Low oxygen levels decrease adaptive immune responses and ameliorate experimental asthma in mice

Abstract Background High‐altitude therapy has been used as add‐on treatment for allergic asthma with considerable success. However, the underlying mechanisms remain unclear. In order to investigate the possible therapeutic effects of high‐altitude therapy on allergic asthma, we utilized a new in vivo mouse model. Methods Mice were treated with house dust mite (HDM) extract over 4 weeks and co‐exposed to 10% oxygen (Hyp) or room air for the final 2 weeks. Experimental asthma was assessed by airway hyper‐responsiveness, mucus hypersecretion and inflammatory cell recruitment. Isolated immune cells from mouse and allergic patients were stimulated in vitro with HDM under Hyp and normoxia in different co‐culture systems to analyse the adaptive immune response. Results Compared to HDM‐treated mice in room air, HDM‐treated Hyp‐mice displayed ameliorated mucosal hypersecretion and airway hyper‐responsiveness. The attenuated asthma phenotype was associated with strongly reduced activation of antigen‐presenting cells (APCs), effector cell infiltration and cytokine secretion. In vitro, hypoxia almost completely suppressed the HDM‐induced adaptive immune response in both mouse and human immune cells. While hypoxia did not affect effector T‐cell responses per‐se, it interfered with antigen‐presenting cell (APC) differentiation and APC/effector cell crosstalk. Conclusions Hypoxia‐induced reduction in the Th2‐response to HDM ameliorates allergic asthma in vivo. Hypoxia interferes with APC/T‐cell crosstalk and confers an unresponsive phenotype to APCs.


| INTRODUC TI ON
Asthma is a complex chronic inflammatory disease with a high prevalence. 1,2 Development and severity of allergic asthma are strongly correlated with exposure to indoor allergens like house dust mite (HDM). 3,4 There is ample evidence for the beneficial effects of inhaled corticosteroids and other specific anti-inflammatory treatments for asthma.
However, there are side effects that may affect quality of life and there are patients with poor response to pharmacologic treatment. [5][6][7] High-altitude climate therapy (HACT) was used before pharmacologic treatments were available and since has still been used to complement pharmaceutical intervention. There are multiple case reports and case series in the literature, describing amelioration of allergic asthma through HACT. A recent meta-analysis including 21 studies and a total of 907 participants suggested that HACT is an effective therapy for allergic asthma, although there are no randomized controlled doubleblind studies. 8 In contrast, asthmatic patients from high-altitude regions, when spending time at lower altitude, experience exacerbation of their disease. 9 Of note, the positive effect of HACT is still detectable after months of return from high altitude. 10 Several complementary mechanisms have been postulated to explain these beneficial effects, including reduced allergen load, 11,12 increased exposure to UV light, 13  Mice were treated intra-nasally with a crude extract of HDM (50 µg protein/25 µl in PBS) once per week for 5 weeks, control mice received PBS. Two groups of mice additionally were subjected to reduced oxygen levels (10% normobaric oxygen) for the last 2 weeks of HDM treatment ( Figure 1A). Analysis of lung function parameters and organ collection was performed 72 h after the last challenge.

G R A P H I C A L A B S T R A C T
Hypoxia ameliorates allergic asthma in vivo. Hypoxia impairs differentiation of APCs. Hypoxia blocks adaptive immune response by interfering with APC/T-cell crosstalk. Abbreviations: APC, antigen presenting cell; HDM, house dust mite

| Preparation of bronchoalveolar lavage fluid (BALF) and lung tissue
After sacrifice, BALF was obtained by flushing the lung with 1ml PBS 1mM EDTA plus protease inhibitor cocktail (ThermoFisher Scientific, Vienna, AT). Afterwards, lungs were perfused with ice cold PBS and one lobe each was taken for different analyses as described below.
For details, see Data S1.

