Influence of macrophages and neutrophilic granulocyte-like cells on crystalline silica-induced toxicity in human lung epithelial cells

Abstract

In many industrial activities, workers may be exposed by inhalation to particles that are aerosolized, To predict the human health hazard of these materials, we propose to develop a co-culture model (macrophages, granulocytes, and alveolar epithelial cells) designed to be more representative of the inflammatory pulmonary response occurring in vivo. Phorbol 12-myristate 13-acetate (PMA)-differentiated THP-1 cells were used as macrophages, All-trans retinoic acid (ATRA)-differentiated HL60 were used as granulocytes and A549 were used as epithelial alveolar type II cells. A crystalline silica sample DQ12 was used as a prototypical particle for its capabilities to induce DNA damage, inflammatory response, and oxidative stress in epithelial cells; its polyvinylpyridine-N-oxide (PVNO)-surface modified counterpart was also used as a negative particulate control. Cells in mono-, bi- or tri-culture were exposed to DQ12 or DQ12-PVNO for 24 h. DQ12 but not DQ12-PVNO induced a significant increase in DNA damage in A549 cells. The presence of differentiated THP-1 reduced the genotoxic effects of this crystalline silica sample. The exposure of A549 to DQ12 but not DQ12-PVNO induced a significant change in interleukin-8 (IL-8) protein levels which was exacerbated when differentiated THP-1, and HL-60, were added. In addition, while no production of TNFα was detected in the A549 monoculture, elevated levels of this cytokine were observed in the co-culture systems. This work shows that a cell culture model that takes into consideration the complexity of the pulmonary inflammatory response might be more dependable to study the toxicological properties of particles than “simple” monoculture models.

Introduction

In various industrial sectors, workers are potentially exposed to fibers and particles. However, their health effects are not always well characterized. Since they may get aerosolized, the major route of exposure to particles is inhalation. Therefore, the most appropriate experimental approach to assess their toxicological properties is to perform inhalation or intratracheal instillation in rodent models.¹ For ethical reasons, and due to the international effort to fulfil the 3R rules to reduce the use of laboratory animals,² there is a need for the development of reliable alternative methods based mainly on cell culture models for the assessment of particle toxicity.^3,4

It is considered that the genotoxic and carcinogenic potential of inhaled particles could be associated i) with primary genotoxic mechanisms, linked to the alteration of the genome directly due to the physico-chemical properties of particles or indirectly via the production of cellular reactive oxygen species (ROS); or ii) with secondary genotoxic mechanisms due to a persistent inflammation associated with particle reactivity and/or pulmonary overload.^5,6 The lung inflammatory response induced by particle inhalation is a complex process, which requires several steps starting with the recognition and phagocytosis of foreign bodies by resident macrophages. It may be followed by the production of cytokines such as Tumor Necrosis Factor α (TNFα) and Interleukins 1β and 8 (IL-1β and − 8) by macrophages or alveolar epithelial cells, leading to the recruitment of neutrophilic granulocytes and macrophages in the alveolar space.^7–9 The persistent inflammation associated with the recruitment of neutrophilic granulocytes and macrophages contributes to a continuous production of ROS and a possible alteration of the genetic material of pulmonary epithelial cells.

Many studies on the genotoxicity of particles used simple in vitro models consisting of the exposure of single cell lines followed by the assessment of the subsequent toxicological effects.¹⁰ These models did not consider the complexity of the pulmonary tissue, particularly the interaction between the different cell types. For instance, in the alveolar space, epithelial cells interact at least with macrophages, and potentially neutrophilic granulocytes during inflammation.¹¹ In addition, these studies only focused on the primary genotoxicity of particles.^12,13

Co-culture models using epithelial cells on one hand and macrophages and/or granulocytes on the other have been described in the literature.^14–16 Wottrich et al (2004) showed an increasing production of pro-inflammatory cytokines IL-6 and IL-8 in a co-culture model of A549 alveolar epithelial cells and activated THP-1 macrophages after exposure to particles, in comparison to monocultures of each cell type. However, the presence of macrophages in co-culture systems did not seem to increase the mutagenicity of crocidolite fibers in Big Blue® fibroblasts¹⁷ nor the cytotoxicity of amosite fibers in rat and human pulmonary epithelial cells.¹⁴ On the contrary, neutrophilic granulocytes derived from mouse bone marrow increased the capacity of crystalline silica to induce DNA breaks in the A549 alveolar epithelial cells.¹⁵ These results could be explained by a higher generation of ROS by neutrophilic granulocytes than macrophages.¹⁸ These radical species, able to play an important role in the induction of oxidative DNA damage, are observed during a persistent inflammatory response.^9,19

The induction of an in vivo pulmonary inflammatory response associated with a significant increase of alveolar neutrophilic granulocytes was correlated with an in vitro production of the pro-inflammatory IL-8 cytokine by the A549 epithelial cells.^12,20 It was also shown that IL-8 stimulates granulocytes migration in vivo and in vitro.^7,21

