Influence of the gut and airway microbiome on asthma development and disease

There are ample data to suggest that early‐life dysbiosis of both the gut and/or airway microbiome can predispose a child to develop along a trajectory toward asthma. Although individual studies show clear associations between dysbiosis and asthma development, it is less clear what (collection of) bacterial species is mechanistically responsible for the observed effects. This is partly due to issues related to the asthma diagnosis and the broad spectrum of anatomical sites, sample techniques, and analysis protocols that are used in different studies. Moreover, there is limited attention for potential differences in the genetics of individuals that would affect the outcome of the interaction between the environment and that individual. Despite these challenges, the first bacterial components were identified that are able to affect the transcriptional state of human cells, ergo the immune system. Such molecules could in the future be the basis for intervention studies that are now (necessarily) restricted to a limited number of bacterial species. For this transition, it might be prudent to develop an ex vivo human model of a local mucosal immune system to better and safer explore the impact of such molecules. With this approach, we might move beyond association toward understanding of causality.


| BACKG ROU N D
An ounce of prevention is worth a pound of cure.The first 1000 days of life, between the start of a woman's pregnancy and her child's second birthday, are a unique window where the foundations for a future healthy development are established. 1Epidemiological data of risk factors for asthma suggest that the above applies particular well for asthma development. 2,3e Hygiene Hypothesis, formulated in the 1980s, originally described that the reduced exposure to infections in first years of life improved not only children's health and survival but was also responsible for the increased prevalence of allergic diseases in the Western world.The modern equivalent of the hygiene hypothesis is more nuanced and considers that the reduced exposure to common bacteria (and helminths) is more relevant.These "old friends" have been part of our natural environment throughout evolution so that changes in the exposure will affect the composition of the microbiomes of the gut and the airways.These changes, together with an increased pollution is now held responsible, not only for the increase in allergic Th2-mediated diseases, but among others even for the increase in sis protocols that are used in different studies.Moreover, there is limited attention for potential differences in the genetics of individuals that would affect the outcome of the interaction between the environment and that individual.Despite these challenges, the first bacterial components were identified that are able to affect the transcriptional state of human cells, ergo the immune system.Such molecules could in the future be the basis for intervention studies that are now (necessarily) restricted to a limited number of bacterial species.For this transition, it might be prudent to develop an ex vivo human model of a local mucosal immune system to better and safer explore the impact of such molecules.With this approach, we might move beyond association toward understanding of causality.

K E Y W O R D S
asthma/airway microbiome, asthma/gut microbiome, asthma/immunology, asthma/ microbiome, microbiome/immunology Th1-mediated autoimmune diseases. 4,5Although (changes in) environmental exposures are important, these cannot be the sole factor influencing asthma development.Since children of asthmatic parents are more likely to become asthmatic themselves, 6 it is reasonable to assume that genetic factors contribute to disease development.The genetic background will affect the immune interaction with the environment.So, even under similar environmental conditions, the healthrelated outcome of this interaction could differ between individuals.
A better understanding of these early interactions will allow us to explore how early-life interventions can help reduce or prevent the burden caused by asthma.To transition from bacterial species that associate with either healthy or asthmatic development to causality, it would be important to understand asthma and its pathogenesis, the (influence of the) microbiome, host immunity, and their interaction.
In this position paper, we summarize the current knowledge and understanding of the influence of the gut and airway bacterial microbiome on asthma and suggest directions for future research.

| G ENER AL CON CEP TS IN MI CROB I OME ANALYS IS S PECIFI C TO A S THMA RE S E ARCH
In asthma, at least part of the observed microbiome differences between the diseased and the healthy state may be held responsible for the disease, 7 as opposed to an asthma state that creates an environment in which certain bacterial species can thrive.Many environmental and lifestyle factors have been associated with both a decreased microbial exposure and the development of asthma. 8,9ese include formula feeding, cesarean section, having no elder siblings, no day-care attendance, and antibiotics use. 10 Moreover, an increased asthma prevalence has been linked to an affluent lifestyle, an urban versus country lifestyle and migration to industrialized countries, especially when this happens in infancy. 3,11,12oving beyond associations toward elucidating causality and using ex vivo models may pave the way for targeted interventions tailored to individual microbiome profiles and genetic makeup, offering new avenues for asthma prevention and treatment strategies.

