Nose-only inhalation exposures to alumina nanoparticles / hydrogen chloride gas mixtures induce strong pulmonary pro-inflammatory response


 The authors have withdrawn this preprint due to author disagreement.

Nose-only inhalation exposures to alumina nanoparticles / hydrogen chloride gas mixtures induce strong pulmonary pro-in ammatory response Results: Iterative co-exposures (IE) increased total proteins and lactate dehydrogenases (LDH) concentrations in BALF indicating alveolar-capillary barrier permeabilization and cytolysis. Early pulmonary in ammation was induced after IE to Al 2 O 3 NPs ± HCl g resulting in polymorphonuclear neutrophils (PMN) and pro-in ammatory cytokines increases (TNF-α, IL-1β, GRO/KC) in BALF. Moreover, after exposure to Al 2 O 3 NPs ± HCl g aerosols, both exposure scenarios induced early pulmonary histopathological lesions, among which vascular congestions, bronchial pre-exfoliations, vascular and interalveolar septum edemas. Lung oxidative damages (8-hydroxy-2'-deoxyguanosine ; 8-OHdG) were observed in situ following UE in each experimental condition, suggesting early oxidative stress induction by aerosols inhalation. However, no 8-isoprostane concentration increase was simultaneously found in animals BALF.
Conclusions: Biological effects of the studied aerosols are dependent on both aerosol content and exposure scenario. Results showed an important early pro-in ammatory effect of Al 2 O 3 NPs/HCl g mixtures on rats lungs following iterative inhalations (IE). Taken together these data raise concerns towards potential long term pulmonary toxicity of combustion mixtures aerosols, and highlight the importance for workers to wear individual protections.

Background
Combustion reactions from different sources can produce high volumes of complex aerosols containing various components as gases and particles in the environment [1,2]. Solid composite propellants use in aerospace and defense elds lead to emission of alumina (Al 2 O 3 ) particles and hydrogen chloride gas (HCl g ) [3,4]. Moreover, these pollutants can also be emitted respectively by industrial activities such as aluminum manufacturing [5], and wastes incineration [6] increasing the risk of workers exposure.
Alumina (Al 2 O 3 ) is the oxidized form of aluminum and exists under several crystalline phases (γ, δ, θ and α) [7]. Among them, a majority of γ then δ polymorphs were found following solid composite propellants use [4,8]. Alumina is mainly used for aluminum production and enters in the composition of many everyday products (paper, plastics, ceramics, ame retardants and coatings) due to its hardness.
Otherwise, alumina can also be found in foods and cosmetics [9]. To determine health hazard related to alumina particles exposure, only few studies were conducted in vivo by inhalation. 28 days nose-only inhalation of Al 2 O 3 NPs (11.94 nm, 0.2 to 5 mg/m 3 , 5 days a week) showed pulmonary pro-in ammatory properties of Al 2 O 3 NPs on rats. This was characterized by total cells, polymorphonuclear neutrophils (PMN), lymphocytes, lactate dehydrogenase (LDH), TNF-α and IL-6 increases in bronchoalveolar lavage uids (BALF). Alveolar macrophages accumulation was also observed in lungs after 28 days exposure [10]. In mice, pulmonary pro-in ammatory effects were also recorded after 7 days exposure to 40  instillation on rats (6.3 nm, 0.5 to 300 cm²/mL) also induced acute pulmonary in ammation resulting in PMN concentrations increases in BALF [12]. Moreover, mice IT instillation exposures to Al 2 O 3 particles (4.37 μm, 40 mg) increased macrophages, PMN and bronectin concentrations in BALF until 12 months after exposure suggesting pulmonary in ammation persistence over time [13]. Another study, carried out by nasal instillation on rats highlighted dose dependant in ammation and alveolar-capillary barrier permeabilization after Al 2 O 3 particles (size unknown, 1 to 40 mg/kg) exposure [14]. Besides proin ammatory properties, several in vitro Al 2 O 3 particles toxicity studies results suggest oxidative stress induction and genotoxic potential (DNA single and double-strand breaks induction) of these particles [15,16]. These effects were also found in vivo after oral, intravenous or peritoneal injection exposures to Al 2 O 3 particles [16,17,18,19].
Hydrogen chloride gas is a known irritant and caustic agent able to cause severe ocular and cutaneous lesions [20]. Once in water, HCl g dissolves to form hydrochloric acid solution (HCl). HCl g and HCl exposure risk is non negligible as these compounds are widely used in industries and laboratories and as they can be found in some household products [21]. After HCl g inhalation by rats and mice, eyes, mucous membranes and skin irritations were observed associated with respiratory system lesions (emphysema, atelectasis and edema) [22]. Respiratory system necrosis as well as PMN increases were also observed on rats after 30 min inhalation (1284 and 1293 ppm of HCl g ) [23]. Nevertheless, HCl g long term inhalation (10 ppm ; 6h a day 5 days a week) performed during rats entire life did not induce neoplastic or preneoplastic lesions [24]. Similar respiratory symptoms were found in Guinea pigs and rabbits associated with severe and persistent in ammatory reactions [25]. Mitogenic effect of HCl (40 mM) was also showed on rabbit esophageal mucosa after 1h incubation [26]. Moreover, respiratory rate increase and oxygen partial pressure in blood decrease were observed after HCl g head-only inhalation in baboons (500, 5 000 and 10 000 ppm ; 15 min). However, no alteration of pulmonary function was reported 3 months post-exposure [27]. Pro-in ammatory effect of HCl was also demonstrated in vitro on esophageal mucosa and mice macrophages (cell line RAW 264.7), resulting in substance P, IL-8, IL-6, IL-1β, plateletactivating factor (PAF) and nitric acid levels increases [28,29,30,31].
Taken together, these toxicity data regarding Al 2 O 3 particles and HCl g raise questions towards their potential combined toxicity due to possible workers exposure. Mixtures effects study is an actual major concern in toxicology. Only few scienti c data are available in literature, especially for mixtures of compounds with different physical forms (solid and gas for example). However, population is daily exposed to multiple pollutants simultaneously and a recent study highlighted mixtures effects (synergy, addition, antagonism or inhibition) unpredictability [32]. Therefore, it is currently necessary to study mixtures effects in a case-by-case basis. Biological effects and resulting hazard of complex combustion aerosols exposure are poorly described in literature. To date only one study was published concerning combined effect of Al 2 O 3 particles and HCl g . This study was carried out in vivo on rats and mice by whole body exposure [33]. However, during exposures, HCl g was always associated with hydrogen uoride gas (HF g ) and the only studied endpoint was mortality. Conclusions did not highlight any additive or synergistic effect of Al 2 O 3 particles addition to the gases mixture on animals mortality.
As the main route of exposure to these mixtures is inhalation, this study objective was to assess pulmonary toxicity after inhalation of Al 2 O 3 NPs / HCl g mixture in vivo. Wistar rats were exposed by noseonly inhalation to aerosols containing Al 2 O 3 NPs (size and crystalline phase: 13 nm ; γ/δ) and/or HCl g .
Blood, BALF and lungs were collected 24h after exposures. Cells counts, pro-in ammatory cytokines, lactate dehydrogenases, total proteins and 8-isoprostane concentrations were measured in bronchoalveolar lavage uids (BALF). Lungs histopathological lesions were analyzed and presence of oxidative nucleic acids damages (8-OHdG) was studied in situ in lung tissue.

