Self-limiting Sp19F infections, resulting from two Sp19F instillations into the left lung lobe spaced 1 week apart, confer left lung-localized heterotypic immunity against an otherwise fatal Sp3 infection occurring a month or more later (12). We have previously demonstrated roles for Th17⁺ T_RM and B_RM cells in mediating heterotypic protection, with B_RM cells remaining in the ipsilaterally-experienced lobe for up to 6 months after two Sp19F infections (12, 14, 27). Mice that receive only 1 respiratory Sp19F exposure (denoted as Sp19F x1), saline, or contralateral lung Sp3 challenge are not protected (12). We hypothesized that the second Sp19F infection (denoted as Sp19F x2) is necessary for the recruitment and phenotypic diversity of B cells in the lung, including their progression to resident B_RM cells. To test this, we assembled a 22-color lymphocyte surface marker panel for full spectrum flow cytometric analysis of lungs collected at 8 different timepoints from day 0 to day 35, including groups that received either one or two Sp19F infections ( Figure 1A ) (29). Intravascular (IV) administration of a fluorescently-tagged CD45.2 antibody shortly before euthanasia of each animal allowed for distinction between extravascular (EV, ivCD45.2^-) and IV (ivCD45.2⁺) leukocytes (30). After a single Sp19F infection, lung EV B cells increased rapidly over 3 days before returning to baseline by day 10 ( Figure 1B ; gating strategy, Supplementary Figure 1A ). In mice that received a second Sp19F infection on day 7 [to facilitate local antigen presentation in the lung (12, 31, 32)], we observed another, more substantial spike in B cell numbers after the second infection, peaking at day 10, and followed by contraction from day 14 to day 35. Thus, the second infection triggers robust B cell accumulation and is essential for B cell persistence in the lungs.

We leveraged machine learning to distinguish the types of B cells in the lungs after infection. Multidimensional data extracted via expert-driven gating of EV B cells at the day 10, day 14, and day 35 timepoints after 1 or 2 pneumococcal instillations were concatenated and clustered using unsupervised Louvain clustering via PhenoGraph algorithm (33). 21 unique subpopulations were resolved and overlaid onto the optimized parameter T-distributed stochastic neighborhood embedding (opt-SNE) projection of the multidimensional EV B cell cytometry data, revealing differences in Louvain cluster densities between timepoints ( Figures 1C, D ) (34). A heatmap of normalized marker expression for each subpopulation was generated for clustering and subset identification. Unsupervised hierarchical clustering dendrograms based on the relative mean fluorescence intensity of each marker enabled categorization of the 21 clusters into 6 major populations. The lowest numbered clusters (including 1 through 5), representing the most abundant cells, were IgD⁺ naïve B cells; a total of 10 clusters fell into this category ( Figure 1E ) (35, 36). 5 clusters were IgD^-IgM⁺, two of which were further segregated by relative CD43 and B220 expression into subsets of IgM⁺ B2 and B1 cells (37). The remaining 5 clusters were IgD^-IgM^- CSW B2 cells, which could be further segregated into non-GC B2 B cells and GC/pre-GC B2 cells based on their GL7 binding (28). Clusters expressing elevated GL7 were termed as GC (CD38^lo) and pre-GC (CD38^hi) B cells (38). Whether other notable B cell populations such as CD23^loCD21/35⁺ marginal zone (MZ) B cells and CD138⁺ TACI⁺ plasma cells were present were assessed in separate cytometry experiments ( Supplementary Figure 1B ). The lung was not observed to harbor MZ B cells, and plasma cells were present but likely underestimated due to the enzyme sensitivity of plasma cell defining markers and ongoing phenotyping challenges (39, 40). Specific markers for these populations were therefore excluded. These high dimensional clustering analyses guided the supervised gating strategies of various lung B cell subsets analyzed with flow cytometry throughout this study. Further, these data reveal previously underappreciated lung B cell population heterogeneity following respiratory exposures to pneumococcus.

