FMRP protects the lung from xenobiotic stress by facilitating the integrated stress response

The epithelial lining of the respiratory tract is continually challenged by a diverse array of environmental toxicants, including gases, particulates and biological agents. Exposure to these agents leads to increased oxidative, genotoxic and endoplasmic reticulum stress. Such stresses lead to cellular damage, inflammation and, in the long term, to lung damage and decreased lung functionality. The goal of this study was to probe the mechanisms by which lungs cope with environmental insults.

The capacity of the lung to manage xenobiotic stress is dependent on stress response proteins that are induced upon insult. In this regard, the integrated stress response (ISR) pathway is an evolutionarily conserved pathway that is integral to how the lung copes with environmental challenges (Pakos-Zebrucka et al., 2016; van ‘t Wout et al., 2014; Konsavage et al., 2012). The ISR is triggered by the activation of one or more of the four stress-responsive kinases GCN2 (also known as EIF2AK4), PKR (EIF2AK2), PERK (EIF2AK3) and HRI (EIF2AK1). The activation of these kinases, in turn, sets in motion two separate but interdependent processes that enable cells to mount a restorative response (Wong and Wispe, 1997). First, these kinases phosphorylate eukaryotic initiation factor 2α (eIF2α, also known as EIF2S1) and shut off ongoing programs of protein synthesis. The inhibition of translation leads to the sequestration of translationally active mRNAs into stress-induced condensates of various types. Second, activation of the kinases also induces specialized modes of protein translation, leading to the expression of stress response proteins. More specifically, these specialized translation regimes upregulate expression of activating transcription factor 4 (ATF4) (Pakos-Zebrucka et al., 2016; van ‘t Wout et al., 2014) and, in turn, ATF4 targets such as ATF3. ATF4 also synergizes with other transcription factors activated in response to stress, such as Nrf2 (NFE2L2), to induce the expression of stress response genes (He et al., 2001; Sarcinelli et al., 2020).

The fragile X mental retardation protein (FMRP, encoded by FMR1) is a multifunctional protein that is expressed in the brain and other organs, in humans and other animals alike. Deficiencies in FMRP lead to fragile X mental retardation syndrome (FXS), a disease characterized by mild-to-moderate intellectual disability (Zhou et al., 2014). FMRP function has been most intensively studied in the neuronal context, wherein the protein has been shown to regulate synaptic plasticity by multiple mechanisms (Santoro et al., 2012). Aside from this well-established role, several studies indicate that FMRP also has a role in facilitating stress responses. At a cellular level, FMRP has been shown to play an essential role in genesis of stress granules in response to arsenite and heat shock (Didiot et al., 2009; Linder et al., 2008). A recent study on fibroblasts derived from Fmr1 knockout (KO) mice showed that FMRP is required for a specialized DNA damage response (DDR) in response to agents such as aphidicolin, 5-hydroxyurea (5-HU) and UV (Alpatov et al., 2014). The central finding of this study is that FMRP has a chromatin-dependent role in resolving stalled replication forks and single-strand breaks in DNA (Alpatov et al., 2014), but not in response to other types of genotoxic stress.

The lung is routinely exposed to a variety of environmental toxicants that cause many different types of stress. Our interest in mechanisms that regulate the pulmonary stress response led us to explore the role of FMRP in the lung. We immunostained murine and human lungs for FMRP to find that the protein is expressed in the airway epithelium and more broadly. To probe the role of FMRP in stress responses in the airways, we subjected Fmr1 KO mice to naphthalene (Nap) injury, a well-established model for oxidative and genotoxic stress. We found that the airways of Fmr1 KO mice exhibited higher expression of markers of oxidative and genotoxic stress, and greater cell death, than wild type. These findings led us to investigate the role of FMRP in airway stress responses, and in the ISR pathway, in mice and in the human lung.

To characterize the role of FMRP in the pulmonary stress response, we examined the expression of the protein in adult lungs from wild-type (WT) and Fmr1 KO animals. Lung sections from WT mice were stained with anti-FMRP antisera and examined under a confocal microscope (5 μm, n=8). FMRP expression was detected throughout the lung (Fig. 1A). We detected widespread protein expression in airway epithelium, both in secretory club cells (CCs, marked by expression of Scgb1a1, Fig. 1A,C) and in ciliated cells (marked by expression of acetylated tubulin, Ac-tub, Fig. 1A,C). Outside the airways, we noted intermittent expression in the alveolar parenchyma (Fig. 1A). Lung sections of Fmr1 KO mice stained with the same anti-FMRP antisera did not show any specific staining (airways shown in Fig. 1B,D, n=3). Taken together, these experiments showed that FMRP is expressed in the murine lung, in the airways and more broadly. Next we examined hematoxylin and eosin (H&E)-stained lung sections from WT and Fmr1 KO mice to compare morphologies of the lungs. We found that lungs from WT and Fmr1 KO were comparable (Fig. S1A,B).

