Alveolar macrophages transport pathogens to lung draining lymph nodes

Abstract

The first step in inducing pulmonary adaptive immunity to allergens and airborne pathogens is antigen acquisition and transport to the lung draining lymph nodes (dLN). Dendritic cells (DC) sample the airways, and active transfer of antigen to the lung dLN is considered an exclusive property of migratory DC. However, alveolar macrophages (AM) are the first phagocytes to contact inhaled particulate matter. Although having well-defined immunoregulatory capabilities, AM are generally considered as restricted to the alveoli. We show that murine AM constitutively migrate from lung to dLN and that following exposure to Streptococcus pneumoniae, AM rapidly transport bacteria to this site. Thus AM, and not DC, appear responsible for the earliest delivery of these bacteria to secondary lymphoid tissue. The identification of this novel transport pathway has important consequences for our understanding of lung immunity and suggests more widespread roles for macrophages in the transport of antigens to lymphoid organs than previously appreciated.

Introduction

Alveolar macrophages (AM²) (1) are highly adapted to their role as the first cell of the immune system to encounter inhaled particulates and pathogens (2, 3). AM are not only excellent phagocytes capable of rapidly clearing large numbers of bacteria from the lung (1), but have well-defined immuno-suppressive capabilities. AM are able to suppress the responses of both T cells and dendritic cells (DC) in vitro (4-7), and several studies have demonstrated the constitutive immunosuppressive activity of AM in vivo (6, 8).

Adaptive immunity to inhaled material, including pathogens, is developed in the lung draining lymph nodes (dLN) following transport of antigen to this site. Despite their position as the primary antigen-exposed immune cell population and their range of immunoregulatory functions, AM are not commonly thought to contribute to adaptive immune responses. This is due mainly to the perceived inability of AM to migrate from the alveolar spaces to lung dLN. Instead, constitutive CCR7-dependent migration of pulmonary dendritic cells (DC) is thought to be the only mechanism by which particulate antigen is transported from the lungs to draining lymph nodes (dLN) (9, 10). While migration of AM to lung dLN has been proposed in three previous studies (11-13), two were prior to the ability to discriminate between AM and DC, and the methodologies employed were insufficient to unequivocally exclude DC-mediated transfer of labelled material. In one recent study where AM were investigated alongside DC no migration of AM was reported (9), but this study did not specifically attempt to examine the migratory potential of these cells.

Several recent studies have demonstrated that LN macrophages have a hitherto unrecognised role in the acquisition and presentation of antigen to B lymphocytes in the dLN (14-16). These studies, which examined the transport of soluble and particulate antigen to the dLN by lymph and from the circulation, demonstrated antigen acquisition by resident subcapsular sinus macrophages. However, these experiments were not designed to address the potential contribution of migratory macrophages to antigen transport. In contrast to other tissue sites however (14, 15), soluble antigens do not normally reach lung dLN through passive mechanisms (17, 18). Therefore, at early time points following intranasal challenge it is highly unlikely that particulate antigens such as non-invasive bacteria have free access to the pulmonary lymphatics. Thus, the early transport of particulates from the lung to the dLN is likely to be an active and host-cell mediated process.

Given the position of AM as the primary exposed immune cell, and the potential ramifications of an alternate mode of antigen transport on our understanding of pulmonary immunity, we have readdressed the question of whether AM exit the lung and migrate to the dLN. We demonstrate that AM constitutively migrate to lung dLN. Following pathogen challenge, AM hold a temporal advantage over DC in antigen acquisition and AM containing S. pneumoniae appear in lung dLN prior to the onset of pathogen-induced DC migration.

