Synergistic Effect of WTC-Particulate Matter and Lysophosphatidic Acid Exposure and the Role of RAGE: In-Vitro and Translational Assessment

World Trade Center particulate matter (WTC-PM)-exposed firefighters with metabolic syndrome (MetSyn) have a higher risk of WTC lung injury (WTC-LI). Since macrophages are crucial innate pulmonary mediators, we investigated WTC-PM/lysophosphatidic acid (LPA) co-exposure in macrophages. LPA, a low-density lipoprotein metabolite, is a ligand of the advanced glycation end-products receptor (AGER or RAGE). LPA and RAGE are biomarkers of WTC-LI. Human and murine macrophages were exposed to WTC-PM, and/or LPA, and compared to controls. Supernatants were assessed for cytokines/chemokines; cell lysate immunoblots were assessed for signaling intermediates after 24 h. To explore the translatability of our in-vitro findings, we assessed serum cytokines/chemokines and metabolites of symptomatic, never-smoking WTC-exposed firefighters. Agglomerative hierarchical clustering identified phenotypes of WTC-PM-induced inflammation. WTC-PM induced GM-CSF, IL-8, IL-10, and MCP-1 in THP-1-derived macrophages and induced IL-1α, IL-10, TNF-α, and NF-κB in RAW264.7 murine macrophage-like cells. Co-exposure induced synergistic elaboration of IL-10 and MCP-1 in THP-1-derived macrophages. Similarly, co-exposure synergistically induced IL-10 in murine macrophages. Synergistic effects were seen in the context of a downregulation of NF-κB, p-Akt, -STAT3, and -STAT5b. RAGE expression after co-exposure increased in murine macrophages compared to controls. In our integrated analysis, the human cytokine/chemokine biomarker profile of WTC-LI was associated with discriminatory metabolites (fatty acids, sphingolipids, and amino acids). LPA synergistically elaborated WTC-PM’s inflammatory effects in vitro and was partly RAGE-mediated. Further research will focus on the intersection of MetSyn/PM exposure.