To investigate the role of FMRP in the pulmonary stress response, we focused our attention on FMRP-expressing airway CCs. Airway CCs are highly sensitive to the polycyclic hydrocarbon Nap (Stripp et al., 1995; Van Winkle et al., 1995). Nap administration leads to the loss of the vast majority of CCs from the airway epithelium within 24–48 h and is a well-established model for lung injury (Guha et al., 2014, 2017). The susceptibility of airway CCs to Nap is a result of the expression, in CCs, of the cytochrome P450 enzyme Cyp2f2 (Buckpitt et al., 2002). Cyp2f2 converts Nap to naphthalene oxide, which causes DNA damage. Naphthalene oxide is also converted to naphthoquinones, which cause oxidative stress (Buckpitt et al., 2002). Thus, to probe the role of FMRP in the pulmonary stress response in mice, we decided to utilize the Nap injury model. Interestingly, the Cyp2f2 isoform that converts Nap to cytotoxic derivatives is not expressed in humans and consequently Nap does not affect humans in the same way.

We exposed WT and Fmr1 KO animals to Nap and harvested lungs for analysis at different timepoints post injury (regimen shown schematically in Fig. 1E). To assess the extent of injury, we quantified frequencies of CCs across timepoints and examined expression of markers of oxidative and genotoxic stress. We found that the frequencies of CCs in WT were significantly higher than in Fmr1 KO at 12 h, 24 h and 48 h, respectively (Fig. 1F, n=3 mice per genotype per timepoint). In other words, cell loss was greater and faster in Fmr1 KOs. Next, we stained sections from mouse lung prior to and post Nap injury with two antisera: anti-4-hydroxynonenal (4HNE, a product of lipid peroxidation and a marker of oxidative stress) and anti-γ-H2AX (a phosphorylated histone variant that is a marker of double-stranded DNA breaks and genotoxic stress). We did not detect expression of either marker in the lungs from uninjured WT and Fmr1 KO mice (Fig. 1Gi,ii,Hi,ii). In contrast, the expression of both markers was dramatically increased in CCs in Nap-injured lungs (Fig. 1G,H). Pertinently, we noted that the levels of 4HNE and γ-H2AX expression in CCs were lower in WT than in Fmr1 KOs at all timepoints examined (Fig. 1Giii–vi,Hiii–vi; see also the quantification in Fig. S1C,D, n=3 mice per genotype per timepoint). Based on these data, we concluded that CCs in Fmr1 KO animals are more susceptible to Nap-induced stress.

To further probe the role of FMRP in stress responses in CCs, we turned to the murine club-cell-like cell line C22. C22 cells were derived from H-2Kb-tsA58 mice expressing a temperature-sensitive isoform of the SV40 large T antigen under the H-2Kb promoter (Demello et al., 2002). To characterize these cells, we stained C22 cells with markers of CCs and other airway and alveolar lineages. Consistent with previous reports, these cells expressed the CC marker Scgb1a1 (Fig. 2A) and did not express markers of other lineages (data not shown). We then performed a series of experiments to determine whether C22 could be utilized as a model for Nap injury, and to probe the role of FMRP therein.

First, C22 cells were stained with antisera against Cyp2f2 and FMRP. We found that C22 cells expressed modest levels of Cyp2f2 (Fig. 2B, n=6 experiments) and expressed FMRP (Fig. 2C, n=9 experiments). Next, we optimized methods for the knockdown of gene expression in C22 cells using siRNAs and methods for Nap challenge. We established that treatment with two different Cyp2f2 siRNAs and three different FMRP siRNAs was sufficient to reduce Cyp2f2 and FMRP expression, respectively, by 80% or greater (see the Materials and Methods; compare Cyp2f2 or FMRP expression in scrambled siRNA-treated cells, Sc, and Cyp2f2 or Fmr1 siRNA-treated cells, Si, in Fig. S2A–C, n=3 experiments, and Fig. 2D, respectively, n=9 experiments). Careful titration of Nap dosage and time of exposure (see the Materials and Methods) showed that a 1 h pulse of Nap was sufficient to induce expression of oxidative and genotoxic stress markers in C22 cells and marginally increase cell death 24 h post exposure (see the Materials and Methods). We subsequently incubated control (scrambled siRNA-treated cells, Sc) and Cyp2f2-depleted (Cyp2f2 siRNA-treated cells, Si) cells with Nap for 1 h and harvested cells at different timepoints for analysis (regimen shown schematically in Fig. S2D; see the Materials and Methods). Levels of 4HNE and γ-H2AX increased in Sc cells within 6 h and returned to baseline by 24 h (Fig. S2E–H, n=3 experiments each). In contrast, levels of expression in Si-cells remained at baseline levels at all timepoints (Fig. S2E–H, n=3 experiments each). These data showed that C22 cells are susceptible to Nap-induced stress in a Cyp2f2-dependent manner like CCs in vivo.

