Development of novel Copper(II)–hydrazone complexes for targeting the inhibition of human osteosarcoma growth

1 Introduction

Cancer remains a leading cause of mortality worldwide, with osteosarcoma (OS) standing out as the most prevalent primary bone malignancy in children and young adults (Bray et al., 2024). OS is characterized by rapid proliferation, high invasiveness, and the production of immature osteoid matrix, which together contribute to poor prognosis (Kansara et al., 2014). Current treatment strategies combining surgical resection with multi-agent chemotherapy, commonly including methotrexate, doxorubicin, and cisplatin, have improved survival rates only modestly (Li et al., 2023). Despite these efforts, relapse and metastasis occur in a substantial fraction of patients, underscoring the urgent need for novel and more effective therapeutic approaches (Moreno et al., 2017).

In this context, metal-based drugs have emerged as a promising alternative, leveraging the unique chemical and biological properties of transition metals (León, 2024; Santa Maria de la Parra et al., 2024a). Their versatility stems from redox activity, coordination geometry, and the capacity to engage in diverse interactions with biomolecules, thereby enabling selective cytotoxicity in tumor cells (Ferretti and León, 2022; Lee et al., 2020). Among these, copper-based complexes have garnered significant attention due to their potent antitumor activity observed across multiple cancer models (Balsa et al., 2023a; Balsa et al., 2025). Copper compounds exert their anticancer activity through multiple mechanisms, including generation of reactive oxygen species (ROS), induction of oxidative DNA damage, proteasome inhibition, and selective targeting of cancer stem cells (Balsa et al., 2023b; Balsa et al., 2023c; Brooks et al., 2026; Del Bello et al., 2022; Fernández et al., 2024; Levín et al., 2021; Rodríguez et al., 2019; Romo et al., 2021). Importantly, the bioactivity of these compounds is often enhanced by rational ligand design, highlighting the critical role of metal–ligand coordination chemistry in modulating reactivity and specificity (Burgos-López et al., 2022).

Despite their promising attributes, only a limited number of copper complexes have been explored against OS, leaving a considerable gap in the development of bioinorganic strategies for this aggressive malignancy (Burgos-Lopez et al., 2019; Cadavid-Vargas et al., 2017). Therefore, the rational design, synthesis, and biological evaluation of copper-based metallodrugs represent a compelling avenue to overcome the limitations of conventional therapies and offer novel, mechanism-driven interventions for osteosarcoma.

N-acylhydrazones constitute a versatile class of ligands exhibiting notable pharmacological properties, including anticancer, antidiabetic, and antioxidant activities, largely attributed to their strong metal-chelating capabilities. Within the framework of a research program aimed at elucidating the relationship between ligand structure and anticancer activity, we have designed and synthesized a series of N-acylhydrazone ligands (L3–L9 ligands in Figure 1) and their corresponding copper (II) complexes. These compounds have demonstrated significant cytotoxicity against a range of solid tumors, with particularly pronounced activity in osteosarcoma and breast cancer models (Balsa et al., 2021; Balsa et al., 2023b; Balsa et al., 2023d; Balsa et al., 2024; Burgos-Lopez et al., 2019; Santa Maria de la Parra et al., 2026a; Santa Maria de la Parra et al., 2026b; Santa Maria de la Parra et al., 2024b; Santa Maria de la Parra et al., 2026).

Our previous studies have shown that structural differences in the hydrazide group of the ligand are reflected in the potency of their anticancer activity (furan = methoxy-benzohydrazide > hydroxy-benzohydrazide). Furthermore, we found that coligands such as o-phen and bipy enhance the biological activity of the complexes in osteosarcoma cells. Notably, the chlorine substitution in Cu₄(L6)₄ was shown to increase the potency of the cytotoxic effect by four times compared to CuL5, surpassing the effect of the change from furan to thiophene (Santa Maria de la Parra et al., 2026a). Since we found that the cytotoxic activity of the complexes increases with greater nuclearity (Espindola-Moreno et al., 2025), the observed effect of Cu₄L6₄ may be partially due to its higher Cu nuclearity than that of the monomeric complexes. Additionally, we found that tridentate compounds exhibited higher anticancer activity than structurally related bidentate complexes (Santa Maria de la Parra et al., 2026).

Building on these findings, we report here the synthesis and characterization of two novel Cu(II) coordination complexes (CuL1 and Cu₂L2₂) with the hydrazone ligands (E)-N'-(2-hydroxybenzylidene)furan-2-carbohydrazide (L1) and (E)-N'-(2-hydroxybenzylidene)-4-methoxybenzohydrazide (L2), together with their X-ray crystal structures. Furthermore, our study focuses on the assessment of the in vitro anticancer activity of the ligands and the new Cu(II) complexes using complementary 2D monolayer and 3D multicellular spheroid OS models derived from MG-63 osteosarcoma cells.

2 Materials and methods

2.1 Materials

Dulbecco’s modified Eagle medium (DMEM) and TrypLE™ were purchased from Gibco (Gaithersburg, MD, USA). Fetal bovine serum (FBS) was bought from Internegocios S.A. (Mercedes, Argentina). OS (MG-63) cell culture line was acquired from the American Type Culture Collection (ATCC). Tissue culture materials were purchased from Jet Bio-Filtration Co. (Guangzhou, China). Tetrazolium salt MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazoliumbromide) was supplied by Invitrogen Co. (Buenos Aires, Argentina). Agarose was purchased from Inbio Highway, Argentina. Resazurin was purchased from Santa Cruz Biotechnology, Dallas, United States. Other reagents were of analytical or HPLC grade, obtained from commercial sources, and used as received from Merck (Darmstadt, Germany) or similar brands.

