Perfluoroalkyl substance pollutants activate the innate immune system through the AIM2 inflammasome

Perfluoroalkyl substances (PFAS) are widely used in various manufacturing processes. Accumulation of these chemicals has adverse effects on human health, including inflammation in multiple organs, yet how PFAS are sensed by host cells, and how tissue inflammation eventually incurs, is still unclear. Here, we show that the double-stranded DNA receptor AIM2 is able to recognize perfluorooctane sulfonate (PFOS), a common form of PFAS, to trigger IL-1β secretion and pyroptosis. Mechanistically, PFOS activates the AIM2 inflammasome in a process involving mitochondrial DNA release through the Ca2+-PKC-NF-κB/JNK-BAX/BAK axis. Accordingly, Aim2−/− mice have reduced PFOS-induced inflammation, as well as tissue damage in the lungs, livers, and kidneys in both their basic condition and in an asthmatic exacerbation model. Our results thus suggest a function of AIM2 in PFOS-mediated tissue inflammation, and identify AIM2 as a major pattern recognition receptor in response to the environmental organic pollutants.

Perfluoroalkyl Substances (PFAS) are also referred to as Perfluorochemicals (PFCs) are a large family of thousands of synthetic chemicals that are widely used throughout society and found in the environment. They all contain carbon-fluorine bonds, which are one of the most stable chemical bonds and remain in the environment for a long period of time. PFAS has been frequently shown to induce tissue damage and is associated with inflammatory diseases in human and experimental systems. Cellular death and secretion of cytokines including IL-1β have been previously associated with PFAS exposure. The current study investigates the role of inflammasome and showed that PFOS (one of the common types of PFASs) can engage AIM2 inflammasome and induce IL-1β production and pyroptosis in macrophages. Mechanistically, they showed that PFOS induced mitochondrial DNA (mtDNA) release and activate AIM2. Moreover, they show that Aim2 deficient (Aim2-/-) mice, but not NRRP3 deficient (Nlrp3-/-) mice, were protected from PFOS-induced tissue inflammation together describing a novel mode of how PFAS can induce inflammatory disorder.
In general, the study is very well structured and executed. The finding that Aim2-/-, but not Nlrp3-/-, mice resist the PFOS-induced inflammation is in my view the central and the most intriguing observation in this study. However, there are some concerns about the physiologic relevance of these findings considering the real environmental exposure to PFAS and the concentrations and experimental setups (IP injection) used in this study. Mechanistic analysis and results aiming to discover how PFOS induces AIM2 inflammasome are at some points incomplete. In particular, the data showed that PFOS involves a number of generic cellular stress responses including JNK, mitochondrial dysfunction and perturbation which then lead to the release of mtDNA and specifically involves AIM2 inflammasomes. Based on all these observations one would suggest that all cellular responses leading to mitochondrial perturbation, BAX/BAK dependent mitochondrial membrane permeabilization or also the therapeutic treatment of cancer patients using venetoclax (Bcl2 antagonist) should also induce AIM2 inflammasome and cause inflammation.
Major points: 1. All experiments are done using mock treatment as control. I was wondering whether any other comparable compound could be used in control groups to show that the observed effects are solely seen, when cells or mice are exposed to PFOS. 2. One central issue is also the concentrations used for treatment of animals and cells. Could authors provide some example in patients with long-term exposure to PFAS. Are the known (detected) concentrations equivalent to the experimental setups in the current study? 3. As already mentioned, authors should show whether JNK activation or mitochondrial damage, induced by other stress cues (including apoptotic and non-apoptotic stimuli), can also induce AIM2 inflammasome 2. Lane 241-242: "Cytochrome c (cyt C) is a protein normally resided in the inner and outer mitochondrial membrane, participates in the mitochondrial electron-transport chain" cyt C resides in mitochondrial intermembrane space.
In this ms. Wang and colleagues investigate the effects of Perfluoroalkyl substances (PFAS) on cells, proposing, based on extensive experimentation that PFAS drive an inflammatory response dependent on MPTP, BAX/BAK dependent mtDNA release causing AIM2 dependent inflammasome activity and IL-1 release. They then demonstrate in vivo protection from PFAS inflammatory effects in AIM2 ko mice. The study is timely and offers potential insight into some of the toxic effects of PFAS. In the main, the data support the authors' conclusions, however further clarification of how mtDNA is released from mitochondria is required.