To explore the population kinetics of different B cell subsets as they occupied the lung, we applied manual gating strategies ( Supplementary Figure 1A ) to lung EV B cells at all tested timepoints after 1 or 2 Sp19F infections from Figure 1 and tracked quantities and frequencies of B cell subsets over time. We observed that CSW B2 B cells ( Figure 2A ), CSW B1 B cells ( Figure 2B ), and GC/pre-GC B cells ( Figure 2C ) numbers hardly deviated from baseline after only 1 infection, but all reached peak numbers at day 14 when a second infection had been administered. There were few to no CSW B1 and GC/pre-GC B cells found at later timepoints. Naïve B cells ( Figure 2D ), IgM⁺ B2 B cells ( Figure 2E ), and IgM⁺ B1 B cells ( Figure 2F ) acutely increased within the first 3 days after Sp19F re-exposure before numbers tapered. Other than innate-like IgM⁺ B1 B cells, each population experienced an exaggerated increase after repeat Sp19F instillation, whereas IgM⁺ B1 B cells populated the lung in nearly equal amounts 3 days after each exposure. In sum, we observe that the second infection leads to the accumulation of all EV B cell populations, with naïve and IgM⁺ B cells peaking earlier than GC/pre-GC and CSW B cells.

We previously used this model of anti-pneumococcal immunity to demonstrate that PD-L2⁺ MBCs are present in the lungs of experienced mice and confer anti-pneumococcal heterotypic immunity, likely via the local secretion of pneumococcus-specific antibodies (14). The presence of multiple MBC markers including PD-L2 and CD73 is associated with greater antibody affinity and likelihood to differentiate into antibody-secreting cells upon stimulation (41–43). To examine the kinetics of B_RM cell accumulation in the lung, we manually gated EV B cells bearing the memory phenotype IgD^-PD-L2⁺CD73⁺ and determined their proportions among total EV B cells across the time course. After a single infection, the proportion of EV B cells that were non-naïve and expressing PD-L2 and CD73 remained between 0-5%, whereas numbers increased sharply 1 week after the second instillation and remained elevated ( Figure 2G ). EV B cells bearing memory markers were sparse in lungs that received only 1 exposure to Sp19F at the recovery time-point ( Figure 2H ). We similarly examined proportions of EV B cells bearing the non-naive activated/resident phenotype IgD^-CD69⁺ after 1 and 2 pneumococcal infections. CD69 is a marker of resident memory in resting lymphocytes (9, 44), but is recognized as a marker of lymphocyte activation in non-resting states such as acutely after infection (45). Activated IgD^-CD69⁺ B cells increased in proportion between days 10 and 14 and remained elevated thereafter ( Figure 2I ). As in the case of memory markers, EV B cells bearing the CD69 marker of residency in resting lungs were minimally present in the lungs that received only 1 Sp19F exposure by day 35 ( Figure 2J ). These data indicate that the second infection with Sp19F in this model system is essential for a transient burst of multiple B cell subsets in the lung immediately after the second infection, as well as for the establishment of lung B_RM cells in the recovered lung.

Among the Sp19F x2 groups, three timepoints stood out. Lungs at day 10 harbored the greatest overall number of B cells, lung CSW and pre-GC/GC B cells peaked at day 14, and lungs at day 35 contained EV B cells bearing markers of B cell memory and residency (12). We chose to next focus our examination of B cell populations to these three stages post-respiratory infection.