2.2 Methods

2.2.1 Synthesis

2.2.1.1 Synthesis of L1 and L2 ligands

The ligand L1 was synthesized by the condensation reaction of 2-furoic hydrazide (2-FH) with 2-hydroxybenzaldehyde (2-HBA), as shown in Figure 2. Similarly, to obtain L2, hydrazide 2-FH was replaced by 4-methoxybenzohydrazide (4-MBH), using the same aldehyde as for L1. For both syntheses, equimolar ethanolic solutions of hydrazide and aldehyde were heated and stirred separately. After adding the aldehyde solution dropwise to the hydrazide solution, the mixture was stirred under mild heating for 1 h. It was left to gradually evaporate at room temperature for 3 weeks. The needle-shaped light-yellow crystals were filtered off, washed with cold ethanol, and dried in a desiccator.

L1·H₂O: Yield 72% (0.179 g). Anal. (%) calc. for C₁₂H₁₂N₂O₄ (MW: 248.2): C, 58.07; H, 4.87; N, 11.29. Found: C, 58.24; H, 5.13; N, 11.64. ¹H NMR (400 MHz, DMSO-d _6, δ/ppm): δ 12.11 (s, 1H, H10), 11.14 (s, 1H, H7), 8.64 (s, 1H, H8), 7.97 (d, J = 1.6 Hz, 1H, H15), 7.54 (d, J = 7.7 Hz, 1H, H6), 7.36–7.22 (m, 2H, H3, H13), 6.93 (d, J = 8.0 Hz, 2H, H4, H5), and 6.72 (dd, J = 3.5, 1.8 Hz, 1H, H14). ¹³C NMR (101 MHz, DMSO-d _6, δ/ppm): δ 157.81 (C11), 148.59 (C8), 144.73 (C15), 131.88 (C13), 129.77 (C6), 119.85 (C7), 116.85 (C4, C5), 115.70 (C3), and 112.64 (C14). HRMS (ESI+): m/z calc for C₁₂H₁₁N₂O₃ [M + H]⁺: 231.0764; found [M + H]⁺: 231.0751.

L2·H₂O: Yield 85% (0.122 g). Anal. (%) calc. for C₁₅H₁₆N₂O₄ (MW: 288.3): C, 62.49; H, 5.59; N, 9.72. Found: C, 62.60; H, 5.81; N, 9.78¹H NMR (400 MHz, DMSO-d _6, δ/ppm): δ 12.03 (s, 1H, H10), 11.42 (s, 1H, H7), 8.64 (s, 1H, H8), 7.96 (d, J = 8.5 Hz, 2H, H14, H16), 7.54 (dd, J = 7.7, 1.7 Hz, 1H, H6), 7.37–7.26 (m, 1H, H3), 7.14–7.05 (m, 2H, H13, H17), 7.01–6.88 (m, 2H, H4, H5), and 3.85 (s, 3H, H18). ¹³C NMR (101 MHz, DMSO, δ/ppm): 162.67 (C11), 157.93(C2), 148.33 (C8), 131.68 (C3), 130.06 (C14, C16, C6), 125.23 (C15), 119.77 (C1), 116.87 (C4, C5), 114.28 (C12, C13, C17), and 55.90 (C18). HRMS (ESI+): m/z calc for C₁₅H₁₅N₂O₃ [M + H]⁺: 271.1077; found [M + H]⁺: 271.1064.

2.2.1.2 Synthesis of [Cu(L1-H) (NO₃)]·H₂O and [Cu₂(μ-L2-H)₂(NO₃)₂(CH₃CH₂OH)₂] complexes

The complexes [Cu(L1-H) (NO₃)] and [Cu₂(μ-L2-H)₂(NO₃)₂(CH₃CH₂OH)₂], abbreviated as CuL1·H₂O and Cu₂L₂ respectively, were synthesized following a common protocol, in which an ethanolic solution of copper (II) nitrate (0.25 mmol) was added dropwise to a solution containing 0.25 mmol of ligand L1 or ligand L2 in absolute ethanol under continuous stirring (Figure 3). The reaction was conducted under stirring at a moderate temperature (<50 °C) for 1 h, after which the solvent was allowed to evaporate slowly at room temperature for 6 weeks until dark-green crystals formed. The crystals were collected by filtration, washed with cold ethanol, and dried at room temperature.

CuL1·H₂O: Yield 72% (0.067 g). Anal. (%) calc. for C₁₂H₁₁CuN₃O₇ (MW: 372.78): C, 38.66; H, 2.97; N, 11.27. Found: C, 38.69; H, 3.00; N, 11.33. HRMS (ESI-): m/z calc for C₁₂H₈CuN₃O₆ [Cu(L1-2H) (NO₃)]^-: 352.9715; found [Cu(L1-2H) (NO₃)]^-: 352.9720.

Cu₂L2₂: Yield 85% (0.094 g). Anal. (%) calc. for C₃₄H₃₈Cu₂N₆O₁₄ (MW: 881.78): C, 46.31; H, 4.34; N, 9.53. Found: C, 46.35; H, 4.33; N, 9.55. HRMS (ESI+): m/z calc for C₃₀H₂₆Cu₂N₄O₆ [Cu₂(μ-L2-H)₂]²⁺: 332.0217; found [Cu₂(μ-L2-H)₂]²⁺: 332.0206.

2.2.2 Crystal data and structure refinement

2.2.2.1 X-ray diffraction data

The measurements were performed on a Rigaku Oxford Gemini Eos CCD diffractometer with graphite-monochromated CuKα (λ = 1.54184 Å) radiation. X-ray diffraction intensities were collected (ω scans with ϑ and κ-offsets), integrated, and scaled using the CrysAlisPro suite of programs (Rigaku, 2015). The unit cell parameters were obtained by least-squares refinement (based on the angular setting for all collected reflections with intensities greater than seven times the standard deviation of measurement errors) using CrysAlisPro. Data were corrected empirically for absorption employing the multi-scan method implemented in CrysAlisPro.

