Recruitment of dendritic cell progenitors to foci of influenza A virus infection sustains immunity

Protection from infection with respiratory viruses such as influenza A virus (IAV) requires T cell-mediated immune responses initiated by conventional dendritic cells (cDCs) that reside in the respiratory tract. Here, we show that effective induction of T cell responses against IAV in mice requires reinforcement of the resident lung cDC network by cDC progenitors. CCR2-binding chemokines produced during IAV infection recruit pre-cDCs from blood and direct them to foci of infection, increasing the number of progeny cDCs next to sites of viral replication. Ablation of CCR2 in the cDC lineage prevents this increase and results in a deficit in IAV-specific T cell responses and diminished resistance to re-infection. These data suggest that the homeostatic network of cDCs in tissues is insufficient for immunity and reveal a chemokine-driven mechanism of expansion of lung cDC numbers that amplifies T cell responses against respiratory viruses.


Introduction
Conventional dendritic cells (cDCs) in tissues act as immune sentinels to detect viruses and microbes and migrate to draining lymph nodes to initiate innate and adaptive immune responses against pathogens (1). Although the number of cDCs in any given tissue is small, the cells can form a semi-continuous network at mucosal surfaces that allows surveillance of a large surface area, increasing the likelihood that pathogen entry will not go undetected. However, it is unknown whether this pool of resident cDCs in the tissue suffices for effective immunity or whether the cells need to increase in number at sites of infection in order to maximise local sampling of pathogen antigens and sustain induction of adaptive immune responses in lymph nodes. An inflammation-dependent increase in cDC numbers in tissues has been reported, including during corneal infection with herpes virus (2), chronic Salmonella infection (3), Plasmodium infection (4), cerebral ischemia and infection with pneumonia virus of mice or influenza A virus (IAV) (5)(6)(7)(8). The latter is of particular interest as IAV is a major human pathogen and often used as a model for studying immunity to respiratory viruses (9). cDCs are derived from proliferating pre-cDCs that leave the bone marrow (BM) and migrate via the blood to seed tissues (1,10,11). There, pre-cDC1s and pre-cDC2s terminally differentiate while maintaining residual mitotic activity, giving rise to small clones of sister cDC1s and cDC2s (7). The inflammation-dependent increase in tissue cDCs could reflect an increase in proliferative capacity but, in the case of IAV infection, proliferation does not account for local cDC expansion, which correlates instead with accelerated BM pre-cDC egress into the blood and an increase in pre-cDC influx into the lungs (7). Similarly, a population termed inflammatory cDC2s (inf-cDC2s), presumably derived from pre-cDC, was recently shown to be recruited to lungs after infection with pneumovirus in mice (5). Thus, inflammation can expand the network of tissue-resident cDCs by augmenting recruitment of blood-borne progenitors, which we have termed "emergency cDCpoiesis" (7). However, how pre-cDCs are recruited to infected lungs, whether they home to infection sites and the significance of emergency cDCpoiesis for antiviral immunity remain unclear.
Here, we set out to understand the mechanisms that underlie lung recruitment of pre cDCs during IAV challenge and assess its impact on adaptive immunity to the virus. We report directional recruitment of pre-cDCs to specific foci of IAV infection in lungs. This directional recruitment is depended on pre-cDC expression of the chemokine receptor CCR2, which was upregulated in BM pre-cDCs upon IAV infection, enabling them to respond to CCR2 ligands. Of these ligands, we found that CCL2 is mainly produced by activated monocytes, which are positioned at infection foci and thereby contribute to reinforcement of the cDC network next to the antigen source. Notably, ablation of CCR2 expression in the cDC lineage selectively compromised expansion of the cDC network in IAV-infected lungs and limited the size of the migratory cDC (mig-cDC) population in lung draining lymph nodes (LNs). This resulted in diminished T cell responses to IAV, compromised establishment of a T cell memory population and reduced protection from re-infection. These findings suggest that the number of resident cDCs in tissues is insufficient for sustaining immune responses to respiratory virus infections in the absence of BM backup, highlighting the importance of emergency cDCpoiesis in immunity.

Defining lung pre-cDCs by flow cytometry
To assess pre-cDC recruitment to the lungs in uninfected and IAV (X31 strain)-infected mice, we first validated strategies to identify the cells in lung cell suspensions by flow cytometry (12,13). Mouse pre-cDCs are generally defined as leukocytes that are negative for many lineage-restricted markers (CD3, Ly6G, SiglecF, B220, CD19, NK1.1 and Ter119), as well as negative/low for surface expression of MHC class II, but positive for CD11c, CD135 and CD43 (Fig. 1A, gates 1-3) (14,15). CD11b is often included as one of the lineage markers used as exclusion criteria (7,12,15). However, we found that the pre-cDC gate (Fig. 1A, gate 3) contained cells that expressed variable levels of CD11b, ranging from negative, low/intermediate to high levels (Fig. 1A, gates 4). We also found that the same pre-cDC gate contained cells that expressed Ly6D (Fig. 1A, gates 4), a marker of progenitors committed to giving rise to plasmacytoid cells (PCs) (16,17).

