The H₄L ligand was prepared according to the literature methods (Lin et al., 2008). All commercially available chemicals of analytical grade were used directly. The FT-IR spectra as the KBr disk were recorded on a Nicolet-IS10 spectrometer. The elemental analysis of CHNO was conducted by using a EA1110 CHNO-S micro analyzer. Luminescence properties were analyzed by using a Hitachi F-7000 fluorescence spectrophotometer. Powder X-ray diffraction (PXRD) measurements were carried out by using a Rigaku D/MAX 2200 diffractometer. Scanning electron microscopy (SEM) was performed using a Hitachi S-4800. UV–Vis spectroscopy was conducted using a Shimadzu UV-3600 spectrophotometer. Single crystal X-ray diffraction was carried out using a Bruker SMART APEX II DUO diffractometer. The cellular images were recorded by using a Bruker In Vivo Imaging System Fx Pro.

0.1 mmol (0.0268 g) H₄L was dissolved in 6 ml distilled water. The solution was adjusted to pH = 6 using 2% KOH. Then, 0.4 mmol (0.0451 g) Gd(NO₃)₃·6H₂O, 0.1 mmol (0.0242 g) Cu(NO₃)₂·3H₂O, and 1 ml of EtOH were added to it. Then, the solution was heated for 48 h at 130 °C in a stainless steel reactor lined with Teflon. After cooling to room temperature, blue block-shape crystals of [Gd₂Cu(L)₂(H₂O)₁₀]·6H₂O were obtained by filtration and washed with EtOH. Yield: 59% based on Cu²⁺. Anal. calculated for C₁₄H₄₀CuGd₂N₁₆O₂₄: C 14.08; H 3.38; N 18.76% found: C 14.32; H 3.36; N 18.48%. IR (KBr, cm⁻¹) 3412.95 (s), 2367.31 (w), 1566.55 (m), 1411.36 (m), 1295.05 (s), 1237.85 (w), 1142.94 (w), 1027.44 (w), 922.03 (w), 660.89 (w), 624.05 (w).

The structure of the complex synthesized in the experiment was studied by using X-ray single-crystal diffraction. Crystallographic data for the complex were recorded on a Bruker Apex II DUO diffractometer with a Mo–Kα source (λ = 0.71073 Å) at room temperature. The relevant data were collected and full matrix least squares correction was performed with the direct method (SHELXTL-2014) to analyze the crystal structure (Sheldrick, 2008). Table 1 shows the crystallization parameter data of the target complex, and main bond lengths and angles and relevant hydrogen bond parameter data are shown in Table S1 and Table S2 (ESI†).

Due to the poor solubility of the complex, the NPs with good dispersibility in water were prepared by nanocoprecipitation with PEG_-2000 by a “bottom–up” method, and this can be universally found in the literature (Li et al., 2017; Yang J et al., 2018). For a typical experiment, tetrahydrofuran (THF, 1 ml) solution containing the complex (2 mg) and PEG_-2000 (5 mg) was sonicated into distilled water. After stirring for 10 min, N₂ was bubbled for 20 min to remove THF. The solution was then stored in the dark for further use (Yang J et al., 2018). Finally, NPs of the complex in the solution were obtained by centrifugation. The as-prepared NPs have good dispersibility in water. In addition, nanoparticles have enhanced permeability and retention (EPR) effect and can passively target the tumor to enhance uptake.

HeLa, HepG2, and HT29 cell lines were obtained from the Cell Bank of SIBCB, CAS (China). These cells were cultured in the minimum essential medium (DMEM, Gibco; Thermo Fisher Scientific) with 10% fetal bovine serum (FBS; Gibco; Thermo Fisher Scientific). All the culture media contain 100 units/mL penicillin and 100 μg/ml streptomycin. The cells were cultured at 37°C in a humidified incubator with 5% CO₂.

HeLa, HepG2, and HT29 cell lines were seeded in 96-well plates at a density of 1 × 10⁵ cells/ml, 5 × 10⁴ cells/ml, and 5 × 10⁴ cells/ml, respectively, for 12 h to get attached. Cell viability assays of the NPs were conducted by first dissolving in distilled water, which were diluted with DMEM to various concentrations and transferred in the 96-well plate with the same volume (200 µL) in each well for 24 h. Then, the plate was irradiated with a xenon lamp (30 mW/cm²) for 5 min. Cell viability was determined by the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assay. MTT in PBS (5 mg/ml, 20 μL) was added to each well and incubated for 3 h under the same conditions at 37°C. Then, the medium was removed and 200 ml DMSO was added. The plate was agitated on a Bio-Tek microplate reader at ambient temperature. The average absorbance of the no cell line was subtracted from the readings of the other wells. The cell viability was calculated by the following equation: cell viability (%) = mean absorbance in each group incubated with different concentrations of NPs/mean absorbance in the control group. The average absorbance of the blank well (no cells) was removed from the data of the other wells.