L1·H₂O. Most of the diffraction pattern was interpreted in terms of a two-component monoclinic twin. The unit cell parameters for both crystal domains were equal to within experimental accuracy. The domains were rotated from each other in 180° around the crystal a-axis. The reflections were indexed in the reciprocal unit cell of the corresponding domains. It was resorted to the twin crystal data reduction procedure of CrysAlisPro to generate two data sets, a regular one with the diffraction data indexed in the reciprocal unit cell of the largest domain (with approximately 70% of diffracting power) and a second one including all reflections from both domains with the overlapping ones flagged for structure development and refinement. The first dataset was employed to solve the structure using the dual procedure of SHELXT (Sheldrick, 2015) and its non-H atoms were refined with isotropic displacement parameters by a full-matrix least-squares procedure using SHELXL (Sheldrick, 2008).

Despite a correct molecular model, however, the refinement evidenced the presence of overlap between reflections from both crystal domains. The evidence included (besides the visual rendering with CrysAlisPro of the weighted reciprocal space) many observed violations of systematic extinctions (Ferraresi-Curotto et al., 2015), a relatively large agreement R1-factor (0.21), and the list of the most disagreeable reflections showing F (obs) values systematically larger than the corresponding F (calc) values. We, therefore, refined the initial molecular model against the second dataset, which included all collected reflections for both crystal domains, employing the untwining procedure of SHELXL. Now that there were no systematic extinction violations, the R1-factor dropped to 0.11, and the sign of F (obs)-F (calc) difference for the most disagreeable reflections was more evenly distributed. Following anisotropic refinement, a difference Fourier map showed all H-atoms except the one of a heterocycle ring. All but the water H-atoms were positioned at their expected locations and refined with the riding model. The hydroxyl H-atoms were refined by treating the O-H group as a rigid body allowed to rotate around the >C (ph)-OH bond to maximize the residual electron density at their calculated positions. A difference Fourier map phased on all determined atoms showed the water H-atoms among the first four peaks. These were refined at their found positions with isotropic displacement parameters. The diffracting weight of the minor crystal domain was 0.285 (1).

CuL1·H₂O. The ligand hydrogen atoms were positioned at their expected locations and refined with the riding model. The water H-atoms were in a difference Fourier map and refined at their found positions with isotropic displacement parameters and Ow-H and H … H distances restrained to target values of 0.86 (1) and 1.36 (1) Å, respectively.

L2·H₂O. The H-atoms were in a different Fourier map phased on the heavier atoms and refined at their found positions with isotropic displacement parameters.

Cu₂L2₂. The hydrogen atoms, except the one of the alcohol groups, were positioned at their expected locations and refined with the riding model. The methyl angular conformation was optimized by treating -CH₃ as a rigid group allowed to rotate around the O-CH₃ bond, such as maximizing the sum of the residual electron density at the calculated hydrogen positions. The alcohol H-atom of the EtOH ligand was in a different Fourier map and refined with an isotropic displacement parameter at its found position, with O-H and H₂C(eth)…H distances restrained to target values of 0.86 (1) and 1.83 (1) Å, respectively.

Crystal data and structure refinement results are summarized in Tables 1, 2. Crystallographic structural data have been deposited at the Cambridge Crystallographic Data Centre (CCDC). Any request to the Cambridge Crystallographic Data Centre for this material should quote the full literature citation and the reference numbers CCDC 2306862 (L1.H₂O), 2392963 (CuL1.H₂O), 2392965 (L2.H₂O), and 2392966 (Cu₂L2₂).

TABLE 1. Crystal data and structure refinement results for the L1 ligand and its Cu(II) complex.

R₁ = Σ||F_o|-|F_c||/Σ|F_o|, wR₂ = [Σw (|F_o|2-|Fc|2)2/Σw (|F_o|2)2]1/2.

TABLE 2. Crystal data and structure refinement results for the L2 ligand and its Cu(II) complex.

R₁ = Σ||F_o|-|Fc||/Σ|F_o|, wR₂ = [Σw (|Fo|2-|Fc|2)2/Σw (|Fo|2)2]1/2.

2.2.2.2 Physicochemical characterization

Elemental analyses for C, N, and H were performed using a Thermo Scientific Flash 2000 analyzer equipped with an Eager 200 data processor. The FT-IR spectra (4,000–400 cm^–1) were recorded using KBr pellets on a Bruker Invenio-R. FT-IR spectrometer. The electronic spectra were recorded in a Shimadzu UV2600 Spectrophotometer. The 1D (¹H and ¹³C) and 2D (¹H–¹H COSY, ¹H–¹³C HSQC, and ¹H–¹³C HMBC) NMR spectra were recorded in DMSO-d6 at 25 °C on a Bruker Avance III 400 MHz spectrometer. Chemical shifts are reported in ppm using the solvent residual peak as an internal standard for ¹H NMR (δH 2.50 ppm) and ¹³C NMR (δC 39.52 ppm). HRMS was obtained in a Q Exactive Plus mass spectrometer (Thermo Scientific, United States) by direct injection using methanol as solvent and an Ion Max API source with a HESI-II probe. The mass spectrometer was operated in negative and positive modes; the ion spray voltage was set at 4.5 kV, and the capillary temperature was set at 250 °C.

2.2.3 Cell culture conditions

The human OS (MG-63) cell line was cultured in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 IU/mL penicillin, and 100 μg/mL streptomycin at 37 °C in 5% CO₂ atmosphere.

