A paradox: Fe2+-containing agents decreased ROS and apoptosis induced by CoNPs in vascular endothelial cells by inhibiting HIF-1α

Abstract Cobalt nanoparticles (CoNPs) released from hip joint implants are known to have a toxic effect on several organs probably through increasing reactive oxygen species (ROS). Ferrous ion (Fe2+) is well-known to enhance oxidative stress by catalysing the production of ROS. However, in our pilot study, we found that Fe2+ conversely inhibited the ROS production induced by CoNPs. To elucidate the underlying mechanism, the present study treated vascular endothelial HUVEC and HMEC-1 cells with CoNPs alone or in combination with ferrous lactate [Fe(CH3CHOHCOO)2], ferrous succinate [Fe(CH2COO)2], and ferrous chloride (FeCl2). CoNP toxicity was evaluated by measuring cell viability, rate of apoptosis and lactose dehydrogenase (LDH) release, and intracellular ROS levels. Treatment with CoNPs decreased cell viability, LDH release, and ROS production and increased apoptosis. CoNPs increased hypoxia-inducible factor-1α (HIF-1α) protein level and mRNA levels of vascular endothelial growth factor (VEGF) and glucose transporter 1 (GLUT1) downstream of HIF-1α signalling. Silencing HIF-1α attenuated CoNP toxicity, as seen by recovery of cell viability, LDH release, and ROS levels and reduced apoptosis. CoNPs caused a pronounced reduction of Fe2+ in cells, but supplementation with Fe(CH3CHOHCOO)2, Fe(CH2COO)2, and FeCl2 restored Fe2+ levels and inhibited HIF-1α activation. Moreover, all three Fe2+-containing agents conferred protection from CoNPs; Fe(CH3CHOHCOO)2 and Fe(CH2COO)2 more effectively than FeCl2. In summary, the present study revealed that CoNPs exert their toxicity on human vascular endothelial cells by depleting intracellular Fe2+ level, which causes activation of HIF-1α signalling. Supplements of Fe2+, especially in the form of Fe(CH3CHOHCOO)2 and Fe(CH2COO)2, mitigated CoNP toxicity.


Introduction
Cobalt-chromium (CoCr) alloy casting implants have been extensively used in total hip arthroplasties, especially in younger and more active patients. Compared with early metal-on-polyethylene bearing implants, CoCr metal-on-metal (MOM) resurfacing implants improve wear characteristics, thereby reducing wear-induced osteolysis, the main cause of loosening and dislocation of prostheses [1,2]. However, numerous studies have demonstrated that CoCr implants release a large amount of very small wear particles and metal ions [1,2]. Particles smaller than 50 nm in size have been reported. CoCr alloy is composed of 62% cobalt (Co) and 28% chromium (Cr). Cobalt nanoparticles (CoNPs) are the most common degradation products of MOM implants. Once the diameter of CoCr materials reaches the nanoscale, the nature culture medium was sonicated for 15 min using ultrasonic oscillators (Ningbo Sklon Lab Instrument Co., Ltd; Shanghai, China) to completely disperse CoNPs in the culture medium. The CoNPs sample was reliable and effective until CoNPs were like fine sand uniformly deposited on the bottom of the sample well. To ensure reliable repeatability, three replicate wells were set for each concentration. Based on the results from our pilot experiments, concentrations of CoNPs ranging from 0-1000 μM were selected.

Electron microscope observation
Cell pellets were fixed in 0.1 M phosphate buffer and 2% glutaraldehyde and embedded in epon resin with Epoxy Embedding Medium Kit (Sigma), following instructions from the kit provider. The sections were analysed with a Field Emission Gun-Environmental Scanning Electron Microscope (Quanta 200, FEI Company, Netherlands) in STEM mode.

Treatment with ferrous agents
HUVEC and HMEC-1 cells were cultured in the cell medium containing CoNPs. Various concentrations of Fe(CH 3 CHOHCOO) 2 , Fe(CH 2 COO) 2 , and FeCl 2 were added to the cells. Cells went through all the measurements 24 h after the treatments were done.