Aerosols physico-chemical characterization
Transmission Electron Microscopy (TEM) observations showed micronic spherical particles agglomerates on Al 2 O 3 bulk powder ( g. 1 A). To characterize alumina crystalline structure, X-Ray diffraction (XRD) experiments were performed on bulk powder. Results con rmed majority of γ and δ polymorphs as indicated by the supplier (Fig. 1 B).
Once nebulized, Al 2 O 3 NPs formed spherical agglomerates ( Fig. 1 C). Presence of HCl g did not modify agglomerates shape in aerosols (data not shown). Energy Dispersive X-ray (EDX) analysis con rmed aluminum and oxygen atoms presence without chemical impurities. The copper spike on EDX spectra correspond to TEM grids used for aerosol sampling (Fig. 1 D). Results obtained with Sioutas (SKC Inc.) cascade impactor revealed that a majority of particles agglomerates in generated aerosols had a size below 250 nm, and presence of HCl g 5 ppm did not signi cantly modify particles concentrations and size distribution ( Fig. 1 E).
To deepen and complete these analytical data, aerosol particles content was assessed using mass

Lungs aluminum content increasing after iterative exposures
Pulmonary aluminum load was measured by ICP-OES after aerosols inhalation. Following unique exposures (UE) to Al 2 O 3 NPs alone or associated with HCl g 5 ppm, concentrations around 100 μg of aluminum per g of lung were measured (Fig. 2). However, these mean quantities were not signi cantly different from air exposed animals. Iterative exposures (IE) to Al 2 O 3 NPs alone or in association with HCl g 5 ppm induced signi cant aluminum quantities increases in lungs compared to air exposed rats or to those exposed to the same conditions in UE (Fig. 2). IE to Al 2 O 3 NPs with HCl g 5 ppm signi cantly increased pulmonary aluminum load compared to iterative exposures to each component alone (Fig. 2).

