Respiratory tract infections with influenza A or COVID-19 can lead to severe respiratory illnesses due to dysregulated inflammatory responses, including overshooting production of proinflammatory cytokines such as TNFa (so-called “cytokine storm”) by cells of the innate immune system. Interstitial lung macrophages play a particular role as elaborators of pro-inflammatory cytokines, as targets of COVID-19 viral infection, and as cellular organizers of the local healing response1,2. On the other hand, TNFa production by macrophages has been associated also with increased viral control after influenza infection, highlighting the need to balance innate immune activation for viral clearance with that of control for overshooting inflammation3–7. Despite recent advances in understanding of inflammatory pathways, identifying control mechanisms of pulmonary inflammation, especially those associated with viral infections, remains an urgent need where poor prognoses are often associated with dysregulated inflammatory responses8–13.
B cells are known foremost for their antibody production14–16, a function critical for controlling respiratory tract pathogens. Yet they can act also as antigen-presenting cells and produce cytokines and metabolites. GM-CSF and IL10-production by B cells were shown to regulate early immune responses, including in the lung17–24. Production of metabolites and neurotransmitters g-aminotubiric acid (GABA) and acetylcholine (ACh)25–29,30,31 are additional but less well understood effector functions of B cells. ACh is of interest, as it is a neurotransmitter generated from choline and acetyl coenzyme A via action of choline acetyltransferase (ChAT)32,33 that functions as both, controller of autonomic body functions34–38 and as immunoregulator, functions that are increasingly explored therapeutically39–43,44–53. ChAT-GFP reporter mice demonstrated ChAT expression by a variety of leukocytes31,54–60 and T cell derived ACh was shown to regulate macrophage function in the spleen 55,56. However, the most abundant ChAT expression among leukocyte in the steady state was seen among B cells31, yet their immune modulatory role is little explored. Here, ChAT-expressing B cells are identified as critical immediate early modulators of interstitial lung macrophages during influenza virus infection.
TNFa regulates viral loads and is a target of ACh
During viral respiratory tract infections, interstitial (IMs) and alveolar macrophages (AMs) are critical cellular components of the innate line of immune defense1. Overshooting activation can cause a local and systemic cytokine storm with detrimental effects on host survival, while following insufficient activation of the innate response or infection with high viral loads may result in a failure to control virus replication, leading to enhanced CD8 T cell-mediated immunopathology potentially resulting in host death61–68.
To assess the effects of ACh on lung macrophage inflammatory responses, we applied ACh to either total lung leukocytes or enriched lung-derived IMs, which were also stimulated with LPS. Application of ACh to lung cell suspensions led to a more than 50% reduction in the frequencies of TNFa generating CD64 + F4/80 + total macrophages, as well as CD11b- CD11c + SiglecF + AMs and CD11b + CD11c- SiglecF- IMs (Fig. 1a). Consistent with a direct effect on macrophages, ACh reduced TNFa production by enriched lung IMs in a dose-dependent manner (Fig. 1b). Next, ACh and its stabilizer, the acetylcholinesterase (AChE) inhibitor pyridostigmine bromide (PB) were applied intranasally to mice 12h prior to, at the time of, and 24h after intranasal infection with influenza A/Puerto Rico/8/34 (A/PR8) (Fig. 1c). In this sublethal viral infection model, virus replication and innate-driven inflammation peaks within the first 2–3 days, and virus-dose dependent weight loss peaks around 7–9 days post-infection (dpi) (Extended Data Fig. 1a). Differences in macrophage responses following intranasal ACh application were observed consistently in AMs and IMs at 24h after infection in a ACh dose-dependent manner. Intranasally applied ACh affected macrophages in both, the airways and the lung parenchyma. Application of ACh enhanced AM numbers in the lung parenchyma (Fig. 1d), suggesting their migration from the airways into the tissue. Both, the frequencies (Fig. 1e) and expression levels (Fig. 1f) of co-stimulatory marker CD86 and of MHCII by total lung macrophages decreased in an ACh dose-dependent manner, while surface expression levels of the activation inhibitor CD206 + increased (Fig. 1e). IMs showed dose-dependent decreases in CD86, MHCII, and CD64 expression (Fig. 1g). Moreover, in vitro LPS stimulation of total lung single cell suspensions from ACh-treated mice showed decreased macrophage TNFa expression levels in an ACh dose-dependent manner (Fig. 1h). Consistent with decreased inflammation, qRT-PCR analysis of lung homogenates from ACh-treated mice showed significant decreased expression of pro-inflammatory genes tnfa, il6, il1b, and chemokines ccl2, cxcl1, ccl5 and ccl7 compared to controls, while Il10, ifng, ifna1 and ifnb were unaffected (Fig. 1i). Thus, ACh can modulate lung tissue AMs and IMs, reducing their levels of activation, as well as chemokine and cytokine expression within 24h of a respiratory tract viral infection.
