Two gas metal arc welding fume samples were kindly generated and provided by Lincoln Electric Co. (Cleveland, OH). Bulk samples of the fumes were generated in a cubical open front fume chamber (volume = 1 m³) by a skilled welder using a semi-automatic technique appropriate to the electrode and collected on 0.2 µm Nuclepore filters (Nuclepore Co., Pleasanton, CA). The two fume samples were generated in following ways: (1) gas metal arc welding using a mild steel E70S-3 electrode (GMA-MS) and (2) gas metal arc welding using a stainless steel ER308L Si electrode (GMA-SS) with argon and CO₂ shielding gases to protect the weld from oxidation. The welding fume generated from the Ni-Cu-based consumable (Ni-Cu WF) was produced by the Welding and Joining Metallurgy Group at The Ohio State University (Columbus, OH) during shielded metal arc (SMA) welding using a consumable with a target composition of Ni-7.5Cu-1Ru that contained other alloying agents. The fume samples were collected onto electrostatic filter medium (PE 1306NA; Hollingsworth and Vose, East Walpole, MA). For cellular exposure experiments, particles were suspended in sterile filtered 1X phosphate-buffered saline (PBS) at a 1 mg/ml stock, vortexed, and diluted into cell culture media at a low (50 µg/ml) or high dose (250 µg/ml), based on previous in vitro studies of GMA-MS and GMA-SS fumes [7], [16]. Welding fume stocks were prepared freshly for experiments. Equal volumes of 1X PBS were used as control conditions for each experiment.

Particle sizes of the welding samples collected onto filters in bulk were characterized by scanning electron microscopy and all three were of respirable size with count mean diameters of <2 µm. Replicate welding fume samples were analyzed for Cr(VI) levels using NIOSH method 7605 [17]. Briefly, 5 ml of extraction solution (3% Na₂CO₃/2% NaOH) were added to each 5 mg sample, and the tubes were sonicated in a bath for 30 min. This procedure extracts both soluble and insoluble Cr(VI) present in the fumes. Analysis used a Dionex HPIC-AS7 column with 250 mM (NH₄)₂SO₄/100 mM NH₄OH mobile phase and a post-column reagent (2.0 mM diphenylcarbazide/10% methanol/1N H₂SO₄) with absorbance detection at 540 nm. Four concentrations of standards were made from a certified Cr(VI) solution, covering a range of 0.4–4 µg/ml. The estimated limit of detection is 0.02 µg, and the method range is 0.05 to 20 µg of Cr(VI). Analysis of other elements present in the welding fume samples was done using inductively coupled plasma-atomic emission spectroscopy (ICP-AES) using NIOSH method 7300 modified for microwave digestion (NIOSH 1994).

In addition, portions of the different welding fume samples (GMA-SS, GMA-MS, and Ni-Cu WF) were suspended in distilled water, pH 7.4, and sonicated for 1 min with a Sonifier 450 Cell Disruptor (Branson Ultrasonics, Danbury, CT, USA) to determine particle/metal solubility. The three particle suspensions (total samples) were incubated for 24 hours at 37°C, and the samples were centrifuged at 12,000 g for 30 min. The supernatants of the samples (soluble fraction) were recovered and filtered with a 0.22 µm filter (Millipore Corp., Bedford, MA, USA). The pellets (insoluble fraction) were resuspended in water. The sample suspensions (total, soluble, and insoluble fractions) were digested, and the metals analyzed by ICP-AES according to NIOSH method 7300 [18]. All three welding fumes were observed to be relatively insoluble with soluble/insoluble ratios of 0.006, 0.020, and 0.004 for the GMA-MS, GMA-SS, and Ni-Cu WF samples, respectively.

The adherent mouse monocyte-derived macrophage cell line, RAW 264.7, was obtained from American Type Culture Collection (Manassas, VA). The cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with 2 mM L-glutamine, 10% fetal bovine serum, and 50 mg/ml penicillin/streptomycin (Invitrogen Life Sciences, Grand Island, NY). Cells were grown at 37°C in a 5% CO₂ incubator and passaged by scraping into medium or PBS, depending on the experiment.

Cells were plated at 2×10⁵ cells/well in 24-well dishes, treated with suspensions of welding fumes (50 µg/ml or 250 µg/ml), and incubated at 37°C for 24 hours. Wells were washed twice with 1X PBS, scraped into 250 µl PBS, mixed 1∶1 with trypan blue (vol:vol) and cell counts were calculated using a Countess Automated Cell Counter (Life Technologies). Data from 4 independent experiments are reported as the mean number of viable cells and percent viability out the cell population for each condition.