G R A P H I C A L A B S T R A C T
There is an association between microbial dysbiosis, in the gut and airways, and asthma development.Microbial interventions may prevent asthma development or contribute to its cure.
These aspects suggest a causal role for environmental exposures in asthma development.This notion is further supported by observations in mouse studies, with OVA-induced asthma.Researchers showed that perinatal vancomycin treatment reduced microbial diversity, changed the microbial composition, and exacerbated disease severity.This effect was not seen when vancomycin was given to adult mice. 13In another study, germ-free mice were more prone to develop the asthma phenotype. 14This could be prevented by neonatal colonization with "healthy" gut microbiota. 15[15]

| Study design and outcome measurements in microbiome research
Modern molecular techniques allow the identification of bacterial species in samples based on DNA sequencing, specific amplification of bacterial ribosomal 16S clusters, or interspace regions that are unique for bacterial genera or species. 16,17These techniques have limitations, such as different techniques for DNA isolation and sequencing, and different bacterial libraries exist to translate your sequences.These differences can affect outcomes and make comparisons between studies challenging, 18,19 hampering translation to clinical practice.
Most studies compare microbiome composition between disease states in cross-sectional design or try to predict disease development in a longitudinal design.Microbiome composition can be analyzed in terms of abundance (the total number of species) and/or diversity (the relative composition) of the microbiome, and this can be done from species to phylum level.Different types of diversity are used as follows: alpha diversity (α-diversity), the within sample diversity, and beta diversity (β-diversity), the between-sample diversity (Figure 1).In most research, relative abundances are used, instead of total bacterial burden, this can lead to a high false discovery rate, which is not completely solved by sophisticated statistical methods. 20For diversity, many different metrics are used throughout different studies, and these metrics can influence outcomes. 21In general, the healthy state is associated with a stable core microbiota (abundance of different phyla, genera, and species) and a high diversity. 22For asthma, this generalization might be less clear.

| Limitations in asthma diagnosis as a reference standard
Asthma in (young) adults is a collection of heterogeneous diseases with different phenotypes and different underlying pathophysiologic mechanisms. 2,23,24The phenotypes include (not exhaustively): allergic asthma, intrinsic asthma, neutrophilic asthma, obesity-related asthma, (pediatric) early-onset asthma, and lateonset asthma. 24,25However, there is no proper clinical asthma diagnosis in children under the age of five, because these children are often unable to perform a reliable lung function test. 2 As a consequence, there is no standard "asthma" diagnosis in young children.Between different studies, "asthma" can be defined as (not exhaustively): (i) parent reported, (ii) doctor's diagnosis, (iii) prescription of corticosteroid inhalation medication, or (iv) multiple wheezing episodes.This wide diagnosis range can make comparison between studies difficult and also highlights the problem of linking these surrogate diagnoses in children to correct asthma diagnosis in adults.An example: In cases where (recurrent) wheezing at a young age is used as a surrogate, only 35% of these infants are diagnosed with asthma later in life. 26The heterogeneity of asthma poses another challenge in children.Early-onset asthma is often allergic asthma, mediated by a Th2 response, but exploration of the other endotypes is limited.Th17 and neutrophilic asthma have been reported in children, but its temporal stability is poorly understood and requires invasive measurements. 27The asthma endotype could significantly impact the interaction with the microbiome, which is often overlooked in current research.

