Conventional dendritic cells (cDCs) play an important role in both innate and adaptive immune responses. In the lungs, cDCs are critical sentinel cells that capture, process and present antigens to activate naive T cells in lymph nodes. In addition, cDCs are involved in the innate immune response via cytokine and chemokine production (Macri et al., 2018). Lung conventional DCs comprise a variety of subsets that are typically subdivided into two sub-populations named cDC1s and cDC2s (Guilliams et al., 2014; Guilliams et al., 2016). cDC1s express the surface marker XCR1, and the expression of IRF8 and BATF3 transcription factors is required for their development. Additionally, they can express the alpha-E integrin CD103 in non-lymphoid organs like the lung, while they express CD8α in lymphoid tissues. cDC2s are characterized by IRF4, Sirpα and CD11b expression (Crozat et al., 2011; Guilliams et al., 2014; Gurka et al., 2015).

The majority of cDC development occurs in the bone marrow and requires the presence of the FMS-like tyrosine kinase 3 ligand (FLT3L) cytokine (Ginhoux et al., 2009). During this process, the commitment of cDC precursors to the cDC1 or cDC2 lineage happens relatively early in cDC development (Grajales-Reyes et al., 2015; Schlitzer et al., 2015). cDC precursors then leave the bone marrow at the pre-DC stage and migrate through the bloodstream to various organs such as the lung (Liu et al., 2009). Pre-DCs committed to the cDC1 lineage do not express CD103 (Brassard et al., 2019). The exact mechanisms by which cDC1s acquire CD103 expression upon their entrance in the lung remain unclear, but in vivo and in vitro studies suggest that exposure to GM-CSF, present in the lung, is a potent inducer of cDC CD103 expression (King et al., 2010; Greter et al., 2012; Mayer et al., 2014). To date, there is no information concerning transcription factors involved in cDC CD103 expression. However, the RUNX family of transcription factors is involved in the induction of CD103 expression in T cells (Grueter et al., 2005). The only reported ligand of αE integrin (CD103) is E-cadherin, which is expressed by epithelial cells (Corps et al., 2001). While the role of CD103 expression on cDC1s remains unclear, reports demonstrate that T cells CD103 expression facilitates lymphocyte localization and induces intracellular signaling (Pauls et al., 2001; Franciszkiewicz et al., 2013).

The two cDC subpopulations have distinct and often opposite functions (Schlitzer et al., 2015; Guilliams et al., 2016). cDC1s are particularly important for IL-12 production, antigen cross-presentation to CD8 T cells and CD4 T cell polarization into T_H1 (Hildner et al., 2008; Mashayekhi et al., 2011; Martínez-López et al., 2015). The specific function of cDC2s is more controversial, but some studies suggested that they are important for T_H2 polarization (Gao et al., 2013; Plantinga et al., 2013). To date, there is no consensus regarding the roles of cDC1s or cDC2s in the efficacy of antibacterial immune responses. Nevertheless, some interesting data suggest that, as observed in several other types of immune responses, cDC1s and cDC2s elicit distinctive functions in the fight against bacteria. Indeed, in most bacterial infections, polarization of naive T cells into T_H1 improves bacterial clearance and lung function, which suggests a beneficial role for cDC1s (Moser et al., 1997; Moser et al., 2002). In accordance, in a mouse model of lung infection with the Gram-negative bacteria Chlamydia muridarum, CD103⁺ cDC1s induced a stronger T_H1 polarization compared to cDC2s, and CD103⁺ cDC1 injection improved bacterial clearance (Shekhar et al., 2018). It should be noted, however, that cDC1s are usually involved in infections to intracellular bacteria, and that their role in response to Pseudomonas aeruginosa is not well-described. cDC2s, on the other hand, could be important for the innate immune response. Indeed, in response to LPS lung exposure, lung CD11b⁺ DCs produce more KC (CXCL1) and MIP–2 (CXCL2), two chemokines involved in neutrophil recruitment, compared to CD103⁺ cDC1s (Beaty et al., 2007).