| Flow cytometry
Bronchoalveolar lavage fluid and lung single-cell tissue homogenates (see Data S1) were analysed using a LSRII flow cytometer and analysed with the FACSDiva software (BD Biosciences) or a Beckman Coulter Cytoflex S with CytExpert 2.4, as previously described. 17 Antibody details and the gating strategy are provided in Table S1 in Data S1 and Figures E1, respectively. Cells in the BALF are shown F I G U R E 1 Reduced oxygen concentrations decrease house dust mite (HDM) induced goblet cell hyperplasia and airway hyperreactivity. (A) Experimental outline. Mice were sensitized and challenged with a crude extract of HDM or PBS once a week over 4 weeks; after 2 weeks, mice kept under room air (21% oxygen) or reduced oxygen conditions (10% oxygen) for an additional 2 weeks. Analysis was performed 72 h after the last challenge. (B) Serum levels of HDM-specific antibodies. (C) Periodic acid-Schiff (PAS) staining of mouse lung tissue isolated from mice exposed to HDM or PBS for 4 weeks with or without the additional exposure to 10% O 2, scale bar indicates 100 µm. as total cell counts, while those in lung tissue were normalized to weight of the lung piece.

| Lung immunohistochemistry and quantitative histology
Three µm lung sections were deparaffinized in xylene followed by decreasing concentrations of ethanol and then stained by periodic acid-Schiff staining (PAS) according to standard protocols. The percentage goblet cells (GC) and mucus volume were quantified using the NewCast software (Visiopharm, Hoersholm, Denmark) as described in the Data S1. 15,18 Pulmonary vascular muscularization was quantified in sections viously described. 19 Slides were scanned with an Aperio or Olympus slide scanner and analysed using the Visiopharm software. On average 200 ± 90 (mean ± SD) vessels were analysed per slide, the degree of muscularization was calculated as non-, partially and fully muscularized, depending on coverage of the smooth muscle layer.

| Western blotting
Total proteins were isolated from frozen lung samples via mechanically disassociating tissue in liquid nitrogen, then dissolving in RIPAbuffer. Protein content was measured via Bradford-assay, adjusted and separated via non-reducing polyacrylamide-gels. Proteins were wet-blotted, blocked with 1% BSA in TBS-T and detected with anti-MHC-II or anti-alpha tubulin overnight.

| RNA isolation and real-time PCR analysis (qPCR)
RNA isolation from mouse lung was performed via the peqGOLD . Primer sequences are given in Table S2 of the Data S1.

| Mouse in vitro assays
Mouse splenocytes were isolated and enriched for immune cells with For OT-II co-culture, BM was harvested from OT-II-mice and dif-

| Human in vitro assays
Peripheral blood was collected by venipuncture from 7 donors with self-reported HDM-allergy and confirmed HDM-specific IgE, as well as 8 healthy controls according to a protocol approved by the Ethics Committee of the Medical University of Graz (17-291 ex 05/06).
Control blood donors reported no allergy-related complaints (details are included in Table S3). Written and informed consent was obtained from all patients included in this study. Oral cortical steroid (OCS) users were excluded, and inhaled cortical steroid (ICS) use was abstained for 24 h prior to blood sampling. In brief, blood cells and plasma were separated by centrifugation, erythrocytes were removed by dextran sedimentation and PBMC (buffy coat) were separated from PMNLs (pellet) by density gradient centrifugation using Histopaque 1077. 20 PBMCs were treated like mouse splenocytes, except HDM was used at 25, 50 and 100 µg/ml and cells incubated for 2, 5 or 8 days, as indicated.

| Statistics
Data are presented as individual data points and box and Tukey whisker plots or mean +/-SEM using R statistical environment Statistical significance is indicated in the figures as follows; *p < .05, **p < .01, ***p < .001.