Methods

Mice, antibodies and bacteria

C57BL/6J and B6J.CD45.1 mice were bred in house. B6.hCD2-DsRed mice, generated at NIMR and backcrossed for 10 generations prior to intercrossing to homozygosity, were a kind gift of D. Kioussis and A. Patel (NIMR, London). Animal experiments were carried out with local ethical approval and under UK Home Office License. All antibodies were from BD Pharmingen (Oxford, U.K.), except anti-CCR7 and anti-F4/80 (eBioscience, San Diego, USA), anti-CD68 (AbD Serotec, Oxford, UK), anti-7/4 (Caltag Labs., Burlingame, USA). Biotinylated anti-Ly49B and isotype control mAb were generous gifts of Colin Brooks (University of Newcastle). The culture and intranasal delivery of encapsulated isolate of S. pneumoniae serotype 6B was as described previously(19). Where required for tracking assays, bacteria were labeled with PKH26 (Sigma-Aldrich, Poole, UK) according to manufacturers instructions.

Flow cytometry and cell sorting

BAL, lung and dLN were harvested, prepared as single cell suspensions and stained for the expression of surface and intracellular markers as previously described (19). Samples were acquired on a CyAn ADP cytometer and analyzed using Summit v4.3 software (both Beckman-Coulter, High Wycombe, UK). Staining for the expression of CCR7 (clone 4B12 or isotype control) was carried out at 37°C. All other flow cytometric staining procedures were carried out on ice. Purified AM and lung DC were obtained by initial enrichment using anti-CD11c magnetic beads (Miltenyi Biotec) prior to surface labeling and sorting on a MoFlo cell sorter (Beckman-Coulter) utilizing a cooled, automated SmartSampler, with cells collected into PBS supplemented with 30% FCS, on ice. AM and DC were sorted based on differential expression of CD11c and MHCII in conjunction with size, granularity and autofluorescence characteristics. B220 expression was used to discriminate plasmacytoid DCs. The identity of purified populations was confirmed by morphology in cytospins (500 rpm, 10 min; Cytospin (Shandon)) of sorted cells which were air-dried, methanol fixed and Giemsa-stained. Light microscopy was carried out on a Zeiss Axioplan and imaged with an Optronics CCD camera using MagnaFire software (Optronics).

Immunofluorescence and confocal microscopy

Paraformaldehyde-fixed cytospins or acetone-fixed 8 μm frozen tissue sections were stained for CD11c and MHCII using directly conjugated mAbs. Siglec-F was detected using purified anti-Siglec-F (or isotype control), followed by biotinylated anti-mouse IgG2a and visualized with streptavidin-Alexa647. Samples were mounted in ProLong Gold (Invitrogen, Paisley, UK) and imaged using a Zeiss Axioplan LSM 510 NLO confocal microscope as single optical slices (0.8-1.0 μm). Images were analyzed using Zeiss LSM Image Browser software and Adobe Photoshop CS.

Results

AM migrate to lung dLN in the steady state

Since AM share some phenotypic characteristics with other cell types, most notably with DC (20), it was critical to be absolutely precise in our phenotypic definition of AM. Therefore, before examining dLN, we first reconfirmed the phenotype of AM in naïve animals based on previously defined characteristics of AM (19, 20). In naïve mice, AM are autofluorescent, CD11c⁺ and regarded as MHCII^LO (20). We report that while the majority of autofluorescent, CD11c⁺ AM are indeed MHCII^LO (Fig. 1B; R1-gated), a fraction (5.75 ±0.96%; n=12) of AM in BAL from naïve animals express significant levels of MHCII at the cell surface (Fig. 1A, B; R2-gated). In comparison, the few DC found in the alveolar spaces were non-autofluorescent and MHCII^HI (Fig. 1B; R3-gated). Using additional phenotypic parameters, including FSC and SSC, and by examining the morphology of each population following flow cytometric sorting, the identity of MHCII^LO and MHCII⁺ cells as AM was confirmed (Fig. 1C, D and data not shown). Furthermore, not only did MHCII^LO and MHCII⁺ AM share an identical phenotype and morphology, but these differed significantly from that associated with alveolar DC (Fig. 1E).

Figure 1. Identification of AM in BAL fluid.