CD11b -, CD11b lo and inf-cDC2s belong to the cDC lineage
We validated our cDC and pre-cDC identification strategy in three ways. First, we confirmed that pre-cDCs and cDCs, as identified by our gating strategy, were greatly reduced in mice lacking the essential cDC growth factor Flt3L (1), regardless of whether the mice were infected or not with IAV ( Fig. 1F and G). In contrast, numbers of CD16/32 + (CD11c + MHC-II -CD11b high ) cells that contaminate the pre-cDC gate, of MCs that fall in the CD11c + MHC-II + gate and of canonical Ly6C + monocytes were the same in Flt3l sufficient and -deficient mice ( Fig. 1F and G). Second, using cDC lineage-tracing mice (Clec9a Cre/Cre × Rosa26 tdTomato/tdTomato , abbreviated Clec9a tdTomato ; (15)), we found that all pre-cDCs, cDC1s, cDC2s as defined by our gating strategy were equally labelled with tdTomato, confirming that they all belong to the cDC lineage (Fig. 1H). In contrast, Ly6C + monocytes, CD16/32 + (CD11c + MHC-II -CD11b high ) cells and CD11c + MHC-II + MCs were poorly labelled with tdTomato but expressed CD88, a marker recently proposed to allow identification of monocytic cells (5) (Fig. 1H, I). These observations held in IAV infected mice, which further revealed the recently-described CD64 + CD26 high CD88inf-cDC2 population to be tdTomato + , as expected from a bona fide cDC type (5) (Fig. 1H, I).
Third, to test the cDC-generating capacity of CD11b lo and CD11bpre-cDC populations in an inflammatory setting, we sorted them (as indicated in Fig 1A, gates 4), as well as CD16/32 + (CD11c + MHC-II -CD11b high ) cells (Fig. 1E) and Ly6C + monocytes ( Fig. S1A; used as a control), from the lungs of CD45.1 mice at 1 day post-infection (dpi) with IAV. We adoptively transferred the sorted cells i.v. into CD45.2 congenic mice infected 1 day earlier with IAV and analysed the recipient mice at 5 dpi when cDC lung numbers peak. We observed that CD11band CD11b lo pre-cDCs generated all cDC subsets, including inf-cDC2s, in both lungs and the mediastinal lymph nodes (mdLN) that drain the lung ( Fig.  1J and K). In contrast, CD16/32 + (CD11c + MHC-II -CD11b high ) cells and Ly6C + monocytes did not give rise to cDC1s, cDC2s or inf-cDC2s ( Fig. 1J and K) but to CD88 + CD64 + macrophage-like cells ( Fig. S1B and C).
The latter were found predominantly in the lung ( Fig. S1B and C), consistent the limited capacity of macrophage-like cells to migrate from lungs to lymph node during respiratory virus infection (5). In mdLNs, cells derived from transferred CD11band CD11b lo pre cDCs contributed to both the resident and mig-cDC pool (Fig. S1D). Altogether, these data reassure that we can identify by flow cytometry most lung pre-cDCs and cDCs, including cells that have been previously ignored because of expression of CD11b or CD64. Furthermore, we confirm that CD11band CD11b lo pre-cDCs and inf-cDC2s are bona fide cDC lineage cells. CD11band CD11b lo pre-cDCs likely represent different stages of differentiation as suggested by their relative proximity to fully differentiated cDCs in UMAP analysis (Fig. 1G).

cDC subsets increase in lungs of IAV-infected mice through increased recruitment of pre-cDCs
After validating our flow cytometry strategy, we sought to analyse myeloid cell infiltration in lungs during IAV infection. Ly6C + monocyte infiltration occurred early during infection, as expected (19), giving rise to an increase in MCs at 2 dpi ( Fig. 2A). The number of PCs (gating strategy in Fig. S1A) did not change appreciably during the course of infection ( Fig. 2A). In contrast, we observed a net increase in total lung cDCs and cDC subsets that became significant at 5 dpi (Fig. 2B) and was preceded by an earlier increase in total pre-cDCs and pre-cDC1/2 subsets at 4 and 5 dpi (Fig. 2C). Uncommitted pre-cDCs were almost absent from the lungs irrespective of infection (Fig. 1E), consistent with the notion that, even though such cells can be found in BM (12,13), peripheral tissues are colonised predominantly by committed pre-cDC1s and pre-cDC2s (7). The increase in pre-cDCs and cDCs in the lung was mirrored in the lung mdLN ( Fig. S2A) but not in the spleen (Fig.  S2B). This tissue tropism suggests directional movement of pre-cDCs to the lung and lung-associated lymphoid tissues, likely guided by soluble mediators.
To investigate the latter point, we examined which chemokines were induced in the lung at 3 dpi, before pre-cDC recruitment is apparent. Using a semi-quantitative chemokine array, we found 17 chemokines increased in the bronchoalveolar lavage fluid (BALF) from infected animals compared to non-infected controls (Fig. 2D). To narrow down which might be relevant, we analysed the repertoire of chemokine receptors in pre-cDCs sorted from bone marrow and found highest expression of Ccr2 transcripts, followed by Cx3cr1 and Cxcr4 ( Fig. 2E), in line with previous observations (12). Mouse CCR2 recognises CCL2, CCL7 and CCL12, of which, CCL2 has the highest affinity for the receptor (20). Ccl2, Ccl7 and Ccl12 transcripts were all elevated in lung homogenates as measured by RT-qPCR ( Fig. 2F) after IAV infection. CCL2 and CCL12 protein were represented in the chemokine array and elevated in BALF at 3 dpi (Fig. 2D). We confirmed that levels of CCL2 protein in lung tissue and BALF increased after infection, peaking at 5 dpi ( Fig. 2G and H), the time of maximal pre-cDC influx (Fig. 2C). Interestingly, all BM pre-cDC subsets upregulated CCR2 after IAV infection, reaching highest intensity at 2dpi, which coincided with the peak of pre-cDC exit from the BM ( Fig. 2I and S2C) (7). We therefore hypothesised that CCR2 mediates lung recruitment of pre-cDCs during IAV infection.