To explore the possibility that FMRP regulates susceptibility to Nap in C22 cells, we incubated control (scrambled siRNA-treated cells, Sc) and FMRP-depleted (Fmr1 siRNA-treated cells, Si) cells with Nap for 1 h and then harvested cells at different timepoints for analysis (shown schematically in Fig. 2E). To assess levels of oxidative and genotoxic stress, we again stained cells with anti-4HNE and anti-γ-H2AX, respectively (Fig. 2F,H). In order to assess the cytotoxicity of Nap, we subject the cells to a WST-1 assay 24 h post exposure. We found that levels of 4HNE and γ-H2AX (Fig. 2F–I) were elevated in Fmr1-depleted cells at all timepoints (n=3 experiments each) and that Fmr1-depleted cells exhibited greater cell death in response to Nap (Fig. 2J). These data correlated well with the increased susceptibility of CCs to Nap in Fmr1 KO animals and demonstrated that FMRP has a cell-intrinsic role in protecting cells from Nap.

A role for FMRP in mediating stress responses has been reported previously. One of these studies has pointed towards a role for the protein in stress granule biogenesis in response to arsenite or heat shock (Didiot et al., 2009). This study suggests that FMRP may have a role in the induction of the ISR, a pathway necessary for stress granule biogenesis, or a more specific role in the induction of stress granule biogenesis, or both. As the ISR has been shown to be important for stress responses in the lung, we decided to examine the possibility that FMRP may be required for the induction of the ISR in C22 cells post Nap. We note that FMRP-deficient cells also fail to recruit γ-H2AX to stalled replication forks and single-strand breaks in response to aphidicolin, 5HU and UV (Alpatov et al., 2014). Although it is plausible that FMRP serves a similar role in Nap-treated cells, we noted that the nuclear accumulation of γ-H2AX in FMRP-deficient CCs and C22 cells post Nap was greater than in the respective controls. This suggested to us that the DDR was at least partially active in FMR-deficient cells and, more importantly, that extent of DNA damage (as reported by nuclear γ-H2AX accumulation) was greater in FMR-deficient cells than in controls (see the Discussion). Taken together, the findings led us to investigate the role of FMRP in the ISR pathway.

As previously mentioned, the ISR is induced when one of four stress-responsive kinases (GCN2, PERK, HRI, PKR) phosphorylate eIF2α at serine-51. Phosphorylation of eIF2α arrests conventional translation, promotes sequestration of mRNAs being actively translated and enables specialized translation of mRNAs encoding stress response proteins such as ATF4 (Pakos-Zebrucka et al., 2016; van 't Wout et al., 2014). To probe the status of the ISR in C22 cells post Nap, we examined the phosphorylation state of eIF2α. We exposed C22 cells to Nap for 1 h, harvested cells at various timepoints and quantified the levels of expression of both eIF2α and phosphorylated eIF2α (p-eIF2α). Western-blot-based ratiometric quantification of total and p-eIF2α in Sc cells showed that p-eIF2α levels increased 3 and 6 h post injury and decreased to baseline levels thereafter (Fig. 3A; Fig. S3A,B, n=5 experiments). We inferred that the ISR is induced in C22 cells in response to Nap. We then exposed Si cells to Nap for 1 h and found that, contrary to controls, the levels of p-eIF2α did not increase post Nap (Fig. 3A; Fig. S3A,B, n=5 experiments). The analysis of p-eIF2α suggested that FMRP depletion might inhibit the ISR.

Next, we examined the expression of ATF4 and its target, ATF3, in control and FMRP-deficient cells. Sc and Si cells were stained with an anti-ATF4 antibody prior to and post Nap treatment. In Sc, the expression of ATF4 was undetectable in untreated cells, increased dramatically at 3, 6 and 12 h post Nap treatment and then approached baseline levels at 24 h (Fig. 3B,C, n=5 experiments). In Si cells, levels of ATF4 were negligible in untreated cells and showed no appreciable increase post Nap treatment (Fig. 3B,C, n=5 experiments). Next, we assayed ATF3 levels by quantitative real-time PCR (qPCR). For this, RNA was isolated from Sc and Si cells at different timepoints and subjected to qPCR analysis. In Sc, levels of ATF3 mRNA increased at 3 and 6 h post Nap and returned to baseline thereafter (Fig. S3C, n=3 experiments). In Si, the ATF3 levels did not rise appreciably above baseline post Nap (Fig. S3C). These findings were also validated with anti-ATF3 immunostaining (data not shown). Based on these data, we concluded that both ATF4 and ATF3 expression are perturbed in FMRP-deficient cells post Nap. Taken together, the findings showed that the ISR is perturbed in FMRP-deficient cells post Nap.