To determine if the recruitment of lymphocytes from draining lymph nodes was necessary for lung B cell accumulation after the second Sp19F infection, we administered the sphingosine-1-phosphate receptor 1 (S1P1R) inhibitor FTY720 or vehicle treatment intraperitoneally (i.p.) before, during, and after the second infection ( Figure 3A ) and used flow cytometric analysis to quantify B cell populations at 3 days post second infection (dpsi; corresponding to day 10 in Figures 1 , 2 ), 7dpsi (corresponding to day 14), and 28dpsi (corresponding to day 35). This treatment schedule allowed effectors to be established normally in the lung and peripheral secondary lymphoid organs after the first infection but prevented lymphocyte egress from draining lymph nodes upon the second Sp19F exposure in FTY720-treated mice (46). A less comprehensive surface marker panel was used to analyze B cell populations in this experiment as we were predominantly interested in determining whether increases in lung EV non-naive B cells would still occur in the presence of a lymphocyte egress inhibitor, whether local GC reactions would take place (indicated by the presence of GC and pre-GC B cells), and whether antigen-activated B cells accumulate in the lung (indicated by CD69 at acute timepoints). Suppression of lymphocyte egress expectedly reduced circulating B cell numbers during the FTY720 treatment ( Figure 3B ). This treatment significantly reduced the total numbers of lung B cells by 28dpsi, but differences were more subtle at earlier timepoints ( Figure 3C ). Lung non-naive (IgD^-) B cells were significantly reduced from 7dpsi onwards in the FTY720-treated groups ( Figure 3D ), whereas lung naïve (IgD⁺) B cells trended towards decrease but were not statistically different at any timepoint tested ( Figure 3E ). CD69⁺ B cells were significantly reduced at 7dpsi and 28dpsi in the FTY720-treated groups, but also exhibited a trending (p=0.06) decrease acutely at 3dpsi relative to vehicle-treated mice ( Figure 3F ). Similarly, GC B cells (GL7⁺CD38^lo) were found to decrease significantly at 7dpsi with FTY720 treatment ( Figure 3G ), while pre-GC B cells (GL7⁺CD38^hi) did not ( Figure 3H ), suggesting that the initiation of local GC reactions also requires intact lymphocyte circulation during the second exposure to pneumococcus. GC and pre-GC B cell populations are scarce at 28dpsi in the Sp-experienced lung and were thus excluded from data collection at the lattermost timepoint. These data suggest that the immune cells established initially in the lung following the first Sp19F infection were insufficient in driving the phenotypic changes in the lung B cell populations associated with B_RM cell development upon subsequent pneumococcal exposure, and that continuous recruitment from the periphery during antigen exposure is necessary for lung EV B cell maintenance.

Local secretion of the follicular chemokine C-X-C motif chemokine ligand 13 (CXCL13) is necessary to recruit naïve CXCR5⁺ B cells from the periphery to form lung ectopic GCs, iBALT, and lung MBCs after IAV infection (22). On the other hand, Pseudomonas aeruginosa infection induces iBALT with CXCR4-expressing B cells attracted to CXCL12 rather than CXCL13 (21). The secretion of these chemokines is triggered by IL-17 and serves to attract B cells into follicles and stabilize contact with T follicular helper (T_FH) cells (47). Outside of homing to follicles, CXCL13 also has a role in B1 B cell compartmentalization in the pleura and peritoneum (48). Although pneumococcal infections do not support persisting iBALT, we wanted to test whether a conserved mechanism for B cell recruitment existed in this Th17 cell-driven model of anti-bacterial immunity (12). CXCL13 and CXCL12 protein levels in whole lung homogenates were measured using ELISAs at various time points after Sp19F infections ( Supplementary Figure 2A ). From days 8-10, CXCL13 protein increased significantly after 2 Sp19F infections and correlated with the accelerated increase in EV B cell numbers observed in Figures 1 , 2 . In contrast, CXCL12 levels did not change. Although this implicated CXCL13 as a potential driver of B cell recruitment and follicular B cell proliferation in the lung, administration of an anti-CXCL13 blocking antibody did not impact B_RM cell quantities or proportions when administered i.p. and intranasally (i.n.) at the same timepoints defined in Figure 3A ( Supplementary Figures 2B–D ) (49). Redundant mechanisms besides CXCL13 likely serve to recruit B cells to the lung, and this chemokine may be less important in models that do not support lasting mature iBALT. Of parallel note, the CXCL13/CXCR5 axis has been implicated in B cell homing to the central nervous system in several neuroinflammatory diseases, but it is dispensable for B cell recruitment, antiviral antibody responses, and overall disease course in alphavirus encephalitis, despite strong induction of CXCL13 (50). These studies highlight pathogen-dependent variations in B cell responses that can occur across tissues.

