Bioactivity of Circulatory Factors After Pulmonary Exposure to Mild or Stainless Steel Welding Fumes Free

Abstract

Studies suggest that alterations in circulating factors are a driver of pulmonary-induced cardiovascular dysfunction. To evaluate, if circulating factors effect endothelial function after a pulmonary exposure to welding fumes, an exposure known to induce cardiovascular dysfunction, serum collected from Sprague Dawley rats 24 h after an intratracheal instillation exposure to 2 mg/rat of 2 compositionally distinct metal-rich welding fume particulates (manual metal arc welding using stainless steel electrodes [MMA-SS] or gas metal arc welding using mild steel electrodes [GMA-MS]) or saline was used to test molecular and functional effects of in vitro cultures of primary cardiac microvascular endothelial cells (PCMEs) or ex vivo organ cultures. The welding fumes elicited significant pulmonary injury and inflammation with only minor changes in measured serum antioxidant and cytokine levels. PCME cells were challenged for 4 h with serum collected from exposed rats, and 84 genes related to endothelial function were analyzed. Changes in relative mRNA patterns indicated that serum from rats exposed to MMA-SS, and not GMA-MS or PBS, could influence several functional aspects related to endothelial cells, including cell migration, angiogenesis, inflammation, and vascular function. The predictions were confirmed using a functional in vitro assay (scratch assay) as well as an ex vivo multicellular environment (aortic ring angiogenesis assay), validating the concept that endothelial cells can be used as an effective screening tool of exposed workers for determining bioactivity of altered circulatory factors. Overall, the results indicate that pulmonary MMA-SS fume exposure can cause altered endothelial function systemically via altered circulating factors.

Epidemiology studies suggest pulmonary exposure to particulate matter is associated with adverse cardiovascular outcomes (Brook et al., 2010). The precise mechanism of how pulmonary exposure to particulate matter translates to cardiovascular dysfunction is an area of continued exploration. Several processes were hypothesized to explain how pulmonary exposure induces systemic dysfunction. These include direct transfer of particulate matter from the lung to systemic circulation (Husain et al., 2015; Miller et al., 2017; Mills et al., 2009) or assisted through translocation within macrophages (Hussain et al., 2013; Meiring et al., 2005; Rothen-Rutishauser et al., 2007), trace metals and metal ions formed due to dissolution (Costa and Dreher, 1997; Wallenborn et al., 2007), alteration in circulating factors due to pulmonary exposure, (Aragon et al., 2016, 2017; Erdely et al., 2009; Mandler et al., 2017; Seaton et al., 1995) and finally, through perturbation of the autonomic nervous system (Carll et al., 2013; Farraj et al., 2012; Perez et al., 2015). This work aims to evaluate if pulmonary exposure to various welding fumes alters the bioactivity of the circulation.

Serum is a complex mixture consisting of cytokines/chemokines, proteins, damaged/oxidized proteins, peptide fragments, lipids, and much more. Alteration of any of these components may have profound effects on endothelial cells, the major regulatory cell type of the vasculature. Adding to the complexity, altered components individually can be present at subthreshold levels to not cause activation or dysfunction by themselves but when in conjunction with other mediators may result in additive, synergistic, or antagonistic effects (Channell et al., 2012). With respect to all factors present and potentially altered following a pulmonary exposure, a straight forward and practical approach, termed the serum cumulative inflammatory potential assay, uses endothelial cells as biosensors to evaluate serum bioactivity (Aragon et al., 2016; Channell et al., 2012; Cung et al., 2015; Kvietys and Granger, 1997; Zychowski et al., 2016). From an occupational exposure perspective, this approach has significant utility in determining whether an exposure has the potential to alter endothelial, and subsequently, cardiovascular function.

The International Agency for Research on Cancer of the World Health Organization estimates that there are approximately 11 million workers worldwide that have a job title of welder and about 110 million additional workers incur welding related exposure (IARC, 2018). Welding aerosols are a complex mixture of metal-rich particulate matter (the fume) and gases generated during joining of 2 base metal pieces with an electrode heated to extreme temperatures as high as 2000°C (IARC, 2018). Workers exposed to welding aerosols are at risk from several hazardous components, including chromium (Cr), copper (Cu), manganese (Mn), iron (Fe), and nickel (Ni). Inhalation of welding aerosols has been known to cause short- and long-term pulmonary effects including inflammation, cytotoxicity, bronchitis, siderosis, increased risk of infection, asthma, and cancer. Although the primary route of exposure is via inhalation, the health effects of welding extends systemically to include neurological and cardiovascular dysfunction (Antonini, 2003, 2014; Fang et al., 2009; Graczyk et al., 2015; Kim et al., 2005; Sjögren et al., 2002; Sriram et al., 2010; Yu et al., 2003; Zeidler-Erdely et al., 2014). In vivo studies have shown systemic oxidative stress, increased atherosclerotic plaque progression, aortic and cardiac inflammation, and vascular dysfunction following welding exposure (Zeidler-Erdely et al., 2014; Erdely et al., 2011a,c, 2014). Increased blood pressure, systemic inflammation, systemic oxidative stress, adverse cardiovascular outcomes, and altered heart rate variability were observed in welders (Cavallari et al., 2007a,b, 2008a,b; du Plessis et al., 2010; Fang et al., 2009; Ibfelt et al., 2010; Kim et al., 2005; Lai et al., 2016; Li et al., 2015; Shen et al., 2018; Umukoro et al., 2016).