Introduction
Particulate matter (PM) exposure has been increasingly linked to systemic disease and lung pathology [1][2][3][4]. Furthermore, PM exposure significantly increases low-density lipoprotein (LDL), cholesterol, and triglycerides, and reduces high-density lipoprotein (HDL) levels [5,6]. Cross-sectional studies have found strong associations between PM exposure, loss of lung function, and metabolic syndrome (MetSyn), a cluster of risk factors for cardiovascular diseases such as hypertension, dyslipidemia, and insulin resistance [7]. We have previously identified that dyslipidemia and insulin resistance in World Trade Center (WTC)-PM-exposed Fire Department of New York (FDNY) first responders increased the risk of developing WTC Lung Injury (WTC-LI) [8,9]. Our murine and FDNY WTC-PM-exposed cohort studies suggest that the receptor for advanced glycation end-products (RAGE; also known as the advanced glycation end-product receptor (AGER) when referring to the human or murine protein) and lysophosphatidic acid (LPA) have key roles in the development of WTC-LI [9][10][11][12][13][14][15][16][17].
RAGE is a multi-ligand receptor of lipid and glucose metabolism intermediates, such as advanced glycation end-products (AGEs) and LPA, and contributes to inflammation, vascular injury, and pulmonary disease [18][19][20][21][22][23][24][25]. Stimulation of RAGE results in the activation of some proteins, such as NF-κB, STATs and Akt [25,26]. RAGE is expressed at low baseline levels in most other end-organs except the lung, where it is most highly expressed. It was also found to have increased expression in explanted lungs of COPD subjects [27]. Single-nucleotide polymorphisms within the RAGE gene (called AGER in these reports) were associated with FEV 1 /FVC ratio (p < 10 −15 ) in two genome-wide association studies of 74,564 and 20,820 individuals [20,28].
As a member of the immunoglobulin super family, RAGE exists both in a cell-membrane-bound and a soluble form. Airway inflammation in asthma and COPD is associated with reduced levels of circulating soluble RAGE or sRAGE [29,30]. Thus, RAGE is an attractive target for potential therapeutic agents in treatment for obstructive airway disease, especially since current treatments that minimize disease progression are limited. A finer understanding of the underlying inflammation and a focus on new therapeutic targets such as the LPA/RAGE axis is crucial.
WTC-PM exposure, dyslipidemia, and elevated LPA have been associated with the development of WTC-LI [15,16,31]. LPA is a phospholipid and is soluble in both cell membranes and in aqueous fluid that can activate pathways involved in vascular injury [32][33][34][35]. The production of LPA results from various pathways, but most involve the catabolism and oxidation of low-density lipoprotein (LDL) [36,37]. Autotaxin (Atx)/lysoPLD is a secreted enzyme that is responsible for the vast majority of LPA synthesis [38,39]. Tissue LPA concentrations are tightly regulated via modulation of Atx. [40,41]. Circulating LPA is rapidly turned over by lipid phosphate phosphatases (LPPs), which terminate its signal by dephosphorylation [42]. Pulmonary vascular injury occurs early in COPD with pulmonary perfusion abnormalities and reduced blood return to the heart observed prior to development of abnormal FEV 1 [43,44]. Pulmonary arteriopathy was present in 58% of lung biopsies from non-FDNY WTC-PM-exposed individuals and in 74% with constrictive bronchiolitis after inhalational exposures during military service [45,46]. G-protein-coupled receptors (GPCRs) have been identified as specific for LPA (LPA 1-5 ) [47]. Although most reported cellular responses to LPA have been attributed to cell surface GPCR activation, not all LPA activities can be explained by GPCR signaling [48][49][50]. LPA can also bind the nuclear peroxisome proliferator-activated receptor gamma (PPARγ) and initiate early stages of atherosclerosis [51]. LPA is an agonist of PPARγ [51]. Exogenous LPA might also enter cells to activate PPARγ. Specifically, when LPA enters RAW264.7 cells, it activates a reporter driven by a PPARγ expression vector [50].
Alveolar macrophages comprise approximately 90% of alveolar immune cells and can release inflammatory cytokines/chemokines when exposed to PM. Alveolar macrophages express RAGE, making them potentially relevant in the MetSyn/lung injury pathways [52,53]. In our earlier work, we utilized an in-vitro model of macrophages and showed that WTC-PM exposure produced comparable inflammatory profiles to those of firefighters with WTC-LI [9,54].
We now focus on the intersection of WTC-PM/lipid co-exposure [55,56]. Using a multiomic approach and murine and human in-vitro exposures, we identified key cytokines/chemokines and transcription factor profiles of WTC-PM exposure. In addition, we determined which biomarkers were synergistically induced by PM/LPA co-exposure in both human and murine macrophages. To further our translational understanding, we then integrated identified biomarkers from our in-vitro analyses with our metabolomics analysis of the PM-exposed firefighters. The incorporation of in-vitro models allows us to further understand the translatability of PM exposure responses and of pathways that may be key to loss of lung function in vivo due to PM exposure.

WTC-PM and LPA Preparation
WTC-PM 53 (≤53 µm) was collected in bulk, aerosolized, sieved using a 53-µm diameter mesh, analyzed as previously published, and stored at room temperature as previously described [15,58]. LPA (lysophosphatidic acid 18:1; CAS 65528-98-5) was purchased from Santa Cruz ® (Dallas, TX, USA) (sc-222720) and supplied as a solution in ethanol. After ethanol evaporation, LPA was reconstituted in media to a final concentration of 500 µM per well. The 500 µM LPA dosing was chosen based on preliminary dose-finding experiments that showed a synergistic elaboration that was not present at lower doses (data not shown). This dose also has clinical plausibility, as we have previously evaluated firefighters with WTC exposure and found that their LPA levels exceed 200 µM, higher than previously recorded levels of healthy patients [31,59]. This dose has also been found to inhibit humoral immunity signaling [60].

Exposures
THP-1-derived macrophages and RAW264.7 cells were exposed to WTC-PM 53 suspensions (50 µg/mL or 100 µg/mL) and/or 500 µM of LPA. All subsequent experiments utilized only the 100 µg/mL dose of WTC-PM since prior work showed it could induce a profound inflammatory response and closely emulate real-world delivery of WTC-PM [9,15,61]. Macrophages were plated at 1 × 10 6 cells/mL in 12-well plates, 1.5mL total volume. Adherent macrophages were exposed to PBS, 100 µg/mL suspensions of WTC-PM 53 , 500 µM of LPA, (Santa Cruz ® ) or both WTC-PM 53 and LPA. After 24 h, supernatants collected and stored at −80 • C [62,63]. All experiments done in triplicate and controlled for volume.

