Antitumor activity and selectivity of a nanostructured peroxo polyoxoniobate

Abstract

In vitro and in vivo experiments were carried out to evaluate the activity and antitumor selectivity of new niobium oligomers. A polyoxoniobate (PONb) structure linked to the methylene blue dye (PONb–MB) showed promising results for some types of tumor cells. Molecular structures were determined for PONb and PONb–MB based on spectroscopic analyses using density functional theory calculations. Both caused a significant cell viability reduction in tumor cells (HeLa) with high effectiveness, because of the reactivity oxygen species (ROS) generated on their chemical structure. Experimental quantitative analysis demonstrated high selectivity for tumor cells (apoptosis 3× higher compared to MRC-5 healthy cells). Lower toxicities in healthy cells (e.g., MRC-5 and L929) confirm the selectivity of the synthesized compounds. It is noteworthy that PONb and PONb–MB are physiologically soluble compounds and can be suitable for intravenous treatments. Biodistribution studies in mice revealed very low accumulation in organs such as the spleen, heart, liver, lungs, and kidneys. Additionally, almost no toxicity was observed in the mice post-injection, which was confirmed by biochemical analysis.

---

Introduction

Polyoxometalates (POMs) are anionic metal oxo clusters usually composed of d⁰ species of Groups VI and V (e.g., Mo, W, V, Ta, and Nb). They are oligomeric aggregates of early transition metal cations linked by oxide anions that are formed by self-assembly processes.^1–3 The POMs have high-water solubility, low toxicity in humans, and their redox and photochemical capacities can present diversified usages, mainly in catalysis and medicine.^4,5 Mo-based POMs have been applied for antitumor activity while those based on V are known for their anti-viral effect. Although POMs based on W, Mo, and V have been deeply investigated in the last few decades, the chemistry of Nb-based POM (called polyoxoniobate, PONb) has been less explored.¹ Nevertheless, it must be emphasized that PONb has shown distinctive properties such as high negative charge, base surface oxygen atoms, and a wide range of cluster sizes that are suitable for several applications, for example inhibiting tumor development.⁶

The exploration of nanostructured materials for biomedical applications, particularly cancer therapy, has garnered significant interest over recent decades. Among these materials, polyoxometalates (POMs) have emerged as promising candidates due to their unique structural, electronic, and redox properties.⁷ Specifically, polyoxoniobates (PONbs) have demonstrated noteworthy potential owing to their unique ability to form diverse structures and exhibit significant chemical stability and bio-compatibility.⁸

Research into the antitumor activity of these materials has been motivated by the limitations of traditional chemotherapy, such as non-selectivity and side effects.⁹ Nanostructured POMs, particularly PONbs, have shown potential to overcome these limitations by providing targeted delivery and controlled release of therapeutic agents.¹⁰ Studies have indicated that POMs can induce apoptosis in cancer cells, inhibit tumor growth, and enhance the efficacy of conventional chemotherapy agents.^11,12 Furthermore, the potential of PONbs to selectively target cancer cells while minimizing damage to healthy cells remains a critical area of investigation.¹³

Recent advancements have highlighted the role of surface modifications and functionalization in enhancing the antitumor efficacy and selectivity of nanostructured PONbs. For example, surface modifications with organic moieties have been shown to improve their solubility, stability, and biocompatibility, thereby enhancing their therapeutic efficacy.¹⁴ Methylene blue (MB) is a potent sensitizer that can be incorporated into the PONb surface. It has been used to generate reactive oxygen species (ROS). ROS is highly electrophilic and can directly oxidize electron-rich double bonds in biological molecules and macromolecules.¹⁵ The cytotoxicity of ROS has been demonstrated in several disease models caused by tumor cells. In cancer cells, increased oxidative stress results from increased ROS production and/or impaired ROS scavenging capacity.¹⁶ A slight elevation in ROS levels within cancer cells relative to normal cells can exceed a critical threshold, leading to cancer cell death and suppressing tumor growth.¹⁷ Thus, the incorporation of agents that induce ROS generation, such as MB, into PONb is considered a targeted strategy to eradicate malignancies.

The structural diversity and multifunctionality of PONbs enable their interaction with various biological targets, including proteins and DNA, which can lead to targeted therapeutic effects.¹⁸ This multifunctionality is especially valuable in the context of combination therapies, where PONbs can be used to synergistically enhance the effects of other anticancer agents.¹⁹ Considering the promising preliminary results, there is a growing need for detailed investigations into the mechanisms of action, optimal formulations, and clinical potential of PONb-based nano-therapeutics.²⁰

Particularly, antitumor activities of PONb can be associated with the apoptosis process and oxidation of cellular components.^4–8 Its bio-application and understanding its behavior in a biological environment is one great challenge. Furthermore, it is also important to investigate how size, shape, and composition can influence its toxicity profile. In general, the experimental determination of molecular structures is not an easy task. From this point of view, molecular modeling techniques based on computational methods of quantum chemistry appear as an alternative to calculating the structural, electronic, and spectroscopic properties. Thus, it is possible to establish the structure–reactivity–activity relationship in the design of new drugs for various applications in medicinal chemistry.^21–24 Particularly, we have recently prepared a PONb complex by proposing a new class of materials to be used as anticancer agents.⁵ Experimental and theoretical analysis demonstrated that the Nb complex species presented reactivity and stability when docked into the DNA crystallographic structure.

Herein we report the synthesis of unique PONb species with emphasis on its application to cancer treatment. Two strains of carcinogenic cells were used to demonstrate the versatility of PONb for inhibiting tumor cells, with quite low toxicity to healthy ones. Furthermore, ex vivo biodistribution and toxicological studies with healthy mice revealed negligible damage, showing the potential application of PONb as a drug candidate for tumor treatment with particular attention paid to the achievements of several human cell carcinomas in vitro.

Results and discussion

Characterization of polyoxoniobates

Polyoxoniobate solutions were characterized by UV-Vis (blue lines in Fig. 1a and b). The spectrum of pure PONb shows an intense band in the UV region, centered at about 270 nm, typical of peroxide groups (ROS) that are linked to Nb atoms in the PONb structure (Fig. 1a). However, the MB dye addition to PONb generates a new species (PONb–MB) that absorbs radiation in the visible light region of the spectrum, with maxima at 574, 694, and 775 nm, besides the high-intensity UV peak centered at 243 nm (Fig. 1b). It is known that the MB dye has absorption bands at 612 and 665 nm,^8,25 which are not observed in the spectrum of PONb–MB (Fig. 1b). This result indicates the formation of a new species of polyoxoniobate rather than a sole mixture of the two compounds of PONb and MB dye. It is also important to point out that the ROS observed (bonded peroxide group) is not a radical species, and for this reason, the niobium compound might not be toxic to healthy cells.