For viability measurements in the presence of an antioxidant, cells were plated at 1×10⁵ cells/ml in 96-well dishes, and half the wells were pre-treated with 20 mM N-Acetyl-L-cysteine (Sigma-Aldrich, St. Louis, MO) for 1 hour prior to treatments with suspensions of welding fumes. Changes in cell viability at 24 hours were assessed using the MultiTox-Fluor Cytotoxicity Assay according to the manufacturer’s directions (Promega, Madison, WI). A cell-permeable protease substrate, GF-AFC (glycyl-phenylalanyl-amino-fluorocoumerin), is cleaved by live cells to produce fluorescent AFC. Thus, the fluorescent signal is proportional to the number of viable cells. To ensure the fluorescent signal was due to substrate products and not autofluorescence or interference by the welding fume suspensions, separate wells of medium and fume suspensions were included in the plates, and these readings were subtracted from their respective wells that contained treated cells. All conditions were run in triplicate wells and three independent experiments were performed.

Cells were plated at 12,500 cells/well in 96-well Seahorse XFe-96 assay plates for 24 hours before being treated with welding fume suspensions (50 µg/ml or 250 µg/ml) for another 24 hours. On the day of metabolic flux analysis, cells were changed to unbuffered DMEM media (DMEM supplemented with 25 mM glucose, 10 mM sodium pyruvate; pH 7.4) and incubated at 37°C in a non-CO₂ incubator for 1 hour. All media was adjusted to pH 7.4 on the day of assay. Six baseline measurements of oxygen consumption rate (OCR) were taken before sequential injection of mitochondrial inhibitors, oligomycin (10 µM), FCCP (1 µM; carbonilcyanide p-triflouromethoxyphenylhydrazone) and rotenone (5 µM). Four measurements were taken after each addition of mitochondrial inhibitor before injection of the next inhibitor. Five independent experiments were performed. Oxygen consumption rate, an indicator of oxidative phosphorylation, was automatically calculated and recorded by the Seahorse XFe-96 software (Seahorse Bioscience, North Billerica, MA, USA).

To detect and measure short-lived free radical intermediates, electron spin resonance (ESR) spin-trapping was used. To assess whether the welding fume suspensions are capable of producing hydroxyl radicals (•OH) after exposure to H₂O₂ through a Fenton-like reaction, final concentrations of 1 mg/ml compounds, 1 mM H₂O₂, and 100 mM DMPO spin trap (5,5′-dimethylpyrroline N-oxide, Sigma-Aldrich) were mixed in test tubes for 3 min at room temperature and transferred to a quartz flat cell for ESR measurement in a Bruker EMX spectrometer (Bruker Instruments Inc., Billerica, MA). For each sample, the instrument was set to run 10 scans with a 41 sec scan time, a receiver gain of 2.5×10⁴, a 40 msec time constant, 1.0 G modulation amplitude, 126.9 mW power, 9.751 frequency, and 3475±100 G magnetic field. Signal intensity (peak height) from the 1∶2∶2∶1 spectra, which is characteristic of •OH [19] was used to measure the relative amount of short-lived radicals trapped.

For cellular ESR, final concentrations of 2×10⁶ RAW 264.7 cells/ml, 1 mg/ml welding fume suspensions, and 200 mM DMPO in PBS were mixed in test tubes. Experiments were carried out in the presence and absence of the H₂O₂ scavenger catalase (2000 U/ml), as previously described [20]–[22]. Samples were incubated at 37°C for 5 min and were loaded into a flat cell and scanned as in previous cellular ESR experiments [20], [23]–[25]. Again, peak heights represent relative levels of trapped hydroxyl radicals.

Cells were plated at 1×10⁵ cells/ml in 96-well dishes and treated with the cell-permeable fluorogenic probe DCFH-DA (2′,7′-dichlorodihydrofluorescein diacetate, Cell Biolabs, Inc., San Diego, CA) at a final concentration of 1 mM in serum-free DMEM for 45 min at 37°C. Cells were washed twice with 1X PBS and DMEM was added back to the wells, along with 50 µg/ml welding fume suspensions. Upon ROS production, DCFH-DA is oxidized to form DCF, which is highly fluorescent. The cells were incubated at 37°C for 7 hours, and plates were read at 485 nm excitation/530 nm emission each hour to measure any change in fluorescence, indicating ROS production. To ensure the fluorescent signal was due to DCF product and not any autofluorescence that may emit from the welding fume suspensions, separate wells of DMEM and each fume were included in the plates, and these readings were subtracted from their respective wells that contained treated cells. All conditions were run in triplicate wells and three independent experiments were performed.