Since the in vivo genotoxicity of particles may be the consequence of persistent inflammation associated with the recruitment of neutrophilic granulocytes, it is important to use in vitro models that can consider this pathophysiological process to better decipher the genotoxic and inflammatory effects of these chemicals. Therefore, the aim of this study was to develop a co-culture model involving pulmonary epithelial cells, macrophages and neutrophilic granulocytes and assess its suitability to detect the effects (nano)particles. For this purpose we used a reference material, crystalline silica that has been extensively studied other the past decades and its toxicological mechanisms of action are well described and include cellular (geno)toxicity, persistent inflammation as well as lung fibrosis and cancer.^22–24 In addition, in order to use a non-toxic version of this material, we coated it with polyvinylpyridine-N-oxide which acts as a scavenger for radicals produced by crystalline silica exposure within the cells.^25,26

Material and methods

Particles

The crystalline silica DQ12 was obtained from DMT GmbH (Essen, Germany, batch number: 280808), with an average diameter of 2 μM. In order to quench the toxicological effects of the DQ12, particles were coated with polyvinylpyridine-N-oxide (PVNO, Tebu-Bio SA, France) as described in the literature.¹⁹ DQ12 (75 mg) were suspended in 15 mL of ultrapure water with 1% PVNO while stirring for 5 h. Then, particles were rinsed three times by centrifugation at 12,000 g and dried at 37 °C. To measure the amount of PVNO adsorbed on the particle surface, the DQ12-PVNO was suspended in 1 N sodium hydroxide, centrifuged at 12,000 x g and the optical density was measured in the supernatant at 240 nm. The blank control was realized with the supernatant of DQ12 without treatment by PVNO. The quantity of PVNO adsorbed on the DQ12 surface was 6.8 ± 0.2 μg/mg of particles.

Cell culture and treatment

The cell lines were obtained from ATCC (Manassas, VA, USA). The A549 lung epithelial cell line (ATCC #CCL-185) was from human alveolar adenocarcinoma. Cells were cultured in DMEM glutamax (Pan Biotech #P04–04510), 10% decomplemented Fetal Calf Serum (FCS), 50 U/mL penicillin and 50 μg/mL streptomycin.

The THP-1 human monocytic leukemia cell line (ATCC #TIB-202) was cultured in RPMI 1640 with 2 mM glutamine, 1 mM Hepes, 1 mM sodium pyruvate, 10% decomplemented FCS, 50 U/mL penicillin and 50 μg/mL streptomycin. Cells were differentiated into macrophage-like cells by treatment with 200 nM 12-O-tetradecanoylphorbol-13-acetate (TPA, Sigma #P1585) for 24h²⁷. To collect the adherent differentiated cells, a culture was performed using Upcell™ 6 well plates (Thermo Scientific #174901) and cells were detached with Versene buffer (PBS with 0.48 mM EDTA). Cell differentiation was assessed by flow cytometry using CD11b and CD11c antibodies as published previously²⁸ (Supplementary Fig. 1A). After PMA treatment, 64.1% of cells were CD11b-positive and 87.7% were CD11c-positive (in comparison to 12.5% and 4.1% in untreated cells, respectively).

The HL-60 human promyelocytic leukemia cell line (ATCC #CCL-240) was cultured in RPMI 1640 with 2 mM glutamine, 1 mM Hepes, 1 mM sodium pyruvate, 10% decomplemented FCS, 50 U/mL penicillin and 50 μg/mL streptomycin. Cells were differentiated in cells with mature granulocyte properties by treatment with 1 μM all-trans retinoic acid (ATRA, Sigma #R2625) for 7 days. Cell differentiation was assessed by nitro blue tetrazolium (NBT) (Sigma #84010) reduction (Supplementary Fig. 1B).^29,30 After ATRA treatment, 80% of HL-60 cells were differentiated (formazan-positive cells).

Based on the literature describing the relative proportion of each cell type in the lung, in the co-culture set-up, cell lines A549:THP-1:HL-60 were mixed with a 5:1:1 ratio.³¹ We also kept the same proportion of each specific cell line medium. Cells were maintained at 37 °C in a humid atmosphere with 5% CO₂.

Cells were treated with DQ12 or DQ12-PVNO for 24 h with 50, 100 or 200 μg/cm² of particles. During treatment, cells were cultured without FCS.

Cell number and viability

A cell count with acridine orange (Sigma Aldrich #A8097) and propidium iodide (Sigma Aldrich #P4864) was performed 24 h after treatment to determine the number of cells and their viability using a Cellometer Vision (Nexcelom Bioscience).

Cytotoxicity assay by LDH release dosage

LDH content in a supernatant was analyzed using a Cytotoxicity Detection LDH kit (Roche Life Science #11644793001) following the manufacturer’s instructions. The percentage of cytotoxicity was calculated based on the 100% toxicity (LDH release) induced by Triton X-100 treatment.

Immuno-magnetic cell separation

After particle treatment, cells were collected and incubated 15 min at 4 °C with anti-CD45 antibodies coupled with magnetic beads (Miltenyi Biotec #130–045-801), specific to THP-1 leucocytes and HL-60 differentiated cells.³² THP-1 and HL-60 cells were retained on the column, whereas the eluted A549 cells were used for genotoxicity and gene expression analyses. The flow-through was analyzed by flow cytometry and revealed that it contained 99.7% A549 cells (Supplementary Fig. 2).