Fig. 1 UV-Vis absorption spectra (Experimental and simulation using TD-DFT–PCM–Water single-point calculation) for PONb (a), PONb–MB system (b), fully optimized DFT structures (gas phase) of the PONb monomer (c), PONb–MB system (d) and polyhedral representation of PONb, and representation of the coordination polyhedron of Nb₂O₅ structure simulated from CIF 51 176 that was used to calculate the distances between atomic pairs in the PDF analysis.

The theoretical spectra in the UV-Vis region (red lines in Fig. 1a and b) show that the PONb species has an intense band at 287 nm, i.e., 6.3% shifted to the low-energy region compared to the experimental band (Fig. 1a). After the formation of the PONb–MB system (Fig. 1b), two absorption bands are observed: (i) an intense band at 274 nm assigned to an intramolecular transition between HOMO−2 (Fig. 1d) and LUMO+5 (Fig. S2a, ESI†) orbitals and (ii) a weak and broad one in the 550 nm region assigned to a transition between HOMO−2 (Fig. 1d) and LUMO (Fig. 1) that can be attributed to the strong intermolecular interaction between the PONb and MB species. More details on the TD-DFT calculations observed in Fig. 1c and d are found in the ESI† (Fig. S3a and b).

Thermodynamic parameters of the PONb–MB interaction were calculated according to eqn S1–S3 (ESI†). All calculated parameters are shown in Table S2 (ESI†). The interaction between PONb and MB molecules forms a strongly bound compound with an energy stabilization value of around −55.0 kcal mol⁻¹. The TΔS term is negative and, according to eqn S3 (ESI†), it contributes repulsively to ΔG_F of the PONb–MB system. According to eqn S2 (ESI†), the ΔE_F value is negative enough to produce a large negative ΔH_F (−63.4 kcal mol⁻¹), even though the thermal correction to enthalpy is negative (−8.4 kcal mol⁻¹). Thus, these results suggest that the formation of the PONb–MB system is thermodynamically favorable for ΔG_F equal to −37.1 kcal mol⁻¹.

Moreover, to estimate the negative charge of PONb and PONb–MB molecules, the solutions were characterized using zeta potential measurements. The results demonstrated a high negative value of the PONb cluster (−50.6 mV). These negative charges possibly allowed the binding of MB cationic dye to generate a new stable species of PONb–MB, in good agreement with the theoretical calculations. This hypothesis was further attested by zeta potential measurements of PONb–MB that presented a lower value (−36.5 mV) than the pure PONb, due to the electrostatic interactions between the cationic dye and the negative charge oxygen species in the PONb structure. This molecular interaction resulted in an increase in the calculated HOMO–LUMO gap energies compared to the value obtained for the isolated PONb, rising from 0.45 eV to 0.56 eV. This effect can be attributed to the simultaneous destabilization of both HOMO (−0.23260 E_h to −0.19405 E_h) and LUMO (−0.21612 E_h to −0.17343 E_h) energy levels. The prepared polyoxoniobates (PONb and PONb–MB) were characterized via DFT calculations, which indicate the formation of the structures shown in Fig. S2 (ESI†). The optimized geometry of PONb (Fig. S2a, ESI†) has short- and medium-range hydrogen bonds in the range of 1.48–1.87 Å between the hydrogen atoms of peroxide ligands and neighboring oxygenated ligands. The bond lengths Nb(V)–O are in the range of 1.8–2.10 Å (μ₂-O), 2.14–2.22 Å (μ₂-OH) and 1.75–1.80 Å (η-OOH) and (η-OO⁻). Similar results for these bond types were found by Grimme et al.²⁶ for the crystalline structure of the peroxopolioxoniobate ion salt [As₂Nb₄(O₂)₄O₁₄]⁶⁻ and by Si et al.²⁷ also through an electronic DFT study based on the salt crystal of the peroxohexaniobate ion salt [H₃Nb₆O₁₉]⁵⁻. The latter theoretical analysis used B3LYP as in the present study.

The optimized geometry of the PONb–MB system (Fig. S2c, ESI†) demonstrated no significant changes in the Nb(V)–O bond lengths, but there are some dihedral angle rearrangements involving the terminal oxygenated groups. Fig. S2c (ESI†) shows that the PONb–S–MB interaction appears to be electronically stronger (shorter intermolecular distance, ≈2.11 Å) than the PONb–N–MB interaction (≈2.47 Å). Since each MB structure is cationic (charge +1) and the niobate model is anionic (charge −2), we can point out that both modes of interaction can be favorable and form a neutral PONb–MB system.

XPS analysis was conducted to assess the surface properties of the materials. The high-resolution Nb 3d for the PONb and PONb–MB is presented in Fig. 2. Both samples exhibit similar surface niobium, with peaks at approximately 210 eV for 3d_3/2 and 207 eV for 3d_5/2, which are characteristic binding energies of electrons coming from highly oxidized niobium species, such as Nb(V).^28–30 It is noteworthy that there is a band shift in PONb–MB, caused by the binding of MB to the PONb structure. However, this shift does not cause a change in the oxidation state of niobium. The XPS results corroborate those presented by UV-Vis (Fig. 1) and Raman (Fig. S10, ESI†) analyses.

Fig. 2 XPS spectra of Nb 3d for PONb and PONb–MB.

The EPR analyses were carried out to characterize the reactive oxygen species, which are the active groups in anticancer activity. No paramagnetic signal was observed in the EPR spectra (not shown) for PBN experiments (Fig. 3a), i.e. the studied compounds cannot produce radicals under such experimental conditions. On the other hand, TEMPO was detected only for PONb–MB and MB. As presented in Fig. 3b and c, the singlet oxygen production initiated at five minutes and, after that, evolved until the end of the test (fifteen minutes). Moreover, the time-dependent signal was four times more intense for PONb–MB compared to MB, which indicates that the electrostatic compound produced from the PONb and MB reaction catalyzes the conversion of molecular oxygen from the triplet to the singlet state. These results show that the species formed by the reaction between PONb and MB dye to form PONb–MB has the potential to generate reactive species of the singlet oxygen type, indicating high potential for use as an anticancer agent. In addition, even though the PONb species did not present this type of oxygenated species, the peroxo groups (observed in experimental characterization studies and theoretical calculations) can exhibit positive effects on anticancer tests.^5,7

Fig. 3 Time-dependent EPR spectra of TEMPO for PONb (a), MB (b), and PONb–MB (c). Raman spectra of PONb and MB dye (d), PONb–MB, and MB dye (e). Differences in vibrational modes between PONb–MB and MB are in the range of 680–880 cm⁻¹.