A, CD11c^HI cells comprise >80% of naïve BAL cells. B, Among CD11c^HI-gated cells AM were identified as predominantly MHCII^LO autofluorescent (AF) cells (R1), with a minority being MHCII⁺ (R2). DC in BAL samples were AF⁻ and MHCII^HI (R3). AM and DC were further differentiated based on forward and side scatter characteristics (not shown). C-E, Purified, giemsa-stained MHCII^LO (R1-gated) and MHCII⁺ (R2-gated) AM exhibited distinct AM morphology compared with MHCII^HI DC (R3-gated). Original magnification x630.

We hypothesized that constant exposure to inhaled antigens would be sufficient to drive constitutive migration of AM from unchallenged lungs, should this pathway exist. Using the basic parameters defined above, we examined naïve lung dLN for cells with a phenotype corresponding to AM found in bronchoalveolar lavage. Pooled, enriched CD11c⁺ cells from lung dLN were divisible initially into 3 populations based on MHCII expression (Fig. 2A). Additional phenotypic analysis indicated that the MHCII^HI and MHCII^INT populations (Fig. 2A; G1 and G2, respectively) were homogenous, whereas MHCII^LO cells (Fig. 2A; G3) were further divisible. G3-gated cells comprised a non-autofluorescent, FSC^LO, B220⁺ population likely to be plasmacytoid DC (pDC) (Fig. 2B, C; G4-gated) and an autofluorescent, FSC^HI, B220⁻ population (Fig. 2B, C; G5-gated). Cells defined by G5 confirmed that a population of dLN cells corresponding precisely to our initial 6-parameter definition of naïve BAL-derived AM (CD11c^HIMHCII^LOCD11b^LOFSC^HISSC^HI and autofluorescent) was present at low frequency in naïve lung dLN (Fig.2A-C). We refer to these cells hereafter as “migratory AM” (mAM).

Figure 2. Identification of mAM in dLN.

A, CD11c⁺ MACS-enriched cells from pooled naïve lung dLN comprised 3 distinct populations separable based on expression of CD11c and MHCII. We identified homogenous mature (MHCII^HI; G1) and immature (MHCII^INT; G2) conventional DC populations, and a mixed MHCII^LO population (G3). B, C, G3-gated cells were divisible into small, non-autofluorescent, B220⁺ cells (G4) and large, autofluorescent B220⁻ cells (G5). D-G, Populations phenotypically defined by regions G1, G2, G4 and G5 were purified and giemsa-stained. Bar, 10 μm. Original magnification x630. H, I, CD11c-enriched cells from inguinal, mesenteric and mediastinal (lung-draining) LN were examined in parallel for the presence of cells with an AM phenotype. H, Within each LN the CD11c⁺ population comprised 3 distinct subpopulations based on MHCII expression. I, Only within the CD11c⁺MHCII^LO cells of lung dLN (R3-gated)was an autofluorescent FSC^HI population corresponding to AM detectable. 3-8×10⁵ CD11c⁺ cells were analysed in each case. Representative of 3 separate experiments in which LN from 4 non-lung draining sites were examined.

Following flow sorting, morphological examination confirmed the identity of G1-, G2- and G4-gated cells as mature conventional DC (cDC), immature DC (iDC) or pDC (Fig. 2D-F). Moreover, the morphological appearance of mAM was clearly distinct from these DC populations and further demonstrated their comparability with BAL-derived AM. Together, cDC, iDC, pDC and mAM populations appeared to comprise the entire CD11c⁺ population (of 1.51 ± 0.91 ×10⁴ cells / dLN; n=12) in dLN, with each naïve dLN containing 286 ± 124 (n=9 pooled samples) mAM.

Constitutive migration of AM would be expected to selectively populate lung dLN with mAM. We therefore examined lymph nodes draining other tissue sites for the presence of cells with a phenotype corresponding to that of mAM seen in lung dLN. In none of the examined peripheral dLN, including superficial inguinal, deep inguinal and cervical LN (Fig. 2H, I and data not shown), nor in the mesenteric LN, selected as a site of mucosal drainage, was such a population observed (Fig. 2H, I). Together these data demonstrate the presence of a population of cells exhibiting the precise phenotype of AM in the lung dLN, and confirm that this population is not present in LN other than those draining the lungs.

