin ring of the compounds occasionally formed π–π contacts with residues in the HDAC surface-recognition region, thereby contributing to the stabilization of binding modes. It was noteworthy that the three substituents shared binding interactions during simulations, including consistent π–π bonds with Pro34 (for series 4a–c), or His33 (for series 7a–c and 10a–c) within HDAC2. Therefore, these aromatic groups might be potential fragments for future design of compounds aimed at improving selectivity of various derivatives against HDAC2.

Compared with SAHA, 4a–c exhibited more negative docking scores when docked against HDAC6, whereas the opposite was true for the remaining isoforms, indicating their weaker inhibitory activity as a whole. Series 10a–c performed better across most HDAC isozymes, except for HDAC3; meanwhile, 7a–c barely showed any discernible trend in inhibitory activity relative to SAHA. These findings implied the progressive increase in HDAC binding affinity for compounds 4a–c, 7a–c, and 10a–c, sequentially. The docking results were accordingly in substantial concordance with biological assessment data, as compounds 10a-c exhibited the greatest HDAC inhibition activity, reflected by their lowest IC₅₀ values. Briefly, our isatin-based derivatives, especially those containing the N-hydroxyheptanamide moiety, could serve as scaffolds for structural optimization towards more potent HDAC inhibitors.

3.4. Molecular dynamics simulations

An integrated analysis combining biological activity data and molecular dynamics (MD) simulations was performed to elucidate the binding behavior and structure–activity relationships of SAHA, 7c, 10b, and 10c. Across all simulated systems, the protein backbone RMSD remained stable throughout the 500 ns trajectories, with average values ranging from 1.30 to 2.43 Å, indicating that ligand binding did not perturb the global fold of the enzyme (Table 3). The radius of gyration (R_g) was similarly conserved (20.45–20.64 Å), confirming that overall protein compactness was maintained in all complexes. Among the ligands, the 10c-bound system exhibited the lowest R_g and smallest fluctuations, suggesting a slightly enhanced stabilizing effect on the protein structure.

Table 3. Summary of structural and energetic properties of protein–ligand complexes calculated from the last 500 ns of MD simulations.

	Cpd. 7c	Cpd. 10b	Cpd. 10c	SAHA
Protein RMSD (Å)	2.03 ± 0.57	2.27 ± 0.58	1.30 ± 0.20	2.43 ± 0.36
Ligand RMSD (Å)	10.00 ± 2.49	17.60 ± 14.21	6.40 ± 1.97	6.90 ± 3.57
Radius of gyration (Rg – Å)	20.56 ± 0.25	20.56 ± 0.17	20.45 ± 0.12	20.64 ± 0.18
Solvent accessible surface areas (SASA – Å2)	722 ± 25	723 ± 24	700 ± 25	560 ± 20
–OH – Zn2+ distance (Å)	5.35 ± 3.74	23.0 ± 10.4	4.94 ± 2.46	2.0 ± 0.01
Number of hydrogen bonds	2.50 ± 1.41	1.31 ± 1.36	1.45 ± 0.82	0.53 ± 0.69

In contrast to the protein backbone, ligand RMSD revealed differences in binding dynamics among the compounds. SAHA and 10c showed relatively low and well-converged ligand RMSD values (Table 3), consistent with persistent binding modes. Compound 7c displayed moderate fluctuations (10.0 ± 2.49 Å), indicative of increased flexibility within the binding pocket. Compound 10b exhibited the largest ligand RMSD (17.60 ± 14.21 Å), reflecting higher conformational mobility and transient positional rearrangements during the simulation rather than complete dissociation from the active site. Overall, these results suggest a gradual decrease in binding persistence following the order 10c > 7c > 10b, in line with the observed activity trend within this subset of compounds.

Hydrogen bond analysis indicated that 7c formed the highest average number of hydrogen bonds, followed by 10c and 10b, whereas SAHA exhibited comparatively fewer interactions. However, the total hydrogen bond count did not directly correlate with inhibitory potency, suggesting that the nature of metal coordination and the stability of ligand positioning are more decisive factors. Regarding metal interactions, SAHA maintained a highly stable O1–Zn distance of 2.0 ± 0.01 Å, reflecting strong and persistent chelation. In contrast, the potent compound 10c exhibited a dynamic binding mode with an average distance of 4.94 ± 2.46 Å. Although lacking the rigid coordination observed in SAHA, the zinc-binding group of 10c remained spatially confined within the catalytic vicinity, likely anchored by hydrophobic interactions. Compound 7c displayed a similar but more fluctuating profile (5.35 ± 3.74 Å), whereas 10b showed a large average distance (23.0 ± 10.4 Å), indicating significant displacement from the metal center. Consistent with these structural dynamics, SASA analysis revealed that SAHA was the most deeply embedded ligand, while 7c, 10b, and 10c were more solvent-exposed, with 10b showing the highest average exposure.