Cellular studies in immune, epithelial, and endothelial cells as well as animal work studying the effect of lung, liver, and circulation after pulmonary exposure confirmed that stainless steel welding fume caused more cytotoxicity, inflammation, and epigenetic changes compared with mild steel fume (Antonini, 2003, 2014; Antonini et al., 1996, 2010; Channell et al., 2012; Shoeb et al., 2017; Taylor et al., 2003). Transcriptomic studies further confirmed these findings and revealed significant up-regulation of inflammatory and chemotactic genes in mice exposed to stainless steel fume compared with mild steel fume (Zeidler-Erdely et al., 2010). Similar profiles were observed systemically with aortic and cardiac inflammation following stainless steel welding fume exposure (Erdely et al., 2011c). Inhalation of stainless steel fume was also shown to advance atherosclerotic lesion development (Erdely et al., 2011a). Compositional difference, specifically the presence of Cr, Mn, and Ni in the stainless steel fume but not in mild steel fume, has been hypothesized to be the primary driver for the observed health hazards (Antonini, 2003; Taylor et al., 2003; Yu et al., 2003).

In this work, Sprague Dawley rats were exposed to 2 mg/rat of GMA-MS (gas metal arc welding using mild steel electrodes), MMA-SS (manual metal arc welding using stainless steel electrodes) or PBS via intratracheal instillation (ITI). At 24 h postexposure the pulmonary toxicity, inflammation and cytokine profiles were evaluated in bronchoalveolar lavage fluid (BALF). The same cytokine profile was measured in the serum, and the molecular and functional responses of the serum were examined on naïve primary cardiac microvascular endothelial cells (PCMEs). Functional predictions based on the transcriptional responses of the endothelial cells were validated in a single cell type and on the endothelial function in a more complex multicellular environment of a naive aorta.

MATERIALS AND METHODS

Welding fume collection and characterization

The 2 welding fumes, GMA-MS E70S-3 and shielded MMA-SS ER308–16-1 were generated and kindly gifted by Lincoln Electric Co. (Cleveland, Ohio). The fume particulate samples were generated in an open fume chamber (volume of 1 m³) by a welder using the electrodes specified above and collected on a 0.2 µm Nuclepore filters (Nuclepore Co.; Pleasanton, California). The welding fume composition was analyzed using ICP-AES previously reported and described in detail (Shoeb et al., 2017). The hydrodynamic agglomerated size of the particulate was evaluated using dynamic light scattering (DLS). DLS was performed on a Malvern Zetasizer Nano ZS90 (Worcestershire, UK) equipped with a 633 nm laser at a 90° scattering angle. The DLS measurements were performed by dispersing the welding fume particulate in PBS at 50 µg/ml. After 2 min of equilibration inside the machine, 5 measurements, each consisting of at least 5 runs were recorded. Scanning electron microscope images were obtained as described in detail previously (Kodali et al., 2017). Briefly, welding fume samples were dispersed in PBS and diluted in water. The particle suspension was withdrawn and filtered under vacuum onto a 0.2-μ polycarbonate filter. The filter was then mounted onto an aluminum stub using carbon double-stick tape. The sample was sputter coated with gold-palladium and imaged using a Hitachi S4800 field-emission scanning electron microscope (Hitachi, Japan).

Animal exposure and serum collection

An 8-week-old male Sprague Dawley rats (250–300 g) were acquired from Hilltop Lab Animals (Scottdale, Pennsylvania). The rats were acclimated for 1 week and were provided HEPA (High-efficiency particulate air)-filtered air, irradiated Teklad 2918 diet, and tap water ad libitum. The rats were exposed to PBS, 2 mg/rat GMA-MS or MMA-SS (n = 4 PBS, n = 6 GMA-MS, and n = 6 MMA-SS) one time via ITI. Welding fumes were dispersed in 300 µl sterile PBS and sonicated for 30 s before ITI. The animals were anesthetized by intraperitoneal injection of methohexital sodium (Brevital 500 mg; JHP Pharmaceuticals, LLC, Rochester, Michigan) before ITI. The dose 2 mg/rat represents an occupational exposure equivalent to 52.1 days or approximately 10.4 weeks of exposure (based on 5 days work per week) in humans based on calculations from (Sriram et al., 2010). The 2 mg/rat was a known dose to induce systemic oxidative stress (Erdely et al., 2014). At 24 h postexposure, whole blood was drawn using an 18-gauge needle from the abdominal vena cava and collected in BD Vacutainer tubes (Becton, Dickinson, and Co.; Franklin Lakes, New Jersey). The collected serum was aliquoted into 300 µl aliquots and stored at −80°C. All animal procedures used during the study were reviewed and approved by the CDC-Morgantown Institutional Animal Care and Use Committee. The animal facilities are specific pathogen-free, environmentally controlled, and accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International.