TNFa and Nfkb signaling pathways are known regulators of early innate responses to influenza infection69. Consistent with those findings, lung IMs but interestingly not AMs, isolated from A/PR8 infected C57BL/6 mice at 1 dpi showed increased TNFa production after in vitro restimulation compared to those from non-infected mice. Increasing the dose of influenza A/PR8 used for infection also increased TNFa expression by IMs but again not AMs (Extended Data Fig. 1b). Early elaborated TNFa significantly contributed to control of influenza A/PR8 viral replication, as in vivo blockade with anti- TNFa mAb administered prior to and at the time of infection (Fig. 1j) significantly increased lung viral loads at 1 dpi compared to infected and sham treated C57BL/6 controls (Fig. 1k).
B cells are the dominant ChAT expressing leukocyte population in the lung controlling Influenza A virus infection and TNFa production by interstitial macrophages
Given previous findings that T cell-derived ACh can modulate macrophage responses, we performed flow cytometry on pleural cavity lavage fluid, lung parenchyma, mediastinal lymph nodes (MedLN) and spleen of ChAT-GFP reporter mice31 to assess for the presence of leukocytes capable of generating ACh in the respiratory tract (Fig. 2a-d and Extended Data Fig. 2a-d). In all tissues analyzed, by frequency and total cell count, most ChAT + cells were CD19 + B cells (Fig. 2c, d and Extended Data Fig. 2c, d). In pleural cavity and lung parenchyma, nearly 50% and 10% of B cells, respectively, expressed ChAT (Fig. 2b). Unsupervised clustering of flow cytometry data revealed distinct clusters of ChAT-expressing B cells in the lung and pleural cavity (Extended Data Fig. 2e-g). ChAT + B cells were predominantly CD5+/-, CD19+, CD43+, IgMhi and IgDlo, CD138-, thus mostly, albeit not exclusively, B-1 cells (Fig. 2e and Extended Data Fig. 2h-l), consistent with a previous study31. While B-1 cells dominated ChAT expression in the pleural cavity by frequency and total numbers, due to their larger total cell numbers, conventional mature B cells outnumbered B-1 cells in spleen, lung, and MedLN (Extended Data Fig. 2h-l).
In the bone marrow (BM), ChAT expression was observed only rarely among pro- and pre-B cells (Hardy fractions A-D) but was seen at increased frequencies at the immature B cell stage (B220hi IgM+ IgD−/lo CD93+; Hardy Fraction E) (Extended Data Fig. 3a, b). Similar to peripheral tissues, BM cells with a mature, B-1-like cell phenotype (CD45R+/lo CD93-, CD43+/-, IgDlo, IgM+) displayed the highest frequencies of ChAT expression in the BM (Extended Data Fig. 3a, b).
ChAT-GFP was induced in spleen FO B cells with LPS but only marginally after anti-IgM stimulation, consistent with previous data suggesting ChAT-expression was MyD88-dependent31 (Fig. 2f and Extended Data Fig. 4a). This may explain the relative large frequencies of ChAT-GFP + B-1 cells, as they respond more strongly to TLR-mediated signaling compared to conventional B cells and require MyD88 expression for survival and differentiation70. Stimulation of splenic B cells under conditions driving B cell differentiation reduced ChAT-GFP expression. This was seen following stimulation with both, LPS plus IL-4 and IL-5, as well as stimulation with anti-IgM, CD40L, IL4 and IL5 (Extended Data Fig. 4b, c). Similar results were obtained with sorted ChATneg B-1 B cells, ruling out preferential activation of ChAT negative B cells (Extended Data Fig. 4d). Consistent with these findings, plasmablasts and plasma cells obtained from respiratory tract draining MedLN of ChAT-GFP mice infected for 7 days with influenza A/PR8 lacked ChAT-GFP expression (Extended Data Fig. 4e). The data suggest an innate-like, immediate-early role for ChAT-expressing B cells, independent of further differentiation.
To determine whether B cell derived ACh regulates early respiratory tract responses to viral infections, we A/PR8 infected mice with a B cell-specific deletion of ACh (mb-1Cre+/− ChATfl/fl mice, ChatBKO). ChatBKO mice showed 10-fold reductions in lung viral loads at 1 dpi with influenza A/PR8 compared to mb-1Cre−/− ChATfl/fl control mice (Control, Fig. 2g), consistent with relatively high constitutive expression of ChAT by B cells prior to infection. While leukocyte-derived ACh effects in the spleen have been shown previously to require T cell-mediated ACh production55,56, T cell-specific deletion of ChAT (Chatfl/fl-LckCre+/−) did not affect influenza virus loads at 1 dpi (Fig. 2h), consistent with their low frequencies in all tissues prior to infection (Extended Data Fig. 2).