To further characterize binding mode heterogeneity, ligand pose clustering based on heavy-atom RMSD was performed using a GROMOS-like algorithm. SAHA and 10c were dominated by a single, highly populated cluster, indicating a well-defined and persistent binding pose. Compound 7c exhibited several medium-sized clusters, reflecting moderate conformational variability within the pocket. In contrast, 10b populated multiple smaller clusters, indicative of enhanced pose diversity and dynamic sampling of alternative binding orientations rather than a single rigid binding mode (Fig. 10).

Fig. 10. Representative ligand sampling within the binding pocket obtained from MD simulations. The protein is shown as a semi-transparent surface, the ligand conformational space is depicted in green, and the catalytic Zn²⁺ ion is shown in gray.

Collective ligand–pocket motions were further explored using principal component analysis (PCA) on combined coordinates of ligand heavy atoms and surrounding pocket residues. Projection onto PC1–PC2–PC3 space revealed compact and well-defined basins for SAHA and 10c, consistent with restricted conformational sampling. Compound 7c explored a broader PCA region, in agreement with the presence of multiple metastable poses (Fig. 11). Notably, compound 10b displayed the widest PCA distribution, reflecting large-amplitude ligand and pocket motions and frequent transitions between dynamically accessible binding states. Representative structures extracted from the extrema of PC1 illustrated that, while SAHA and 10c undergo only subtle conformational adjustments, 10b samples distinct binding geometries, including states with reduced Zn²⁺ coordination and increased solvent exposure.

Fig. 11. Principal component analysis (PCA) of ligand–pocket collective motions obtained from MD simulations. The top panels show three-dimensional projections onto PC1–PC2–PC3 space, while the bottom panels display two-dimensional PC1–PC2 projections colored by PC3. The distributions illustrate differences in conformational sampling and binding dynamics among the studied complexes.

The interaction timelines shown in Fig. 12 provide residue-level insight into the persistence and nature of contacts formed during the simulations, including hydrophobic interactions, van der Waals contacts, hydrogen bonds, π–π stacking, and π–cation interactions. SAHA exhibited a highly persistent interaction pattern characterized by long-lived contacts with key active-site residues, including stable hydrophobic and van der Waals interactions with Phe144, Phe199, Tyr198, and Tyr297, as well as sustained coordination with the catalytic Zn²⁺ ion that were also as observed in our prior work.⁴⁶ These interactions were maintained throughout most of the trajectory, consistent with the low ligand RMSD, compact PCA basin, and strong inhibitory activity of SAHA. Compound 10c exhibited a largely SAHA-like interaction profile, characterized by recurrent and long-lasting contacts with key catalytic-pocket residues, including Phe144, Asp258, Asp263, Arg264, Leu265, and Tyr297. Although short-lived disruptions were observed, both hydrogen bond and hydrophobic interactions were rapidly re-established, indicating a resilient and persistent binding mode that accounts for its high binding stability and biological potency despite weaker average Zn²⁺ coordination. Compound 7c showed a more heterogeneous interaction pattern. While it formed the highest average number of hydrogen bonds, these interactions were predominantly intermittent and redistributed among multiple residues over time. Contacts with aromatic residues such as Phe144 and Tyr198 were maintained only transiently, consistent with its moderate ligand RMSD, multiple medium-sized clusters, and broader PCA distribution. This dynamic interaction behavior supports an intermediate activity profile driven by flexible yet partially productive binding.

Fig. 12. Protein–ligand interaction profiles obtained from MD simulations for compounds 7c, 10b, and 10c. Each row represents an interacting residue, and the colored bars indicate the presence of specific interaction types over the trajectory frames, including hydrophobic contacts, van der Waals contacts, hydrogen bond acceptor/donor interactions, π–cation interactions, and π–π stacking. The interaction timelines highlight differences in interaction persistence and dynamic contact redistribution among the compounds studied.

Compound 10b displayed the most dynamic interaction behavior among the three compounds, with contacts redistributed across different residue subsets throughout the trajectory. Importantly, 10b did not completely disengage from the binding pocket; instead, it maintained recurring hydrophobic and van der Waals interactions with catalytically relevant residues, including Asp258, Asp263, Arg264, Leu265, and Tyr297. These interactions appeared in distinct temporal windows, indicating that 10b samples multiple transient yet productive binding orientations rather than a single dominant pose.