Cells were plated at 2×10⁵ cells/ml in 24 well plates, treated with or without 50 µg/ml welding fume suspensions for 3 hours, washed twice and scraped into 1X PBS, added to glass slides with agarose, then lysed and subjected to electrophoresis according to manufacturer’s instructions (Trevigen Inc., Gaithersburg, MD). This causes fragmented DNA to migrate out of the nuclear region, forming a comet-like tail, which was labeled with SYBR green (binds double-stranded DNA). Images were acquired using an Olympus AX70 microscope equipped with an Olympus DP73 digital camera (Olympus, Center Valley, PA). Two independent experiments were performed and at least 50 total cell comets per condition (23–30 comets per replicate) were measured for the percentage of DNA in comet tails per condition. This was calculated by determining the background-corrected fluorescence from: the nuclear region of interest in a cell (nuclear ROI) and the total cell fluorescence for a defined region of interest (total cell ROI). ImageJ software [26] was used to measure the “integrated density” (the sum of the values of the pixels in the selection) for regions of interest, and Microsoft Excel was used to calculate the corrected fluorescence values and ratios of nuclear ROI to total cell ROI. This was converted to a percentage of DNA in the “tail” (100 – [(nuclear ROI/total cell ROI)×100]), with elevated percentages indicating DNA damage.

Cells were plated at 1.5×10⁵ cells/ml in 96-well dishes and pre-treated with 50 µg/ml welding fume suspensions for 3 or 6 hours prior to the addition of pHrodo Red E. coli BioParticles, which were prepared according to the manufacturer’s instructions (Molecular Probes, Carlsbad, CA). Briefly, E. coli BioParticles were suspended in Live Cell Imaging Solution at 1 mg/ml and sonicated for 5 min. At each time point, wells were washed twice with warm Live Cell Imaging Solution, and 100 µl of suspended BioParticles were added to all wells, including blanks (no cells) to be subtracted from all wells. Following 2 hours incubation at 37°C, plates were read at 560 nm excitation/600 nm emission in a plate reader to measure changes in fluorescence. Cytochalasin D (Sigma-Aldrich) was added to triplicate wells 1 hour prior to the end of each time point as a positive control for preventing phagocytosis. It was added back to those wells with the BioParticles at a final concentration of 10 µM, allowing 3 hours total incubation with the phagocytosis inhibitor. All conditions were run in triplicate wells and four independent experiments were performed. The percentage phagocytosis was calculated as a ratio of the net experimental phagocytosis to the untreated control cell phagocytosis.

Cells were plated at 5×10⁵ cells/ml in 12-well dishes and treated with 50 µg/ml or 250 µg/ml welding suspensions for 24 hours. Lipopolysaccharide (LPS) at 1 µg/ml was used as a positive control (from E. coli 055:B5, Sigma-Aldrich Corp., St. Louis, MO). Media were collected from wells at the time points and frozen at −80°C. Cytokine levels were measured via BD OptEIA ELISA kits (mouse TNFα; cat. no. 555268, mouse IL-6; cat. no. 555240, mouse IL-1β; cat. no. 559603, BD Biosciences, San Diego, CA) according to manufacturer’s directions.

All data are represented as the mean ± standard deviation (SD). A one-way or two-way analysis of variance (ANOVA) with a Tukey post-test was performed using GraphPad Prism 6 software (GraphPad Software, Inc., La Jolla, CA) for each experiment to compare the responses between groups, and statistical significance is shown when p<0.05. For the mitochondrial stress assay results, data were first log-transformed prior to analysis.

The welding fumes used in this study were generated from either gas metal arc welding with mild steel (GMA-MS) or stainless steel (GMA-SS) materials or by shielded metal arc welding using a nickel-copper-based consumable (Ni-Cu WF). Table 1 lists the three welding samples and their respective metal content. The GMA-MS consisted mostly of Fe (82.8%) and Mn (15.2%), while GMA-SS had similar Mn content (13.8%) but less Fe (57.2%) and a large percentage of Cr (20.3%). The newly-developed Ni-Cu WF had a very different metal profile compared to the GMA-MS and GMA-SS fumes, as it contained very low levels of Fe, Mn, and total Cr. This welding fume contained mostly potassium (29.9%), aluminum (20.7%), Ni (13.4%), and titanium (11.7%). All three particles were similar in size and in the respirable range (<2 µm count mean diameters), and tests of particle and metal solubility showed all samples to be relatively insoluble in water. Therefore, any observed differences are likely to be due to the metal profiles of the fume samples.

