Protection against influenza infection requires early recognition by inflammatory dendritic cells through C-type lectin receptor SIGN-R1

Abstract

The initial phase of influenza infection occurs in the upper respiratory tract and the trachea. Yet, little is known about the initial events whereby the immune system recognises the virus and controls viral dissemination. Here, we report that inflammatory dendritic cells (IDC) are recruited to the trachea early on following influenza infection through type I interferon-mediated production of the chemokine CCL2. We further show that recruited IDC express the C-type lectin receptor SIGN-R1 that mediates direct recognition of the virus by interacting with the N-linked glycans present in the glycoproteins of the virion envelope. Activation of IDC via SIGN-R1 triggers the production of the chemokines CCL5, CXCL9 and CXCL10 that initiate the recruitment of protective natural killer (NK) cells in the infected trachea. In the absence of SIGN-R1, the recruitment and activation of NK cells is impaired, which leads to uncontrolled viral proliferation. Altogether, our results shed light on the orchestration of the early cellular and molecular events involved in the immune protection against influenza.

How the innate immune system initially recognises the replicating influenza virus is an important determinant of the infection outcome¹. However, there is limited knowledge about these early immune defence mechanisms and how they affect disease progression. Influenza is first encountered by the host immune system in the upper airways in which the initial virus replication is restricted². Later, progression of the infection to the lower respiratory tract occurs through the airway epithelium of the trachea and is essential for the development of the acute phase of the disease³. Most of the knowledge about the immune response in the trachea has been gained from studies focused on the pulmonary phase of the infection² and, therefore, a better characterisation on the early response in the trachea is needed.

The immune response to airborne pathogens involves their recognition by macrophages and dendritic cells (DC) located underneath the mucosal cell layer⁴. Tracheal DC have been divided in different subsets according to their anatomic location and expression of surface markers^4,5. One of the best-studied subtypes is the resident conventional DC (cDC) that play an important role in initiating the adaptive response by migrating to regional lymph nodes and presenting antigen to T cells⁶. Conversely, a second group of DC, namely the inflammatory DC (IDC), perform their function at the site of infection⁷. IDC are known to be an important source of cytokines, such as tumour-necrosis factor, and inflammatory molecules, such as nitric oxide⁸. However, the precise function of IDC during influenza infection have not been yet characterised.

The activation of these cells is associated with the detection of viral components by pattern-recognition receptors (PRR) commonly expressed by DC⁹. C-type lectin receptors are PRR known to bind specific pathogen-associated carbohydrate structures¹⁰. Amongst them, the DC-SIGN murine homologue SIGN-R1 is known to directly recognise bacterial¹¹, viral¹² and fungal¹³ pathogens. Previous studies have described the presence of cells expressing SIGN-R1 in the skin and the lungs, and in secondary lymphoid organs such as the spleen¹⁴ and the lymph nodes¹². However, the specific function of this receptor in the response against infectious virus is not known.

In this study, we found that IDC are recruited into the tracheal epithelium and recognise influenza virus via SIGN-R1. The production of inflammatory chemokines by these cells induces the recruitment of NK cells, which is essential for the control of the viral replication in the trachea.

Results

Influenza infection induces the recruitment of IDC to the trachea

To characterise the inflammatory response that occurs in the trachea following influenza infection, we infected mice with 200 PFU of the mouse-adapted influenza virus strain PR8. We confirmed, using confocal microscopy, that the virus was able to initially infect the tracheal epithelial cell layer (Supplementary Fig. 1a). Next, we characterised by flow cytometric analysis tracheal cell populations according to the expression of different surface markers^4,7,15: Inflammatory dendritic cells (IDC), plasmacytoid DC (pDC), CD103+ conventional DC (CD103+ cDC) CD11b+ conventional DC (CD11b+ cDC) and resident respiratory tract macrophages (RTMϕ) (Fig. 1a). We observed a prominent recruitment of IDC at 3 days post infection (d.p.i.) (Fig. 1b). Previous studies have reported CCR2+ Ly6c^high monocytes as the precursors of IDC¹⁶. Therefore, we confirmed the monocytic origin of these cells by studying their expression of CCR2 (Fig. 1c), in contrast to the other subtypes of DC (Supplementary Fig. 1b). Furthermore, we observed that, in IDC, CCR2 is eventually downregulated, as the expression of the maturation marker MHC class II increases (Supplementary Fig. 1c). Next, we studied the specific effect on IDC when infecting CCR2-KO animals. We confirmed by flow cytometry that at 3 d.p.i. IDC were significantly reduced in KO animals, compared to their WT counterparts (Fig. 1d). However, we did not detect any significant variation in the total number of the other cell types analysed (Supplementary Fig. 1d).

Figure 1. Influenza infection promotes early recruitment of IDC to the trachea.

a, Flow cytometric characterisation of inflammatory dendritic cells (IDC; CD45+ / MHCII+ / CD11c+ / CD11b+ / Ly6c^hi), plasmacytoid DC (pDC; CD45+ / MHCII+ / CD11c^Int / B220+ / CD11b- / Bst2+/ Ly6c+), CD103+ conventional DC (CD103+ cDC; CD45+ / MHCII+ / CD11c+ / CD11- / Ly6c- / CD103+), CD11b+ conventional DC (CD11b+ cDC; CD45+ / MHCII+ / CD11c+ / CD11b+ / Ly6c- / CD103-) and resident respiratory tract Macrophages (RTMϕ; CD45+ / MHCII+ / CD11c- / B220- / Ly6c- / CD11b+ / F4/80+) in mouse trachea. b, Flow cytometric analysis of DC subsets and RTMϕ in trachea at 0, 3, 5 and 7 days after intranasal infection with influenza virus PR8 (n = 3, 4, 4 and 4 mice per respective group). c, Representative histogram showing surface expression of the chemokine receptor CCR2 in tracheal IDC at 0 and 3 days p.i.. d, Flow cytometric analysis showing a significant reduction in the number of IDC in tracheas from CCR2-KO mice compared to the WT group at day 3 p.i.. Left panel shows a representative scatterplot at day 3 p.i.. Right panel shows IDC quantification (n = 4 mice per group). The presented data are representative of at least three independent experiments (a-d). Results are given as mean ± SD. Data were analysed using the two-tailed Student’s t-test.

