Safe-by-Design of CuO Nanoparticles via Fe-Doping, Cu-O Bond Lengths Variation, and Biological Assessment in Cells and Zebrafish Embryos

Abstract

The safe implementation of nanotechnology requires nanomaterial hazard assessment in accordance with the material physicochemical properties that trigger the injury response at the nano/bio interface. Since CuO nanoparticles (NPs) are widely used industrially and their dissolution properties play a major role in hazard potential, we hypothesized that tighter bonding of Cu to Fe by particle doping could constitute a safer-by-design approach through decreased dissolution. Accordingly, we designed a combinatorial library in which CuO was doped with 1–10% Fe in a flame spray pyrolysis (FSP) reactor. The morphology and structural properties were determined by XRD, BET, Raman spectroscopy, HRTEM, EFTEM and EELS, which demonstrated a significant reduction in the apical Cu-O bond length while simultaneously increasing the planar bond length (Jahn-Teller distortion). Hazard screening was performed in tissue culture cell lines and zebrafish embryos to discern the change in the hazardous effects of doped vs. non-doped particles. This demonstrated that with increased levels of doping, there was a progressive decrease in cytotoxicity in BEAS-2B and THP-1 cells, as well as an incremental decrease in the rate of hatching interference in zebrafish embryos. The dissolution profiles were determined and the surface reactions taking place in Holtfreter’s solution were validated using cyclic voltammetry (CV) measurements to demonstrate the Cu⁺/Cu²⁺ and Fe²⁺/Fe³⁺ redox species playing a major role in the dissolution process of pure and Fe doped CuO. Altogether, a safe-by-design strategy was implemented for the toxic CuO particles via Fe doping and has been demonstrated for their safe use in the environment.

Graphical Abstract

The global consumption of copper (industrial and R&D) has more than tripled in the last 50 years and currently amounts to more than 20 million tons/year. ^1–2 A large proportion of this production volume is in the form of Cu oxide, which is increasingly being used for superconducting, semiconducting, thermoelectric, sensor, ceramics, energy storage, antifouling paints, photoconductive, and photothermal applications, including in the automotive, ship, and aerospace industries. ^3–14 Due to the large scale production and wide range of applications, CuO NPs are likely to come into contact with humans and the environment. ^15–16 The propensity of these materials to dissolve and shed toxic copper ions has been shown to play a key role in the hazard potential of these materials. ^17–24

It is known that the cellular toxicity of CuO NPs is induced through dissolution and release of Cu²⁺ions after their entry to the cellular interior or lysosomes. ^25–26 The Cu²⁺ions are involved in triggering series of downstream adverse effects such as ROS generation and subsequent oxidation of biological components including DNA damage. ^27–29 Increased cellular uptake and the release of Cu ions disrupt metal homeostasis, trigger cell cycle progression, induce metabolic alkalosis or the production of reactive oxygen species (ROS) in the cell through the initiation of Fenton chemistry.^{26, 30–35} The entry of Cu²⁺ ions and the generation of ROS in the brain causes imbalance in the oxidation/anti-oxidation process resulting into neuronal damage, learning and memory impairment.³⁶ The Cu²⁺ toxicity has also been observed during treatment of patients having Wilson’s disease with neurological dysfunctions. The Cu-chelators used in the treatment reduce Cu²⁺ to Cu⁺ and enhance copper accumulation inducing toxicity in the brain. ³⁷ Human exposure to CuO NPs is capable of inducing oxidative stress and acute inflammatory effects in the lungs of exposed welders, giving rise to a condition known as metal fumed fever. ^38–41 Studies on zebrafish embryos also demonstrated that CuO NPs are capable of interfering in embryo hatching as a result of the inhibitory effects of released copper ions on the metalloprotease enzyme, ZHE1, secreted by the hatching gland. ^{20, 24, 42–43} Moreover, the dissolution of CuO NPs has been shown to induce autophagy, which serves as a homeostatic mechanism for protecting the cell against starvation, oxidative and ER stress. ^{38, 44}

Against this background, we hypothesized that changing the dissolution kinetics of CuO NPs by Fe doping may present a safer nanomaterial that has a less hazardous impact by decreasing copper ion shedding. ^{18–20, 45–46} The expected stronger binding interaction between Cu and Fe in the particle matrix could conceptually lead to decreased particle solubility in aqueous biological environments, with the potential to reduce toxicity by a decreased dissolution. ^47–48 A flame spray pyrolysis process was used for doping CuO with 1–10% Fe. The NPs were subjected to toxicological profiling in a human macrophage-like cell line (THP-1), a human bronchial epithelial cell line (BEAS-2B), and zebrafish embryos. ^49–50 The cell lines represent the cell types (macrophages and epithelial cells) that serve as the 1^st line of defense against inhaled NPs in the lung. Similarly, zebrafish embryos are used as a sensitive screening platform for nanomaterial toxicity in the water matrix. ²⁰

RESULTS

Physicochemical characterizations

The ultrafine particles obtained using versatile flame spray pyrolysis were characterized using physicochemical techniques. ^49–50 The specific surface areas and BET primary particle diameters (d_BET) of pure and Fe doped CuO NPs were in the range from 77.6 (±1.8) to 93.6 (±4.5) m²/g and 10–12 nm, respectively. Iron doping had very less influence on particle size.

The crystallite sizes extracted from Rietveld refinement of the XRD patterns are in the range of 9–12 nm (Figure S1 and Table 1), closely agreeing with the BET primary particle diameters. The Fe content in CuO after FSP was determined using energy-dispersive X-ray spectroscopy (EDX), showing 0.2, 2.2, 4.5, 6.3, 8.5 and 9% of Fe at calculated doping levels of 1, 2, 4, 6, 8 and 10% Fe [Figure 1(A) and S2]. The low resolution TEM of pure and/or Fe doped CuO NPs show spherical particles. Pure and Fe doped CuO particles are indistinguishable in morphology with primary particle sizes in the range of 10–20 nm. The crystalline nature of these particles is evident from high resolution images with distinct hexagonal arrangements of the octahedra [Figure 1(B) insets]. The crystal structure of the NPs was studied by probing a single particle with HRTEM imaging. The middle and lower columns in Figure 1(B) show single crystalline particles of pure and/or Fe doped CuO with well-developed lattice fringes. The lattice plane distances of 0.2529 (±0.003), 0.2514 (±0.003), 0.2308 (±0.003), 0.2323 (±0.003), 0.2528 (±0.003), 0.2518 (±0.003) and 0.2545 (±0.003) nm were observed for pure, 1, 2, 4, 6, 8 and 10% Fe doped CuO, respectively. According to these data, 2 and 4% Fe doped CuO NPs have lattice distances of ~0.23 nm while the other particles have ~0.25 nm.