Cell viability assay
The effect of CoNPs on the viability of vascular endothelial cells was assessed using the MTT assay. HUVEC and HMEC-1 cells were plated into 96-well culture plates (5 × 10 3 cells per well) and then exposed to CoNPs. Culture medium served as the control in each experiment. Optical density was measured at 570 nm using a microplate reader.

Lactose dehydrogenase release assay
A lactose dehydrogenase (LDH) assay kit (Beyotime Biotechnology, Shanghai, China) was used to test LDH release into culture medium. The method was according to the manufacturer's protocol. The samples were assessed using a microplate reader.

Apoptosis assessment using flow cytometry
An apoptosis detection kit provided by Beyotime Biotechnology was used to detect and quantify apoptosis in vascular endothelial cells. Briefly, cells were trypsinised and resuspended at a concentration of 1 × 10 6 /ml in diluted binding buffer and labelled with 10 μl of annexin V-FITC. Cells were incubated for 30 min at room temperature in dark, followed by a 5-min incubation with 5 μl of PI. Subsequently, 400 μl of 1× binding buffer was added to each tube. Flow cytometric analysis was performed to monitor the annexin V and DNA-bound PI. Data acquisition and analysis were performed using the FlowJo software.

Hoechst33258 and PI staining
Cell apoptosis was also observed by a dual staining with Hoechst33258 (Beyotime Biotechnology) and PI. HUVECs were stained with 5 mg/l Hoechst 33258 (Sigma) for 30 min at 37 • C, followed by a 5-min incubation with 5 μl of PI. The cells were visualised under a fluorescence microscope with standard excitation filters.

Detection of intracellular ROS, lipid hydroperoxide and malondialdehyde
Intracellular ROS levels were evaluated using 2 ,7 -dichlorofluorescin diacetate (DCFHDA) (Beyotime Biotechnology). DCFHDA can form the fluorescent compound dichlorofluorescein in the presence of ROS. DCFH-DA (10 μM) was added to cells. After the incubation for 20 min at 37 • C, the cells were washed using MEM without serum, at least five times. Labelled cells were trypsinised, resuspended in PBS supplemented with 5% FBS, and analysed by flow cytometry (Fortessa, BD Biosciences). A minimum of 10000 cells were analysed per condition.
Lipid peroxidation causes the formation of the highly unstable and reactive intermediary product, LPO, and the end product, malondialdehyde (MDA). LPO level was measured using a Lipid Hydroperoxide Assay Kit (Caymen Chemical Co., Shanghai, China). LPO was initially extracted from the cell sample to avoid the interference of intracellular iron ions in the assay. LPO then reacted with a Fe 2+ agent provided by the kit to yield Fe 3+ . Fe 3+ further reacted with thiocyanate anion (SCN − ) to produce a purple-coloured Fe(SCN) 5 2− product that can be quantified spectrophotometrically at 500 nm. The level of LPO was expressed as nmol/g protein. For MDA measurement, cell sample were lysed and then heated in boiling water bath for 15 min, followed by thawing on ice, and the optical density of the solution was measured at 532 nm. The results were expressed as nmol MDA/mg protein.

Iron assay
Fe 2+ and total iron levels in cells were measured using an iron assay kit (ab83366, Abcam). Briefly, samples were homogenised in 5× volumes of iron assay buffer on ice and then the supernatant was collected. Iron buffer and iron reducer were added respectively for testing Fe 2+ and total iron levels. Consequently, the iron probe was added to each sample before mixing and incubating for 60 min. Optical density was immediately measured at 593 nm on a colorimetric microplate reader.

