Per- and polyfluoroalkyl substances (PFASs) modify lung surfactant function and pro-inflammatory responses in human bronchial epithelial cells

No competing interests are declared. Abstract The toxicity of some per- and polyfluoroalkyl substances (PFASs), such as perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) has been studied thoroughly, showing that systemic PFASs targets the lungs. However, regulators lack data to assess the impact of other PFASs on the lungs and alternative methods to test substances for lung toxicity are needed. We combined two in vitro models to assess toxicity to the respiratory system; i) a lung surfactant (LS) function assay to assess the acute inhalation toxicity potential, and ii) a cell model with human bronchial epithelial cells to study pro-inflammatory potential and modulation of inflammatory responses. We tested salts of four PFASs: perfluorobutane sulfonate (PFBS), perfluorohexane sulfonate (PFHxS), PFOS, and PFOA as well as the fluorotelomer 8:2 FTOH. The results show that PFHxS, PFOA and PFOS can inhibit LS function. High PFOS concentrations induced a pro-inflammatory response, measured as increased IL-1α/β release. Moderate concentrations of PFOS suppressed release of the chemokines CXCL8 and CXCL10, whereas both PFOS and PFOA stimulated the release of the pro-inflammatory cytokine IL-1β in immune stimulated human bronchial epithelial cells. These findings support the concern that some PFASs may increase the risk of acute lung toxicity and of airway infections.


Introduction
Per-and polyfluoroalkyl substances (PFASs) have diverse technical applications due to their superior surface-active properties. Emerging evidence shows a link between exposure to indoor air pollutants, like PFASs, and lung diseases (Qin et  Although PFAS exposure and the risk of developing asthma has been studied, the relation is not clear (Humblet et al., 2014,Impinen et al., 2018,Impinen et al., 2019).
The respiratory system is a target for PFAS toxicity, both by inhalation and systemically from the circulating blood. Volatile PFASs, such as fluorotelomer alcohols (FTOHs), can be measured at relatively high levels in indoor air. In addition, consumer products containing PFASs such as fluorinated ski wax, cosmetic sprays or impregnation products may lead to inhalation exposure and add to the total PFAS body burden (Freberg et  PFASs have long terminal half-lives in humans and are persistent in the environment (Haug et al., 2009,Olsen et al., 2017. Consequently, high levels of PFASs can be measured in blood from heavily exposed populations, which leads to chronic systemic exposure of organs, including the lungs (Haug et al., 2009,Haug et al., 2011a. In a mouse study, the level of PFOS was shown to be higher in lung tissue than in blood in both perinatal and adult mice, suggesting the lung to be a target of PFOS toxicity (Borg et al., 2010).
The production and use of perfluorooctanoic acid (PFOA) and its salts and precursor substances are regulated in a REACH restriction that will apply from 2020 (European Chemicals Agency, 2017b). In addition, perfluorooctane sulfonate (PFOS) and In the present study, we have used two in vitro screening strategies to measure effects of PFASs on central mechanisms of the respiratory system. Firstly, we used a LS function assay to screen PFASs for acute inhalation toxicity potential. Dynamic LS function was measured in the constrained drop surfactometer (CDS), where the effect of a chemical can be screened rapidly and at low cost (Da Silva and Sørli, 2018). The assay has previously been used to show a link between inhibition of LS function and induction of acute inhalation toxicity in vivo by impregnation products and inhaled pharmaceutical enhancers (Sørli et al., 2015,Sørli et al., 2017. Secondly, we studied the immunomodulatory effects of PFASs. Bronchial epithelial cells were exposed to PFASs with and without priming with the highly pro-  PFASs (Tab. 1) were purchased from Sigma-Aldrich (Oslo, Norway). Gibco LHC-9 medium was purchased from Thermo Fisher Scientific (Oslo, Norway).