Integration
A subset of maximally discriminative metabolites previously identified via a machine learning algorithm, random forests, were integrated with the human analogues of significantly altered cytokines/chemokines identified from this manuscript's in vitro studies [15,31,67,73,74]. Agglomerative hierarchical clustering was performed on standardized data with Spearman's rank correlation matrices and average linkage (Matlab, Mathworks).

Statistical Analysis
SPSS-23 (IBM, Armonk, NY, USA) and Prism 5.0 (GraphPad, San Diego, CA, USA) were used for data storage, handling and analysis. Cytokine/chemokine concentrations were expressed as medians (IQR) and compared by the Mann-Whitney U-test. For categorical data, counts and proportions were used, while the Pearson-χ 2 test was applied for comparison. The interaction (additive and synergistic) between WTC-PM 53 and LPA was assessed by two-way ANOVA, corrected for multiple comparisons by Sidak's test [57,76,77]. Results were deemed statistically significant at p < 0.05.

Ethics Approval and Consent to Participate
We have reviewed the journal's guidelines involving ethical publication and we confirm here that we have abided by said guidelines. All subjects, at the time of enrolment, consented to analysis of their information and samples for research according to Institutional Review Board approved protocols at Montefiore Medical Center (#07-09-320) and New York University (#16-01412).

Synergistic Response to WTC-PM/LPA Co-Exposure in Human-THP-1-Derived Macrophages
A dose-dependent elaboration response was observed in the WTC-PM/LPA co-exposure cohort for GM-CSF, IL-8, and IL-10 ( Figure 1A-C). When comparing independent WTC-PM to co-exposure of WTC-PM/LPA, there was a significant increase in GM-CSF, IL-8, VEGF, MCP-1 and MIP-1α. However, there were also inhibitory effects seen in Eotaxin, IP-10, and MIP-1β (Table 1).

WTC-PM and LPA Induce RAGE/PPARγ Protein Production in Human-THP-1-Derived Macrophages
Levels of RAGE and PPARγ were assessed to determine the direct cellular effects. Compared to MA, WTC-PM independently induced RAGE and PPARγ protein production in differentiated THP-1-derived macrophages ( Figure 1E, Lanes 1,2). LPA independently induced RAGE ( Figure 1E, Lanes 1,3). WTC-PM/LPA co-exposure reduced RAGE and PPAR-γ production relative to independent exposures ( Figure 1E, Lanes 1-4). WTC-PM and LPA independently induced significantly higher RAGE protein production compared to MA, whereas co-exposure decreased RAGE by densitometry ( Figure 1F). Moreover, WTC-PM and LPA each had significantly higher RAGE production compared to co-exposure. PPAR-γ did not change after WTC-PM exposure, whereas it was decreased after LPA and co-exposure ( Figure 1F). Similarly, co-exposure significantly reduced PPAR-γ production compared to WTC-PM alone.

Synergistic Response to WTC-PM/LPA Co-Exposure in RAW264.7 Cells
In contrast to differentiated THP-1-derived macrophages, murine macrophages co-exposed to WTC-PM/LPA induced a synergistic expression of only IL-10 (75.11(<0.0001)) ( Figure 2B). WTC-PM caused a significant increase, while co-exposure caused a decreased production of NF-κB fold change ( Figure 2D). When comparing WTC-PM exposure to co-exposure of WTC-PM/LPA, there was a significant increase in IL-1α, -10, and MCP-1 ( Table 2). In contrast, there was a significant reduction in the expression of IL-6, IL-17, IP-10, and RANTES (Table 2).

Synergistic Response to WTC-PM/LPA Co-Exposure in RAW264.7 Cells
In contrast to differentiated THP-1-derived macrophages, murine macrophages co-exposed to WTC-PM/LPA induced a synergistic expression of only IL-10 (75.11(<0.0001)) ( Figure 2B). WTC-PM caused a significant increase, while co-exposure caused a decreased production of NF-κB fold change ( Figure 2D). When comparing WTC-PM exposure to co-exposure of WTC-PM/LPA, there was a significant increase in IL-1α, -10, and MCP-1 ( Table 2). In contrast, there was a significant reduction in the expression of IL-6, IL-17, IP-10, and RANTES (Table 2).