Studies via Raman spectroscopy corroborated to identify the peroxo groups in the PONb species. In fact, the Raman spectra (Fig. 3d) showed an intense band in the lower frequencies for niobium compounds. Such a band is usually associated with the Nb–O lattice band.^22,31,32 In the theoretical Raman spectrum, this band was observed in the range of 680–800 cm⁻¹ (see the bands highlighted in Fig. S4d and Table S3, ESI†). Furthermore, the PONb shows a band around 880 cm⁻¹ that is characteristic of the vibrational mode of peroxo species.³³ However, in the PONb–MB system, this signal was partially superimposed by MB dye bands. Our DFT calculations showed that the stretching modes of the peroxo groups in the PONb model are observed in the range of 1230–1400 cm⁻¹ (see the bands highlighted in Fig. S4d and Table S3, ESI†). For the peroxo groups involved in the intermolecular interactions between PONb and MB, these bands are displaced to 740–770 cm⁻¹ (see the bands highlighted in Fig. S4f and Table S3, ESI†). Experimentally, it was observed that the PONb–MB presents all the vibrational modes of MB, but they are more intense, and their frequencies slightly deviated (see Fig. 3e), which again indicates that the PONb–MB is a new compound, not just a simple mixture of PONb and MB dye.^34–36 The differences found between theoretical and experimental frequencies for the analyzed vibrational modes are related to the theoretical unimolecular model proposed. The experimental solid sample has PONb–PONb interactions that can affect the absorption spectrum of the analyzed groups and that were not considered in our theoretical model. Nevertheless, the B3LYP calculations of the vibrational modes showed good agreement with the experimental data. Details of the theoretical assignments of the IR and Raman spectra (Fig. S4 and Table S3) are provided in the ESI.†

Cell viability associated with PONb and PONb–MB in HeLa (tumoral), MRC-5 and L929 (healthy) cell lines

Fig. 4 displays the cytotoxic effects of PONb (a) and PONb–MB (b) on HeLa and MRC-5 cells (treated for 24 h), evaluated by the MTT assay. There was a significant reduction in cell viability for polyoxoniobate compounds at concentrations above 1 mg L⁻¹, in the two tested strains. For PONb, this decrease is more significant and the viability for tumor cells is less than 4% from the concentration of 5 mg L⁻¹ (Fig. 4a). The viability of the healthy cell, incubated at the same concentration is about 40%, indicating certain tumor selectivity in the PONb treatment. This cytotoxic profile can be verified also for PONb–MB, although to a lesser extent (Fig. 4b). At the concentration of 5 mg L⁻¹, the cell viability is 15% for tumor cells and 39% for healthy cells. On the other hand, the compounds used in this study are still poorly understood regarding their biocompatibility; therefore, assays with different cell lines are required to evaluate cytocompatibility. In order to confirm the low toxicity in healthy cells, which is a highly desirable property for a potential drug, the action of PONb and PONb–MB on the cell viability of L929 healthy cells was also investigated (Fig. 4). The PONb (Fig. 4c) and PONb–MB (Fig. 4d) solutions exhibited no significant cytotoxic effect at concentrations lower than 10 mg L⁻¹, after 24 and 48 h of incubation. These results further demonstrate that the two polyoxoniobate molecules have no cytotoxic effect on L929 healthy cells (fibroblasts).

Fig. 4 Cell viability in HeLa (tumor) and MRC-5 (healthy) cells in the presence of PONb (a) and PONb–MB (b). Cell viability in L929 healthy cells with PONb (c) and PONb–MB (d) and solutions at different concentrations and incubation times. Fluorescence microscopy images of the cells after 24 h of incubation with PONb and PONb–MB (e) and cell apoptosis (f).

Fluorescence microscopy images of the cells after 24 h of incubation with the prepared materials are shown in Fig. 4e. Detection of propidium iodide (DNA-specific red fluorescence) shows dead cells while calcein-MB makes the cell membrane of living cells visible (green fluorescence). A larger number of dead cells are observed in the images of the tumor cells incubated with the materials, mainly in the PONb group (Fig. 4e). The percentage of apoptosis calculated after counting the cells is shown in Fig. 4f. For the two studied cell lines, the control group had a death rate of less than 1%. Concerning the groups incubated with PONb and PONb–MB (10 mg L⁻¹), the apoptosis was statistically comparable to tumor cells. Nevertheless, it should be emphasized that an apoptosis percentage of approximately 3× higher was obtained in tumor cells compared to healthy cells, for both PONb and PONb–MB (Fig. 4f). This result confirms the effect observed in the cell viability assay, indicating a greater action of the prepared polyoxoniobates on tumor cell lineage. The ability to cause damage to tumor cells with significant selectivity makes both compounds promising materials for cancer therapy.

The action mechanism of the polyoxoniobates selectivity related to tumor cells was investigated. For that, studies involving the formation of intracellular reactive oxygen species (ROS) were performed. Representative fluorescence microscopic images of the cells after 4 h of incubation are presented in Fig. 5. The cells with higher ROS species are shown as red dots (Fig. 5a). As such, the redder the dots, the greater the ROS generation. It is known from the literature that when cells are subjected to ROS, these species can affect the vascular system of the tumor cell causing its death.³⁷ The ratio of dead to live cells was determined by counting. In both cell lines, ROS species increased for PONb and PONb–MB groups compared to control cells (Fig. 5b). Moreover, the death of tumor cells showed a statistically significant increase (about 50%). This result indicates that the capacity to promote higher cell death signaling in tumor cells is associated with a greater generation of ROS species. Habtetsion et al.³⁸ showed that adoptive immunotherapy deeply altered tumor metabolism, resulting in glutathione depletion and accumulation of ROS in tumor cells. Raza et al.³⁹ reported that a certain level of ROS is required by cancer cells, which can lead to cytotoxicity in them. In the present work, the new Nb-containing molecules are rich in labile oxygen species which might be interesting as a cancer therapy strategy.