| Influence of gut and airway microbiome on asthma development
Asthma is by definition a disease of the lower airway, yet historically, the gut microbiome was the first studied in relationship to F I G U R E 1 Explanation of alfa and beta diversity measurements.asthma development.The ease of sampling of the gut microbiome and other diseases researched in context to the gut microbiome probably played a significant role in this choice.Although the symptoms of asthma are associated with the lower airways, one should realize that asthma is a systemic disease.In allergic asthma for instance, there is clear progression from early childhood into adulthood that has been named the atopic march. 28This progression begins with a food allergy (egg and/or milk), progressing to atopic dermatitis, followed by allergic rhinitis, and finally allergic asthma. 28,29In this case, a direct involvement of the gut, the skin, the upper, and the lower airways is seen, respectively.In addition to this link in time, there are also direct mechanistic links between the lower airways and other anatomical sites.These links include, the gut-lung axis 30 and the concept of the united airways. 31,32e gut-lung axis describes that microbes in both the gut and the lower airways influence each other, via local immune modulation and the mesenteric lymphatic pathway. 30The concept of the united airways centers around the notion that in allergy, infection, inflammation, and asthma, there is a pathological continuum due to interaction between the upper and lower airways. 31,32Research has shown that triggers and interventions in the upper airways can modulate asthma severity. 33,34lated to the microbiome, attention shifted from the gut to the lower airways upon discovery that the lower airways are not sterile as, previously believed. 35However, particularly in young children, sampling the lower airways poses greater challenges than the upper airways.Analyzing the airway microbiome presents its own challenges due to the plethora of locations (nasal cavity, naso-, oro-and hypopharynx, trachea, and bronchi) and diverse sampling techniques (brush, swab, nasal wash, induced sputum, or bronchial alveolar lavage).These variations in sampling methods may inadvertently introduce differences in the microbiome composition.[37][38] Another challenge in airway microbiome research pertains to temporal changes (it could be argued that this is also important for the gut microbiome).0][41][42] Considering the sampling moment concerning age, time of year and in between infections becomes crucial, especially when comparing across different clinical states.

| G UT MI CROB I OME AND A S THMA
In Table 1, we highlighted the data from 12 longitudinal studies examining the microbiome in stool samples of infants (0-3 years) and its correlation with an asthma diagnosis later in infancy (1-11 years).
With their longitudinal design, these studies aimed to predict asthma development based on microbiome composition at an earlier age.4][45] The observed variations likely result, in part, from differences in study design, including time of sampling, asthma diagnosis, and sequencing techniques.
Of the 12 studies discussed in this article, 11 reported α-diversity.7][48][49] Conversely, one study, while reporting no difference in Shannon diversity, identified an association between a higher microbial maturity at 5 weeks and an increased risk of asthma development. 502][53][54][55][56] Only four out of the 12 reported ß-diversity.Two studies reported a difference in ß-diversity in the gut microbiome at age 1 or 3 between children that became asthmatics later in infancy and healthy children. 52,55The remaining two studies found no differences. 53,5756,57 However, while results are consistent for certain genera, conflicting results emerge for others.For instance, a low relative abundance of Dialister in infants was associated with asthma development in school-aged children in one study, 50 whereas another study (in the same age group) showed a higher relative abundance. 48Differences also arise concerning Ruminococcus and Bifidobacterium, with one study that associated a higher relative abundance of both genera with asthma development at school age, 53 while two other studies reported a lower relative abundance of Ruminococcus associated with asthma development at school age. 46,48Yet, another study found that children with the highest asthma risk had a lower relative abundance of Bifidobacterium. 57Two studies in the same research group, different cohort, show conflicting results for Veillonella.The study in the CHILD cohort, with a wheeze diagnosis at age 1, showed a lower relative abundance of Veillonella. 56ile the study in Ecuador, with recurrent wheeze at age 5, showed a higher relative abundance of Veillonella. 53A study in another cohort, with an asthma diagnosis at age 5, also found a higher relative abundance of Veillonella. 52

| AIRWAY MI CROB I OME AND A S THMA
Research on the potential role of the airway microbiome in asthma development is emerging.However, due to variations in sampling locations, techniques, and time points, generalizing outcomes from these studies has proven challenging. 58,59Akin the studies of the gut microbiome, there are six studies of the upper airways with a longitudinal design that evaluate the microbiome in infants (0-2 years) and correlate this with an asthma diagnosis later in infancy (2-18 years) 39,40,[60][61][62][63] (Table 2).Three of the six studies in the airways, reported diversity.One study in hypopharyngeal aspirates described a higher Shannon TA B L E 1 Schematic overview of articles discussing the gut microbiome studies in association to asthma.Note: Order year of publication.