Cell culture and cytokine analysis

Rat PCMEs from Sprague Dawley rats were obtained from Cell Biologics Inc (Chicago, Illinois). All experiments were performed between passages 3–6. Cells were grown by coating sterile culture dishes or flasks with gelatin-based coating solution (Cell Biologics, Catalog No. 6950) for 2 min and then aspirating the excess solution before seeding cells. Cells were cultured using rat endothelial cell medium supplemented with epidermal growth factor (EGF), vascular endothelial growth factor (VEGF), L-glutamine, antibiotic-antimycotic, and Fetal Bovine Serum (FBS; Cell Biologics, Catalog No. M1266) in a humidified, 5%-CO₂ incubator at 37°C. Cells were passaged using a 0.25% trypsin containing 1 mM EDTA in Hank’s balanced salt solution without calcium and magnesium (Cell Biologics, Catalog No. 6914). During experimental challenge with serum collected from welding fume exposed rats, cells were cultured in basal endothelial medium without supplements and 10% serum collected from exposed rats. During serum starvation, cells were cultured in basal endothelial medium without supplements.

The cytokine/chemokines from serum and BALF of rats exposed to 2 mg/rat of GMA-MS, MMA-SS, or saline was assessed using the Rat Cytokine Array/Chemokine Array 27 Plex (Eve Technologies, RD27; Calgary, Canada). The cytokine/chemokines evaluated were eotaxin, EGF (Epidermal growth factor), fractalkine, interferon (IFN)-γ, Interleukins IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12(p70), IL-13, IL-17A, IL-18, IP-10 (Interferon γ-induced protein 10 kDa), GRO/KC (Growth-regulated oncogen/Keratinocyte chemoattractant), TNF-α (Tumor necrosis factor alpha), G-CSF (Granulocyte colony-stimulating factor), GM-CSF (Granulocyte-macrophage colony-stimulating factor), MCP-1 (Monocyte chemoattractant protein-1), Leptin, LIX (Liposaccharide-Induced CXC chemokine), MIP-1α (Macrophage inflammatory protein 1 alpha), MIP-2 (Macrophage inflammatory protein 2), RANTES (Regulated upon activation, normal T cell expressed and presumably secreted), and VEGF (Vascular endothelial growth factor). The assay sensitivities for these markers ranged from 0.1 to 33.3 pg/ml.

Endothelial transcriptional array

mRNA was extracted from PCMEs serum starved overnight and challenged for 4 h with endothelial media containing 10% serum from Sprague Dawley rats exposed to 2 mg/rat of GMA-MS, MMA-SS, or PBS for 24 h using the RNeasy Mini Kit (Qiagen Inc, Maryland). RNA was measured using a NanoDrop UV-vis Spectrophotometer (Wilmington, Delaware). Purity of the RNA preparation was determined as the 260/280 nm ratio with expected values between 2 and 2.3. cDNA was prepared from RNA samples (1 µg) using RT² First Strand Kits (Qiagen Inc). The real-time polymerase chain reactions were conducted using RT² qPCR SYBR Green Master Mix (Qiagen Inc). Rat Endothelial Cell Biology RT² profiler PCR array (Qiagen Inc, Ref. PARN-015za) containing a panel of 84 endothelial cell biology-related genes and 5 housekeeping control genes was used to profile the transcriptional response. Gene amplification was performed on an Applied Biosystems StepOne Plus (Applied Biosystems, California) with a 10 min at 95°C activation step followed by 40 cycles with a 15 s at 95°C denaturation and 1 min at 60°C extension step. The data were analyzed on Qiagen RT² Profiler PCR Array Data Analysis web-based platform. mRNA level expression of each gene was normalized to the average expression of the 5 housekeeping genes Actb, B2m, Hprt1, Rplp1, and Ldha. All experiments were conducted in triplicate. The p values are calculated based on a Student’s t test of the replicate 2^ (−Delta CT) values for each gene in the control group and treatment groups. Results were expressed relative to cells challenged with serum from saline exposed animals and the expression was considered statistically significant when fold change was >1.5 and p < .05.