Fig. 5 (a) Fluorescence microscopic images of the cells after 4 h of incubation and (b) ROS intensity for groups studied compared to control cells.

The results indicate that the formed niobium species show anti-cancer action through two different mechanisms: (i) the PONb species did not present radical species or the formation of singlet oxygen. Thus, the peroxide-type bonds (or the peroxo group) present in the molecule (Fig. 6a) must be directly responsible for the effect observed in the cell viability tests. An indication that the peroxo group in the PONb forms an interaction with the tumor cells decreasing their cell viability can be observed by the ³¹P NMR spectra of the PONb in contact with the 5-GMP molecule (used to simulate the possible interaction between the active species and the DNA).

Fig. 6 (a) Representative scheme of the peroxide-type bond (or peroxo group) and how it acts on tumor cells, (b) ³¹P NMR spectrum for the 5-GMP molecule, (c) ³¹P NMR spectrum showing the interaction between PONb and 5-GMP and, (d) the representative scheme of singlet oxygen generation promoted by the action of light in the compound PONb–MB.

Fig. 6b shows the ³¹P NMR spectrum for the 5-GMP molecule. It is observed that it has a single P signal with a chemical shift of 5.95 ppm, corresponding to the P of the phosphate group. However, in Fig. 6c, the ³¹P NMR spectrum of the molecule in contact with PONb is shown. The presence of another signal was observed at a smaller chemical shift, 5.26 ppm, possibly corresponding to the interaction of P with the peroxo group. The presence of this new group affects the chemical environment of the P (phosphate group) from the 5-GMP molecule, because the P atom that is interacting with the peroxo group is more shielded. After all, this group is more electronegative, which causes the appearance of a signal with a smaller chemical shift.⁴⁰ The PONb–MB species showed a capacity for generating singlet oxygen (Fig. 6d) probably due to the action of ambient light promoting the electronic transition for the conduction band of the species formed by the reaction of PONb with the MB dye.

The innovative results presented in this work indicate that ROS are formed and have their generation modulated when the dye is chemically bound to the polyoxoniobate species.³⁹ We believe that these results open a new line of possibilities since the dyes of various natures can present even better results. In addition, inorganic cations may also interact with the negative charges of the polyoxoniobate generating more versatile anticancer candidates. In this context, our new achievements might contribute to the development of new drugs with antitumoral and antiviral proposals.

In vivo tests for biosafety assessment

The promising results obtained for the polyoxoniobates for tumor cells motivated further investigation through in vivo tests with mice. Among several questions that can be raised, one can point out the concern of pharmaceutical accumulation, i.e., the accumulation of the polyoxoniobates in mice organs. The results showed a biphasic clearance profile, indicated by the red curve in Fig. 7a, with a fast half-life of 7.7 min and a slow half-life of 9.1 min.

Fig. 7 In vivo studies after PONb–MB intravenous administration into healthy Swiss mice. Blood clearance assays (a) and the biodistribution profile post-injection of the PONb–MB determined by the Nb loading in the mice organs (b).

The biodistribution profile of the possible drug candidate was determined by the Nb loading in mice organs (Fig. 7b). Some uptake was observed in kidneys and liver, which may represent the contribution of these organs in the elimination of the Nb compound. Noteworthy is the higher uptake in the heart at 60 min and 240 min post-injection, which indicates the affinity of PONb–MB towards this organ and suggests potential cardiotoxicity.

Thus, to better analyze and prove the safety profile of PONb–MB, a toxicological assay was performed (Table 1). The data suggest that the treatment with PONb–MB may not cause tissue damage. The CKMB value indicates no cardiac tissue damage, even with the higher uptake presented in the biodistribution study. Therefore, the values found for all the tests are statistically like those of the control groups, the occurrence of acute toxicity not being indicated.

Table 1 Biochemical parameters of healthy Swiss mice after intravenous administration of PONb–MB 1 h post-injection (n = 6)

	Control	PONb–MB
CKMB (U L^−1)	0.32 ± 0.13	0.26 ± 0.12
AST (U L^−1)	184.43 ± 55.98	178.26 ± 65.96
ALT (U L^−1)	22.17 ± 4.56	23.75 ± 4.39
Creatinine (mg dL^−1)	0.61 ± 0.05	0.56 ± 0.06
Urea (mg dL^−1)	47.54 ± 13.26	60.18 ± 14.95

---

Experimental

Preparation of polyoxoniobates

For the synthesis of polyoxoniobate, 46.2 mmol of the NH₄[NbO(C₂O₄)₂(H₂O)](H₂O)₁ (provided by CBMM) was dissolved in 100 mL of distilled water at 90 °C. Then, an NH₄OH solution (5 mol L⁻¹) was slowly added to the salt solution for precipitation of the niobium oxyhydroxide. The suspension was aged under magnetic stirring for 12 h and the obtained solid was filtered. Afterward, 8 mL of H₂O₂ (30%, v/v) and 100 mL of distilled water were added to 300 mg of niobium oxyhydroxide. After 12 h, the solid was centrifuged and the liquid supernatant contained the leached species of the peroxoniobate complex.⁸

The reaction yields to obtain PONb was around 50%. The PONb yellow solution was used to prepare the PONb–MB after reaction with a methylene blue solution (MB). For that, 10 mL of a 50 mg L⁻¹ solution of the MB dye was added to 10 mL of the PONb. The cationic dye strongly binds to the structure of the PONb by electrostatic attraction, forming the PONb–MB compound, of blue color.

Crystallization was carried out using the slow solvent evaporation method. For this, 30 mL of acetone PA and 5 mL of polyoxoniobate were used. The mixture was left to rest for 7 days at 25 °C until total evaporation. The material obtained was macerated and taken for X-ray diffraction. As usual, we first carried out quick measurements (exploratory: sweep 2 at 140 degrees/step 0.05 degrees/sampling time 82 s/measurement duration 8 min) to check what would be the ideal parameters to be used in the main measurements, carried out only in the angular range between 5 and 60 degrees. The increase in intensities was prioritized, increasing the sampling time (600 seconds) by approximately 7.5 times in relation to that of exploratory measurements and maintaining the same step as exploratory measurements (0.05 degrees). In this way, the adjustment of the reflections obtained suggests the presence of the H₈Nb₆O₁₉·10H₂O structure [JCPDS 14-0288], orthorhombic, with cell constants: a = 10.01 Å, b = 12.12 Å, and c = 12.61 Å (Fig. S7, ESI†).