3.5. In silico ADME and toxicity predictions

Based on its notable inhibitory and antiproliferative activity compared with other newly synthesized derivatives, compound 10c was selected to investigate its pharmacokinetic and drug-likeness properties utilizing the SwissADME online tool. The most critical descriptors are summarized in Table 4. The results revealed that the structure of 10c satisfies various drug-likeness benchmarks, including those suggested by Lipinski,⁴⁷ Ghose,⁴⁸ Veber,⁴⁹ Egan,⁵⁰ and Muegge.⁵¹ This agreement might be attributed to the ligand's calculated physicochemical parameters, which generally fall within the acceptable ranges defined by such standards. In addition, the compound was predicted to be effectively absorbed through the gastrointestinal tract, as substantiated by its favorable lipophilicity and water solubility, reflected by its log P_o/w and log S values, respectively. The chemical structure of 10c, however, suffers from certain shortcomings. For instance, this compound was predicted to inhibit CYP3A4 and CYP2C9, two predominant cytochrome P450 isoforms,⁵² potentially increasing the risk of drug interactions with other active pharmaceutical ingredients, given the frequent use of combination therapies in cancer treatment. Assessments based on the more stringent lead-likeness criteria proposed by Teague identified several violations, including values exceeding the recommended thresholds for molecular weight, number of rotatable bonds, and X log P.⁵³ Although clinically approved substances do not necessarily comply with such strict criteria, such as the renowned pan-HDAC inhibitor vorinostat (SAHA), those observations still provide possible areas for structural enhancements during the development of promising HDAC-targeted inhibitors in future studies.

Table 4. Physicochemical and drug-likeness characteristics of compound 10c.

Properties		Cpd. 10c
Physicochemical properties	Molecular weight (g mol−1)	399.42
	Number of heavy atoms	29
	Fraction of Csp3	0.29
	Number of rotatable bonds	9
	Number of hydrogen bond acceptors	6
	Number of hydrogen bond donors	3
	Topological polar surface area (Å2)	102.23
Lipophilicity	Consensus log Po/w	3.27
Water solubility	log S	−4.43 (ESOL)
		−5.65 (ali)
		−6.21 (SILICOS-IT)
Pharmacokinetics	Gastrointestinal absorption	High
	Blood–brain barrier permeant	No
	P-gp substrate	No
	CYP1A2 inhibitor	No
	CYP2C19 inhibitor	No
	CYP2C9 inhibitor	Yes
	CYP2D6 inhibitor	No
	CYP3A4 inhibitor	Yes
Druglikeness	Lipinski	Yes; 0 violation
	Ghose	Yes
	Veber	Yes
	Egan	Yes
	Muegge	Yes
	Bioavailability score	0.55
Medicinal chemistry	Pan-assay interference compounds	0 alerts
	Brenk	4 alerts: hydroxamic_acid, imine_1, oxime_1, oxygen-nitrogen_single_bond
	Leadlikeness	No; 3 violations: MW > 350, Rotors > 7, XLOGP3 > 3.5

4. Conclusions

In summary, a series of novel 2-oxoindoline-based hydroxamic acids was successfully synthesized and evaluated for HDAC inhibitory activity and anticancer potential. The results demonstrate that structural variations in both the linker and substituents on the isatin scaffold markedly influence HDAC inhibition and cytotoxic profiles. Among the evaluated compounds, the N-hydroxyheptanamide derivatives exhibited superior HDAC inhibitory activity, with several compounds showing potency comparable to or greater than that of SAHA. Notably, compound 10c emerged as a promising lead, displaying broad antiproliferative effects against multiple cancer cell lines while maintaining lower toxicity toward non-malignant cells. Mechanistic studies in SW620 colorectal cancer cells confirmed that 10c induced S-phase cell cycle arrest and promoted apoptosis, supporting its HDAC-mediated anticancer activity. Molecular docking analyses suggested favorable binding modes within the HDAC active site, which were further elucidated by molecular dynamics simulations. The MD results demonstrated stable protein–ligand interactions without perturbation of the overall enzyme structure and highlighted distinct binding dynamics among the most active derivatives, with 10c exhibiting the highest binding persistence, followed by 7c and 10b. Furthermore, in silico ADME and toxicity predictions for compound 10c suggested acceptable drug-like properties and a favorable safety profile. Collectively, these findings highlight 2-oxoindoline-based hydroxamic acids, particularly those bearing extended alkyl linkers, as promising scaffolds for the further development of HDAC-targeted anticancer agents.

Conflicts of interest

The authors report no conflicts of interest.

Data availability

The datasets supporting this article have been uploaded as part of the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d6ra04679g.