Migration and aortic ring functional assays

The migratory induction potential of serum collected from welding fume exposed rats was assessed in Sprague Dawley rat PCMEs. Cells were suspended in a 96-well plate coated with 0.1% gelatin matrix (Cell Biologics, Catalog No. 6950) for 24 h. After 12 h of serum starvation, a scratch was made in each well using a p200 pipette tip. The debris were washed with warm PBS followed by challenge with endothelial medium without supplements and containing 10% serum from PBS, GMA-MS, or MMA-SS (2 mg/rat) exposed rats. The migratory behavior of endothelial cells while closing the scratch was followed for 6 h using a 4× objective on an ImageXpress high-content imaging system (Molecular Devices, California). Images obtained were quantified for percentage change in area by segmenting the images using NIH Image J (Weka Segmentation Plugin; Arganda-Carreras et al., 2017). The serum from each animal was challenged on 3 wells in a 96-well plate, each treatment group had serum collected from 4 to 6 animals. Serum from saline treated animal’s augmented with 10 ng/ml rat VEGF (VEGF Recombinant Rat Protein, Invitrogen, PRG0114) or TNF-α (TNF-α Recombinant Rat Protein, Invitrogen, PRC3014) were used as positive control and cell culture media with no serum or supplements was used as negative controls.

The multifactorial angiogenesis process was evaluated in the 3D ex vivo aortic ring model, where all the key intermittent steps of angiogenesis occurring in vivo are replicated (Baker et al., 2012). After euthanasia, rat aorta was excised from 12-week-old naïve Sprague Dawley rats. The aorta was cleaned by removing the adipose and adventitial tissue around the aorta. The aorta was further cleaned by placing the aorta in a petri dish filled with opti-MEM (ThermoFisher, Massachusetts; 51985034) and the blood in the lumen was flushed with 1 ml opti-MEM using a 27 g needle (Becton, Dickinson and Co.). The cleaned aorta was sectioned into 1 mm rings and stored at 4 ⁰C in opti-MEM until embedding in a 3D matrix. 100 µl of 2–3 mg/ml growth factor reduced basement membrane matrix (Matrigel; Corning Inc, New York; 354230) on ice was placed in each well of a precooled 24-well plate. Each 1 mm sectioned aortic ring was placed in a well followed by addition of 300 µl of 2–3 mg/ml growth factor reduced basement membrane matrix. The 24-well plate was placed in a 37°C incubator for 30 min to solidify the 3D matrix. The wells containing the rings were fed with 0.5 ml of opti-MEM containing 10% serum from PBS, GMA-MS, or MMA-SS (2 mg/rat) exposed rats. The serum from each animal was tested on 3 rings and each treatment group had serum collected from 4 to 6 animals. Serum from saline treated animals augmented with 10 ng/ml rat VEGF (VEGF Recombinant Rat Protein, Invitrogen, PRG0114) was used as positive control and cell culture media with no serum or supplements was used as negative controls. The rings were monitored daily and after 4 days the vessels in the 3D matrix were fixed with 10% formalin in phosphate buffer (ThermoFisher, 9990244). After excluding wells with damaged rings, the microvessel sprouts from the center of the rings was captured using a 10× objective on an Olympus phase contrast microscope. The % change in area was evaluated by segmenting the images using NIH Image J (Weka Segmentation Plugin; Arganda-Carreras et al., 2017). Alternative commercial algorithms exist for quantification of various parameters in the aortic ring assay (Blatt et al., 2004).

Statistical methods

The in vitro and ex vivo functional assays were performed in triplicates using serum from each animal exposed. The heat maps were presented as log2 transformed mean of the experimental data. Fold regulation and statistical significance (student t test) for the transcriptional array was analyzed on Qiagen RT² Profiler PCR Array Data Analysis web-based platform. Other statistical analyses were performed either using JMP version 12.1.0 (SAS Institute, Cary, North Carolina) or using Prism (Graphpad Inc, California). Normality of the data was analyzed using Shapiro-Wilk test. All the treatments were compared for statistical significance using a one-way analysis of variance (ANOVA) followed by Tukey’s or Newman-Keuls multiple comparison post hoc test. Differences were considered statistically significant at p < .05.