AM in lung and dLN share a detailed and exclusive phenotype

Given the phenotypic similarities of AM and DC (20), and the known ability of lung DC to migrate to dLN (9, 10), it was critical that we identified further properties of AM to facilitate their explicit identification and to unequivocally distinguish them from other populations.

Having found that a proportion of AM in BAL express significant levels of MHCII at the cell surface, we carried out a more detailed examination of MHCII expression by lung CD11c⁺ cells. Surprisingly, flow cytometric analysis revealed constitutive expression of intracellular MHCII in AM, as well as in DC, from naïve BAL (Fig. 3A). This was confirmed using confocal microscopy, demonstrating intracellular MHCII in AM from naïve mice, but not in peritoneal macrophages from the same animals (Fig. 2B, C). The expression of intracellular MHCII is therefore a further phenotypic characteristic of AM, as well as of lung DC. Confocal microscopy examination of cells sorted from each of the 4 defined CD11c⁺ populations in lung dLN (Fig. 2) revealed that mAM shared the characteristic expression of intracellular MHCII with BAL AM.

Figure 3. AM in BAL and mAM in dLN constitutively express intracellular MHCII.

A, AM and DC from BAL and lung (defined as in Fig. 1) express intracellular MHCII by flow cytometric evaluation. Cells were fixed and stained immediately for MHCII (surface MHCII only) expression, or stained following permeabilization to detect both surface and intracellular MHCII. B, Purified CD11c^HIMHCII^LO AM from naïve BAL express intracellular MHCII by confocal microscopy, whereas, C, CD68⁺ naïve peritoneal macrophages do not. MHCII^HI cells in peritoneal samples are B lymphocytes as defined by CD19 expression (data not shown). D, The four distinct CD11c⁺ populations from naïve lung dLN (G1, G2, G4 and G5; Fig. 2A-G) were purified and assessed by confocal microscopy for surface and intracellular MHCII expression. Surface MHCII was detected using AlexaFluor-488 conjugated mAb, cells were then permeabilized and intracellular MHCII detected using mAb conjugated to AlexaFluor-546. Bar, 10 μm. Original magnification x630.

To identify additional markers which were potentially AM-restricted, rather than shared between AM and DC, we carried out whole genome Affymetrix expression profiling of AM from naïve BAL (data not shown). Among the panel of surface molecules thus obtained were Siglec-F (21) and Ly49B(22), whose expression by AM has recently been independently confirmed. In flow cytometric analyses, both Siglec-F and Ly49B were expressed by naïve BAL AM and by mAM in lung dLN (Fig. 4A), but neither molecule was expressed by other CD11c⁺ populations in either lung or dLN (Fig. 4B). As we found that AM and DC can also be differentiated based on expression of 7/4 antigen and F4/80 (see Fig. 6), we have therefore identified 12 phenotypic characteristics (CD11c⁺, surface MHCII^LO, intracellular MHCII⁺, CD11b^LO, FSC^HI, SSC^HI, autofluorescent, B220⁻, Ly49B⁺, Siglec-F⁺, 7/4⁺, F4/80⁺), in addition to morphology, shared exactly by AM in BAL and mAM in lung dLN.

Figure 4. mAM share an extended phenotype with AM.