Table 1.

Specific surface area, primary particle size, hydrodynamic size of pure and Fe doped CuO NPs.

Properties/Materials	CuO	1%Fe-CuO	2%Fe-CuO	4%Fe-CuO	6%Fe-CuO	8%Fe-CuO	10%Fe-CuO
Specific surface area (m2/g), primary particle sizes (nm) and crystallite sizes (nm)
SSA (m2/g)	80.9 (±2.5)	79.2 (±3.2)	77.6 (±1.8)	89.6 (±4.2)	92.9 (±3.6)	93.6 (±4.5)	90.4 (±1.2)
dBET (nm)	11.8 (±1.3)	12.0 (±1.5)	12.3 (±1.2)	10.7 (±1.8)	10.3 (±1.6)	10.3 (±1.9)	10.7 (±1.0)
dXRD (nm)	9.4 (±0.1)	11.9 (±0.4)	9.2 (±0.1)	10.5 (±0.1)	10.8 (±0.1)	9.8 (±0.4)	9.6 (±0.9)
Hydrodynamic sizes (nm)
DI water (pH~7)	210 ± 6	284 ± 11	287 ± 2	312 ± 5	270 ± 1	318 ± 8	434 ± 12
Holfreters Medium (pH~7.6)	264.9±7.1	279.3±4.2	456.7±3.0	423.1±9.6	381.4±14.6	450.6±7.1	224.9±3.2
RPMI 1640 medium (pH~7.6)	394 ± 29	388 ± 16	375 ± 12	664 ± 42	388 ± 25	366 ± 12	349 ± 48
BEGM medium (pH~7.6)	366 ± 1	520 ± 59	474 ± 4	477 ± 36	467 ± 48	638 ± 36	546 ± 17
Zetapotential (ξ) mV
DI H2O (pH~7)	41.5 ± 0.8	38.1 ± 0.4	38.4 ± 0.9	37.2 ± 1.1	32.6 ± 3.0	−14.9 ± 3.0	−21.6 ± 0.5
Holfreters Medium (pH~7.6)	−12.63±0.95	−22.74±4.49	−23.16±5.81	−17.63±1.7	−22.58±3.1	−19.41±0.8	−16.45±3.2
RPMI 1640 medium (pH~7.6)	−16.6 ± 1.0	−8.5 ± 2.9	−11.1 ± 2.5	−9.8 ± 3.0	−9.4 ± 0.4	−9.4 ± 2.9	−17.2 ± 4.2
BEGM medium (pH~7.6)	−15.6 ± 1.7	−16.5 ± 3.8	−11.3 ± 3.2	−20.3 ± 5.1	−19.2 ± 5.5	−10.2 ± 6.1	−13.4 ± 6.8
PdI
DI H2O	0.256	0.285	0.321	0.243	0.244	0.195	0.275
Holfreters Medium	0.261	0.182	0.272	0.254	0.206	0.25	0.113

Figure 1.

(A) EDX analyses of the Fe doped CuO NPs. The results show good agreement with the Fe loading during FSP and the amount of Fe present in each sample. (B) Overview and high resolution TEM images of pure and Fe doped CuO NPs.

The XRD patterns and the Rietveld refinement of as-prepared CuO NPs show sharp and well resolved reflections, indicating a monoclinic system (ICSD 69757) with a = 0.46927 nm, b = 3.4283 nm, c = 5.1370 nm, α = γ = 90°, β = 99.546°. Increasing Fe loading, results in an increasing peak shift indicating Fe incorporation into the CuO lattice [Figure 2(A)]. A peak shift of 0.35° is observed for 10% Fe doped CuO across the entire 2θ angles. At 63.62° a minor second phase (< 5% intensity) appears for ≥ 6% Fe doped CuO which agreed with the pure CuFe₂O₄(Figure S3).

Figure 2.

(A) XRD patterns of Fe doped CuO. The shift in the patterns at different 2θ angles shows Fe incorporation within the CuO lattice. The lower XRD intensity with increasing Fe doping supports such incorporation. (B) The a parameter of the CuO crystal structure consistently increases while the b parameter decreases with increased Fedoping.

The postulation of Fe incorporation is based on the following findings: (1) the a parameter of the CuO unit cell consistently increases while b and c parameters decrease with increasing Fe doping [Figure 2(B) and Figure S4] with a linear trend, agreeing with Vegard’s rule ⁵¹; (2) CuO has a square planar geometry with each Cu coordinated to 4 oxygen atoms. ^52–53 The ionic radii of Cu²⁺ in square planar geometry is 0.71 Å, making it unlikely that Cu²⁺ can be substituted with Fe²⁺ having much larger ionic radii (0.77Å). However, Fe³⁺ with lower ionic radii (0.63Å) could be incorporated in the CuO lattice. ⁵⁴ The structural analyses of Fe doped CuO NPs show an increase in the planar Cu-O bond length from non-doped to 2% Fe doped CuO. For iron loading of more than 2%, such increase in the bond length is only marginal. Looking at the apical Cu-O bond lengths, the result is just the reverse to the planar Cu-O indicating strong structural distortion (Jahn-Teller effect) through Fe incorporation [Figure 3(A)]. ^55–56 This observation is also supported by the linear increase and decrease of a and b parameters of the CuO unit cells after Fe doping [Figure 2(B)].

Figure 3.

(A) The structural analyses of Fe doped CuO NPs. The increase in planar Cu-O and decrease in apical Cu-O bond length between 2% and 4% Fe doping shows strong distortion of the octahedra and after 4% such distortion is rather slow. The results show Fe is incorporated in CuO with lattice distortion of almost 4.1 % at higher Fe doping. (B) The energy filtered transmission electron microscopy of 10% Fe doped CuO NPs: (left panel) zero loss image of the experimental probe, (middle panel) image of the Cu mapping and (right panel) image of the Fe mapping of the Fe doped CuO NPs distribution. The results show homogeneous distribution of the Fe in CuO NPs.