| Hypoxia decreases airway remodelling and hyper-responsiveness
In order to investigate the direct effects of hypoxia on the progression of experimental asthma, mice were first sensitized and challenged with HDM, then randomly divided into two groups, one at room air (21%) and one at 10% O 2 ( Figure 1A). HDM sensitization was confirmed by the presence of HDM-specific immunoglobulins in the serum ( Figure 1B). As expected, 2 weeks of exposure to 10% While expression of the goblet cell differentiation factor Spdef was slightly decreased at 10% O 2, the difference to 21% O 2 was not significant ( Figure 1E). Analysis of lung function revealed pronounced AHR in normoxic HDM-treated mice while Hyp-HDM mice exhibited significantly less AHR, as shown by attenuated airway resistance, compliance and elastance ( Figure 1F). Interestingly, Hyp-mice exhibited increased inspiratory capacity ( Figure 1G), which is indicative of reduced air trapping. There was a significant interaction between HDM treatment and Hypoxia (p for interaction <.001), suggesting that hypoxia specifically inhibited the asthmatic response to HDM.

| Hypoxia reduces inflammation upon HDM treatment
We and IL-17 upon HDM-stimulation, which again was almost completely suppressed in HDM-10% mice ( Figure 2E).

| Induction of antigen-specific adaptive immune response in vitro is critically dependent on oxygen
In order to examine the reasons for dampened immune response, we

| Up-regulation of MHC-II in APCs of the lung is impaired by hypoxia
In order to verify our in vitro observations, we returned to our in vivo were also confirmed via Western blot of total lung tissue ( Figure 4B).

