The Effects of Inhaling Hydrogen Gas on Macrophage Differentiation, Fibrosis and Lung Function in Mice with Bleomycin-induced Lung Injury

Acute respiratory distress syndrome, which is caused by acute lung injury, is a destructive respiratory disorder caused by a systemic inammatory response. Persistent inammation results in irreversible alveolar brosis due to excessive activation of M2 macrophages. Because hydrogen gas possesses anti-inammatory properties, we hypothesized that daily intermittent inhalation of hydrogen gas could suppress persist acute inammation by inducing functional changes in macrophages, and consequently inhibit lung brosis during late-phase lung injury.


Introduction
Excessive, non-speci c in ammation in the lungs initiates pathological processes leading to acute lung injury (ALI) and acute respiratory distress syndrome (ARDS), which directly and indirectly cause destruction of lung tissue including alveolar structures [1]. The development of ALI/ARDS is triggered by an immune response that leads to activation of classically activated macrophages, or M1 macrophages, and accumulation of neutrophils in the alveoli. As a result, alveolar epithelial cells and vascular endothelial cells are extensively damaged, and pulmonary edema develops (exudation phase) [1,2]. Seven to 10 days after onset, proliferation of type II alveolar epithelial cells and broblasts is observed in the destroyed alveoli (proliferative phase) [3]. This process of reconstruction is accompanied by persistent in ammation can promote alveolar brosis and decrease of alveolar compliance ( brotic phase). Alternatively activated macrophages, or M2 macrophages, which differentiate as a result of persistent in ammation, play a key role in promoting disordered alveolar brosis [2]. Overexpression and persistence of M2 macrophages in the alveolar interstitium is a hallmark of the transition to the brotic phase [4,5]. Therefore, novel therapies to reduce persistent in ammation prior to the establishment of irreversible alveolar brosis are required and have attracted great interest.
Molecular hydrogen has potent antioxidant and anti-in ammatory properties [6,7]. The mechanisms underlying the anti-in ammatory effects of hydrogen are becoming clearer with known inhibition of proin ammatory cytokines and upstream signaling molecules [7][8][9]. Previous studies demonstrated that hydrogen ameliorates lung brosis caused by a variety of insults by inhibiting in ammatory signaling by the innate immune system and regulating signaling cascades that impact macrophages [8, 10,11]. Inhalation of hydrogen gas may be a straightforward and promising therapeutic option, because inhaled gaseous molecules can directly reach the alveoli. Additionally, inhaled hydrogen has a low chemical toxicity [11][12][13]. Thus, this gaseous therapy has good clinical feasibility, as long as its ammability can be controlled.
A previous investigation demonstrated that hydrogen inhalation suppressed increases in oxidative stress and in ammation induced by intratracheal bleomycin administration in mice, and suppressed the malignant cycle toward lung brosis initiated by transforming growth factor (TGF)-β1 and in ammation [14]. This study was initially aimed to examine the hypothesis that hydrogen could reduce brosis in the ARDS brotic phase through suppression of lung in ammation and oxidative stress. However, Gao et al. reported hydrogen inhalation mitigated oxidative stress and reduced lung brosis using similar animal model after we initiated our experiments [14]. Therefore, to make the rationale of doing our experiment to ll the gap in knowledge, we tried to clarify the involved upstream mechanisms in the protective effects of hydrogen and focused on macrophage differentiation, which may be the target of hydrogen. Also, we tried to make a physiological assessment of the effects of hydrogen in clinically relevant evaluation using basal respiratory function parameters, since the protective effects of hydrogen were never physiologically tested in previous studies [7,15]. Our results rst showed that the primary effect of hydrogen was inhibition of in ammatory signaling in the innate immune system, which ameliorated lung brosis via regulation of signaling cascades that regulate M2 macrophage differentiation [8, 10,16].