Using energy filtered transmission electron microscopy (EFTEM), the results of elemental mapping of the Fe and Cu appeared to be homogeneously distributed throughout the nanoparticles at all Fe loadings [Figure 3(B)].

The room temperature Raman spectrum of reference CuO shows three signals at 291, 340 and 624 cm⁻¹, respectively. ⁵⁹ With the reference to the pure CuO signals, the 291 cm⁻¹ peak of our 10% Fe doped CuO was found to shift by 17 cm⁻¹ towards lower frequencies due to Fe incorporation [Figure 4(A), Figure S5]. Similarly, the characteristic signal of CuO at 340 cm⁻¹ becomes progressively weaker and almost undetectable at 10% Fe, while new signals appear at 442, 514, 685, 813 and 926 cm⁻¹. The intensity of these peaks is relatively constant except for the signals at 813 cm⁻¹ and 926 cm⁻¹ which consistently increase with increasing Fe doping [Figure 4(B)]. For higher Fe content, the characteristic Raman signal of CuO at 291 cm⁻¹ occurs at a lower Raman shift resulting from the increased structural disorder closely agreeing with the XRD analyses [Figure 2(A)]. Xu et al. demonstrated that with a decrease in particle size, the Raman peaks shift toward lower wavenumber and vice versa. ⁶⁰ Furthermore, Chapelle et al. reported Raman signal at 808 cm⁻¹ for CuFe₂O₄ arising from ν (Cu-Fe-O). ⁶¹ The Raman signal (increasing intensity) observed for the doped particles at 813 cm⁻¹ is also assigned to ν (Cu-Fe-O), a clear indication of CuFe₂O₄with increasing iron doping.

Figure 4.

(A) Room temperature Raman shift at 291 cm⁻¹ of CuO with Fe doping. (B) Increasing intensity of Raman signals at 813 and 926 cm⁻¹ are assigned to ν (Cu-Fe-O). (C) Black, red and blue EELS signals of Fe²⁺/Fe³⁺ are adapted from the literature ^57–58 while the green pattern was measured for 10% Fe doped CuO NPs. (D) The comparison of the EELS spectra occurring at 708.8 eV for 1, 4 and 10% Fe doped CuO demonstrating Fe³⁺oxidation state for all the dopings.

The oxidation states of Fe (Fe²⁺ or Fe³⁺) in Fe doped CuO were determined using EELS spectroscopy since multiple oxidation states and coordination modes might be present in Fe doped particles. Materials containing Fe²⁺ or Fe³⁺ are most commonly octahedral-coordinated although tetrahedral coordination of Fe is also known. Usually Fe is expected to be in the high-spin state with strong field ligands causing spin pairing. Fe L_2,3 edges acquired from a number of reference materials containing Fe²⁺ and Fe³⁺ in both octahedral and tetrahedral coordination [chromite (Fe²⁺), siderite (Fe²⁺) hematite (Fe³⁺) and leucite (Fe³⁺)] show different edge shapes, as well as a distinct chemical shift between the divalent and trivalent oxidation states [Figure 4(C)]. ^{57–58, 62} The L₃ edges for the divalent Fe minerals are ~ at 706.4 eV, and the trivalent Fe L₃ edge is 1.7 eV higher at 708.1 eV. The separation of the L₃ and L₂ peak maxima, due to spin orbit splitting is ~13 eV. The EELS spectra of 1%, 4% and 10% Fe doped CuO show Fe L₃ edges occurring at 708.8 eV (d-d transitions of d electrons into the unoccupied 3d-states) agreeing with the trivalent Fe of hematite (Fe³⁺) and/or leucite (Fe³⁺) [Figure 4(D)]. ^{57–58, 62} The spin orbit coupling constant for the Fe doped CuO NPs was found to be 13.1 eV as for hematite and/or leucite [Figure 4(C) and (D)]. ^63–64 The presence of Fe³⁺ is a clear indication of the stabilized Cu–Fe–O phase ⁶⁵ on the CuO nanoparticle surface which prevents dissolution when these particles are exposed to the cellular medium.

Characterization of CuO dispersions and dissolution properties

Hydrodynamic sizes of pure and/or Fe doped CuO NPs were found to be in the range of ~210–434, ~225–457, ~349–664 and ~366–638 nm in DI water, Holtfreter’s Medium, RPMI 1640 medium and BEGM medium respectively. The zeta potential measurements in DI water (at pH ~ 6.5–7.4) indicate that 2–6% Fe doped CuO have surface positive charges while 8% and 10% Fe doped CuO have negative surface charges. A measurement of the pure CuFe₂O₄ NPs showed a zeta potential of −19.6±1.8 mV in DI water, which agrees with the data reported in the literature. ^66–67 The negative surface charge of 8% and 10% of iron in CuO is due to increasing CuFe₂O₄ on the surface, which is evidenced through cyclic voltammetry (CV) discussed in the next section. NP dispersion in Holtfreter’s medium, RPMI 1640 medium and BEGM medium (at pH ~7.6–7.8), resulted in the negative surface charge of the particles, ranging from −8.5±2.9 to −23.16±5.8 mV, which is likely caused by serum albumin in RPMI 1640 medium and alginate that is used as a dispersing agent in Holtfreter’s medium (Table 1). ^19–20 Clear differences in the NP dissolution profiles in Holtfreter’s medium, DI water, and the various cell culture media are shown in Figure 5. While Fe doping generally decreases the amount of dissolved ions, the concentrations were much higher in cell culture compared to Holtfreter’s medium. The dissolved ion concentration in DI water was much lower and almost insignificant compared to Holtfreter’s or the cell culture media [see Figure 5(A) and (B)].

Figure 5.

The experimental results of Cu²⁺ release from 0–10% Fe doped CuO NPs. (A) Dissolution profile of Fe doped CuO in Holtfreter’s medium at 0.5 ppm. The dissolution profile shows Cu²⁺ release is significantly reduced via Fe doping. (B) The release data in RPMI and BEGM cell medium exposed to 50 ppm of NPs show consistent decrease in the Cu²⁺ release while in pure water the dissolution is not significant.