To investigate the •OH generation, the solution of the complex NPs (0.5 ml, 10 μg/ml) was added into methylene blue (MB, a dye that can be faded by •OH) solution (0.5 ml, 20 μg/ml) to obtain a mixed solution and allowed to stand at room temperature for 30 min. The absorbance value of the MB and mixed solution of MB with NPs was recorded at about 665 nm via UV–Vis spectroscopy.

In weak acidic conditions, hydrothermal reactions of H₄L and metal salt with a ratio of 1:5 in MeOH-H₂O at 130°C for 2 days provided the complex in moderate of 59%. Crystals of the complex are all air-stable. The elemental analysis reveals that the component is in well-agreement with the results of the X-ray diffraction analysis.

The main IR peaks of the complex are as follows: the absorption peak at 3412.9 cm⁻¹ is attributed to water molecules, the peak at 2367.3 cm⁻¹ belongs to the carboxyl peak, and peaks within 1566.5 cm⁻¹ to 1411.3 cm⁻¹ are attributed to the tetrazole ring conjugate system of the ligand. The results of the infrared analysis are in well-accordance with the composition of the complex.

X-ray diffraction (XRD) was used to characterize the stability of the complex NPs in water. As shown in Figure 1A, the patterns are exactly identical to those simulated from single-crystal analysis, indicating the high phase purity of the bulk crystal. Notably, the PEG-free complex NPs were prepared by the same method, maintaining the structural integrity of the framework after exposure to water at 25°C for 1 day and demonstrating excellent stability of these complex NPs.

The fluorescence properties of the ligand H₄L and complex NPs were studied by using a fluorescence spectrophotometer at room temperature. As shown in Figure 1B, through the exploration of its emission peak, the ligand H₄L has the largest characteristic emission peak at 426 nm with 374 nm excitation. With 311 nm excitation, the complex has a maximum emission peak at 469 nm. The observed red shift of 43 nm for the complex is in good agreement with the previously reported complex (Guo et al., 2018). The shifts are probably due to π*–π transitions of the corresponding ligand because similar peaks also appear for the free ligand.

Based on the Fenton-like reaction mechanism, Cu(II) complex NPs can catalyze H₂O₂ to •OH (Liu C et al., 2019; Bao et al., 2020). In order to study the catalytic performance of the complex NPs, MB was selected as an indicator of ROS in the presence of H₂O₂. As shown in Figure 1C, the absorbance of MB decreased via oxidation when the complex NPs were added, indicating the effective generation of •OH. In contrast, no obvious absorbance decrease was observed without NPs even after 30 min. The Gd(III)–Cu(II) complex can also oxidize GSH to GSSG to further enhance the yield of ROS.

Nanocoprecipitation was used to improve the aqueous solution dispersability of the NPs. SEM (scanning electron microscope) and DLS (dynamic light scattering) were used to characterize the size and diameter of the complex NPs. As shown from Figure 1D, it can be seen that the complex can self-assemble NPs with good dispersion, and the average diameter is about 50 nm, suggesting the suitability for EPR. In addition, the stability of the complex in PBS (pH = 5.5 and 7.4) and DMEM with 10% FBS was also determined. The complex NPs without PEG_-2000 were retained in the aqueous solutions at RT for 1 day. The diffraction patterns are similar to the simulated ones (Supplementary Figure S1, ESI†), suggesting the high-phase purity of the bulk products. In addition, no obvious changes were observed after incubation with the NPs in PBS or in DMEM, indicating their good stability. Moreover, the DLS results showed the diameter of the NPs with a slight aggregation in water, PBS, and DMEM (Supplementary Figure S2, ESI†). This may be due to the low solubility of NPs in aqueous solution and absence of PEG₋₂₀₀₀ coating.

