Background
Vitamin D has been proposed as an important regulator of both inflammation [
1] and the microbiome [
2,
3]. Indeed, it is hypothesised that there is a protective role for vitamin D in regulating the gut microbiome, which may shape predisposition for the development of inflammatory autoimmune and allergic diseases [
2‐
4]. Vitamin D is commonly acquired through skin exposure to ultraviolet B radiation found in sunlight and through the diet. Circulating 25-hydroxyvitamin D (25(OH)D) is typically used as a measure of vitamin D status, although 1,25-dihydroxyvitamin D (1,25(OH)
2D) is the most active vitamin D metabolite.
Our understanding of the effects of vitamin D on the microbiome and its associations with reduced tissue inflammation is currently limited to the gastrointestinal tract. Colitis severity and bacterial numbers in the colons of mice were increased during vitamin D deficiency [
5]. Dietary-induced vitamin D deficiency altered the composition of the fecal microbiome of C57Bl/6 mice, which had increased relative quantities of
Bacteroidetes,
Firmicutes,
Actinobacteria, and
Gammaproteobacteria [
6]. These mice also had increased colonic injury in comparison to mice fed a vitamin D-sufficient diet, but exhibited signs of hypocalcemia [
6]. In other studies, the colons of 21-day-old vitamin D-deficient mice were enriched for
Bacteroides/
Prevotella colony forming units, although this difference disappeared with age [
7]. CYP27B1
−/− mice, which lack the expression of the 1α-hydroxylase enzyme, responsible for converting circulating 25(OH)D to active 1,25(OH)
2D, had an increased fecal burden of
Proteobacterium phylum (including
Helicobacteraceae species) in a colitis model [
8]. Treatment of CYP27B1
−/− mice with 1,25(OH)
2D (1.25 μg/100 g diet) suppressed colitis severity and
Helicobacteraceae numbers [
8]. Vitamin D supplementation (980 IU vitamin D
3/kg per week for 4 weeks) of 16 healthy young adults reduced the relative abundance of Gammaproteobacteria such as
Pseudomonas spp. and
Escherichia/Shigella spp., increasing bacterial richness in the upper gastrointestinal tract, but not in ileum, colon nor in stool samples [
9]. Changes in
Bacteroides spp. were identified in the stool samples of African American men (
n = 115) with prediabetes and hypovitaminosis D after weekly supplementation with a vitamin D analogue (ergocalciferol; 50,000 IU) [
10]. Together, these data suggest that dietary vitamin D can alter the gut microbiome, reducing the abundance of potentially pathogenic species with these changes linked to reduced gastrointestinal inflammation and injury.
Vitamin D may shape the microbiome through a number of interdependent mechanisms. Firstly, vitamin D regulates innate immune responses. Vitamin D can induce the expression of antimicrobial peptides and proteins, such as cathelicidins and ß-defensins, which are produced by monocytes, macrophages and epithelial cells in the skin and lung [
1,
3]. Vitamin D-deficient mice had reduced colonic expression of the antimicrobial protein, angiogenin-4, which was associated with increased bacterial load, and tissue inflammation [
5]. Such antimicrobials may directly kill microbiota or be involved in activating innate immune processes such as autophagy in macrophages, promoting the ingestion of microbes into phagolysosomes for neutralization [
11]. Vitamin D may promote tolerant adaptive immune responses by modulating the gut microbiome. As highlighted above, there is evidence linking the modified gut microbiome and increased gut inflammation of CYP27B1
−/− mice, with fewer tolerogenic CD103+ dendritic cell in lamina propria [
8]. These are cells responsible for shaping the regulatory T cell repertoire of the gut [
12]. Ongoing presentation of bacterial antigens by dendritic cells may be required for tolerance towards gut microbes, with an essential role for tolerogenic dendritic cells and T cells to limit inflammation [
13,
14]. Finally, vitamin D may up-keep epithelial integrity. Colonic epithelial cells from CYP27B1
−/− mice expressed reduced levels of the cell-to-cell adhesion protein, E-cadherin [
8]. Assa et al. [
6] demonstrated that vitamin D-deficient mice had reduced colonic epithelial barrier function, with increased permeability that associated with increased proinflammatory cytokine expression, signs of colitis and relative quantities of
Bacteroidetes, Firmicutes,
Actinobacteria and
Gammaproteobacteria in the intestinal microbiome.