Characterization of the compounds

The Nb concentration in the polyoxoniobate solutions was determined via inductively coupled plasma mass spectrometry (ICP-MS Agilent 7700), using an analytical curve for the Nb element. The polyoxoniobate solution (PONb) and the solution after the methylene blue dye addition (PONb–MB) were characterized via UV-Vis spectroscopy (Shimadzu – 2600/2700) and zeta potential measurements (Zetasizer Nano ZS). The analytical procedure was conducted in a zeta-sizer Nanoseries Zs apparatus (Malvern Instruments, Malvern, UK) after the adequate dilution of the samples in ultra-pure MilliQ® water. The results were expressed as mean ± standard deviation for at least three different batches of each sample.

The polyoxoniobate solution was placed in contact with the 5-GMP molecule (sodium guanosine 5-monophosphate, which simulates the DNA) to analyze the possible interaction between the 5-GMP molecule and the PONb compound. This interaction was verified using NMR of ³¹P (Bruker spectrometer 200 MHz), using 1024 scans and phosphoric acid as the standard. The spectrum of the 5-GMP pure molecule was obtained using 20 mg of 5-GMP dissolved in 600 μL of D₂O. Afterward, 400 μL of D₂O was added to 200 μL of a 5-GMP and PONb mixture. Then, the system was taken to the spectrometer. Then X-ray photoemission spectroscopy (XPS) analyses were performed on an Amicus spectrometer (Kratos Analytical, UK) using Al Kα as the excitation source operating at 240 W (12 kV and 20 mA). The high-resolution regions were recorded with 0.05 eV step. The C 1s signal at 284.6 eV, from the contaminating carbon, was used to calibrate the instrument. Spectra were analysed using the Vision software from Kratos Analytical with peaks fitted using Gaussian–Lorentzian functions and the Shirley background.

The reactive oxygen species (ROS) were characterized by using the EPR spin trapping technique using PBN (N-tert-butyl-α-phenylnitrone) and TEMP (2,2,6,6-tetramethyl-4-piperidone) as spin trapping molecules. The PBN is known to react with radicals and produces very stable spin adducts,¹² while the TEMP reacts with singlet oxygen producing a nitroxide radical called TEMPO.¹³ One milliliter of PBN (85 mM) or TEMP solution (1 M) was added to 1 mL of PONb, PONb–MB, and MB aqueous solutions. These mixtures were placed in a 5 mL beaker and magnetically stirred for 20 min. Aliquots of the samples were extracted every 5 min using a capillary glass tube and the data were acquired using an X-band Magnettech miniscope 400 spectrometer. The tests were performed at room temperature and illumination. Moreover, Raman data were collected on a Bruker Senterra spectrometer with a 532 nm laser operated at 20 mW.

The structure of the materials was further investigated by X ray powder diffraction and PDF patterns. The XRPD patterns were collected using a STADI-P diffractometer (Stoe®, Darmstadt, Germany) with MoKα1 radiation (λ = 0.7093 Å), operated in transmission geometry. The measurements for the application of the low energy PDF method were performed using 0.5-mm diameter special glass capillaries no. 14 (Hilgenberg®, Malsfeld, Germany), which reduce the container contribution to the measurement. The data were measured from 2.0 to 136.4° (2θ). An empty capillary was measured with the same conditions as the filled capillaries to remove the instrumental contributions of the measurements. A Q-range from 0.1 to 16.4 Å⁻¹ was considered for the S(Q) (total structure-function) that was Fourier transformed to obtain the PDF pattern. S(Q) and PDF patterns were obtained using the PDFGetX3 software.¹⁴ In this software, all non-structural contributions are determined through an ad hoc approach; then they are parameterized, and the parameters are used to obtain the structure function S(Q).

The main distances up to 10 Å were calculated using the Bond_Str software¹⁸ based on the crystallographic information framework file (CIF file) for the structure of Nb₂O₅ obtained in the Inorganic Crystal Structure Database®, under the reference code 51176. The calculated distances for individual atomic pairs were used to compare the distances observed in the PDF patterns.

In vitro comparative tests using HeLa (tumoral) and MRC-5 (healthy) cells

The assays were performed using the cell lines HeLa (ATCC®-CCL-2™), a tumor cell derived from human cervical adenocarcinoma, and MRC-5 (ATCC®-CCL-171™), a healthy cell line derived from human lung fibroblast. The cells were cultured in a CO₂ incubator (5% CO₂ Cole Parmer) with a humid atmosphere at 37 °C in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin/amphotericin B (complete DMEM medium). Upon reaching 80% confluency, cells were harvested using sterile PBS to wash the plate and sterile 1% Trypsin solution to detach the cells to be collected in the falcon tube. The cells were then counted with the aid of a Neubauer chamber.

The cell viability assay with sulforhodamine B (SRB) was performed according to the literature.¹⁹ For this assay, the cells were seeded in 96-well culture plates (1 × 10⁴ per well) and incubated at 37 °C/5% CO₂. After 24 h, the cells were treated with PONb and PONb–MB at concentrations of 1, 5, 10, 50, and 100 mg L⁻¹ (concentrations are in terms of the niobium content present in the compound). The group defined as cell control received only full DMEM medium and was maintained under the same conditions as the treated groups. After 24 h, the cells were fixed through protein precipitation with 10% trichloroacetic acid at 4 °C (100 μL per well) for 1 h. After five washing steps, the cells were stained with 0.04% SRB (100 μL per well) at room temperature for 1 h. Afterward, the plate was washed with 1% acetic acid four times to remove the unbound stain. The plate was air-dried at room temperature and the bound protein stain of each well was solubilized with 100 μL of 10 mM Tris base [tris(hydroxymethyl)aminomethane]. The optical density was measured by spectrophotometry on a UV-Visible Microplate Reader (Molecular Devices) at 510 nm. The survival fraction was calculated as a percentage of the control (Absorbance in control = 100% survival). The experiments were carried out in triplicate.