To determine the degree of welding fume cytotoxicity, viability assays were carried out in RAW 264.7 cells following 24 hour treatments with 50 µg/ml and 250 µg/ml doses. All three welding samples at the higher dose caused significant reductions in the concentration of live cells and overall viability using trypan blue exclusion (Figure 1). However, only treatment with Ni-Cu WF caused a significant reduction in the live cell concentration at the 50 µg/ml dose (9.5×10⁵ cells/ml, Ni-Cu WF vs. 1.5×10⁶ cells/ml, PBS), suggesting that this newly-developed welding material is more cytotoxic to RAW 264.7 macrophages than the GMA-MS and SS fumes.

Further, an additional cytotoxicity assay was performed to measure the activity of a protease that is only active in viable cells. In this assay, cells were pre-treated with or without the antioxidant N-acetyl-L-cysteine (NAC) to assess whether the observed reduced cell viability was due to excessive ROS production. This appeared to not be the case, as there were no significant differences between wells treated with and without NAC for the same welding fume dose (Figure 2). However, compared to PBS controls, there were significant reductions in the live cell signal at the 250 µg/ml dose, confirming cytotoxicity for all three welding fumes. Also in agreement with Figure 1 results, only Ni-Cu WF induced significant reductions at the 50 µg/ml dose.

As an additional approach to examine welding fume cytotoxicity and effects on viability, we measured mitochondrial function. Using the Seahorse Biosciences Extracellular Flux (XF) Analyzer, real-time changes in cellular bioenergetics were measured via a mitochondrial stress test. Figure 3A shows oxygen consumption rate (OCR) traces after 24 hour exposures to the welding fume samples from a representative experiment. We found that 50 µg/ml GMA-MS- and GMA-SS-treated cells in general had slightly higher basal OCRs than PBS-treated controls. Following a period of basal readings, compounds used to estimate bioenergetic function were added to the wells. Oligomycin inhibits the ATP synthase (Complex V) and was used to determine the extent by which OCR is linked to ATP production, FCCP is a proton uncoupler that forces maximal respiratory capacity, and antimycin A/rotenone inhibit Complexes III and I, respectively, to dramatically suppress the OCR. The resulting rates were used to calculate ATP production, maximum respiration, and spare capacity compared to control cells. Reductions in these parameters at the 250 µg/ml dose confirm the loss of viability induced by all three fumes as in Figures 1 and 2.

Pre-treatment of cells with 50 µg/ml Ni-Cu WF not only caused lower basal OCRs, but also prompted changes in the way mitochondria were able to deal with pharmacological inhibitors (Figure 3B). For example, following treatment with FCCP, the OCR spiked to provide maximal respiration, but this response was mitigated in cells pre-treated with Ni-Cu WF (65.4±15.4% of PBS control, mean ± SD). Whether this was due to fewer mitochondria because of proceeding cell death or actual mitochondrial dysfunction remains unclear; however, it is likely a combination of both factors that led to these results. Although it was not statistically significant for GMA-MS and GMA-SS, the reserve or spare capacity (maximum OCR – basal rate) was reduced compared to controls following 50 µg/ml treatments with all three fume samples (74.1±24.6% GMA-MS, 78.9±45.0% GMA-SS, and 45.1±13.5% Ni-Cu WF, mean ± SD). This measurement represents the respiratory capacity available for cells when responding to various stressors that increase bioenergetic demands. Its decrease has been implicated in oxidative stress conditions [27], [28], which may be induced by these welding fumes.

Due to the previously-documented evidence of ROS production from welding fumes [7], [8], [16], [29] and effects on mitochondrial bioenergetics, the samples were tested for their ability to generate hydroxyl radicals via electron spin resonance (ESR) using spin trapping with DMPO (5,5′-dimethylpyrroline N-oxide). The acellular Fenton-like reactions were carried out with 1 mM H₂O₂ and 1 mg/ml welding fume suspensions to determine if the particles have the potential to convert H₂O₂ into •OH as in previous studies [16], [24], [30]. Each of the samples induced ESR peaks in a 1∶2∶2∶1 spectra, indicating •OH radical generation [19], but the GMA-SS fume produced significantly higher peaks compared to the other two samples (94.7±5.6 mm GMA-SS vs. 68.2±10.6 mm GMA-MS and 52.3±3.6 mm Ni-Cu WF, mean ± SD) (Figure 4A). ESR was also carried out with the particles incubated for 5 minutes at 37°C with RAW 264.7 cells. The goal was to examine whether these inflammatory cells produce free radicals as a quick, oxidative burst reaction to the welding fumes. Compared to PBS-treated control cells, all 3 welding fumes produced peaks in the spectra (93.7±17.2 mm GMA-MS, 79.5±16.5 mm GMA-SS, 56.8±8.0 mm Ni-Cu WF, mean ± SD) (Figure 4B). However, both the GMA-MS and GMA-SS samples caused significantly greater hydroxyl radical production compared to Ni-Cu WF. To confirm the role of endogenous H₂O₂ from cells as the source by which the welding fumes generate hydroxyl radicals, additional samples were run in the presence of catalase, which is a strong H₂O₂ scavenger. This addition abolished peaks with all three welding fume samples, as in previous ESR studies with metal-containing particles [20], [22], [31].