Materials And Methods
Animals Eight-week-old, C57BL/6 male mice (21-23g, speci c-pathogen free) were purchased from CLEA Japan Inc (Tokyo, Japan). Mice were kept on a 12-hour light / dark cycle at 20℃ to 22℃ and fed sterile food and water. Every effort was made to minimize the number of experimental animals and minimize pain or distress during the experimental procedures. All protocols followed the principles of laboratory animal care (NIH Publication No. 86-23, revised 1985), and all research protocols were reviewed and approved by the Animal Care and Use Committee, Okayama University (OKU-2018876). This study was conducted in compliance with the ARRIVE guidelines (https://arriveguidelines.org/). Animals were sacri ced at de ned endpoints using CO 2 asphyxiation or by exsanguination under deep anesthesia with intraperitoneal administration of 0.75 mg / kg medetomidine hydrochloride (Domitor, Meiji Seika Pharma, Tokyo, Japan), 4 mg/kg midazolam (Dormicum, Astellas Pharma, Tokyo, Japan), and 5 mg/kg butorphanol (Vetorphale, Meiji Seika) as previously described [14]. Collected samples were snap-frozen using liquid nitrogen and stored at -80°C until use. Animals were checked twice daily after the administration of bleomycin. Dying animals were sacri ced using a humane endpoint.

Generation of ALI/ARDS with bleomycin administration and inhalation of hydrogen gas
This study was conducted using a well-established mouse model of ALI/ARDS and idiopathic pulmonary brosis [15,17]. In summary, lung injury was generated by administrating bleomycin (bleomycin hydrochloride, Nippon Kayaku, Tokyo, Japan) dissolved in saline intratracheally via tracheotomy [18]. The bleomycin causes persistent in ammation pharmacologically in the bronchus and alveoli, eventually resulting in alveolar brosis. Mice were anesthetized, and an incision was made through the neck into the front of trachea. Bleomycin dissolved in saline (50 µl, 1 mg/kg) was injected using a Hamilton syringe and a 32G needle, then the wound was closed by cyanoacrylate glue. In sham controls, saline without bleomycin was administrated in the same manner. We conducted a preliminary pathologic assessment, which con rmed that bleomycin administration induced lung injury with temporal changes in pathology that mimicked those observed during ALI/ARDS (data not shown).
Mice were randomly assigned to 1 of 4 experimental groups: 1) saline administration and air inhalation (SA group), 2) saline administration and hydrogen inhalation (SH group), 3) bleomycin administration and air inhalation (BA group), and 4) bleomycin administration and hydrogen inhalation (BH group). A gas cylinder containing 4% hydrogen and 96% nitrogen blended gas was prepared (Taiyo Nissan, Tokyo, Japan For gas administration, 5 or fewer mice were placed in a sealed acrylic box (L 40 cm x W 20 cm x H 20 cm) for mixed gas exposure while temperature (acceptable range 22-24℃) and humidity (acceptable range 40-70%) were monitored. Mice were exposed to either air or 3.2% hydrogen in air for 6 hours every day for either 7 or 21 days.

Respiratory physiological examination
The respiratory physiology was evaluated using a FlexVent® small animal ventilator with spirometer (SCIREQ, Montreal, PQ, Canada). The programs for examination of respiratory function were already programmed into the device and were performed according to the manufacturer's instructions. Mice were anesthetized as described above, and 1 cm of a 18-gauge endotracheal tube was inserted into the trachea by tracheostomy. The endotracheal tube was attached to the FlexVent. Then, mechanical ventilation is started at 150 respirations per minute, 10 mL/kg of tidal volume, and an inspiratory:expiratory ratio of 2:3 for 1 minute. Inspiratory capacity (IC) was measured in mL using the

Computed tomography
The lung computed tomography (CT) images were taken using a small-animal CT system, Latheta LCT200® (Hitachi, Ltd. Tokyo. JAPAN). The mice are sedated, then inserted into the CT machine and imaged with the following settings: Imaging condition, lung; Pixel size, 48 μm; Slice thickness, 192 μm; Slice interval, 192 μm; X-ray voltage, Low; Scale of tomographic image, -700 to +100; and Respiratory synchronization, "Yes". Of the 70 slices taken of the whole lung eld, 40 slices in the center were used for analysis.
After the CT images were saved as JPG les, images of the inside of the thorax were extracted and converted to 8-bit grayscale. To trace the areas in the lung containing air, the the "Threshold" program was set at "Range: 0-136". To measure the area, the following settings in "Analyze Particles" program were used: Size (inch^2), 0 -In nity; Circularity 0.00-1.00; and Show Bare Outline. The air-containing area of the whole lung eld was calculated by integrating the slice width and the area containing air as detected above.