Recruited IDC are located in close proximity to the tracheal epithelial cell layer during influenza infection

To study the position of the DC network within the mucosal tissue, using confocal microscopy, we produced three-dimensional (3D) reconstructions of mouse tracheas from CD11c-YFP animals. We observed that in uninfected mice, CD11c+ DC were distributed homogeneously above the basement membrane in proximity to the epithelial cell layer (Fig. 2a). To study the specific location of the different subsets of DC and RTMϕ before and after influenza infection, we stained explants from tracheas of CD11c-YFP animals with antibodies against CD11b and Ly6c (Fig. 2b and Supplementary Fig. 2a). Consistent with the flow cytometry data, we could not detect any IDC in uninfected animals (Supplementary Fig. 2b). Instead, in the infected mice, we found that recruited IDC were located towards the lumen of the trachea unlike the other cell types that were mainly located in the lamina propria (Fig. 2c).

Figure 2. Recruited IDC are located in close proximity to the epithelial cell layer during influenza infection.

a, (Top and Middle) 3D surface reconstruction of confocal images from a trachea showing the network of CD11c+ cells (green) located in the lamina propria in close contact with the basement membrane (violet). Epithelial cells (red), located in the luminal part, are stained with CMTMR. (Bottom) Schematic drawing indicating the anatomical distribution of the reconstructed area. Scale bar represents 20 μm. b, Confocal micrographs showing the 3D reconstruction of the DC subsets and RTMϕ, located in the tracheal mucosa at day 3 p.i.. Cells are characterised based on the expression of CD11c (green), CD11b (light blue) and Ly6c (red). IDC are defined as: CD11c+ / CD11b+ / Ly6c+; CD103+ cDC: CD11c+ / CD11b- / Ly6c-; CD11b+ cDC: CD11c+ / CD11b+ / Ly6c-; and RTMϕ: CD11c- / CD11b+ /Ly6c-. Scale bar represents 16 μm in IDC panels and 20 μm in the remaining. c, (Left) representative 3D reconstruction from confocal micrographs of a trachea at day 3 p.i. showing the location of the IDC (red), CD103+ cDC (green), CD11b+ cDC (yellow) and RTMϕ (light blue) with respect to the basement membrane (violet). Scale bar represent 20 μm. (Right) Box plot representing the position of the different group of cells with respect to the basement membrane in the tracheal mucosa at day 3 p.i.. The value of the position of the basement membrane (BM) was considered 0, cells located towards the lamina propria received positive values while cells located towards the luminal side received negative values (data are based on the analysis of 3 confocal reconstructions; IDC, n = 61; CD103+ cDC, n = 54; CD11b+ cDC, n = 40; RTMϕ, n =40). The presented data are representative of at least three independent experiments (a-c). In c, the box plots show 25^th to 75^th percentiles and median, and the whiskers show 5^th to 95^th percentiles. Data were analysed using the two-tailed Student’s t-test.

Type I interferon (IFN) is required for the production of CCL2 that recruits IDC precursors to the trachea

To characterise the anti-viral response that leads to the recruitment of IDC, we assessed the expression of type I IFN in the trachea. We observed a significant upregulation of the transcript levels of both IFN-α and IFN-β during the first 2 d.p.i. (Fig. 3a). Moreover, the cytoplex analysis confirmed that the protein levels of IFN-β were also significantly elevated in both WT and CCR2-KO mice at 3 d.p.i. (Fig. 3b). We also detected a significant increase in the RNA levels of the CCL2 chemokine, a CCR2 ligand, during the first 3 d.p.i. (Fig. 3c), which coincided with an increase in its protein levels (Fig. 3d). Regarding the source of this cytokine we found that RTMΦ, together with pDC and IDC, were the mainly producers of CCL2 (Supplementary Fig. 3a and Supplementary Fig. 3b). Interestingly, we observed a similar pattern of expression of CCL2 and Type I IFN during the first 7 d.p.i. (Supplementary Fig. 3c) concurrent with the IDC recruitment to the trachea. To examine whether there is a connection between the two cytokines, we analysed the CCL2 protein levels in type I IFN receptor (IFNAR)-KO mice, which showed that signalling through IFNAR was required for the secretion of CCL2 (Fig. 3d). In addition, we observed a prominent decrease in the numbers of CCR2+ monocytes in tracheas from IFNAR-KO mice at 3 d.p.i. (Fig. 3e) and a similar reduction in the number of IDC (Fig. 3f), consistent with the role of CCR2+ monocytes as IDC precursors.

Figure 3. Type I interferon (IFN) is required for the production of CCL2, which recruits IDC precursors to the trachea.

a, Time course showing RNA expression of the cytokines IFN-α and IFN-β in mouse trachea during the first 3 days p.i. with influenza virus (IFN-α, n = 3, 4, 4 and4; IFN-β, n = 3, 3, 4 and 4 mice per respective time point). b, Levels of secreted IFN-β in trachea from CCR2-KO mice at day 3 p.i. compared to WT (WT, n = 5 and 5; CCR2-KO, n = 4 and 6 mice per respective time point). c, Time course showing RNA expression levels of the chemokine CCL2 in trachea during the first 3 days p.i. (n = 4, 4, 4 and 5 mice per respective time point). d, Levels of secreted CCL2 in trachea from CCR2-KO and IFNAR-KO mice at day 3 p.i. compared to WT (WT, n = 9 and 10; CCR2-KO, n = 4 and 6; IFNAR-KO, n = 3 and 5 mice per respective time point). e, Flow cytometric quantification of total number of CCR2+ monocytes (CD45+ / MHCII- / CD11c- / CD11b+ / Ly6c^hi / Ly6g- / CCR2+) in trachea from IFNAR-KO mice at day 3 p.i. compared to uninfected controls (WT, n = 3 and 5; IFNAR-KO, n = 3 and 4 mice per respective time point). f, Flow cytometric quantification of total number of IDC in trachea from IFNAR-KO mice at day 3 p.i. compared to uninfected controls (WT, n = 3 and 5; IFNAR-KO, n = 3 and 4 mice per respective time point). The presented data are representative of at least three independent experiments (a-f). In a-d, the box plots show 25^th to 75^th percentiles and median, and the whiskers show minimum and maximum values. In e and f, results are given as mean ± SD. Data were analysed using the two-tailed Student’s t-test.