| Human adaptive immune response is dependent on oxygen level
In order to confirm our findings in a human system, we repeated our in vitro experiments with human PBMCs collected from HDM-allergic and non-allergic donors. Total IgE and HDM-specific IgE levels are shown in Figure 5A. As expected, PBMCs from allergic humans showed a robust adaptive immune response to HDM. PBMCs from non-allergic donors also reacted to HDM, but to a far lesser extent ( Figure 5B). In line with our mouse data, human adaptive immune (both T-and B-cell) response to HDM was strongly suppressed by hypoxia ( Figure 5C). Again, the polyclonal T-cell response to direct stimulation via aCD3/CD28-beads was unaffected. Interestingly, the B-cell response upon aCD3/28 was also reduced under hypoxia; since aCD3/28 acts on the T cells, which in turn activate B cells, again suggesting that hypoxia interferes with cell-cell communication. sults support previous studies, where similar effects were observed in asthma patients when moved from lower altitude to high altitude. 23,24 Historically, three main hypotheses were held as to the mechanism underlying HACT: Allergen avoidance, 11 increased UVradiation 24 or avoidance of psychosomatic triggers. 14 Our data suggest that reduced oxygen concentration is a very important factor explaining at least some of the beneficial effects of HACT. Hypoxia has been investigated in the context of asthma before; one study found intermittent hypoxia to aggravate asthma, 25 while another study reported the opposite. 26 Older experimental models used intermittent hypoxia, resembling conditions that occur in patients in sleep apnoea. In contrast to previous studies, our model used sustained hypoxia, resembling the situation of high-altitude therapy.
However, it should be noted that our study uses severe hypoxia, which differs from the mild hypoxia that patients would experience in a HACT-clinic. Nevertheless, our study identifies several principle hypoxia-driven molecular pathways that may play a role in HACT. While previous studies speculated about the role of APCs in the conveyance of the positive effect of HACT, most studies have only used peripheral blood samples to explain the effects. 24,28 In recent years, however, we have learned that there are striking differences between circulating and tissue-resident immune cells. 29 Therefore, we must investigate the actual immune status of the lung. of Hif1α signalling has a potent anti-inflammatory effect 35 ; however, it also renders mice susceptible towards bacterial inflammation and impairs wound healing. 36 The current state of knowledge depicts oxygen conditions as a key regulator of macrophage function, but how exactly hypoxia affects macrophages, and how this in turn affects the outcome of inflammatory disease, is strongly dependent on the context. Much more research is needed to decipher the complex interplay of macrophage subtype, mode of activation, oxygen levels and additional environmental influences such as cytokines, nutrients, mechanical pressure or others.
Our in vivo data show that lung APCs are strongly affected by hypoxia, failing to upregulate MHC-II and CD80 upon HDM challenge. Furthermore, supporting previous findings, 37 our data show that hypoxia affects the development of APCs from BM precursors, both reducing total amount but also causing a shift in generated subsets of BM-APCs. Finally, we could show that APCs generated under hypoxia fail to initiate a T-cell response, irrespectively of oxygen conditions under stimulation or T-cell-co-culture. Therefore, we hypothesize that hypoxia during differentiation of APCs causes a 'hypoxic imprinting', and thereby a lasting reduction in their ability to initiate adaptive immune responses. The exact genetic or epigenetic mechanisms governing this imprinting will still need to be discovered. Our observations might offer an explanation as to why the positive effect of HACT persists after the patient has returned from the high-altitude stay.
Even under steady-state conditions, APCs permanently leave their original tissue to migrate to the next draining lymph node.
Under inflammatory conditions, this migration is massively accelerated. As a consequence, the tissue is re-populated with APCs; this happens partially through multiplication of remaining APCs in situ, partially through immigration and differentiation of blood monocytes/BM-derived precursor cells (reviewed in 38 ). We propose that a stay at lower oxygen conditions alters the differentiation of repopulating APCs and therefore changes composition and function of the tissue-resident APCs. This, in turn, results in an altered immune response when the tissue is again exposed to antigen. After exposure to higher oxygen, these 'hypoxically imprinted tissue APCs' would persist for weeks or months, until finally again being replaced by normoxically generated APCs.
The immune system is incredibly complex; its abilities are dependent on a multitude of cell types to differentiate and interact with each other, as well as with the surrounding tissue and eventual pathogens. Therefore, an immune response, particularly one underlying a pathological condition, can only be understood in an in vivo setting. We suggest our model for further investigations of hypoxia effects on allergic asthma. Especially the 'hypoxic imprinting' of APCs needs to be studied in further detail and in vivo. Further research into the hypoxia-mediated effects on the immune cells and their crosstalk may lead to durable effects on allergic asthma that can be attained by small interfering molecules or biologicals.
Our study has some limitations: First, our principle findings are based upon mouse experiments; while our data very well recapitulated the human situation, some details are not directly translatable.
The perfect way to test translatability would be lung biopsies from patients before, at the end and sometime after HACT to study APCs in the tissue. Obtaining such samples, however, will be a challenge for itself.
Next, our in vivo experiments were performed at 10% oxygen; outside of a laboratory, these conditions are found at altitudes of 5500 metres. However, most patients undergoing HACT reside at 2500 metres altitude or below. However, our study provides excellent proof of concept new data, a novel hypothesis and possible molecular mechanisms to investigate.
Numerous studies have described a positive effect of HACT on humans; however, altitude of the treatment facilities and duration spent there varies. Whether a longer stay at moderate altitude confers benefits comparable to a stay at higher altitudes and what is the minimum time and altitude to robustly confer a longer lasting effect still needs to be determined. This paper introduces an in vivo model to investigate the impact of hypoxia on adaptive immunity; we hope that similar protocols as ours will be of use to further unravel the molecular mechanisms governing this immune-dampening effect. We propose to use our model to further dissect monocyte-to-APC differentiation and cell migration/retention in the lung. Further research in this direction is necessary to deepen our understanding, but hardly possible in humans. Ultimately, we hope that our work contributes to a deeper understanding of adaptive immune modulation and possibly reveals new therapeutic targets and treatment strategies for immune driven diseases.

ACK N OWLED G EM ENTS
We would like to thank Sabrina Reinisch, Camilla Götz and Thomas Fuchs for their excellent technical assistance; and Slaven Crnkovic and Chandran Nagaraj for helpful discussions.