These distinctive roles for cDC1s and cDC2s in the response to bacteria suggest that the tight balance between cDC lung subsets is important to support effective local immune responses. Recently, we demonstrated that lung exposure to lipopolysaccharide (LPS), which induces a strong local and peripheral inflammatory response, modulates the proportions of cDC populations by decreasing the percentage of CD103⁺ cDC1s and increasing CD11b⁺ DC proportions (Brassard et al., 2019). Since LPS is one of multiple bacterial components that may influence cDC populations and with the potential crucial role of cDC1s in the fight against bacterial infections in the lung, the impact of whole bacteria on the lung cDC signature remained an important unanswered question. We therefore set out to analyze the influence of an acute exposure to the Gram–negative bacteria P. aeruginosa on the local lung cDC signature. P. aeruginosa is ubiquitously found in nature and causes opportunistic acute and chronic infections in immunocompromised patients, such as those suffering of cystic fibrosis (Green et al., 1974; Lyczak et al., 2000; Moradali et al., 2017). We observed that P. aeruginosa modulated the proportions of lung cDC1 and cDC2 populations in favor of cDC2s, which was in part explained by a higher propensity of bone marrow cDC precursors to differentiate towards the cDC2 lineage, and by an incapacity of lung cDC1s to fully express CD103 in response to GM-CSF. The latter was linked to reduced GM–CSF receptor internal localization, following exposure to P. aeruginosa and LPS, and alterations in RUNX1 expression, which regulate CD103 expression. Finally, we report that the lack of CD103 expression on cDCs leads to an exacerbated airways neutrophilia, supporting the idea that the absence of CD103 expression on cDC1s promotes the lung innate response to P. aeruginosa. We therefore shed a light on a possible mechanism demonstrating that the blockade of cDC CD103 expression by Gram-negative bacteria is a crucial step in promoting the initial innate immune response to this potentially infectious agent in the lung.

DCs take part in the induction of innate and adaptive immune responses, and the efficacy of these responses is influenced by the nature of local cDC subsets (Macri et al., 2018). Until now, few studies focused on the influence of bacterial infection on cDC populations. In this report, we demonstrated that lung exposure to live P. aeruginosa bacteria decreases the proportions of CD103⁺ cDC1s in favor of CD11b⁺ cDC2s/monocyte-derived DCs. We determined that this was in part modulated by modifications in bone marrow pre-DC populations and an altered CD103 expression on XCR1⁺ cDC1s ( Figure 6 ). Furthermore, we demonstrated that the absence of CD103 expression increases neutrophils recruitment in the lung in response to LPS and P. aeruginosa.

First, we realize that our gating strategy to identify cDC2s does not discard monocyte-derived DCs, so this population must be considered as potentially upregulated in response to P. aeruginosa. Independently of this, the altered proportions of cDC populations are not restricted to lung exposure to LPS or P. aeruginosa. Indeed, similar results were reported in a mouse model of lung infection with Chlamydia muridarum, an LPS⁺ bacteria (Shekhar et al., 2018; Brassard et al., 2019). Interestingly, in the C. muridarum model, a decrease in CD103⁺ cDC proportions was observed up to 7 days post-infection, suggesting that this phenomenon is not limited to acute immune responses (Shekhar et al., 2018). This is also supported by similar alterations observed by our group in a 21-day mouse model of chronic allergic inflammation to Saccharopolyspora rectivirgula antigen (Bernatchez et al., 2017).

Our results suggest that lung exposure to P. aeruginosa altered bone marrow cDC precursors towards pre-DCs that are committed to the cDC2 lineage, which likely contributes to the accumulation of lung cDC2s. A modification of bone marrow pre-DCs towards the cDC1 or cDC2 lineage was observed in other immunological contexts including cancer, viral infections and lung exposure to LPS (Beshara et al., 2018; Meyer et al., 2018 Brassard et al., 2019), suggesting that this mechanism is conserved across various diseases and types of inflammatory responses. We previously observed that lung exposure to LPS leads to a decrease in IRF8 expression in pre-DCs, whereas an increase of IRF4 expression is reported here in response to P. aeruginosa (Brassard et al., 2019). Some studies suggest a sequential order between IRF8 and IRF4 expression in which the decrease in IRF8 expression precedes the increase in IRF4 expression in pre-DCs committed to the cDC2 lineage (Bajaña et al., 2016; Sichien et al., 2016). Therefore, the differences reported in both studies could be explained by a difference in the timing of induction for these transcription factors in these models. Despite this difference, both results report an imbalance towards the cDC2 lineage differentiation.