The ability of the material to damage the cell membrane and promote cell death signaling was assessed by fluorescence microscopy imaging. For the cell dead assay, the cells were seeded in 96-well culture plates (1 × 10⁴ per well) and incubated at 37 °C/5% CO₂. After 24 h, the cells were treated with PONb and PONb–MB at a 10 mg L⁻¹ concentration. The group defined as control cells received only a complete medium and were maintained under the same conditions as the treated group. After 24 h of incubation at 37 °C, the cells were supplemented with 1 μM calcein-MB (Life Technologies) and 2 μM propidium iodide (Life Technologies) in DMEM culture medium. The images were collected with 4× objective lens on an inverted fluorescence microscope Olympus IX70 (Olympus America, Melville, NY). The percentage of apoptosis was calculated from the ratio between the number of dead and living cells, which were counted with the aid of ImageJ software. Each treatment was performed in triplicate, three different images were counted, and the means were used for statistical analysis.

Reactive oxygen species (ROS) are the result of oxygen reduction during aerobic respiration by various enzymatic systems within the cell. The assay detects the formation of intracellular ROS (especially superoxide and hydroxyl radicals) in live cells employing the fluorometric method. For this assay, the cells were seeded in 96-well culture plates (1 × 10⁴ per well) and incubated at 37 °C/5% CO₂. After 24 h, 100 μL of ROS Detection Reagent was added to each well. The plate was incubated for another 1 h and then the cells were treated with PONb and PONb–MB in PBS medium at the concentration of 10 mg L⁻¹. The group defined as control cells received only medium and were maintained under the same conditions as the treated group. Multi-Mode Reader Cytation 5 Cell Imaging (Biotek) was used to measure the fluorescence intensity and to collect the images with a 4× objective lens. The detection of ROS was performed using an excitation wavelength of 490 nm and collecting the fluorescence emission of 525 nm. Each treatment was performed in triplicate, three different images were counted, and the means were used for statistical analysis.

In vitro tests using L929 (healthy) cells: cell viability with MTT assay

Healthy cell lines L929 (connective tissue fibroblasts) were cultured; Manassas, VA, USA in Dulbecco's modified Eagle's medium, supplemented with 10% FBS (fetal bovine serum) and 1% amphotericin B-streptomycin (100 IU mL⁻¹). Cells were maintained under the following conditions: 37 °C and 5% CO₂, with medium exchange every two days. After reaching confluence, the cells were seeded in 96 well plates in concentrations around 1 × 10⁵ cells per well for the tests. The tests were performed after the cells reached 70% confluence in which different concentrations of PONb and PONb–MB were added in each well: 3.0; 5.0; 8.0; 10.0; 12.0; 15.0 and 20.0 μg mL⁻¹, in six replicates. An aqueous solution of methylene blue dye (MB) at 12.5 μg mL⁻¹ was used as a positive control. The solutions were sterilized by filtration (0.22 μm membrane). The contact time (incubation) with the PONb, PONb–MB, and MB solutions was 24 and 48 h at 37 °C and 5% CO₂. Then, the reagent 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide MTT was added to the plates. After 4 h of incubation at 37 °C for the formazan crystals formation, a sodium dodecyl sulfate (SDS) detergent solution was added. After 14 h, the UV-vis spectrophotometer readings were taken at a wavelength of 570 nm. The respective negative controls (containing only untreated cells) and blank controls (no cells and no treatment, containing all solutions involved in the test) were analyzed on all plaques. The effects on cell viability were expressed as a percentage of the viability detected for the treated cells versus the viability of the cells incubated with culture medium alone. Data are expressed as the mean ± standard deviation (n = 6).

In vivo experiments

The study was carried out in compliance with the ARRIVE guidelines. Third Swiss CF1 female mice (6–8 weeks old, 24–28 g) were obtained from the Faculty of Pharmacy of the Federal University of Minas Gerais (UFMG). The animals had free access to water and food and were kept under controlled environmental conditions. All animal procedures were approved by the Institutional Animal Care and Use Committee (CEUA/UFMG), under protocol 118/2019. Mice were randomly separated into groups of 6 animals for the in vivo studies. The number of mice was defined similarly to other papers from our group.^21,41

For the blood clearance experiments, aliquots of 100 μL of a tenfold diluted PONb–MB solution were injected intravenously into the healthy Swiss mice (n = 6, 6–8 weeks old, 20–24 g). An incision was made in the tail of the animals and blood was collected in preweighed tubes at 15, 30, 60, 120, 240, 480, 1440, and 2880 min after administration. The mice were anesthetized with a mixture of xylazine (15 mg kg⁻¹) and ketamine (80 mg kg⁻¹). Blood was digested using HCl and HNO₃ solution, using a ratio of tenfold solution diluent to blood.

In the biodistribution assays, aliquots of 100 μL of a ten-fold diluted PONb_MB solution were injected intravenously into healthy Swiss mice (n = 6, 6–8 weeks old, 20–28 g). At the post-injection times of 30, 60, and 240 min, organs such as spleen, heart, liver, lungs, kidneys, and blood were removed and weighed.²¹ The organs were macerated using tissue macerator, and the supernatant was used for the determination of the niobium content present in each organ. For the analysis by ICP-MS, the samples went through acid digestion, using distilled HNO₃ (Merck), 65% (w/v). For sample preparation, 1 mL of HNO₃ was added to each 1 mL of the supernatant resulting from the digestion of each organ. The mixture was left under constant magnetic stirring at 50 °C for two hours and then centrifuged using a QUIMIS centrifuge, model 0222E24. The supernatants were transferred to polyethylene tubes, completed to 10.0 mL, and analyzed via ICP-MS.

For the biochemical analyses, aliquots of 100 μL of a tenfold diluted PONb_MB solution were injected intravenously into healthy Swiss mice (n = 6, 6–8 weeks of age, 20–28 g). After 24 h of the injection, blood was collected using an anticoagulant (0.18% w/v EDTA). The blood was centrifuged at 5000 rpm for 10 min to obtain the plasma used to perform the biochemical analyses. The biochemical tests were performed using commercial kits from Labtest® (Lagoa Santa, Brazil) using Bioplus BIO-2000 semiautomatic analyzer equipment (São Paulo, Brazil). Heart (CK-MB), liver (AST, aspartate aminotransferase, and ALT, alanine aminotransferase), and kidney (urea and creatinine) functions were tested.

It is worth mentioning that all the reagents used in in vitro and in vivo experiments have >95% purity using HPLC analysis. The methods presented in the work were carried out in accordance with relevant guidelines and regulations.

Statistical analysis

Data are expressed as mean ± SD. Statistical analyses were performed using GraphPad PRISM, version 6.00 software (GraphPad Software Inc., La Jolla, CA, USA). Differences between the experimental groups were tested by one-way analysis of variance (ANOVA), followed by the Tukey test, or Student's t-test when two groups where compared. All data showed normal distribution and homoscedasticity. P values <0.05 were considered statistically significant.