Cyclic voltammetry of pure and Fe doped CuO NPs

The cyclic voltammetry (CV) experiments were conducted in the Holtfreter’s medium for all the samples [Figures 6 and 7(A)–(D)]. The results of the pure CuO showed two anodic signals (A₁ and A₂) occurring at −150 mV and 100 mV, respectively, assigned to redox processes on the CuO surface [Figure 6]. The FTO glass (used as a conducting substrate coated with CuO particles [inset of Figure 6]) itself does not interfere in the Holtfreter’s medium [Figure S6(A)].

Figure 6.

CV of pure CuO in Holtfreter’s solution recorded at 25 mVs⁻¹. The pure CuO showed two anodic peaks, A₁ is related to the corresponding cathodic signal C₁ indicating redox reaction with the carbonate species and A₂ corresponds to Cu⁺/Cu²⁺ redox couple with the reduction occurring below −400 mV.

Figure 7.

Cyclic voltammetry of (A) 10% Fe doped CuO, (B) 33% Fe added CuO, 1–8%Fe doped CuO (only 2^nd cycle) recorded at 25 mVs⁻¹ and (D) 1–8% Fe doped CuO (30^th cycle) recorded at 100 mVs⁻¹. While two new signals A_CuFe2O4 and C_CuFe2O4 were observed for 10% Fe doped CuO, their intensities were highest for 33% Fe doped CuO due to CuFe₂O₄. The quasi-stationary 30^th cycles shows the spinel CuFe₂O₄ formation at the surface and decreasing oxidation of CuO involving sodium bicarbonate (A₁ signal) with increasing Fe.

The interaction of the carbonate species in Holtfreter’s medium with the CuO NPs surface was demonstrated by varying the NaHCO₃ concentration (2.4 mM to 36 mM in the Holtfreter’s medium). The A₁ signal intensity at 36 mM NaHCO₃ is much larger compared to the signal intensity at 2.4 mM NaHCO₃(oxidation reaction involving CO ₃²⁻), while the respective reduction signal C₁ occurs at −200 mV [Figure S6(B)]. The oxidation reaction (A₁) clearly demonstrates the interaction of the CuO particles with the carbonate species. This interaction results in a change in pH, which might be responsible for the dissolution of CuO. ⁶⁸ The second oxidation signal (A₂) is likely due to Cu⁺/Cu²⁺ redox couple triggered through the formation of Cu₂O (reduction of CuO) in the negative slope below −400 mV.^69–70 The negative potential is too low for a complete reduction (limited by the electrochemical window of the FTO glass) and hence, the A₂ signal vanished with increasing number of cycles (Figure 6). At potentials above 400 mV, further oxidation of the CuO occurs in the anodic slope (A₃) with the corresponding reduction signal (C₃) at similar potentials. Both processes are influenced by the change in pH, indicating the role of OH⁻in the redox reaction [Figure S6(C)].

The results of the cyclic voltammetry of the Fe doped CuO NPs showed two additional redox signals A_CuFe2O4 and C_CuFe2O4 [overlapping with C₁, see Figure 7(A)–(D)]. These signals were observed at the potential of 200 mV and −200 mV (oxidation and reduction) due to surface reaction with the spinel CuFe₂O₄.⁷¹ Both signals are clearly visible for the 6–10% Fe doped CuO NPs [Figure 7(A) and (C)]. To demonstrate the presence of this phase (CuFe₂O₄), a new material with high Fe loading (33%) was synthesized with the flame spray pyrolysis. The redox signals of 33% Fe added CuO NPs showed enhanced A_CuFe2O4 and C_CuFe2O4 signal intensities clearly indicating the presence of the CuFe₂O₄ that is corroborated by the characteristic XRD peak of CuFe₂O₄ occurring at (62° 2θ) [Figure 7(B) and S3]. While the first anodic scans from 0 to 600 mV vs. Ag/AgCl for all the particles tested had no spinel oxidation signal, it appeared only after reducing at a sufficiently negative potential in the subsequent scans indicating the presence of CuFe₂O₄ prior to CV cycling. The emerging A_Fe signal was attributed to Fe²⁺/Fe³⁺ redox couple where the intensity increased with increasing Fe [Figure 7(C) and (B)]. To exclude that the Fe²⁺/Fe³⁺ redox couple is from Fe₃O₄, the cyclic voltammetry experiments were conducted for pure Fe₃O₄. In these experiments, the A_Fe signal was not detected [Figure S6(D)]. The quasi-stationary 30^th cycle were recorded with a higher scan rate at 100 mVs⁻¹ to enhance the spinel A_CuFe2O4 and C _CuFe2O4 signals [Figure 7(D)].

While the spinel oxidation signals were only visible after 4% Fe doping, the signal resulting from the copper oxidation involving carbonate was found to decrease and vanish completely for more than 6% Fe loading. The results indicate the inhibition of the reaction involving CO₃²⁻ is due to the presence of CuFe₂O₄ (reduced dissolution) [Figure 5(A)]. In addition, the oxidation process in the anodic region which might contribute to the dissolution of the CuO NPs, is shifted to higher potentials with increase in Fe content suggesting dissolution hindrance due to doping. ⁶⁹

Cytotoxicity, autophagy and embryo hatching

Previous studies demonstrated that the intra and extracellular dissolution of CuO particles is responsible for biological outcomes such as cytotoxicity and autophagy. ^{24, 31, 72} We asked, therefore, whether a change in nanoparticle dissolution characteristics of the NPs could impact these responses in mammalian cells.

To test the hypothesis, the bioresponse of pure and doped CuO NPs were assessed in THP-1 and BEAS-2B cells (representative of the responses for macrophages and bronchial epithelial lung cells). ¹⁹ While non-doped CuO NPs induced dose-dependent cytotoxicity in BEAS-2B cells, iron doping resulted in a significant reduction in cell death [Figure 8(A) and (B)]. This protective effect was incremental with the level of doping, such that doping at >6% resulted in the abrogation of cell death at a particle concentration of 100 μg/mL. However, a physical mixture of CuO and 10% Fe₃O₄ NPs showed similar biological response (extent of cell death) compared to non-doped particles [Figure 8(A) and (B)] clearly suggesting the efficient doping effect during flame spray pyrolysis. This indicates that the doping rather than physical mixture plays a key role in reducing CuO cytotoxicity. Albeit less prominent, iron doping resulted in similar protective effects in THP-1 cells [Figure 8(C)–(D)]. In addition, the cell death was minimal when the same concentration of CuCl₂ salt was exposed to the cellular system studied compared to the CuO NPs [Figure 8(B) and (D)] reasonably agreeing with the literature. ^{25, 73}

Figure 8.