Hematoxylin and Eosin and Elastica Masson staining
The left upper lung lobes were xed with 4% paraformaldehyde dissolved in phosphate buffered saline (PBS) for 2 days, embedded in para n, then sliced into 4-µm sections. Hematoxylin and eosin (HE) staining and Elastica Masson (E-M) staining were performed using standardized protocols by skilled technicians in the Central Research Laboratory at Okayama University. Images were automatically captured using the Nano-Zoomer 2.0RS slide scanner (Hamamatsu Photonics, Shizuoka, Japan) and analyzed using NDP.view2 software, (Hamamatsu Photonics, Shizuoka, Japan). For the analysis of collagen type I (COL1), bronectin and α-smooth muscle actin (αSMA), proteins (10μg) from lung tissue were separated by electrophoresis on 8% acrylamide gels without SDS and transferred to Immobilon®-P polyvinylidene di uoride (PVDF) membrane (0.45 µm) (Merck, Darmstadt, Germany). For the analysis of TGFβ, proteins (10μg) from lung tissue were separated by electrophoresis on 12% acrylamide, 0.1% SDS gels.
PVDF membranes are blocked with 5% non-fat dry milk to prevent non-speci c binding of antibodies. Primary antibody against bronectin, αSMA, COL1, and TGFβ were diluted with Can Get Signal immunoreaction enhancer solution 1 (Toyobo, Osaka, Japan) (Supplementary Information), and incubated with the membranes overnight at 4°C. Horseradish-peroxidase-conjugated secondary antibodies against mouse IgG and rabbit IgG were diluted with Can Get Signal immunoreaction enhancer solution 2 (Toyobo, Osaka, Japan) and membranes were incubated for 2 hours at room temperature. Chemiluminescence detection was performed with ECL Prime Western Blotting Detection Reagents (Cytiva, Tokyo, Japan) and a WSE-6100 LuminoGraph I (ATTO Corporation, Tokyo, Japan).

Immunohistochemistry
Para n-embedded lung tissue sections (4 µm) were immunostained for TGF-β using an ABC Kit (Vector laboratories INC., Burlingame, California). Information on the primary and secondary antibodies used is shown in the Supplementary Information. Sections were depara nized, rehydrated, and treated for antigen retrieval with 10 mM citric acid pH 6.0 at 120°C for 10 minutes in a pressure cooker. Endogenous peroxidase inhibition was performed with 0.3% hydrogen peroxide in PBS for 20 minutes at room temperature. Blocking treatment was performed with 10% goat serum in tris buffered saline with 0.1% Tween 20 (TBS-T) to prevent non-speci c binding of antibodies. The primary antibodies were diluted by Can Get Signal immunostaining Solution A (Toyobo, Osaka, Japan), applied to the sections, incubated overnight at 4°C, and then washed with TBS-T. Biotin-conjugated secondary antibodies were diluted by Can Get Signal immunostaining Solution A, applied on the sections, and incubated for 2 hours at room temperature. After washing, ABC reagent was applied to the sections then incubated for 30 minutes at room temperature as per the manufacturer's instructions. For 3,3'-diaminobenzidine (DAB) staining, one DAB tablet was dissolved in 50 mL of 0.05mol/l Tris-HCl buffer pH 7.6 with 10 μL of 30% hydrogen peroxide as per the manufacturer's instructions. Sections were incubated in DAB solution for 10 minutes at room temperature, then washed under running water, counterstained with hematoxylin, dehydration, clearing, and coverslipping Immuno uorescence Para n blocks were sectioned, depara nized, dehydrated, and treated for antigen retrieval using the technique described above. The multiplex uorescent immunostaining was used for staining with antiionized calcium binding adaptor molecule 1 (Iba-1) antibody and anti-CD163 antibody. Information on primary and secondary antibodies is given in the Supplementary Information. Blocking treatment was performed with Super Block® (SCY AAA125, Cosmo Bio Co., Ltd. Tokyo, Japan). Anti-Iba-1 antibody and anti-CD163 antibody were diluted in Can Get Signal immunostaining Solution A, then incubated on the tissue section overnight at 4°C. After washing, the sections were incubated with uorescently labeled secondary antibodies with Alexa Flour (AF) 488 or 594. DAPI-Fluoromount G® (0100-20, SouthernBiotech, Birmingham, AL) was used for nuclear staining and sealing.
Fluorescent images were taken by the Mantra TM Quantitative Pathology Imaging System (PerkinElmer Inc., Waltham, Massachusetts), and cells were counted in the alveoli and interstitium were automatically using the InForm® 2.4.10 software (Akoya Biosciences, Inc., Menlo Park, California). Three images were taken randomly from each section with a 200x image. Fluorescence imaging was performed at 488 nm, and 594 nm wavelengths. In the InForm software, a computer learning system was used to learn the characteristics of alveolar epithelium and alveolar interstitum tissues and exclude tracheal epithelial cells. The cells were identi ed by DAPI staining, and the immunostaining was visualized at 488 nm (Iba-1) or 594 nm (CD163) wavelength. The intensity thresholds for Iba-1-positive and CD163-positive cells were carefully adjusted and identi ed, and all images were analyzed according to the same rules. The median of the 3 results obtained for each section was then analyzed.