While the FTY720 data suggest that B cell recirculation is required for the maximal accumulation of B cells observed after 2 doses of Sp19, they do not preclude an additional contributory role for B cell proliferation in the lungs. Given the findings in Figure 3 , it is possible that S1P1R inhibition prevented other lymphocytes from entering the lung, such as CD4⁺ T cells, which could facilitate B cell proliferation in situ despite a lack of mature GC structures or known T_FH cells. This would be in line with recent findings using an IAV vaccination model and adoptive transfer strategy to determine that the formation of certain CD4⁺ T cell effectors and memory populations require cognate antigen/antigen-presenting cell exposure 6-8 days after initial antigen exposure (51). This included lung T_FH cells, which have been demonstrated in mature iBALT-inducing models, but thus far remain elusive in Sp infection models (32, 52–54). To capture only the cells in the lung that are dividing over the short period after infection, rather than those that had proliferated earlier in lymph nodes, we administered 5-ethynyl-2′-deoxyuridine (EdU) i.p. to mice 4 hours prior to euthanasia on day 9, day 14, or day 35 and used flow cytometric analysis to enumerate actively proliferating B cells. This was accomplished by fluorescently tagging EdU⁺ cells via copper-catalyzed click reactions ( Figure 4A ; gating strategy Supplementary Figure 3A ). Panel breadth was limited due to the interference of copper with certain fluorophores. Quantities of EdU⁺ IV B cells were relatively low and unchanged between Sp19F x1 and Sp19F x2 groups ( Figure 4B ). Among EV B cells, EdU was detected in groups that received 2 infections at each tested timepoint, but not in those that received only 1 exposure to Sp19F ( Figure 4C ). By day 35, the Sp-experienced lung is at a resting immune state and B cell proliferation is expectedly minimal. After two Sp19F infections, EdU is incorporated into proliferating non-naive B2 B cells predominantly at day 14 ( Figure 4D ). EV naive B cells did not exhibit significant increases in proliferation at any timepoint ( Figure 4E ). EdU⁺ B1 B cells were only detected at day 9, reflecting their properties as proliferating acute responders ( Figure 4F ). Finally, as with non-naive B2 B cells, B cell proliferation was significantly increased among lung pre-GC/GC B cells ( Figure 4G ). These findings demonstrate that the pneumococcus-experienced lung undergoes changes that can support non-classical GC driven proliferation of EV B cells after repeat exposure to T-independent (TI) and T-dependent (TD) antigens.

Although lacking mature tertiary lymphoid structures, known lung T_FH cells, or dependence on the follicular chemokine CXCL13, we observed proliferative B2 and GC B cells in situ in the lung after two Sp19F infections, suggesting that follicular signaling such as CD40-CD40L interactions between T and B cells may take place locally after the second exposure to Sp19F (38, 55). This led us to question the extent to which B cells in this model may depend on CD4⁺ T cell interactions for establishment in the lung. At 2dpsi, B cells marked by CD19 and T cells marked by CD3 were observed both as aggregates around bronchovascular areas as well as across alveoli ( Figure 5A ). Although found to be colocalized around bronchovascular bundles, the patterns of B and T cells observed at this timepoint were not consistent with those of organized follicular structures and were likely instead reflective of acute lymphocyte infiltration.

Since we observed B and T cells together in the lung at 2 dpsi, we tested whether CD4⁺ cells impacted initial B cell accumulation after the second Sp exposure. To do so, we administered the anti-CD4 antibody clone GK1.5 or IgG2b as an isotype control to C57BL/6J (B6) mice as previously described (12), using the treatment timepoints depicted in Figure 5B . This allowed the usual establishment of lymphocytes in the lung and periphery after the initial infection, but the second exposure to antigen occurred after CD4⁺ cell depletion. Lungs were extracted at 2 dpsi for flow cytometric analysis of early lung B cell populations. Depletion of CD4⁺ cells ( Figure 5C ) just before the second Sp19F infection did not impact IV blood B cells but diminished EV lung B cells ( Figures 5D, E ). Non-naive B2 B cells and especially CD69⁺ acutely activated B2 B cells were significantly reduced among the GK1.5-treated mice, relative to isotype control ( Figures 5F, G ). Not only were there fewer CD69⁺ B cells, but the CD69 median fluorescence intensity (MFI) levels were also lower on those expressing this activation marker in the GK1.5-treated group, serving as evidence for decreased activation during acute infection ( Figure 5H ). On the other hand, CD4⁺ cell depletion did not significantly affect B1 B cells, CD69⁺ B1 B cells, or expression of CD69 on B1 B cells 2 days post-infection ( Figures 5I–K ). This may be because B2 B cells are equipped with adaptive B cell receptors (BCRs) that respond to TD protein antigens and have the capacity to undergo follicular reactions, whereas B1 BCRs are germline-encoded with reactivity to TI antigens including pneumococcal capsular polysaccharide and cell wall phosphorylcholine (29, 56). These findings demonstrate that the initial accumulation of antigen-activated B2 B cells (whether via peripheral recruitment and/or in situ proliferation) following repeat antigen exposure exhibits some reliance on CD4⁺ cells, while B1 B cell accumulation and activation do not.