Iron doping reduced the toxicity of CuO NPs in BEAS-2B (panels A and B)) and THP-1 cells (panels C and D). BEAS-2B or THP-1 cells were incubated with 0–100 μg/mL (0–100 ppm) (A and C) or 50 μg/mL (B and D) pure or doped CuO NPs for 24 h. The cell viability was determined by MTS assay exposed to 50 μg/mL (50 ppm) of NPs.

The effects of Fe doping on autophagy, a stress-induced homeostatic response was also studied using the autophagosome assembly marker, LC-3II, which can be visualized by confocal microscopy in THP-1 cells that are stained with FITC-labeled anti-LC-3II antibodies. ⁴⁰ As shown in Figure 9, non-doped CuO NPs induced autophagosome formation, which was significantly less prominent in cells treated with 10% Fe doped CuO NPs. Chloroquin (CQ) was used as a positive control. The quantification of cellular content of Cu and Fe in BEAS-2B and THP-1 cell lines exposed to 50 ppm of 0–10% Fe doped CuO after 24 h using ICP-OES showed increased cellular Cu and Fe uptake with the increase in Fe doping [see Figure 10(A)–(E)]. The similar experiment using copper salt (CuCl₂) showed no Cu²⁺ uptake. Flow cytometric analysis (early and late cellular apoptosis/necrosis) was used to study the apoptotic effects of 50 ppm 0–10% Fe doped CuO and CuCl₂ in BEAS-2B cells at 24 h by using Annexin V-FITC and propidium iodide (PI) double-labeling technique (see Figure S7). Results showed doping effectively reduced early and late apoptosis as well as increased live BEAS-2B cells. Similarly, flow cytometric analysis of the apoptotic effects at 50 ppm of 1–10% Fe doped CuO and CuCl₂ in THP-1 cells also indicated reduced early and late apoptosis/necrosis (see Figure S8). Despite higher cellular uptake of Fe-doped CuO, there were no increase in the cell death; on the contrary, the cell death decreased as the doping percentage increases. This strongly suggests that reduced particle dissolution by Fe doped CuO decreased CuO-induced cytotoxicity. Taken together, iron doping could greatly reduce the cell death as well as autophagosome accumulation.

Figure 9.

(left panel) Accumulation of autophagosomes as determined by fluorescent staining with FITC-labeled anti-LC-3II. The images show extensive autophasosome induction by 50 ppm chloroquin (CQ) as a result of its effects of interfering in autophagic flux. CuO had the same effect on autophagosome accumulation, which disappeared for 10% Fe doped CuO (50 ppm). (right panel) Quantification of the Cu²⁺ by exposing CuO and 10%Fe doped CuO in the cells.

Figure 10.

(A)–(D) The quantification of Cu and Fe in BEAS-2B and THP-1 cells using ICP-OES after incubating with 0–10% Fe doped CuO NPs and CuCl₂ (control) for 24 h at 50 ppm. (E)–(F) Results of the early apoptosis, late apoptosis/necrosis, and live BEAS-2B and THP-1 cells demonstrating effective reduction of apoptosis via Fe doping.

Furthermore, results showed ~50% late apoptosis and 25% early apoptosis induction via pure CuO exposure in BEAS-2B cells, while 1%–10% Fe doped CuO induced less than 40% late apoptosis and 10 % early apoptosis [see Figure 10(D)]. The percentage of live cells increased from 25% for CuO to more than 60% for 10% Fe doped CuO. THP-1 cells showed similar results as BEAS-2B cells, especially for the effects onearly apoptosis [see Figure 10(E)].

It was previously demonstrated that ionic dissolution of specific transition metal oxide NPs, including CuO, can interfere in embryo hatching due to an effect on the metalloprotease hatching enzyme, ZHE1.^19–20 Figure 11(A) and (B) show the effects of pure and doped NPs on embryo hatching. The data clearly show the reduction of the hatching interferences through Fe doping at low concentrations [Figure 11(B)].

Figure 11.

(A) Dose-dependent effects of non-doped and doped (1%, 6% and 10% Fe) CuO on zebrafish embryo hatching (%). Embryo hatching was studied in 96 well plates with automated high content imaging, demonstrating progressive hatching interference by CuO at doses ≥ 0.25 ppm. While Fe doping lead to reduced hatching interference at 0.25, 0.5 and 0.625 ppm CuO, the doping effect disappeared at a nanoparticle concentration ≥1.25 ppm. (B) Representative light optic images obtained in the experiment.

DISCUSSION

Almost a decade ago, Karlsson et al. reported cellular toxicity of CuO NPs via dissolution. They observed Cu²⁺ released from CuO NPs were much more toxic than Cu²⁺ ions from CuCl₂ salt. ^{25, 73} The CuO NPs passing through the cell membrane penetrate lysosomes and release Cu²⁺ ions via Trojan-horse mechanism. ²⁶ Similarly, two systematic studies on CuO dissolution and the involvement of Cu²⁺ in triggering series of adverse effects on algal growth showed Cu²⁺ shedding a key factor for toxicity induction. ^{27, 74} They suggested Cu²⁺ internalization generates ROS which activates superoxide dismutase gene for unspecific oxidation of various biomolecules including DNA. The simultaneous determination of ROS, NP uptake and their correlation with genotoxicity after exposing the cells with various nanoparticles such as ZnO, CuO, Fe₃O₄, TiO₂, and Ag showed the former two (ZnO and CuO) produced high ROS triggering phosphorylated histone γ-H2AX, a signal for DNA damage. ^28–29 The toxicity of CuO NPs on the cognitive functions of rats caused learning and memory impairment due to imbalance of oxidation/anti-oxidation homeostasis. ³⁶ In addition, Cu²⁺ ions are known to regulate genes associated with Mitogen-Activated Protein Kinase (MAPK) pathway, cell cycle progression, DNA damage human lung cells apoptosis indicating the role of Cu²⁺ in adverse biological responses. ^{38, 73, 75} The most significant toxic result was observed during the stabilization of neurological functions in the patients with Wilson’s disease when excess copper was chelated. These Cu-chelators were found to reduce Cu²⁺ to Cu⁺ and enhance copper accumulation as well as induce brain cells toxicity. They suggested the attempts of rescuing brain cells against copper-induced ROS exploiting Cu-chelators are highly risky. ³⁷ In very recent reports, Hedberg et al. showed cellular membrane damage due to NP corrosion which helped accelerate dissolution. ⁷⁶ In addition, the dissolution of CuO NPs caused specific developmental effects in sea urchin embryos due to disruption in the aboral-oral axis related to altered gene expression induced by changes in the redox environment. ⁴³ Similar to these investigations, the flame sprayed CuO NPs in the present investigation were found to be toxic to THP-1 and BEAS-2B cells as well as hatching interference in zebrafish embryos. While extensive use of CuO based NPs in the industry and R&D world-wide is a clear indication for the potential exposure of the particles, there is an urgent need to re-design CuO NPs to increase safety margin when these aerosol CuO NPs are deliberately/undeliberately exposed to the environment. Hence, the CuO nanoparticles were doped with Fe to reduce the dissolution and to increase such safety margin.