Short interfering RNA transfection for HIF-1α silencing
The vascular endothelial cells were seeded in six-well plates at a density of 5 × 10 6 cells/well, 16 h before transfection. Cells were transfected with three pairs of HIF-1α short interfering RNAs (siRNAs) (50 nM, GenePharma, Shanghai, China), using Lipofectamine 2000 (Invitrogen). A scrambled siRNA was used as control. Six hours post-transfection, cells were cultured in fresh growth medium with or without treatment with CoNPs. Cells were then harvested for detection of HIF-1α mRNA and protein levels by real-time polymerase chain reaction (PCR) and Western blotting.

Quantitative real-time PCR
mRNA was extracted from the HUVEC and HMEC-1 cells using the TRIzol-based method. The extracted mRNA (approximately 1 μg) was reverse-transcribed into the first-strand of cDNA using RevertAid First Strand cDNA Synthesis Kit (Thermo Fischer Scientific, Waltham, Massachusetts, U.S.A.). By using SYBR ExScript RT-PCR kit (TaKaRa, Dalian, China), the synthesised cDNA was used for quantitative real-time PCR on an ABI 7300 Real-Time PCR System (Applied Biosystems, Foster City, CA, U.S.A.). All gene expression data are normalised to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) expression levels. The reaction conditions were as follows: initial 94 • C for 30 s, followed by 35 cycles at 94 • C for 5 s, 58 • C for 15 s, 72 • C for 10 s, and final extension at 72 • C for 10 min. Specific primers are shown in Table 1.

Western blot
Cells were lysed on ice for 30 min using a lysis buffer (Thermo Fischer Scientific). The supernatants of the lysates were denaturation and then separated on a 4-12% Bis-Tris gel (Invitrogen). Proteins in gel was next transferred to a nitrocellulose membrane, and probed with anti-HIF-1α and anti-GAPDH primary antibodies (Abcam) overnight at 4 • C. Membrane-bound primary antibodies were detected using the appropriate secondary antibodies. Equal loading of protein was ensured by measuring GAPDH expression.

Statistical analysis
All experiments were performed in triplicates in each cell line and the data are shown as mean and standard deviation, of three separate experiments. All statistical analyses were performed with GraphPad Prism 5 software (La Jolla, California, U.S.A.) using one-way analysis of variance followed by Dunnett's test to evaluate significance relative to control.

The toxic effect of CoNPs on vascular endothelial cells
Using electron microscope, we observed that CoNPs entered into HUVEC cells after treatment with CoNPs for 24 h ( Figure 1A). HUVEC and HMEC-1 cells were exposed to varying concentrations of CoNPs (0, 100, 200, 400, 800, and 15000 μM) for 24 h. MTT analysis showed that CoNPs decreased the viability of HUVEC and HMEC-1 cells in a dose-dependent manner ( Figure 1B). Treatment with 800 μM CoNPs decreased the viability of HUVEC and HMEC-1 cells to approximately 50% (P<0.01). The present study also evaluated the toxic effect of 800 μM CoNPs at different time points (0, 2, 6, 12, 24, and 48 h). We found that 800 μM CoNPs marginally increased the viability of HUVEC and HMEC-1 cells after a very short incubation period (2 h). However, the viability HUVEC and HMEC-1 cells significantly decreased after incubation for 12 h with 800 μM CoNPs (P<0.05) and this decrease occurred in a time-dependent manner. Treatment of HUVEC and HMEC-1 cells with 800 μM CoNPs for 24 h increased apoptosis (P<0.01, Figure 1C). The concentration of LDH in the culture medium was also found to be higher 24 h after treatment with 800 μM CoNPs (P<0.05, Figure 1D).