2.2.4 2D studies

2.2.4.1 Cell viability

A cytotoxic study was performed according to Mosmann (1983). Cells were seeded in a 96-well plate at a density of 3.5 × 10⁴ cells/mL, allowed to attach for 24 h, and then treated with different concentrations of ligands (10–100 µM) and complexes (0.5–3 µM) at 37 °C for 24 h. Afterward, the medium was replaced, and the cells were incubated with 0.5 mg/mL MTT under normal culture conditions for 3 h. Cell viability was assessed by the conversion of the tetrazolium salt MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazoliumbromide) to a colored formazan by mitochondrial dehydrogenases. Color development was measured spectrophotometrically with a microplate reader (Multiplate Reader Multiskan FC, Thermo Scientific) at 570 nm after cell lysis in DMSO (100 µL per well). Cell viability was plotted as the percentage of the control value.

2.2.4.2 Reactive oxygen species study

Oxidative stress in MG-63 cells was evaluated by measuring the intracellular production of ROS after treating the cell monolayer with different concentrations of Cu(II) complexes. In brief, the cells were grown in 24-well plates (10⁵ cells/well), incubated for 30 min in the dark with dihydrorhodamine-123 (DHR-123), and then treated for 3 h with increasing concentrations of the Cu(II) complexes at 37 °C in darkness. ROS generation was determined by the oxidation of DHR-123 to rhodamine by spectrofluorescence at 530 nm. Protein content was measured using the Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, Massachusetts, United States), and the results were adjusted accordingly.

2.2.5 3D studies

2.2.5.1 Multicellular tumor spheroid formation

MG-63 cells were used to create multicellular tumor spheroids using the modified hanging drop technique that we had previously presented (León et al., 2016; Ruiz et al., 2019). In brief, on the cell culture plate lid, 25 µL drops of cell suspension containing 400 cells each were suspended. Following the 72-h period needed for cell aggregation, the spheroids were transferred to 96-well plates coated with agarose (one droplet per well) and cultivated using 150 µL of complete medium. Spheroids were allowed to grow for another 48 h until reaching a diameter of 400 µm.

2.2.5.2 Spheroid cell viability studies

Multi-cellular spheroids (MCS) were treated in 96-well plates with 0.5% DMSO in DMEM (control) and with Cu(II) complexes in a range of concentrations from 5 to 50 µM in DMEM for 24 h. Afterward, cell viability was evaluated using the resazurin reduction assay, in which the resazurin dye is irreversibly reduced by intracellular oxidoreductases to a pink-red fluorescent dye known as resorufin (Anoopkumar-Dukie et al., 2005). Fluorescence was measured using a Shimadzu RF-6000 fluorometer (excitation at 560 nm, emission at 590 nm). Cell viability was plotted as a percentage of the basal condition (solvent control).

2.2.5.3 MCS spreading assay

To evaluate the migratory and proliferative capacity of spheroid-derived cells following 24 h exposure to CuL1 and Cu₂L2₂, spheroids were transferred to 48-well plates containing 500 µL of DMEM either alone (control) or supplemented with different concentrations of the complexes (1–10 µM) and incubated under standard culture conditions. Upon attachment of the spheroids to the plastic surface, cells initiated radial migration, progressively expanding the adhesion area. After 24 h of treatment, cells were fixed with methanol, stained with 20% Giemsa solution, and cell migration was assessed using an inverted microscope and ImageJ v1.53t software.

2.2.6 Statistical analysis

Three independent experiments were performed, and the results are presented as the mean ± standard error of the mean (SEM), unless otherwise stated. Data normality was assessed using both the Shapiro–Wilk and D'Agostino–Pearson normality tests. Data that satisfied the normality assumption were analyzed using Student’s t-test (for comparisons between two groups) or one-way analysis of variance (ANOVA) followed by Tukey’s multiple comparisons test (for comparisons among three or more groups), as appropriate. Differences were considered statistically significant at p < 0.05. All statistical analyses were performed using GraphPad Prism version 8.0.1 (GraphPad Software Inc., San Diego, CA, United States).

3 Results

3.1 Crystallography

Crystal data and structure refinement results for L1·H₂O, L2·H₂O, CuL1·H₂O, and CuL2 were described below.

L1·H₂O. There are two closely related but independent molecules per crystal asymmetric unit (Z’ = 2, Z = 8), as shown in the ORTEP (Farrugia, 2012) drawing of Figure 4. Because of extended π-bond delocalization, both L1·H₂O molecules are nearly planar (rms deviation of non-H atoms from the best least-squares plane less than 0.13 Å). Planarity is further favored by an intra-molecular OH … N bond [O … N bond distances of 2.671 (3) and 2.693 (3) Å and corresponding O-H … N bond angles of 146 and 144°]. Observed bond distances and angles conform to stablished organic chemistry knowledge and are listed in Supplementary Table S1. They are in close agreement with the corresponding data of the solid-state anhydrous organics reported by Bikas et al. (2010). The lattice is further stabilized by a complex network of medium-strength H-bonds. Neighboring organic molecules are bridged by crystallization water molecules acting as donor in intermolecular > C=O … HO_wH … O=C< bonds [O_w … O_carb distances of 2.873 (3) and 2.807 (3) Å and corresponding O_w-H … O_carb angles of 179 (5) and 166 (5)° for molecule #1 and Ow … O_carb distances of 2.827 (3) and 2.755 (3) Å and O_w-H … O_carb angles of 167 (4) and 170 (4)° for #2] and as acceptor in >N-H … O_w bonds [N … O_w distances of 2.808 (3) and 2.864 (3) Å and N-H … Ow angles of 164 and 166°]. The H-bonding structure is further detailed in Supplementary Table S2.