Safe-by-designed CuO based nanoparticles

The CuO NPs were doped using in-house build flame spray pyrolysis (FSP), a key synthesis process to re-engineer CuO NPs at elevated temperatures. The Fe doping at such high temperatures in symmetrical CuO results in structural distortion producing stable material (Jahn-Teller effect). ^77,78 The XRD analysis of the Fe doped CuO shows inward movement of the trans oxygen ligands by 2.19 Å for pure CuO and 2.05 Å for 10% Fe doped CuO while the planar oxygen ligands moves outward with 2.02 Å for pure CuO and 2.19 Å for 10% Fe doped CuO making almost 6.4% of plane elongation and apparent distortion of 4.1% (Figure 3). Such lattice distortion explains the material stability due to the formation of CuFe₂O₄on the surface. ⁶⁵

The exposure of these pure and Fe doped particles in test models (THP-1 and BEAS cells as well as zebrafish embryos) showed reduced Cu²⁺ release from the doped particles and almost no release for 10% Fe doped CuO NPs. The dissolution of the CuO NPs in Holtfreter’s medium is likely triggered by a redox reaction taking place on the CuO surface in presence of sodium bicarbonate. Hence, the use of bicarbonate species as an electron acceptor in the biological medium triggers increased pH and enhances further dissolution.⁶⁸ The reduced dissolution of Fe doped CuO is due to the formation of a stable CuFe₂O₄ phase at the surface of the Fe doped CuO NPs. The insoluble CuFe₂O₄ (supported by our extensive physicochemical characterizations and the cyclic voltammetry, Figure 6 and 7) on the Fe doped CuO prevents Cu²⁺ release in the cellular interior. ^79–80 While the redox signals only appeared after reducing the spinel, it is obvious that the spinel is already present.

The biological response assessment in THP-1 and BEAS-2B cells, showed dose-dependent cytotoxicity for CuO NPs, while CuO NPs with >6% Fe had protective effects even at the highest dose (100 μg/mL) [Figure 6(A) and (B)]. Similarly exposure of CuO based library in zebrafish embryos clearly showed the reduction of the hatching interferences with respect to Fe doping. The striking observation is that the physical mixture of CuO and 10% Fe₃O₄ showed no decrease in apoptosis compared to pure CuO NPs confirmed acquiring autophagy response for 10% Fe doped CuO. Hence, pure CuO NPs were toxic to all the test models and the reduced dissolution (increased formation of CuFe₂O₄ spinel) via Fe doping was proportional to the reduction in toxicity and/or hatching interference clearly indicating the importance of safe designed CuO particles before they can be released in the environment(Figure 12).

Figure 12.

Structural-activity-relationship describing the structural and chemical properties of pure and Fe doped CuO NPs and their influence to the biological response assessment.

The dissolution driven cytotoxicity of pure ZnO in cell models, zebrafish, rat, mouse (using high throughput system) and variety of organisms at different trophic levels in the environment, including bacteria, algae, crustaceans, ciliates, yeast, fish, and nematodes was previously demonstrated. ^{23, 46, 81–83} The biological response was then delineated by engineering ZnO NPs via Fe doping. ^{46, 84–86} In the present investigation, the decreased toxicity at cellular level and the reduced hatching interference in the zebrafish embryo (safe-by-design strategy) is established for Fe doped CuO NPs (Figure 12). The role of physicochemical properties of pure and Fe doped CuO playing a decisive part in NP toxicity (particle dissolution and Cu²⁺ release) is demonstrated. Although Fe doping was considered to be a key strategy to reduce ZnO toxicity in the environment,^{46, 87} there are major differences in the toxicity reduction mechanism between Fe doped CuO and/or ZnO. In the case of Fe doped CuO NPs, the Fe incorporation has two effects: (1) stabilizes the CuO crystal lattice through distortion (4.1% when 10% Fe is doped in CuO) Considering d⁹ configuration for CuO, Cu²⁺ finds itself in the center of the six octahedral oxygen ligands (2 in the apical positon and 4 in the planar position) [see Figure 3(A)]. When Fe is doped in CuO, one of the d sub-orbitals of Cu is vacant making the electronic states of e_g orbital degenerate, resulting in a Jahn-Teller distortion. ^77–78 (2) formation of surface CuFe₂O₄ spinel (only detectable for higher Fe loading). The stable material and increasing surface CuFe₂O₄ spinel with increase in doping prevents excess Cu²⁺ release to significantly mitigate the cytotoxicity. However, the stabilization in Fe doped ZnO implies Fe²⁺ to be more strongly bound than Zn²⁺ in the crystal structure. For Fe doped ZnO, the central metal Zn has d¹⁰ configuration and cannot undergo Jahn–Teller distortion. From the earlier investigation, it is known that there are no detectable structural differences even after loading 10% of Fe due to the possibility of doping Fe beyond the solubility limit in the FSP process. ²² Taken all together, doping does offer a safer alternative for toxic CuO NPs when they enter the living system, but still requires careful investigation with respect to the specific application and model organisms.