Comet assay

The comet assay enables the detection of DNA simple and double strand breaks induced by chemical agents. After 24 h of particle treatment, epithelial cells were separated from other cell types using immuno-magnetic beads, as described above, prior to their utilization for the comet assay in alkaline conditions.³³ A549 treated with 100 μM of Methyl methane sulfonate (MMS, Sigma # 129925) for 24 h was used as a positive control. Following embedding in low-gelling agarose (Sigma Aldrich # A9414), cells were lysed overnight at 4 °C. Electrophoresis was performed the next day at a voltage of 0.7 V/cm for 40 min in BioRad Sub-Cell Model 192 electrophoresis systems. Comets (100 events per sample) were scored using Comet assay IV software (Perceptive Instruments Ltd) after propidium iodide DNA staining.

Reactive oxygen species detection

In order to detect ROS production by A549 cells after treatment in the co-culture model, A549 cells were labelled with 0.5 μM of the cell proliferation dye, eFluor™ 670 (eBioscience #65–0840) as described previously,²⁸ prior to the addition of THP-1 and HL-60 cells.

Briefly, cells were incubated with the dye solution for 10 min at 37 °C in the dark and the labelling was stopped by adding cold complete medium and then incubated on ice for 5 min. After 3 washes with PBS, the cell pellets were suspended in complete DMEM with 10% FBS and seeded into 6-well culture plates. The 2′,7′-dichlorodihydrofluorescein diacetate probe (H₂DCFDA, Life Technologies #D399) is a permeable cell membrane dye cleaved by esterase and transformed into a fluorescent dye by ROS.²⁸ This probe was used as a marker of intracellular ROS production in A549 cells. eFluor™ 670-labeled A549, cultured alone or with THP-1 and/or HL-60 cells, were treated with particles for 24 h without FCS and phenol red followed by a removal of the medium and addition of H₂DCFDA 10 μM for 30 min. Then, the analysis was performed with flow cytometry (BD Accuri™ C6) on the gate of eFluor™ 670 positive cells.

Gene expression analysis by RT-qPCR

After particle treatment, epithelial cells were separated using immuno-magnetic beads, as described above, and pellets were used for RNA extraction (RNeasy mini kit, Qiagen #74134). The RNA (500 ng) was reverse transcribed (iScript™ cDNA Synthesis Kit, Bio-Rad #1708891) and subsequently amplified by quantitative PCR (SsoAdvanced Universal SYBR® Green Supermix, Bio-Rad #172–5,275). Genes studied were: β-actin (ACTB Forward: AATGCCAGGGTACATGGTG, Reverse: TTCCTTCCTGGGCATGGAGT) as housekeeping gene, interleukin-8 (IL8, CXCL8 Forward: ATTTCTGTGTTGGCGCAGTG, Reverse: AGCTCTGTGTGAAGGTGCAG), heme oxygenase 1 (HMOX1 Forward: TCTTGCACTTTGTTGCTGGC, Reverse: CCTGCTCAACATCCAGCTCT) and cyclooxygenase 2 (COX2, PTGS2 Forward: AGGGCTTCAGCATAAAGCGT, Reverse: CAAATTGCTGGCAGGGTTGC).

Cytokine dosage

TNFα, IL-8 and IL-1β quantifications were performed on the cell supernatant after 24 h of treatment by the ELISA assay following the manufacturer’s instructions (Thermo Scientific #EH3TNFA5, Invitrogen #KHC0081C, Invitrogen #KHC0011C respectively).

Proteome profiler

The Proteome Profiler Human XL Cytokine Array Kit (R&D SYSTEMS #ARY022B) is a membrane-based sandwich immunoassay. Captured antibodies corresponding to 105 different proteins, spotted in duplicate on nitrocellulose membranes bind to specific target proteins present in the sample. Captured proteins are detected with biotinylated detection antibodies and then visualized using chemiluminescent detection reagents. The signal produced is proportional to the number of bound analytes. This assay was conducted on A549 cell supernatants with and without THP-1 or HL-60 cells and with or without DQ12 (100 μg/cm²) treatment.

Supernatants were diluted in blocking buffer 6 for a final volume of 1.5 mL. They were incubated overnight with a pre-blocked Proteome Profiler Array cytokine membrane, on a rocking platform at 4 °C. After the incubation period, the membrane was washed and incubated with a cocktail of biotinylated detection antibodies for 1 h. Finally, the membrane was incubated with streptavidin-HRP reagent for 30 min. Chemiluminescent spots were visualized using chemiluminescent detection reagents provided in the kit. The quantification of cytokines was performed with a ChemiDoc MP Imaging System (Bio-Rad).

Statistical analysis

Except for the Proteome profiler immunoassay, statistical analysis was performed using R-studio software (version 2.8.2, RStudio Inc). A linear regression model was applied followed by a multiple-comparison post-hoc test (Dunnett).

For the immunoassay, a statistical analysis was performed using Stata Statistical Software (Release 14.2, StataCorp LLC, College Station, TX, USA). To test the particle effect, for each cytokine and each mono or co-culture conditions, an ANOVA was applied followed by a Bonferroni post-hoc test. The statistical significance threshold was set at 5%.