IDC express the lectin receptor SIGN-R1 that recognises glycosylated influenza viral proteins

To assess the expression of SIGN-R1 receptor in the mononuclear phagocytic cell compartment of the trachea, we performed flow cytometric analysis. We found that SIGN-R1 was highly expressed by IDC at 3 d.p.i. (Fig. 4a and 4b). Moreover, 3D confocal microscopy revealed that SIGN-R1 colocalised with DIO-labelled influenza virus after 1 h of incubation (Fig. 4c), suggesting that this C-type lectin expressed in IDC is able to bind influenza virus. To examine this possibility, we performed ELISA that confirmed that SIGN-R1 binds with high affinity to whole influenza virus (EC50 = 1.67 μg/ml; Supplementary Fig. 4a). To further investigate this interaction, we performed Surface Plasmon Resonance (SPR) assays to investigate whether SIGN-R1 binds the hemagglutinin (HA) glycoprotein of influenza PR8 virus with a KD of 142 nM (Fig. 4d, upper panel). Next, to evaluate whether the binding of influenza virus to SIGN-R1 could be abrogated by interfering with glycosylation, we removed glycan moieties from the hemagglutinin using different glycosidases. We used Endoglycosidase H (Endo H) to eliminate mannose glycosylation from HA. However, since all later oligosaccharide structures are resistant to Endo H, we also used the Peptide:N-glycosidase F (PNGase F), which is cleaving between the innermost N-Acetylglucosamine and asparagine residues of high mannose regardless of the complexity of the glycan. We monitored glycosylation removal by immunoblot after treatment and we observed that PNGase F was the most efficient enzyme to remove all N-linked glycan (Supplementary Fig. 4b). After N-linked glycan removal of HA, we could not detect any binding between SIGN-R1 and the treated HA (Fig. 4d, lower panel). Of note, binding to the mAb H36-7¹⁷ was preserved before and after N-linked glycan removal (Supplementary Fig. 4c), indicating that HA was correctly folded after PGNAse F treatment. Next, we performed binding assays with different HA subtypes to explore the possibility of SIGN-R1 also recognising other flu strains. In agreement with the SPR affinity measurement, we observed high binding titres; EC₅₀ = 0.66 μg/ml, 0.81μg/ml, 1.19 μg/ml and 1 μg/ml for H1 (PR8 and N. Caledonia), H3 and H5, respectively (Fig. 4e). In addition, SIGN-R1 binding was also observed with neuraminidase subtype 1 (N1) (EC₅₀ = 0.63 μg/ml; Supplementary Fig. 4d). In this assay FI6, an antibody specific against the stem region of HA¹⁸, was included to confirm that the observed binding was not due to a HA contamination in the N1 purification. Interestingly, we also observed the binding of SIGN-R1 to other respiratory viruses such as the human respiratory syncytial virus (RSV)¹⁹ and the human metapneumovirus (MPV)²⁰ (EC₅₀ = 0.98 μg/ml and EC₅₀=1,14 μg/ml, respectively; Supplementary Fig. 4a). Taken together, these results indicated that SIGN-R1 recognises several viruses by targeting the carbohydrate part of glycoproteins. Moreover, we observed that addition of an αSIGN-R1-blocking antibody significantly reduced the binding of DIO-labelled influenza PR8 to SIGN-R1+ IDC (Fig. 4f), confirming the role of this receptor in the attachment of the virus on IDC.

Figure 4. IDC express the lectin receptor SIGN-R1 that recognises influenza PR8 through binding to glycosylated viral proteins.

a, Flow cytometric analysis showing the expression levels of SIGN-R1 in tracheal DC and RTMϕ at day 3 p.i. (n = 5 mice per group). b, Representative scatterplots showing the percentage of SIGN-R1+ IDC in trachea at day 3 p.i.. c, Representative 3D reconstruction of confocal micrographs showing the colocalisation between DIO-labelled PR8 (PR8-DIO) and SIGN-R1 (white arrows), expressed in an IDC one hour after in vitro culture. Bar indicates 2 μm (left) and 0.5 μm (right). d, Sensogramms of Surface Plasmon Resonance (SPR) analysis of the HA-binding affinity of SIGN-R1 Fc chimera protein (SIGN-R1-Fc.) SIGN-R1-Fc with or without glycosylation (after treatment with N-glycosidase F (PNGase F)), with different concentration of H1, below their respective kinetic parameters, calculated by SPR using purified proteins are shown. N.B., no binding detected. e, ELISA analysis of the HA-binding titre from different subtypes of the recombinant mouse SIGN-R1-Fc (n = 2 replicates per group). f, (Left) Representative scatterplots from flow cytometric analysis showing the capture of PR8-DIO by SIGN-R1+ IDC after an incubation period of 1 h with the virus. In the αSIGN-R1 group, the binding was inhibited by administration of a blocking antibody previous to the incubation with the virus. (Right) Mean fluorescence intensity (MFI) analysis showing differences in the capture of PR8-DIO by IDC treated with αSIGN-R1 blocking antibody with respect to the control groups (n = 4, 4 and 3 replicates per respective group). The presented data are representative of at least three (a-c and f) or two (d-e) independent experiments. Results are given as mean ± SD (a, b and f), as mean± SEM (d) or mean alone (e).