RESULTS

Welding Fume Characterization, Pulmonary Toxicity, and Inflammation

The 2 welding fumes tested were generated with GMA-MS E70S-3 and with MMA-SS ER308-16-1. Inductively coupled plasma-atomic emission spectroscopy (ICP-AES) showed a varied composition in the welding fumes generated (Antonini et al., 2011). GMA-MS was composed of 83% Fe, 15% Mn, 2% Cu, and 0.017% Al, whereas MMA-SS was made up of 57% Fe, 14% Mn, 20% Cr, 8.5% Ni, 0.16% Cu, and 0.085% Al. These welding fume samples had a varied dissolution rate as previously reported (Antonini et al., 2011) with soluble-to-insoluble ratios 0.020 and 0.345 for GMA-MS and MMA-SS, respectively. Morphologically, electron micrography showed both welding fumes to have a chain-like agglomerate structure of primary particles in the ultrafine size range. The formed welding fume particulate can range from ultrafine to fine sizes (Figure 1A). In aqueous dispersions, the hydrodynamic diameter of the particulate was 913 and 710 nm for GMA-MS and MMA-SS, respectively (Figure 1B). The pulmonary toxicity and cell influx (Figures 1C and 1D) after welding fume exposure reported here was previously described by Shoeb et al. (2017). When compare with control animals, there was a statistically significant increase in BALF lactate dehydrogenase (LDH), a marker for pulmonary toxicity, after exposure to both welding fumes. MMA-SS exposure caused greater pulmonary toxicity compared with GMA-MS exposed animals. Polymorphonuclear leukocyte (PMN) influx, a hallmark for pulmonary inflammation, was significantly greater in the lungs of both welding fume exposed animals compared with control, primarily contributing to the increase in total bronchoalveolar lavage cells. There was no difference in PMN influx between GMA-MS and MMA-SS exposed animals. There was no difference in alveolar monocyte/macrophage influx in lungs of animals exposed to either welding fume or PBS exposed animals. Serum was collected from these animals at sacrifice 24 h postexposure. The oxidative status of serum is a known risk factor for cardiovascular disorders (Gawron-Skarbek et al., 2014). We evaluated the total antioxidative capacity of serum from animals exposed to 2 different welding fumes using the ferric reducing ability of serum assay (FRAS). When compared with animals exposed to PBS, there was approximately 34% increase in serum antioxidants with GMA-MS exposure and approximately 14% increase with MMA-SS. However, there was no statistical difference between the antioxidant status due to high coefficient of variation (25% for GMA-MS and 18% for MMA-SS) within the animals in the groups (Figure 1E).

Cytokine Profiling

As expected, there was significant cytokine production measured in the BALF due to exposure to both welding fumes (Figure 2). When compared with control, out of 27 cytokines screened, GMA-MS exposure caused a significant induction in 13 cytokines, whereas MMA-SS induced 16 cytokines. The upper Venn diagram shows 10 statistically significant proteins were commonly altered with GMA-MS and MMA-SS in BALF. This suggests commonality in pathways perturbed with both the welding fumes. The lower Venn diagrams illustrate common significantly expressed proteins in BALF and serum with each welding fume. With MMA-SS exposure, apart from IL-6, 3 of the proteins were also common between BALF and serum. In general, the effects were slightly more pronounced in the MMA-SS exposed animals compared with GMA-MS exposed animals. Exposure to GMA-MS and MMA-SS caused an increase in proinflammatory cytokines, like IL-1β and TNF-α. The proinflammatory and antiangiogenic chemokine IP-10 (IFN-γ inducible protein, CXCL10) was highly expressed (approximately 30-fold change compared with control) in the lavage fluid of both GMA-MS and MMA-SS exposed animals. Another angiogenic neutrophil chemokine LIX, that regulates neutrophil homeostasis, and, like IP-10, is induced following simulation with proinflammatory cytokines, like TNF-α, was also highly expressed in lavage fluid of both GMA-MS (13-fold from control) and MMA-SS (18-fold from control) exposed animals. MMA-SS caused a statistically significant and approximately 2-fold change in IL-10 levels. Both welding fumes caused an increase in monocyte/macrophage chemokines CCL2/MCP-1, CCL5/RANTES, and CCL3/MIP1a. Exposure to MMA-SS caused approximately 2- to 10-fold increase in expression levels of monocyte/macrophage chemokines compared with GMA-MS exposure. There was a 50- and 80-fold change in IL-17a production in the lavage fluid due to GMA-MS and MMA-SS exposure, respectively, compared with control rats, an indication that there was a substantial Th17 response.

Although there was a statistically significant change in approximately 48% and 60% of the total 27 cytokines screened after exposure to GMA-MS and MMA-SS, respectively in BALF, the same proteins screened in the serum revealed no change or a minor change in 4 proteins. MMA-SS pulmonary exposure caused a statistically significant alteration in serum levels of IP-10, CCL5/RANTES, leptin, and IL-6, and only 2 of those proteins (leptin and IP10) were directionally similar to changes in the BALF.

Screening Primary Endothelial Cells for Evaluating Functional Changes

Endothelial cells form a continuous monolayer on the interior surface of the blood vessels with a steady exposure to the contents of the circulation. These cells play an important role in maintaining the physiological balance of the vasculature by regulating permeability, vascular tone, recruitment of circulatory cells, as well as modulating clotting and formation of new blood vessels. Due to their pivotal role in sensing and responding to various physiological changes (Givens and Tzima, 2016; Oberleithner et al., 2010; Pohl, 1990; Zaragoza et al., 2012), endothelial cells have been used as a translational screening tool for evaluating cellular activation and signaling and vascular dysfunction (Agarwal et al., 2013; Channell et al., 2012; Cung et al., 2015; Kvietys and Granger, 1997). In order to evaluate at a molecular level the effects of circulatory factors after pulmonary exposure to various welding fumes on the endothelium, naïve PCMEs were challenged for 4 h with serum collected from welding fume exposed animals. The transcriptional response of the endothelial cells was determined using the rat endothelial cell biology RT² profiler PCR Array. This endothelial pathway-specific array allowed us to screen 84 genes related to various endothelial functions. The welding fume serum treated endothelial cells differentially expressed (fold change >1.5, p < .05) in total 38 of the 84 genes screened. Serum from MMA-SS and GMA-MS exposed animals induced an orthogonal activation profile in endothelial cells (Figure 3A). Although serum from MMA-SS exposed animals caused a change in 31 genes (approximately 37% of the total genes screened), serum from GMA-MS animals caused a differential expression of only 7 genes (Figure 3B). Figures 3C–F depict heat maps of fold regulation from control of the differentially expressed genes, clustered with respect to their role in endothelial function. In all 4 clusters of cell adhesion and migration, vasoconstriction and vasodilation, inflammatory response, and angiogenesis, serum from the MMA-SS exposed animals caused upregulation in gene expression suggesting possible functional changes. Serum from GMA-MS exposed animals caused no change or slight down regulation compared with endothelial cells treated with control serum from rats exposed to the PBS control.

