Metabolic alterations and mitochondrial dysfunction in human airway BEAS-2B cells exposed to vanadium pentoxide

Abstract

Vanadium pentoxide (V⁺⁵) is a hazardous material that has drawn considerable attention due to its wide use in industrial sectors and increased release into environment from human activities. It poses potential adverse effects on animals and human health, with pronounced impact on lung physiology and functions. In this study, we investigated the metabolic response of human bronchial epithelial BEAS-2B cells to low-level V⁺⁵ exposure (0.01, 0.1, and 1 ppm) using liquid chromatography-high resolution mass spectrometry (LC-HRMS). Exposure to V⁺⁵ caused extensive changes to cellular metabolism in BEAS-2B cells, including TCA cycle, glycolysis, fatty acids, amino acids, amino sugars, nucleotide sugar, sialic acid, vitamin D₃, and drug metabolism, without causing cell death. Altered mitochondrial structure and function, and disruption in glycolysis were observed with as low as 0.01 ppm (0.2 μM) V⁺⁵ exposure. In addition, decreased level of E-cadherin, the prototypical epithelial marker of epithelial-mesenchymal transition (EMT), was observed following V⁺⁵ treatment, supporting potential toxicity of V⁺⁵ at low levels. Taken together, the present study shows that V⁺⁵ has adverse effects on mitochondria and the metabolome which may result in EMT activation in the absence of cell death. Furthermore, results suggest that high-resolution metabolomics could serve as a powerful tool to investigate metal toxicity at levels which do not cause cell death.

Graphical Abstract

1. Introduction

Vanadium (V) is one of the leading heavy metals causing environmental contamination, primarily from oil refinery and mining activity [1], steel processing and manufacturing [2], smoking [3], and vehicle emission [4]. V is identified as a hazardous material according to the U.S. Department of Transportation [5], and may be poisonous by ingestion, inhalation, or skin absorption [6, 7]. Attention has been drawn to the human health risks of V compounds over the past few decades due to the increased release of anthropogenic V into air, water, and soil [6–12]. Mean exposure to V in Eastern cities in United States was found to be 620 ng/m³[13]. Much higher concentrations were measured in occupational settings. For instance, vanadium exposure was found to be between 10 and 500 μg/m³ for boilermakers[14]. The presence of V in food and water is generally low, but higher concentrations have been observed in drinking water with a maximum greater than 100 ppb (2 μM) depending on geographical location [8], and up to 899 ppm in drinking water as a result of iron pipe corrosion [15]. Over 1000 ppb of V were found in a variety of daily foods including many vegetables and grains [9]. Moreover, exposure to V could be substantially increased through dietary supplements [16], leading to thousands times higher than the normal daily intake.

Vanadium pentoxide (V⁺⁵) is the most relevant environmental form of V, and it possesses strong oxidative potential [17]. V⁺⁵ is classified by the International Agency for Research on Cancer (IARC) as possibly carcinogenic to humans. V⁺⁵ can cause severe health issues in animal and human lungs [18–26], with decreased pulmonary reactivity [20], increased respiratory tract irritation [27], pulmonary inflammation [18, 19, 22, 24], airway fibrosis [23] and interstitial pulmonary fibrosis [28]. Earlier studies showed that the LD₅₀ of V⁺⁵ ranged from 10 to over 100 mg/kg body weight (bw) in rats and mice [13, 29]. Specifically, acute LD₅₀ values for V⁺⁵ were 41 mg V/kg bw in rats and 31.2 mg V/ kg bw in mice following 14 days oral exposure [30]. A 96-h LD₅₀ was estimated to be 119.0 mg/kg bw in rabbits through drinking water exposure [31]. Moreover, recent studies show that V⁺⁵ exhibited considerable cytotoxicity to lung carcinoma epithelial A549 cells [32], human lung fibroblasts [33–36], Chinese hamster lung fibroblast V79 [37], and pulmonary macrophages [38]. Earlier study showed that the exposure to V₂O₅ at 10 ppm V and 20 ppm V, which ionizes to hydrated forms of V⁺⁵ in solution at 100 μM and 200 μM, respectively, caused induction of cyclooxygenase-2 expression in human bronchial epithelial cells (BEAS-2B) [39]. However, those studies heavily focused on high level exposure, and little information is available on the V⁺⁵ at sub-ppm levels.