In healthy human lungs, the bronchial tree holds approximately 2000 distinct bacterial genomes per cm
2 [
15]. There are diverse microbial communities, with
Bacteroidetes,
Proteobacteria and
Firmicutes the mostly commonly identified at the phylum level [
16]. Similarly, the healthy mouse lung microbiome is dominated by species of these phyla and also
Actinobacteria and
Cyanobacteria [
17]. The lung microbiome changes from infancy through to adulthood, with the ratio of
Bacteroidetes:Firmicutes/Gammproteobacteria increasing with age in specific-pathogen-free mice [
18]. However, dysregulation of the microbiome of the lungs and associated tissues could contribute towards respiratory inflammation.
Indeed, microbial communities shift in the respiratory tract during respiratory disease [
15,
19‐
23]. Hilty et al. [
15] demonstrated that pathogenic proteobacteria were more common in bronchial brushings of the left upper lung lobe of adult asthmatics (
n = 11) or patients with chronic obstructive pulmonary disease (COPD) (
n = 5) than controls (
n = 8), with similar findings in the bronchoalveolar lavage fluid (BALF) of asthmatic children (
n = 13). Lung tissue samples from patients with COPD (
n = 8) had increased bacteria of the
Lactobacillus genus with bacterial communities distinct to healthy controls (
n = 8), smokers without COPD (
n = 8) or patients with cystic fibrosis (
n = 8) [
20].
Lactobacillus,
Pseudomonas, and
Rickettsia species were enriched in endobronchial brush samples from patients with chronic persistent asthma (
n = 39) in comparison to healthy controls (
n = 19) [
21]. The bronchial brushings of patients with severe asthma (
n = 30) were enriched with
Actinobacteria and a member of the
Klebsiella genus in comparison to healthy controls (
n = 8) and mild-to-moderate asthmatics (
n = 41) [
22]. Together, these findings suggest that pathogenic bacteria are more common in the lungs and associated respiratory tissue of patients with respiratory disease, which may vary depending on the site examined.
Animal models offer the opportunity of dissecting various factors that change the lung microbiome to initiate and/or exacerbate inflammatory or allergic lung diseases. The effect of vitamin D deficiency on the lung microbiome has not been specifically investigated in vivo. We previously observed that inflammatory cell numbers in the BALF of vitamin D-deficient male mice with allergic airway disease were associated with increased bacterial load in the lungs [
24]. Low levels of circulating 25(OH)D were associated with increased numbers of eosinophils and neutrophils in BALF in male mice with allergic airway disease [
24]. Increased levels of BALF inflammation and microbial load in the lungs of initially vitamin D-deficient mice were reversed by dietary vitamin D supplementation [
24]. In the current study, we hypothesised that vitamin D deficiency would alter the lung microbiome of naïve mice, potentially contributing towards increased respiratory inflammation, prior to the initiation of allergic airway disease.
Discussion
We observed modest effects of dietary vitamin D on the bacterial composition of the lung microbiome. Serum 25(OH)D levels inversely correlated with the number of OTUs detected, and more specifically the presence of an unidentified Pseudomonas OTU in the lungs of naïve mice. These results suggest that vitamin D sufficiency limited the number of respiratory pathobionts (like Pseudomonas) rather than increasing the number of protective commensals in the lungs. We also observed a sex difference in mice fed the vitamin D-containing diet throughout the experiment, with differences observed largely limited to a single Acintobacter OTU. The induction of allergic airway disease altered the lung microbiome composition to a degree that far exceeded any of the more subtle effects of vitamin D deficiency. More external environmental microbes, such as Cyanobacteria and Chloroplast, were detected in mice with allergic airways disease as well as OTUs previously identified in the lung microbiome, such as Staphylococcus, Peptostreptococcaceae and Acinetobacter. BALF neutrophil numbers were increased in mice of both sexes by ~10-fold with the induction of OVA-induced allergic airway disease, while the effects of vitamin D deficiency on these cell numbers in BALF were modest.