Clinical symptoms and body weight
Animals' clinical follow-up did not highlight any treatment-related clinical symptom following exposures. Signi cant weight losses were observed after each 4-hour exposure compared to weight before exposure ( Table 2). Rats did not always recover their initial weight between two exposures, especially after IE, even in air exposed group. However, these weight decreases were always less than 10% of rats' initial weight. Table 2 -Rats body weight monitoring during the experimental period. Rats were exposed uniquely or iteratively to air, HCl g 5 ppm, Al 2 O 3 13 nm or to the mixture Al 2 O 3 13 nm + HCl g 5 ppm. Rats body weight (in gram, mean ± SD) was measured before and after each aerosol exposure. Obtained results during unique exposures and during iterative exposures are respectively presented in table A and table B. Tables show initial body weight before exposures (BE, t0, t24h (B), t48h (B), t72h (B)), after exposures (AE, t4h, t28h (B), t52h (B), t76h (B)) and before sacri ce (BS, t24h (A) and t96h (B)).

Alveolar-capillary barrier and pulmonary in ammation
No change in total proteins or lactate dehydrogenases (LDH) concentrations was observed following UE compared to air exposed rats ( Fig. 3 A and B). Only IE to Al 2 O 3 NPs alone or associated to HCl g 5 ppm induced signi cant concentrations increases of total proteins and LDH in rats bronchoalveolar lavage uids (BALF) compared to air exposed control animals ( Fig. 3 A and B). Total proteins concentrations of 0.18 ± 0.03 g/L and 0.22 ± 0.04 g/L and LDH concentrations of 99 ± 43 U/L and 112 ± 42 U/L were measured in rats BALF following respectively IE to Al 2 O 3 NPs and to Al 2 O 3 NPs + HCl g 5 ppm.
Total cells number increased in rats BALF after IE to Al 2 O 3 NPs. Moreover, signi cantly higher total cells numbers were counted following IE than UE to Al 2 O 3 NPs and to Al 2 O 3 NPs + HCl g aerosols ( Fig. 3 C). IE to same aerosols also induced signi cant polymorphonuclear neutrophils (PMN) number increases in BALF compared to air exposed control animals ( Fig. 3 D). In contrast, no signi cant change in macrophages and lymphocytes numbers was highlighted by the carried out counts (Fig. 3 E and F).
IE to Al 2 O 3 NPs or Al 2 O 3 NPs + HCl g aerosols leaded to signi cant TNF-α and IL-1β concentrations increases in rats BALF (Fig. 4 A and B). TNF-α concentrations of 42.86 ± 13.07 pg/mL and 45.73 ± 11.77 pg/mL, and IL-1β concentrations of 456.2 ± 163.2 pg/mL and 341.6 ± 58.11 pg/mL were respectively measured. Signi cant increased concentrations of GRO/KC and MIP-2 were also measured in rats BALF after respectively IE to Al 2 O 3 NPs (Fig.4 C) and UE to Al 2 O 3 NPs (Fig. 4 D) compared to air exposed control rats .

Histopathology observations
Lung parenchyma observations revealed the presence of pathological lesions following experimental aerosols inhalation (Fig. 5). However, no brosis, resorption hyperplasia or pleura in ammation was reported in this study.
In control rats lungs, vascular congestions were occasionally found and considered as physiological lesions (Fig.5 A and B). HCl g 5 ppm exposures mainly induced vascular lesions. UE increased vascular congestions (Fig. 5 C), while IE induced vascular edema and, in some cases, velamentous epithelium was observed (Fig. 5 D). Al 2 O 3 NPs inhalation mainly induced interstitial and in ammatory lesions (Fig. 5 E and F). Indeed, marked interalveolar septum edemas were found within rats pulmonary epithelium after UE and IE. Bronchial pre-exfoliation phenomena (Fig. 5 E) and vascular edemas (Fig. 5 F) were also observed. In some cases, localized in ammatory areas, composed of polymorphonuclear cells and lymphocytes, were also found in rats lungs (Fig. 5 F). Co-exposures to Al 2 O 3 NPs and HCl g 5 ppm induced vascular and interstitial lesions at a time ( Fig. 5 G and H). Bronchial pre-exfoliation phenomena (Fig. 5 G), vascular and interalveolar septum edemas (Fig. 5 G and H) were found in rats lungs after both UE and IE.
Lesions scoring (Novotec society) was performed in order to compare early pulmonary lesions induced by the different experimental aerosols and exposure scenarios (Fig. 6). Obtained results showed higher mean lesion scores in animals exposed to aerosols containing Al 2 O 3 NPs (alone or associated with HCl g ).
Results did not suggest major in uence of exposure scenario towards histopathological lesions induction as close scores were calculated for UE and IE to the same aerosol. Moreover, adding HCl g to Al 2 O 3 NPs aerosol did not seem to worsen lesions comparing to those resulting from Al 2 O 3 NPs inhalation alone.