Our recent study found that dietary exposure to V⁺⁵ can significantly increase the V deposition in mouse lungs, further causing oxidative stress and eliciting inflammatory signaling [40]. Importantly, metabolic perturbation in mouse lungs aligned well with perturbation seen in lung fibroblasts, including major lipid metabolism and mitochondrial and redox pathways, which might be related to the oxidative stress, protein S-glutathionylation, and cellular senescence induced by V⁺⁵ exposure [33]. However, little is known about the contributing role of airway epithelia in the process of mitochondrial oxidation and inflammatory signaling following V⁺⁵ exposure despite the critical role of airway epithelia in maintaining normal lung physiology [41] as well as the critical function of mitochondria in airway epithelia [42]. In an earlier study on nano-sized V₂O₅ particulate exposures [43], we used BEAS-2B cells, an established airway epithelial cell model for cellular and molecular mechanisms involved in lung inflammation and other lung injuries [44–48] and responses to hazardous exposures [49, 50]. Expanding on our prior endeavors of low-level V⁺⁵ effects in cell and lung models[33, 40], the present study was designed to test effects of dissolvable V⁺⁵ on mitochondrial stress and metabolism in BEAS-2B cells and compare to mouse lung and human lung fibroblast cell models [33, 40]. In this study, we exposed BEAS-2B cells to V⁺⁵ at 0.01, 0.1, and 1 ppm for 24 h, a range of concentrations we previously determined to mimic human exposures from air, drinking water, food, and dietary supplements [40]. A dose-response study with the same V⁺⁵ exposures levels (0.01, 0.1, and 1 ppm) was performed with measures for absolute cellular V⁺⁵ content, cell viability, high resolution metabolomics, mitochondrial measurements and morphology examination by transmission electron microscopy.

2. Methods and Materials

2.1. Chemicals

Vanadium pentoxide (V⁺⁵), trichloroacetic acid, perchloric acid, sodium iodoacetate, KOH, potassium tetraborate, dansyl chloride, methanol, and H₂O₂ were purchased from Sigma-Aldrich (St. Louis, MO). Dithiothreitol was purchased from Bio-Rad (Hercules, CA). Boric acid was purchased from J.T.Baker^™ (Phillipsburg, NJ). γ-Glutamylglutamate (γ-Glu-Glu) was purchased from MP Biomedical, Inc. (Solon, OH). Nitric acid was purchased from MilliporeSigma (Burlington, MA). An internal standard mixture for mass spectral analysis consisted of [¹³C₆]-D-glucose, [¹⁵N]-indole, [2-¹⁵N]-L-lysine dihydrochloride, [¹³C₅]-L-glutamic acid, [¹³C₇]-benzoic acid, [3,4-¹³C₂]-cholesterol, [¹⁵N]-L-tyrosine, [trimethyl-¹³C₃]-caffeine, [¹⁵N₂]-uracil, [3,3-¹³C₂]-cystine, [1,2-¹³C₂]-palmitic acid, [¹⁵N, ¹³C₅]-L-methionine, [¹⁵N]-choline chloride, and 2’-deoxyguanosine-¹⁵N₂,¹³C₁₀-5’-monophosphate, all purchased from Cambridge Isotope Laboratories, Inc (Andover, PA). All reagents were analytical grade or above unless otherwise stated.

2.2. Cell culture

Human bronchial epithelial cells (BEAS-2B, ATCC Catalog# CRL-9609) were cultured at 37 °C and 5% CO₂ in serum-free Bronchial Epithelium Basal Medium (BEBM, Catalog # CC-3177, Lonza, Walkersville, MD) containing BEGM^® SingleQuots^® (Catalog # CC-4175, Lonza, Walkersville, MD) until 80–90% confluent. Cells were treated with media containing 0, 0.01, 0.1, or 1 ppm added vanadium pentoxide and incubated at 37 °C and 5% CO₂ for 24 h. Cells were collected at the end of exposure for assessment of cell viability, metal analysis using inductively-coupled plasma mass spectrometry (ICP-MS), high-resolution metabolomics using liquid chromatography-mass spectrometry (LC-MS), measurement of GSH, GSSG, and mitochondrial membrane potential, and evaluation by transmission electron microscopy (TEM).

2.3. Cell Viability Assay

Cytotoxicity of V⁺⁵ in BEAS-2B cells was quantified using the Cell Proliferation Reagent WST-1 (Roche, Catalog # 11644807001, Rotkreuz, Switzerland) following the manufacturer’s instruction. In brief, BEAS-2B cells were seeded in 96-well plate with 8–16 replicates for each treatment and grown to 70–80% confluency. Cells were then exposed to 0 to 200 ppm V⁺⁵ for 24 h. After removing cell media, 100 μL fresh media containing 10 μL WST-1 reagent were added to each well. Cells were further incubated at 37 °C and 5% CO₂ for 4 h. Cells were then rinsed with PBS and stained with 100 μL 1:2500 Hoechst 33342 in PBS at 5% CO2 and 37 °C for 5 min. Next, cells were rinsed with PBS again and read using an ELISA plate reader (SpectraMax M2, Molecular Devices, San Jose, CA). The absorbance of the formazan product of WST-1 reagent was measured at 450 nm relative to a reference wavelength at 610 nm. Cell nuclei staining was measured at excitation and emission wavelengths of 350 and 461 nm, respectively. The ratio of measurements from WST-1 assay and Hoechst 33342 staining was used for final cytotoxicity.