There were sex-specific differences in the lung microbiomes of mice fed a vitamin D-containing diet throughout the experiment. The
Acinetobacter OTU, which accounted these effects, was distinct from the
Acinetobacter OTU modulated by OVA-induced allergic airway disease. The nature of the sex-specific effect of vitamin D on the
Acinetobacter OTUs differed, depending on whether mice were naïve or had OVA-induced allergic airway disease. We have previously shown sex-specific clustering of the lung microbiome in a different mouse strain (C57Bl/6) using denaturing gradient gel electrophoresis [
37].
Acinetobacter may be a common commensal of human [
38] and mouse lungs [
17]. This genus may colonise the lungs early in life and has been linked to protection from allergy [
39‐
41]. The switching of the direction of the sex-specific effects may reflect differences in the severity of OVA-induced allergic airway disease in male and female mice that are regulated by vitamin D, as previously observed [
24]. The sex-specific effects on the lung microbiome were not reproduced by supplementing initially deficient mice with vitamin D, supporting the hypothesis that colonization with
Acinetobacter species in the lungs occurs early in life.
As we have previously observed, 25(OH)D levels were reduced in male mice as compared to female mice fed a vitamin D-containing diet (see also [
24,
25,
42,
43]). This observation is explained by increased renal expression of 24-hydroxylase (CYP24A1), the enzyme responsible for breaking down active 1,25(OH)
2D, in male mice [
42]. Hormones influence other effects of vitamin D, with reports of functional synergies between 1,25(OH)
2D and 17-β-estradiol in T cells [
44,
45], and differential regulation of 1,25(OH)
2D-modulated pathways by testosterone in males and females [
45,
46]. Testosterone increased Foxp3 (regulatory gene) expression in T cells from women, but not men [
46] and 1,25(OH)
2D more potently induced CD4 + CD25 + Foxp3+ (regulatory) T cells from PBMCs of female than male subjects [
45], although we observed similar reductions in T
Reg cell percentages in the skin-draining lymph nodes of vitamin D-deficient male and female mice [
47]. These effects of estrogen and testosterone on immune cell function suggest that females may be more resilient to the effects of vitamin D deficiency than males.
The inverse relationship between lung
Pseudomonas and serum 25(OH)D could be related to increased levels of mBD2 in BALF of vitamin D-sufficient mice. mBD2 is a member of the defensin family of peptides, which have well-established killing effects on Gram-negative bacteria like
Pseudomonas. mBD2 exhibits significant protein sequence homology with human ß-defensins 1 and 2 [
48] and is expressed by a variety of epithelial cells (including tracheal) as well some immune cells (reviewed in [
49]). mBD2 expression is induced during infection with Gram-negative bacteria, their products (e.g. lipopolysaccharide), and various proinflammatory cytokines (e.g. tumour necrosis factor) [
48,
50]. Lipopolysaccharide may also drive the production of active 1,25(OH)
2D from circulating 25(OH)D, resulting in synthesis of ß-defensins [
51]. mBD2 is an immunoregulatory molecule, and its expression improves bacterial resistance by promoting proinflammatory cytokine expression through activation of toll-like receptor-4 [
50]. Using immunohistochemistry, other researchers have detected only minimal concentrations of mBD2 in the lungs of BALB/c mice with OVA-induced allergic airway disease [
52]. In addition, reduced levels of cathelin-related antimicrobial peptide (the mouse equivalent of cathelicidin) were detected in mice with OVA-induced allergic airway disease in comparison to non-sensitised mice [
53]. These findings suggest that OVA-induced allergic airway disease may reduce the expression of antimicrobials like mBD2 levels in the lungs.