Oxidative stress analysis
To assess oxidative stress induction following experimental aerosols exposures, 8-OHdG (8-hydroxy-2'deoxyguanosine, oxidative nucleic acids damages marker) was labeled in situ in rats lung tissue using immuno uorescence. Microscopic observations revealed 8-OHdG uorescent signal after UE ( Fig. 7 A-D), whereas no speci c signal was observed after IE to the different aerosols. After UE to the experimental aerosols, cytoplasmic and nuclear labeling was found both in alveolar ( Fig. 7 B and D) and bronchial epithelia (Fig. 7 C). These labeling localizations referred respectively to RNA (cytoplasmic) and DNA (nuclear) oxidations.
To complete our investigations on oxidative stress induced by inhalations, 8-isoprostane (lipid peroxidation marker) concentrations were measured in rats BALF 24h after the last exposure. Results did not reveal any signi cant increase in 8-isoprostane concentrations following inhalations (Fig 7 E). Signi cant 8-isoprostane concentrations decreases were measured after IE to Al 2 O 3 NPs and to Al 2 O 3 NPs + HCl g 5 ppm compared to control, with respective mean values of 26.6 ± 1.5 pg/mL and 24.4 ± 1.7 pg/mL versus 34.9 ± 4.7 pg/mL for control group (Fig. 7 E).