CONCLUSION

In the present investigation pure and Fe doped CuO NPs homologues library has been designed using in-house built flame spray pyrolysis. The extent of dissolution has been significantly lowered by incorporating Fe in CuO NPs. Cyclic voltammetry demonstrated the formation of CuFe₂O₄ at the surface for Fe concentrations above 6%. With an increase in Fe doping, the oxidation reaction of sodium bicarbonate a probable cause for the CuO dissolution, is strongly reduced. The doped particle exposed to zebrafish embryos showed reduced interference of ZHE1 enzyme and enzyme activity (improved safety profiles) due to reduced dissolution. The biological response assessment in pure CuO exposed THP-1 and BEAS-2B cells showed dose-dependent cytotoxicity while CuO doped with ≥6% Fe showed protective effects even at high concentration. Altogether, the data suggested Fe doped CuO could serve as an example for extending safer-by-design rule to many other particles (posing hazard potential to the human and the environment)used in commercial applications.

METHODS

Synthesis of pure and/or Fe doped CuO NPs

Flame spray pyrolysis was used for the production of ultrafine powders of pure and/or Fe doped CuO NPs. ^{22, 49} The synthesis was performed using copper naphthenate (Strem Chemicals, 99.9% pure) with/without iron naphthenate (12% Fe by metal, Strem Chemicals, 99.9% pure). For Fe doping in CuO, the required amount of iron naphthenate was mixed with the copper naphthenate solution before combustion. For instance, the synthesis of 10% Fe is calculated as: %Fe = Weight of Fe in 5.7 mL of a 0.5M solution/(Weight of Fe in 5.7 mL of 0.5M solution + weight of Cu in 50 mL of a 0.5M copper naphthenate solution). All precursors were diluted with xylene (Strem Chemicals, 99.95% pure) to obtain a total metal concentration of 0.5 M. To synthesize the complete library of Fe doped CuO, a 50 mL portion of 0.5 M copper naphthenate was separately mixed with 0.57–5.7mL of 0.5 M iron naphthenate to obtain 1–10 %Fe in CuO. During synthesis, the liquid precursor was delivered to the two-phase nozzle at the rate of 5 mL/min using a syringe pump and was atomized with oxygen at flow rate of 5 L/min O₂ (pressure drop of 1.5 bar at the nozzle tip). ⁸⁸ The spray was ignited with a premixed flame of CH₄ and O₂ flowing at a rate of 1.5 and 3.2 L/min, respectively. The ultrafine particles were collected from the 257 mm glass filters placed at a distance of 60 cm from the nozzle. ^{50, 83, 89}

X-ray diffraction (XRD) and Rietveld analysis

The in-house synthesized pure and/or Fe doped CuO NPs were placed in circular sample holders (diameter of 16 mm) and loaded in Bruker D8 diffracting system. The diffractometer was configured in Bragg-Brentano geometry equipped with a primary Johansson monochromator producing Ni-filtered Cu-K_α (λ=0.154 nm) radiation. A 0.1° fixed divergence, 4° primary, 2.5° secondary soller slits, and LynxEye detector (position sensitive in a range of 3° 2θ with 192 channels, yielding a channel width of 0.01563° 2θ) was used. Continuous scans were applied with an integration step width of 0.03° 2θ and 30 s per step. The structural and microstructural parameters were extracted using Rietveld refinement by applying BRASS program. ⁹⁰ Background, scale factor, unit cell parameters, Gaussian as well as Lorentzian peak width parameters were simultaneously refined to obtain the crystallite size. For the pattern refinement, the structural model for CuO (ICSD-69757) with monoclinic space group C1C1 [a =4.6927(4), b=3.4283(4), c=5.1370(6), β=99.546(9)] was used. The quality of Rietveld refinement was evaluated in terms of the usual R factor (R_wp) and the background corrected residual R’_p. ⁹¹ A volume weighted average crystallite size (d_XRD) and the root-mean-square lattice micro strain for each of the doped CuO samples were determined. The instrumental contribution to the peak broadening was taken into account using instrumental parameters derived from standard crystalline LaB₆.

Brunauer-Emmett-Teller (BET) measurements

BET measurements were carried out at 77K using a Quantachrome NOVA 4000e Autosorb gas sorption system to determine the specific surface areas of the samples. The powders were placed in a test cell and allowed to degas for 2 hours at 200°C in flowing nitrogen to remove water vapour and adsorbed gases from the samples. Data were obtained by exposing or removing a known quantity of adsorbing gas in or out of a sample cell containing the solid adsorbent maintained at the constant liquid nitrogen temperature. BET surface area measurement is related to the average equivalent primary particle size as d_BET = 6000/(ρ·S_A), where d_BET is the average diameter of the spherical particles (nm), S_A the measured specific surface area (m²/g), and ρis the theoretical density in g/cm^{3. 49}

Transmission Microscopy (TEM) imaging

For the TEM specimen preparation, a small portions of the FSP-generated pure and/or Fe doped CuO powder (~1–2 mg) were dispersed in 5mL of ethanol (Strem Chemicals, AR grade) in an ultrasonic bath and sonicated for 15 minutes. A drop of the dispersed sample was placed on a nickel grid coated with carbon film (Cu grid was not used as it would interfere with the Cu in the sample during extraction of the quantitative elemental data). The samples were dried at RT followed by scanning the large regions of the grid. The low and high resolution TEM of the sample were examined using transmission electron microscopy (TEM) on a FEI Titan 80/300 microscope equipped with a Cs corrector for the objective lens, a Fischione high angle annular dark field detector (HAADF), GATAN post-column imaging filter and a cold field emission gun operated at 300 kV as an acceleration voltage. For the electron energy loss spectroscopy (EELS) measurement, the Gatan parallel EELS spectrometer was operated at 0.2 eV per channel as an energy dispersive detector. Imaging was performed at each sampling point in order to identify the homogeneity of the samples.⁹² EFTEM images of pre and post edges of elemental mappings were obtained from the data by energy calibration and background subtraction. Zero loss and low loss spectra were also taken at each sampling point for energy calibration using the energy of the zero loss, low loss or core loss at the lower energy.

Raman measurements

The Raman spectra were collected using pressed powder samples on glass plates (~5 mm diameter). The spectra were recorded on a Horiba LabRam Aramis spectrometer equipped with a solid state Nd-YAG laser with wavelength centered at 532 nm. Data were collected by scanning the wide spot in the range of 50–2000 cm⁻¹ with a spectral resolution of ~3.2 cm⁻¹ using a grating of 1800 grooves/mm.