Regulation of mBD2 by vitamin D may significantly inhibit specific bacteria like
Pseudomonas. Wu et al. demonstrated that mBD2 was required for host resistance against corneal infection with
Pseudomonas aeruginosa [
49]. These bacteria were more frequently detected in the sputum of vitamin D-deficient patients with bronchiestasis than -replete patients [
54]. However, in another study there was no difference in the incidence of
P. aeruginosa infection in children with cystic fibrosis that were vitamin D-deficient or -sufficient (25(OH)D > 30 μg/L) [
55]. Active 1,25(OH)
2D (10 nM) induced antimicrobial activity in bronchial epithelial cells against
P. aeruginosa [
56]. While we observed a significant inverse relationship between total OTUs or numbers of an unspecified
Pseudomonas OTU and serum 25(OH)D, there was no inverse association with mBD2 in BALF. It is likely that another explanation, such as compromised epithelial integrity [
6,
8], is responsible for the observed inverse relationship between serum 25(OH)D and lung OTUs.
In addition to
Pseudomonas, we identified microbes in the lungs of our naïve mice that were previously detected in the lungs of humans [
57] and other mice [
17] including OTUs of
Micrococcus,
Staphylococcus,
Cupriavidus and
Streptococcus. The microbiome dataset from the lung tissue of naive mice (in particular) was very sparse, with few OTUs found among all lung samples (~130–140 OTUs/mouse). We obtained lung samples directly from the thoracic cavity of euthanised specific-pathogen-free mice, limiting possible contamination from the upper respiratory tract and oral cavity. It is possible that the modest effects of dietary vitamin D were due to the scarcity of the collected DNA. However, we believe this is not likely, due to the rigorous use of negative controls to identify OTUs of contaminating bacteria [
33], our observations of association between circulating 25(OH)D levels and lung OTUs, and the significant microbial shift induced by the induction of allergic airway disease.
Methodological considerations are important, as we have previously shown differences in the microbial diversity of BALF and lung tissue [
17]. With the small biomass of bacteria expected in the lungs of specific-pathogen-free naïve mice [
58], there could also be problems associated with low yield DNA introduced through bacterial contamination of commercially purchased products (eg. in molecular-grade water or PCR reagents, [
34]). As described by Salter et al. [
34], reagents and extractions kits used for 16S amplicon library preparation can mask real biological effects and or increase false results. This is particularly a problem with lung samples as the bacterial yield is low, Therefore we analyzed data both with and without OTUs considered as contamination by sequences at low levels in our negative controls, allowing us to observe the sex-specific differences in mice fed only a diet containing vitamin D.
A limitation of the current study was that the naïve mice did not receive a placebo treatment (i.e. Alum bolus and nebulisation). We also did not examine the microbiome of other locations including the BALF [
15,
17], or the oro- [
59] or naso-pharynx [
60], which could limit comparisons with human studies. We could not discern cause and effect in our modeling: for example, with allergic airway disease did the induced inflammation change the microbiome (or vice versa)? A detailed time-course analysis and use of further control groups (i.e. Alum only) could help determine the cause and effect relationships in future studies.