L2·H₂O. A view of the crystal asymmetric unit content is shown in Figure 5 and its metrics are detailed in Supplementary Table S3. It is a monohydrate conformer of the molecule reported by Tang (2008), differing mainly by a 180° rotation of the methoxy -OCH₃ group. In turn, L2 differs from L1 by the replacement of the furan cycle with the bulkier methoxybenzene group. Here, the above-mentioned tendency toward planarity due to π-bond delocalization is hampered by steric hindrance, and the methoxybenzene plane subtends a dihedral angle of 37.0 (1)° with the rest of the backbone skeleton. As for L1, the planarity of this molecular fragment is further favored by an intra-molecular OH … N bond [d (O … N) = 2.648 (6) Å and ∠(O-H … N) = 145 (10)°]. Neighboring molecules are bridged by the crystallization water molecule, acting as a donor in O_wH … O (carb) bonds [O_wH … O (carb) distances of 2.843 (5) and 2.843 (5) Å, and corresponding O_w-H … O (carb) angles of 175 (8) and 157 (6)°] and as acceptor in NH … O_w interactions [d (N … O_w) = 2.832 (5) Å and ∠(N-H … O_w) = 168 (6)°]. H-bonds are further detailed in Supplementary Table S4.

CuL1·H₂O. The copper complex is shown in Figure 6, and bond distances and angles are in Supplementary Table S5. The Cu(II) ion is in a distorted square environment, coordinated to a L1^- anion acting as a tridentate ligand through his hydroxyl oxygen atom [d (Cu-O_hyd) = 1.884 (3) Å], the carbonyl oxygen [d (Cu-O_carb) = 1.971 (3) Å], and the N-atom [d (Cu-N) = 1.929 (3) Å] and a nitrate oxygen atom [d (Cu-O) = 2.002 (3) Å], all along electron lone-pair orbital lobes of the ligand atoms. The copper ion is capped by a slanted and much weaker Cu … O interaction with another nitrate oxygen atom [d (Cu … O) = 2.466 (3) Å]. Trans L1-Cu-L1 bond angles in the coordination plane are ∠(O_carb-Cu-O_hyd) = 173.9 (1)° and ∠(N-Cu-O_nitr) = 171.2 (1)° and cis L1-Cu-L1 angles are in the 81.9(1)–93.1(1)° range. The crystallographic metrics around the metal indicate the copper d (x²-y²) orbital as the ground-state orbital for the unpaired electron (or hole).

As expected, the major bond changes of the ligand L1^- upon hydroxyl deprotonation of L1 and coordination to the metal occur at the hydroxyl and carbonyl groups. >C (ph)-O bond length shortens from 1.353 (4) Å in unbonded L1 to 1.318 (4) Å in coordinated-to-metal L1^- (−0.035 Å = 9 times the standard deviation error σ) and to a much less degree > C=O bond distance lengthens from 1.240 (3) to 1.251 (4) Å (+0.011 Å = 3σ).

The crystallization water molecule links neighboring complexes in the lattice, as H-acceptor in a NH … O_w bond [d (H … O_w) = 1.85 (4) Å and ∠(N-H … O_w) = 168 (4)°] and as H-donor in O_wH … O_hyd and O_wH … O_nitr interactions [d (H … O_hyd) = 1.93 (2) Å, ∠(O_w-H … O_hyd) = 176 (7)° and d (H … O_nitr) = 2.07 (2) Å, ∠(O_w-H … O_nitr) = 164 (5)°, respectively]. H-bonds are further detailed in Supplementary Table S6.

Cu₂L2₂. The substance crystallizes as a centrosymmetric binuclear Cu(II) complex (Figure 7), and bond distances and angles are in Supplementary Table S7. The inversion-related Cu(II) ions are in a distorted square bipyramidal environment; each equatorially coordinated to a L2^- anion acting as a tridentate ligand through his hydroxyl oxygen atom [d (Cu-O_hyd) = 1.960 (1) Å], the carbonyl oxygen [d (Cu-O_carb) = 1.999 (1) Å], and the N-atom [d (Cu-N) = 1.938 (2) Å]. The fourth ligand is the hydroxyl oxygen atom of a neighboring L2^- ligand which bridges de metals in the binuclear complex [d (Cu-O_hyd’) = 1.970 (1) Å, ∠(Cu-O_hyd’-Cu’) = 99.74 (6)°, d (Cu … Cu’) = 3.0052 (5) Å], all nearly along electron lone-pair orbital lobes of the ligand atoms. The six-fold coordination around copper is completed by a nitrate oxygen atom at one pyramid axis [d (Cu-O_nitr) = 2.637 (3) Å] and a solvent ethanol molecule occupies the opposite apical site at 2.371 (2) Å from the metal, hence completing a strongly elongated square bipyramidal coordination around the copper ions. Trans L-Cu-L bond angles in the coordination plane are ∠(O_carb-Cu-O_hyd) = 171.08 (6)° and ∠(N-Cu-O_hydr’) = 171.75 (7)° and cis L-Cu-L angles are in the 80.26(6)-107.71(6)° range. The Trans L-Cu-L bond angle involving the axial ligands is 164.12 (7)°. Full bond distances and angles are provided in Supplementary Table S7. The crystallographic metrics indicate a copper d (x ²-y ²) ground state and, according to Anderson–Goodenough–Kanamori rules, the copper binding structure favors an AF exchange coupling between the unpaired electrons of neighboring copper (II) ions (Anderson, 1950; Goodenough, 1955; Goodenough, 1958; Kanamori, 1959).

Neighboring binuclear units in Cu₂L2₂ are linked through a single NH … O_nitr bond [d (N … O_nitr) = 2.769 (2) Å and ∠(N-H … O_nitr) = 165°]. Further details of this bond along with other much weaker inter-dimeric EtOH … O_nitr and CH … O interactions are provided in Supplementary Table S8.