Validation of the Predicted Functional Outcomes

Based on the transcriptional response in primary microvascular endothelial cells, it was predicted that the serum circulatory factors from MMA-SS exposed rats, but not from GMA-MS exposed rats, can cause altered endothelial function. These predictions were evaluated by performing functional assays for cell migration and angiogenesis in a single endothelial cell culture system as well as in a more complex multicellular architecture of a tissue using ex vivo culture, respectively. The scratch assay was employed to test the migration potential of endothelial cells challenged with serum from the different welding fumes (Figure 4). A second functional validation for angiogenesis was done by culturing, ex vivo, a naïve aorta in a 3D extracellular matrix challenged with serum from exposed animals (Figure 5).

Cell Migration

As depicted in Figure 4A, the scratch assay was adapted for high-throughput screening wherein the primary microvascular endothelial cells were plated on a gelatin-coated 96-well plate left to grow and adhere for 24 h to form a monolayer. After 12 h of serum starvation, a scratch then was made in the monolayer of each well using a p200 pipette tip. After the scratch was made, the wells were washed with warm PBS to remove the floating cell debris and challenged with fresh endothelial media containing 10% rat serum from animals exposed to GMA-MS, MMA-SS, or PBS control for 6 h. Serum from each animal was challenged on triplicate wells, and each treatment group had serum from 4 to 6 animals. As a positive control, cells were challenged with rat serum from saline exposed animals augmented with 10 ng/ml of VEGF or 10 ng/ml of TNF-α, 2 known stimulators for endothelial cell migration. As a negative control, cells were challenged with media containing no serum. The scratches were evaluated after 6 h.

Although the gap varied, serum from all treatment groups triggered endothelial cells to migrate and close the scratch. As predicted by the endothelial transcriptional response earlier, challenge with serum from GMA-MS exposed animals did not change the migratory response compared with serum from control animals (Figure 4B). Serum from MMA-SS exposed animals was a potent motogen, causing endothelial cells at the leading edge of the scratch to migrate at a faster pace, and the % change in the gap over 6 h was significantly greater compared with serum from PBS challenged animals as well as serum from GMA-MS challenged animals (Figs. 4B and 4C). As expected, the positive controls of serum from PBS treated animals augmented with 10 ng/ml of either VEGF or TNF-α caused enhanced endothelial cell migration.

Angiogenesis

Angiogenesis in vivo is a complex function involving endothelial cells, smooth muscle cells, and pericytes. Together, these cells coordinate and execute various distinct phases in the angiogenic process like degradation of extracellular matrix, cell migration, proliferation, and structural reorganization. We validated the functional prediction of angiogenesis from the endothelial cell transcriptional response in an ex vivo organ culture model (aortic ring assay) to test the response of the serum in a more realistic multicellular architecture where all cell types are represented and all angiogenic phases are recapitulated (Figure 5).

The aortic ring angiogenesis assay previously described (Baker et al., 2012; Go and Owen, 2003; Nicosia and Ottinetti, 1990) was slightly altered to adapt to our screening approach. In this quantitative method of studying angiogenesis, the thoracic aorta from a naïve rat was extracted, and after removing the fat and connective tissue, the aorta was segmented into 1 mm rings and embedded into a 3D growth factor reduced basement membrane extracellular matrix in a 24-well plate. The embedded aortic rings were then challenged with fresh medium containing 10% (volume/volume) serum from rats exposed to GMA-MS, MMA-SS, or PBS control (Figure 5A). As a positive control, serum from saline exposed rats with 10 ng/ml rat VEGF was used. After 4 days of incubation, photomicrographs of the vascular outgrowth from the rings were captured using a phase contrast microscope. Morphometric evaluation of the change in area due to sprouting vessels was segmented and quantified.

VEGF, a known angiogenic factor (Zhu et al., 2003), caused an increased sprouting and vascular networks. As a negative control, embedded vessels with cell culture media only, containing no serum or supplements, resulted in no microvessel sprouting. All the serum challenged rings caused sprouting and vascular formation (Figure 5B). There was no statistically significant change in angiogenesis between aortic rings challenged with serum collected from animals exposed to PBS vehicle or GMA-MS fume (Figure 5C). As predicted from the endothelial array, serum from MMA-SS exposure caused a significant change in angiogenesis in the aortic rings.