A, B Parallel examination of (A) AM and (B) DC populations in lung and dLN for expression of Siglec-F and Ly49B was carried out by flow cytometry. Histograms show (A) gated AM and mAM, as defined above (R1+R2, Fig. 1A; and G5, Fig. 2A, respectively) or (B) total cDC in lung tissue digests (CD11c⁺MHCII^HIFSC^LOSSC^LO, non-autofluorescent cells) or lung dLN (G1+G2, Fig. 2A) stained with anti-Siglec-F, anti-Ly49B or appropriate isotype control mAb. C, (Left panels) AM purified from naïve BAL and prepared as cytospins, and (right panels) frozen sections of lung dLN from naïve wild-type animals, were stained for expression of CD11c and Siglec-F by confocal microscopy. mAM were localized by their dual expression of CD11c and Siglec-F. In the left panels the gate indicates the location of a CD11c⁺Siglec-F⁺ (mAM) cell. Adjacent DC are apparent as CD11c⁺Siglec-F⁻ cells. 12 wild-type and 6 B6.CD2DsRed lung dLN were assessed and representative images are shown. Original magnifications x630 (left panels) and x400 (right panels). D-F, Representative images from frozen sections of lung dLN from naïve B6.CD2DsRed mice stained for expression of CD11c and Siglec-F are shown to demonstrate that (D), CD11c⁺Siglec-F⁺ cells were not observed within T cell-rich (densely CD2⁺) areas of dLN, (E) CD11c⁺Siglec-F⁺ cells were not observed at the subscapsular sinus region, or within the immediate subcapsular areas (dotted line indicates edge of dLN), and (F) occasional SiglecF⁺CD11c⁻ cells, most likely eosinophils, were observed in dLN (indicated by box). All images are representative and original magnifications are x400 (D, E) and x630 (F).

Figure 6. AM rapidly transport intranasally delivered S. pneumoniae to dLN.

Mice were challenged intransally with PKH26-labelled S. pneumoniae and 2-4 h later (A) lungs (whole lung digest) and (B) lung dLN pooled from groups of 3-6 mice were examined for the presence of pathogen-containing (PKH26⁺) cells. Left plots show CD11c+ cells stained for expression of 7/4 and F4/80 antigens. 7/4⁻F4/80⁻ (G1-gated) and 7/4⁺F4/80⁺ (G2-gated) cells were further defined as DC (G3) and AM (G4) based on FSC and SSC characteristics (upper centre and right plots) and autofluorescence (not shown). G3-gated DC and G4-gated AM were subsequently examined for association with PKH⁺ bacteria, here shown against CD11b expression (lower plots). Identical gating strategies were used for (A) total lung cells and (B) lung dLN from the same animals. MHCII was not employed due to potential overlapping expression arising from S. pneumoniae challenge (see Fig. 5A). All panels are representative of at least 3 separate experiments.

We next addressed the localization of mAM in the naïve dLN. In naïve mice, Siglec-F is predominantly expressed by eosinophils (23, 24), which being CD11c⁻, do not overlap phenotypically with AM (25). As we observed no other CD11c^HI (DC) populations expressing Siglec-F by flow cytometry, and other lung and dLN resident macrophage populations are CD11c⁻, dual CD11c⁺Siglec-F⁺ expression was used as a minimal phenotype to examine mAM by fluorescence microscopy. We calculated an expected frequency of slightly over one CD11c⁺Siglec-F⁺ cell per 8μm tissue section, given a dLN diameter of approximately 2 mm and approximately 300 mAM per dLN. CD11c⁺Siglec-F⁺ cells were indeed scarce, but present in naïve lung dLN (Fig. 4C). No CD11c⁺Siglec-F⁺ cells were observed in >100 fields of DC-rich (T cell) zones of wild-type dLN (data not shown), or in >100 fields of DsRed-positive areas of dLN in mice expressing T-cell restricted DsRed (Fig. 4D). Neither were any Siglec-F⁺ cells present among CD11c⁺ cells found in proximity to the subcapsular sinus (Fig. 4E). All clearly identifiable CD11c⁺Siglec-F⁺ cells observed by fluorescence microscopy were present in non-T cell (most likely, B cell) zones. Finally, occasional CD11c⁻Siglec-F⁺ cells were found at similar or lower frequency than that of dual-positive cells (Fig. 4F). These are likely to be eosinophils, which are very scarce in naïve dLN. Thus, we have shown that AM in BAL and mAM in dLN share a much extended and unique phenotype, and that mAM appear to be most commonly observed in T cell scarce regions of dLN.