Results

DQ12 dose selection

The doses used in this study were determined based on their toxicity in A549 cells alone. For the duration of the assay, A549 cells were cultured without serum and treated with increasing concentrations of DQ12 for 24 h. Results presented in Fig. 1A show that DQ12 treatment decreased significantly cell growth (measured following AO/PI staining) starting at 50 μg/cm² with the highest effect at 200 μg/cm² (15, 55 and 90% inhibition of cell growth, respectively, Fig. 1A). On the contrary, DQ12-PVNO treatment did not decrease cell growth, a significant slight increase of cell numbers was even observed with 50 μg/cm². In addition, cell viability was significantly affected by DQ12 treatment (about 10, 20 and 40% decrease following treatment with 50, 100 and 200 μg/cm², respectively, Fig. 1B), whereas DQ12-PVNO treatment slightly affected A549 viability. Based on these results, particle concentrations of 50 and 100 μg/cm² were kept for the following experiments.

Fig. 1.

Effect of particles on A549 cell growth and viability. A549 cells were treated for 24 h with the indicated concentrations of DQ12 and DQ12-PVNO. Cells were numbered (A) and the viability (B) was determined. The bar graph represents the mean ± standard error of the mean (SEM) of at least three independent experiments. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 were significantly different from the control.

Cytotoxicity of silica

LDH activity in a cell culture supernatant was used to determine the particle toxicity in the different cell culture conditions. The treatment of A549 cells with 100 μg/cm² of DQ12 induced a significant 7-fold increase of LDH release, whereas DQ12-PVNO did not significantly affect cytotoxicity (Fig. 2). When epithelial cells were cultured with THP-1 and/or HL-60 differentiated cells, as compared to A549 cells alone, a significant decrease of DQ12-induced cytotoxicity was observed. This protective effect was more pronounced in the presence of THP-1 than HL-60 cells. However, at 100 μg/cm² DQ-12 was still inducing a significant release of LDH whatever the cell culture conditions. Treatment with the two concentrations of DQ12-PVNO did not induce any cytotoxicity when A549 cells were cultured alone or in a co-culture with the two other cell lines, except a slight but significant 1.3-fold increase with the highest concentration of DQ12-PVNO when A549 cells were co-cultured with THP-1 cells.

Fig. 2.

Cytotoxic effect of DQ12. A549, THP-1 and HL-60 cells were cultured alone or in di- or tri-cultures and treated for 24 h with 50 and 100 μg/cm² of DQ12 or DQ12-PVNO. LDH activity was quantified in the supernatant. The bar graph represents the mean ± standard error of the mean (SEM) of nine independent experiments. ^*P < 0.05, ^***P < 0.001 were significantly different from the control.

Genotoxicity of silica in A549 cells

The DQ12 treatment of A549 cells induced a significant increase of DNA strand breaks regarding the two concentrations evaluated, whereas treatment with DQ12-PVNO did not induce any significant increase in DNA damage (Fig. 3 and Supplementary Fig. 3). The induction of DNA breaks observed with the two concentrations of DQ12 was slightly decreased when A549 cells were co-cultured with THP-1 cells. Interestingly, when A549 cells were co-cultured with HL-60 cells, DQ12 treatment induced a comparable level of DNA strand breaks compared to A549 cells alone (Fig. 3). The protective effect due to the presence of THP-1 cells was more pronounced in the tri-culture model, whereas treatment with DQ12 did not induce any significant DNA strand breaks. DQ12-PVNO treatment only slightly modulated the DNA strand break level regardless of the culture conditions.

Fig. 3.

Effect of silica particle on DNA damage. A549 cells were cultured alone and in di- or tri-cultures with THP-1 and HL-60 cells. Cells were treated for 24 h with DQ12 or DQ12-PVNO and a comet assay was performed. The bar graph represents the mean ± standard error of the mean (SEM) of at least four independent experiments. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 were significantly different from the control.

Intracellular reactive oxygen species production

The study of intracellular ROS production was performed using the H₂DCFDA probe (Fig. 4 and Supplementary fig. 4). The experiments showed that DQ12 treatment (100 μg/cm²) increased their A549 cell levels in the monoculture (5.6-fold, Fig. 4). In the presence of THP-1 cells, the basal level was much higher (18-fold compared to A459 cells alone) and the treatment with 100 μg/cm² DQ12 did not induce ROS overproduction. In the presence of HL-60 cells, the basal level was like that of A459 cells alone, but an increase of the ROS level was observed only with the highest concentration (3.9-fold compared to cells without treatment). In the tri-culture model, the basal level was close to that observed in the presence of THP-1 cells and DQ12 treatment did not induce ROS overproduction. We noticed that DQ12-PVNO treatment decreased the ROS production compared to untreated cells, in mono-, di- or tri-culture conditions.

Fig. 4.

Overproduction of reactive oxygen species following DQ12 treatment. After 24 h treatment of A549 cells cultured alone and in di- or tri-culture with THP-1 and HL-60 cells with DQ12 or DQ12-PVNO, the production of reactive oxygen species was determined. The bar graph represents the mean ± standard error of the mean (SEM) of nine independent experiments. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 were significantly different from the control.