SIGN-R1 is involved the production of inflammatory chemokines by IDC and the recruitment of NK cells into the tracheal mucosa

To investigate the role of IDC as inducers of the inflammatory response, we studied the production of different chemokines in tracheal lavage from CCR2-KO and SIGNR1-KO mice at 3 d.p.i.. We found that the absence of IDC or SIGN-R1 resulted in a significant reduction in the levels of the chemokines CCL5, CXCL9 and CXCL10 (Fig. 5a, 5b and Supplementary Fig. 5a, respectively). Furthermore, the reduced chemokine expression on SIGNR1-KO was not related with a deficiency in DC and RTMΦ compared to WT animals (Supplementary Fig. 5b). The specific expression of these chemokines by IDC, as well as their reduced levels in SIGNR1-KO mice, was also confirmed by intracellular flow cytometric staining (Fig. 5c-f). To examine whether the observed differences were due to the direct interaction of SIGN-R1 with the viral proteins, we isolated IDC generated by intranasal administration of Poly (I:C) which had a similar expression levels of SIGN-R1 compared to tracheal IDC at 3 d.p.i. (Supplementary Fig. 5c), and we incubated them in vitro with a vaccine suspension containing HA. The analysis of the secreted chemokine levels at 12 h post stimulation showed that IDC treated with HA were able to increase the production of CCL5, compared to untreated controls, while SIGNR1-KO IDC were not (Supplementary Fig. 5d). We also confirmed that IDC were the major producers of these chemokines in the trachea at 3 d.p.i. and that SIGN-R1 deficiency reduced their expression levels only on IDC (Fig. 5g and 5h, Supplementary Fig.5e). Considering the role of CCL5, CXCL9 and CXCL10 in the recruitment of NK cells to the infection site²¹, we developed a flow cytometry assay (Fig. 5i) to detect the infiltration of NK in the trachea (Fig. 5j). We observed an early peak in the number of NK cells in infected trachea, which coincided with the aforementioned peak of IDC at 3 d.p.i.. Moreover, confocal analysis indicated that the majority of NK cells were positioned towards the luminal side, in close proximity to IDC (Fig. 5k and 5l) with an average distance of 66.2 µm (Supplementary Fig. 5f). We could also observe that on average, 25.5% of the NK cells analysed were establishing contacts with IDC (Supplementary Fig. 5g). Furthermore, we observed that NK cells expressed the receptors CCR5 and CXCR3, which bind to the chemokines CCL5 and CXCL9/10 respectively²², but not CCR2 (Fig. 5m). The recruitment of tracheal NK cells was partially dependent on the IDC, as demonstrated by the reduced number of these cells in CCR2-KO and SIGNR1-KO mice (Fig. 5n).

Figure 5. SIGN-R1 is involved the production of inflammatory chemokines by IDC and the recruitment of NK cells into the tracheal mucosa.

a,b, Cytokine levels of secreted CCL5 (a) and CXCL9 (b) in tracheas (WT, n = 5; CCR2-KO, n = 4; SIGNR1-KO, n = 4 mice per group). c-f, Representative histogram and mean fluorescence intensity (MFI) analysis showing CCL5 (c,e; WT, n = 3 and 6; SIGNR1-KO, n = 3 and 7 mice per respective time point) and CXCL9 (d,f; WT, n = 3 and 5; SIGNR1-KO, n = 3 and 4 mice per respective time point) expression by tracheal IDC at day 3 p.i.. g,h, Mean fluorescence intensity (MFI) analysis showing expression levels of CCL5 (g; WT, n = 6; SIGNR1-KO, n = 7 mice per group) and CXCL9 (h; WT, n = 5; SIGNR1-KO, n = 4 mice per group) by different cell types at day 3 p.i.. i,j, Flow cytometric quantification of NK cells (CD3- / B220- / NK1.1+) in mouse trachea (n = 4 mice per group). k, Representative 3D reconstruction from confocal micrographs of a trachea at day 3 p.i. showing the proximity between the IDC (white) and NK cells (green) with respect to the basement membrane (violet). White arrows indicate IDC interaction with NK cells. Scale bar represents 40 μm. Magnified area bars, 5 μm. l, Box plot representing the anatomical position of IDC (white) and NK cells (green) relative to the basement membrane in the tracheal mucosa at day 3 p.i. The value of the position of the basement membrane was considered 0, cells located towards the lamina propria received positive values while cells located towards the luminal side received negative values (data are based on the analysis of 3 confocal reconstructions; IDC, n = 63; NK cells, n = 102). m, Representative histograms showing surface expression of different chemokine receptors in tracheal NK cells at day 3 p.i. (n = 3 mice per group). n, Flow cytometric quantification of total number of NK cells in tracheas at day 0 and 3 p.i. (WT, n = 8 and 9; CXCR3-KO, n = 3 and 4; SIGNR1-KO, n = 7 and 8; CCR2-KO, n = 4 and 4 mice per respective time point). The presented data are representative of at least three independent experiments (a-m) or at least two independent pooled experiments (n). In a,b and l, the box plots show 25^th to 75^th percentiles and median, and the whiskers show minimum and maximum values (a, b) or 5^th to 95^th percentiles (l). In c, d, g, h, j and n, results are given as mean ± SD. Data were analysed using the two-tailed Student’s t-test.

SIGN-R1+ IDC are required to control influenza infection

Next, we investigated the involvement of IDC in the activation and functional properties of the recruited NK cells. We detected a significant decrease in the number of cells positive for the markers CD69, granzymeB, perforin, NKG2D, NKp46, CD11b and IFN-γ in CCR2-KO and SIGNR1-KO mice, with respect to the WT at 3 d.p.i. (Fig. 6a). The observed reduction also correlated with a significant decrease in the frequency of positive NK cells for the examined markers and the expression levels of the same markers in NK cells from CCR2-KO and SIGNR1-KO mice, with the exception of perforin, NKG2D and NKp46 (Supplementary Fig. 6a, Fig. 6b,). Considering the role of IL-18 in the activation of NK cells, we detected a significant increase in the transcription levels of these molecule in WT animals during the first 3 d.p.i (Fig. 6c), which were yet significantly reduced in the CCR2-KO and SIGNR1-KO animals (Fig. 6d). To study how the absence of IDC, NK cells or SIGN-R1 affected the progression of influenza infection, we measured viral titres in tracheas from CCR2-KO, SIGNR1-KO and in mice treated with αNK1.1 depletion antibody at 3 d.p.i.. We observed an increase in the influenza titres in all of the examined groups, in relation to the WT control (Fig. 6e). These results correlated with an increased mortality (Fig. 6f) and weight loss (Fig. 6g) in all of the deficient groups. Finally, we analysed the presence of influenza-specific antibodies in WT and SIGNR1-KO mice at day 5 and 10 p.i. (Supplementary Fig. 6b). The results indicated that at the time of death (5-6 d.p.i.) mice had very low or undetectable titers of influenza-specific antibodies and that SIGNR1-KO mice showed no deficiency in antibody production at later time points p.i.