Generation of a conditional CCR2 knockout mouse specific to the cDC lineage
To test this hypothesis, we crossed Clec9 Cre/Cre mice (15) with Ccr2 fl-eGFP/fl-eGFP mice (21). In progeny animals (Clec9a Cre/+ Ccr2 fl-eGFP/+ ), cells expressing Clec9a recombine the Ccr2 locus to excise exon 3 from the Ccr2 gene, allowing expression of enhanced green fluorescent protein (eGFP), which reports on targeted cells (Fig. S3A). Mice carrying two copies of the engineered Ccr2 locus generate a non-functional truncated CCR2 protein in Cre-expressing cells (21). Therefore, Clec9a Cre/+ Ccr2 fl-eGFP/fl-eGFP (here termed Clec9a ΔCCR2 or C9a ΔCCR2 ) can be used to ablate CCR2 in Clec9a-expressing cells and compared to Clec9a Cre/+ Ccr2 fl-eGFP/+ (here termed Clec9a WTCCR2 or C9a WTCCR2 ) littermates whose Clec9a-expressing cells express normal levels of CCR2 because the locus is haplosufficient (22). Clec9a-expressing cells include cDC1s, PCs, conventional DC progenitors (CDPs) and pre-cDCs (7,15). To identify targeted cells in Clec9a WTCCR2 and Clec9a ΔCcr2 mice, we analysed the myeloid compartments of BM and lung for eGFP + cells. Pre-cDCs and cDCs were prominent among labelled cells in the BM, both in frequency (> 75% penetrance) and mean fluorescence intensity ( Fig. 3A and B), which remained unaltered by IAV infection (Fig. 3B and S3B). As expected, monocytes were not labelled (Fig. 3A, B). Consistent with these results, a large reduction in CCR2 positivity, at its peak of expression in pre-cDCs (2dpi; Fig. 2I), was seen in all BM pre-cDC (Fig. 3C) and lung cDC subsets (Fig. S3C) from Clec9a ΔCCR2 mice when compared to the same cells in Clec9a WTCCR2 controls. In contrast, expression of CCR2 by PCs (Fig. 3C) or monocytes (Fig. S3D) was not significantly altered in Clec9a ΔCCR2 mice vs Clec9a WTCCR2 controls. Further, CCR2-dependent exit of monocytes from BM was similar in the two strains, both at steady-state and after IAV infection (Fig. S3D). We conclude that Clec9a WTCCR2 mice allow eGFP marking of pre cDCs and cDCs while Clec9a ΔCCR2 mice constitute a useful tool to selectively ablate CCR2 expression in the cDC lineage.

CCR2 mediates recruitment of pre-cDCs to infected lungs
Comparison of uninfected Clec9a WTCCR2 and Clec9a ΔCCR2 mice revealed a similar number of pre-cDCs and cDCs in the BM and lungs ( Fig. 3D-F). This is in line with a report that CCR2 is dispensable for lung seeding by pre-cDCs at steady state (23) and consistent with the low levels of CCR2 ligands in that organ in the absence of inflammation (Fig.  2F, G). We next asked next whether the reduction in CCR2 expression in pre-cDCs from Clec9a ΔCCR2 mice affects their lung recruitment upon IAV infection. Interestingly, CCR2 ablation in Clec9a ΔCCR2 mice did not affect pre-cDC BM exit after IAV infection, as BM pre-cDCs were equally reduced in mice of both genotypes (Fig. 3D). However, at 5dpi, we observed significantly fewer total pre-cDCs, including pre-cDC1 and pre-cDC2 subsets, in lungs of Clec9a ΔCCR2 mice compared to Clec9a WTCCR2 controls (Fig. 3E). This correlated with reduced infection-induced expansion in lung cDC1, cDC2 and inf-cDC2 numbers in Clec9a ΔCCR2 mice (Fig. 3F). Dimensionality reduction analysis of flow cytometry data confirmed that eGFP + cells lacking lineage-restricted markers (CD3, Ly6G, SiglecF, B220, CD19, NK1.1, Ly6D and Ter119) expressing low levels of CD11b (corresponding to pre cDCs and cDCs) increased to a smaller extent in lungs of infected Clec9a ΔCCR2 mice compared to Clec9a WTCCR2 controls ( Fig. 3G, H). This could not be explained by changes in eGFP labelling during infection as BM pre-cDCs and lung pre-cDCs were equally targeted in both genotypes regardless of infection status (Fig. 3B). PCs, Ly6C + monocytes and CD11c + MHC-II + MCs were found in equal numbers in the lungs of uninfected Clec9a WTCCR2 and Clec9a ΔCCR2 mice (Fig. 3I). Importantly, the latter two cell populations markedly increased in abundance after infection irrespective of mouse genotype (Fig. 3I), confirming that monocyte mobilization is not affected in Clec9a ΔCCR2 mice.
When examining the lung draining mdLNs, we observed that CCR2 ablation in Clec9a ΔCCR2 mice also compromised the infection-associated increase in mig-cDCs (Fig.  S3E). This is likely an indirect consequence of the decreased recruitment of Clec9a ΔCCR2 pre-cDCs to infected lungs, which results in fewer differentiated cDCs able to emigrate via afferent lymph to mdLNs, a process driven by CCR7. To establish this point (and exclude a direct effect of CCR2 ablation on the ability of lung cDC to migrate from lung to mdLNs), we measured surface expression of CCR2, CCR7 and the activation marker CD86 on lung cDCs at 3dpi, around the point of maximal cDC migration to mdLN. As expected, all cDC subsets upregulated CD86 upon IAV infection (Fig. S3F). Cells poised to emigrate also upregulated CCR7 and a clear fraction of cDC1, cDC2 and inf-cDC2 was positive for CCR7 in lungs from IAV-infected animals (Fig. S3G, H). Importantly, CCR7 + cells downregulated CCR2 (Fig. S3G, H), indicating that CCR7 and CCR2 expression in cDC subsets are mutually exclusive and implying that the CCR2 axis might actually oppose cDC migration to lymph nodes. Supporting this notion, we observed that induction of CCR2 ligands in mdLN was much lower than in lungs (Fig. S3I), suggesting that infection induces a CCR2 ligand gradient that is high in the lungs and lower in mdLN, thereby retaining CCR2-expressing cells in the infected tissue rather than promoting their migration to draining mdLNs.
Finally, we tested whether CCR2 was required for pre-cDC homing to the lungs in different types of inflammation. We infected Clec9a WTCCR2 and Clec9a ΔCCR2 mice with Nippostrongylus brasiliensis, a lung-migrating nematode that induces acute tissue injury followed by a strong type 2 inflammatory response (24). Like IAV, parasite infection induced pre-cDC recruitment to lungs ( Fig. S4A) but, in contrast to infection with the virus, this was unaffected by loss of CCR2 in cDC precursors (Fig. S4B).
Overall, these data suggest that, during IAV infection, pre-cDCs in blood follow a gradient of CCR2 ligands to find their way into IAV-infected lungs where they expand the network of cDC1s, cDC2s and inf-cDC2s. Those cDCs can then become activated to downregulate CCR2 and upregulate CCR7 in order to migrate to draining mdLNs, presumably carrying IAV antigens.