2.4. Measuring ⁵¹V levels using Inductively Coupled Plasma Mass Spectrometer (ICP-MS)

ICP-MS (iCap Q, ThermoFisher Scientific, Waltham, MA) was used for quantification of ⁵¹V. Briefly, cells were plated and treated in 10-cm plates. After 24 h of exposure, 1 mL media were collected, and cells were rinsed with ice-cold PBS. Cells were lysed with 400 μL pure water by freeze–thaw cycling and homogenization (EpiShear Q120AM probe sonicator, Active Motif, Carlsbad, CA). Cell lysate and media were randomized within sample type and microwave digested in closed vessels using 40% HNO₃ and 9% H₂O₂ (MARS 6, CEM Corp., Matthews, NC). Samples were digested alongside a blank sample, and samples were supplemented with scandium (Sc), indium and lutetium for a final concentration of 16 ppb for each standard.

The samples were then cooled to room temperature, diluted to 10 mL with 2% nitric acid, capped to prevent contamination, and analyzed the following day. Samples were analyzed in triplicate. Washes using 1% nitric acid (reagent grade) were performed at the beginning and the end of the run, as well as every 14 samples. The iCAP Q was equilibrated in collision cell mode using kinetic energy discrimination (KED) with helium as the collision gas (30 psi) and passed quality control sensitivity tests before each run. Sample intensities were normalized relative to blank. Measured ⁵¹V intensities were normalized relative to detected ⁴⁵Sc internal standard to account for any shift in sensitivity. Average recovery of ⁴⁵Sc was 115% (standard deviation: 7%) relative to the analytical blank. A linear standard curve for ⁵¹V was prepared using a vanadium standard (1000 mg/L, 2% HNO₃; Ricca Chemical; Arlington, TX) and prepared on the range of 0.25 to 64 ppb in the same run as the samples with a linear correlation coefficient of > 0.9999. A quality control sample of 4 ppb was utilized at the end of the analyses to determine accuracy and precision of the standard curves. ICP-MS procedures conformed to previously stated accuracy (100 ± 10%) and precision standards (relative standard deviation < 12%). Cell lysate data were normalized to total protein.

2.5. Liquid chromatography-high-resolution mass spectrometry (LC-HRMS)

BEAS-2B cells were collected in 300 μL extraction solution containing 200 μL acetonitrile and 100 μL H₂O and internal standards as described previously [51, 52]. Extracted metabolites were analyzed using a High Field Q Exactive mass spectrometer (85–1275 m/z, Thermo Fisher, Waltham, MA) with an ultra-high-field Thermo Scientific^™ Orbitrap^™ detector, coupled with hydrophilic interaction liquid chromatography with positive electrospray ionization (ESI) [HILIC (+)] and C18 reversed phase chromatography paired with negative ESI [C18 (−)]. Samples were run in triplicate and in batches of 20 samples.

Metabolic features were obtained and characterized by high-resolution mass to charge (m/z), retention time (rt) and abundance, which were extracted using apLCMS [53] and xMSanalyzer [54] and filtered to retain features with nonzero values in >70% in all samples and >80% in each group. Data were median summarized among replicates, log2 transformed and quantile normalized for biostatistics and bioinformatics analyses. NIST SRM1950 and Q-Standard 3 (Qstd3) were used for quality control and quality assurance [55].

2.6. Glutathione (GSH), glutathione disulfide (GSSG) and steady-state redox potential

GSH and GSSG were measured using high-performance liquid chromatography (HPLC) with a gradient HPLC module 2695 (Waters, Milford, MA), a multiwavelength fluorescence detector 2475 (Waters, Milford, MA) at excitation wavelength of 335 nm and emission wavelength of 518 nm and Empower software using SUPELCOSIL^™ LC-NH2 HPLC Column (25 cm × 4.6 mm, 5 μm) [56]. In brief, cells were rinsed twice with ice cold PBS and collected in 375 μL ice cold 5% perchloric acid containing 0.2 M boric acid and 10 μM γ-Glu-Glu internal standard. Cells were centrifuged at 14,000×g, 4 °C for 2 min. 300 μL of the supernatant were then transferred to new tubes containing 60 μL of 50 mM sodium iodoacetate. pH was adjusted to 9 using KOH-tetraborate (1 M KOH with 0.38 M potassium tetraborate) and incubated at room temperature for 20 min. 300 μL of 75 mM dansyl chloride were added, and all tubes were left in the dark overnight. The next day, 500 μL of chloroform were added into the solution. Extracted solutions were vortexed and centrifuged at 14,000×g, 4 °C for 2 min. 50 μL of the upper fraction were injected and concentrations of GSH and GSSG were calculated relative to the internal standard γ-Glu-Glu and normalized to the protein concentration. The individual concentrations, estimated for cellular molar values, were used with the Nernst equation to calculate steady-state redox potentials for GSH/GSSG (E_hGSSG/GSH) with E_o = −240 mV, pH 7.0.