We also probed whether the upstream kinases that phosphorylate eIF2α and induce the ISR become activated (phosphorylated) in FMRP-deficient cells. We probed the expression of GCN2, PERK, HRI, PKR and their phosphorylated isoforms in Nap-treated C22 cells using commercially available antibodies (see the Materials and Methods). Among all pairs of antisera tested, antisera for PKR and p-PKR provided reproducible results. Western-blot-based ratiometric quantification of p-PKR and total PKR in Sc and Si cells showed that the p-PKR levels increase in both Sc and Si 3 h post Nap (Fig. S3D–F, n=3 experiments). Importantly, we noted that levels of p-PKR returned to baseline in Sc at 6 h and later timepoints, but remained significantly higher in Si at later timepoints (Fig. S3D–F). This suggested that at least one of the stress-responsive kinases (PKR) is activated in FMRP-deficient cells post Nap.

Perturbations to the ISR provided a plausible explanation for why FMRP-deficient C22 cells are more susceptible to Nap. To test this, we decided to probe how perturbing the ISR, by knocking down levels of Atf4, would impact susceptibility to Nap. Control (scrambled siRNA) and Atf4 siRNA-treated C22 cells were exposed to Nap as described earlier and cells were harvested at different timepoints for analysis. ATF4 immunostaining of control and Atf4 siRNA-treated cells showed that siRNA treatment eliminated ATF4 expression in cells post Nap exposure (Fig. 3D). We also found that ATF4-depleted cells exhibited increased expression of 4HNE (Fig. 3E; representative images shown in Fig. S3G) and γ-H2AX (Fig. 3F; representative images shown in Fig. S3H) at all timepoints examined (n=3 experiments each, quantification of cell fluorescence based on n≥25 cells per experiment) and increased cell death 24 h post injury (Fig. 3G). We concluded that the increased levels of oxidative and genotoxic stress and increased cytotoxicity observed in FMRP-deficient cells could be the result of a failure to induce the ISR and upregulate ATF4.

In light of the findings in C22 cells, we examined whether perturbations to the ISR are also observed in FMRP-deficient CCs in Nap-treated mice. We counterstained sections from control and Fmr1 KO lungs post Nap with antisera to both ATF4 and ATF3. Although ATF4 immunostaining was inconclusive, we noted that the levels of ATF3 were negligible in CCs in the control lung and upregulated post Nap (Fig. S3I,J, sections from n=3 mice). Pertinently, the levels of ATF3 in CCs in Fmr1 KO did not increase post Nap. These results are consistent with a role for FMRP in the induction of ISR in CCs post Nap.

The findings in the murine lung led us to ask whether FMRP has a conserved role in the human lung. To investigate this possibility, we first examined the distribution of FMRP in the human lung. Paraffin sections stained with FMRP antisera showed that FMRP is expressed throughout the airways and more broadly (Fig. 4Ai,ii,iv, n=2 sections each from n=5 independent lung biopsies). Triple labeling experiments with markers for ciliated cells and CCs showed that FMRP is expressed in both ciliated and non-ciliated cells, including CCs. Based on the distribution of FMRP, we surmised that the protein could also play a role in the airways in the human lung.

The BEAS-2B cell line is derived from normal human airways. These cells do not express markers of ciliated cells and, akin to CCs, have characteristics of non-ciliated cells. We stained BEAS-2B cells with FMRP antisera to find that these cells expressed FMRP (Fig. 4B, n=6 experiments). We then proceeded to develop an assay to probe the role of FMRP in stress responses in these cells.

Since the susceptibility of airway CCs to Nap is not recapitulated in the human lung or in BEAS-2B cells (data not shown), we utilized a different injury model to probe the role of FMRP in stress responses in human cells. 9,10-Phenanthrenequinone (PQ) is an air pollutant that is present at high levels in diesel exhaust particles and is known to trigger oxidative and genotoxic stress (Lavrich et al., 2018). As part of our characterization of PQ, we first exposed control (scrambled siRNA-treated cells) and FMRP-depleted (Fmr1 siRNA-treated cells) C22 cells to a pulse of PQ for 1 h and harvested cells at different timepoints for analysis (data not shown). Consistent with our findings in the Nap model, we found that FMRP-depleted C22 cells exhibited increased expression of 4HNE and γ-H2AX and increased cell death 24 h post exposure (data not shown). We then examined ATF4 expression to find that although ATF4 levels increased in Sc cells at 3 h and 6 h post PQ, no expression was detected in Si cells (data not shown). These experiments showed that PQ treatment does lead to oxidative and genotoxic stress, and that FMRP-deficient C22 cells are more susceptible, and led us to examine the effects of PQ on BEAS-2B cells.

To test the role of FMRP in BEAS-2B cells, we determined that the protocol for the knockdown led to a 90% reduction in the levels of FMRP post treatment (Fig. 4B,C, n=6 experiments). Next, we exposed control (scrambled siRNA-treated cells) and FMRP-depleted (FMR1 siRNA-treated cells) BEAS-2B cells to a pulse of PQ for 1 h and harvested them at different timepoints for analysis (shown schematically in Fig. 4D). We found that FMRP-depleted cells exhibited increased expression of 4HNE (Fig. 4E,F) and γ-H2AX (Fig. 4G,H) at all timepoints examined (n=3 experiments each, quantification of cell fluorescence based on n≥25 cells per experiment) and increased cell death 24 h post injury (Fig. 4I). These experiments showed that FMRP-deficient BEAS2B cells are more susceptible to PQ.