Gene expression

Interleukin-8 (IL8), cyclooxygenase-2 (COX2) and heme oxygenase 1 (HMOX1) gene expressions were studied in A549 by RT-qPCR after immune-magnetic cell separation. In the A549 monoculture, the IL8 expression increased 2.6- and 3.4-fold after DQ12 treatment with 50 and 100 μg/cm², respectively, but was not modified by DQ12-PVNO (Fig. 5A). In the presence of THP-1 cells, and to a much lower extent with HL-60 cells, the basal expression level of this gene was significantly increased (88- and 1.6-fold increase, respectively). In addition to this higher basal level of IL8 mRNA, an increase of its expression was observed with the higher concentration of DQ12 in the di-culture A549-THP-1 cells (2-fold). The DQ12 treatment of di-culture A549-HL-60 cells induced a dose-dependent increase of the IL8 gene expression (1.9- and 5.4-fold with 50 and 100 μg/cm², respectively). In the tri-culture model, expression of this cytokine was also increased in epithelial cells without treatment, at a similar level as in the di-culture of A549-THP-1 cells. The addition of DQ12 or DQ12-PVNO did not modify the IL8 expression.

Fig. 5.

Increase of inflammatory mediator expression. The mRNA levels of IL-8 (A), COX2 (B) and HMOX1 (C) were analyzed in A549 cells cultured alone and in di- or tri-culture with THP-1 and HL-60 cells after 24 h of treatment at the indicated concentration of DQ12 or DQ12-PVNO. The bar graph represents the mean ± standard error of the mean (SEM) of at least four independent experiments. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 were significantly different from the control. IL: Interleukin, COX2: Cyclooxygenase, HMOX1: Heme oxygenase.

Similarly, the COX2 expression analysis showed an increase of its level in A549 cells after DQ12 treatment (1.8- and 1.9-fold with 50 and 100 μg/cm², respectively), whereas treatment with 100 μg/cm² of DQ12-PVNO decreased its expression (Fig. 5B). Even if the basal level of COX2 expression was higher in the di-culture with THP-1 cells than with HL-60 cells, only 100 μg/cm² of DQ12 significantly increased its gene expression (1.3- and 2.2-fold increase after treatment with 100 μg/cm² of DQ12 with THP-1 and HL-60 cells, respectively). In the tri-culture model, an increase of the COX2 basal expression level was observed to a similar extent as that observed with the di-culture of A549 with THP-1 cells (22-fold). DQ12 treatment of the tri-cultured cells did not increase the COX2 expression.

The HMOX1 gene expression was weakly modulated by particle treatment in A549 cells cultured alone (1.7 and 1.8-fold increase after treatment with 50 and 100 μg/cm² of DQ12, respectively, Fig. 5C). However, the presence of THP-1 cells in di- or tri-culture models led to a significant decrease of the expression of this gene (2-fold) compared to the basal level of A549 cells alone. DQ12 treatment (100 μg/cm²) of A549 di-cultured with THP-1 cells slightly increased the HMOX1 expression (1.5-fold). When cells were co-cultured with HL-60 cells, with or without THP-1 cells, DQ12 treatment did not modulate the HMOX1 gene expression. DQ12-PVNO treatment did not affect the HMOX1 expression regardless of the cell culture conditions, except a slight increase (1.3-fold) after treatment of A549 cells alone with 50 μg/cm².

Cytokines expression

The Tumor Necrosis Factor α (TNFα) dosage in a cell supernatant by ELISA showed that A549 and HL-60 cells did not express this cytokine without particle treatment, whereas the co-culture with THP-1 cells induced the production of TNFα even without DQ12 treatment (Fig. 6A). DQ12 treatment of A549 cells alone did not induce the expression of this protein. However, this protein was overexpressed after treatment with 100 μg/cm² of DQ12 in A549 cells co-cultured with THP-1 (1.4-fold) or HL-60 (4.7-fold). In the case of the tri-culture, without any treatment, the level of this protein was higher (2.7-fold compared to di-cultured A549-THP-1 cells). Nevertheless, DQ12 treatment significantly decreased its expression (2.2- and 1.6-fold decrease with 50 and 100 μg/cm², respectively). DQ12-PVNO treatment of the tri-culture decreased the production of TNFα as well (1.4- and 1.2-fold with 50 and 100 μg/cm², respectively).

Fig. 6.

Silica particle treatment increased cytokine production. The levels of TNFα (A), IL-8 (B) and IL-1β (C) cytokines were analyzed in A549 cells cultured alone and in di- or tri-culture with THP-1 and HL-60 cells after 24 h of treatment at the indicated concentration of DQ12 or DQ12-PVNO. The bar graph represents the mean ± standard error of the mean (SEM) of at least three independent experiments. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 were significantly different from the control. IL: Interleukin, TNF: Tumor necrosis factor.