DISCUSSION AND CONCLUSION

Although studies conducted over the past decade have established that pulmonary exposure to particulate insults can induce systemic toxicity, the precise mechanism of how this occurs is still under debate. Previous work from our group showed pulmonary exposure to stainless steel welding fume caused an increase in oxidative stress, a decrease in immune response to secondary challenge of bacterial components, and a general dysfunction in circulating leukocytes, highlighting effects of welding fume exposure outside the lung (Erdely et al., 2014; Shoeb et al., 2017). Additionally, aortic and cardiac inflammation with increased atherosclerotic lesion area was measured following stainless steel welding exposure (Erdely et al., 2011a,c). Several human health effects studies indicate an increased risk for cardiovascular disease following exposure to welding fumes (Fang et al., 2009; Kim et al., 2005; Lai et al., 2016; Li et al., 2015; Sjögren et al., 2002). In this study, we set forth to examine if pulmonary exposure to distinct welding fumes alters the composition and bioactivity of circulating factors in the serum.

Exposure to GMA-MS and MMA-SS led to a range in the degree of pulmonary injury, with MMA-SS causing the maximum pulmonary response in terms of cytotoxicity and inflammatory cytokine release, although both exposures resulted in significant injury and inflammation compared with control. This was not unexpected as stainless steel welding fume has consistently been more inflammatory compared with mild steel welding fumes (Taylor et al., 2003). Of the 27 BALF proteins measured, 16 and 13 proteins were differentially induced in the lung lavage of rats exposed to MMA-SS and GMA-MS, respectively. The 10 proteins that were differentially expressed in the lung were common with both MMA-SS and GMA-MS, suggesting that these were expressed as a response to particulate matter exposure to the lung. The pulmonary results indicate that general lung inflammation (eg, cytotoxicity, cellular influx, and cytokine production) was not a great predictor for establishing whether circulating factors were present to induce altered endothelial function, given the striking difference observed in the endothelial cell response between the GMA-MS and MMA-SS groups. These results were consistent with studies measuring cardiovascular dysfunction with minimal to no pulmonary inflammatory response (Nurkiewicz et al., 2008).

Although lung inflammation was ongoing, complementary cytokine induction in the circulation was not observed. Certainly, a temporal effect was appreciated as we have previously shown that primary cytokine levels were increased at 4 h, but not 24 h, in the serum after a pulmonary exposure (Erdely et al., 2009, 2011b). Given that both exposures can induce pulmonary injury and inflammation, but only MMA-SS causing altered endothelial cell function was enigmatic. Although a temporal effect was plausible, vascular studies with serum from animals exposed to high aspect ratio multi-walled carbon nanotubes (MWCNTs) showed that there was no dose dependency at 4 and 24 h following exposure and the responses were inconsistent with general pulmonary inflammation and serum levels of primary cytokines (Aragon et al., 2016). Mostovenko et al. (2019) demonstrated that pulmonary exposure to MWCNT caused an upregulation in matrix proteases with an increase of related peptide fragments in the circulation. These peptide fragments were shown to be the bioactive component that influenced endothelial and vascular function. It should be noted that serum is a complex dynamic mixture consisting of cytokines, chemokines, peptides, growth factors, lipoproteins, hormones, extracellular vesicles, etc. Identifying a single causative agent may be futile. This is further complicated by the fact that several of these stimulatory components fluctuate around basal concentrations. These components may be below the threshold required for activating a signaling cascade to induce a measurable effect in endothelial cells, but a combination of ligands can achieve activation of the same signaling cascade or multiple cascades (Channell et al., 2012; Delgoffe et al., 2011).

Endothelial cells are robust and sensitive sensors that continually sample and control vascular function by responding to stimuli from the blood. Evaluating the molecular and functional response of naïve endothelial cells or vascular tissue using total serum provides a quick screening strategy for determining the bioactivity arising due to an alteration in circulatory factors. The transcriptional response of 84 genes related to various endothelial functions in naïve rat PCMEs challenged for 4 h with serum obtained from rats exposed to GMA-MS, MMA-SS, or PBS illustrated an orthogonal mechanism of activation by the 2 welding fumes. The cells exposed to serum from MMA-SS exposed animals had 31 differentially expressed genes, whereas cells challenged with serum from GMA-MS exposed animals had 7 differentially expressed genes. Unlike the cytokines in the lavage fluid, there were no common differentially expressed genes between MMA-SS and GMA-MS groups. Clustering the genes, with respect to their endothelial function, illustrated an upregulation of genes in all the functional clusters when challenged with serum from MMA-SS exposed animals but not with GMA-MS or PBS exposed animals. The inflammatory response cluster was consistent with previous studies of diesel or engineered nanomaterial exposures, which indicated an inflammatory potential of the serum following a pulmonary exposure Aragon et al., 2017; Channell et al., 2012; Shen et al., 2018). Additionally, the vascular function cluster was not unexpected as welding studies in vivo and in humans suggest vascular dysfunction (Erdely et al., 2011a; Fang et al., 2009; Wong et al., 2015; Zeidler-Erdely et al., 2014). Two additional clusters of predicted functional outcomes, cell migration and angiogenesis, were further validated in this study using functional assays.