TA B L E 1 (Continued)
TA B L E 2 Schematic overview of articles discussing the airway microbiome studies in association to asthma.

Source Age at sample collection.
Location and technique.index, richness, and difference in ß-diversity in 1-month-old children, who became asthmatic at age six. 39The other two studies in nasal 40 or oropharyngeal swabs 62 found no differences in either αor ß-diversity.

Relative abundance
Concerning relative abundance, the interpretations suffer from the same issues as for diversity in the section above.A limited number of studies, different sampling time points, and different anatomical sites will affect the outcome and, as a consequence, consistency.A higher relative abundance of Haemophilus 40,60 and Streptococcus 61,63 in the upper airways of infants has been associated with wheeze or asthma later in infancy.For Prevotella, results are less consistent.In one study with nasal swabs, a lower relative abundance of Prevotella was associated with wheeze at age 2. 62 While, in a study with hypopharyngeal aspirates, a higher relative abundance of Prevotella at 1 month was associated with asthma at age 6. 39 Data for Moraxella also seem inconsistent.Two studies with nasal swabs found a higher relative abundance associated with asthma development. 60,63ereas, in another study with nasal swabs, a Moraxella sparsity profile was associated with asthma at school age. 40Veillonella, which is often associated with asthma in the gut of infants, was also higher in hypopharyngeal aspirates at 1 month for children that became asthmatic at age 6. 39