WTC-PM and LPA Induce RAGE Protein Production in Murine RAW264.7 Cells
Immunoblot assays show a reduction in p-Akt, p-STAT3, and STAT5b expression after coexposure ( Figure 2E, Lane 4). WTC-PM alone did not modify production in any of the measured proteins when compared to MA ( Figure 2G). LPA and co-exposure independently decreased p-STAT3 and STAT5b production when compared to MA by densitometry. Moreover, WTC-PM and LPA each had significantly higher p-STAT3 and STAT5b production compared to co-exposure. Additionally, co-exposure decreased p-Akt when compared to MA ( Figure 2G). Similarly, co- . * denotes p < 0.05 between LPA and no LPA, Student's t-test; ** p < 0.05 for interaction of WTC-PM/LPA co-exposure, multiple t-tests by row; † p < 0.05 compared to PBS, Student's t-test; ‡ p < 0.05 compared to media alone, Student's t-test; § p < 0.05 compared to WTC-PM/LPA co-exposure, Student's t-test.

WTC-PM and LPA Induce RAGE Protein Production in Murine RAW264.7 Cells
Immunoblot assays show a reduction in p-Akt, p-STAT3, and STAT5b expression after co-exposure ( Figure 2E, Lane 4). WTC-PM alone did not modify production in any of the measured proteins when compared to MA ( Figure 2G). LPA and co-exposure independently decreased p-STAT3 and STAT5b production when compared to MA by densitometry. Moreover, WTC-PM and LPA each had significantly higher p-STAT3 and STAT5b production compared to co-exposure. Additionally, co-exposure decreased p-Akt when compared to MA ( Figure 2G). Similarly, co-exposure significantly reduced p-Akt production compared to WTC-PM alone. LPA induced greater RAGE expression compared to WTC-PM ( Figure 2F). Moreover, unlike that found in our human macrophages, co-exposure induced RAGE production greater than that of either WTC-PM or LPA alone ( Figure 2F,G). This is reflected in densitometry showing LPA and co-exposure independently inducing significantly higher RAGE production when compared to MA ( Figure 2G). Additionally, co-exposure significantly increased RAGE production compared to WTC-PM alone.

MultiOMIC (Metabolome and Chemome) Integrated Biomarker Analysis
Serum analyte data collected from FDNY WTC-exposed first responders for GM-CSF, IL-10, -8, -1α, MIP-1α, TNFα, LPA, and RAGE were of interest due to our in-vitro model findings in our prior work (Table 3) [9,15,31,67,68,73,78]. Serum MCP-1 exhibited a significant median difference (Table 3). These were integrated with the metabolite data assayed in the same population (n = 15/group). Linkage thresholds determined by inspection of the dendrograms were used to highlight clusters of metabolites that may reflect mechanistic relations. For the data matrix, a linkage threshold of one was used to identify three distinct clusters of metabolites (A-C) ( Figure 3A). Cluster A consisted largely of amino acids and their metabolites, which were generally decreased in cases of WTC-LI compared to controls. These metabolites included acyl/acetylated, BCAAs, and those of the urea cycle. Meanwhile, cluster B consisted mostly of lipids and their metabolites, including the glycerol-3-phosphocholines, arachidonate, and LPA. Interestingly, in WTC-LI cases, elevated sRAGE and other metabolites were found in cluster B. Cluster C was composed of the cytokines/chemokines involved in the innate acute inflammatory response and tissue repair, and did exhibit significant differences in cases of WTC-LI compared to controls.