B cells remained the predominant population of ChAT-GFP + cells at early infection timepoints (Extended Data Fig. 5a-e). In addition, significant increases in absolute, but not relative, numbers of ChAT-GFP + B cells were measured at 1 and 3 dpi in the lungs of ChAT-reporter mice (Extended Data Fig. 5a-c). No accumulation of ChAT-GFP + T cells in the lungs were observed until 7 dpi, a time when influenza-specific CD4 T cells are known to enter the lung in large numbers and virus is largely cleared from the lungs (Extended Data Fig. 5d, e). Together these data indicate that innate signal-induced ChAT expressing respiratory tract B cells rapidly respond to innate stimulation to affect influenza A virus replication in the lung. If primed and activated ChAT + T cells affect immune responses to influenza infection via ACh release, they would not do so until later timepoints71,58.
Given the decreased control of influenza virus replication in mice in which TNFa signaling was blocked (Fig. 1j), the effects of B cells on control of macrophage function during influenza infection were evaluated by measuring TNFa production in µMT−/− mice, which lack mature B cells in all tissues, including the lungs (Extended Data Fig. 6a). Cells isolated from the lungs of control and µMT-/- mice at 1 dpi with A/PR8 showed significant increases in TNFa production by IMs but not AMs from µMT mice following in vitro restimulation compared to controls, supporting a modulating effect of B cells on IMs but not AMs immediately early after infection (Extended Data Fig. 6b). µMT−/− mice also showed increased lung monocyte infiltration and decreased AM numbers, suggesting that the absence of B cells caused enhanced inflammation and perhaps increased AM apoptosis (Extended Data Fig. 6c)72–74. Furthermore, µMT−/− mice had reduced frequencies of macrophages expressing the inhibitory receptor CD206 and decreased CD206 expression levels, but higher surface expression of activation markers F4/80, CD11b and CD64 (Extended Data Fig. 6d) and in IMs (Extended Data Fig. 6e). Thus, B cells regulate the activation state of IMs, but not AMs, in the respiratory tract immediately early after influenza virus infection.
To determine the extent to which these effects of B cells were facilitated by their secretion of ACh, A/PR8 infected ChATBKO mice were analyzed at 1 dpi. Consistent with a major role for B cell derived ACh, ChatBKO derived IMs showed increased TNFa production as well as increased expression of CD86 very similar to IMs in µMT−/− mice, while AMs were unaffected (Fig. 2i, Extended Data Fig. 6f). Moreover, qRT-PCR analysis of lung homogenates from ChatBKO revealed increased expression of pro-inflammatory cytokine genes like il6, tnfa, il1b, and csf2, while il10 showed minimal change or downregulation compared to controls. ifna1 expression was reduced in the absence of B cell derived ACh, consistent with reductions in lung viral loads in the ChatBKO mice compared to controls (Extended Data Fig. 6g and Fig. 2g).
Depletion of ChAT in B cells did not significantly affect neutrophil numbers at 1dpi with A/PR8 (Extended Data Fig. 6h), a population that was shown previously to be affected by B cell derived ACh in a sepsis model31. Slight increases in lung monocyte and neutrophil cell counts, however, were noted by 7 dpi in ChatBKO mice (Extended Data Fig. 7a). Histolopathological evaluation of the respiratory tract at 7 dpi suggested enhanced epithelial degeneration in the nasal cavities and poorer overall health of ChATBKO mice compared to the controls (Extended Data Fig. 7b, c). The analysis also showed heightened CD8 T cell infiltration into the lungs and increased NK cells in the spleen, indicating enhanced systemic inflammation (Extended Data Fig. 7d, e). This was despite similar lung viral loads of control and ChatBKO mice by 7 dpi (Extended Data Fig. 7f), suggesting insufficient control of lung inflammation as main causes of increased pathology and immune cell activation. Altogether, the data demonstrate that B cell derived ACh acts on IMs but not AMs to inhibit lung inflammatory responses to influenza infection at 1 dpi, inhibiting early control of lung viral loads, but significantly modulating production of various pro-inflammatory cytokines and chemokines and reducing the impact of the respiratory tract infection both locally and systemically.