CoNPs caused the activation of HIF-1α signalling and ROS production but decreased intracellular Fe 2+ level
To evaluate the effect of CoNPs on HIF-1α signalling, we measured the protein level of HIF-1α and mRNA levels of genes, vascular endothelial growth factor (VEGF) and glucose transporter 1 (GLUT1), which are pre-transcriptionally regulated by HIF-1α. As seen in the Western blot, HIF-1α protein levels were higher in HUVEC and HMEC-1 cells under normoxia after treatment with 800 μM CoNPs for 24 h (P<0.001, Figure 2A). CoNPs also increased VEGF and GLUT1 mRNA levels (P<0.01 or P<0.001, Figure 2B). Moreover, treatment with CoNPs dramatically increased ROS level in HUVEC and HMEC-1 cells (P<0.001, Figure 2C). LPO and MDA were increased by CoNPs as well (P<0.001, Figure 2D and P<0.01, Figure 2E, respectively). However, intracellular Fe 2+ and total iron levels were found to be decreased after CoNPs treatment (P<0.01 or P<0.001, Figure 2F).

Activation of HIF-1α signalling was associated with the toxic effect of CoNPs on vascular endothelial cells
To determine whether HIF-1α activation mediates the toxic effect of CoNPs on vascular endothelial cells, we knocked down HIF-1α before treating vascular endothelial cells with CoNPs (P<0.001, Figure 3A,B). HIF-1α knockdown attenuated the reduction in cell viability caused by CoNPs (P<0.05 vs. CoNPs group, Figure 3C). The increase in apoptosis and release of LDH resulting from treatment with CoNPs were also suppressed with HIF-1α knockdown (P<0.05 vs. CoNPs group, Figure 3D,E).

Ferrous agents conferred protection against the toxicity of CoNPs in vascular endothelial cells
CoNPs caused significant reduction of Fe 2+ level in HUVEC and HMEC-1 cells, where Fe 2+ plays important role in maintaining HIF-1α signalling in an inactive state. We added three ferrous agents, Fe(CH3CHOHCOO) 2 , Fe(CH 2 COO) 2 , and FeCl 2 to vascular endothelial cells to compensate for the reduction in Fe 2+ level caused by CoNPs, which was supposed to attenuated the toxic effects of CoNPs. All three ferrous agents improved HUVEC and HMEC-1 cell viability that was diminished by CoNPs (P<0.05, Figure 5A), while FeCl 2 could not improve cell viability to the extent the other two ferrous agents could. All three ferrous agents decreased apoptosis in HUVEC and HMEC-1  cells and LDH concentrations in culture medium (P<0.05, Figure 5B,C), with more effective effects observed in Fe(CH 2 COO) 2 and Fe(CH3CHOHCOO) 2 .
The morphological changes of apoptosis were observed through a dual staining with Hoechst33258 and PI. The morphological changes of apoptosis included chromatin condensation (in early stage), and cell membrane fragmentation (in later stage). After Hoechst33258 staining, the nuclei of normal cells were normal blue, while the nuclei of apoptotic cells were dense stained or fragmented, and some of them turn white from blue colour. The nuclei of normal cells were not stained by PI, because PI can not pass through the complete cell membrane. However, the integrity of cell membrane was broken in the later stage of apoptosis, thus the nuclei of apoptotic cells can be stained. We observed that the cell number was reduced after treatment with CoNPs, with notable increase of PI staining ( Figure 6). Silencing HIF-1α blocked the reduction in cell number caused by CoNPs, whereas a portion of cells showed chromatin condensation as indicated by Hoechst33258 staining. Treatments with Fe(CH3CHOHCOO) 2 , Fe(CH 2 COO) 2 , and FeCl 2 attenuated the increase in PI staining induced by CoNPs.
Moreover, intracellular ROS, LPO and MDA levels, which increased upon treatment with CoNPs, were reduced in the presence of all three ferrous agents (P<0.05 or P<0.01, Figure 8). All three ferrous agents decreased CoNPs-mediated increase in NOX1, NOX2, NOX4 and NOS3 expression.