Figure 6. SIGN-R1+ IDC are required to control influenza infection in the trachea.

a,b, Flow cytometric quantification of total numbers and mean fluorescence intensity (MFI) of CD69+, GranzymeB+, Perforin+, NKG2D+, NKp46+, CD11b+ and IFN-γ+ NK cells at day 3 p.i (WT, n = 9, 9, 8, 9, 9, 9 and 8; CCR2-KO, n = 5, 3, 3, 5, 5, 5, and 5; SIGNR1-KO, n = 9, 9, 9, 9, 9, 9, and 8 mice per respective marker). c,Time course showing RNA expression levels of IL-18 in trachea during the first 3 days p.i. (n = 4 mice per group). d, RNA expression levels of IL-18 in trachea form WT, CCR2-KO and SIGNR1-KO mice at day 3 p.i. (WT, n= 3 and 5; CCR2-KO, n = 6; SIGNR1-KO, n = 6 mice per group). e, Virus titre in trachea form WT, CCR2-KO, SIGNR1-KO and αNK1.1-treated mice determined by TCID₅₀ at day 3 p.i. with influenza virus (n = 4 mice per group). f, g, Mortality and morbidity analysis of WT, CCR2-KO, SIGNR1-KO, αNK1.1-treated and isotype control-treated mice after infection with influenza virus (n = 8 mice per group). The presented data are representative of at least two independent pooled experiments (a,b) or at least three independent experiments (c-g). In a, b and g, results are given as mean ± SD. In c-e, the box plots show 25^th to 75^th percentiles and median, and the whiskers show minimum and maximum values. Data were analysed using the two-tailed Student’s t-test (a-e) and the Mantel-Cox method for the survival curve (f).

Discussion

In this study, we aimed to elucidate the mechanism by which IDC contribute to the control of influenza infection. Initially, we characterised the origin of the recruited IDC. Former studies had confirmed Ly6c^hi CCR2+ monocytes as the precursors of IDC at the site of infection¹⁶. In this work, we observed that, following their recruitment to the trachea, monocytes differentiated into IDC, exhibited a high expression of MHCII and CD11c and a downregulation of CCR2. The prominent expression of CCR2 in IDC progenitors suggested a critical role of its ligands, CCL2 and CCL7, in the regulation of their migration to the inflamed tissue^23,24. The observed increase in the expression of CCL2 described in this study is probably associated with the recruitment of monocytes from the bone marrow to the bloodstream²⁵ and facilitates monocyte trafficking across an epithelial cell monolayer²⁶.

Previous studies demonstrated that type I IFN signalling is necessary for the enhancement and sustainment of CCL2 levels²⁷. Accordingly, we observed an early and pronounced expression of type I IFN in the trachea upon influenza infection. Moreover, we showed that this signalling pathway was necessary for CCL2 production in the trachea, as IFNAR-KO mice presented a severe reduction in the levels of this chemokine. This correlated with a complete inhibition in the recruitment of CCR2+ monocytes and IDC in the infected trachea. A similar effect on CCL2 production was previously reported in the draining LN of IFNAR-KO mice after vaccination with a non-replicative UV-inactivated PR8²⁸. However, we cannot discard the possible effect that lack of viral control in IFNAR-KO mice²⁹ might have on the production of CCL2 and cell recruitment in the infected trachea.

Furthermore, we demonstrated that, following their recruitment in the trachea, IDC are positioned in close proximity to the epithelial cell layer, which suggests a role of IDC in sensing viral particles generated by infected epithelial cells. Moreover, it has been shown that IDC from lungs have the capacity to internalise viral particles during influenza infection³⁰. Therefore, we might expect a similar function of IDC capturing influenza virus in the trachea.

C-type lectin receptors have recently emerged as inducers of the expression of specific genes in response to pathogens³¹. During the course of our investigation, we studied the expression of the C-type lectin receptor SIGN-R1 in IDC in the trachea. Our data showed that SIGN-R1 participates in the recognition of influenza virus, as blocking of this receptor affected the capture of PR8 by IDC. We confirmed this hypothesis using SPR and demonstrated that SIGN-R1 binding to HA is glycosylation-dependant. Moreover, SIGN-R1 was able to recognise another surface protein of influenza virus, the neuraminidase, and other viruses with a comparable glycosylation pattern. These results are related with the capacity of SIGN-R1 to bind the mannose present on HA from all influenza virus subtypes. Thus, we demonstrated that SIGN-R1 is a key receptor able to bind a number of viral glycoproteins by targeting the carbohydrate moiety. This property could be used for the development of therapeutic agents that might block the infection of a wide range of influenza virus subtypes. In addition, it has been previously demonstrated that SIGN-R1 recognises other airborne pathogens such as Streptococcus pneumoniae¹¹ or Mycobacterium tuberculosis³², suggesting its potential therapeutic use to control airborne bacterial infections.

We have also demonstrated that SIGN-R1 is necessary for the secretion of inflammatory cytokines after binding to influenza virus, as expression in IDC enhanced the production of CCL5, CXCL9 and CXCL10. Furthermore, we observed that CCL5 expression was not increased in SIGNR1-KO IDC when they were stimulated in vitro with HA. However, we did not observe a significant production of CXCL9 and CXCL10. We believe that this might be related to the technical limitations of the in vitro model that prevents IDC from additional activation signals such as IFN-γ, which are known to be required for the expression of these two chemokines³³.

CCL5, CXCL9 and CXCL10 are known to be involved in the recruitment of T cells and NK cells to the site of infection^34,35. In the present study, we concluded that the reduction in the number of NK cells observed in SIGNR1-KO animals was due to the lower expression of these cytokines as a result of the lack of activation of tracheal IDC. However, due to limitations in the in vivo model of infection, such as the absence of a mutant influenza virus lacking glycosylation, we cannot rule out that an alternative mechanism, independent of SIGN-R1 binding to viral proteins, might be responsible for the observed phenotype. The importance of NK cells during viral infection has been previously demonstrated in humans³⁶, but also in mice, where the absence of NK cells was associated with a higher susceptibility to influenza virus³⁷. In this study, we postulate that newly recruited NK cells perform effector functions intended to control the viral spread of the virus. In support of this hypothesis, we detected a robust expression in the recruited NK cells of the cytotoxic receptors NKG2D and NKp46, which are involved in the recognition of viral-associated molecules and the activation of these cells³⁸.