-The authors propose that MPTP is required for mtDNA release, this is based entirely upon CsA expts. They then propose that BAX/BAK are required for mtDNA release. The two processes leading to outer membrane permeabilisation have been shown to be separate events (i.e. MPTP does not require BAX, BAK). Two questions arise, is MPTP actually occurring ? (CsA can inhibit multiple targets in addition to CypD, most notably calcineurin), does this step lie upstream of BAX,BAK activity ? To test both possibilities, would suggest deletion of cyp D (through CRISPR) should enable this.
-BAX and BAK are usually considered redundant with one another, yet the authors find that deletion of either prevents mtDNA release, which is somewhat unexpected, and should be commented upon. Is deletion of either also preventing cell death in response to PFAS ?
Minor point -Mitotracker deep red (used in 4a) is not a potentiometric dye (unlike TMRE), the reduced signal after treatment suggests loss of mitochondrial mass upon PFAS treatment. This should be commented upon.
Reviewer #3 (Remarks to the Author): Background: This MS examines the cellular and in vivo mechanisms by which perfluoroalkyl substances (PAFS) trigger inflammation and exacerbation of inflammatory disease in murine models. Due to their surfactant properties, PAFS are present in many consumer products (e.g. stain repellents, paints, non-stick cooking surfaces) but have also been identified as persistent environmental pollutants that can accumulate in living organisms to perturb a wide range of normal biological functions. This study used perfluoroctane sulfonate (PFOS) as a model PAFS which is known to induce accumulation of inflammatory cytokines and cytotoxic responses in multiple tissues, cell types, and animal models.
Experimental Models: Given the ability of PFOS to induce production of IL-1 beta, the authors specifically focus on characterizing the actions of PFOS on inflammasome signaling in: 1) primary murine bone-marrow-derived macrophages (isolated from various inflammasome component genetic knockout strains); 2) human THP1 macrophages (with engineered knockouts of various inflammasome components); 3) an acute (5 day) in vivo model of intraperitoneal PFOS-induced tissue inflammation (in control C57Bl/6 mice and the various inflammasome component knockout strains); and 4) an OVA-induced model of murine asthma with consequent airway/lung inflammation. In the cell-based experiments, they assayed standard readouts of inflammasome signaling including: 1) IL-1b secretion; 2) accumulation of active caspase-1; 3) proIL-1b upregulation and cleavage; 4) ASC oligomerization; 5) gasdermin D cleavage and LDH release (as indices of pyroptotic cell death). In the animal-based experiments, they assayed: 1) inflammatory tissue damage (histology of lungs, liver, kidney); and 2) accumulation of IL-1b, TNFa, and IL-6 in serum, peritoneal fluid, or bronchiolar lavage fluid. In general, the experiments are technically well-designed, are quantified with appropriate statistics, rigor and reproducibility, and are thoughtfully interpreted.
Major findings and conclusions: Based on much supporting data, the authors propose that acute (6 hrs) PFOS treatment induces selective activation of the AIM2 inflammasome (in macrophages) to drive processing/ release of IL-1b and pyroptotic cell death. Because AIM2 is a DNA-sensing inflammasome initiator, additional experiments were performed to support a signaling pathway involving: 1) PFOS-induced upregulation of pro-apoptotic BAX/BAK and downregulation of antiapoptotic Bcl-2; 2) JNK-mediated activation of BAX/BAK oligomerization and consequent mitochondrial outer membrane permeabilization (MOMP); 3) release of both pro-apoptotic cytochrome C and mitochondrial DNA (mtDNA) into the cytosol; and 4) binding of mtDNA to AIM2 to drive AIM2/ ASC/ caspase-1inflammasome assembly. Consistent with these ex vivo results, AIM2 knockout mice treated with PFOS (in the acute 5 day peritoneal injection model) exhibit markedly reduced inflammatory damage to their lungs, liver, and kidney that correlates with markedly lower levels of serum/ peritoneal IL-1b but NOT TNFa or Il-6. Likewise, AIM2 knockout mice exposed to PFOS exhibited reduced lung damage and accumulation of inflammatory cytokines in the OVA-induced/PFOS-exacerbated asthma model. Thus, the major finding and conclusion is that activation of the AIM2 inflammasome is a major innate immune response to tissue accumulation of pathogenic amounts of the environmental pollutant PFOS. This is a mechanistically novel finding with regard to the immunological consequences of very high dose perfluoroalkyl substance exposure and tissue accumulation. However, the study lacks broader mechanistic novelty and pathophysiological significance for the specific reasons noted below.
Specific Concerns (Conceptual): 1. Multiple previous studies have demonstrated that activation of the intrinsic (mitochondrial) apoptotic cascade by various sterile stimuli (e.g. acute ionizing radiation-induced genomic DNA damage [PMID2786608; PMID31862870; PMID32760056] or chemotherapeutic drugs [PMID2840962;PMID24078693; PMID31372985] can trigger assembly of AIM2 or NLRP3 inflammasomes in both myeloid and non-myeloid cells. Importantly, a recent study (PMID30965677) by Bae et al demonstrated that circulating cell-free mtDNA in the serum of type 2 diabetic subjects induces AIM2 inflammasome activation in macrophages (after internalization) to drive chronic sterile inflammation. Thus, the finding that primary induction of intrinsic apoptosis by PFOS can induce AIM2 inflammasome assembly and sterile inflammation is an incremental advance.
