that were designed to target the host also affect the microbiome. Accordingly, inhaled corticosteroids and proton pump inhibitors affect the lung microbiota [115–117], as well as subsequent pneumonia risk [118, 119]. Macrolide antibiotics, broadly effective across chronic lung conditions such as COPD [120], affect both lung microbiota and host immunity. Perhaps most provocatively, baseline differences in lung microbiota appear to predict patient responsiveness to therapies such as inhaled corticosteroids [10], suggesting that variation in lung microbiota may represent an untapped phenotype of “precision medicine” in the lung. This opens new possibilities to exploit this important cross-talk in therapeutic interventions, but in order to do so, we first need to improve our understanding of the molecular mechanisms.

Increasing evidence of the importance of the microbiome raises the concept of restoring “diseased” microbiomes to prevent or treat diseases using microbiota-directed therapies or host-targeted therapies, such as probiotics, metabolites, lung microbiota transplantation or vitamin D therapy, which we discuss here in more detail.

Many clinical studies have investigated the efficacy of probiotic bacteria, which are supposedly beneficial for the host, to prevent chronic diseases such as asthma or allergic rhinitis. However, these studies have largely produced contradictory outcomes [121], which might be due to the fact that the used probiotics were not selected based on potential mechanistic effects and may not be ideal. The microbiome is a complex ecosystem comprised of a variety of different inhabitants and influenced by many external (host-derived) factors; thus, the addition of single strains may not make a profound difference. Thus, the transfer of whole microbiomes via faecal microbiota transplantation (FMT) could be a more promising approach. The introduction of healthy microbiota into diseased hosts has restored immunity and physiology [122], demonstrating that intestinal microbiota and their products can modulate host immunity locally and systemically, and that FMT can replace disease-related microbiomes with healthy ones [40, 107]. FMT is remarkably (90%) successful in treating antibiotic-resistant Clostridium difficile-induced colitis [123], and is now being used as treatment in selected patients [124, 125]. The new microbiome engrafts quickly and lasts for at least a month, indicating a potential difficulty in inducing long-term beneficial changes in the microbiome via only targeting the microbial side. While there are encouraging data, questions remain whether FMT may also affect lung health, and whether lung microbiota transplantation is feasible. Furthermore, recently, severe complications of FMT have been reported due to transfer of drug-resistant bacteria [126]. Thus, it is essential to determine whether such approaches, that so far only transiently change the microbiome, can be used for the required long-term treatments of chronic (lung) diseases that coincide with a variety of structural changes, aberrant mucociliary clearance, and many more [40].

Along with living bacteria, specific microbial molecules such as lipopolysaccharide (LPS) and peptidoglycan can induce or modulate inflammatory responses [127–129]. In addition, culture supernatants of probiotic bacteria display anti-inflammatory effects, which have been ascribed to the presence of secreted immune-modulatory metabolites [130]. For example, culture supernatants of certain probiotic Bifidobacterium species decreased the secretion of type 2 cytokines from immune cell lines and the expression of costimulatory molecules on primary dendritic cells [131]. A likely mechanism is quorum sensing. Quorum sensing is a means of communication among bacteria of the same species to coordinate effector functions such as biofilm formation, sporulation or toxin secretion. The best-described quorum sensing molecules are acyl homoserine lactones (AHLs) [132]. Several AHLs are targeted by the host to interfere with growth of pathogens [133], and are in turn exploited by bacteria to regulate host gene expression for their benefit. Some AHLs can bind to distinct bitter taste receptors expressed on the airway epithelium [134] and innate and adaptive immune cells [135, 136], thereby modulating barrier and immune functions. The most-studied AHL, 3-O-C12-HSL, can activate phagocytes to increase phagocytosis, expression of adhesion receptors and chemotaxis [137, 138], but is itself cleaved and inactivated by airway epithelial cells [133].

Another example of bacterial-derived modulators of the host's immune responses are outer membrane vesicles (OMVs) [139]. OMVs are spherical bilayered membrane vesicles released from the surface of both Gram-negative and Gram-positive bacteria, and contain much of the biological material from the parent bacterium, but in a nonreplicative form [140, 141]. Evidence suggests that the release of OMVs provides bacteria with competitive advantages when exposed to acute and chronic host-associated stressors [142]. They may protect bacteria against innate and adaptive immune responses [143–145], antimicrobial peptides and antibiotics [146, 147]. Moreover, OMVs contain factors (e.g. siderophores) aiding in the acquisition of nutrients in an environment devoid of crucial elements such as iron [147]. Besides supporting the survival of the parent bacteria, OMVs may also play role in the progression of pulmonary diseases. Bacteria frequently associated with COPD exacerbations are known to release OMVs [148]. Furthermore, macrophages stimulated with OMVs derived from prominent airway pathogens such as P. aeruginosa, H. influenzae or M. catarrhalis release higher amounts of tumour necrosis factor-α and IL-6 [148]. Legionella-derived OMVs significantly enhanced bacterial replication in macrophages [149], and bacteria-free P. aeruginosa OMVs have been shown to potently induce pulmonary inflammation in mice [150], strengthening the idea that OMVs exert disease-promoting activities. In addition, OMVs have been shown to induce tolerance and hyporesponsiveness, thereby facilitating bacterial adherence to and internalisation by macrophages, which may contribute to clearance of the infection [149, 151]. Thus, despite our increasing knowledge on OMVs and their potential role in interkingdom communication, there is a need for further research to better understand their pathogenic properties and possible therapeutic or prophylactic implications (e.g. novel vaccines).

An interesting alternative to FMT or probiotic bacteria might be to use immune-modulatory microbial metabolites or “beneficial” OMVs. Such chemically defined bacterial substances could be produced at large scale under controlled conditions, applied in defined (effective) doses both systemically or locally, and may have fewer adverse side-effects, i.e. in immunocompromised patients, compared to live bacteria [152]. Several studies have already used defined bacterial metabolites to treat AAD in mice. LPS from Escherichia coli O111 [153], bacterial polysaccharide A [154], oligodeoxynucleotides with bacterial CpG motifs [155], flagellin B [156], short-chain fatty acids [47], D-tryptophan [157] and the neutrophil-activating protein from Helicobacter pylori [158] all suppressed airway eosinophilia and type 2 T-helper cell-mediated immune responses in murine models of AAD. Likewise, several molecules from the parasite helminths have now been identified, among which excretory–secretory (ES)-62 and cystatin (AvCystatin) from Acanthocheilonema vitae, TGF-β mimic (TGM) and Heligmosomoides polygyrus alarmin release inhibitor (HpARI) from H. polygyrus and IL-4-inducing principle from schistosome eggs (IPSE) are the most promising, with effects in allergic asthma models [159–163].

Nevertheless, there are significant limitations concerning the use of bacterial or parasite metabolites for therapeutic or preventative approaches. These molecules have a very narrow window of beneficial and therapeutic doses, which might be difficult to control in local tissues. Furthermore, the detailed mechanisms of action remain largely unknown, hampering translation to human studies. And finally, single metabolites, just as single bacterial strains, might not be sufficiently effective to treat or prevent complex chronic diseases with highly aberrant gut and lung microenvironments. Thus, further studies are urgently needed to determine the mode of action, but also the efficacy of single or combined metabolites to develop novel metabolite-based treatment options.