Computational details

All calculations were performed using the Gaussian 09 package48 employing the density functional theory (DFT) methodology²¹ with the B3LYP functional⁴² using the 6-31G(d,p) basis set⁴³ for O, H, N, S, and C atoms and the LANL2DZ⁴⁴ effective core potential (ECP) for Nb(V) ions. B3LYP/LanL2Dz/6-31G(d,p) level of theory used in the present study has been widely used in the literature^45,46 and has demonstrated good agreement against experimental data for Nb compounds. Therefore, we point out that this level of theory can contribute to experimental analysis in elucidating the structural properties of the system.

Firstly, the monomers corresponding to the structures of PONb and MB (Fig. S1, ESI†) were fully optimized in the gas phase and their geometries were used to build the structure of the PONb–MB system. The monomers (PONb and MB) and the PONb–MB system were characterized by harmonic frequency analysis indicating the optimized geometries as true minima. Finally, electronic excitations (UV-Vis spectra) were calculated using the B97-D functional²⁷via TD-DFT formalism⁴⁷ and the molecular orbital analysis was computed, both by using the Polarizable Continuum Model (PCM)56 to describe solvent effects – water solvent (dielectric constant, ε = 78.3553).

Conclusions

In this study, we describe the synthesis of unique polyoxoniobate compounds obtained through a new synthetic route from species of niobium oxide treated with hydrogen peroxide solution (PONb). The addition of methylene blue dye (MB) to PONb solution generated new oligomer species (PONb–MB). Theoretical analyses and thermodynamic properties contributed to propose a new molecular structure for the polyoxoniobate compounds. Both PONb and PONb–MB present reactive oxygen species (ROS), which resulted in high activity and selectivity as an antitumoral drug candidate. The results indicate that the proposed polyoxoniobate molecule can be useful in the case of selective cytotoxic effects on tumor cells.

It is noteworthy that the viability of healthy cells was not strongly affected by the presence of both compounds, which may indicate that these Nb compounds can be widely and safely used as new possible drugs. Additionally, in vitro safety outcomes are corroborated by in vivo assays, which showed minimal uptake in the major organs with no sight of toxicity in the renal, liver, and heart biochemical analyses. In summary, all these findings open new perspectives for this class of compounds to be a promising alternative for cancer treatment with particular attention as antitumor agents and quite low toxicity activities.

Further investigations on therapeutic efficacy and new developments of delivery systems are certainly necessary for future clinical tests. One can also envision that these new polyoxoniobate molecules can be tested in conjunction using other conventional approaches such as chemotherapy to treat cancer cells, instead of simply replacing other traditional drugs since it has demonstrated low toxicity for healthy cells that eventually surround the tumors.

Author contributions

Luiz C. A. Oliveira: term, supervision, conceptualization, project administration, funding acquisition, resources. Cinthia C. Oliveira: conceptualization, methodology, writing – original draft, writing – review & editing, visualization. Samuel M. Breder: visualization, writing – original draft, data curation. Klaus Krambrock: methodology, formal analysis, resources. Poliane Chagas: investigation, data curation. Ana P. Heitmann: investigation, data curation. Estella G. da Mota: visualization, writing-original draft, writing – review & editing. José B. Gabriela: investigation, data curation. Jadson C. Belchior: formal analysis, writing – original draft. Leonardo A. De Souza: investigation, formal analysis, writing – original draft. Tiago H. Ferreira: investigation, methodology, writing – original Draft. Sued E. M. Miranda: investigation, methodology. André L. B. Barros: methodology, writing – original draft. Cynthia L. M. Pereira: writing – original draft, visualization. Vinícius D. N. Bezzon: methodology, writing – original draft. Fábio F. Ferreira: methodology, writing – original draft.

Data availability

The authors declare that the data supporting the findings of this study are available within the paper and its ESI.† Should any raw data files be needed in another format, they are available from the corresponding author upon reasonable request. Source data are provided with this paper.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors are immensely grateful for the financial support of CAPES, CNPq, and FAPEMIG. Furthermore, they acknowledge the Companhia Brasileira de Metalurgia e Mineração (CBMM) for the donation of niobium compounds. Leonardo A. De Souza thanks the FAPERJ for the financial support (E-26/211.911/2021).