B cells regulate innate immune cells in the lung parenchyma via ACh production
Previous studies demonstrated that immediate early influenza virus infection is controlled in part by natural IgM production, secreted mostly by B-1 cells75,76, which we show here to be a ready source for ACh. However, lack of ChAT expression by B cells did not significantly affect total or virus-binding IgM levels (Extended Data Fig. 8a, b), further supporting an antibody independent role for B cells in inhibiting IMs’ ability to secrete TNFa and in promoting viral replication. Similarly, the lack of B cell-expressed ChAT had no effect on extrafollicular plasmablast development or the frequencies of germinal center B cells (Extended Data Fig. 8c), influenza-specific IgM or IgG antibody-secreting cells in the MedLN at 7 and 14 dpi (Extended Data Fig. 8d), nor on serum influenza-specific IgM or IgG subclasses over a 4-week time-course (Extended Data Fig. 8e). Deletion of ChAT in B cells also did not affect total numbers of innate leukocytes, T or B cell subsets in the spleen, or bone marrow B cell development when compared to control mice77 (Extended Data Fig. 9a-e). The exception was a slight but significant reduction in the number of CD5 + B-1 cells in ChATBKO mice compared to ChATflx/flx Cre-negative controls (Extended Data Fig. 9f). Studies with non-floxed and Cre-expressing mb-1Cre−/− ChAT+/+ mice showed similar reductions, suggesting that this effect on B-1 cell numbers was driven by mb-1 haploinsufficiency rather than ChAT expression (Extended Data Fig. 9g)78–80,81.
To assess the impact of B cell-specific ACh generation on respiratory tract leukocytes we conducted single-cell RNA sequencing (scRNA-Seq), comparing cells from lung parenchyma of Control and ChatBKO mice prior infection (n = 4/group) (Fig. 3a). Post sample integration analysis revealed 16 distinct cell clusters within the lung parenchyma (Fig. 3b). Clusters 7, 9, 11, and 13 were CD45neg and classified as non-immune cells, and the remainder were CD45 + leukocytes (Extended Data Fig. 10a and Supplemental Data 1). Among CD45 + leukocytes, B cells were identified as clusters 0, 6, 8, and 12, while T/NKT cells were present in clusters 15, 5, 2, and 3 and NK cells in cluster 1 (Extended Data Fig. 10b-d and Supplemental Data 1). Clusters 10, 14, and 4 represented myeloid cell compartments (Extended Data Fig. 10e-h and Supplemental Data 1). Cluster 10 exhibited markers indicative of AMs, cluster 14 granulocytes, and cluster 4 monocyte/monocyte-derived macrophages or IMs (Extended Data Fig. 10e-h). The cluster 4 IMs expressed markers of activation, including tnf, socs3, cd80, and cd86 (Supplemental data 1).
No notable differences were observed in lung cell subset frequencies between Control and ChatBKO mice (Extended Data Fig. 10i). To identify potential targets of B cell derived ACh, we compared the transcriptional profiles of lung parenchyma cell clusters from Control and ChatBKO mice. Amongst CD45neg clusters, only cluster 11 showed significant differences in gene expression, as assessed by GSEA. Those differences included pathways associated with an IFN-g responses, IFN-a response and the PI3K-AKT- mTOR signaling pathway, all of which showed increases in the absence of B cell derived ACh (Supplemental data 2). Amongst the CD45pos clusters, most B cells (cluster 0, 6 and 8) and CD8 T cells lacked significant changes (Supplemental data 2). NK and NKT cells (clusters 1 and 3) showed differential expression of immune-related pathways, including lower TNFa signaling via Nfkb pathway (Supplemental data 2). Amongst the myeloid clusters, AMs (cluster 10) showed no differences, while both granulocytes (cluster 14) and the biggest myeloid cluster (cluster 4), consisting of monocytes and IMs, showed the strongest differences between Control and ChatBKO mice with statistically significant differences in expression levels of 700 genes (Fig. 3c, Extended Data Fig. 10j and Supplemental Data 2). GSEA demonstrated upregulation of several hallmark pathways in IMs of ChatBKO mice, including kras signaling, myc targets V1, and notably, the apoptosis and the tnfa signaling via nfkb pathways (padj<0.05 and padj<0.0001) (Fig. 3d, e). The apoptosis genes dap, pmaip1, anxa1, mcl1, and activation and immune-related genes pnrc1, nfe2l2, ccl4, rel, tnf, il1b, and cd83 were genes driving the difference, consistent with the flow cytometric data, suggesting that inhibition of viral replication via TNFa may occur through increased apoptosis82,83. Also consistent with the functional data obtained after influenza infection, transcriptional analysis revealed no significant differences in the subset classified as AMs (Fig. 3f, g and Extended Data Fig. 10k). Thus, the lack of B cell derived ACh affected IMs but not AMs, further demonstrating the specific impact of ChAT + B cells on monocyte/monocyte-derived macrophages/IMs.