Discussion
Due to the lack of scienti c data in literature concerning biological effects of mixtures containing NPs and other environmental pollutants, leading in vitro and in vivo studies is a major challenge to improve knowledge on toxicological impact on human health. Therefore, our research focused on exposure to mixture aerosols containing hydrogen chloride gas (HCl g ) and alumina nanoparticles (Al 2 O 3 NPs). These two pollutants of interest are especially retrieved in high concentrations after solid composite propellants use in aerospace and defense elds. Inhalation is the main route of exposure to these combustion aerosols, and little is known about combined effects of HCl g and Al 2 O 3 NPs on the respiratory tract.
Consequently, our objective was to assess early pulmonary effects of model aerosols mimicking for instance inhalation exposure to combustion aerosols from solid composite propellants on rats. To investigate toxicological effects, animals were nose-only exposed 4h (unique exposures ; UE) or 4h a day for 4 days (iterative exposures ; IE) to aerosols. Wistar rats were selected for this project as they are often used in inhalation studies [34] and recommended by different OECD guidelines [35,36,37] for pulmonary toxicity assessment. Nose-only inhalation was used to insure better control of the inhaled dose. To assess pro-in ammatory response and oxidative stress induction, bronchoalveolar lavage uids (BALF) content and lungs histopathology were analyzed 24h after exposures.
Speci c physico-chemical characterization is a mandatory step to better understand the biological effects of mixture aerosols. Their characterization revealed the presence of spherical Al 2 O 3 NPs agglomerates with sizes smaller than 250 nm. Al 2 O 3 NPs concentrations in aerosols were high and comprised between 20.0 mg/m 3 and 22.1 mg/m 3 . Indeed, Paur and colleagues reported in their literature review that 5 mg/m 3 could be considered as an upper limit for workplace NPs mass concentration, according to Occupational Safety and Health Administration (OSHA) standard for respirable nuisance dust [38]. Based on this value, they estimated a mass of NPs per surface area deposition in human lungs of 5 x 10 -3 μg/cm²/h, which would correspond in our exposure scenarios to theoretical particles deposition in human lungs of 0.02 μg/cm² after UE and of 0.08 μg/cm² after IE. Considering alveolar surface in rats of 4000 cm², exposures to NPs aerosols led to higher measured pulmonary depositions.
Indeed, for Al 2 O 3 NPs exposures values of 0.04 μg/cm² and 0.13 μg/cm² were respectively obtained for UE and IE, while concentrations for co-exposures with HCl g 5 ppm were respectively 0.03 μg/cm² and 0.18 μg/cm². Therefore, these experimental values are higher than theoretical values but consistent with aerosols high NPs concentrations. Even if we observed high NPs pulmonary depositions, these scenarios are relevant in a context of accidental exposures where speci c workers (i.e. military personnel) could be exposed to combustion aerosols.
To quantify Al 2 O 3 NPs biodistribution in lungs after exposures, aluminum concentrations were measured using ICP-OES. Obtained values were approximately four times higher in lungs after IE compared to UE. These results are consistent with exposure scenarios as IE correspond to four repetitions of UE. NPs accumulated in lungs after exposures and were still retained by lungs after 96 hours in the case of IE, suggesting absence or slow clearance mechanisms after Al 2 O 3 NPs aerosols inhalation. This result corroborates with a study carried out on Sprague-Dawley rats by intratracheal instillations, which already concluded to a very slow clearance, gradual accumulation and pulmonary retention of alumina particles [39]. Moreover, aluminum concentration was signi cantly increased in case of co-exposure to Al 2 O 3 NPs with HClg compared to Al 2 O 3 NPs aerosols, while NPs concentration and size distribution were not modi ed. Therefore, we hypothesized that acidi cation of mixture aerosols may increase pulmonary deposition.
Pro-in ammatory properties of Al 2 O 3 particles have already been reported in literature following inhalation [10,11,40]. To date and to our knowledge, no literature data were found concerning the potential in ammatory response induced by HCl g alone or Al 2 O 3 NPs / HCl g inhalation exposures.
Therefore, we studied pro-in ammatory biomarkers following Al 2 O 3 NPs, HCl g or Al 2 O 3 NPs / HCl g mixtures nose-only exposures. Different secretion pro les were observed for each cytokine, depending on aerosol composition and exposure scenario. No increase of immune cells or cytokines concentrations was measured in animals BALF exposed to air or HCl g . Our results showed early in ammation triggered by exposures to aerosols containing Al 2 O 3 NPs. Signi cant amount of alveolar macrophages and PMN were measured in BALF only after IE with increased secretion of IL-1β, TNF-α and GRO/KC. However, iterative co-exposures to Al 2 O 3 NPs / HCl g only induced PMN in ux and secretion of IL-1β and TNF-α.
These results corroborate with histopathological observations with mainly interstitial and in ammatory lesions (localized areas composed of PMN and lymphocytes) after Al 2 O 3 NPs inhalation, while Al 2 O 3 NPs / HCl g mixtures induced vascular and interstitial lesions. Therefore, IE induced more potent proin ammatory response compared to UE but different immune cells and cytokine pro les were observed after Al 2 O 3 NPs and mixtures exposures. Increased cytokine concentrations observed in BALF corroborate with PMN in ux as they have chemoattractant properties [10,41,42]. IL-1β and TNF-α seem to orchestrate the pulmonary pro-in ammatory response. On the one hand, it has been described that Al 2 O 3 NPs could stimulate the nuclear factor NFκB pathway [43]. IL-1β concentrations increases could thus be associated with this hypothesis, suggesting the in ammasome pathway involvement in the pulmonary pro-in ammatory response initiated by Al 2 O 3 NPs. Indeed, the nuclear factor NFκB can contribute to the in ammasome regulation [44] and together they can lead to IL-1β synthesis [45]. This conjecture is reinforced by the fact that this pathway is also known to be activated by pulmonary in ammation in patients with chronic obstructive pulmonary disease [46]. On the other hand, Al 2 O 3 NPs can also induce TNF-α release in BALF. This result corroborate with the study of Kim et al. which showed neutrophils, LDH, TNF-α and IL-6 increased concentrations in BALF after Al 2 O 3 NPs nose-only exposure of rats for 28 days (5 days/week) [10]. TNF-α secretion after exposure to Al 2 O 3 NPs may be linked to IL-1β release. Indeed, in a context of acute pulmonary in ammation, Saperstein and colleagues demonstrated that IL-1β contributes, in part, to TNF-α mediated chemokine release, and neutrophil recruitment to the lung, potentially associated with altered soluble TNFR1 release into the BALF [47]. Besides, in our study GRO/KC and MIP-2 secretion in BALF were only measured after Al 2 O 3 NPs exposure. We hypothesized that co-exposure to Al 2 O 3 NPs / HCl g mixtures could modify kinetic of these cytokines secretions in BALF.
Signi cant increase of MIP-2 concentrations was found in BALF only after UE to Al 2 O 3 NPs. As a consequence, MIP-2 may act as an early step contributing to PMN, macrophages and lymphocytes recruitment. Pirela and colleagues also demonstrated down-regulation of GRO/KC and MIP-2 in the nasal lavage uid after iterative whole-body inhalation exposure to printer-emitted engineered NPs in rats (containing aluminum oxide) [48]. Otherwise, IL-6 and INF-γ were not detected in BALF after UE and IE to the different aerosols of the study (respective detection limits: 13 pg/mL and 6.2 pg/mL). For IL-6 secretion, this result is contradictory with the study of Kim and colleagues as they showed that IL-6 was signi cantly increased in BALF after iterative nose-only inhalation (28 days; 5 days/week) of Al 2 O 3 NPs by rats [10]. However, Adamcakova-Dodd and colleagues exposed male mice (C57Bl/6J) to aluminum oxide-based nanowhiskers 4h/day, 5 days/week for two or four weeks in a dynamic whole-body exposure chamber and did not found neutrophilic in ammation nor IL-6 secretion in BALF [49]. Moreover, IL-6 has been shown to be increased in some pathological conditions involving the respiratory system when there is sustainability of lesional processes [50]. INF-γ promotes pulmonary in ammation through oxidative stress induction [51] and could have been secreted earlier than 24h post-exposure to the aerosols, inducing nucleic acids oxidative damages (8-OHdG), which were still detected at that time. Therefore, we hypothesized that aluminum oxide-based nano-objects are notably able to induce secretion of different cytokine secretion pro les in BALF depending of their shape, time of exposure and used inhalation system.
As oxidative stress is a physiological mechanism closely linked to pulmonary in ammation induced by NPs exposure, and indirectly to the increase of PMN number potentially releasing free radicals in the alveoli, this endpoint was investigated [52,53,54,55]. Nucleic acids oxidation (8-OHdG were only detected in situ following UE to HCl g , Al 2 O 3 NPs or mixtures exposures while no increase of lipid peroxidation was detected in BALF (8-Isoprostane). Interestingly, preliminary results for DNA doublestrand breaks immuno uorescent detection gave similar results: γ-H2AX was detected only after UE to the same aerosols. Absence of 8-OHdG following IE could be explained by lesions repairing systems induction after 24h which would limit their later appearance and persistence over time. The main mechanism implied in these oxidative damages repair is base excision repair [56], but other intracellular actors such as p53 protein [57] or PARP-1 enzyme [58] may also contribute to damages repair. As previously shown, NPs inhalation could also initiate increased production of enzymes such as endonuclease III homolog 1 (NTH1) or apurinic/apyrimidinic endonuclease 1 (APE1) [59], p53 [57] or poly [ADP-ribose] polymerase 1 (PARP-1) [58], involved in various repair mechanisms and consequently limit side effects on nucleic acids. Similar hypothesis could be made involving γ-H2AX itself as inductor of DNA repair mechanism. Indeed, γ-H2AX is DNA double-strand breaks marker, but it also initiates these speci c lesions repair [60,61,62,63]. Lack of lipid peroxidation with 8-isoprostane detection in BALF might be due to activation of the antioxidant response following exposures. Signi cant decreases of 8isoprostane after IE to Al 2 O 3 NPs or mixtures may also be due to changes of antioxidant balance. It would be interesting to perform glutathione peroxidase dosage in BALF in further studies to assess these antioxidant mechanisms [64].
As obtained results highlighted Al 2 O 3 NPs lung retention after aerosols exposures, early pro-in ammatory and oxidative stress induction, the aerosols effect on alveolar-capillary barrier permeability was also studied. Signi cant elevations of LDH and total protein concentrations were measured in BALF only after IE to Al 2 O 3 NPs and mixture aerosols containing HCl g , indicating impairment of air-blood barrier permeability and cell cytolysis. These biological effects could be linked to pro-in ammatory mechanisms and PMN in ux observed in lungs following IE to Al 2 O 3 NPs alone or combined with HCl g . Indeed, as previously demonstrated, PMN recruitment and pulmonary in ammation trigger alveolar-capillary barrier permeabilization [65,66]. Air-blood barrier permeabilization could promote translocation of inhaled particles into bloodstream leading to potential systemic effects or local toxicity on other organs such as the brain, the liver, the spleen or kidneys. A study is currently performed on these organs to assess local histopathological effects induced by aerosols inhalation. Mainly vascular, interstitial and in ammatory lesions were rapidly observed after UE and IE to aerosols. Similar lesion scores were obtained after the two exposure scenarios and the addition of HCl g to Al 2 O 3 NPs aerosol did not worsen lesions, meaning that it is mainly the presence of Al 2 O 3 NPs in aerosols, which induces early pulmonary histopathological modi cations.