2. It's clear that primary responses to high-dose PFOS exposure are: 1) activation of NFkB signaling (by an undefined mechanism) which induces not only proIL-1b expression as inflammasome-dependent inflammatory cytokine, but also TNFa and IL-6 as inflammasomeindependent cytokines; 2) transcriptional regulation of pro-and anti-apoptotic Bcl2-family members by undefined signaling pathways; and 3) the acute activation of JNK signaling (by an undefined mechanism) to induce oligomerization of the upregulated BAX and BAK. Induction of AIM2 inflammasome assembly (in macrophages) is a secondary (albeit significant) consequence of exposure to high dose PFOS. As noted, the mechanisms by which high-dose PFOS triggers these primary responses (upstream of AIM2 activation) are not investigated in this study. It can be reasonably argued that characterization of the NFkB activation mechanism is beyond the scope of the present study. However, delineation of the underlying mechanism (s) for the primary intrinsic apoptotic induction is central to understanding how two regulated cell death pathways (primary AIM2-independent apoptosis and secondary AIM2-dependent pyroptosis), which are triggered by PFOS, are temporally and spatially integrated at the cellular and systemic levels to coordinate systemic inflammation in mice (or humans) exposed to low-dose or high-dose PFOS. Minimally, experiments that explore the presumed post-transcriptional mechanism(s) by which PFOS triggers JNK activation and JNK-dependent BAX/BAK oligomerization are required.
3. As noted in the paper's introduction and discussion, the broad exposure of most human subjects to environmental PFOS is limited to consumer products and results in serum levels of ~100 ng/ml. Given PFOS's molecular mass of 500, this translates to 0.2 uM concentration which is a 1000-fold lower than the 100-300 uM concentrations used in the cell culture experiments of this study. The authors acknowledge that the latter concentrations would translate to serum concentrations of ~90,000 ng/ml which may only be observed in some human subjects (such as fluorochemical production workers) exposed to extraordinarily high levels of environmental PFOS. This raises important issues regarding the broader pathophysiological/ clinical significance of the observed AIM2 inflammasome pathway in understanding the mechanisms by which chronic inflammatory disease and tissue dysfunction is induced by the much lower PFOS environmental exposure to consumer products in general human population cohorts.
Specific Concerns (Experimental): 1. The cell death analyses in control and AIM2-ko macrophages are limited to 6 hr PFOS treatments. Longer term treatments with PFOS should be included (particularly in the AIM2-ko cells) to analyze: 1) the kinetics of progression to secondary necrosis/lytic cell death; 2) roles of secondary GSDMD-independent but GSDMDE-dependent pyroptosis. (GSDME can be cleaved and activated by the apoptotic executioner caspase-3.) 2. Related to above concern, the cell death and AIM2 experiments are limited to macrophages. However, AIM2 inflammasome signaling and cell death in non-myeloid cells, particularly epithelial cells (PMID27846608; PMID29973713; PMID31919014),can be a major driver of acute or chronic inflammatory responses to sterile stimuli. Experiments that analyze PFOS-induced AIM2 inflammasome signaling in an epithelial cell model should be included to establish whether PFOS drives similar AIM2 responses in non-myeloid cells.
3. Related to the above issue of non-myeloid AIM2 inflammasome signaling: The in vivo experiments described in Figures 6 and 7 shows that: 1) PFOS-induced tissue inflammation is suppressed in AIM2-deficient mice; and 2) PFOS-induced accumulation of IL-1b, but not TNFa or IL-6, is also suppressed in the AIM2-deficient mice. This raises very relevant questions as to whether production of IL-1b is, or is not, the major driver of inflammatory tissue damage in the PFOS-treated wildtype mice. Experiments using IL-1b-knockout or IL-1 receptor-knockout mice are required to address this important mechanistic issue. It is possible that AIM2-dependent pyroptosis and/or AIM2-dependent IL-18 production, independently of AIM2-dependent IL-1b accumulation, may be the major driver of inflammation in the PFOS-exposed mice.
Dear Referees, Thank you very much for providing us a valuable opportunity to revise our paper entitled "Innate immune activation through AIM2 inflammasome sensing of perfluoroalkyl substances". We really appreciate the positive and thoughtful comments regarding our manuscript, and we have performed additional experiments and now present new data addressing all of the critical points raised by the Referees.
A point-by-point response to the Referees' concerns is included below.
We hope that the revised manuscript meets the requirements for publication in "Nature Communications", and we look forward to hearing from you.