Next, we determined whether FMRP is required for the induction of the ISR in BEAS-2B cells. As described previously, we probed the phosphorylation status of eIF2α (Fig. 5A; Fig. S4A,B, n=5), the levels of ATF4 induction (Fig. 5B,C, n=5) and the levels of ATF3 induction (Fig. S4C, n=3), and the ratio of phosphorylation status of PKR (Fig. S4D–F) at different times post PQ. These experiments showed that although p-PKR levels were increased in both Sc and Si post PQ, all of the downstream processes of the ISR were perturbed in Si.

We then investigated whether the loss of ATF4 would recapitulate the loss of FMRP post PQ. Control (scrambled siRNA) and ATF4 siRNA-treated BEAS-2B cells were exposed to PQ as described previously, and cells were harvested at different timepoints for analysis. Consistent with expectations, ATF4 siRNA-treated cells showed no anti-ATF4 immunostaining post PQ exposure (Fig. 5D, n=3 experiments). We found that ATF4-depleted BEAS-2B cells exhibited increased expression of 4HNE (Fig. 5E; representative images shown in Fig. S4G) and γ-H2AX (Fig. 5F; representative images shown in Fig. S4H) and increased cell death 24 h post injury (see the Materials and Methods, Fig. 5G). These data indicated that the loss of ATF4 largely phenocopies the loss of FMRP in PQ-treated BEAS-2B cells.

The findings in BEAS-2B cells suggested that the role of FMRP in the actuation of the ISR pathway is conserved. To probe whether this finding is more broadly applicable to the lung, we performed the assays described above in another cell line of epithelial origin: A549 cells. We found that although A549 cells are of alveolar origin, they also express FMRP (Fig. S5A). Importantly, PQ exposure assays showed that A549 cells lacking FMRP exhibit higher levels of oxidative and genotoxic stress and fail to actuate ATF4 expression (Fig. S5B–I). Taken together, the studies in BEAS-2B and A549 strongly suggest that FMRP also regulates the induction of the ISR in the human respiratory epithelium.

The aim of this study was to probe the role of FMRP in stress responses in the lung. We report that FMRP plays an essential role in protecting the airways in mice, and potentially in humans, from the deleterious effects of xenobiotic stress. Our studies provide strong evidence that FMRP protects the lung by facilitating the induction of the ISR (see model, Fig. 6). In the paragraphs that follow we will discuss the plausible mechanism(s) by which FMRP may regulate the ISR, the possibility that FMRP regulates stress response pathways in addition to the ISR, and the clinical implications of the findings reported here.

A major finding of our study is that FMRP is required for the actuation of the ISR pathway. The mechanism by which FMRP regulates this step is currently unknown. We find that the stress-responsive kinase PKR is activated in FMRP-deficient cells but that the phosphorylation of the PKR substrate, eIF2α, is perturbed. This suggests that the role of FMRP may be downstream to the activation of stress-responsive kinases. The analysis of FMRP-binding proteins in neuronal and other tissues has identified numerous interacting partners. Among these interacting partners are the proteins Caprin1 and G3BP1, which have independently been implicated in the induction of the ISR pathway in response to stress (Taha et al., 2020; Wu et al., 2016). Pertinently, both Caprin1 and G3BP1 have been shown to be important for eIF2α phosphorylation (Reineke et al., 2015; Solomon et al., 2007). Thus, it is plausible that FMRP acts in concert with Caprin1 and G3BP1 to facilitate eIF2α phosphorylation. Although eIF2α phosphorylation is an early event in the ISR pathway and perturbations at this stage are likely to affect all downstream processes, our data do not allow us to rule out the possibility that FMRP has independent roles either upstream or downstream. Our future experiments will probe these possibilities.

Studies that have examined the role of FMRP vis-à-vis stress responses suggest that FMRP could protect cells from stress in myriad ways. For example, it has been demonstrated that FMRP plays a chromatin-dependent role in inducing the DDR. This could be relevant in the context of the lung. Along the same lines, there is also evidence that FMRP regulates the expression of superoxide dismutase 1 (SOD1) in the brain (Bechara et al., 2009). Levels of SOD1 are reduced in the brains of Fmr1 KO animals. As SOD1 has an important role in protecting cells from stress, FMRP could alter the susceptibility of tissues to stressful stimuli by altering the baseline levels of SOD1. To investigate this possibility, we probed levels of SOD1 in the brain and lung using both western blot and immunohistochemical approaches (Fig. S6). Although we did observe that SOD1 levels in the brain were lower in Fmr1 KO than wild type, the levels of SOD1 in the lung were comparable (Fig. S6A,B). Moreover, we also analyzed SOD1 levels in the bronchial cell lines (C22 and BEAS-2B) with or without FMRP to find that SOD1 levels were comparable (Fig. S6D–I). Taken together, these data show FMRP is unlikely to regulate SOD1 expression in the lung. Nevertheless, the role of FMRP in the DDR (Alpatov et al., 2014), and in the regulation of SOD1 expression in the brain, show that FMRP can contribute towards protecting tissues from stress by ISR-independent mechanisms as well.