Another consideration is the model of asthma used. This well-characterised model [
61] involved sensitisation of mice with a ‘low dose’ of the allergen OVA (1 μg) with Alum (0.2 mg), which caused methacholine-induced airway hyperresponsiveness, airway eosinophilia and neutrophilia, increases in BALF levels of IL-5, and circulating allergen-specific IgE and IgG [
24]. This model induces a phenotype similar to allergic asthma typified by a T helper type-2 (Th2) immune response, which is induced by a single well-defined allergen [
61]. There is no ideal animal model of allergic asthma [
62]; commonly known as allergic airway disease in mice. House dust mite (HDM) models offer an advantage of using a human allergen; however, HDM preparations can be contaminated with varying quantities of bacterial-derived lipopolysaccharide. The induction of neutrophilic and eosinophilic inflammation in HDM-induced allergic airway disease models are dependent on toll-like receptor-4 [
63], complicating the interpretation of the inflammatory response [
62], especially with regards to examining the lung microbiome. We have shown here that the induction of OVA-induced allergic airway disease significantly altered the lung microbiome and induced lung inflammation. We hypothesise that factors that modulate airway inflammation per se, including those induced by other allergens, such as OVA with Alum or HDM extract [
61], will also modulate the lung microbiome to perpetuate pulmonary inflammation and disease.
In previous studies, we have shown that the effects of vitamin D deficiency on airway inflammation in male mice were dependent on the dose of OVA and Alum used to sensitise mice [
24]. However, vitamin D deficiency did not significantly modify airway resistance, tissue elastance or damping in male mice with OVA-induced allergic airway disease induced by this low-dose sensitisation [
24]. In other studies we examined the lung function of naïve vitamin D-deficient and -replete BALB/c mice and observed increased airway resistance and tissue damping in female (but not male) vitamin D-deficient but otherwise naive mice [
43]. These observations suggest that any protective effect of vitamin D on lung function is limited to female mice, and combined with a trend for an inverse correlation between total OTUs in the lungs and serum 25(OH)D levels in female mice, may suggest that the lung microbiome could regulate lung function in a sex-dependent fashion. In addition, vitamin D may maintain optimal lung function by preventing airway remodeling through a process dependent on transforming growth factor-β [
43].
In addition to the sex-dependent effects of vitamin D deficiency on the severity of lung inflammation in mice with allergic airway disease [
24], we have previously shown that vitamin D deficiency increased the capacity of airway-draining lymph node cells from male and female mice to proliferate and produce Th2 cytokines [
24,
25]. The effects of deficiency were reversed by subsequent supplementation with dietary vitamin D
3 [
24]. Vitamin D deficiency increased the influx of lymphocytes into BALF in response to exposure to HDM; however, deficiency was protective and reduced airway smooth muscle mass and airway resistance induced by HDM [
65]. Increased OVA-specific IgE and IgG1 were detected in vitamin D-deficient female BALB/c mice following OVA/Alum sensitisation (without further respiratory challenge) [
66]. Increased lung eosinophil numbers and CD4 + T1ST2+ cells (Th2 cells), and reduced CD4 + IL-10+ (regulatory) cells were observed in young adult offspring born to vitamin D-deficient BALB/c dams following chronic intranasal instillation of HDM to offspring from 3 days of age [
67]. However, as observed in our studies with OVA-induced allergic airway disease, there was no effect of vitamin D deficiency on airway hyperresponsiveness [
67]. Collectively, these studies suggest that vitamin D deficiency promotes Th2 responses and the accumulation of eosinophils and neutrophils in the lungs of mice with allergic airway disease, without further compromising lung function.
We detected a significant negative relationship between circulating 25(OH)D and total OTUs detected in the lungs, suggestive of reduced bacterial diversity with increasing serum 25(OH)D. Others have noted that bacterial diversity is increased in healthy lungs when compared to diseased lungs [
23]. Patients with poorly controlled asthma (
n = 30) had reduced bacterial diversity and species richness, with increased
Haemophilus influenzae in sputum samples, particularly in younger males with increased neutrophils (
n = 7) [
23]. Changes in bacterial diversity were associated with oral corticosteroid intake, airway obstruction, and eosinophilia in lung lavage fluid [
21]. Conversely, bacterial diversity in bronchial epithelial brushings was positively correlated with bronchial hyperresponsiveness of adults with sub-optimally controlled asthma (
n = 65) [
19]. However, those with increased baseline diversity had greater improvements in bronchial hyperresponsiveness in response to clathriomycin treatment [
19].