As predicted, MMA-SS serum challenged endothelial cells had an enhanced migratory response and closed the cellular “scratch” faster in a 2D scratch wound healing assay. The result contrasted with what was observed for endothelial cells challenged with serum collected from mice exposed to MWCNT (Aragon et al., 2017). Although the difference is not inherently obvious, the reason may be due to mechanism of action. Although exposure to MWCNT induced specific bioactive proteolytic circulating byproducts due to induction of matrix proteases (Mostovenko et al., 2019), metal oxide particulate exposure does not induce a pronounced matrix protease response (Andujar et al., 2014; Zeidler-Erdely et al., 2010). We validated the angiogenesis prediction using an ex vivo, 3D aortic ring assay. The aortic ring assay for evaluating angiogenic potential is a physiologically relevant model that bridges the gap between single cell in vitro models and more multicellular-physiologically complex in vivo models, considering all the relevant cell types that play a key role in angiogenesis (Baker et al., 2012). The aortic ring assay results indicated that serum from MMA-SS exposed animals was more potently angiogenic. Increased angiogenesis was not necessarily indicative of an adverse outcome but could be contraindicated in the context of welding fume being a Group 1 carcinogen. It should be noted that the angiogenic potential of serum from MMA-SS was predicted based on the transcriptional response from 2D naïve primary microvascular endothelial cells. Although endothelial cells are the primary drivers in angiogenesis, other cells such as smooth muscle cells have an important role in the process, yet the prediction from endothelial cells was valid even in a more complex multicellular environment. This once again affirmed the use of endothelial cells as a quick screening tool for evaluating bioactivity of circulating factors.

MMA-SS and GMA-MS welding fumes have a varied composition of metals with a soluble-to-insoluble ratio of 0.345 and 0.020, respectively (Antonini et al., 2011). Kinetic studies indicated that the soluble Cr from MMA-SS welding fumes quickly translocated to the circulation and is significantly elevated at 1 d after exposure (Antonini et al., 2010). For the MMA-SS fume, 87% of the soluble fraction was found to be Cr. The toxicity of soluble Cr is well documented (Cohen, 2004; Dayan and Paine, 2001). Soluble metal components like Cr in MMA-SS may have contributed to the differential endothelial response and functional outcomes observed. Oxidatively damaged proteins and posttranslational modifications like nitrosylated or glutathionylated proteins can alter cellular signaling, change endothelial function and cause several vascular disorders (Chen et al., 2010; Duan et al., 2016; Esterbauer et al., 1992; Hoffmann et al., 2001; Irani, 2000; Stamler et al., 1992; Witztum and Steinberg, 1991). Studies in Cr-exposed workers showed that plasma Cr levels correlated with increased redox-sensitive lipid peroxidation products and decreased in thiol antioxidants (Goulart et al., 2005). Soluble, reactive metals, from welding fumes similar to what was classically described for particulate matter (Costa and Dreher, 1997), may indirectly contribute to endothelial dysfunction by increasing posttranslation protein modifications in the circulation. This study only interrogated the bioactivity of circulatory factors after 24 h of pulmonary exposure to various welding fumes or PBS. Welding fumes, contingent on composition, are known to dissolve and translocate systemically at various rates. Evaluating the circulation for endothelium bioactivity temporally can help us understand better the circulatory alteration kinetics.

The observed alteration in endothelial function that occurred from only the MMA-SS exposure was likely due to factors present in the circulation. Our study was not designed to pinpoint a single causative agent. The outcomes were likely due to a combination of new or altered levels of cytokines, chemokines, or peptides and modified redox sensitive proteins or lipids due to MMA-SS exposure. Alternatively, or additionally, soluble metals due to dissolution of metal components from the welding fume in the lung may quickly move into the circulation and directly act on the endothelial cells or circulating components contributing to altered endothelial function. In this work, we showed that pulmonary exposure to some welding fumes can alter circulatory factors likely contributing to systemic toxicity. Irrespective of the precise mechanism of altered endothelial function induced by circulating factors, the methodological approach offers a screening tool of exposed populations to predict functional endothelial-dependent changes occurring due to various occupational pulmonary exposures.

SUPPLEMENTARY DATA

Supplementary data are available at Toxicological Sciences online.

FUNDING

NIOSH-Nanotechnology Research Center (Project No. 939ZXFL).

DECLARATION OF CONFLICTING INTERESTS

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Disclaimer: The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention. Mention of brand names does not constitute product endorsement.

REFERENCES