Flow cytometry supported the gene expression differences, with IMs from non-infected ChatBKO mice displaying increased TNFa production upon short-term restimulation with LPS in vitro, both by frequency and MFI, compared to controls (Fig. 3h and Extended Data Fig. 11a-c). Notably, the scRNA-sequencing data confirmed that changes among lung macrophages were restricted to IMs, as no significant differences were observed in AMs from ChatBKO and control mice regarding frequencies or total cells of TNFa producers, albeit a slight difference in TNFa MFI was observed (Fig. 3h and Extended Data Fig. 11a-c). Further characterization revealed significant increases in F4/80 expression on IMs and subtle differences in surface expression of CD11b and CD206 (Extended Data Fig. 11b, c) in cells from ChatBKO mice. These findings indicate that B cell derived ACh significantly alters the functionality of lung IMs, but not AMs, revealing a distinct regulatory pathway by which B cells regulate inflammatory responses in the respiratory tract.
B cell derived ACh directly inhibits IMs via the α7 nicotinic ACh receptor (a7nAchR)
Given the overall ability of AMs to respond to ACh (Fig. 1a), the data suggest that the distinct and selective effects of ChAT + B cells on some myeloid cell clusters was driven by the location of B cells in the respiratory tract, rather than their differential ability to respond to ACh. Indeed, B cells are readily found in the lung interstitium, in fact they constituted the largest cluster of leukocytes by scRNAseq (Fig. 3b), but they are not present in the airways of not previously infected mice84,85. Given the extremely short half-life of ACh, this suggests that ChAT + B cells may exert a direct effect on IMs.
To assess this, allotypically marked lung IMs from CD45.1 + C57BL/6 wildtype (WT) mice were enriched to > 75% by magnetic cell separation and adoptively transferred intranasally into either CD45.2 + C57BL/6 WT controls or CD45.2 + ChatBKO, followed by infection with influenza A/PR8. 24h later, lung single cell suspensions were stimulated in vitro for 4h in the presence of LPS and Brefeldin A to assess cytokine production by the adoptively transferred macrophages (Fig. 4a). Significantly greater TNFa responses were seen from the transferred IMs placed into ChatBKO compared to those placed into Control mice (Fig. 4b). Thus, ruling out differences in IMs development and/or epigenetic changes in ChatBKO mice as a reason for their enhanced inflammatory responses following influenza infection. In further support of direct B cell – IMs interaction, confocal microscopy revealed the co-localization of B220 (CD45R) + B cells and F4/80 + macrophages in the lung interstitium of ChAT-GFP reporter mice at 1 dpi with influenza (Fig. 4c and Supplemental Fig. 3). GFP + B cells were often seen among small clusters of GFP- B cells, and in close proximity to one or more F4/80 + macrophages (Fig. 4c and Supplemental Fig. 3). These small clusters do not represent bronchus associated lymphoid tissues, which are not typically observed this early after infection of naïve mice.
Cells respond to ACh via several nicotinic and/or muscarinic cholinergic receptors. The inhibitory effects of T cell derived ACh on splenic macrophages were shown to depend on the a7 nicotinic (n)ACh receptor (R)86–92. Revealing a role for a7nAChR also in controlling inflammatory responses of lung macrophages by ACh, increased TNFa generation was observed in both AMs and IMs from 1 day influenza A/PR8 infected mice that lacked this receptor (acra7-/-) compared to controls (Fig. 4d).
Additional IM cell adoptive transfer experiments were conducted to probe further for a direct impact of B cell derived ACh on IMs (Fig. 4e-f). For that equal numbers enriched IMs from allotype disparate 45.2 + acra7-/- mice and CD45.1 + WT controls were transferred i.n. into WT CD45.1/2 double-positive C57BL/6 mice which were then infected with influenza A/PR8 (Fig. 4e). 24h later, lung cells were stimulated in vitro with LPS, demonstrating significantly enhanced TNFa production in the CD45.2 + acra7-/- IMs compared to the co-transferred WT CD45.1 + cells (Fig. 4e). In contrast, when cell tracker dye labeled WT (CTV) and α7nAChR-deficient (eF670) CD45.2 + IMs were co-transferred into ChatBKO mice, which were subsequently infected for 24h, no significant difference in TNFa production was seen between acra7-/- and WT cells (Fig. 4f). We conclude that B cells directly modulate cytokine production by lung IMs via ACh production immediately early after a respiratory tract infection.