Clec9a Cre Ccr2 fl-eGFP mice can be used to visualise pre-DCs by microscopy
IAV infection of the lung is known to be heterogeneous and virally-infected cells remain localised to discrete foci (7,25). We wondered whether recruited pre-cDCs were directed to such foci. We confirmed that most pre-cDCs arriving in lungs of Clec9a Cre Ccr2 fl-eGFP mice express eGFP (Fig. S5A), as noted in BM (Fig. 3B). We therefore used this eGFP expression to visualise pre-cDCs in situ in lung tissue sections from Clec9a Cre Ccr2 fl-eGFP mice, excluding any cells that stained with AF594-conjugated anti-MHC class II, anti-B220 and anti-CD64 antibodies (AF594 + cells) to eliminate eGFP + cDCs, PCs, monocytes and macrophages from our analysis ( Fig. 4A; S5B, C). Co-staining for IAV matrix (M) and nucleoprotein (NP) identified the foci of infection, and visual inspection suggested that they contained higher numbers of pre-cDCs than adjacent areas (Fig. 4B, C). This was confirmed by quantitating pre-cDC numbers in defined lung 3D volumes with high vs low virus burden (Fig. 4D). Volumes with high virus burden also contained more MHC-II + CD64 + and/or B220 + cells irrespective of eGFP marking ( Fig. 4D; referred to as AF594 + ), likely corresponding to inflammatory cells, such as CD64 high monocytes and MHC-II + cDCs.

CCR2 specifically recruits pre-cDCs to infection foci in the lung
We then tested whether CCR2 is necessary for pre-cDC localisation at infection foci by measuring co-localisation between pre-cDC surfaces (eGFP + MHC-II -CD64 -B220 -) and IAV-infection foci (M + NP + ) in Clec9a WTCCR2 versus Clec9a ΔCCR2 lung slices (Fig.   4E). We used a form of statistical analysis that is robust within the range of population sizes observed in images and therefore unaffected by lower pre-cDC numbers in the infected lungs of Clec9a ΔCcr2 mice (Fig. 4E). Notably, the co-localisation score of pre cDCs and IAV was consistently higher in Clec9a WTCCR2 compared to Clec9a ΔCCR2 lung sections when quantified by two statistical distance metrics (Jensen-Shannon and Hellinger distances; Fig. 4F). Conversely, using the same distance metrics, the "randomness" of pre cDC positioning (measured by comparison of actual pre-cDC distribution to a theoretical distribution uniformly spread across the whole tissue) was greater in Clec9a ΔCCR2 compared to Clec9a WTCCR2 lungs ( Fig. 4F) even though the infection foci were positioned to an equally random extent between genotypes (Fig. S5D, E). In contrast to pre-cDCs, there was no difference between the two mouse strains in the number of AF594 + cells in regions of high or low IAV infiltration (Fig. S5F). These data suggest that CCR2 is used by pre-cDCs not only to migrate into IAV-infected lungs but specifically to home to the areas of active infection.

Ccl2 is predominantly expressed by activated monocytes during IAV infection
To investigate sources of CCR2 ligands that might underlie recruitment of pre-cDCs to infection foci, we used PrimeFlow analysis to quantify Ccl2 transcripts by flow cytometry. At 5dpi, more than 90% of Ccl2 + cells were of hematopoietic origin (CD45 + ) (Fig. 5A), in contrast with earlier time points at which CCL2 is produced by CD45 − epithelial cells (19). Among CD45 + Ccl2-expressing cells, the majority corresponded to Ly6C + monocytes ( Fig. 5B, C), which are recruited earlier than pre-cDCs during IAV infection ( Fig. 2A). This is consistent with the notion that monocytes recruited to inflammatory sites can produce CCL2 in a positive feedback amplification loop that recruits additional cells (26). Around 40% of Ly6C + monocytes expressed Ccl2 at 5dpi (Fig. 5D) but there was a noticeable skew in staining towards monocytes with higher CD64 expression (Fig. 5E). CD64 + monocytes correspond to those with a more activated phenotype (27) and might explain the observation that pre-cDCs co-localise at foci of infection with higher numbers of MHC-II, B220 and CD64-positive cells (see above; Fig. 4D). Overall, these data suggest that early recruited monocytes, activated adjacent to IAV infection foci, produce CCL2 that attracts not only additional monocytes but also pre-cDCs.