Interleukin-8 (IL-8) dosage showed that A549 cells weakly expressed this cytokine, however treatment with DQ12 led to a significant increase of its expression (46-fold with 100 μg/cm²); DQ12-PVNO did not have this effect, the highest concentration decreased the IL-8 expression (Fig. 6B). The presence of THP-1 cells with epithelial cells increased the basal level of this protein (486-fold) but also stimulated its overexpression in the presence of a higher concentration of DQ12 (2.1-fold). This phenomenon was not observed with DQ12-PVNO. To a lesser extent, a similar effect was noticed when adding HL-60 cells (4.4- and 11.6-fold increase following treatment with 50 and 100 μg/cm² of DQ12, respectively). In the tri-culture model, the basal level of this protein was comparable with that of di-culture A549-THP-1 cells, however treatment with DQ12 slightly affected this protein level (between 1.1 and 1.3-fold). DQ12-PVNO treatment did not modulate or slightly change the protein level of IL-8 in co-cultured conditions.

An analysis of interleukin-1β (IL-1β) in the cell supernatant revealed that A549 cells cultured alone or di-cultured with HL-60 cells did not express this cytokine with or without particle treatment (Fig. 6C). When epithelial cells are in the presence of THP-1 cells, the basal level of cytokine was increased and exacerbated after 100 μg/cm² of DQ12 treatment (1.9- and 4.2-fold in di- and tri-culture, respectively). DQ12-PVNO treatment did not modify the level of this cytokine, regardless of the cell type considered.

Proteome profiler

The level of various proteins was analyzed in mono-, di- and tri-cultures after treatment with 100 μg/cm² of DQ12 or DQ12-PVNO. Treatment of A549 cells alone induced the downregulation of 4 proteins (Angiogenin, GM-CSF, MCP-1 and PDGF-AA) and up-regulation of 3 proteins (ENA-78, IL-8 and MIF, Table 1). The treatment of the di-cultured A549 and THP-1 cells modulated the level of 12 proteins, 6 were down-regulated and 6 were up-regulated. In the di-culture of A549 with HL-60 cells, 2 proteins were down-regulated (Angiogenin and uPAR) and 7 proteins were up-regulated. Finally, in the tri-culture model, 7 proteins were down-regulated and 4 were up-regulated after silica treatment. The presence of HL-60 and/or THP-1 reversed the modulation of the protein level after the treatment of A549 cells in most cases or at least attenuated the observed effect. After 24 h of treatment with DQ12-PVNO, 7 proteins were down-regulated when A549 cells were cultured alone, 10 proteins were down-regulated and 1 was up-regulated in the di-culture with THP-1 cells, 5 proteins were down-regulated and 1 was up-regulated in the di-culture with HL-60 cells, and in the tri-culture, DQ12-PVNO treatment down-regulated 7 proteins and up-regulated 1 protein.

Table 1.

Protein level.

	A549		A549 - THP-1		A549 - HL-60		A549 - THP-1 - HL-60
Proteins	DQ12	DQ12- PVNO	DQ12	DQ12- PVNO	DQ12	DQ12- PVNO	DQ12	DQ12- PVNO
Chitinase 3 like-1	0.78	4.08	1.21a	1.09	1.52	0.38a	1.08	0.99
DKK-1	1.01	0.40b	1.31a	0.56b	1.32 a	0.53a	1.01	0.77a
ENA-78	3.76c	0.95	1.26a	0.86	4.53 c	1.05	1.19a	0.86
GM-CSF	0.14a	0.11a	1.76c	0.76a	0.18	0.01	1.21a	0.76b
GRO-α	0.46	0.22	1.03	1.22a	19.79c	0.61	1.15a	1.27b
MIF	3.78c	1.11	2.00b	1.01	1.79 c	1.06	1.28a	1.05
MIP-3α	1.66	0.00a	1.38b	0.58b	0.40	0.25	1.09	0.91
SERPIN E1	0.97	0.68c	1.01	0.62b	1.21 c	1.01	1.02	0.89
IL-8	20.40c	0.10	0.97	0.79	11.54b	2.31	0.73b	0.99
MCP-1	0.12b	0.27a	0.84c	0.69c	5.74c	1.74a	0.51b	1.00
Angiogenin	0.30b	0.47b	0.69b	1.09	0.31c	0.48c	0.94	0.74a
GDF-15	3.11	2.53	0.46c	0.61b	0.65	0.08	0.52c	0.62c
IL-1ra	0.00	0.84	0.92	0.94	0.27	0.32	0.60c	0.73b
MIP-1α/MIP-1β	0.34	0.73	1.03	0.78b	1.14	3.33	0.76b	0.88a
MMP-9	5.41	0.51	0.76b	0.65c	0.55	1.18	0.85	0.98
PDGF-AA	0.26b	0.26b	0.45b	0.48b	0.72	0.58a	0.30b	0.53a
uPAR	0.40	0.18	0.76a	0.70b	0.00a	0.15a	0.87b	1.02

A549 cells cultured alone or in a co-culture with THP-1 and/or HL-60 were treated for 24 h with 100 μg/cm² of DQ12 or DQ12-PVNO. Protein levels were quantified using Proteome Profiler. Results were presented as fold changes compared to the concurrent cell culture control.

^a P < 0.05; ^bP < 0.01; ^cP < 0.001 were significantly different from the control (untreated matched mono, bi or tri-cell cultures).