3.2 Characterization of L1, L2, and complexes CuL1 and Cu₂L2₂

The synthesized ligands and complexes were characterized by elemental analysis, ¹H and ¹³C NMR, FT-IR, electronic spectra, and high-resolution mass spectrometry (HRMS). Elemental analysis of carbon, hydrogen, and nitrogen showed percentage values in agreement with those calculated for the proposed molecular formulas (see Materials and Methods section). FT-IR spectra of the solid ligands and their corresponding complexes were recorded in the 4000–400 cm^-1 frequency range (Supplementary Figures S1 and S2). The vibrational assignments were based on the reported data (Balsa et al., 2021; Rodríguez et al., 2021; Santa Maria de la Parra et al., 2024b) and are summarized in Table 3. The compounds L1, L2, and CuL1 crystallize with one lattice water molecule, as we previously determined by X-ray diffraction. The bands assigned to the antisymmetric and symmetric stretching vibrations of hydrogen-bonded water molecules were observed between 3,550 and 3,300 cm^-1 in the three spectra. The vibrational modes associated with the NH group remain largely unaffected in the complexes’ spectra compared to those observed for the free ligands, indicating retention of this proton upon coordination, which is also consistent with the results obtained from X-ray diffraction. The bands observed at 3,001, 1,369, and 848 cm^-1 in the spectrum of L1 and 2999, 1375, and 849 cm^-1 for L2, assigned to the OH group in the ligands, are absent in the spectra of the corresponding complexes. This finding confirms the deprotonation of the phenolic oxygen, leading to the formation of a phenoxide group coordinated to the Cu(II) center. The bands observed at 528 and 556 cm^-1 in the spectra of CuL1 and Cu₂L2₂, respectively, indicate coordination of the metal center to this oxygen atom. Furthermore, the bands assigned to νC = O of the amide moiety (1,642 and 1,623 cm^-1) and imine group (1,609 and 1,605 cm^-1) in the L1 and L2 spectra, respectively, are overlapped and shifted to lower frequencies in the complex spectra, thus confirming the coordination to the metal center through these groups. These findings indicate tridentate coordination of ligands L1 and L2 to the metal center through the imine N-atom, the O-atom from the carbonyl group, and the O-atom from the phenolic group.

TABLE 3. Assignment of the most significant vibrational modes for ligand L1 and L2 and complexes CuL1 and Cu₂L2₂ acquired experimentally.

References: ph, phenyl ring; νas, antisymmetric stretching mode; νs, symmetric stretching mode; δ, bending mode; γ, out-of-plane bending mode. Intensities of the experimental bands vs, very strong; s, strong; m, medium; m-s, medium–strong; m-w, medium–weak; w, weak; b, broad; sh, shoulder.

High-resolution mass spectrometry (HRMS) further confirmed the composition of the compounds in solution. For the free ligands, the spectra recorded in positive ionization mode cleanly displayed the characteristic protonated molecular ions [L + H]⁺ (Supplementary Figures S3, 4). Conversely, the mononuclear copper complex CuL1 exhibited significantly better ionization efficiency and resolution in negative mode, where it was detected as a deprotonated species (Supplementary Figure S5). In the case of the dimeric complex Cu₂L2₂, ESI + -HRMS analysis revealed the predominant presence of the dicationic core [Cu₂(µ-L2-H)₂]²⁺ (Supplementary Figure S4). These results indicate that nitrate anions and coordinated ethanol molecules, although explicitly observed in the solid-state crystal structure, readily dissociate in methanol solution during electrospray ionization.

The ¹H and ¹³C NMR spectra of the free ligands (L1 and L2) were recorded in DMSO-d₆ solution, and two-dimensional NMR experiments (¹H–¹H COSY, ¹H–¹³C HSQC, and ¹H–¹³C HMBC) were employed to achieve complete signal assignment. The ligand atom numbering scheme used for the ¹H and ¹³C NMR discussion is depicted in Supplementary Figure S7. The chemical shifts, signal multiplicities, and integrations are consistent with the proposed structures of the ligands, confirming the integrity of the hydrazone framework (Supplementary Figures S8-S17). The complete assignment including chemical shifts (δ), multiplicities, and coupling constants (J) for all studied compounds are presented after the description of the synthesis. ¹H NMR signal integration and signal multiplicities agree with the proposed molecular formula. Both free ligands displayed similar features to the structurally related compounds previously reported by us (Santa Maria de la Parra et al., 2026). Signals corresponding to the aromatic protons were clearly detected in the 7.0–8.7 ppm region, and the key singlet signal corresponding to H8 was identified in both molecules. Furthermore, heteroatom-labile protons (H7 and H10) appeared upfield without any coupling with another proton or carbon. The signal corresponding to the methyl group in L2 was identified at 3.85 ppm. The ¹³C NMR signals were assigned using the bidimensional spectra. All ¹³C were identified with exception of quaternary carbon in L1 since they do not appear in the spectra (Supplementary Figure S9). Overall, signal integrals and multiplicities matched the stoichiometry derived from the synthetic procedure and elemental analysis, further supporting structural consistency.

Electronic absorption spectra in the UV−VIS region of the CuL1 and Cu₂L2₂ complexes, as well as their corresponding ligands, were recorded between 280 and 900 nm both in the solid state and DMSO solution at different concentrations, to identify transitions with different molar absorptivities.

Considering the square pyramid geometry of the coordination environment observed for CuL1 by crystallography, absorption bands assignable to d–d electronic transitions are expected to be observed in the visible region of the spectrum. As shown in Figure 8a, this band was observed at 694 nm, consistent with the green color of the solution and characteristic of d–d electronic transitions of Cu(II). The band centered at 392 nm, with a shoulder at 408 nm, is attributed to ligand-to-metal charge transfer transitions (LMCTs) between σO and dCu orbitals (Burgos-Lopez et al., 2019). The set of bands observed at 303, 317, and 332 nm, assigned to intraligand transitions (ILCTs), appears at longer wavelengths than those corresponding to L1 (291, 302, and 331), which is consistent with the decrease in their energy upon complex formation. The band assignments were based on results previously reported for similar compounds (Rodríguez et al., 2021; Santa Maria de la Parra et al., 2024b).