Conclusion
Consequently, we conclude that early pulmonary effects induced by Al 2 O 3 NPs and/or HCl g aerosols inhalation are in ammation, oxidative stress and alveolar-capillary barrier permeabilization. Obtained results underline that the exposure scenario mainly in uence toxicological effects of mixtures. Al 2 O 3 NPs presence in the aerosols also seem to play essential role driving toxicity. Indeed, despite an increased pulmonary aluminum biodistribution in case of IE to mixtures compared to Al 2 O 3 NPs, similar proin ammatory responses were observed in the two conditions. UE induce oxidative stress resulting in nucleic oxidative damages, while IE essentially promote pulmonary in ammation. Special attention should be paid to pulmonary pro-in ammatory and oxidative stress induction as these phenomena are known to be closely linked to long-term pathologies such as pulmonary brosis or cancers [67,68,69].
Therefore, this study raises concerns towards potential long-term pulmonary toxicity of combustion aerosols, and emphasizes necessity to precise health hazard resulting from these aerosols exposures.

Animal exposure
Two exposure scenarios were tested in this study. Unique exposures (UE) of 4h and iterative exposures (IE) of 4h a day for 4 days were performed. The 96 animals were distributed in 8 groups of 12 rats (6 rats for biochemistry analysis, 3 rats for histopathology analysis and 3 rats for aluminum quanti cation in lungs among each group) respectively exposed to air, HCl g , Al 2 O 3 NPs or Al 2 O 3 NPs/HCl g for each exposure scenario. They were exposed in a nose-only inhalation tower (randomized placement on the inhalation tower) to nebulized Al 2 O 3 NPs suspensions in ultrapure water (10 g/L) using a PALAS AGK 2000E (Palas) and/or to 5 ppm of HCl g . Exposure ow was controlled using mass ow meters (TYLAN).
Both compounds were mixed (Al 2 O 3 NPs/HCl g exposures) prior entering in the inhalation tower at a 20 L/min ow rate. Samples were taken on bubblers or cellulose lters using mass ow meters to allow aerosols characterization. A neutralization system (bubblers with caustic soda and desiccant) was placed at the exposure device exit to eliminate generated aerosols. Rats were euthanized by iso urane inhalation 24h after the last exposure and biological samples were collected for further analysis. Aluminum quanti cation in lungs Rats (n=3) were euthanized 24h after the last exposure and perfused with PBS to remove blood from lungs. Aluminum burden was then quanti ed using ICP-OES. Brie y, tissue samples were digested in a mixture of concentrated nitric acid (HNO 3 ; Analpure® Analytika 69% for UE and Normatom VWR 67-69% for IE) and hydro uoric acid (HF; Analpure® Analytika 48% for UE and Normatom VWR 47-51% for IE) using a microwave oven (Mars Xpress CEM for UE and Ethos One Milestone for IE). After cooling to a room temperature, boric acid solution was added to neutralize digests (H 3 BO 4 ; Ultra-pure Chem Lab 99.9% for UE and Suprapur® Merck 99.9999% for IE) and samples were diluted in ultrapure water ( nal volume of 50 mL). Aluminum quanti cation was then performed using ICP-OES 5110 (Agilent, for UE, LLOQ = 0.25 µg/g of lung) or ICP-OES Optima 8300 (Perkin Emler, for IE, LLOQ = 1.25 µg/g of lung).

Monitoring and characterization of aerosols
Software used to realize spectra analysis were ICP Expert II 7.3 (Agilent, for UE) or WinLab32 ICP v5.5 (Perkin Emler, for IE).

Clinical symptoms and body weight
Daily monitoring of clinical symptoms was performed throughout the experiment. Respiratory, behavioral and dermal visible changes were watched. Type, date of occurrence and symptoms severity were individually recorded. Rats were weighted at the arrival, before and after each 4h exposure and before euthanasia.

Bronchoalveolar lavage uid (BALF) analysis
Rats (n=6) were euthanized 24h after the last exposure and bronchoalveolar lavages were performed with PBS. A rst bronchoalveolar lavage was performed using 5 mL of PBS. Two other lavages were then performed with 10 mL of PBS to collect more cells. Collected BALF were centrifuged at 350 g (4°C) for 10 min.

Lungs histopathology
After euthanasia, rats (n=3) were perfused with PBS then with 4% formalin. Injections of 4% formalin were also performed directly inside the lungs to in ate them. After 48h incubation in 4% formalin, lungs were transferred to PBS and included in para n. 5 μm slices were cut from para n blocks with Autocut 2045 (Leica) and stained with hematoxylin, phloxine and saffron (HPS staining). Lungs histopathology study was carried out on cranial lobes of right lungs. Samples were observed under DMI6000B light microscope (Leica) with a DFC 450 C color camera (Leica).
After staining, microscopy slides were sent to Novotec society who performed anatomo-histopathological reading and lesions scoring. Localization (alveolar, vascular or bronchial) and severity of pulmonary lesions resulting from exposures was assessed (Table 3) and lesion scores were calculated individually. Lesion scores means for each experimental group are presented in arbitrary units (A.U.).