Discussion

The study of particle toxicity could be performed using different approaches with the use of animal models or cell cultures. Nowadays, in vivo models are still the most dependable to predict the effects of particles on human health. However, development of alternative in vitro models is a necessity. Indeed, these models help to reduce the use of laboratory animals and decrease the cost and the time needed to assess the toxicological effects of particles.

In workplaces, the major route of exposure is inhalation. Therefore, the first tissue in contact with particles is the respiratory system and in particular, the lung tissue. Different cell types are present in this tissue including epithelial cells and macrophages.¹¹ For the evaluation of particle toxicity, the mono-culture model of human epithelial cells (A549, HBE. . .) was used and considered as representative of the target cells in the lung.^3,4,34–36 However, the interaction between the different cell types in the lung suggests the need to improve the in vitro model. In the lungs, particles are mainly phagocytized by alveolar macrophages. To address this process, in vitro models using macrophages have been developed. THP-1 cells differentiated by TPA have often been used as surrogates for human alveolar macrophages. Numerous studies used a co-culture model with differentiated THP-1 cells and A549 cells to analyze the interaction and communication between these two cell types and their consequences on cell response.^36–42

However, such co-culture models do not take into consideration the subsequent inflammation induced by particle inhalation. To try to understand the mechanism of the action of particles and specifically that associated with pulmonary inflammatory response, we wanted to evaluate the role of neutrophilic granulocytes, considered as an in vivo hallmark of inflammation.⁴³ Therefore, HL-60 cells were differentiated by ATRA treatment, a model well described in the literature as an alternative to the use of granulocytes from blood donors.^29,44 To the best of our knowledge, this study is the first to try to mimic an in vitro lung inflammatory response using these three cell lines. In the context of co-culture, the ratio of A549:THP-1:HL-60 cells is 5:1:1, which corresponds to the ratio between epithelial cells and macrophages in a normal lung, and the proportion of granulocytes expected during a pulmonary inflammation induced by crystalline silica.^11,15,31

The performance of the co-culture model was evaluated with the reference crystalline silica DQ12 material, known to induce genotoxic and pro-inflammatory effects in vitro. In parallel, the coated DQ12 by PVNO was used as a particulate negative control. In vivo studies showed that exposure to DQ12 by intratracheal instillation induced a pulmonary inflammatory response whereas DQ12-PVNO exposure induced a lesser effect.^19,45 Similarly, treatment of A549 cells by DQ12 induced IL-8 production and a DNA strand break, whereas DQ12-PVNO did not.^46,47 The DQ12 doses for this study were chosen according to their capabilities to induce DNA damages in A549 cells. Then, they were much higher than the concentrations sufficient to induce IL-8 expression in epithelial cells or pulmonary inflammation in rats. In addition, based on the literature, cell exposure to particles was done without serum since its constituents could inhibit the toxicological effects of particles.⁴⁷ Even if these culture conditions could artificially increase the model sensitivity, Gonzales et al showed that this protocol did not significantly modify A549 cell survival.⁴⁸ In addition, the absence of serum did not induce striking THP-1 or HL-60 cell death for the duration of the experiment (24 h).

The evaluation of the genotoxicity of silica in epithelial cells was performed using the comet assay.³³ DQ12 but not DQ12-PVNO induced DNA strand breaks in A549 cells in monocultures, as previously shown in the literature.⁴⁷ However, in the presence of THP-1 cells with or without HL-60, a decrease in DQ12-induced DNA breaks was observed along with a lower cell toxicity. These results agree with the data from the literature, which also show the protective effect of THP-1 cells in co-culture models with epithelial cells in response to particle treatment.^{16,35,36,41,49} The phagocytosis of particles by macrophages preventing their interaction with epithelial cells could be an explanation of the lower toxicity of particulate matter in such co-culture models. One may also argue that DQ12 treatment is able to induce deleterious ROS production in A549 cells alone while in the presence of THP-1; this particle has no effect on the production of these reactive species as compared to the untreated cells, suggesting that ROS could be involved in their toxicity. However, the levels of ROS in A549 cells are 20 times higher when they are cultivated with THP-1. The method used does not provide any information regarding the nature of the detected ROS. Indeed, 2′,7′-dichlorofluoresceine fluorescence is proportional to the formation of ROS including peroxynitrite radicals and OH°, as well as molecules containing peroxyl, alkoxyl, carbonate (CO₃°-) and NO₂° groups,⁵⁰ whereas their cellular effects are different. In this context, some of the ROS produced in the co-culture model could be the consequence of cellular signaling and act as secondary messengers.^39,51 Therefore, this method was unable to discriminate ROS induced by chemicals and deleterious to the cells, from those produced as secondary messengers.