To further explore the influence of the welding fumes on ROS production in cells, an intracellular ROS assay was used. Utilizing DCFH-DA (2′,7′-dichlorodihydrofluorescein diacetate) as a fluorescent probe for ROS within cells, we were able to monitor changes in fluorescence over a time course of 7 hours. Only the 50 µg/ml dose was used for these experiments, because the 250 µg/ml dose appears to be cytotoxic with all three samples (Figures 1–3) and mostly viable cells were needed for ROS measurements. Figure 5A shows that GMA-MS and GMA-SS induced strong intracellular ROS production starting at the 2 hour time point and this continued to increase over several hours, whereas the Ni-Cu WF only induced minimal ROS over the time course. This agrees with the ESR data, in which Ni-Cu WF was less capable of producing free radicals compared to the GMA-MS/SS fumes.

An alkaline comet assay was also performed to determine whether the intracellular ROS from fume exposures caused oxidative damage to nuclear DNA. The comet assay has been long used to visualize DNA strand breakage as an indicator of genotoxic insult [32], [33]. Our findings indicated that cells treated for 3 hours with 50 µg/ml welding fumes had significant DNA damage with all exposures (43.5±3.6% GMA-MS, 43.6±0.6% GMA-SS, and 49.7±1.3% Ni-Cu WF vs. 24.8±2.2% PBS, mean ± SD) (Figure 5B). This suggests that the persistence or redox cycling of ROS within welding fume-exposed cells can lead to downstream oxidative damage.

Macrophages infiltrate the lung following welding fume inhalation and appear to play a role in the pulmonary clearance of these particles [34]–[36]. However, it has also been shown that GMA welding fumes can negatively affect macrophage function [7], [37]. Thus, we examined the ability of RAW 264.7 cells to properly phagocytize E. coli particles following 3 or 6 hour welding fume exposures at 50 µg/ml. The shorter time points were chosen, because exposure to Ni-Cu WF for 24 hours reduced cell viability at this lower dose (Figures 1, 2), and we wanted the cells to be viable in order to observe changes in function prior to cytotoxicity. Figure 6 shows that only pre-treatment with Ni-Cu WF significantly impaired subsequent phagocytosis of a pathogen-based particle compared to PBS controls (72.3±2.1% and 74.5±10.8% at 3 and 6 hours, respectively, mean ± SD) while GMA-MS and GMA-SS had no effect at these time points and at this concentration. These results suggest that perhaps pre-uptake of GMA-MS and GMA-SS cause cells to produce more ROS, whereas Ni-Cu WF negatively affects macrophage function by reducing the phagocytic ability prior to cytotoxicity. We hypothesize that whatever negative effect these particles had on the RAW 264.7 cells’ ability to phagocytose bacteria was already occurring by the 3 hour time point. Thus, at 6 hours and perhaps later times, the cells would still not be able to properly take up the E. coli BioParicles.

Based on the above results showing ROS production and previous in vivo studies that have shown GMA welding fumes to cause pulmonary inflammation [34], [37]–[39], we hypothesized that cytokine production may occur in cells following welding fume exposures. Unprimed RAW 264.7 cells were exposed to welding fumes for 24 hours and ELISAs were used to measure TNFα, IL-6, and IL-1β (Figure 7). Although GMA-SS treatment appeared to induce some cytokine production (TNFα and IL-6 at 250 µg/ml after 24 hours exposure), the results were not statistically significant from PBS controls. The cellular conditions at 24 hours included at least some portion of the cells undergoing death at the higher dose (Figures 1 and 2), so it is possible that this triggered some mild production and release of cytokines. However, in general, it appears that the welding fumes are not able to induce robust cytokine production from unprimed macrophages in vitro.