Furthermore, in CCR2-KO and SIGNR1-KO we observed a reduced infiltration of NK cells expressing IFN-γ, granzyme B and perforin. These molecules are critical for the effector function of NK cells^39–42. However, we also observed a decrease in the expression level of some of the previous markers, which indicates that SIGN-R1+ IDC are contributing to the activation of NK cells during influenza infection. Regarding the way in which this may occurs, previous studies have demonstrated that IL-18 plays an important role in NK cell activation⁴³ and IDC have the capacity to produce IL-18⁴⁴. Here, we have reported reduced levels of IL-18 in infected SIGNR1-KO and CCR2-KO mice. Therefore, we believe that IDC, secrete IL-18 that promotes the activation and effector function of NK cells during influenza infection.

The increase in the viral titres observed in CCR2-KO and SIGNR1-KO mice might be associated with a deficiency in the activation and recruitment of NK cells in these mouse models. This data supported the role of IDC in controlling the initial replication of the virus, as suggested by previous studies with influenza-infected CCR2-KO mice⁴⁵. Interestingly, CCR2-KO mice exhibited increased viral levels and mortality after infection compared to SIGNR1-KO mice, which coincided with a reduced chemokine production and NK cell activation. We believe that this might be related with the severe reduction in IDC numbers in this strain rather than a deficiency in other cell subtypes or immune mechanisms. IDC are known to also produce other cytokines such as TNF-α and NO⁸ which limit influenza virus infection^46,47. Conversely, other reports have observed a lower susceptibility of CCR2-KO mice compared to wild-type animals after infection with influenza virus^48,49. This discrepancy has been proposed to be related to the differences in the experimental conditions, such as the concentration of the virus used for the initial infection⁵⁰.

In conclusion, we demonstrated that IDC play an important role in the initial events of the immune response against influenza infection through the direct recognition of viral particles by SIGN-R1 and the secretion of chemokines that recruit and activate NK cells in the tracheal mucosa. Consequently, NK cells contribute to the early elimination of virus-infected cells in the trachea (Supplementary Fig. 6c).

Methods

Mice

C57BL/6 mice were bred in-house or acquired from Janvier Labs. SIGNR1-KO³² mice were kindly donated by Dr. Olivier Neyrolles. The following transgenic strains with a C56BL/6 background were used: CD11c-YFP⁵¹, CCR2-KO⁵², IFNAR-KO⁵³. Mice were maintained in specific pathogen-free facilities at the Institute for Research in Biomedicine, Bellinzona. Experiments were performed in accordance with the Swiss Federal Veterinary Office guidelines and animal protocols were approved by the local veterinarian authorities.

Antibodies

The fluorescently labelled antibodies for cell surface phenotypic staining, cell sorting, and intracellular staining are listed in Supplementary Table 1. In addition, the following depletion antibodies were used: for NK cells, αNK1.1 (PK136, BioXcell) and mouse IgG2a, κ (C1.18.4, BioXcell) as isotype control. The following antibodies were used for ELISA: FI6 (HUMABS BioMed); alkaline phosphatase AffiniPure goat anti-human IgG, Fcγ fragment specific (Jackson Immunoresearch); human anti-F (RSV) IgG1 (MPE8, HUMABS BioMed) as isotype control; and alkaline phosphate-conjugated goat anti-mouse total IgG, IgG1, IgG2a, IgG2b, IgG3, IgM and IgA antibodies (Southern Biotec). For the SPR assays, mAb H36-7¹⁷ was used as positive control.

Virus production, generation of recombinant influenza virus, infection and survival assay

The influenza virus strain A/PR/8/34 was grown, purified, inactivated and labelled as described previously¹². To generate the recombinant influenza virus PR8 carrying a mCherry reporter (rPR8-NS1-mCherry), a plasmid encoding the NS1 of PR8 fused with mCherry gene was constructed by overlapping fusion PCR. Briefly, the NS1 ORF without the stop codon was fused to the N- terminal of a mCherry via a GSGRRQAGGT linker region (NS1-mCherry). The mCherry was followed by a short GSGSG linker, a 19-aa 2A autoproteolytic site (ATNFSLLKQAGDVEENPGP) derived from porcine teschovirus-1 and by the NEP ORF. The first 147 nucleotides of the NEP were silently mutated. Also, silent mutations in the endogenous splice acceptor site in the NS1 ORF were introduced to prevent splicing. The rPR8-NS1-mCherry virus was rescued using standard reverse genetics techniques. Age-matched (six to eight-week-old) female mice were anesthetised with a mix of ketamine (100 mg/kg bodyweight, Parke Davis) and xylazine (10 mg/kg bodyweight, Bayer) and intranasally inoculated with 40µl (20µl on each nare) containing 200 PFU of influenza virus. In survival studies, mice were monitored daily for up to 12 days and sacrificed when weight loss was superior to 20%. For NK cell depletion 200 µg of the depletion antibodies αNK1.1 (see antibody section) was administered intraperitoneally at day 2 before infection and at day 1 after infection. For analysis of influenza-specific antibodies, mice were infected with 70 PFU of influenza virus. The human RSV virus (A2 strain) was obtained from ViraTree, propagated and expanded in HEp-2 cells. The MPV virus (strain A1/6621/01) was propagated and expanded in LLC-MK2 cells.

Flow cytometry

Trachea was collected from the larynx to the tracheal bifurcation, mechanically disrupted with scissors and digested for 45 min at 37°C in an enzyme mix composed of: DNase I (0.28 mg/ml, Amresco), and 0.26 units / ml of Liberase TL Research Grade (Roche) in RPMI 1640 Medium (Gibco) followed by a stop solution of 2 mM EDTA (Sigma Aldrich) and 2 % heat-inactivated filter-sterilised Fetal Calf Serum (Thermo Fisher Scientific) in PBS (Sigma Aldrich). Single cell populations were obtained by forcing the remaining tissue pieces through a 40-µm strainer followed by lysis of red blood cells. Fc receptors from the isolated cells were blocked (αCD16/32, Biolegend) followed by surface staining and analysis by flow cytometry on a LSRFortessa™ (BD Biosciences). Where indicated, intracellular staining was performed according to eBioscience™. Intracellular Fixation & Permeabilization Buffer Set (eBioscience) following the manufacturer’s instructions. Dead cells were excluded using ZombieAcqua fixable viability dye (Biolegend) and data were analysed using FlowJo software (TriStar Inc).