Furthermore, comparison with the spectrum obtained for the solid sample reveals the same bands as those observed in solution, with only minor differences attributable to the distinct measurement techniques (Figure 8b). The experimental electronic spectra, together with the proposed assignments, are detailed in Table 4.

TABLE 4. Experimental electronic spectra of the complexes CuL1 and Cu₂L2₂ in DMSO solution and the solid state. Band maxima are expressed in nm. Molar absorptivity (in M^-1cm^-1) is indicated in parentheses. The proposed assignment is also included.

Similarly, the spectra obtained for Cu₂L2₂ and L2 in the solid state and in DMSO solution were analyzed (see Figure 9) and compared with the spectra obtained for similar compounds (Balsa et al., 2021; Ferraresi-Curotto et al., 2015). The band corresponding to the d–d transition of the Cu(II) ion appears at 708 nm in the spectrum obtained for the 2.5 × 10⁻⁵ M solution of Cu₂L2₂ and at 703 nm in the spectrum of the solid sample. The band appearing with a maximum at 392 nm in the solution spectrum and at 371 nm in the solid sample was assigned to ligand-to-metal charge transfer. Finally, the set of bands between 300 and 335 nm in both spectra of the complex is attributed to intraligand processes, consistent with those observed in the spectrum of L2.

Furthermore, stability studies performed over 24 h (Supplementary Figures S18, 19) demonstrated that both CuL1 and Cu₂L₂ remained stable in DMSO. In PBS, CuL1 also remained stable, whereas Cu₂L₂ exhibited a gradual decrease in signal intensity over time. This decrease is attributed to its lower solubility in aqueous media at this concentration, rather than to chemical degradation as no shift in the characteristic absorption bands was observed throughout the experiment. Although Cu₂L₂ displays lower aqueous solubility, the solutions used for the biological assays were freshly prepared immediately before use, thereby minimizing any potential impact of this limitation. Moreover, because cell culture medium is substantially more complex than PBS and cannot be accurately reproduced by this simplified system, both complexes were considered sufficiently stable for subsequent in vitro biological studies.

3.3 2D studies

3.3.1 Cell viability

Cytotoxicity studies for ligands (L1 and L2) and their copper (II) complexes (CuL1 and Cu₂L2₂) on the human OS MG-63 cell line were performed using the MTT assay. Supplementary Figure S20 shows that both complexes exhibited anticancer activity against MG-63 cells at low-micromolar concentrations from 0.5 to 3 µM (p < 0.0001) in a dose-dependent manner with IC₅₀ values of 2.3 ± 0.1 and 0.9 ± 0.1 µM for CuL1 and Cu₂L2₂, respectively (Table 5). Moreover, both complexes exhibited stronger anticancer activity than CDDP (IC_50,CDDP = 39 ± 1.8 µM). However, both ligands and Cu(NO₃)₂ did not affect the cell viability until 100 μM, highlighting the significance of metal coordination to enhance the anticancer activity of the inactive ligands, as we previously reported (Espindola-Moreno et al., 2025; Rodríguez et al., 2021; Santa Maria de la Parra et al., 2026a; Santa Maria de la Parra et al., 2026b; Santa Maria de la Parra et al., 2024b; Santa Maria de la Parra et al., 2026).

TABLE 5. IC₅₀ (µM) values of L1, L2, CuL1, Cu₂L2₂, and structurally related Cu(II) complexes in MG-63 and L929 cells (2D) after 24 h of incubation. n.d, not determined.

Several scientific reports indicate that copper complexes with IC₅₀ values lower than 10 µM are efficient cytotoxic agents against several human cancer cell lines (Denoyer et al., 2018; Santini et al., 2014). As discussed above, our group reported a series of Cu(II)-hydrazone complexes with potent cytotoxic activity against MG-63 cells (see Figure 1; Table 5). The complexes CuL3 and CuL7 are structurally related to CuL1 and Cu₂L2₂, respectively, via the hydrazide moiety of the ligand (furan and methoxy-benzohydrazide), as shown in Figure 1. However, the structural difference in the aldehyde group of the ligand (2-hydroxy-3-methoxybenzaldehyde > 2-hydroxy-benzaldehyde) is reflected in the lower cytotoxic potency of CuL1 compared to CuL3. In contrast, Cu₂L2₂ exhibits enhanced anticancer activity relative to CuL7, as evidenced by an almost 3-fold reduction in the IC₅₀ value. This difference may be attributed to the fact that Cu₂L2₂ is a dimer, whereas CuL7 is a monomeric complex; notably, increased nuclearity has previously been associated with enhanced cytotoxic activity, as earlier demonstrated for Cu₄(L6)₄ (Santa Maria de la Parra et al., 2026a).

We also evaluated the effect of ligands and complexes on the viability of non-tumor cells, such as the L929 mouse fibroblast line (see Table 5; Supplementary Figure S21). The selectivity index (SI = IC₅₀ non-tumoral cells/IC₅₀ tumor cells) of both complexes exceeded that of CDDP (SI = 0.3), with CuL1 exhibiting the greatest selectivity. Given these promising results, we continued the study of the cytotoxic activity of both complexes.

3.3.2 Oxidative stress

Numerous studies have demonstrated a correlation between redox potential and the anticancer activity of copper compounds, which is largely attributed to the accumulation of reactive oxygen species (ROS) generated through Fenton-like reactions induced by these complexes (Garcia‐Bosch and Siegler, 2016; Jungwirth et al., 2011). In this context, the DHR-123 probe, which selectively reacts with hydrogen peroxide and peroxynitrite, was employed to evaluate the potential effect of the complexes on ROS formation (Kooy et al., 1994; Qin et al., 2008). Following cellular uptake, intracellular ROS oxidize DHR-123 to fluorescent rhodamine 123. The incubation of MG-63 cells with CuL1 for 3 h caused a marked increase in intracellular ROS levels from 340% to 450% over basal in the range of 5–20 μM, as shown in Figure 10. Similarly, incubation with Cu₂L2₂ triggered an increase in ROS production from 1 µM (215%) to 20 µM (515%) over basal (p < 0.0001).