Statistics
Statistical analysis was performed using IBM SPSS Statistics version 23.0 (IBM, Armonk, New York). Statistically signi cant differences between groups were determined using the Kruskal-Wallis test followed by Dunn's multiple comparison test. All values are presented as mean ± 95% con dence interval (CI). Results were considered signi cant at P < 0.05.

Results
Hydrogen inhalation mitigates respiratory physiological dysfunction during brotic phase after bleomycin-induced lung injury.
To determine the impact of hydrogen inhalation on respiratory function, several basal parameters were examined after lung injury with bleomycin and 21 days of hydrogen inhalation or sham/ air therapy. While IC and Rs were not signi cantly different between mice that received hydrogen therapy and mice that received sham/ air therapy (BH group vs BA group, Fig. 1a, b) . 1d). In brotic phase after lung injury, brotic changes progress in the alveolar interstitum, and the lung tissues become hardened. These results suggests that hydrogen inhalation therapy preserved the ability of the lung to expand and reduced lung stiffness. There were no differences in any of the respiratory parameters examined between hydrogen and sham/ air therapy when lung injury was not induced (SH and SA groups), suggesting that hydrogen has no effects on respiratory physiological function in individuals without alveolar damage (Fig. 1).
Hydrogen inhalation can attenuate the reduction in lung capacity typical of bleomycin-induced lung injury.
After lung injury and 21 days of hydrogen therapy or sham/ air therapy, lung capacity was measured directly using CT volumetry (Fig. 2a, left column). The aerated area in each CT section was identi ed (Fig. 2a, right column), quantitated, and used to calculate the volume of aerated space in lung (Fig. 2b). Although bleomycin-induced lung injury signi cantly reduced aerated lung capacity (BA group), hydrogen treatment partially ameliorated this reduction, and the BH group had higher aerated lung capacity than the BA group (BH 296µL [95% CI: 245-347] vs BA 207µL [95% CI: 128-286], p = 0.02). Hydrogen treatment had no effect on aerated lung capacity in the absence of lung injury (SH and SA groups) (Fig. 2b).
Hydrogen reduced alveolar brosis in brotic phase after bleomycin-induced lung injury.
Histopathological evaluation for brosis was performed 21 days after lung injury. There were many cytoplasm-rich cells, which might include broblasts, myo broblasts, and in ammatory cells, in the alveolar interstitium in lungs with bleomycin-induced ALI. The presence of these cells in the alveolar interstitium was attenuated by hydrogen treatment (BH group) (Fig. 3, H&E staining). Collagen bundles were seen in the interstitium of the mice with bleomycin-induced ALI (BA group) and were less frequently observed in mice treated with hydrogen (BH group) (Fig. 3, E-M).