Although FMRP is expressed in many tissues in humans and mice alike (https://www.genecards.org/cgi-bin/carddisp.pl?gene=FMR1), historically, FMRP has almost exclusively been studied in a neural context because of its connection with intellectual disability. An important finding of this study is that it demonstrates a role for FMRP in the lung. Although the data implicating a role for FMRP in the human lung is based on findings in cell lines and requires validation in more physiologic assays, the data clearly point towards a potential vulnerability in individuals with an FMR1 deficiency. Clinically, the bulk of the case studies on FXS patients are derived from geographic regions where the load of pulmonary environmental stressors is low. Our study suggests that individuals with FXS living in areas of higher pollutant load may be more susceptible to lung damage and disease, and FMRP status in the lung may be a strong correlate of resilience to pulmonary insults.

All animal work reported here has been approved by the Internal Animal Users Committee (IAUC) and the Institutional Animal Ethics Committee (IAEC) at inStem. Any procedure that could conceivably cause distress to the animals employed pre-procedural anesthesia with isoflurane gas (Baxter Healthcare Corp.), delivered by an anesthetic vaporizing machine. All animals were monitored for signs of distress and killed if in distress. The analysis of human biopsies was approved by the Institutional Ethics Committee of JSS Medical College.

An Fmr1 knockout (Fmr1 tm1 Cgr) mouse (Mus musculus) strain was maintained on a C57BL/6J background at Brain Development and Disease Mechanisms (BDDM), inStem. Genotyping of the animals was performed using established protocols (Bakker et al., 1994).

Human (Homo sapiens) lung tissue was obtained from five subjects at autopsy by a forensic pathologist from JSS Medical College, Mysore. The cause of death was not attributed to lung trauma. De-identified samples were fixed in 4% paraformaldehyde (PFA) at 4°C overnight, embedded in paraffin, and processed for immunohistochemical analysis. All human tissues were obtained following due protocols and all clinical investigations have been conducted according to the principles expressed in the Declaration of Helsinki.

Human lung (BEAS-2B) non-ciliated airway epithelial origin cell line and human alveolar basal epithelial adenocarcinomic (A549) cell line were obtained from Johns Hopkins University (kind gifts from Prof. Shyam Biswal, Department of Environmental Health and Engineering, Johns Hopkins Bloomberg School of Public Health, USA; Singh et al., 2009, 2013). The murine club cell line (C22) was purchased from ECACC, UK (cat. no. 07021401, #07D022). All cell lines were tested for mycoplasma contamination and found to be negative. BEAS-2B were grown in DMEM: F12K (Gibco, USA, 21127030) (1:1) medium and A549 cells were grown in DMEM, supplemented with 10% FBS (Gibco, USA, 10082147) and penicillin-streptomycin (Gibco, USA, 1540122) at 37°C and 5% CO₂. The C22 cell line was maintained in a proliferative state as per the supplier’s instructions, and experiments were performed 24 h post differentiation. Experiments were conducted within the 3rd to 7th passages for BEAS-2B and A549, and within the 3rd to 12th passages for C22.

For Nap injury in mice, wild-type or Fmr1 KO mice aged ≥8 weeks of age (2 males and 1 female per timepoint per genotype) were injected intraperitoneally with corn oil (vehicle, Sigma, USA, C8267) or with Nap dissolved in corn oil (300 mg kg⁻¹, Sigma, USA, 147141) using established protocols (Guha et al., 2014, 2012). Animals were killed 12 h, 24 h or 48 h after injection for analysis.

To establish an assay for Nap injury in C22 cells, we first determined that these cells expressed Cyp2f2, which converts Nap to stress-inducing derivatives. Having established this, we tested a range of concentrations of Nap (50 μg ml⁻¹ to 500 μg ml⁻¹, in DMSO in DMEM). Nap was found to be stable in solution at concentrations up to 100 μg ml⁻¹ and unstable at higher concentrations, leading to cell death within 3 h post exposure. Nap exposure at 50–75 μg ml⁻¹ (DMSO in DMEM, DMSO final concentration 0.7%) for short (1 h) and long (24 h) duration led to a progressive increase in expression of stress markers and mild cytotoxicity after a 24 h period. To probe the effects of FMRP or ATF4 deficiency on susceptibility to Nap, cells were exposed to Nap at 75 μg ml⁻¹ (DMSO in DMEM, DMSO final concentration 0.7%) for a period of 1 h. Cells were then washed in PBS and chased for varying periods of time in complete medium.