Response to the comments of Reviewer#1
In general, the study is very well structured and executed. The finding that Aim2-/-, to show that the observed effects are solely seen, when cells or mice are exposed to PFOS.

but not Nlrp3-/-, mice resist the PFOS-induced inflammation is in my view
Response: As the matter of fact, at the beginning of this project, we also detected the IL-1β production in PFOA (another comparable compound)-treated THP-1-derived macrophage. However, the results showed that PFOA could not promote the production of IL-1β in THP-1-derived macrophages (see New Figure 1a  However, these highly exposed workers didn't appear significant clinical symptoms, and such phenomenon indicates these workers may be tolerated to PFAS exposure. Taken together, although the epidemiological and clinical studies have confirmed the adverse effects of PFAS on human health, so far there is no exact "threshold" or "range" for PFAS exposure levels related to certain disease syndrome. Indeed, we have measured the serum PFOS level in the same mouse model (C57BL/6) in another study and the results showed that PFOS concentration in mice exposed to 7 mg/kg for 28 days (total 196 mg/kg) was about 8430 ng/ml 11 . Using this PFOS value, we can calculate that every 1 mg/kg PFOS exposure in mouse is equivalent to 43 ng/ml days. At 5 days post-treatment, mice liver, lung and kidney tissue were stained with hematoxylin-eosin (H&E) and assayed using a light microscope (a). Scale bar, 100 μm. The tissue (liver, lung, and kidney) injury score was determined in 5 randomly selected nonoverlapping fields from respective individual mouse tissue sections at x400 magnification. The scores were averaged in respective organs (liver, lung, and kidney) and individual animals. All histology analyses were conducted in a blinded manner. At 5 days post-treatment, the liver, lung and kidney of treated mice were isolated and cultured for 24 h, and supernatants were analyzed by ELISA for IL-1β (b), TNF-α (c) and IL-6 (d). In (a-d), data are mean values ± SD, ** P < 0.01; ***P < 0.001 (Student's t test). IL-1β and LDH release was measured in the supernatants. In a-c, data are mean values ± SEM, ns (non-significant), P > 0.05; * P < 0.05; *** P < 0.001 (Student's t test).  Response: We thank the reviewer for noticing this and changed the reference in the manuscript.

Comment 7: Lane 241-242: "Cytochrome c (cyt C) is a protein normally resided in the inner and outer mitochondrial membrane, participates in the mitochondrial electron-transport chain" cyt C resides in mitochondrial intermembrane space.
Response: We thank the reviewer for noticing this and changed the description of cytochrome c (cyt C) as "cyt C is a protein normally resided in the mitochondrial intermembrane space, participates in the mitochondrial electron-transport chain" in the manuscript. (Fig. 4d, e)." this is not supported by Fig. 4D. Please double check the labeling.

Response:
We have changed it accordingly.

Response to the comments of Reviewer#2
In this ms. Wang    inflammasome activation. Hence, given that PFOS-induced mtDNA release through two BAX-dependent pathways, it would be to understand why deletion BAX or BAK could decrease the mtDNA release.