2.7. Mitochondrial membrane potential assay

JC-1 probe (Invitrogen, Waltham, MA) was used for measuring mitochondrial membrane potential. BEAS-2B cells were seeded in 96 well plates at a density of 1 × 10⁴/mL and grown to 70–80% confluency. After exposure, cells were rinsed twice with PBS and then incubated with JC-1 (10 μg/mL) for 20 min at 37 °C. The green JC-1 monomer fluorescence was observed at emission 530 nm with excitation 485 nm, and the red JC-1 aggregate fluorescence was observed at emission 590 nm with excitation 535 nm using an ELISA plate reader (SpectraMax M2, Molecular Devices, San Jose, CA). Fluorescence images were taken with fluorescent microscopy (BioTek Lionheart FX, Winooski, VT). BEAS-2B cells treated with H₂O₂ (20 μM) for 20 min were used as positive controls.

2.8. Western blotting

The level of E-cadherin in BEAS-2B cells was examined using Western blot analysis. In brief, after exposure to V⁺⁵ at the indicated conditions, BEAS-2B cells were collected and lysed in Pierce^® RIPA buffer containing protease inhibitor (cOmplete^™, Mini, EDTA-free Protease Inhibitor Cocktail, Roche) and 1 mM PMSF. Protein concentration was quantified using BCA assay. 20 μg of protein was loaded for Western blot analysis with primary antibodies specific for E-cadherin and β-actin (Cell Signaling Technology, Inc., Danvers, MA), and protein levels were imaged and analyzed as previously described [57].

2.9. Transmission Electron Microscopy

Cellular organelles of BEAS-2B were imaged using TEM. After exposure, cells were rinsed, fixed, and processed as described previously [43]. Ultrathin sections (80nm) were cut and then stained with 5% uranyl acetate and 2% lead citrate and imaged on a JEOL JEM-1400 TEM (JEOL Ltd., Japan) equipped with a Gatan US1000 2k×2k CCD camera (Gatan, Pleasanton, CA) at 80 kV.

2.10. BCA protein assay

Protein concentration was measured in 5 μL of cell samples using Pierce^™ BCA Protein Kit (Thermo Scientific) in a 96-well plate reader. After adding 200 μL of working reagent to each well, mixing and incubation for 30 min, absorbance was measured at 562 nm (SpectraMax M2, Molecular Devices, San Jose, CA).

2.11. Statistics

Quantification data were analyzed using Matlab R2021a (MathWorks, Inc., Natick, MA) and graphs were made using OriginPro 2021b (OriginLab Corp., Northampton, MA). Results are presented as mean ± standard error. One-way ANOVA with post hoc testing using Fisher’s Least Significant Difference (LSD) was performed to obtain statistical significance. Three replicates were used for ICP-MS, six replicates were used for metabolomics and GSH redox measurements, 16 replicates were used for mitochondrial membrane potential assay, and eight replicates were used for cytotoxicity testing. Statistical p values < 0.05 were considered significant.

Statistical significance of metabolic features between groups was computed using one-way ANOVA (LIMMA) and a linear regression model (LMREG) based feature selection using xmsPANDA (https://github.com/kuppal2/xmsPANDA, last accessed Oct. 15, 2022). Disturbed metabolites from LIMMA analysis were selected at p < 0.05 and fold change > 2 for initial comparison between control and each individual exposure group, while metabolites from LIMMA and LMREG analyses were selected at p < 0.05 for follow-up multiple group comparison. Selected significant metabolites were then used for hierarchical cluster analysis (HCA), partial least squares-discriminant analysis (PLS-DA), and correlation analysis using MetaboAnalyst [58], and metabolic pathway enrichment using mummichog v.2.6 [52, 59] with 10 ppm maximum tolerance in m/z and 1000 permutations [60]. Metabolite annotation was carried out using xMSannotator [60]. xMSannotator employees a multilevel scoring algorithm to determine annotation for metabolic features, with confidence score from 0 (only accurate mass match) to 3 (high confidence), based on adduct/isotope patterns and elemental or abundance ratio checks. Annotations were made against Kyoto Encyclopedia of Genes and Genomes (https://www.kegg.jp, last accessed Oct. 15, 2022) and the Human Metabolome Database (http://www.hmdb.ca/, last accessed Oct.15, 2022). Annotated metabolites with a score of 2 or greater were selected for further analyses [61].

3. Results

3.1. Dose-dependent Cytotoxicity Following V⁺⁵ Exposure

Cytotoxicity studies of V⁺⁵ at 24 h showed that LC₅₀ in BEAS-2B cells was 8.03 ppm V⁺⁵ (Fig. 1A), and no cell death was apparent at 1 ppm (20 μM) V⁺⁵. Given this result, to study the effects of V⁺⁵ without cell death, we used exposures at or below 1 ppm V⁺⁵ for 24 h (Fig. 1A). ICP-MS showed a dose-dependent cellular uptake of vanadium, with 6.2 ± 1.4, 25 ± 4, and 160 ± 11 ng V/mg protein following 0.01, 0.1, and 1 ppm V⁺⁵ exposure, respectively, compared to 2.6 ± 0.8 ng V/mg protein in control (Fig. 1B). Measurement of V content in the culture medium showed that most V remained in the culture medium at the end of V⁺⁵ exposure for 24 h, i.e., 0.0098 ± 0.004, 0.094 ± 0.01, 0.870 ± 0.014 ppm V in cell medium following 0.01, 0.1, and 1 ppm V⁺⁵ exposure, respectively, compared to 0.0026 ± 0.0003 ppm V in the control group (Fig. S1).

