Background
Recent advances in high-throughput bacterial sequencing, allowing the large-scale study of the microbial communities in humans and experimental models, led to the study of the bacteria composing this flora, since termed “microbiome” (or microbiota), as well as their pathologic alterations, termed dysbiosis (or dysbioses) [
1]. Dysbiosis has been shown to be responsible for altered immune responses against pathogens [
2]. The disruption of this microbiome-driven immune homeostasis by antibiotics is considered a mechanism for the deleterious effects of antibiotics beyond the gut, such as predisposition to allergic lung diseases [
3]. Non-culture-based methods have shown that the human microbiome includes bacterial communities in body compartments other than the gut, such as the lung. Therefore, the microbiome-driven immune cross-talk between the gut and the lung, or “gut-lung axis,” involves both gut and lung microbiota [
4]. However, these studies are based on models of antibiotic-induced dysbiosis that rely on very broad-spectrum antibiotics using associations of 4 to 5 antibiotics, many with systemic diffusion. Not only are such models far from mirroring clinical situations, but they also result in the form of dysbiosis with significant overall depletion [
5], and deplete both the gut and lung microbiota [
5,
6]. Therefore, it is difficult to decipher the respective roles of the lung and gut microbiota on lung immunity and its impact on lung infection.
We aim to determine the effects of antibiotic-induced gut dysbiosis, without lung dysbiosis, on outcomes in a non-lethal murine model of acute P. aeruginosa lung infection and the underlying immune alterations in order to manage side effects of antibiotics.
Discussion
Our results show that oral non-absorbable antibiotics induce gut dysbiosis, without lung dysbiosis, leading to a widely depressed lung immune cellular response responsible for worse outcomes of subsequent P. aeruginosa lung infection. These effects involve bone marrow progenitor depression and are susceptible to therapeutic immunomodulation by the hematopoietic cytokine Flt3-L as well as fecal microbial transplant.
Our study shares limitations inherent to the experimental model. Ours is a murine model that does not strictly mimic a clinical situation and cannot therefore be extrapolated to clinical practice. However, animal models allow studying complex interactions such as the gut-bone marrow-lung axis without the added complexities related to the heterogeneity of patients. Furthermore, our choice of oral non-absorbable antibiotics (vancomycin and colistin) while not representative of antimicrobial therapies administered to intensive care unit (ICU) patients, allowed us to differentiate the effects of antibiotics on the gut from the lung. Additionally, oral vancomycin and colistin are components of selective digestive decontamination (SDD), which is a relevant ICU situation. Technical limitations to microbiota analysis were a limited number of animals that was mitigated by very clear shifts and/or inversions observed. Last, animals were not randomized to study groups, while this may be a potential study bias; studied mice are a standard breeding lineage from large batches bred by an international supplier where breeding schemes help minimize the risk of systematic differences in baseline characteristics.
The first description linking altered gut microbiota and outcome of bacterial lung infection used germ-free mice infected with
Klebsiella pneumoniae [
8]. Several authors described afterward that the gut microbiota protects against respiratory infection by
Streptococcus pneumoniae,
K.
pneumoniae, and
P.
aeruginosa [
5,
6,
9]. However, these studies used very broad-spectrum antibiotic regimens, which lead to a significant depletion of overall bacterial 16srDNA in the gut. We observed a reduction from 10
6 to 10
2 gene copies/ng DNA [
5], which is closer to “germ-free-like” models than to clinically relevant antibiotic regimens. In comparison, our model’s gut dysbiosis was mainly due to shifts in phyla, not to extreme overall depletion. We observed depletion of only 1 log copies 16 s rDNA (supplemental figure
3) rather than 4 log copies 16 s rDNA [
5]. Therefore, our data suggest that eradicating some phyla and/or expansion of others is sufficient to alter the responses to lung infection independently of significant/complete overall gut microbiome depletion.
Conversely to lung dysbiosis associated with altered outcomes like asthma [
10], our treatment did not significantly modify the lung microbiota, suggesting other links between the observed gut dysbiosis and lung outcomes. Since we used non-absorbable oral antibiotics, this treatment had no direct effect on the lung microbiota. In contrast, this may have been the case in other dysbioses induced by systemically diffusing antibiotics, as in the study of Robak et al. in which lung microbiota depletion was confirmed [
5]. Therefore, lung dysbiosis is not a requirement for the deleterious effects of prior antibiotics, and lung dysbiosis may rather be another collateral damage when antibiotics diffuse systemically to the lung.
In previous studies in which antibiotic-induced gut dysbiosis was shown to worsen bacterial pneumonia outcomes, the study of lung immune responses was focused in scope. Indeed, assessment of lung immunity restricted to macrophages showed that macrophage function was impaired consecutively to antibiotic-induced gut dysbiosis and involved in worse outcomes in an
S.
pneumoniae pneumonia model [
9]. Likewise, lung immunity assessment restricted to IgA-producing cells found that IgA production was impaired in the lung and involved in worse outcomes in a
P.
aeruginosa pneumonia model [
5].
We conducted a broader assessment of the immune response after antibiotic-induced gut dysbiosis and observed widespread lung cellular immune depression. Immune depression included cells crucial to the clearance of
P.
aeruginosa from the lung, such as alveolar macrophages, non-conventional lymphocytes, and neutrophils. Bone marrow progenitors of monocytes and DC were depressed and partially corrected by FMT, suggesting the involvement of immune cell hematopoiesis. A link between antibiotic-induced dysbiosis and depressed hematopoiesis has been demonstrated. In a non-infectious model [
11] and in mice infected by
Flaviviridae [
12].
We found that FMS-like tyrosine kinase 3 ligand (Flt3-Ligand) was significantly decreased by antibiotic-induced gut dysbiosis and partially restored by FMT. Flt3-Ligand is a cytokine that acts as a hematopoietic growth factor for early progenitors through its receptor, Flt3, in synergy with other cytokines [
13]. When administered after antibiotics, Flt3-L (a) restored or stimulated the expansion of several monocytes and dendritic cells (DC) bone marrow progenitors, (b) restored or stimulated immune lung cell populations, (c) restored outcomes of sublethal
P.
aeruginosa lung infection, and (d) increased survival of lethal
P.
aeruginosa lung infection. Flt3-Ligand promotes the generation of a primarily myeloid cell containing colonies. Flt3-Ligand is essential to the generation of DC. Furthermore, Flt3-Ligand transgenic mice show a significant expansion of Flt3-positive cells and progenitors (myeloid cells, DCs, MPP, CMP, granulocyte-macrophage progenitors, CLP, and EPLM progenitors) [
14]. Administration of human Flt3-Ligand into mice increases DC-marker positive cells in the bone marrow, liver, Peyer’s patches, thymus, peritoneum, and lung [
15]. In a murine model of influenza A virus infection, Beshara et al. showed that Flt3-L overexpression reduced the dissemination of
S.
pneumoniae instilled into the lungs by enhancing bone marrow cDC progenitors and restoring lung cDCs [
16]. These results suggest that its protective effect is not limited to
P.
aeruginosa.
Interestingly, recombinant human Flt3-L (CDX-301, Celldex Therapeutics, Hampton, NJ, USA) is currently undergoing over 30 clinical trials to treat various malignancies. Furthermore, given the FMT recent safety issues [
17], it is of particular interest that deleterious effects of prior antibiotics on lung immunity may be modulated beyond the gut microbiome.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.