Discussion
The global burden of co-existing PM exposure and metabolic dysregulation is significant. Systemic inflammation caused by this co-exposure results in significant end organ compromise [82,83]. Individuals with MetSyn are predisposed to PM-exposure-associated lung injury [5,8,9,84]. These modifiable risks have proven challenging to mitigate. The interaction of MetSyn and PM exposure is a topic of considerable importance.
WTC-PM-induced GM-CSF, IL-1α, TNF-α, MCP-1, and MIP-2 further supports our prior studies suggesting that WTC-PM-exposed alveolar macrophages exhibit an inflammatory response [52,54]. This is in line with other literature showing that there were higher MCP-1 levels in COPD patients [85,86]. Both THP-1-derived macrophages and RAW264.7 cells exhibited higher LPA-associated (LPA exposure and WTC-PM co-exposure) analyte levels compared to WTC-PM and control exposures. From this study, we have demonstrated that the PM-induced inflammatory response is reproducible in macrophage lines.
RAGE expression is higher after LPA exposure relative to WTC-PM exposure. Co-exposure decreased RAGE in THP-1-derived macrophages but it was increased in RAW264.7 macrophages. RAGE activation and expression have been demonstrated to complete a positive feedback loop upon RAGE binding, with one of its many known ligands being LPA [87,88]. To date, WTC-PM has not exhibited properties that suggest direct RAGE binding. This distinction is further supported by our observed baseline RAGE levels after WTC-PM exposure in our cell lines. While our human in-vitro model showed an increase in RAGE expression only after LPA exposure, our murine in-vitro model showed an increase in expression after PM, LPA, and joint exposure. Our previous murine models have shown that both PBS-and PM-exposed mice express RAGE, and this remains elevated even one month after exposure [15,17]. WTC-PM contains silica and heavy metals that could induce double-stranded DNA breaks and the production of reactive oxygen species, such as superoxide and peroxide [89][90][91][92][93][94][95][96][97][98][99][100]. It causes cellular stress on lung epithelium and alveolar macrophages, characterized by innate inflammation via pathways including but not limited to those not mediated by RAGE.
We know from the literature that macrophages may exhibit an activated phenotype upon LPA exposure, and can contribute to the formation of atherosclerotic plaques and coronary artery disease [101][102][103]. LPA-activated macrophages have exhibited increased TNF-α, MCP-1, IL-8, and IL-10, among other cytokines/chemokines, and therefore appear to exhibit a mixed inflammatory/anti-inflammatory response [104]. Our results and previous findings suggest that, upon WTC-PM/LPA co-exposure, the synergistic elaboration of cytokine/chemokines with heterogeneous effects may be a consequence of both a constitutive innate response to cellular damage and subsequent macrophage activation. In addition, we know from the literature that COPD patients and those with worse airflow obstruction had higher levels of IL-10 and MCP-1 [85,86,105].
Though cytokine/chemokines characteristic of such an inflammatory response are synergistically upregulated after co-exposure, NF-κB activity does not subsequently correlate. Downregulated NF-κB, p-STAT3, and STAT5b, coupled with elaboration (in some cases synergistic) of inflammatory cytokines/chemokines upon WTC-PM/LPA co-exposure suggests that co-exposure may attenuate the regulatory capacity of these signaling/transcriptional components [24,25]. Future studies that evaluate inflammatory/vascular disease in the WTC-LI cohort would further clarify these relationships.
Establishing clusters of related biomarkers is important. Since these biomarkers may be expressed prior to phenotypic disease expression in humans, each cluster may reflect an independent pathway to disease that could be evaluated in subsequent research. High-dimensional data analysis of metabolites, serum, and clinical biomarkers support our findings, as we have identified pathways associated with the loss of lung function after WTC-PM exposure. Previous work has noted that later-staged atherosclerosis in mammalian models exhibited similar modified serum-derived amino acid metabolites. This modification of amino acid metabolites, specifically, has been associated with hypomethylation epigenetic modifications [106][107][108][109]. Elevated lipid metabolism has been linked to several pulmonary disorders and may be linked to metabolic/inflammatory lung disease and accumulation of activated macrophages. Similar to the COPD literature, we found elevated glycero-phosphatidylcholines (GPC), which are converted to LPA via cellular-damage-induced phospholipase A2 (PLA2) (Figure 4) [110][111][112][113][114][115][116][117]. We suggest that this cytokine/chemokine profile, dependent on macrophages and driven by dyslipidemia, yields subsequent WTC-LI. Based on previous literature, the amino acid cluster (5) may indicate epigenetic modification, resulting in atherosclerotic presentation, which may inhibit lung function; a correlation that could be clarified in future studies [118].