It has been reported previously that PQ causes a sharp decrease in the viability of BEAS-2B cells when administered to cells for 24 h at concentrations greater than 1 μM (Koike et al., 2014). We reconfirmed these findings and determined the LD₅₀ to be ∼1.5 μM (Sigma, USA, 275034; dissolved in DMSO in DMEM, DMSO final concentration 0.00002%). To probe the effects of FMRP or ATF4 deficiency on susceptibility to PQ, cells were exposed to PQ at 1.5 μM (DMSO in DMEM, DMSO final concentration 0.00002%) for a period of 1 h. Cells were then washed with PBS, and fresh complete medium was added and chased for varying periods of time in complete medium. A549 cells were also treated with a 1.5 μM dose of PQ for 1 h. Cells were washed with PBS and kept in fresh complete medium and collected at different timepoints for analysis.

Several studies have demonstrated that multiple siRNAs administered together or sequentially work more efficiently for silencing gene expression than a single siRNA (Wang et al., 2016; Fähling et al., 2009; Zhang et al., 2015; Hatch et al., 2010). For our studies we used three or two distinct siRNAs for each targeted gene. siRNAs were administered to cells sequentially, 12 h apart, to silence the gene expression. siRNA transfections were performed with Lipofectamine 2000 (Thermo Fisher Scientific, USA, 11668027). All xenobiotic stress assays in C22 cells were performed 36 h after treatment with the last siRNA. C22 cells were transferred from proliferative to differentiation-inducing medium 12 h after the last siRNA treatment and utilized for xenobiotic stress assays 24 h thereafter. All xenobiotic stress assays in BEAS-2B cells were performed 12 h after treatment with the last siRNA. All siRNAs were obtained from Ambion, USA: murine Fmr1 (4390771), murine Atf4 (16708), murine Cyp2f2 (4390771), human FMR1 (4392420) and human ATF4 (16708), and scrambled (negative control, 4390843). The assay IDs for each of the siRNAs are as follows: mouse Fmr1 siRNA (Assay ID: 5315, 5317, s66177), Human FMR1 siRNA (Assay ID: 5315, 5316, 5317), mouse Atf4 siRNA (Assay ID: 160775, 160776, 160777), Mouse Cyp2f2 siRNA (Assay ID: s64735, s64734), and human ATF4 siRNA (Assay ID: 122168, 122287, 122372).

C22 and BEAS-2B cells were inoculated into a 96-well plate and treated with Nap or PQ for 1 h, as described above, and harvested for analysis 24 h later. Cell viability was assayed using WST-1 reagent (Sigma, USA, 5015944001). Briefly, cells were incubated with WST-1 for 4 h and absorbance readings were taken and analyzed as per the manufacturer's protocols. Cytotoxicity percentage=100 [(OD (450 nm–650 nm) of untreated cells – OD (450 nm–650 nm) of treated cells)/OD (450 nm–650 nm) of untreated cells].

Lungs were inflated with 4% (w/v) PFA (Alfa Aesar, USA, 30525-89-4) in PBS and fixed for 8 h at 4°C. Fixed lungs were subsequently embedded in paraffin, sectioned (5 μm) and processed for immunohistochemical analysis post heat-mediated antigen retrieval at pH 6.0 (Vector Labs, USA, H-3300) except for sections stained with anti-SOD1 antisera, which were subject to antigen retrieval at pH 9.0 (Vector Labs, USA, H-3301). For cellular immunostaining, cells were seeded on coated coverslips (0.1% gelatin, Sigma, USA, G9391), as per the manufacturer's protocol. Post treatment, cells were fixed with 4% PFA for 30 min and blocked with 2% FBS, 0.2% BSA and 0.1% Triton X-100 in 1× PBS for 1 h and stained. Primary antibodies were diluted using the same blocking solution. Immunohistochemical analysis utilized the following antisera: rabbit anti-FMRP (Abcam, UK, 17722, 1:500), rabbit anti-FMRP (Sigma, USA, F4055 1:200), goat anti-Scgb1a1 (Santa Cruz, USA, Sc365992, 1:500), mouse anti-acetylated tubulin (Sigma, USA, T7451, 1:1000), mouse anti-4HNE (Abcam, UK, ab48506, 1:500), rabbit anti-γ-H2AX (Novus Biologicals, USA, NB100-384, 1:1000), mouse anti-Cyp2f2 (Santa Cruz, USA, SC374540, 1:100), mouse anti-ATF4 (Sigma, USA, WH0000468M1, 1:200), rabbit anti-ATF3 (Sigma, USA, HPA001562, 1:200), rabbit anti-SOD1 (Abcam, UK, ab16831, 1:200) and Alexa Fluor 488, 568 or 647-conjugated donkey anti-mouse, -rabbit or -goat IgG secondary antibodies (Invitrogen, USA, A21447, A21202, A21206, A10037, A10042, A31571, 1:300). Stained sections were mounted in ProLong Diamond (Invitrogen, USA, P36962). All samples were imaged on a FV3000 4-laser and FV3000 5-laser confocal microscope or on a Zeiss LSM-780 laser-scanning confocal microscope (Carl Zeiss AG, Germany). For H&E staining, sections were stained with hematoxylin for 10 s and eosin for 30 s, dried and mounted in DPX and imaged on a Nikon Eclipse Ti2 microscope (Nikon, Japan).