AM rapidly transport pneumococci from lung to dLN

To examine whether mAM can transport pathogens to the dLN, we used a Streptococcus pneumoniae serotype 6B infection model (19), and examined AM over the initial few hours following intranasal challenge. On establishing this model, we first examined BAL to confirm our ability to unequivocally identify AM following S. pneumoniae challenge. Surprisingly, we observed that S. pneumoniae infection resulted in elevated surface MHCII expression on a significant proportion of AM within the first 90 min post-challenge (Fig. 5A). The identity of cells within the increased MHCII⁺ population was confirmed as AM by additional phenotypic parameters, and by morphology following cell sorting (Fig. 5A and data not shown). MHCII upregulation by AM appeared to be independent of phagocytosis, as many cells without visible bacteria were MHCII⁺, and vice versa. Furthermore, we were able to identify MHCII^HI BAL DC as a distinct, non-overlapping population within BAL samples from S. pneumoniae-challenged mice.

Figure 5. Effect of acute S. pneumoniae challenge on AM and DC in BAL.

A, Plots show gated CD11c+ cells within (left) naïve BAL and (right) BAL taken 90 min following intranasal S. pneumoniae challenge. CD11c+ autofluorescent cells (AM) are divisible into MHCII^LO (R1) and MHCII⁺ (R2) populations in each case. Other phenotypic characteristics were used to confirm the identity of AM and to exclude DC. R1- and R2-gated cells were purified in each case and the morphology examined following geimsa staining (lower panels). Representative images are shown, original magnification x630. B, The proportion of MHCII⁺ AM (R2-gated in A, above) among total AM and C, the proportion of CD11c^HI MHCII^HI FSC^LO SSC^LO, non-autofluorescent DC (R3-gated in Fig.1B) among CD11c⁺ BAL cells was determined in BAL samples from naïve mice and from mice 90 min and 12 h following S. pneumoniae challenge. Bars represent mean ±SEM, and n=6-12 at each time point. Numbers over bars indicate p values vs naïve (Student’s t-test). D, Left plot shows CD11c⁺ BAL cells from naïve mice stained for expression of CD11b and MHCII. CD11b⁻ and CD11b⁺ MHCII^HI BAL DC (R1), and total AM were further analysed following staining with isotype control mAb (centre panels) or anti-CCR7 (right panels). Percentages of cells within each quadrant are shown. Plots are representative of at least 2 experiments in which pooled BAL from 6 individuals was examined.

A significantly increased proportion of MHCII⁺ AM was maintained for at least 12 h post challenge, while no significant increase in the proportion of cells phenotypically identified as BAL DC was observed over the same time period (Fig. 5B, C). At up to 4 h post challenge, those AM upregulating MHCII appeared to be previously resident AM, in that they lacked expression of CD11b associated with newly influxing AM (19).

As DC migration to lung dLN is CCR7 dependent (9), and this is the only defined pathway of cell migration from lung to dLN, we also examined the expression of CCR7 by AM. CCR7 expression was not observed on naïve BAL AM, although it was clearly detectable on a subset of DC present in the BAL (Fig. 5D). S. pneumoniae infection did not induce CCR7 expression on AM in BAL at up to 24 h following challenge (data not shown).

Having confirmed that S. pneumoniae challenge did not impair our ability to distinguish AM, the cellular association of fluorescently labelled (PKH26⁺) bacteria was examined at early time points following intranasal challenge. MHCII was excluded as a discriminatory marker in these assays due to the potential effects of S. pneumoniae challenge on MHCII expression by AM. At 2-4 h post challenge, >40% of all CD11c⁺7/4⁺F4/80⁺FSC^HISSC^HI, autofluorescent AM contained bacteria and approximately 95% of all PKH26⁺ cells in whole lung digests were AM (Fig. 6A). The remaining PKH⁺ cells were Ly6C/G⁺CD11b⁺ neutrophils and CD11c⁻7/4⁺ cells (data not shown). Significantly, no association of PKH26⁺ bacteria with lung DC was observed at this time point (Fig. 6A). Together, our data clearly demonstrate that AM hold a significant temporal advantage over pulmonary DC populations with regard to the acquisition of intranasally-delivered bacteria.