References

1. J. T. Rhule, C. L. Hill, D. A. Judd and R. F. Schinazi, Chem. Rev., 1998, 98, 327–357.
2. M. Nyman, Dalton Trans., 2011, 40, 8049.
3. A. P. Heitmann, P. S. Patrício, I. R. Coura, E. F. Pedroso, P. P. Souza, H. S. Mansur, A. Mansur and L. C. Oliveira, Appl. Catal., B, 2016, 189, 141–150.
4. Y. Zhang, J. Q. Shen, L. H. Zheng, Y.-X. Li, Z.-M. Zhang and E.-B. Wang, Cryst. Growth Des., 2013, 14, 110–116.
5. Y. Wang, S. Sun, W. Hu, Q. Yan and H. Zeng, Bioorg. Med. Chem. Lett., 2017, 27, 2357–2359.
6. A. Thomadaki, H. Lymberopoulou-Karaliota, A. Maniatakou and A. Scorilas, J. Inorg. Biochem., 2011, 105, 155–163.
7. D. Zhang, Z. Liang, S. Xie, C. Zhang, J. Wang and J. Niu, Inorg. Chem., 2014, 6, 6–11.
8. E. C. Pereira-Maia, I. P. Souza, K. J. R. C. Nunes, A. A. Castro, T. C. Ramalho, F. Steffler, H. A. Duarte, A. Pacheli, P. Chagas and L. C. A. Oliveira, RSC Adv., 2018, 8, 10310–10313.
9. L. A. De Souza, M. M. Soeiro and W. B. De Almeida, Int. J. Quantum Chem., 2018, 1–15.
10. L. A. De Souza, H. F. Dos, L. T. Costa and W. B. De Almeida, J. Inorg. Biochem., 2018, 178, 134–143.
11. J. C. Kato, R. B. Silva, F. T. Pappa and G. L. Belchior, Phys. Chem. Chem. Phys., 2015, 17, 2703–2714.
12. H. F. Almeida, E. R. De Souza, L. A. De Almeida and W. B. Dos Santos, J. Chem. Inf. Model., 2020, 60, 500–512.
13. G. R. Buettner, Free Radical Biol. Med., 1987, 3, 259–303.
14. J. W. Han, S. K. Hwang, T. M. Yoon and Y. Kang, Chemosphere, 2011, 84, 1095–1101.
15. J. D. Hayes, A. T. Dinkova-Kostova and K. D. Tew, Cancer Cell, 2020, 38, 167–197.
16. B. Perillo, M. Di Donato, A. Pezone, E. Di Zazzo, P. Giovannelli and G. Galasso, Exp. Mol. Med., 2020, 52, 192–203.
17. E. C. Cheung and K. H. Vousden, Nat. Rev. Cancer, 2022, 22, 280–297.
18. J. Rodríguez-Carvajal, Phys. B, 1993, 192, 55–69.
19. A. L. Orellana and E. A. Kasinski, Bio-Protoc., 2017, 6, 1984.
20. G. V. O. Fernandes, G. G. Alves, A. B. R. Linhares, M. H. Prado da Silva and J. M. Granjeiro, Key Eng. Mater., 2011, 493–494, 37–42.
21. J. Munkert, E. R. Gomes, L. L. Marostica, B. B. Cota, C. L. M. Lopes, S. F. Andrade, J. D. D. S. Filho, R. J. Alves, M. C. Oliveira, F. C. Braga, C. M. O. Simões, R. M. Pádua and A. L. B. De Barros, ACS Omega, 2019, 4, 22048–22056.
22. I. E. Jehng and J.-M. Wachs, Chem. Mater., 1991, 3, 100–107.
23. D. J. Frisch, M. J. Trucks, G. W. Schlegel, H. B. Scuseria, G. E. Robb, M. A. Cheeseman, J. R. Scalmani, G. Barone, V. Mennucci, B. Petersson, G. A. Nakatsuji, H. Li, M. Caricato, X. Hratchian, H. P. Izmaylov, A. F. Bloino, J. Zheng and G. Sonnenb, Gaussian 09, Wallingford CT, 2009.
24. W. Parr and R. G. Young, Density Functional Theory of Atoms and Molecules, Oxford University Press, New York, 1989.
25. X. Zhang, D. Feng, M. Chen, Z. Ding and Z. Tong, J. Mater. Sci., 2009, 44, 3020–3025.
26. S. Grimme, J. Comput. Chem., 2006, 27, 1788–1799.
27. Y. L. Si and E. B. Wang, Mol. Phys., 2009, 107, 1521–1526.
28. N. R. M. Sturt, J. C. S. Terra, D. L. Sangiorge, L. C. A. Oliveira and F. C. C. Moura, Fuel, 2020, 262, 116417.
29. J. B. G. Filho, L. D. Almeida, H. F. V. Victória, G. H. M. Gomes, K. Krambrock, P. A. Robles-Azocar, M. C. Pereira and L. C. A. Oliveira, J. Catal., 2024, 436, 115580.
30. Y. Hao, Y. Wang, T. Zhang, Y. Liu, Q. Fan, Y. Jiang, Y. Gao, Z. Mao, X. Gu and S. Zeng, Appl. Catal., B, 2024, 340, 123254.
31. C. N. R. McConnell, A. A. Anderson and J. S. Rao, Spectrochim. Acta, Part A, 1976, 32, 1067–1076.
32. I. Nowak and M. Ziolek, Chem. Rev., 1999, 99, 3603–3624.
33. L. C. A. Prado, N. T. Nogueria, F. G. E. Nogueira, A. E. Nunes, C. A. Diniz and R. Oliveira, Energy Fuels, 2010, 24, 4793–4796.
34. S. Q. Xiao and G. N. Man, Chem. Phys. Lett., 2007, 447, 305–309.
35. S. H. A. Nicolai and J. C. Rubim, Langmuir, 2003, 19, 4291–4294.
36. R. M. Naujok, R. R. Duevel and R. V. Corn, Langmuir, 1993, 9, 1771–1774.
37. J. Kwiatkowski, S. Knap, B. Przystupski, D. Saczko, J. Kędzierska, E. Knap-Czop, K. Kotlińska, J. Michel, O. Kotowski and K. Kulbacka, Biomed. Pharmacother., 2018, 106, 1098–1107.
38. G. Habtetsion, T. Ding, Z.-C. Pi, W. Li, T. Lu, C. Chen, T. Xi, C. Spartz, H. Liu, K. Hao, Z. Mivechi, N. Huo, Y. Blazar, B. R. Munn and D. H. Zhou, Cell Metab., 2018, 28, 228–242.
39. M. Raza, M. H. Siraj, S. Arshad, A. Waheed, U. Aldakheel, F. Alduraywish and S. Arshad, J. Cancer Res. Clin. Oncol., 2017, 143, 1789–1809.
40. D. J. Silverstein, R. M. Webster and F. X. Kiemle, Identificação Espectrométrica de Compostos Orgânicos, LTC, 2006.
41. R. S. Fernandes, J. O. Silva, S. V. Mussi, A. Elaine, G. D. Cassali, V. N. Cardoso, D. M. Townsend, M. Patrick, L. A. M. Ferreira, D. Rubello, U. Federal, D. M. Gerais, M. Gerais, U. Federal, D. M. Gerais, B. Horizonte, M. Gerais, E. Therapeutics, L. Angeles and V. T. Martiri, Mol. Imaging Biol., 2019, 20, 437–447.
42. C. Lee, W. Yang and R. G. Parr, Phys. Rev. B: Condens. Matter Mater. Phys., 1988, 37, 785–789.
43. J. A. Hehre, W. J. Radom, L. Schleyer and P. V. R. Pople, Ab Initio Molecular Orbital Theory, John Wiley & Sons, Inc, New York, 1986.
44. I. You, A. Be and M. A. Y. In, J. Chem. Phys., 1985, 82, 270–283.
45. J. Geng, Q. Liu, Q. Ma, P. Wang and J. Niu, Dalton Trans., 2014, 43, 9843–9846.
46. E. F. F. da, C. T. C. R. T. Silva, K. J. de Almeida, M. Dos Santos Pires, A. A. de Castro and M. Aquino Gonçalves, React. Kinet. Mech. Catal., 2017, 120, 195–208.
47. M. J. Stratmann, R. E. Scuseria and G. E. Frisch, J. Chem. Phys., 1998, 109, 8218–8224.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5tb00396b

---