| P OTENTIAL MECHANIS MS BY WHICH A MI CROB I OME MI G HT AFFEC T A S THMA DE VELOPMENT
Secreted metabolites from bacteria have the capacity to influence the transcriptional state of human (immune) cells, 64 consequently impacting their differentiation and function. 9,14In this way, they might affect asthma development.
One class of bacterial metabolites that is considered as disease modifiers through affecting the epigenome in human cells are short chain fatty acids (SCFAs). 65These molecules are effective inhibitors of histone deacetylases, an important group of enzymes whose action leads to transcriptional inactivation of genomic regions through the local formation of heterochromatin. 66,67Inhibition of deacetylases shifts the equilibrium toward a higher state of acetylation of histones and (re)activation of genes within heterochromatin.Several reviews have highlighted the importance of higher levels of butyrate and other SCFAs in protection against asthma development. 59,65,68A study in infants showed an inverse relation between levels of butyrate (and the abundance of butyrate producing bacteria) with asthma development. 46The increase in asthma in children treated with antibiotics in infancy, is attenuated by breastfeeding.The proposed mechanism involves that breastfeeding leads to an increase of Bifidobacterium infantis in the child's gut microbiome, 69 a species that is highly efficient in metabolizing breastmilk to SCFAs. 70Indeed, it has been shown that B. infantis is most abundant in the gut of breastfed infants (40%-80% of the total gut microbiota). 70Furthermore, supplementation with B. infantis in breastfed infants, silenced the intestinal allergic Th2 and Th17 in favor of a Th1 response. 71The direct influence of SCFAs on the (mammalian) immune system has also been investigated in mice, where multiple studies have shown that oral administration of SCFAs reduced the severity of allergic airway inflammation. 68,72 addition to the potential role of SCFAs, lipopolysaccharide (LPS), a bacterial cell wall component, was also considered as disease modifier.4][75] The effect of LPS seems to be modified by cysteine proteases sometimes present in farm and house dust.
In the presence of these proteases, LPS prevents Th2-cell-allergic responses, while in the absence, it induces Th2-cell-allergic responses. 76Another strategy of gram-negative bacteria to influence the interaction with their host is to change their lipid A structure, by genetic mutations and regulation of gene expression, which can make them more or less virulent. 77This ambiguous effect of LPS might also be why in different studies LPS positive bacteria (Dialister, Veillonella, and Prevotella) show either an association with asthma symptoms and development or with healthy development. 35,39,48,50though SCFA-and LPS-related mechanisms offer important insights into what mechanisms are available to influence immune reactions, they would not explain why only certain species have been linked to asthma development.SCFAs and LPS are ubiquitously produced or present in multiple bacterial species, so from this perspective, host genetics should play a role to explain any specificity in the interaction between the microbiome and the host's immune system.Alternatively, one needs to consider a specific type of metabolite produced by only a limited number of species.An interesting option in this respect are the quorum sensing molecules produced by bacteria.Originally, these molecules were discovered as part of the mechanism by which a bacterium could sense the presence of other members of its species or class and through which they could orchestrate a collective behavior. 78,79ter, it was realized that these quorum sensing molecules are also used to actively compete between (or attack) other bacterial species, 80 thus affecting the composition of the microbiome.In a mouse model, a non-species-specific quorum sensing molecule autoinducer-2 (AI-2) produced by E. coli opposed the effect of streptomycin on the Firmicutes/Bacteroidetes ratio in the gut, by favoring the expansion of Firmicutes. 81Although not extensively studied within the context of asthma, these molecules also appear able to directly affect the transcriptional state of mammalian cells and the differentiation of T cells.Another study in mice showed that acyl homoserine lactones (AHLs), a quorum sensing molecule produced by P. aeruginosa, induced IgG1 and IgE production. 82An in vitro study with AHLs and Pseudomonas quinolone signal (PQS) showed that they both decreased the production of IL-12 in mice dendritic cells. 83These results demonstrate that quorum sensing molecules could be an interesting area for further research, as they might be a potential mechanism by which the abundance and diversity of a microbiome can affect asthma development.
These first glimpses offer an intriguing window into the future that probably will reveal even more active metabolites.Some bacteria even produce histamine themselves providing a more direct link with asthma. 84On the other side of the spectrum, bacteria might damage the epithelial barrier increasing exposure and thus indirectly affecting asthma development. 85,86Regardless of the mechanism, if a small molecule would be important, its application would certainly be simpler than exposing children to protective microbiota for the prevention of asthma.However, the latter option is not an impossibility, given the effectiveness of poop-transplants in the treatment of colitis ulcerosa. 87

| INTERVENTI ON S TUD IE S
Interventions have been used in two study designs: cross-sectional and longitudinal studies.In cross-sectional studies, probiotics or placebo were given to asthmatics, both children and adults, and asthma symptoms, quality of life, and medication use were evaluated.This design evaluates the effect of microbial interventions on asthma symptoms and disease severity.In longitudinal studies, probiotics or placebo were given to pregnant women in the last trimester and from birth to the newborn for some months.Later in life atopy, asthma and asthma symptoms are evaluated.This design evaluates whether microbial interventions can prevent asthma development.
9][90] A study in children from 6 to 18 years old who received L. paracasei, L. fermentum or the combination in a mix for 3 months had a lower asthma severity and higher asthma control score (ACT). 91Children in the same age group who received L. reuteri also had a higher ACT after 8 weeks of probiotic treatment. 92A study in which 3-14 years old asthmatics were supplemented with a mixture of Lactobacillus and Bifidobacterium for 16 weeks found a reduction in asthma exacerbations. 9399][100] In conclusion, the in vivo interventions described above only seem to influence asthma symptoms in children above 3 years and adults and do not prevent asthma development, when given during pregnancy and in newborns.This is contradictory to studies with germ-free mice where an asthma phenotype could be prevented in young mice, but not in elder mice by colonization with 'healthy' gut microbiota. 14,15There are many potential explanations for this Furthermore, it will be easier for a microbiome to populate a germfree gut, than to compete with a microbiome which already inhabits the gut as is the case in the human experiments.Also, the environment of the mice after their interventions was standardized, while the neonates potentially all had a very different microbial exposure in the years after their interventions, which could guide the immune system in either direction.It seems clear that despite the encouraging data form human intervention studies, we still have a long way to go and that the interaction between the microbiome and the hosts immune system needs further elucidation.