Since we observed differences in lung B cell subsets 2dpsi following CD4⁺ cell depletion but did not see complete erasure of initially formed lung B cell subsets 2 dpsi, we wanted to examine how the removal of CD4⁺ cell interactions during the second exposure to Sp19F impacted the maintenance of different lung B cell subsets and the establishment of B_RM cells in the Sp-experienced lung. We have previously observed loose aggregates of B and T cells at this timepoint after Sp19F infections and determined that these were not mature follicular structures based on absent peripheral node addressin staining (14). The presence of B_RM cells in a model lacking mature iBALT structures and known T_FH cells raised the question of whether B_RM cell formation necessitated interactions with CD4⁺ cells during repeat antigen exposure. To test this, we administered GK1.5 or isotype control throughout the second exposure to Sp19F as in Figure 5 , extending the timeline to 28 dpsi ( Figure 6A ).

While circulating IV B cells were again indifferent to CD4⁺ cell depletion ( Supplementary Figures 4A, B ), we observed a trending decrease in total EV B cells ( Figure 6B ). Both non-naive B2 B cells and CSW B cells were decreased in number ( Figures 6C, D ), demonstrating that CD4⁺ cells are important for the establishment of these cells in the recovered lung. In addition, we were surprised that B1 B cell quantities diminished among the GK1.5-treated mice as well ( Figure 6E ). The significant decrease in lung B1 B cells suggests that they may directly or indirectly rely on CD4⁺ cells or other B2 B cells for maintenance in the lung.

B_RM cells with phenotypes associated with memory (IgD^-PD-L2⁺CD73⁺; Figure 6F ) and residency (IgD^-CD69⁺; Figure 6G ) were completely abolished in the GK1.5-treated group. The frequencies of naïve and non-resident EV B cells were unchanged between treatment groups ( Supplementary Figure 4B ), indicative of a selective reliance on CD4⁺ cells for the formation and/or maintenance of B_RM cell populations in the lung. At resting timepoints, CD69 serves as a marker of lymphocyte residency rather than activation, therefore expression measured via MFI was not assessed as in the acute timepoints (45, 57). Taken together, these results demonstrate that lung BRM cells require CD4⁺ cells after recovery from pneumococcal infections.

The shift from IgM to other BCR isotypes resulting from class switch recombination is predominantly facilitated by specific cytokine and CD40 signals received by T helper cells, but can also occur without T cell help (58, 59). To assess the function of IgM⁺ and CSW antibody-secreting cells in CD4⁺-depleted experienced mice, we used bacterial whole-cell ELISAs to measure relative levels of Sp-reactive airspace antibodies in the bronchoalveolar lavage fluid (BALF) of the mice treated with GK1.5 vs. IgG2b. As these mice were experienced but in a resting state, we interpreted antibodies measured in the BALF as being sourced from baseline antibody-secreting cell activity in the lung, rather than from acutely formed antibody-secreting cells responding to antigen re-exposure. Therefore, we could use this as a proxy to determine whether these interventions during the second exposure to Sp19F impacted the formation or retention of lung plasma cells with serotype-independent reactivity against pneumococcal proteins, which we have previously demonstrated as a likely protective mechanism conferred by lung B_RM cells in this model (14). In agreement with the apparent dependence on CD4⁺ cells for the formation or maintenance of CSW B cells, we found that CD4⁺ cell depletion with GK1.5 during the second infection resulted in significantly decreased levels of airspace IgG and IgA reactive to acapsular Sp3 and a trending decrease in IgM ( Supplementary Figure 4C ). These data suggest that the formation or maintenance of lung humoral immunity depends on the presence of CD4⁺ cells while antigen is being presented locally in the lung.