Acute T cell responses are impaired in Clec9a ΔCCR2 mice
To assess the immunological relevance of these observations, we examined T cell priming and viral clearance in Clec9a WTCCR2 versus Clec9a ΔCCR2 mice. In uninfected mice, we did not observe any differences in immune cell composition in spleen and lung draining mediastinal LNs (mdLNs) between Clec9a ΔCCR2 vs Clec9a WTCCR2 mice ( Fig. S6A-D). However, upon IAV infection, Clec9a ΔCCR2 mice displayed less weight loss at 8 and 9 d.p.i compared to controls (Fig. 6A), a timepoint that coincided with higher abundance of mRNA encoding IAV matrix protein in their lungs (Fig. 6B). These two observations are consistent Europe PMC Funders Author Manuscripts with the possibility that Clec9a ΔCCR2 mice mount an impaired T cell response to IAV, which reduces their ability to clear the virus at late infection timepoints but spares them from T cell-induced immunopathology causing weight loss (28).
To examine this in more detail, we re-stimulated cell suspensions taken from mdLNs or infected lungs with peptides corresponding to known H-2 b MHC class I-or II-restricted IAV epitopes. Cultures of mdLN or lung cells from Clec9a ΔCCR2 mice accumulated less IFN-γ than those from Clec9a WTCCR2 controls in response to IAV peptides (Fig. 6C,   S7A). Consistent with those findings, the frequency and number of IAV NP-specific H-2D b -restricted CD8 + T cells were reduced in mdLNs and lungs from Clec9a ΔCCR2 mice compared to Clec9a WTCCR2 controls at 7 and 10 dpi (Fig. 6D, E). In contrast, IgM and IgG antibody responses to IAV were unaffected in Clec9a ΔCCR2 mice (Fig. S7B). To rule out an intrinsic defect in T cell priming in Clec9a ΔCCR2 mice, we measured T cell responses after infection with N. brasiliensis and found that they were similar to those in control mice (Fig. S7C). Therefore, the defect in the T cell response to IAV likely reflects the fact that, in the absence of cDC network expansion around virus infection foci, the number of activated mig-cDCs carrying IAV antigens from lung to mdLNs becomes limiting and, as a consequence, T cell priming is compromised (29). Consistent with this notion, at 5dpi there was a smaller population of mig-cDCs in mdLNs from infected Clec9a ΔCCR2 mice compared to controls (Fig. S3E) and they appeared less activated as measured by CD86 expression (Fig. S7D). These data suggest that acute T cell responses against IAV are curtailed if lung cDC numbers are not sufficiently boosted via CCR2-mediated pre-cDC recruitment.
Clec9a ΔCCR2 mice are more susceptible to IAV reinfection cDCs are also key for the induction of long term memory responses to viral infection, including generating tissue resident memory T cells (TRMs) (30). When examining T cell memory formation in Clec9a ΔCCR2 mice, we found that 1 month after IAV infection the number of IAV-specific CD8 + T cells in circulation and in the lungs, including canonical CD103 + TRMs, was reduced compared to Clec9a WTCCR2 controls (Fig. 7A, B). In contrast, the quantity and neutralising ability of anti-IAV antibodies were either unaffected or, if anything, slightly increased in Clec9a ΔCCR2 mice (Fig. 7C). Importantly, re-infection of mice with a heterologous IAV strain (PR8; H1N1) ( Fig. 7D), which is not neutralised by antibodies against X31 (H3N2), induced increased acute weight loss in Clec9a ΔCCR2 mice compared to Clec9a WTCCR2 controls (Fig. 7D), which was accompanied by reduced ability to clear the re-challenge virus (Fig. 7E). Together, these data indicate that expansion of the lung cDC network by pre-cDCs recruited via CCR2 to infection foci is necessary to support the effective generation of effector and memory T cell responses against IAV that protect from re-infection (29).