Our results also suggest that splenic cDC populations are affected by lung P. aeruginosa exposure. As the cDC turnover in the spleen is fast, even at steady state, an absence of total cDC number increase (such as observed here) does not indicate an absence of new cDC recruitment (Kamath et al., 2002), and the altered cDC populations is likely a reflection of a systemic impact of P. aeruginosa. Thus, our results suggest that bone marrow pre-DCs are biased towards the cDC2 lineage, which in turns influences the subsets of cDC precursors recruited in the spleen, resulting in a modulation of splenic cDC1 and cDC2 proportions ( Figure 6 ). As pre-DCs are precursors for most tissue cDCs, this modification in bone marrow pre–DCs could also influence cDC populations in other organs, and ultimately impact the efficacy of immune responses in case of a systemic infection (Ginhoux et al., 2009). We are aware that other mechanisms besides alterations in cDC precursors could be involved in explaining the differences in the cDC population signature in our models. As our gating strategy also included mo-DCs as CD11b⁺, this population could contribute to the imbalance reported here (Guilliams et al., 2016). Also, a higher migratory rate of cDC1s from the lung to the draining lymph nodes could take part in the observed decrease percentage of this population following P. aeruginosa exposure (Ho et al., 2011; Nakano et al., 2013).

Supported by previously-published results (Brassard et al., 2019) and present data, we propose that under homeostatic conditions, newly-recruited lung pre-DCs acquire CD103 expression in response to GM-CSF stimulation through their final differentiation into cDCs. However, in the context of Gram-negative bacterial exposure, the interaction with the bacteria or LPS prevents CD103 mRNA synthesis resulting in lower GM-CSF-induced CD103 expression on cDC1s. Importantly, our results do not demonstrate that LPS nor P. aeruginosa reduce the expression of CD103 once it has been expressed, but rather block the induction upon contact with GM-CSF. Furthermore, our observations suggest that a reduced internal localization of the GM-CSFR could be one of the mechanisms contributing to the reduction in CD103 expression. Our results linking the GM-CSFR internalization to CD103 expression were obtained using Dynasore a GTPase inhibitor that inhibits dynamin activity, which prevents endocytosis. Although this inhibitor is frequently used to block GM-CSFR internalization (Katz et al., 2016; Zsiros et al., 2019), Dynasore could also exert other effects. For instance, the presence of Dynasore could also block the internalization of other receptors, such as TLR4, express by cDCs (Kagan et al., 2008). However, as our assays with Dynasore were performed using only GM-CSF stimulation, it is very likely that in this case, the inhibition of CD103 expression was linked to a reduced blockade of GM-CSFR internalization. The exact steps linking the activation of GM-CSFR signaling to the induction of CD103 mRNA expression on cDCs remained unknown. For the first time, we demonstrate that RUNX1 expression is crucial for CD103 expression in cDCs. Additionally, we show that LPS and P. aeruginosa alter RUNX1 expression, likely contributing to reduced CD103 expression in response to these stimuli. Of course, other cellular signaling pathways besides GM-CSF are involved in regulating CD103 expression on cDCs. For example, TGFβ and retinoic acid also reportedly induced CD103 expression (Iliev et al., 2009; Roe et al., 2017; Roe et al., 2020). Whether LPS and P. aeruginosa alter CD103 expression in response to these stimuli remains unknown, but definitely of interest.

Despite a wealth of knowledge on CD103 modulation on cDCs in response to various stimuli (Nakano et al., 2013; Bernatchez et al., 2017; Brassard et al., 2019), the physiological function of this phenomenon on the host immune response remained undefined. We propose here that the absence of CD103 expression supports the intensity of the initial host innate immune response. Anecdotally, in a model of skin bacterial infection, cDC1s were crucial for neutrophils recruitment to the site of infection via the secretion of VEGF-α (Janela et al., 2019), which suggested that cDC1s influence neutrophil migration. CD103 signaling in cDC1s could be involved in promoting homeostasis by restraining the production of chemokines or pro-inflammatory cytokines. In contrast, during bacterial infections, the absence of CD103 signaling on cDC1s could increase their capacity to secrete neutrophils-attracting chemokines or induce neutrophil migration via an indirect effect on surrounding cells. All in all, our study suggests that the reduced CD103 expression on cDC1s may be a step in stimulating the innate immune response to Gram-negative bacteria.