As CD40L blockade has commonly been used to demonstrate T-dependent processes in B cells (60), and because we observed decreased B2 B cell and CSW B cells in CD4⁺ cell-deplete mice, we used the anti-CD40L blocking antibody (MR1) or isotype control (IgG) to test if the T-dependent contributions to B cell population establishment could be associated with CD40 signaling. As in the CD4⁺ cell depletion experiments, we tested whether MR1 administered through the second infection period impacted the initial and/or established lung B cell populations ( Figure 7A ). Interestingly, CD40L blockade did not have any impact on cell numbers ( Figures 7B, C ; Supplementary Figure 5A ) or activation state ( Figure 7D ) of early B cell subsets found at 2 dpsi. Synthesis of findings depicted here with those of Figure 5 suggests that CD4⁺ cells mediate acute B cell accumulation and activation in the lung independently of CD40-CD40L interactions.

At 28 dpsi, consequences of administering CD40L blockade during local antigen presentation were evident. Although IV circulating B cells were intact, EV lung B cell numbers were diminished when CD40L had been blocked during infection ( Figure 7E ; Supplementary Figure 5B ). Non-naive B2 B cells ( Figure 7F ) and CSW B cells ( Figure 7G ) were virtually eliminated due to the prior MR1 treatment. We again observed significant decreases in B1 B cells ( Figure 7H ), which do not typically require CD40-CD40L interactions (61). As in GK1.5-treated mice, the proportion of B_RM cells with phenotypes associated with memory (IgD^-PD-L2⁺CD73⁺; Figure 7I ) and residency (IgD^-CD69⁺; Figure 7J ) were significantly lower in the MR1-treated group than in the IgG-group. Thus, while CD40L-CD40 is not essential for the early CD4⁺ cell-dependent accumulation and activation of B cells during infection, these data implicate CD40-CD40L interactions as the mechanism by which CD4⁺ cells contribute to pan-B cell maintenance and B_RM generation in the lung.

Class switch reactions can occur via T helper and CD40L-dependent or independent mechanisms, with the latter providing low-affinity antibodies with specialized effector functions early in the immune response. IgA-secreting plasma cells and T-independent class switch reactions from IgM to IgA among intestinal B cells, but not colonic B cells, were previously found to be intact in CD40 knockout mice that are unable to form GCs (59). To assess the potential effect of CD40L-blockade on the formation and/or maintenance of lung antibody-secreting cells, BALF was collected at 28 dpsi for analysis of Sp-reactive airspace antibodies in the MR1-treated vs. IgG-treated mice ( Figure 7A ). MR1-mediated blockade of CD40L resulted in decreased Sp-reactive IgG, but IgM and IgA were left intact ( Supplementary Figure 5C ). Given that B cells were also globally affected by the loss of CD4⁺ cells and CD40L blockade, it is likely that the reduced supply of CSW B cells to become IgG and IgA-secreting antibody-secreting cells played a role in the observed differences in antibody-secreting cell activity, but it is also possible that CD4⁺ cells could contribute to a milieu that can support antibody-secreting cells.

We next questioned whether CD4⁺ T cells were necessary for optimal B_RM cell recall function. Reactivation of MBCs during secondary exposure to antigen was thought to require primed T cell help (24). Recent studies utilizing respiratory IAV infection described the induction of T_FH-like CD4⁺ T cells expressing BCL6 and/or FR4 in the lung, and showed that deletion of Bcl6 at later resting timepoints, impaired local antibody production upon rechallenge (53, 54).