Discussion
The toxic effects of CoNPs have earlier been reported in liver, kidney, and brain but few studies have investigated the influence of CoNPs on blood vessels [1][2][3][4]. After CoNPs enter the bloodstream, vascular endothelial cells are directly exposed to CoNPs. We found that CoNPs exerted their toxic effects on the vascular endothelium. Since the rate of apoptosis and LDH release also increased, along with inhibition of cell viability, CoNPs probably caused both apoptosis and necrosis of vascular endothelial cells. Most studies attribute the toxic effects of CoNPs to ROS production. However, not all antioxidants have been effective in attenuating toxicity. For example, l-ascorbic acid and α-tocopherol have been found to attenuate apoptosis, induced by CoNPs, by suppressing ROS production [18,19]. In contrast, glutathione failed to enhance cell viability of human macrophages after CoNPs treatment, although it decreased CoNP-induced ROS production [7]. These data suggest that the toxicity of CoNPs cannot be attributed to the adverse effects of ROS alone. Interestingly, we found that vascular endothelial cells had relatively strong resistance to CoNPs compared other types of cells (e.g. lymphocytes) [7]. One possible reason is that endothelial cells have a property that they closely connect to each us due to the tight junction. We guess that the tight junction between endothelial cells prevent the contact of CoNPS to the side and bottom of endothelial cells, only exposing the top Hoechst33258 staining, the nuclei of normal cells were normal blue, while the nuclei of apoptotic cells were dense stained or fragmented, and some of them turned white from blue colour. The nuclei of normal cells were not stained by PI, because PI can not pass through the complete cell membrane. However, the integrity of cell membrane was broken in the later stage of apoptosis, thus the nuclei of apoptotic cells can be stained. We observed that the cell number was reduced after treatment with CoNPs, with notable increase in PI staining. Silencing HIF-1α blocked the reduction in cell number caused by CoNPs, whereas a portion of cells showed chromatin condensation as indicated by Hoechst33258 staining. Treatments with Fe(CH3CHOHCOO) 2 , Fe(CH 2 COO) 2 , and FeCl 2 attenuated the increase in PI staining induced by CoNPs. Fe 2+ and total iron levels were measured following the CoNPs treatment. (A) Expression of HIF-1α, VEGF and GLUT1 were tested using Western blot (B) or PCR assay (C). Treatment with these three agents increased Fe 2+ and total iron levels in HUVEC and HMEC-1 cells that were exposed to CoNPs. All three ferrous agents inhibited increase in HIF-1α, VEGF, and GLUT1 induced by CoNPs. of cells to CoNPS. Previous study showed lymphocytes are vulnerable to CoNPs. Lymphocytes are independently floating in the culture medium [7]. Therefore, CoNPS can contact to every sides of lymphocytes.
A few studies have found abnormal activation of HIF-1α signalling by CoNPs, even under normoxia. The primary function of HIF-1α signalling is to enable the cell to adapt to hypoxia, for example by increasing GLUT expression to elevate the efficiency of glucose transport and by inducing autophagy to derive energy from the degradation of the structural components of the cell [20]. However, activation of HIF-1α under normoxia is associated with many adverse effects. For instance, high glucose levels can trigger HIF-1α activation in H9c2 cardiomyoblasts under normoxia. Silencing HIF-1α attenuated apoptosis and inflammation in H9c2 cells under high glucose stress [21]. In addition, various reports indicate that HIF-1α induces the expression of NOXs and NOSs, which are key enzymes in cellular ROS production [10]. We have confirmed that HIF-1α activation is closely linked with the toxicity of CoNPs in vascular endothelial cells, because HIF-1α depletion attenuated apoptosis, LDH release, and ROS production caused by CoNPs. This is in agreement with the results from a study done by Nyga et al., who found that HIF-1α activation mediated the toxic effects of CoNPs in human macrophages [7].