Nanoparticle dissolution studies

To determine particle dissolution in biological media, particles were dispersed at 50 ppm in cell culture medium and at 0.5 ppm in Holtfreter’s medium for 0, 24 and 48 h. After vortexing of the suspensions and centrifugation at 15 000 rpm for 1 h, supernatants were transferred to clean tubes for acid digestion. Digestion was carried out with concentrated nitric acid (10 mL of HNO₃, 65–70%, Trace Metal grade) at 80 °C for 6 h in a HotBlock (SC100, Environmental Express). The temperature was increased to 95 °C to evaporate the remaining liquid. The samples were cooled to room temperature and resuspended in 2% nitric acid at 80 °C for 3 h. The resulting solutions were transferred to 15 mL ICP-OES analysis tubes and brought to a final volume of 8 mL through the addition of nitric acid. ICP-OES analysis was carried out in a Shimadzu ICPE 9000 to determine the elemental Cu concentration. A calibration curve was established using a standard Cu solution (Elements Inc., 1000 mg/L in 2% HNO₃). Each sample and standard was analyzed in triplicate in the presence of 2% (v/v) nitric acid.

Electrochemical characterization using cyclic voltammetry (CV)

The CV measurements were carried out in a three-electrode system consisting of an Ag/AgCl (3M KCl, SI Analytics) reference electrode, a Pt-coated grid as counter electrode and the working electrode. The working electrodes (8 cm²) were prepared using double-roll laminator (Hot Roll Laminator HL-101, Cheminstruments) where the particles from FSP collecting unit were directly transferred (glass fiber filter) to the conductive FTO (fluorine doped tin oxide) glass substrates (7 Ω cm⁻², Sigma Aldrich). The dry particle transfer process enables crack-free coating of nanoparticle layers on the electrodes. ⁹³ After transferring the particles with a pressure of 3.2 MPa, a subsequent heat treatment at 350 °C for 6 h stabilizes the particles on the FTO surface. The CV’s were recorded with a potentiostat (IMP83, Jassile) at a scan rate of 25 and/or 100 mVs⁻¹ at a fixed potential range of −600 to 600 mV vs. Ag/AgCl using Holtfreter’s solution as an electrolyte. Prior to the measurements, the electrolyte was purged with N₂ for 30 minutes. The experiments were carried out at a constant temperature of 22±1°C.

Preparation of NPs in cell culture media

CuO NPs were suspended in distilled, deionized H₂O at a concentration of 5 mg/mL. A water sonicator (Branson, Danbury, CT, USA, model 2510) was used to sonicate these suspensions at 100 W output with a frequency of 42 kHz for 15 min. The suspensions were used as stock solutions for further dispersion in RPMI1640 or BEGM media. An appropriate amount of each stock solution was added to cell culture media to achieve the desired final concentration. For RPMI 1640, the media were supplemented with 10% fetal bovine serum (FBS). For BEGM, BSA was added to the medium at 0.2 mg/mL, before the addition of CuO NPs to achieve a high dispersion. ^{19–20, 23–24} The diluted CuO suspensions in cell culture media were dispersed using a sonication probe (Sonics & Materials, USA) at 32 W for 15 s before exposure to the cells.

Cytotoxicity Assessment

THP- 1 and BEAS-2B cells, purchased from ATCC (Manassas, VA), were cultured in RPMI 1640 medium supplemented with 10% FBS (c-RPMI 1640 and BEGM, respectively, at 5% CO₂ and 37 °C). Before exposure to the CuO NPs, BEAS-2B cells were seeded at a density of 2×10³/well in 96-well plates (Corning, NY, USA) at 37 °C overnight. For THP-1 cells, aliquots of 5×10⁴ THP-1 cells were seeded in 0.1 mL c-RPMI 1640 medium with 1 μg/mL phorbol 12- myristate acetate (PMA) overnight in 96-well plates (Corning, NY, USA). All the CuO suspensions were freshly prepared in BEGM with 0.2 mg/mL BSA or c-RPMI 1640 before addition to the BEAS-2B and THP-1 cells (120 μL/well), respectively. After 24 h treatment, the cell culture medium was removed by aspiration. Each well received 100 μL of culture medium containing 16.7% of MTS stock solution for 1 h at 37 °C. The plate was centrifuged at 3000 g for 5 min in NI Eppendorf 5430 with a microplate rotor to spin down the cell debris. A 100 μL amount of the supernatant was transferred into a new plate. The absorbance of formazan in culture media was read at 490 nm on a SpectraMax M5 microplate reader (Molecular Devices Corp., Sunnyvale, CA, USA).

Confocal imaging of autophagosomes

BEAS-2B cells were seeded in a 8-well chamber slide at 2×10^4/well. After exposure to pure or Fe doped CuO NPs at 25 μg/mL for 24 h, the cells were fixed in 4% paraformaldehyde for 30 min and permeabilized by 0.2% Triton X-100 for 15 min. Permeabilized cells were subsequently incubated with primary LC-3II antibodies at 4 °C overnight, followed by washing and the addition of secondary antibody or Hoechst for 2 h at room temperature. ⁴⁰ After washing, the cells were visualized under a confocal microscope (Leica Confocal SP2 1P/FCS) in the UCLA/CNSI Advanced Light Microscopy/Spectroscopy Shared Facility. High magnification images were obtained with the 63 X objective.

Zebrafish embryo nanoparticle exposure

Wild type adult zebrafish (Danio rerio) were housed and maintained in the UCLA zebrafish facility. Two pairs of male/female fish were placed in a single tank one day ahead of time and released the next morning to trigger spawning. The embryos were collected at 2 hour post fertilization (hpf) and rinsed with Holtfreter’s medium to remove any residue from the embryo surface. Subsequently, fertilized embryos were robotically pick-and-placed into 96-well plates with one embryo in each well. Exposure started at 4 hpf by adding 100 μL of the nanoparticle suspension into each well of the multitier plates. Twelve replicates for each treatment group were used to allow statistical analysis. The hatching rate and survival percentage were assessed at day 3 of embryo development and used as toxicity endpoints. Bright-field high content imaging (ImageXpress) was performed to capture high resolution microscopic images of embryos as previously described. ^19–20 All procedures were carried out in accordance with the Animal Care and Use Committee guidelines at UCLA.