Discussion
cDCs are positioned at barrier sites to ensure the early detection of infection. Upon pathogen encounter, they migrate to draining lymph nodes to prime T cells and die 1-3 days thereafter (1,31). In order to support and diversify T cell priming, the influx of tissue-derived mig cDCs carrying antigen needs to be sustained (30,32). However, the basal density of cDCs in tissues is low and it is therefore reasonable to speculate that cDC numbers need to increase at the site of pathogen challenge in order to meet dLN demand for mig-cDCs. During inflammation, monocytes can differentiate into cells that have many cDC features, but it is clear that such monocyte-derived DC-like cells are distinct from those that arise from regular cDCpoiesis and cannot substitute for them, for example in anti-viral or anti-tumour immunity (1). We have previously shown that local proliferation of cDCs and pre-cDCs is decreased during IAV challenge (7) and acute demand for cDCs at sites of infection therefore requires increased recruitment of cDC progenitors. Here, we show that, in the context of IAV infection in mice, this increased local demand is met by recruiting pre-cDCs to lung infection foci in a CCR2-dependent manner. The terminal differentiation of recruited pre-cDCs sustains cDC accumulation next to the antigen source, offsetting cDC emigration to draining lymph nodes and maintaining antigen sampling to induce robust T cell immunity.
CCR2 is generally considered a monocyte marker, although it has been previously reported on some cDC2s, especially inf-cDC2s (5,23,33). Ablation of CCR2 in cDCs in Clec9a ΔCCR2 mice, could, in theory, impact cDC migration from lung infection foci to lymph nodes. However, we show that activated lung CCR7 + cDCs downregulate CCR2 as they initiate migration to mdLN, presumably to become desensitized to the high levels of CCR2 ligands in lung tissue that would oppose such migration. As for possible CCR2 dependent mobilization of differentiated cDCs from BM to infected lungs, we cannot formally exclude it due to the lack of genetic models to specifically target cDCs without affecting their precursors (or vice versa). However, tissue seeding with cDCs is mediated by migratory precursors (pre-cDCs) rather than by differentiated cells (11). Our data indicate that expression of CCR2 by pre-cDCs does not affect steady-state colonisation of the lungs but is responsible for the increased seeding observed in inflammatory conditions. Consistent with that notion, CCR2 expression is upregulated in BM pre-cDCs during IAV infection, as are CCR2 ligands in the lung, suggesting the existence of an axis where both the receptor and ligand are induced at distant locations to attract target cells to the site of infection. The finding that CCR2 can mediate pre-cDC recruitment suggests caution is warranted when using CCR2-deficient mice or anti-CCR2 antibodies to interrogate the role of monocytes in immunity and inflammation (34,35).
Interestingly, recent work has revealed a role for CCR2 in increasing numbers of cDCs in lymph nodes draining sites of intramuscular immunisation with AS01-adjuvanted antigens (36). Global loss of CCR2 resulted in decreased CD4 + T cell, CD8 + T cell and antibody responses to the antigens, suggesting a specific role for inf-cDC2s (36). By restricting receptor loss to the cDC lineage, we find that CCR2 is required for lung increases not only in inf-cDC2s but also in cDC1s and cDC2s in a model of IAV infection and that this is essential for effective T cell-based but not humoral immunity. Which cDC type contributes predominantly to the CCR2-dependence of the T cell response remains unclear although CD8 + T cell priming to IAV in mice has been proposed to be particularly dependent on cDC1s (6,37).
It is likely that not all emergency cDCpoiesis is exclusively dependent on CCR2. Indeed, RNAseq analysis revealed that pre-cDCs in BM also express Ccr1, Ccr5 and Cxcr4, the Europe PMC Funders Author Manuscripts ligands for which are also increased in mouse lungs during IAV infection. Those receptors could potentially contribute to pre-cDC recruitment to IAV-infected lungs as accumulation of the cells was not fully abrogated in Clec9a ΔCCR2 mice (although this might also reflect incomplete penetrance of the recombination event (15)). The expression of other chemokine receptors could also explain why CCR2-deficiency does not prevent the increase in lung pre-cDCs numbers after N. brasiliensis infection. Interestingly, the egress of pre-cDCs from BM into blood in IAV-infected mice is not affected by CCR2-deficiency, which is in contrast to monocytes (39). It is therefore likely that BM exit of pre-cDCs and subsequent homing to tissues involves distinct signals and context-specific chemokines, both in steady-state and in emergency cDCpoiesis. The reliance on multiple mechanisms allied to the redundancy of the chemokine system (38) likely ensures robustness and increases resistance to pathogen interference with cDC-driven immunity.
Lung architecture presumably prevents fully-differentiated cDCs scattered around airways and alveoli from migrating through the tissue to coalesce at foci of infection. Therefore, as for other lung resident cells such as macrophages, increasing local cell density requires acute focal recruitment of precursors from the blood (39). Consistent with this hypothesis, we were able to visualise pre-cDCs clustering around IAV infected foci. Our data suggest a model in which challenge of the respiratory tract with IAV rapidly communicates to the BM a need for emergency cDCpoiesis. Pre-cDCs acutely released from BM (7) in a CCR2 independent manner circulate via blood and extravasate from lung capillaries adjacent to foci of infection in response to CCR2 ligands. These include CCL2 produced by CD64 high monocytes that were recruited by a prior wave of CCL2 expression by epithelial cells (19). Pre-cDCs then differentiate locally at sites of infection to expand the number of cDCs in close proximity to the antigen source. Newly-generated cDCs expressing CCR2 are retained at foci of infection by CCL2 until they contact virus or virus-infected cells, which act as a source of antigen and activation signals (1). Activation leads to a switch in chemokine receptor expression characterised by upregulation of CCR7 and downregulation of CCR2, allowing the antigen-laden cells to migrate to mdLNs (1,31). This is dispensable for the antibody response but sustains T cell priming and supports generation of a productive memory response that confers T cell mediated cross-strain protection to heterologous IAV strains. Additional studies will be necessary to establish to what extent pre-cDC back-up is important for immunity to other immune challenges. Such studies could reveal whether emergency cDCpoiesis is an integral part of robust immune responses to infection that could be exploited in vaccine design or, conversely, inhibited in some settings in order to dampen immunopathology caused by T cells.

Study design
The aim of this study was to examine the requirement for cDC expansion to achieve effective adaptive immunity against respiratory viruses. We used IAV as a widely-studied respiratory virus and the mouse as model organism. Mouse experiments were planned in accordance with the principles of the 3Rs (Replacement, Reduction and Refinement) following UK Home Office guidelines. The X31 strain of IAV (H3N2) was used as a mouse-adapted strain for infections and the PR8 strain (H1N1) as a re-infection virus to look at T cell-mediated protection. Different immune cells were examined in lung, blood and bone marrow by flow cytometry and fluorescence confocal microscopy. In addition, chemokines, viral content and antibodies were measured using techniques such ELISA, CBA and qPCR. T cell responses were monitored by ex vivo re-stimulation with defined IAV epitopes. All experiments were performed at least twice. During analysis no individual data points were excluded under any circumstances other than technical failure to process the sample.

Mice
Ccr2 fl-eGFP (21), Clec9a Cre (here sometimes abbreviated C9a Cre ) (15), Flt3l -/-(Taconic Biosciences), Rosa26 LSL-tdTomato (abbreviated R26 LSL-tdTomato ; The Jackson Laboratory) and C57BL/6J mice were bred at The Francis Crick Institute in specific pathogen-free conditions. All genetically modified mouse lines were backcrossed to C57BL/6J. Six to twelve-week-old male and female mice were age and sex-matched in all experiments. Sample sizes were determined so as to include the minimum number of mice necessary to achieve statistical robustness when assessing differences in pre-cDC/cDC numbers caused by influenza A virus infection (7). Mice were not randomized in cages, but each cage was randomly assigned to a treatment group. Investigators were not blinded to mouse identity during necropsy and sample analysis. Male and female mice were used to perform the experiments. However, we did not observe differences between sexes. All animal experiments were performed in accordance with national and institutional guidelines for animal care.