As observed in our work and in agreement with the literature, the increase of ROS production was also associated with the increase of IL-8 synthesis.^52,53 Indeed, DQ12 treatment stimulated the production and release of this cytokine in A549 cells whereas DQ12-PVNO did not. The presence of THP-1 cells induced an exacerbated release of this interleukin as well as an augmented gene expression in A549 cells without any particle treatment. However, this induction was increased by the higher concentration of DQ12 treatment but not with DQ12-PVNO. These results are in accordance with data from the literature which shows that the co-culture of these two cell lines lead to an increased release of cytokines, such as IL-8 and TNFα.^16,36,41 In addition, TNFα, which has been shown to stimulate the production of IL-8 in HTB54 human pulmonary epithelial cells in response to particulate matters,⁵⁴ is constitutively produced by untreated THP-1 cells. Therefore, these results revealed an important communication between the two cell lines, emphasizing the fact that mono-culture models do not consider these crosstalk mechanisms that may occur in the lung. To a lesser extent, similar results were observed in the co-culture of HL-60 and A549 cells where the increase induced by silica treatment is stimulated. They are in agreement with the work from Arnold et al.,⁵⁵ which showed that the secretion of IL-8 was higher in a co-culture of A549 cells with polymorphonuclear granulocytes than in A549 cells alone before or after infection with a respiratory syncytial virus. In addition, the complexity of the cellular signaling and interaction between the three cell lines also observed by others^13,56,57 is strengthened by the proteome profiler assay where we saw that the production of various proteins has been altered in response to particle treatment. The intensity of changes and the name of the proteins may vary depending on the cell combination. However, it is interesting to note that two proteins, MCP-1/CCL-2 and GROα/CINC-1/CXCL-1, are only overexpressed in the co-cultures of A549 and HL60 exposed to DQ12. The two inflammatory cytokines are produced by epithelial cells and are chemoattractant for macrophages and neutrophils, respectively.^58–60 Even though HL60 cells have a lower impact on the (geno)toxicity of DQ12 in A549 cells than THP-1, these results show a specific effect of these granulocytic cells on epithelial cells. Generally, DQ12-PVNO particles usually have fewer effects on protein production compared to DQ12. This is in accordance with what has been mentioned above. Nevertheless, the results observed in vitro must be appreciated with care. Indeed, the IL-8 levels observed should be higher than those present in normal lungs. In addition, the sole presence of THP-1 cells induced the expression of the cyclooxygenase 2 gene in A549 cells, an enzyme playing a role in inflammation.⁶¹ Feedback mechanisms probably present in the lung might not be reproduced in in vitro models. TPA-differentiated THP-1 cells are supposed to be M0 macrophages^62,63 but they remain surrogates of alveolar macrophages. The latter have, from a developmental point of view, a different origin than macrophage-derived monocytes.⁶⁴ Therefore, it cannot be ruled out that they may have different phenotypes and cellular interactions with epithelial cells. In addition, the co-culture model using cancer cell lines, for example, A549 cells, is established from adenocarcinoma, therefore these cells might only keep a proportion of type II alveolar epithelial cell properties. These cells constitutively overexpressed heme oxygenase 1 which is an anti-oxidative enzyme associated with a deregulation of the NRF2 transcription factor.⁶⁵ This alteration could interfere in the response of epithelial cells to stimuli from THP-1 and HL-60 cells. For example, it was shown that an overexpression of the heme oxygenase 1 inhibits the signaling pathway induced by TNFα.⁶⁶

The model developed here uses cell lines cultivated under submerged conditions and then does not totally relate with the in vivo conditions where cells are at the interface between air and liquid. Then various co-culture models involving A549 cells and differentiated THP-1 at the interface air-liquid (ALI) have been published to assess toxicity of particulate matters.^67–69 However, some papers have compared the toxicity of particles in A549-THP-1 co-culture under submerged and ALI conditions. It has been shown that alpha-quartz (Min-U-Sil 5) was more cytotoxic and induced more differential gene expression under submerged conditions than under ALI and its effects on IL-8 production were also more pronounced under submerged conditions.^70,71 In addition, a study comparing the toxicological profile of nanostructured metal oxides showed that A549 and THP-1 cells in submerged conditions have positive correlations for inflammatory responses with pulmonary neutrophils influx in mice following exposure to such nanomaterials. For this set of nanomaterials, the genotoxicity in A549 submerged cells and co-culture in the ALI system is concordant with their genotoxicity in mouse lung tissue. The authors concluded that complex models exposed in ALI systems did not improve remarkably the assessment of inflammatory responses compared to cell type A549 or THP-1 in submerged conditions.⁷²

Although this in vitro model appears to be useful for the detection of early events induced by particle treatment such as oxidative stress or inflammatory response, it was not able to reflect a kinetic effect. Even if it could show the complexity of the interactions between particles and the different cell types, it does not enable an evaluation of the chronic effects and consequently, the progression or the reversibility of the toxicity induced by particles.

Conclusion

In conclusion, this study has tried to decipher, in vitro, the complex interaction between the main actors of pulmonary inflammation in response to particle exposure (i.e. epithelial cells, macrophages, and granulocytic neutrophils). These inter-cellular communications represent an important point in the study of the inflammatory potential of particles. However, it is important to note differences between the lung tissue and the co-culture of cancer cell lines. Developments are still needed to understand the in vitro mechanism of particle action with the perspective to evaluate the risk assessment of fiber and particle inhalation.

Author’s contribution

L.G. conceived the study; S.S., M.L. performed the data collection and produced the figures; S.S., M.L., C.S. and C.D. performed the data analysis and interpretation, with overall guidance from L.G. All authors wrote the manuscript.

Funding

None declared.

Conflicts of interest. The authors have no conflict of interest to declare.