Preparation of native lung surfactant
Native LS was prepared and the phospholipid content determined as described previously (Taeusch et al., 2005). LS was isolated from bronchoalveolar lavage of porcine lungs by multiple centrifugations, including a density gradient ultracentrifugation that cleans potential blood contaminants from the surfactant. The method consistently yields surfactant preparations with highly reproducible batch-to-batch biophysical performance. To perform the described LS function assay, lyophilized LS was diluted to 5 mg/mL in a buffer containing 0.9% NaCl, 1.5 mM CaCl 2 , and 2.5 mM HEPES, adjusted to pH 7.0 (Valle et al., 2015). the CDS was heated to 32°C). The surface tension was recorded for the first 10 cycles, and the minimum surface tension of the 10 first cycles was found.

Control for system interference from other factors than PFASs.
To exclude that the observed inhibition of function was from the DMSO in which all PFASs were dissolved, the solvent effect on surface tension function was assessed. Inhibition caused by DMSO could be excluded as the concentration of DMSO that inhibited LS function was 60 times higher than the concentrations used in any of the inhibitory dilutions of PFASs, and for PFASs that did not cause inhibition of the function, the DMSO concentration approached the concentration where pure DMSO showed inhibition (data not shown). LS function is also affected by extreme pH, and the optimal pH range is between 4 and 7 (Amirkhanian and Merritt, 1995), therefore the dilutions of PFASs in water were tested using pH indicator strips (Alkalit, Merck). All solutions had a pH of 6, therefore pH was excluded as a cause of LS function inhibition. with 5% CO 2 . The medium was changed every 2-3 days. Prior to exposure, cells were plated in 35 mm 6-well culture dishes (220,000 cells per well), grown to near confluence in serum free LHC-9 medium and exposed as described below.

Cell culture exposures
Stock solutions of 100 mM PFAS in DMSO were diluted in culture medium prior to cell exposure. HBEC3-TK cells were exposed to concentrations ranging from 0.13 µM to 60 µM of PFOS, and 0.13 µM to 10 μM of the remaining PFASs. DMSO (0.02%) in culture medium was used as vehicle control. In a pilot experiment the release of CXCL8 after PFOS exposure was investigated after 24 and 48 h, and the 48 h exposure resulted in higher concentrationresponse pattern. Therefore, the remaining exposures were performed for 48 h. In parallel to the exposure to PFASs, the cells were incubated with 5 µg/ml Poly I:C or with Lipopolysaccharide (LPS) 3 h prior to exposure to PFASs.

Cell viability analysis
Cell viability was measured using AlamarBlue cell viability reagent (ThermoFisher Scientific, Oslo, Norway) according to the manufacturer's protocol.
Absorbance was measured using a plate reader (TECAN Sunrise, Phoenix Research Products, Hayward, CA, USA) equipped with Magellan (v1.10) software.

Statistical evaluation
The LS function data were analysed by calculating the statistical difference between the minimum surface tension at the 10 th cycle of the control and the treated sample using ANOVA in the statistical platform R (version 3.4.1). If there was a significant difference J o u r n a l P r e -p r o o f 13 between the control and the treated sample, and the minimum surface tension was higher, a ttest with Bonferroni correction was performed to find the group(s) different from control.
The sample with the lowest concentration and statistical difference from the control was determined to be the lowest observed adverse effect concentration (LOAEC) of the PFAS.
For the cell culture study, statistical analysis of differences in cytokine secretion was performed by one-way ANOVA with Dunnett's post-test for multiple comparisons. Nonnormally distributed data were log-transformed. All calculations were performed using GraphPad Prism 5.04 software (GraphPad Software, Inc., San Diego, CA). The CMC for each PFAS was determined in water at 32°C for all PFAS using the CDS (Tab.

LS function analysis
2) to be able to control for potential influence of micelle formation on the function of LS. The data shows that for the LOAEC was below CMC for the compounds where a LOAEC could be found. For the PFBS and 8:2 FTOH a LOAEC could not be found, the tested concentration of PFAS was at least twice the CMC (Tab. 2).

Human bronchial epithelial cell analyses
The human bronchial epithelial HBEC3-KT cells were exposed to the different PFASs and cytokine release was measured. Furthermore, modulation of cytokine release was analyzed after pre-treatment of the cells with the immune stimulants Poly I:C or LPS.