The higher recruitment of lung neutrophils in Cd103^-/- mice in response to P. aeruginosa was not caused by a difference in the number or proportions of cDC or pre-DC populations. However, a reduced IRF4 expression in bone marrow pre-DCs was observed in Cd103^-/- naive mice, suggesting a lower propensity of cDC2 differentiation at steady state in these mice. As pre-DCs don’t express CD103, an indirect mechanism must be involved to explain this (Brassard et al., 2019). Also, we cannot exclude the possibility that, as WT and Cd103^-/- strains were not co-housed in our experiments, differences in microbiota could be involved in impacting IRF4 expression.

Neutrophils are crucial to eradicate P. aeruginosa infection (Koh et al., 2009). The fact that the increased neutrophil count in Cd103^-/- mice did not in turn change bacterial clearance indicates that the number of neutrophils in WT mice was sufficient to quickly clear bacteria. This is also supported by the fact that most of this mucoid strain of P. aeruginosa was cleared from the lung past 6 h (data not shown) in WT mice. Furthermore, several other studies observed that a higher number of lung neutrophils do not necessarily lead to a better P. aeruginosa clearance (Ballinger et al., 2006; Sen-Kilic et al., 2019). As CD103 expression could restrain chemokine production by cDC1s, mice in which cDC1s constitutively express CD103 would be a great tool to address the importance of cDC-specific CD103 expression on neutrophil recruitment and bacterial clearance with WT mice. Based on current knowledge, it is hard to determine whether the alterations of cDC populations induced by P. aeruginosa promote or impair host immune responses. First, the combination of the important lung recruitment of cDC2s, known to secrete large amount of MIP-2 and KC neutrophil-attractant chemokines, and the possible higher capacity of cDC1s to recruits neutrophils in absence of CD103 expression suggest a positive impact on immune response. On the other hand, current knowledge on adaptive immune responses against bacterial infection endorse the idea that cDC1s support the efficacy of the adaptive immune response compared to cDC2s. To date, no studies have established a specific role for cDC1s or cDC2s in innate or adaptive immune responses to P. aeruginosa, but this report takes a first important step towards better understanding the modulation of local cDC populations in the context of Gram-negative bacterial clearance.

An important conclusion rising from this data and our previously published research is that several factors such as the presence of LPS, TNF or P. aeruginosa can prevent the induction of CD103 expression on cDC1 (Brassard et al., 2019). Currently, in the lung, CD103 remains one of the main markers used to identify cDC1s (Ng et al., 2018; Shekhar et al., 2018; Monaghan et al., 2020). However, our results suggest that CD103 is not an ideal marker for this population in context of lung inflammation as its expression is highly modulated. Furthermore, in a few other particular immunological contexts, some lung CD11b⁺ DCs can also express CD103, which could lead to misinterpretations if used as a specific marker of cDC1s (Shane et al., 2018; Tweedle and Deepe, 2018). Finally, another disadvantage is that CD103 is not a marker of human lung cDC1s (Guilliams et al., 2014; Amon et al., 2020). Thus, the ideal marker to properly identify cDC1s should be constitutively expressed by cDC1s independently of their differentiation stage or activation status, and similarly expressed by human cDC1s.

In summary, we demonstrated that i.n. exposure to the Gram-negative bacteria P. aeruginosa alters the proportions of CD103⁺ cDC1 and CD11b⁺ cDC2 populations in favor of cDC2s, via the modulation of bone marrow cDC precursors and an impact on CD103 mRNA production by cDC1s. Furthermore, we demonstrated that the absence of CD103 expression favors the recruitment of lung neutrophils, suggesting that this phenomenon could be an important step in the early innate immune response to P. aeruginosa.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by Comité de protection des animaux de l’Université Laval.

JB and M-RB conceived the study. JB, JR, M-JB, and EB performed experiments. JB and M-RB drafted and revised the paper. AML and M-JB critically revised the manuscript. MV and CD contributed to the final revision and supported microbiology experiments. M-RB supervised the study. All authors contributed to the article and approved the submitted version.

This work was supported by the Fonds sur les Maladies Respiratoires Bégin/Lavoie de l’Université Laval and by the Fondation de l’Institut Universitaire de Cardiologie et de Pneumologie de l’Université Laval.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.