| FUTURE DIREC TIONS
To understand the interaction between the microbiome and host, it is essential to move beyond associations toward understanding causality.This knowledge can be used to develop intervention strategies for the prevention and treatment of asthma.Therefore, it is necessary to elucidate which bacteria at which anatomical site affect asthma and the mechanism by which this is achieved.This would allow researchers to use the active component, responsible for protection, rather than the (collection of) bacteria.Of course, it is important to standardize microbiome research, since sample site and technique, age and season of sampling, asthma diagnosis, and sequencing protocols can all impact a reliable comparison between clinical states and different studies.At this moment, however, it is far from clear which of these approaches would be better than others.
For microbial interventions, the gut and nose are convenient sites, as they provide an easy route of access.Furthermore, both the concept of the gut-lung axis and the united airways show that an intervention in the microbiome at a distal site can affect the pathogenesis of the lower airways.Fecal microbial transplants have shown great promise in the treatment of colitis ulcerosa, 87 but no data are available whether a similar approach would be useful in the prevention of asthma.An extension of this idea would be a nasal mucus transplant as a treatment for asthma.To reach this point, more mechanistic data on how the microbiome interacts with the local immune system needs to be collected.
Another issue that needs to be addressed is the design of studies where health and disease states are directly compared.Although this allows the identification of common factors, the underlying assumption that all individuals in a group respond in an identical fashion might not be true.When each individual has a "personalized" interaction with the environment, the grouping of individuals will result in a large variance of the observed data.This could mask potentially relevant outcomes.Identifying the composition of distinct clusters of microbiota (independent of the disease state) and mapping the diseased and healthy individuals to these clusters would be a way to mitigate this potential problem.Some studies have already focused on microbiome clusters.[42] The "personalized" interaction between microbiome and individual cannot solely depend on the particular microbiome cluster of that individual, but must also address the genetic makeup of the individual.Potentially, an ex vivo cellular model of the gut or airway could be used or developed for such an approach.Airway models, consisting of a co-culture of differentiated nasal epithelium with the relevant immune cells of the same child, would allow us to dissect the interaction of (parts) of the immune system with bacterial species, full bacterial microbiomes, fungal, and other environmental components.Although modeling the complete interaction might still be a step to far, with potential incompatible conditions required to study combinations of different triggers and cell types.The development of the CRISPR-CAS9 technology even allows to study the impact of genetic variations, found in vivo, on the interaction between the microbiome and the human immune system.These genetic variations could be defined biomarkers identifying children at risk.This can help identify (groups of) children at a young age, at risk to develop asthma at a later age.These children can then receive a personalized intervention using a specifically tailored microbiome and/or their active components to prevent asthma.On a population base, this could be incorporated into preventive care already in place for newborns.Dependent on the asthma risk, deduced from microbiome composition and genetics, a tailored intervention with the "right" microbiome, microbial metabolites, prebiotics, or probiotics can be given to prevent asthma development.
Epidemiological data in relation to asthma development underscores the importance of early-life microbial exposure in shaping immune responses.Despite clear hints on the role of certain bacterial species, genera, or phyla in asthma development, it has been hard to reach consensus.With different anatomical sites (gut, lower, and upper airways) involved and different analysis pipelines, this can hardly be unexpected.Perhaps, because of these challenges in study design and interpretation, understanding the mechanisms underlying this relationship holds promise for personalized interventions to prevent or mitigate asthma burden.
contradiction given our limited understanding of what would be relevant in man versus what would be relevant in mice.For instance, the bacteria used in the interventions in humans are only one or a few species together, while the intervention in mice is a whole microbiome.Moreover, not all human individuals will react in the same way to the same microbial intervention, because of their distinct and individual genetic makeup.Laboratory mice are genetically more similar and thus might react more similarly to an intervention.