Statistical analysis
Results are presented as means ± standard deviations (SD) for each experimental group. Statistical analysis was performed using GraphPad Prism 7 software. Within the same exposure scenario, comparison with control group (air) or other experimental groups means was performed using one-way ANOVA variance tests and Tukey's post-tests. To compare exposure scenarios for the same aerosol twoway ANOVA variance tests and Sidak's post-tests were implemented. Statistical analysis was performed using an α risk of 0.05 and p-value was used to assess differences signi cance. P-values were represented on gures using stars or circles (* or ¤ p-value < 0.05, ** or ¤¤ p-value < 0.01, *** or ¤¤¤ pvalue < 0.001, **** or ¤¤¤¤ p-value < 0.0001).

Consent for publication
Not applicable.

Availability of data and materials
Not applicable.

Competing interests
The authors declare that they have no competing interests.

Funding
The French Government of Defense Procurement Agency, The Direction Générale de l'Armement (DGA), The Agence de l'Innovation Défense (AID), France, research program PROPERGOL funded this work.
Author's contribution    Pulmonary aluminum deposition after nose-only exposures. Rats were exposed to air, HClg 5 ppm, Al2O3 13 nm or to the mixture Al2O3 13 nm + HClg 5 ppm. Clear blue points symbolize unique exposures (UE) and dark blue points symbolize iterative exposures (IE) to each aerosol. Aluminum quantities (μg/g of lung) were measured in lungs 24h after the last exposure using inductively coupled plasma optical emission spectrometry (ICP-OES). One-way ANOVA and Tukey's post-test (n=3, ** p-value < 0.01, **** pvalue < 0.0001, α risk = 0.05) were performed to compare each experimental condition to the control within the same exposure scenario and Two-way ANOVA and Sidak's post-test (n=3, ¤¤¤ p-value < 0.001, ¤¤¤¤ p-value < 0.0001, α risk = 0.05) were performed to compare the two exposure scenarios for the same aerosol. Figure 3 Total proteins, LDH and immune cells populations in bronchoalveolar lavage uids (BALF). Rats were exposed to air, HClg 5 ppm, Al2O3 13 nm or to the mixture Al2O3 13 nm + HClg 5 ppm. Clear blue points symbolize unique exposures (UE) and dark blue points symbolize iterative exposures (IE) to each aerosol. Total proteins (g/L) (A) and lactate dehydrogenases (LDH, U/L) (B) concentrations were measured in rats BALF 24h after the last exposure. Total cells, polymorphonuclear neutrophils (PMN), macrophages and lymphocytes (cells x106) were counted on BALF cytospins (repectively C, D, E and F). One-way ANOVA and Tukey's post-test (n=3, * p-value < 0.05, ** p-value < 0.01, *** p-value < 0.001, **** p-value < 0.0001, α risk = 0.05) were performed to compare each experimental condition to the control within the same exposure scenario and Two-way ANOVA and Sidak's post-test (n=3, ¤ p-value < 0.05, ¤¤ p-value < 0.01, ¤ ¤¤ p-value < 0.001, ¤¤¤¤ p-value < 0.0001, α risk = 0.05) were performed to compare the two exposure scenarios for the same aerosol. Aerosols pro-in ammatory potential on rat lungs after nose-only inhalations. Rats were exposed to air, HClg 5 ppm, Al2O3 13 nm or to the mixture Al2O3 13 nm + HClg 5 ppm. Clear blue points symbolize unique exposures (UE) and dark blue points symbolize iterative exposures (IE) to each aerosol. Proin ammatory cytokines concentrations (pg/mL) of TNF-α (A), IL-1β (B), GRO/KC (C) and MIP-2 (D) were measured in rats BALF 24h after the last exposure by ELISA multiplex. One-way ANOVA and Tukey's posttest (n=3, * p-value < 0.05, ** p-value < 0.01, **** p-value < 0.0001, α risk = 0.05) were performed to compare each experimental condition to the control within the same exposure scenario and Two-way  Pulmonary histopathological lesions scoring. Rats were exposed to air, HClg 5 ppm, Al2O3 13 nm or to the mixture Al2O3 13 nm + HClg 5 ppm. Scoring was performed on HPS stained lungs slices of three rats of each group. Lesion scores (mean ± SD, n=3) were estimated for each experimental aerosol and exposure scenario based on the presence and the gravity of vascular, alveolar and bronchial lesions observed in rats lungs 24h after the last exposure. Empty and striped bars represent respectively unique exposures (UE) and iterative exposures (IE) results.