This complex belongs to the monoclinic space group P2 ₁/n with a crystallographic ally-independent symmetric unit. From Figure 2, each Gd(III) atom is nine-coordinated by four oxygen atoms (O1, O2, O3A, and O4A) from two carboxylate groups and five oxygen atoms (O5, O6, O7, O8, and O9) from five water molecules, thus forming a deformed and distorted tricapped trigonal prism coordination configuration. The Cu(II) cation is coordinated with four nitrogen atoms (N1, N5, N1B, and N5B) of the tetrazole ring and two carboxyl oxygen atoms (O1 and O1B) from two L⁴⁻ ions, forming a deformed octahedron coordination configuration. In addition, four oxygen atoms from two carboxyl groups of L⁴⁻ chelated with two Gd(III) ions (Figure 3), and two N atoms from two tetrazole rings and a carboxyl oxygen atom connected with one Cu(II) ion (Scheme 1). By analyzing the bond lengths of the complex, the bond lengths of Gd–O are 2.391–2.544 Å and that of Cu–O are 2.492 Å. The bond lengths between the Cu atom and N atom are in the range of 1.976–1.992 Å. As shown in Figure 4, two such ligands set up a double bridge between two Cu(II) ions and two Gd(III) ions to generate a bimetallacycle of [Cu₂Gd₂(L)₂]. The adjacent bimetallacycle is further connected to yield a one-dimensional chain structure, and by hydrogen bonding between the chains, a three-dimensional supramolecular structure was formed (Figure 5).

The MTT assay was performed to further investigate the cytotoxicity of the complex NPs. Different concentrations of NPs of the complex were cultured with HeLa, HepG2, and HT29 cells. As shown in Figures 6A–C, the cell viability of the group cultured with H₄L remained high even at high concentration, which demonstrated the low cytotoxicity of the ligand. The complex NPs displayed promising cytotoxic activity against all the three cancer cell lines. As the NPs concentration increases, the cell viability decreases, and the IC₅₀ values of HeLa, HepG2, and HT29 cells are calculated to be approximately 4.9 μg/ml (0.90 μM), 11.1 μg/ml (2.05 μM), and 5.5 μg/ml (1.01 μM), respectively. Among them, the IC₅₀ of the complex for HeLa was the lowest, which is superior to that of the previously reported Cu complexes with tetrazole (triazole)–carboxylate as ligands, such as [Cu(2-pytzipa)₂(H₂O)₂]·2H₂O (7 μM) (Zhai et al., 2017), [Cu(atzpa)₂], [Cu(pytzipa)₂], [Cu₄(Hphtz)₈](ClO₄)₄·4H₂O (Zhang et al., 2021), [Cu(L¹)₂(ClO₄)₂]∙2MeCN [Cu(L¹)₂(MeOH)₂](ClO₄)₂, and [Cu(L³)₂(NO₃)₂]∙3H₂O (Li et al., 2021) (Table 2), and also superior to various transition metal complexes based on tetrazole–carboxylate, such as [Zn (pytzipa)₂(H₂O)₄]·2H₂O (35 μg/ml) (Gu et al., 2018), Ca (2-pytzipa)₂(H₂O)₄] (48 μg/ml), and [Ca(3-pytzipa)₂(H₂O)₂]n (30 μg/ml) (Cao et al., 2019). This result also demonstrated that the Cu(II)-containing complex was able to effectively decrease the GSH level in the solution to induce cell apoptosis (Tang et al., 2019). On the contrary, the overexpression of GSH can reduce the reaction of Cu(II) to Cu(I) via Fenton-like reaction, further improving the production rate of •OH and reducing the antioxidant capacity of tumor cells (Liu Y et al., 2019). From the abovementioned fact, the Gd(III)–Cu(II) complex showed better apoptosis-inducing effect than most copper compounds; this may be because the addition of Gd(III) increased the uptake of NPs by cancer cells, resulting in a more efficient CDT therapeutic effect (Lee et al., 2015; Ren et al., 2019; Zeng et al., 2020; Stinnett et al., 2021).

After trypan blue staining, the cell nucleus incubated with the NPs showed a stronger blue–white color than the control group, indicating that it had a significant apoptosis-inducing effect to HeLa, HepG2, and HT29 cells (Figures 6G–I). From the comparison of these cell morphologies, it can be seen that the morphology of live cells in control groups is irregular, while that of dead cells tends to be regular. In addition, the volume of the dead cells had generally shrunk, possibly due to the mechanism of apoptosis rather than necrosis (Cao et al., 2019).