Our results demonstrate that both complexes are internalized by MG-63 cells and induce ROS production within as little as 3 h, indicating that ROS generation may be a key component of their mechanism of action. Evidence indicates that cancer cells experience oxidative stress that supports proliferation and other processes essential for tumor development (Schumacker, 2006; Valko et al., 2006). Consequently, basal ROS levels are typically higher in cancer cells than in normal cells, rendering them more susceptible to chemotherapeutic and redox-active agents—such as copper (II) complexes—that further enhance ROS production.

3.4 3D studies

Although 2D cultures are the first step in assessing a compound’s cytotoxicity, 3D cultures offer a more accurate representation of the intratumoral environment, including gradients of nutrients and oxygen, as well as cell–cell interactions (Kondo and Inoue, 2019). Multicellular resistance (MCR), a hallmark of solid tumors, can also be reproduced using 3D models, which reduce cellular sensitivity to anticancer therapies and result from cell–cell interaction, limited drug penetration, and the resistance of quiescent cells found in the MCS core (Mitrakas et al., 2023). Due to its impact on apoptotic factor expression, hypoxia has also been implicated as a cause of MCR (Erler et al., 2004).

3.4.1 Spheroid cell viability studies

To further investigate the antitumor activity of CuL1 and Cu₂L2₂, we generated MCS from MG-63 cells by the hanging drop technique (León et al., 2016). Our results indicated that complexes CuL1 and Cu₂L2₂ impaired the cell viability of spheroids derived from MG-63 cells showing IC₅₀ of 25.0 ± 0.4 and 14.7 ± 0.9 µM, respectively, corresponding to cytotoxic activities approximately 2.5-fold and 4.5-fold higher than CDDP (IC₅₀ = 65.1 ± 5.6 µM). As shown in Table 6, comparison of the potency of the CuL1 and Cu₂L2₂ complexes with that of the CuL3 and CuL7 complexes indicates that the removal of the methoxy group from the benzaldehyde moiety diminished the cytotoxicity of the complexes. In addition, the complexes exerted dose-dependent effects on the spheroids’ volume and shape, as shown in Figure 11.

TABLE 6. IC₅₀ (µM) values of CuL1, Cu₂L2₂, and structurally related Cu(II) complexes in MG-63 cells (2D) and spheroids derived from MG-63 cells (3D) after 24 h of incubation. n.d, not determined.

3.4.2 MCS spreading assay

Metastasis represents the leading cause of cancer-related mortality and involves the dissemination of malignant cells from the primary tumor to adjacent tissues and distant organs (Fares et al., 2020). This multistep process is driven by several factors, including cellular migration through blood and lymphatic vessels, invasive growth across the extracellular matrix (ECM), and the establishment of secondary tumor colonies at distant sites (Kramer et al., 2013). In this context, spheroid spreading assays provide a valuable in vitro model for mimicking the migration of cells from small tumor aggregates and investigating processes associated with tumor dissemination. To investigate the effect of Cu(II) complexes on the migration of spheroid cells, MCS derived from MG-63 cells were treated with CuL1 at concentrations ranging from 2.5 to 10 µM and Cu₂L2₂ at concentrations ranging from 1 to 5 µM (see Supplementary Figure S22). Using the migration area of untreated spheroids as 100% cell migration, treatment of the spheroids with the CuL1 complex reduced cell migration to 60% at 5 µM and further to 20% at 10 µM (see Figure 12). A similar reduction in cell migration was observed following treatment with the Cu₂L2₂ complex at concentrations of 2.5 and 5 µM. Importantly, this effect was observed even at sub-IC₅₀ concentrations, a phenomenon we reported previously for CuL3 and Cu₄(L6)₄ complexes on MG-63-derived MCS (Santa Maria de la Parra et al., 2026a; Santa Maria de la Parra et al., 2024b). These findings further support the inhibitory effect of the CuL1 and Cu₂L2₂ complexes on cell migration from OS spheroids.

4 Discussion

In this study, two new Cu(II)-hydrazone complexes were successfully synthesized, structurally characterized, and evaluated for their antitumor activity against osteosarcoma using complementary 2D and 3D cellular models. Metal coordination markedly enhanced the biological activity of the hydrazone ligands, yielding compounds with potent antiproliferative effects that clearly outperformed cisplatin under the experimental conditions evaluated. Both complexes also promoted intracellular ROS generation and significantly impaired tumor cell migration, highlighting their ability to affect multiple processes associated with tumor progression.

Among the investigated compounds, the binuclear complex Cu₂L2₂ consistently exhibited superior biological performance, demonstrating greater cytotoxic potency than its mononuclear analog in both monolayer cultures and multicellular spheroids. The preservation of its activity in the more physiologically relevant 3D model further supports its therapeutic potential.

Overall, these findings reinforce the value of hydrazone-based Cu(II) complexes as promising candidates for osteosarcoma treatment and emphasize the importance of rational coordination design in the development of copper-based metallodrugs. Further mechanistic investigations and in vivo studies will be essential to establish their pharmacological profile and assess their potential for preclinical development.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem.2026.1897476/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Jasmina Dimitrić Marković, Univerzitet U Beogradu, Serbia

Reviewed by: Elif Avcu Altıparmak, Istanbul University-Cerrahpasa, Türkiye

Tianping Chen, Anhui Provincial Children’s Hospital, China