Pre-cDC differentiation assays
Flt3L-driven differentiation of pre-cDCs was carried out by culturing 2 × 10 4 OP9-DL1 (18) cells into twelve-well plates in RPMI medium supplemented with L-glutamine (Gibco), penicillin-streptomycin (Gibco), non-essential amino acids (Gibco), 10% FCS (Sigma) and β-mercaptoethanol (Gibco). The following day, 1-5 × 10 3 sorted cells were added to the OP9-DL1 monolayer after removing the medium and replacing it with fresh medium containing 300 ng/ml of mouse Flt3L. Progeny cells were assessed by flow cytometry three days later. cDC differentiation was assessed by MHC-II upregulation, whereas PC differentiation was quantified by the expression of B220 and SiglecH. cDC1s were defined as XCR1 + and cDC2s as SIRPα + .

Infection
Mice were anesthetised by isoflurane inhalation and were infected intranasally with 35,000 tissue culture infectious doses 50 (TCID50) of influenza A X31 (H3N2) in 30μl PBS. For heterosubtypic challenge, mice were infected with X31 influenza at the same dose followed by i.n. administration of 10,000 TCID50 of H1N1 PR8 IAV (in 30μl PBS), 28 days after primary infection. Mice were monitored daily for weight loss and signs of infection. N. brasiliensis was maintained by serial passage through rats, as described previously (24). Mice were infected subcutaneously with 250 N. brasiliensis L3 larvae.

Preparation of single cell suspensions
Spleens, mdLNs, iLNs and lungs were cut into small pieces and digested with Collagenase VIII (1mg/ml, Sigma) or VI (400U/ml, Worthington) and DNase I (0.4mg/ml, Roche) in RPMI for 15-30 min (spleen and LNs) or 20-60 min (lung) at 37°C. Digested tissues were strained through a 70μm cell strainer (BD Bioscience) and washed with FACS buffer (3% foetal calf serum, 5mM EDTA in PBS). For lung, leukocytes were enriched by Percoll gradient centrifugation (GE Healthcare) as previously described (7). For BM, femur and tibia extremities were cut and spun for 30s at 10,000 rpm. Cells were resuspended in final volume of 500μl.

Flow Cytometry analysis
Cells were preincubated with blocking anti-CD16/32 in PBS for 10 min at 4°C and then stained for 20 min at 4°C with antibody cocktail and LIVE/DEAD Fixable Blue Dead Cell Stain Kit (ThermoFisher) in PBS. Lineage (Lin) markers included CD3, Ly6G, SiglecF, B220, CD19, NK1.1, Ly6D and Ter119, unless otherwise specified. Antibodies (Abs) used for flow cytometry are listed in Table S2. PE conjugated pentamer comprising H2-D b ASNENMETM (ProImmune) was used to detect IAV NP-specific CD8 + T cells. Data was analysed using FlowJo. Uniform manifold approximation and projection (UMAP) analysis (40) of flow cytometry data were generated on the basis of CD11b, CD11c, CD16/32, CD26, CD43, CD64, CD88, CD135, Sirpα, MHCII, Ly6C, SiglecH and XCR1 expression. Annotation of clusters on the UMAP plots was done by using defining markers for each immune population. The accuracy of our manual gating was confirmed on the UMAPs by overlaying different immune populations identified as shown in Figure 1E and S2A.

RNA extraction, cDNA synthesis and RT-qPCR
mdLNs and lungs were collected in Trizol and subsequently homogenised in a TissueLyser LT (Qiagen). RNA was isolated using chloroform and precipitated with isopropanol. After Europe PMC Funders Author Manuscripts washing with ethanol, RNA was resuspended and used to synthesise cDNA using the Superscript II reverse transcriptase (Invitrogen). Quantitative real-time PCR (RT-qPCR) was performed using the TaqMan Universal PCR Master Mix (ThermoFisher) and primers (listed in Table S5). Analysis was performed on a QuantStudio (Thermo Fisher Scientific) using ΔCt quantification. RT-qPCR conditions were as per manufacturer's instructions.

RNAseq
For Figure 2E, pre-cDCs were isolated from BM using a BD FACS Aria Fusion sorter. 1-6 × 10 4 cells were sorted directly into lysis buffer to avoid loss of material. RNA was extracted using the RNeasy Mini Kit (Qiagen). NuGEN Ovation RNA-Seq System (v2) was used for cDNA synthesis followed by NuGEN UltraLow Library System (v2) for library preparation. Samples were normalised to 1 ng RNA for input and the preparation was performed according to the manufacturer's guidelines. Sequencing was performed on the Illumina HiSeq 4000, with 100 base pair single end reads. After sequencing, samples were normalised and analysed.

Microscopy
Lungs were removed and the left lobe was fixed in 4% paraformaldehyde (Electron Microscopy Sciences) in PBS overnight at 4°C. Fixed lungs were incubated in PBS with 30% sucrose at 4°C overnight and embedded in Tissue-Tek OCT compound (Sakura) at -80°C. The tissue was cut in a Leica 3050 cryostat to generate 60μm frozen sections. For antibody staining, sections were hydrated in 0.1M TRIS pH 7.4 and blocked for 1h at 25°C in 1% bovine serum albumin (Sigma), 1% normal mouse serum (Invitrogen) and 0.25% Triton TX-100 (Sigma). Sections were first stained overnight at 4°C with rabbit anti-CD64 (clone 027, Sino, 1:1000), rat anti-CD45R/B220 biotin (RA3-6B2, BD Pharmingen, 1:100), rat anti-I-A b biotin (M5-114.14.2, BD Pharmingen, 1:100) and anti IAV NP+M FITC (Oxoid, 1:100) diluted in blocking buffer. Sections were subsequently stained with anti-FITC AF647 (1F8-1E4, Jackson ImmunoResearch, 1:400) and anti-rat IgG AF594 (Invitrogen, 1:400) or anti-rabbit IgG AF594 (Invitrogen, 1:400). Sections were counterstained in Hoechst 33342 solution (ThermoFisher, 1:500) and mounted in RapiClear 1.47 (Sunjin Lab). Antibodies used for confocal microscopy are listed in Table S4. Imaging was performed in a Zeiss LSM 880 inverted confocal microscope with a 25x oil immersion objectives. Sequential excitation of fluorophores at 405nm, 488nm, 561nm and 633nm was provided by a combination of argon and helium lasers. Tile scans were acquired covering the entire surface area of the section at a step size of 3μm and a pinhole of 1 Airy unit. Images were acquired with 512x512 pixel resolution with a line averaging of 4. Tile stitching was performed using Zen software (Zeiss).