Fig. 1.

Cytotoxicity and cellular uptake of V⁺⁵ in BEAS-2B cells. (A) Dose-response with concentrations from 0 to 200 ppm. (B). Cellular V (ng V/mg protein) measured by ICP-MS. Statistical significance for each condition vs. control is indicated by asterisks, with ***p < 0.001.

3.2. Metabolic Disruption Induced by V⁺⁵ Exposure

Metabolomic analyses for V⁺⁵ exposure at each dose were performed by comparing with vehicle control (without V⁺⁵ exposure) and difference in abundance of metabolites are shown by volcano plots (Fig 2A). As shown in Fig. 2A, 2B, and Fig. S4A, S4B, there were clear metabolic disruption by V⁺⁵ at all doses. Fig 2B and Fig. S4B shows that a total of 265 and 97 metabolites analyzed by HILIC(+) and C18(−), respectively, were altered by 1 ppm compared to control (p < 0.05, fold change > 2). At lower doses, 387 and 163 metabolites of HILIC(+) and C18(−), respectively, were altered by 0.01 ppm while 137 and 55 metabolites were altered by 0.1 ppm compared to control. As shown in Fig. 2C, pathway enrichment analysis on all disturbed metabolites (Fig. 2B and Fig. S4B) showed that four metabolic pathways, including two energy pathways (pyruvate, p = 0.0125; TCA cycle, p = 0.016) and two lipid regulated pathways (glycosphingolipid, p = 0.044; polyunsaturated fatty acid, p = 0.01), were disrupted at 0.01 ppm. While the number of disturbed metabolites was less than 0.01 ppm, V⁺⁵ at 1 ppm resulted in disruptions across multiple metabolic pathways, including amino sugars (p = 0.033), sialic acid (p = 0.0065), TCA cycle (p = 0.012), glycosphingolipid (p = 0.012), glycosylphosphatidylinositol(GPI)-anchor biosynthesis (p = 0.048), nucleotide sugar (p = 0.048), pyrimidine (p = 0.048), and vitamin D3 (p = 0.022); whereas, disruption in heparan sulfate degradation (p = 0.036) was observed by V⁺⁵ treatment at 0.1 ppm.

Fig. 2.

The selection and analyses of metabolite features from HILIC(+) chromatography altered by V⁺⁵ exposure. (A) Volcano plots comparing metabolic features from each V⁺⁵ exposure compared to the controls. The highlighted points represent the significant (p < 0.05, |fold change| > 2) metabolic features. Red indicates increased metabolites in cell exposed to V⁺⁵ and blue indicates decreased in cell exposed to V⁺⁵ compared with control (B) The overlap of metabolites from the significant metabolic features shown in (A). (C) Pathways enriched from each V⁺⁵ exposure comparing to controls. (D). Two-way HCA of selected metabolites at ANOVA p < 0.05 and linear correlation p < 0.05. Warm color and positive z-score indicate higher abundance of metabolites. PLS-DA score plot (E) and loading plot (F) of selected metabolites from (D). Annotated metabolites are labeled in (F) with their m/z and rt. (G) Pathways enrichment analysis revealed a total of 12 pathways following V⁺⁵ exposure based on dose-dependent ANOVA and linear regression analyses. n = 6 per group.

To further elucidate the dose-dependent response in metabolic perturbation following V⁺⁵ exposure, we conducted additional analyses to compare among all V⁺⁵ -treated groups. The disturbed metabolic features were identified based on ANOVA and linear regression model (see Manhattan plots in Fig. S2). A total of 213 HILIC(+) and 75 C18(−) features common to ANOVA and regression criteria (Fig. S3A and S3B; Supplemental Table 1 and Supplemental Table 2) were further selected for 2-way hierarchical cluster analysis (HCA) (Fig. 2D and Fig. S4C). The heatmap showed greatest similarity among cells treated with the highest dose (1 ppm, cyan; Fig. 2D and Fig. S4C) and somewhat more metabolites decreased in abundance than increased in abundance by V⁺⁵ at 1 ppm (Fig. 2D and Fig. S4C). PLS-DA using the 213 selected metabolites showed a trajectory in response to dose, with nearly complete separation of the 1 ppm exposure from control and lower doses (Fig. 2E and Fig. S4D). The PLS-DA loading plot (Fig. 2F and Fig. S4E) showed many metabolites had small contributions to separation of the V⁺⁵ exposures.