Cell viability
The effects of the PFASs on cell viability was measured using the AlamarBlue assay. In an J o u r n a l P r e -p r o o f 15 initial experiment, cells were exposed to a large concentration range of PFOS for 48 h to determine the effect on viability. The data showed no apparent reduction in cell viability at concentrations ≤ 30 µM (Fig. 1A). Furthermore, the other PFASs did not show any reduced cell viability at the highest tested concentration (10 µM) (Fig. 2, Fig. S2).

Cytokine release
Pro-inflammatory responses were analyzed in the initial experiment that included high PFOS concentrations. As shown in Fig. 1, exposure to PFOS alone did not induce any biologically significant changes in CXCL8 secretion at the tested concentrations. However, at concentration ≥30 µM, IL-1α and IL-1ß releases were increased. Subsequently, all PFASs were tested at non-cytotoxic concentrations relevant to human exposures, ranging from 0.13 µM to 10 µM, with and without priming with an immunestimulating agent (Fig. 2). Two pro-inflammatory agents were tested, Poly I:C and LPS, mimicking virus and bacterial infection, respectively. In response to LPS, only a modest increase of cytokine release was observed in the HBEC3-KT cells (data not shown), whereas exposure to Poly I:C resulted in a marked upregulation of cytokine release (Fig. 2). Poly I:C was thus used in the subsequent co-exposure experiments. None of the PFASs tested induced significant changes in release of the cytokines measured at non-cytotoxic concentrations in unstimulated cells ( Fig. 2 and Fig. S2). did not induce any statistically significant changes in cytokine release at the concentrations tested (Fig. S2).

Discussion
PFASs represent a versatile group of ubiquitously occurring chemicals of increasing regulatory concern. For many of the PFASs, public information on their hazardous properties, Thus, a screening approach to determine the potential for immunomodulation of the currently used PFASs and their potential substitutions is needed. In the present study, non-cytotoxic J o u r n a l P r e -p r o o f 22 PFAS concentrations, up to 10 µM, caused no significant effects on cytokine release (Fig. 2,   Fig. S2). However, in immune stimulated cells (with Poly I:C) exposed to PFOS, a dose related suppression of CXCL8 and CXCL10 release was observed. A similar tendency was observed for IL-6, but this was not statistically significant. Cytokine suppressive effects has also been reported by Corsini  The differential regulation of cytokines reported in literature, as well as in the present study, points to a likely cell type dependent regulation of cytokines in response to different PFASs.
The mechanism behind such a complex regulation is not clear, but may be related to differences in regulation of cytokine release. Inhibition of NF-κB activation was shown to be related to the suppressive effects of PFASs on cytokine transcription in human leucocytes (Corsini et al., 2012,Corsini et al., 2014. Notably, the IL-1β is synthesized as a precursor protein, stored in the cytoplasm, and is released in response to proteolytic activity (Ozaki et al., 2015). This suggests that various cell types relevant for lung tissue respond differently to the PFASs, but also that the cytokines are induced and/or released via different mechanisms in the same cell type. and rats were exposed to PFOS or PFOA, the pups died shortly after birth (Lau et al., 2003,Lau et al., 2006. Impaired lung function has been proposed to contribute to the neonatal death (Borg et al., 2010). In line with this hypothesis, it has been shown that the alveolar walls in the lungs of rat pups exposed to PFOS during gestation were thicker than in the controls, but neither the LS composition nor the amount were changed (Grasty et al., 2005

Conclusion
In this project, we have used two in vitro test systems: i) an in vitro surfactant assay that is being developed as an acellular test for acute lung toxicity, and ii) an in vitro human bronchial epithelial cell model of inflammatory responses. These two assays are promising tools for the development of an in vitro integrated screening approach for acute inhalation toxicity and immunotoxicity of PFASs, respectively. However, additional cell types of the respiratory tract should be included in a screening strategy since they may differ in their responses to PFAS exposures. Both test systems support that PFASs may be toxic to the respiratory system and points to PFOS and PFOA as the two most potent substances.

Acknowledgement
The work was supported by the Norwegian Environment Agency. J. P.-G. acknowledges funding from the Spanish Ministry of Science and Universities (RTI2018-094564-B-I00) and the