Hydrogen inhibited increases in bronectin protein expression during brotic phase after bleomycininduced lung injury.
When the expression of extracellular matrix proteins was examined after bleomycin-induced ALI, expression of COL1 and αSMA did not differ between any of the treatment groups. (Fig. 4a) (Fig. 4b).
Hydrogen inhalation attenuates the upregulation of critical interleukins and downregulates bronectin mRNA in lung tissue early after bleomycin-induced lung injury.
IL-6 is a cytokine produced by alveolar endothelial cells and macrophages in response to the innate immune system and is a major mediator of fever and acute reactions. IL-6 also promotes the differentiation of type 2 T helper cells, which produce IL-4, IL-13 and L-10, and M2 macrophages [19][20][21][22]. IL-4 and IL-13 are signals that induce monocytes to differentiate into M2 macrophages [23][24][25]. IL-10 are signals that suppresses the expression of pro-in ammatory cytokines. Expression of the mRNAs for IL-6, IL-4 and IL-13, all of which are considered pro-in ammatory cytokines in the lung, were signi cantly upregulated 7 days after bleomycin administration to induce ALI as compared with saline-treated control lungs. Intermittent hydrogen inhalation signi cantly suppressed upregulation of IL-6, IL-4 and IL-13 in response to bleomycin-induced ALI (Fig. 5a, b, and c) The mRNA expression of brinogen and COL1, components of the extracellular matrix, were also investigated 7 days after bleomycin treatment. The levels of brinogen mRNA were reduced by hydrogen therapy, while the levels of COL1 mRNA were not. These results mirrored the protein expression levels seen 21 days after bleomycin treatment (Fig. 4).
Hydrogen inhalation suppresses the expression of TGF-β in the alveolar interstitium early after bleomycininduced lung injury.
TGF-β1 is secreted by many cell types, including macrophages [26]. Because TGF-β1 plays a central role in brosis by inducing epithelial cells, vascular endothelial cells, and mesenchymal cells to adopt a phenotype that produces extracellular matrix proteins, we examined TGF-β1 expression after bleomycininduced ALI and changes in expression in response to hydrogen therapy. In western blot analysis, the 44 kDa band is likely the latent form of TGF-β1, while the 13 kDa band is the active form. When TGF-β1 protein expression was analyzed by Western blotting, hydrogen treatment did not affect TGF-β1 expression (Fig. 6a). However, when the localization of TGF-β1 was examined using immunostaining, fewer TGF-β1-positive cells were found in the alveolar interstitium after hydrogen therapy than in sham/ air treated controls (Fig. 6b).
Hydrogen inhalation reduces M2 macrophages in the alveolar interstitium after bleomycin-induced lung injury.
Because IL-4 and IL-13 are known to drive the differentiation of M2 macrophages and hydrogen therapy mitigated increases in IL-4 and IL-13 expression that typically accompany bleomycin-induced ALI, we performed immuno uorescent staining identify the phenotype of macrophages in the alveoli and alveolar interstitum in the 4 treatment groups. Anti-Iba-1 antibody was used to detect all macrophages, anti-CD163 antibody was used to speci cally detect M2 macrophages, and the number of macrophages by phenotype was measured. Bleomycin-induced ALI increased the number of Iba-1-positive, CD163-negative macrophages in the alveoli, and hydrogen treatment had no effect on the number of Iba-  (Fig. 7c). These results suggest that one mechanism whereby hydrogen inhalation preserves lung function is blocking the differentiation of macrophages.