Pathway enrichment analysis on these 213 HILIC(+) and 75 C18(−) metabolites showed associations with 12 metabolic pathways (p < 0.05) in response to V⁺⁵ exposure (Fig. 2G). These pathways included aminosugar metabolism (p = 0.004), TCA cycle (p = 0.008), glycolysis and gluconeogenesis (p = 0.014), fatty acid oxidation (p = 0.02), sialic acid metabolism (p = 0.029), and nucleotide sugar metabolism (p = 0.04). Importantly, major amino acid metabolism pathways, including alanine (p = 0.016), glycine, serine, alanine and threonine (p = 0.04), and lysine (p = 0.049) were also altered. Disruption in GSH metabolism (p = 0.01) provided evidence for oxidative stress, and results also showed changes in vitamin D₃ (cholecalciferol) (p = 0.036) and drug metabolism (p =0.047). Notably, abundance of N-acetyl-D-hexosamine 6-phosphate, the leading metabolite shown in Fig. 2F, was increased by V⁺⁵ dose (FDR < 0.05) as measured by C18(−) and HILIC(+) (Fig. S5).

3.3. Disruption in Mitochondria Metabolism by V⁺⁵

The pathway enrichment analysis showed mitochondria energy metabolism, e.g., TCA cycle and fatty acid oxidation, was significant pathway perturbed by V⁺⁵ (Fig. 2C and 2G). Dose-dependent effects of V⁺⁵ on selected metabolites associated with the energy metabolism were examined (Fig. 3A–I). A dose-dependent decrease was observed for glutamate (Fig. 3A), N-acetyl-glutamate (NAG; Fig. 3B), glutamine (Fig. 3C), caprylic acid (Fig. 3D), lactate (Fig. 3E), oxalosuccinate (Fig. 3F), aspartate (Fig. 3G), dihydrothymine (Fig. 3H), and dTTP (Fig. 3I). The relationships among those metabolites are illustrated in a simplified pathway map in Fig. 3J. Additionally, we also observed decreased levels of spermine (Fig. S6A), tridecanoylglycine (Fig. S6B), Se-adenosylselenohomocysteine (Fig. S6C), methionine (Fig. S6D), and pentadecanoyl CoA (Fig. S6E). On the other hand, increased levels of several metabolites including vitamin D₃ (Fig. S7A), sphinganine (Fig. S7B), 6-keto-decanoylcarnitine (Fig. S7C), acetylcarnitine (Fig. S7D), 20-carboxy-leukotriene-B4 (Fig. S7E), dihydroxyacetone phosphate (Fig. S7F), and asparaginyl-hydroxyproline (Fig. S7G) were observed, suggesting that diverse disruptive responses of metabolites associated with energy metabolism occur to V⁺⁵ exposure.

Fig. 3.

Selected metabolites in energy metabolism and pyrimidine pathways altered by V⁺⁵ exposure. (A), Glutamate; (B), N-acetyl-glutamate (NAG); (C), Glutamine; (D), Caprylic acid; €, Lactate; (F), Oxalosuccinate; (G), Aspartate; (H), Dihydrothymine; (I), dTTP. (G), Simplified pathway showing changes in glycolytic/gluconeogenic pathway, TCA cycle, central connections to amino acid metabolism and nitrogen balance, pyrimidine, urea cycle, and fatty acid metabolism. Statistical significance for each condition vs. control is indicated by asterisks, with *p < 0.05, and **p < 0.01. n = 6 per group.

3.4. Dysfunction in Mitochondria Following V⁺⁵ Exposure

Following observations with V⁺⁵-disrupted mitochondrial energy metabolism, we further examined mitochondrial response to V⁺⁵ by analyzing mitochondrial membrane potential and structure. Analysis of mitochondrial membrane potential with JC-1 showed depolarization following V⁺⁵ exposure at 0.01 ppm or higher doses as measured by the ratio of red to green fluorescence intensity (Fig. 4A). Data analysis showed cells exhibited decline of mitochondrial membrane potential by 56%, 58%, and 69% following 0.01, 0.1, and 1 ppm V⁺⁵ exposure, respectively (Fig. 4B). Additionally, the morphological changes in mitochondria were examined by TEM after exposing cells to V⁺⁵ at 1 ppm. In the mitochondria, disrupted cristae with reduced electron density were observed as indicated by yellow arrowheads (Fig. 4C).

Fig. 4.

Mitochondrial dysfunction caused by low-dose V⁺⁵ exposure. (A) Mitochondrial membrane potential measured by JC-1 dye. Quantitative analyses are shown in (B). Cells exposed to H₂O₂ (20 μM) for 20 min were used as positive controls. (C) Representative TEM images of BEAS-2B cells treated with or without V⁺⁵. Note the swollen and dilated mitochondria with disrupted cristae structure (yellow arrows) in cells exposed to V⁺⁵. Normal mitochondria are indicated by black arrows. Scale bars are indicated in μm. Statistical significance for each condition vs. control is indicated by asterisks, with *p < 0.05, **p < 0.01 and ***p < 0.001. n = 8 for mitochondrial membrane potential assay; n = 3 for TEM analysis.