accumulation of activated macrophages. Similar to the COPD literature, we found elevated glycerophosphatidylcholines (GPC), which are converted to LPA via cellular-damage-induced phospholipase A2 (PLA2) (Figure 4) [110][111][112][113][114][115][116][117]. We suggest that this cytokine/chemokine profile, dependent on macrophages and driven by dyslipidemia, yields subsequent WTC-LI. Based on previous literature, the amino acid cluster (5) may indicate epigenetic modification, resulting in atherosclerotic presentation, which may inhibit lung function; a correlation that could be clarified in future studies [118]. Our work has several limitations. The complexity and unique nature of both the WTC-PM and the corresponding immunological response likely incorporate multiple pathophysiological pathways, as suggested by the pathway structure ( Figure 4). As per prior studies, >150 compounds have been identified in WTC-PM [119]. However, in almost all exposure-related diseases, like the sub-cohort WTC-LI, this is not the case. By utilizing WTC-PM, we limit our capacity to isolate one predominate inducer of downstream WTC-LI, but simultaneously acknowledge the complex interplay of an indefinite amount of immunologically active stimuli within the WTC-PM. Alternatively, our study focused on the co-exposure model, which encompasses elevated levels of one phospholipid derivative (LPA) observed in MetSyn as the predominate inflammatory metabolite, Our work has several limitations. The complexity and unique nature of both the WTC-PM and the corresponding immunological response likely incorporate multiple pathophysiological pathways, as suggested by the pathway structure ( Figure 4). As per prior studies, >150 compounds have been identified in WTC-PM [119]. However, in almost all exposure-related diseases, like the sub-cohort WTC-LI, this is not the case. By utilizing WTC-PM, we limit our capacity to isolate one predominate inducer of downstream WTC-LI, but simultaneously acknowledge the complex interplay of an indefinite amount of immunologically active stimuli within the WTC-PM. Alternatively, our study focused on the co-exposure model, which encompasses elevated levels of one phospholipid derivative (LPA) observed in MetSyn as the predominate inflammatory metabolite, which works synergistically with components of WTC-PM to induce WTC-LI. In contrast to our prior studies performed in ex-vivo macrophages, MCP-1 was significantly elaborated while GM-CSF was not in our limited human sample size, Table 3 [9,54].
These multi-analyte and metabolite datasets may yield directionality in our future research associated with characterizing WTC-LI. Future studies, with respect to our approach, should focus on including more endogenous and exogenous contributors to WTC-LI and how it relates to the predisposition to this decrease in FEV 1 in individuals suffering from MetSyn as well. Globally, air pollution contributes to pulmonary and vascular disease and complications which account for seven million annual deaths [120]. Furthermore, the prevalence of MetSyn, a vascular risk, is rapidly increasing globally, as is concurrent global high PM. The impact and resultant physiologic changes that occur in the intersection of PM exposure and MetSyn is a topic of considerable global health importance. Studies of comorbid asthma, air pollution, and MetSyn have noted that obesity-associated asthma often resists conventional therapy, further highlighting the need for novel mechanistic work [121][122][123][124]. Identifying predictive biomarkers may facilitate early treatment after PM exposure, which may be crucial to the preservation of lung function and identification of biologically plausible pathways and therapeutic targets. This work could be generalizable to biomass/PM-exposed populations since RAGE is a determinate of lung function.

Conclusions
In conclusion, we demonstrated that WTC-PM induced inflammation in a dose-dependent manner. LPA and WTC-PM elevate RAGE protein production. Combined WTC-PM exposure and LPA synergistically enhance the observed inflammatory response. Our in-vitro findings and the WTC-LI cohort metabolomics profile suggest that the observed synergistic inflammatory response is partially a result of a dyslipidemia-driven inflammation. This work was partially funded by the NYU-HHC CTSI, supported by grant UL1TR000038 via the Saperstein Scholars Fund. The funding agencies did not participate in the study design; in collection, analysis or interpretation of data; in the writing of the report; or in the decision to submit the article for publication.
Acknowledgments: This work is dedicated to the FDNY rescue workers, and we would like to thank them for their bravery, sacrifice, and continued commitment to our work.

Conflicts of Interest:
The authors declare that they have no competing interests.