Discussion
Hydrogen inhalation therapy has been proven effective in mitigating in several animal models of lung injury including hyperoxic lung injury, hemorrhagic shock-induced lung injury, radiation-induced lung injury, and bronchial asthma [5,7,8,27]. Our study is the rst to prove that hydrogen inhalation therapy effectively attenuates the decline of respiratory physiological function induced by bleomycin in a mouse model of persistent lung in ammation and brosis. We also demonstrated that the protective effects of hydrogen gas inhalation therapy in this lung injury model were accompanied by, and like due in part to, attenuation of pro-in ammatory cytokine expression and inhibition of M2 macrophage differentiation. Identifying macrophage differentiation as a potential underlying mechanism of hydrogen therapy advances our understanding of hydrogen biology.
In this study, we administered an air mixture with 3.2% hydrogen concentration for 6 hours daily beginning in the day of bleomycin administration to induce ALI and continuing for 21 days. Other investigations of inhaled hydrogen therapy have shown effective results with hydrogen concentrations of 2-4% [7,[26][27][28][29][30][31]. Intermittent inhalation was reported to be more effective than continuous inhalation in rat model of Parkinson's disease [1]. Because the endpoint of this study was to assess lung function during brotic phase after lung injury, we targeted treatment to the proliferative phase of ALI/ARDS, which ordinary occurs 7 to 21 days after the onset of lung injury [1]. Our preliminary studies indicated that the bleomycin-induced lung injury model mimicked the temporal changes in pathology observed during ALI/ARDS and dictated our protocol of intermittent hydrogen inhalation for 21 days after bleomycin administration.
During ALI/ARDS, pathogen-associated molecular pattern molecules and damage-associated molecular pattern molecules stimulate type II alveolar epithelial cells and alveolar macrophages to secrete proin ammatory cytokines. The permeability of pulmonary capillaries and alveolar epithelial cells increases, and exudate ows from the blood vessels into alveoli and its interstitium. The immune cells responsible for this are classically activated macrophages (M1 macrophages) and neutrophils [32]. Innate immune sensors, such as Toll-like receptors (TLRs), are expressed not only on macrophages but also on type II alveolar epithelial cells [33]. TLRs sense in ammatory signals, resulting in the recruitment of the adaptor protein myeloid differentiation factor 88 (MyD88), and ultimately in nuclear migration of nuclear factor κB (NF-κB), which binds to its recognition sites in DNA and activates transcription of pro-in ammatory cytokines, such as IL-6 and tumor necrosis factor-α [34]. IL-6, which is expressed in type II alveolar epithelial cells and M1 macrophages, is associated with lung brosis through the differentiation of M2 macrophages [19][20][21][22][23]. Although M2 macrophages are important for tissue repair, an excess of M2 macrophages can cause organ brosis [25,34]. Suppression or deletion of IL-6 suppresses M2 macrophage differentiation and attenuates lung brosis [35,36]. Our nding that hydrogen inhalation therapy reduced IL-6 mRNA expression is consistent with published work demonstrating that hydrogen suppresses the expression of IL-6 [7,23,27,[37][38][39][40][41][42]. Therefore, it is reasonable to hypothesize that the reduction in IL-6 expression in response to hydrogen inhalation therapy caused less M2 macrophage differentiation and less brosis after bleomycin-induced ALI.
IL-4 and IL-13 also induce the differentiation of macrophages to M2 macrophages, which decreases in ammation and encourages tissue repair. Persistent or excessive expression of IL-4 or IL-13 and the accompanying M2 macrophage differentiation leads to abnormal organ brosis [24,25]. In our study, the expression of IL-4, and IL-13 mRNAs were decreased by hydrogen inhalation therapy. The regulation of IL-4 and IL-13 by hydrogen may result in an anti-brotic effect through suppression of M2 macrophage differentiation and thereby reduce alveolar brosis. TGF-β is secreted from M2 macrophages. Hydrogen administration decreased the number of TGF-β-producing cells in the alveolar interstitium, again indicating that control of M2 macrophage differentiation may be an important mechanism underlying the therapeutic bene ts of inhaled hydrogen. The results of this study support models put forth by others that hydrogen therapy regulates upstream signals in a cascade impacting macrophages and innate immunity that leads to in ammation [7,8,10,43].
A previous study using a bleomycin-induced ALI model was reported by Gao and colleagues who found that hydrogen attenuated oxidative stress, increased the expression of the antioxidant glutathione peroxidase, and consequently suppressed the expression of reactive oxygen species in the injured lung [13]. The expression of TGF-β1 was also suppressed and epithelial-to-mesenchymal transition was inhibited, which may be one mechanisms involved in brosis suppression by hydrogen treatment. In our study, there was no change in TGF-β1 protein expression in the lungs, however there were fewer TGF-β1expressing cells in the alveolar interstitium after hydrogen therapy.
Our study adds several novel ndings to the published literature on therapeutic hydrogen in animal models of lung injury. Importantly, we found that hydrogen inhalation suppressed the expression of IL-4 and IL-13 in the lungs in the bleomycin-induced ALI model and suppressed the appearance of M2 macrophages in the alveolar interstitium, likely by suppressing their differentiation. These mechanisms are associated with suppression of lung brosis late after ALI. Another unique aspect of this study was that the degree of preservation of lung function after hydrogen treatment was formally examined using physiological respiratory function tests and CT volumetry. Hydrogen administration increased ventilation and increased alveolar compliance, which strongly suggested that hydrogen inhalation would improve the clinical pro le of ALI/ARDS patients when used as therapy.
The study has some limitations. The pathogenesis of ALI/ARDS in clinical practice is diverse. Bleomycininduced lung injury is only one type of drug-induced lung injury and does not replicate all possible ALI/ ARDS presentations. Determining the most effective hydrogen administration regimen will require additional study and was not a focus of these experiments. Finally, the mechanisms by which hydrogen inhibits the expression of cytokines (IL-6, IL-4, IL-13) were not analyzed. These will need to be clari ed in future studies.

Conclusion
Intermittent hydrogen inhalation therapy with 3.2% hydrogen for 6 hours per day for 21 days inhibited the decline of respiratory physiological function and increase in alveolar brosis typical of ALI. This inhibition of ALI was partially due to via suppression of differentiation of M2 macrophages in the alveolar interstitium in this mouse model of bleomycin-induced ALI.