3.5. V⁺⁵-induced Decreased E-cadherin Expression

Dysfunction in mitochondria can further lead to cellular remodeling and epithelial-to-mesenchymal transition (EMT) activation [62]. We therefore examined whether V⁺⁵ has effect on EMT activation by measuring E-cadherin expression levels. The results show that E-cadherin expression measured by Western blotting was decreased in V⁺⁵ treated cells by a dose-dependent pattern, suggesting a potential EMT activation by V⁺⁵ exposure (Fig. 5A). E-cadherin levels decreased by 30% in BEAS-2B cells exposed to 0.1 ppm V⁺⁵ (p = 0.18), whereas a 57% decrease was observed in cells exposed to 1 ppm V⁺⁵ (p = 0.03). Little changes were observed in cells exposed to 0.01 ppm V⁺⁵.

Fig. 5.

E-cadherin expression and GSH/GSSG redox assay following V⁺⁵ exposure. (A) Quantification of E-cadherin expression level in BEAS-2B cells following V⁺⁵ treatment (left) is obtained by densitometry of the E-cadherin to β-actin intensity (right) analyzed by Western blotting (n=4). (B-D) Cells treated with V⁺⁵ as indicated were measured by HPLC for quantification of GSH (B), GSSG (C) and E_hGSSG/GSH (D). Statistical significance for each condition vs. control is indicated by asterisks, with *p < 0.05 and ***p < 0.001. GSH and GSSG amounts were normalized to total protein content (n=6).

3.6. V⁺⁵ Induced Oxidative Stress and Impaired GSH Homeostasis

Both mitochondrial dysfunction and potential EMT activation are commonly associated with oxidative stress and lung diseases [63–65]. Importantly, among the list of metabolic pathways impacted by V⁺⁵ exposure, glutathione pathway was identified as the 3^rd most significant pathway (Fig. 2G). In addition, our prior study with human lung fibroblasts showed that V⁺⁵ induced oxidative stress. Consequently, we examined for the effects of V⁺⁵ on glutathione redox state by measuring cellular amounts of reduced (GSH) and glutathione disulfide (GSSG). Results showed no changes in GSH concentrations up to 1 ppm (Fig. 5B) while GSSG was substantially increased at 1 ppm V⁺⁵ (Fig. 5C). Consistent with these changes, calculated cellular steady-state redox potential (E_hGSSG/GSH) was substantially oxidized by V⁺⁵ at 1 ppm (Fig 5D, E_h control; −220 mV and E_h 1 ppm V⁺⁵; −208 mV). Consistently, we also observed decreased levels in methionine (Fig. S6D), together with decrease in Se-adenosylselenohomocysteine (Fig. S6C), further suggesting dysregulated cellular redox homeostasis.

Discussion

The present studies with sub-ppm levels of V have widespread relevance to disease processes dependent upon mitochondrial function and oxidative stress. V is one of the most prevalent metals in our environment. Soil V is typically around 150 ppm [10], reaching up to 4800 ppm in contaminated areas [7], and bioaccumulation of V in plants can surpass 1000 ppm [66–68]. Although exposure to high-level V⁺⁵ is well known to lead to impaired lung functions and lung disease [18, 26, 69], little is known of V⁺⁵ effects at lower levels. Our earlier studies have demonstrated that oral exposure to V⁺⁵ resulted in stimulation of inflammatory and fibrotic signaling in lung fibroblasts [40]. Recognizing that the normal mouse lung consists of a significant number of epithelial cells which plays essential roles in maintaining normal lung physiology [41], it is important to identify how V⁺⁵ exposure would affect epithelial cell function. The present study expands prior knowledge of low-level V⁺⁵ effects in lung models, such as oxidative stress, cell-cycle arrest and cellular senescence [33, 40], by providing metabolic characterization of a human airway cell line, BEAS-2Bs, over the same sub-ppm range of exposures. Through metabolomics, mitochondrial membrane potential measurements and TEM, results show that this cell line has mitochondrial perturbations and extensive metabolic changes in response to low levels of V⁺⁵.

Widespread metabolic effects at V⁺⁵ concentrations well below the LC₅₀ for BEAS-2Bs, 8.03 ppm, were observed, including pathways changed in a bleomycin model for fibrotic lungs [70], i.e., TCA cycle, glycolysis and gluconeogenesis and fatty acid oxidation. Mitochondrial responses at sub-ppm levels are especially important because V⁺⁵ could induce mitochondrial oxidative stress at 0.1 ppm (2 μM total V) [71]. Our earlier study of human lung fibroblasts exposed to V⁺⁵ revealed increased mitochondrial ROS at 0.1 ppm without further response at higher doses [33]. Our recent mouse study showed that V⁺⁵ decreased glycolysis but increased energy metabolism and mitochondrial activity, with glycolytic activity of lung fibroblasts decoupled from whole lungs [40]. Studies have shown that mitochondria regulate key metabolic pathways in epithelial cells including surfactant production, senescence, programmed cell death and regeneration [72]. Indeed, we observed decreased respiratory activity in mitochondria following V⁺⁵ exposure. This is consistent with our metabolomic analyses showing TCA cycle pathway was perturbed in cells exposed to V⁺⁵. In addition, decreases in glutamate, NAG, and glutamine can further disrupt mitochondrial function.