Frequencies of club cells/mm of airway, and total cellular fluorescence in club cells, in lung sections, were determined from single tiled optical sections acquired on a confocal microscope using ImageJ software. For club cell frequency analysis, cells attached to the basement membrane were counted per section per animal. Total cellular fluorescence intensity was calculated by subtracting a ‘background’ value per section from the integrated density per cell (outlined using the software) (for FMRP, 4HNE, γ-H2AX and ATF4). The ‘background’ value was determined by sampling integrated density of regions on the section devoid of cells. Total cellular fluorescence of C22 and BEAS-2B cells was estimated from single optical sections on a confocal microscope using ImageJ software. In all experiments involving C22, BEAS-2B and A549 cells, ≥25 cells were analyzed per timepoint, per experiment. The images of Scgb1a1 and Cyp2f2 expression in C22 cells are maximum intensity projection images of z-stacks acquired on a confocal microscope.

Protein was extracted from cell lysates using RIPA buffer (Thermo Fisher Scientific, USA, 89900) containing Sigmafast EDTA-free protease inhibitor cocktail (Sigma, USA, s8830) and PhosSTOP (Merck, USA, 4906845001). Total protein was run on a 12% SDS PAGE gel, transferred onto a nitrocellulose membrane (Amersham, UK, 10600002), and the membrane was stained with reversible MemCode (Thermo Fisher Scientific, USA, 24580) for total protein estimation [imaged on ImageQuant600 (Amersham, UK) and quantified using ImageJ]. The membrane was subsequently destained, blocked with 5% BSA (Sigma, USA, A9418) for 1 h and probed using the following primary antisera: rabbit anti-Phospho-eIF2α (Ser51) (Cell Signaling Technology, USA, 9721S, 1:1000), mouse anti-eIF2α (Cell Signaling Technology, USA, 2103S, 1:1000), rabbit anti-phospho-PKR (Sigma, USA, SAB4504517, 1:3000), mouse anti-PKR (Santa Cruz, USA, Sc-6282, 1:1000), rabbit anti-GCN2 (Cell Signaling Technology, USA, 3302s), mouse anti-phospho-GCN2 (Cell Signaling Technology, USA, 3301S), rabbit anti-PERK (Cell Signaling Technology, USA, 3192s, 1:1000), rabbit anti-phospho-PERK (Cell Signaling Technology, USA, 3179s), mouse anti-HRI (Santa Cruz, USA, sc-365239). Primary antisera was detected using the following secondary antisera: HRP-conjugated anti-rabbit (Abcam, UK, 6721, 1:3000) and HRP-conjugated anti-mouse secondary (Invitrogen, USA, # 62-6520 1:5000) antibodies and ECL (BioRad, USA, 1620177) and analyzed (imaged on ImageQuant600 and quantified using ImageJ). The levels of eIF2α, phospho-eIF2α, PKR and Phospho-PKR were normalized to the total protein content of the respective lanes. For analysis of SOD1 expression from murine brain and lung tissue lysate, tissues were collected after dissection, washed with PBS and protein was extracted with RIPA buffer and protease inhibitor cocktail. Total protein was run on a 12% SDS PAGE gel and transferred onto a nitrocellulose membrane. We used 5% non-fat dry milk (Santa Cruz, USA, Sc-2325) solution in PBS for blocking and probed the blots with anti-SOD1 (Abcam, UK, ab16831) and anti-β-tubulin (CST, USA, 15115S) antisera. Levels of SOD1 were normalized to corresponding β-tubulin levels.

RNA from cell lysates was extracted using TRIzol (Invitrogen, USA, 15596018, as per the manufacturer's protocol) and qPCR was performed using the primers listed in Table S11. The qPCR assays were constituted with the Maxima SYBR Green/ROX qPCR Master Mix (2×) (Thermo Scientific, USA, K0221) and analyzed on a BioRad CFX3 real-time PCR system (BioRad, USA).

Statistical significance of datasets was assessed using unpaired two-tailed t-tests post Shapiro–Wilk tests for normality. Data were also analyzed using a two-way ANOVA with a Sidak post-hoc test to compare changes in two groups with respect to time, genotype and interaction parameters. ANOVA data and normality test results for each figure are presented in a tabular format (Tables S1–S10).

We thank Joseph Jomon, National Centre for Cell Science (NCCS), Pune for sharing antibodies; Aditya Deshpande, inStem, for assistance in animal experiments; Sarfaraz Nawaz and Sudhriti Ghosh Dastidar, inStem, for their assistance with biochemical analyses; Harlin Kaur, Binita Dam, Arnab Karmakar and Mamta Yadav for technical assistance; and the Central Imaging and Flow Cytometry Facility (CIFF) and Animal Care and Resource Center (ACRC) Facility at Bangalore Life Science Cluster (BLiSC) for their constant support.

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.258652.