SIGN-R1 blocking assays

Tracheas were processed as for flow cytometric analysis and tracheal IDC in single-cell suspensions were sorted using a FACSAria IIIu (BD Biosciences) accordingly with their expression markers (CD45+ / CD11c+ / MHCII+ / CD11b+, Ly6c^hi). IDC dispensed into a round-bottom 96 well-plate containing 5x10⁴ cells in 100 µl of RPMI 1640 Medium (Gibco) per plate and cultured for 2 h at 37°C. Following this, cell suspensions were incubated with 1 µg of fluorescently labelled αSIGN-R1-blocking antibody or isotype control antibody for 30 min on ice (see antibody section). After washing, cell suspensions were (apart from Negative Control (NC) wells) incubated with 5x10⁵ PFU of DIO-fluorescently labelled UV-inactivated PR8 (PR8-DIO) for 1 h on ice. Finally, after washing, IDC suspensions were surface stained using 1 µg of fluorescently labelled αSIGN-R1-blocking antibody, fixed and analysed by flow cytometric analysis.

Immunohistology/microscopy of tracheal explants

Immunofluorescence confocal microscopy was performed using a Leica TCS SP5 confocal microscope, using sequential acquisition to limit the signal crosstalk. For immunofluorescence images shown in Fig. 2, 5, Supplementary Fig. 1 and Supplementary Fig. 2, tracheas were collected and fixed in 4% paraformaldehyde at 4°C for 1 h. Then, in case of that the epithelial cell layer was stained, a solution containing 10 µM of CellTracker™ Orange CMTMR (Thermo Fisher Scientific) or 5 µM Cell Trace™ Violet (Thermo Fisher Scientific), was placed in the intraluminal space of the trachea and explants were incubated for 30 min at 7°C and washed 3 times with PBS. After that, they were blocked with proper sera and stained with the indicated antibodies in Triton 0.01% + BSA 1% PBS for 4 h at 4°C with shaking (see antibody section). Tracheas were cut in two halves along the long axis. Pieces were placed on a microscopy slide and embedded in Fluoromount™ Aqueous Mounting Medium (Sigma Aldrich). 3D images were acquired using a HCX PL APO CS 20X/0.7 oil immersion objective, z-step was 1 μm for a total depth ranging from 50 μm to 112 μm. The cell populations were identified assessing the signal intensity in the different channels as follow (or explained in the relative caption for Fig. 2 and Supplementary Fig. 2). In Fig. 5k IDC were defined as CD11c+ / Ly6c+ / NK1.1-, NK cells were defined as CD11c- / NK1.1+. The basement membrane was identified (in the absence of additional specific staining) by manually localising the position of collagen fibres that generated autofluorescence signals detectable in the 412-464 nm channel after excitation with a 405 nm laser during the imaging process. For every reconstruction 10-40 cells of each cell type were analysed. 3D surfaces shown in Fig. 2, 5 and Supplementary Fig. 2 were reconstructed using the Imaris software (Bitplane). Counting and localisation with respect to the collagen layer of basement membrane (Fig. 2d, 5k and Supplementary Fig. 2b) was performed manually using a Cell Counter plugin in FIJI software⁵⁴. In Supplementary Fig. 6a, the average distance between NK and the nearest neighbour cell in the IDC population was calculated measuring the euclidean distance of the centres of the identified cells. In Supplementary Fig. 6b, contact between cells was calculated as described⁵⁵ and was considered when their distance (closest voxels) is less than or equal to 5 µm.

SINGR1-PR8 microscopy colocalisation

In Fig. 4, IDC were isolated from tracheas as mentioned in the previous section and cultured in RPMI 1640 medium in μ-Slide 8 well ibiTreat (ibidi) for 3 h at 37°C to let them attach to the bottom of the slide wells. Cell suspensions were subsequently incubated with 5x10⁵ PFU of PR8-DIO for 1 h on ice. After washing, IDC suspensions were surface stained using 1 µg of fluorescently labelled αSIGN-R1 antibody. Finally, cell were fixed, embedded with Vectashield® mounting media with DAPI (Vector Laboratories) and examined by immunofluorescence confocal microscopy. 3D images were acquired using a HCX PL APO lambda blue 63.0 x 1.40 oil immersion objective, z-step was 1 μm for a total depth 20 μm. 3D surfaces shown in Fig. 4 was reconstructed using the Imaris software (Bitplane).

Recombinant proteins and enzymatic treatment

Recombinant mouse SIGN-R1 Fc chimera protein (SIGN-R1-Fc), extracellular domain was obtained from R&D systems. HA from Influenza A H1N1 (A/Puerto Rico/8/34), Influenza A H1N1 (A/New Caledonia/20/99), Influenza A H3N2 (A/Victoria/361/2011) and Influenza A H5N1 (A/Vietnam/1194/2004), and Neuraminidase from Influenza A H5N1 (A/Hubei/1/2011) were obtained from Sino Biological. Hemagglutinin HA from Influenza A H1N1 (A/Puerto Rico/8/34) was used for glycosylation assay using Endo Hf (NEB) or PNGase F (NEB) using the manufacturer’s instructions.

Western blotting

Recombinant HA were treated or not by glycosidase and boiled in reducing SDS-PAGE loading buffer and subjected to SDS-PAGE and western blotting. Detection was performed with an anti-His tag antibody (Thermo Fisher Scientific, clone 4E3D10H2/E3) in 1% BSA for 2 hours and secondary Goat anti-Mouse IgG (H+L)-HRP conjugated (Thermo Fisher Scientific).