Such a pattern is quite distinguishable from the V⁺⁵ induced mitochondrial response in lung fibroblasts [40], where respiratory activity was increased in fibroblasts exposed to V⁺⁵ at all tested doses. Additionally, lactate was increased in lung fibroblasts exposed to V⁺⁵ [40], but decreased in BEAS-2B cells. Mitochondrial dysfunction and metabolic reprogramming may have distinct characteristics in different lung cell populations, contributing to the development of complex lung diseases such as idiopathic pulmonary fibrosis [73]. Lung epithelia may be more susceptible to damage from oxidative stress [74, 75], and epithelial injury is an initiating event of fibrosis [76]. Therefore, our study provides evidence of sensitive mitochondrial responses to V⁺⁵ with loss of mitochondrial membrane potential and impaired mitochondria activity with V⁺⁵ as low as 0.01 ppm.

As previously reported [77], a metabolic shift from glycolysis to fatty acid oxidation occurs during airway epithelial repair from lung injury. Increased acylcarnitines following V⁺⁵ exposure could further activate proinflammatory signaling and promote age-related disease in association with decreased mitochondrial activity [78, 79]. Fatty acid oxidation and carnitine metabolism are tightly associated with phospholipid and sphingolipid metabolism and interconnected with biosynthetic and redox pathways [80]. Earlier studies show that lipid mediators derived from phospholipids and sphingolipids play an important role in the pathogenesis of pulmonary fibrosis [81, 82]. Observation of elevated carboxyleukotriene-B4, a leukotriene associated with pathogenesis of pulmonary fibrosis [83–86], in the current study also support our previous finding that V⁺⁵ exposure stimulated lung fibrosis signaling.

In addition to shift from glycolysis to fatty acid oxidation, the linear increase in N-acetyl-D-hexosamine 6-phosphate (FDR < 0.05) indicates a redirection of glycolysis into the important hexosamine biosynthetic pathway also likely occurred. As an important hexosamine biosynthetic pathway intermediate, N-acetyl-D-hexosamine 6-phosphate is a precursor for UDP-N-acetyl-D-hexosamine and functionally related to D-hexosamine. Elevated levels of N-acetyl-D-glucosamine 6-phosphate might further lead to abnormal O-GlcNAcylation [87, 88], perturbing energy metabolism through modulating TCA cycle and the mitochondrial respiratory chain[89]. Notably, studies show that there exists a strong correlation between hexosamine biosynthetic pathway/O-GlcNAcylation and EMT [90]. While abnormal O-GlcNAcylation can lead to TGF-β activation and cause fibrosis[91], altered levels of N-acetyl-D-hexosamine 6-phosphate were consistently linked to pulmonary hypertension[92], pulmonary arterial hypertension[93] and idiopathic pulmonary fibrosis [94], further suggesting it could act as a biomarker for V⁺⁵ exposure and associated lung diseases.

Importantly, mitochondrial dysfunction is commonly associated with EMT activation[62, 95]. Indeed, our results indicate that E-cadherin was downregulated in BEAS-2B cells exposed to V⁺⁵, suggesting activation of the EMT in BEAS-2B cells following V⁺⁵ exposure. A prior study reported that activation of EMT could increase the demand for glutamine for fatty acid biosynthesis with a concomitant decrease in glycolytic activity [96]. Glutamine deprivation can further up-regulate EMT[97], consistent with the decreased glutamine level in this study. Central metabolic perturbation also occurred in amino acid metabolism linking glutamine to the urea cycle and pyrimidine metabolism, which are also commonly associated with EMT activation [98, 99]. These processes are connected by dependence upon carbamoyl phosphate, intermediately formed from glutamine by mitochondrial enzyme, carbamoyl phosphate synthetase (CPS). CPS is highly sensitive to oxidative stress [100] and requires NAG for activity. V⁺⁵ inhibited this step by oxidative stress and decreased NAG by interacting with glutamine and aspartate. These link to pyrimidine biosynthesis, another essential pathway linked to mitochondrial function and sensitive to oxidative stress. Consistently, an important precursor for GSH biosynthesis, methionine, was decreased by V⁺⁵; decreased methionine is associated with many lung diseases, including cystic fibrosis [101] and lung inflammation [102]. The altered methionine pool could contribute to observed changes in GSH after V exposure [103]. Additionally, methionine, cysteine, and GSH are all closely associated with vitamin D3 [104], together regulating redox homeostasis and mediating inflammatory responses. Those changes in cellular redox homeostasis [105, 106] as well as vitamin D [107] are oftentimes linked to EMT activation, elevated inflammation, and pathogenesis of lung diseases.

In summary, the current study showed that V⁺⁵ exposure in lung epithelial cells experienced widespread metabolic impacts resultant from V⁺⁵ at sub-ppm levels. Major effects included changes in the amino acid, carbohydrate, vitamin D, energy, and redox regulated pathways, as well as mitochondrial structure and function. Additionally, evidence of oxidative stress, increases in inflammatory lipids, and decreases in E-cadherin provide potential evidence for stimulation of EMT at exposure levels which do not cause cell death.