The dLN of the same animals taken 2-4 h post-challenge were examined flow cytometrically by the same criteria applied to the lung, above. At this time point the only cell population which contained PKH26⁺ bacteria exhibited the phenotype of mAM, precisely matching that of PKH⁺ AM in the lung (Fig. 6B). Based on flow cytometry data, an estimated maximum of 40-50 pathogen-bearing mAM were present in each dLN at this time point. Therefore, not only are AM capable of transporting bacteria to lung dLN, but these pathogen-bearing AM reach the lung dLN prior to significant migration of pathogen-bearing lung DC.

Discussion

This study has readdressed the question of whether alveolar macrophages, phagocytes assumed to be alveolus-restricted, can migrate from the alveolar spaces and contribute to antigen transport to the lung dLN. We have shown that cells phenotypically and morphologically identical to AM in the BAL are found in the dLN of naïve animals, indicating a constitutive migration pathway for these cells. Moreover, we show that AM acquire inhaled pathogens prior to lung DC, and that a fraction of pathogen-bearing AM migrate to the dLN.

Migration of AM to lung dLN has been proposed in three previous studies (11-13). These studies instilled labelled particles or cells into lungs, and demonstrated the appearance of label predominantly within paracortical T cell areas of dLN, to which DC migrate (18). However, the earliest two studies suffered from the inability to distinguish AM from DC, while the later study failed to address the possibility of DC-mediated transport. More recently, Jakubzick et al. clearly demonstrated the CCR7-dependent migration of lung DC to dLN (9) and no migration of AM was reported in their model. However, this study did not set out to address AM migration directly and, as the current has demonstrated, the numbers of mAM in dLN are relatively small and may have been easily overlooked.

In the dLN of unmanipulated, naïve mice we found four populations within CD11c-enriched cells (Fig. 2). Three of these corresponded to known DC populations, while the fourth was identical in phenotype and general morphology to AM found in the BAL, and has been named mAM. Importantly, we found no similar population in LN taken from other peripheral sites, suggesting that this population was not a previously unrecognised population of LN cells. Additionally, this population was absent from mesenteric LN, suggesting that the phenotype was not broadly associated with mucosal draining sites. Together these data demonstrate the specificity of mAM to lung dLN.

Having made our initial observation based on existing, defined phenotypic and morphological parameters associated with AM and lung DC, we subsequently extended the phenotypic characterization to include the constitutive expression of intracellular MHCII by AM and mAM. Flow cytometric observation of intracellular MHCII was confirmed by confocal microscopy of highly purified AM and mAM (Fig. 3). The potential explanations for constitutive expression of intracellular MHCII by AM and mAM are currently under investigation, as most macrophage populations do not constitutively express significant amounts of MHCII intracellularly. With regard to phenotype, the expression of intracellular MHCII was shared by AM and DC. However, we also identified Ly49B and Siglec-F as surface markers expressed at very similar levels by both AM and mAM (Fig. 4A). Ly49B is the only member of the Ly49 family not expressed by NK cells, but has a relatively broad distribution among myeloid cells (22). Siglec-F is predominantly a marker of eosinophils in the naïve mouse(24), although its expression by AM has recently been confirmed in passing by other groups (23, 26). Most importantly for this study, neither Ly49B nor Siglec-F were expressed by any other CD11c⁺ population in either lung or dLN (Fig. 4B). In total, we have defined a 12-parameter phenotype and a distinctive morphology shared exclusively by AM and mAM, which unambiguously distinguishes AM and mAM from any other identifiable DC or tissue macrophage population.