SIGN-R1 ELISA

Maxisorp (Nunc) of ELISA plates where coated with 50ul of UV-inactivated virus (1mg/ml of PR8, RSV and MPV), recombinant HA (2µg/ml) or N1 (2µg/ml) at 4 °C overnight. The plates were washed with PBS, blocked with 1 % BSA in PBS for 1 h at room temperature, then dilutions of recombinant mouse SIGN-R1 Fc chimera protein (SIGN-R1-Fc, R&D systems). For plates coated with N1, Fl6 antibody and the isotype control human anti-F (RSV) IgG1 (see antibody section) were added to the wells. After incubation for 1 h at room temperature, plates were washed, and 50 μl of alkaline phosphatase AffiniPure goat anti-human IgG, Fcγ fragment specific (see antibody section) diluted 1:1000 in 1% BSA/PBS was added to each well. After 1 h incubation at room temperature and washing, 50 μl of the solution (1.0 mg/ml pNPP, 0.2 M Tris buffer) prepared with SIGMAFAST™ p-Nitrophenyl (Sigma-Aldrich) phosphate tablets in deionized water was added to each well, followed by incubation at room temperature for 45 min. Absorbance was read on a microplate reader Biotel PoweWave 340 (Biotek) at 405 nm. EC₅₀ were inferred from fitting binding curves by GraphPad Prism 7 software.

Influenza-specific serum antibodies ELISA

After infection, blood was collected at 5 and 10 d.p.i. and serum was obtained. Enzyme-linked immunosorbent assays were performed as previously described⁵⁶. Absorbance was read on a microplate reader Biotel PoweWave 340 (Biotek) at 405 nm.

Tracheal Cytoplex assay

LEGENDPlex™ assays (Mouse Proinflammatory Chemokine Panel and Mouse Inflammation Panel; Biolegend) were performed to monitor cytokine/chemokine expression. Briefly, tracheas were collected and the luminal side was washed five times with 100 µl of ice-cold PBS. Tracheal washes were centrifuged at 1,500 rpm for 5 min and the supernatant was collected. 25 μl of supernatant were processed following the manufacturer’s instructions. Samples were analysed by flow cytometry on LSRFortessa™ (BD Biosciences) and data were analysed using LEGENDPlex™ software (Biolegend).

Generation of IDC and in vitro culture with influenza HA

IDC were generated by intranasal administration of 40 µg of Poly (I:C) to mimic the replication intermediates present in cells infected with RNA viruses in the lungs⁵⁷. After 2 days of administration lungs were harvested, IDC were sorted as mentioned in the previous section and 5x10⁴ cells were plated per well into a round-bottom 96-well plate in 65 µl complete RPMI media alone or RPMI+ Inflexal V® (60 µl of media+ 5µl of vaccine solution) media. After 12 h, 25 µl of supernatant was used to perform Mouse Proinflammatory Chemokine Panel LEGENDPlex™ assay.

RQ-PCR

For cytokine expression in whole tissue, trachea were collected at the specified time after infection in 700µL of TRIzol™ Reagent (ThermoFisher Scientific), disrupted in lysing matrix D 1.4 mm ceramic sphere tubes using FastPrep®-24 tissue disruption (MP Biomedicals) and RNA was isolated using a RNAeasy Mini kit (Qiagen). For analysis of cytokine expression of IDC and epithelial cells, tracheas were collected at 0 and 3 d.p.i., single cell suspension obtained, IDC (CD45+ / CD11c+ / MHCII+ / CD11b+, Ly6c^hi) and epithelial cells (CD45- / EpCam+) sorted as mentioned before and RNA was isolated using a RNAeasy Mini kit (Qiagen). Due to the absence of IDC at 0 d.p.i., RNA was extracted from uninfected tracheal DC (CD45+ / CD11c+ / MHCII+). After RNA isolation, 0.5-1μg of cDNA was synthesised using a cDNA synthesis kit (Applied Biosystems) following the manufacturer’s recommendations. For the RQ-PCR reaction, a SYBR® Master Mix (Applied Biosystems) was used and samples were run on a 7900HT Fast Real-Time PCR System (Applied Biosystems). To measure the expression levels of the cytokines IFN-α, IFN-β, CCL2, IL-18, CCL5, CXCL9 and CXCL10 a sets of primers listed in Supplementary Table 2 were designed. The Pfaffl method⁵⁸ was used to calculate the relative expression of the transcripts.

Viral Titres

Influenza titres from trachea homogenates were measured by a 50% tissue culture infective dose (TCID₅₀) assay. Briefly, tracheas were aseptically removed from mice, weighed and disrupted in 1ml of ice-cold sterile PBS. The determination of TCID₅₀ was carried out using 96-well plates containing confluent Madine-Darby canine kidney (MDCK) cell monolayers. The MDCK cells were incubated with serial 3-fold dilutions of influenza virus culture supernatant in infection medium (Minimum Essential Medium Eagle+ GlutaMAX ™-I [Gibco] with antibiotics and 1µg/mL of TPCK-Treated Trypsin [Sigma Aldrich] and without FCS) at 37°C. After 1 h, the monolayer was rinsed with PBS, overlaid with infection medium, and incubated at 37°C for 4 days. To identify influenza virus-positive wells, the monolayers where stained with Crystal Violet (Sigma Aldrich) in 70% methanol. Titres were expressed as the dilution of tracheal extract at which 50% of the MDCK cultures revealed virus growth, as calculated by the Spearman and Karber method.

Surface Plasmon Resonance (SPR)

The experiments were carried out at 25°C on a ProteON XPR-36 instrument (Bio-Rad Laboratories) in PBS (GIBCO, Thermo Fisher Scientific) and 0.05% Tween-20 (Sigma). The indicated proteins were immobilised on a GLM sensor chip surface through amine coupling at 100 nM and a blank surface with no protein was created under identical coupling conditions for use as a reference. Analyte proteins (HA with different glycosylation pattern), were injected at a flow rate of 100 μl/min, at concentrations of 100, 75, 50, 25 and 12.5 nM in different sensor channels. The data were processed using Proteon Manager Software and double-referenced by subtraction of the blank surface and buffer-only injection. k_on, k_off, and K_D were calculated by two-state fitting.

Statistics

For statistical analyses and data presentation, Prism 7 (GraphPad Software) was used. Group comparisons were assessed using the unpaired two-tailed Student’s t-test. For statistical analysis between survival curves the Mantel-Cox method was used.