HIF-1α activity is tightly controlled by Fe(II)-dependent PHD that induces the rapid degradation of HIF-1α under normoxia. However, we found that Fe 2+ levels decreased in HUVEC and HMEC-1 cells after being treated with CoNPs. Although HIF-1α is involved in the regulation of intracellular iron levels, HIF-1α knockdown did not result in the recovery of Fe 2+ levels, suggesting that reduction of Fe 2+ level by CoNPs is not related to HIF-1α activation. Conversely, supplementation with Fe 2+ -containing agents to restore Fe 2+ level in cells resulted in the degradation of HIF-1α and down-regulation of genes downstream of HIF-1α. These data suggest that the activation of HIF-1α induced by CoNPs is associated with the reduction of Fe 2+ levels. The mechanisms by which CoNPs caused a reduction of Fe 2+ level in cells, are not fully understood. According to the literature, CoNPs-induced ROS depletes large amounts of vitamin C in cells required for the oxidation of Fe(II) from Fe(III) [7]. Another possible reason is that Co 2+ from CoNPs competes with Fe 2+ for divalent metal-ion transporter-1 binding [22]. Divalent metal-ion transporter-1 is important for transporting iron ions but can also transport other divalent metal ions such as, Co 2+ and Mn 2+ [22]. However, both these reasons cannot fully explain all the results from CoNPs treatment. The present study helps confirm that ROS production induced by CoNPs is closely linked with HIF-1α activation, and HIF-1α activation is associated with Fe 2+ reduction caused by CoNPs. Thus, ROS production is supposed to occur after Fe 2+ reduction. Of course, existence of other mechanisms underlying CoNPs-induced ROS production before Fe 2+ is reduced, cannot be ruled out. Further, the present study confirms that CoNP toxicity cannot be attributed to induce ROS alone; its toxicity is also associated with the induction of iron deficiency. Therefore, further work is needed to elucidate the mechanism behind CoNPs-induced reduction of Fe 2+ level in cells.
In the present study, we have reported for the first time, that supplementing media with Fe 2+ -containing agents protected vascular endothelial cells from CoNPs toxicity. This finding might overthrow following traditional concept: CoNPs exert their toxic effects by triggering ROS overproduction; it is well-known that Fe 2+ strengthens the oxidative capacity of ROS by a Fenton-type reaction (O 2 ᭹− + H 2 O 2 → O 2 + OH − + OH). Thus, it is necessary to decrease Fe 2+ levels to suppress the oxidative capacity of ROS. However, according to our data, CoNPs caused a notable reduction in Fe 2+ levels, which causes HIF-1α activation and subsequently, ROS production. Therefore, supplementation with Fe 2+ -containing agents, within a proper range of concentrations, inhibited ROS induced stress and protected from CoNPs toxicity.
The present study also revealed that the protective effects of Fe(CH3CHOHCOO) 2 and Fe(CH 2 COO) 2 were more pronounced than those of FeCl 2 , although the efficiency of FeCl 2 in enhancing intracellular iron levels was higher than the other two. After being released from Fe(CH3CHOHCOO) 2 and Fe(CH 2 COO) 2 , Fe 2+ forms weak acids with H + , whereas FeCl 2 probably forms a strong acid. We found that the dosage range of FeCl 2 for protecting cells was very narrow. Using high doses of FeCl 2 would leave many Cl − in the culture medium, thereby destroying the acid-base balance in the extracellular environment. In addition, CH3CHOHCOO − and CH 2 COO − might chelate Co 2+ like they chelate Fe 2+ . This reaction might also attenuate CoNPs toxicity, but proving this hypothesis requires further investigation. Thus, the present study recommends the use of Fe(CH3CHOHCOO) 2 and Fe(CH 2 COO) 2 to protect vascular endothelial cells from CoNPs toxicity.
In summary, the present study revealed that HIF-1α activation mediated the toxic effects of CoNPs in human vascular endothelial cells. The activation of HIF-1α is associated with the reduction of Fe 2+ level in cells, which is caused by CoNPs. Supplementation with Fe 2+ -containing agents, especially Fe(CH3CHOHCOO) 2 and Fe(CH 2 COO) 2 mitigated the toxic effects of CoNPs.

Data Availability
The data used to support the findings of the present study are available from the corresponding author (Fan Liu) upon request.