In a model in which labelled AM were adoptively transferred intratracheally, AM have been proposed to migrate to paracortical T cell areas of dLN (13). However, this study did not determine the nature of the cells containing label which were found in dLN subsequent to transfer. Utilising highly purified donor AM and rigorous phenotyping of dLN populations, we were unable to detect significant AM migration in adoptive transfer models similar to those of Thepen et al. (13), despite viable transferred AM being detectable in BAL up to 48 h post-transfer (our unpublished observations). The inability to detect migration of transferred AM is most likely due to the extremely low frequency of AM migration. In contrast, using un-manipulated naïve mice, mAM were detectable by both flow cytometry and confocal microscopy. Furthermore, the microscopy data suggest that constitutively migrating AM are predominantly found in B cell regions in dLN, rather than to the paracortical T cell areas (Fig. 4C-E). In this respect, our data more closely reflect the migration described for peritoneal macrophages to the subcapsular sinus and medullary regions of dLN (27). It is therefore tempting to speculate that AM migrating to dLN may interact with B cells in a similar way to that recently demonstrated for resident subcapsular sinus macrophages (14-16). However, the specific function of mAM in dLN during homeostasis remains under investigation.

We have demonstrated that AM in the lung have a significant temporal advantage over lung DC in the acquisition of intranasally-delivered bacteria (Figs. 5, 6). Moreover, S. pneumoniae challenge did not induce the influx of DC into the alveolar spaces, with no significant increase in the proportion of DC within the alveolar spaces observed up to 12 h following S. pneumoniae challenge (Fig. 5C). This contrasts with the reported increase in alveolar DC 48 h following instillation of latex beads (9) or following repeated allergen challenge in an asthma model (28). It is most likely that differences in the models used are responsible for these observations. Importantly, AM did not express CCR7 in naïve animals, nor was CCR7 expression on AM induced in the 12 h following S. pneumoniae challenge (Fig. 5D). This indicates that, in contrast to lung DC, AM migration to dLN utilises an as-yet unidentified CCR7-independent pathway.

Affymetrix whole genome analysis of naïve AM did not reveal any obvious candidate receptors which may direct AM migration to dLN (our unpublished data). However, given the low level migration of AM to dLN during homeostasis, it is possible that migratory receptor expression is restricted to a very small proportion of AM in BAL and consequently difficult to identify among unfractionated populations using this approach. Although a recent study (29) demonstrated that inflammatory stimuli enhanced macrophage migration to lymph nodes following i.v. transfer, factors directing macrophage migration remain relatively undefined. Further studies of enriched subsets of BAL AM are clearly required and are currently underway.

Finally, we show that AM and not DC are the first pathogen-bearing population to arrive in the lung dLN (Fig. 6B). The proportion of AM containing S. pneumoniae which reach the dLN in the first 4 h following challenge is very low (>0.1%). However, this implies an influx equivalent to 15-20% of total mAM in the initial 4 h following challenge. S. pneumoniae induce widespread apoptosis in AM as a host-protective mechanism (30, 31) regulated by nitric oxide levels (32). At present, the putative trigger allowing a specific AM to avoid local apoptosis, and instead initiating migration remains unidentified at present. Moreover, it has proved difficult to further quantify the turnover of mAM during homeostasis and following pathogen challenge due to the limited cell numbers involved. Nevertheless, the kinetics of AM migration across alveolar epithelium to dLN are in broad agreement with similar studies of inflammatory peritoneal macrophage migration to dLN subcapsular sinus (27, 33, 34). While it is not possible to formally exclude the free passage of labelled bacteria from lung to dLN, studies examining lung physiology suggest this is highly unlikely at early time points after challenge (17, 18). Moreover, should free transport occur, one may expect to find bacteria associated with multiple populations, including subscapsular sinus macrophages (14), and we do not find this to be the case (data not shown). Our data support a model in which pathogen-bearing AM reach the dLN prior to DC, providing a potential window of opportunity in which mAM may exert influence over downstream immune responses. How this may occur, and how any influence may be manifest, will be the subject of future studies.

In conclusion, we have demonstrated not only the constitutive migration of AM, but also a novel pathway for bacterial transport from lungs to dLN, operating before lung DC have the opportunity to acquire bacteria. Identification of this pathway for bacterial transport may influence our understanding of pulmonary immunity and may open new avenues for manipulating the pulmonary immune response to infection and other inhaled particulates.