Pre-cDC localisation analysis
Three dimensional volumetric surfaces corresponding to lung pre-cDCs, IAV + and lineage + cells (lin = CD64, B220, I-A b ) were generated in Imaris 9.2 (BitPlane). Thresholding for pre-cDC surface generation was based on eGFP intensity and then an additional filter was added to exclude lin + cells. Statistics were exported and analysed using GraphPad Prism. Surface coordinates were exported and plotted in two dimensions (X and Y). Non-parametric probability density functions (PDFs) for pre-cDCs and IAV + cells in a given lung section were generated using the ksdensity function from the Statistics and Machine Learning toolbox in MATLAB (MathWorks). Boundary correction of reflection and a bandwidth of 100 were selected; additional bandwidths were tested for robustness. Statistical distance metrics (Jensen Shannon distance and Hellinger distance) for continuous distributions were used to compare the similarity between pre-cDC and IAV PDFs and determine pre-cDC/IAV co-localization (42). All measures are bounded between zero and one. To assess the randomness of the distribution of pre-cDCs and IAV + cells, these distance metrics were used to compare pre-cDC and IAV PDFs to tissue specific uniform distributions generated for each lung section based on saturated Hoechst staining. Each particular point within the stained tissue has an equal (uniform) probability of a pre-cDC or IAV + cell being located there.

Haemagglutination inhibition assay
Haemagglutination inhibition (HAI) activity was evaluated using previously described protocols (41). IAV (X31 strain) diluted in PBS to give 8 haemagglutination activity units and serum samples (diluted 1:5 in Seiken Receptor Destroying Enzyme (Cosmos Biomedical Ltd.) followed by 1:2 serial dilutions in PBS) were incubated in V-bottom 96 microtitre plates for 30 min at room temperature. Turkey red blood cells (0.75% (v/v); supplied by PHE, Colindale) were added to the antibody-virus mixture, mixed gently, and incubated for 30 min at room temperature. Wells were scored visually for haemagglutination. HAI titres were recorded as the reciprocal of the highest serum dilution that prevented haemagglutination.

Quantification of chemokines in lung homogenates and BALF
Lung homogenates were generated by mechanical tissue lysis in PBS with Halt protease inhibitor cocktail (ThermoFisher) and supernatants were stored at −80°C. For isolation of fluid from the bronchoalveolar space, mice were sacrificed and bronchoalveolar lavage was carried out by flushing with 2 × 0.6ml cold PBS using a needle inserted into the trachea. Cells were removed by centrifugation at 800g for 8 min at 4°C and supernatants were used for analysis. Chemokines and cytokines were analysed using the mouse chemokine array ARY020 kit (R&D). Altenatively, mouse CCL2 Flex Set (BD Biosciences) was used for CCL2 quantification by cytometric bead array (CBA, BD Biosciences) in lung homogenates and BALF. Events were acquired on an LSRFortessa or FACSymphony flow cytometer (BD Biosciences). Analysis was carried out using FlowJo 10 software (TreeStar Inc). Flow Cytometry Standard (FCS) files were exported from FlowJo for analysis by Uniform Manifold Approximation and Projection (UMAP) performed in R (40).

Statistical Analysis
Statistical analyses were performed using GraphPad Prism software (GraphPad) or MATLAB (Mathworks). Results are depicted as means +/-SEM. The statistical test used is specified in each figure legend. When Student's t test was perfomed, a two-side level was used and all data were checked for normality. When ANOVA was used, Tukey correction was performed.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.

One-sentence summary
Effective T cell-dependent immunity to influenza A virus requires recruitment of cDC progenitors to foci of infection via CCR2   PC-generating potential (a combination of cell recovery and subset differentiation) of these populations is shown in the left (see also  -8) and data were pooled from 2 experiments. Cells were sorted using the gating strategy depicted in Fig. 1A, excluding the subset gating step. One-way ANOVA statistical test was used comparing day 0 (non-infected mice) to each day post-infection. * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001. Not significant when not indicated.  of PCs, Ly6C + monocytes and CD11c + MHC-II + MCs in lungs from naïve and 5dpi IAV infected C9a WTCCR2 (grey) or C9a ΔCCR2 (teal) mice. Each dot in (B-F, H, I) represents one mouse (n = 5-10) and data were pooled from 2 experiments. Statistical analysis was done using a two-way ANOVA test comparing infected C9a WTCCR2 and C9a ΔCCR2 mice. * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001. Not significant (ns).   (A, B) Number of IAV NP-specific pentamer + CD8 + T cells in lung (A) or in blood (B) during the memory phase of the response to IAV infection. Lung T cells gated as in Figure 6 were separated into CD103 -(left) and CD103 + (right) populations. at the specified time points after PR8 infection. Each dot represents one mouse (n = 7-10) and data were pooled from 2 experiments. Error bars represent SEM. Statistical test was done using one-way ANOVA test in A, B, E (weight loss) and ttest in C, F (IAV matrix). * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001. Not significant (ns) or when not indicated in E.