Minor point: Comment 3: Mitotracker deep red (used in 4a) is not a potentiometric dye (unlike TMRE), the reduced signal after treatment suggests loss of mitochondrial mass upon PFAS treatment. This should be commented upon.
Response: We thank the reviewer for noticing this and confirmed the description accordingly. As a matter of fact, the reduced signal of Mitotracker deep red and Mitotracker green suggested that the decreased mitochondrial respiration and mitochondrial mass, respectively 17,28 .

Response to the comments of Reviewer#3
Major findings and conclusions: Based on much supporting data, the authors propose that acute (6 hrs Response: Thanks for this suggestion. We have discussed several studies reporting the roles of NLRP3 and AIM2 involved in sterile stimuli (mtDNA and genomic DNA)-mediated inflammation (Discussion part, 3 rd paragraph) [29][30][31] . We have also added the discussion about the findings reported by Bae et al 32 . It seems that there are not many studies reporting mtDNA sensed by AIM2 inflammasome; in addition, the detailed mechanism of mtDNA release was not elucidated in previous works.
Recently, only a few studies have investigated the association of inflammasome and environmental stimulants (such as silica, particulate matters, and nanoparticles), and they have mainly focused on NLRP3 inflammasome [33][34][35] . Until now, the host sensing mechanisms in response to organic toxins have not been reported yet. Our study for the first time demonstrated that AIM2 activation is required for PFAS-induced acute or chronic inflammatory response and tissue damage, in which the process is dependent on mitochondrial dysfunction and release of mtDNA. Based on the suggestions of the reviewer, we performed a series of new experiments and revealed that PFOS triggered the mtDNA release through BAX/BAK-mediated MOMP and BAX/CypD-mediated MPTP pathway, resulting in AIM2 inflammasome activation but not NLRP3 inflammasome. Nonetheless, our findings reveal that mtDNA-AIM2 axis is a critical innate mechanism for PFAS-induced inflammation (see New Figure   8, related to Figure 8  We agree with the reviewer's comments that there might be a kinetics dependency between PFOS-induced AIM2-independent apoptosis and AIM2-dependent pyroptosis. In this study, we detected the PFOS-induced apoptosis with Annexin V/7AAD staining by flow cytometry analysis and the PFOS-induced pyroptosis using immunoblot analysis, respectively. We found that PFOS could trigger evident apoptosis and pyroptosis in a dose-dependent manner (see Figure 1d in the manuscript and Extended Data Figure 11e in the Supplemental material).
Moreover, we found that in macrophages depleted mtDNA (ρ 0 ) with PFOS treatment still underwent apoptosis (see New Figure 9c). However, the PFOS-induced pyroptosis was nearly eliminated in ρ 0 macrophages (see New Figure 9d). In addition, we knocked down caspase-3 in the THP-1-derived macrophages, and found that the cleavage of GSDMD was not impaired in caspase3 KD THP-1-derived macrophages, suggesting that PFOS-induced caspase3-mediated apoptosis is irresponsible for AIM2-dependent pyroptosis (see New Figure 9e). Collectively, caspase-3-mediated apoptosis and AIM2 inflammasome-dependent pyroptosis are two separate cell death pathways.    Response: As the reviewer suggested, we treated human nasal epithelial progenitor cells (hNEPCs) with PFOS. The results showed that PFOS could significantly induced both the secretion of the proinflammatory cytokine IL-1β and cell death in hNEPCs (see New Figure 11a and b). We then knocked down AIM2 and found that PFOS-induced IL-1β production and cell death were apparently decreased in AIM2

New
KD hNEPCs (see New Figure 11 c), further confirming that PFOS could also induce AIM2 inflammasome activation in non-myeloid cells. Consistently, in the bone-marrow transfer experiment, we found that despite its major role in the bone marrow-derived cells, AIM2 might also function in other cell types such as epithelial cells, which contributes to the PFOS-induced tissue inflammation and damage.
lung tissue and kidney tissue of these mice were stained with hematoxylin-eosin (H&E) and assayed using a light microscope. Scale bar, 100 μm. The tissue (liver, lung, and kidney) injury score was determined in 5 randomly selected nonoverlapping fields from respective individual mouse tissue sections at x400 magnification. The scores were averaged in respective organs (liver, lung, and kidney) and individual animals. All histology analyses were conducted in a blinded manner. Data are mean values ± SED, *** P < 0.001 (Student's t test).