To test whether CD4⁺ T cell help was required for lung MBC heterotypic recall responses to Sp3 challenge, we treated Sp-experienced B6 mice with the CD4⁺-cell depleting antibody GK1.5 or IgG2b isotype control before and during respiratory Sp3 infection to deplete circulating and parenchymal CD4⁺ cells ( Figure 8A ) (12). BALF was collected 48 hpi for ELISA quantitation of secreted crossreactive antibodies recognizing acapsular Sp as we have previously observed differences from baseline at this timepoint in Sp3-challenged experienced mice for Sp-reactive IgG and IgA ( Figure 8B ). We did not observe differences in airspace Sp-reactive IgG and IgA titers between the GK1.5- and isotype-treated groups at 48 hpi. Sp-reactive IgM was found to have a trending decrease in CD4⁺ T cell-depleted mice compared to isotype control-treated mice ( Figure 8C ), although airspace Sp-reactive IgM levels are not significantly elevated above baseline at this timepoint. As IgM secretion is associated with acute B cell activation, the trending differences seen in airway Sp-reactive IgM may reflect indirect changes in the inflammatory milieu secondary to CD4⁺ cell absence that prevented the recruitment or activation of acutely responding B cells (56, 62).These findings suggest that, like systemic MBCs, T cell dependence for lung B_RM cell reactivation is likely context-dependent. Whether the lack of access to follicular support is a factor in T-independent lung B memory recall will require more studies of B_RM cells in different infectious and autoimmune contexts.

In lungs recovered from pneumococcus, B cells are most prominent in loose leukocyte aggregates within the bronchovascular bundles (14). However, after IAV infection, lung B_RM cells are also numerous and active throughout the alveolar septae (16). Given that CD4⁺ cells were not required for memory recall, we hypothesized that B_RM cells might also be found outside of lymphocyte aggregates in lungs recovered from pneumococcus. We first sought to quantify B cells in these different anatomic compartments of lungs recovered from pneumococcal infections. At day 0, alveoli were patent, and the lung was without infiltrates or obvious immune cell aggregates. B cells were diffusely spread throughout the alveolar septae, but we were unable to determine whether they were intravascular or extravascular ( Figure 9A ). At day 35, we observed clusters of CD19⁺ cells within loose lymphoid aggregates in the bronchovascular compartment, with preserved alveolar architecture akin to that observed in uninfected lungs ( Figure 9B ).

CD19⁺ cells were quantified in peri-bronchovascular and alveolar compartments in a semi-automated manner using the imaging analysis platform QuPath (63) ( Figure 9C ). This allowed us to normalize differences in anatomic phenomena and exclude luminal airspaces across sections from different mice that may otherwise bias B cell quantification. Across lung sections, a mean of 77% of tissue was classified as alveolar while 33% was bronchovascular. Our analysis revealed that the number of CD19⁺ cells per square millimeter (mm²) of alveolar tissue in the experienced mice increased significantly compared to numbers in naïve uninfected lungs at day 0 ( Figure 9D ). This may represent increases in EV B cells that populate alveolar areas after Sp experience in addition to a consistent number of IV B cells circulating through the lung. Changes in B cell density were more pronounced between naïve and experienced mice in the bronchovascular compartment, which saw a >4-fold increase in CD19⁺ cells per mm² between naive and experienced lungs. Combining the relative densities of B cells in each tissue-type with the fractions of lung comprised of each tissue type, we approximate that the increased B cells in the recovered lung are similarly numerous across the tissue types (distributed as 55% in the alveolar septae and 45% in bronchovascular regions).

We previously used the transgenic CD19-Cre PD-L2-ZsGreen-TdTomato-diphtheria toxin receptor (PZTD) mouse line, which operates as both an inducible-depletion system and as a fluorescent reporter to show that 28 days after 2 pneumococcal infections, PD-L2⁺ B cells are only present in the EV spaces of the lung, where they function as protective B_RM cells (14, 64). This system therefore allows us to differentiate the anatomic location of EV B cells in the lungs, if we localize these PD-L2⁺ B_RM cells in the Sp-experienced lung to distinct anatomic sites. After IAV infection, the B cells in the alveolar region are motile when examined using precision cut lung slices (16). Therefore, in the present study, we used ex vivo confocal microscopy in the crystal ribcage (65) to visualize TdTomato⁺ cells in the peripheral parenchyma as PD-L2⁺ B_RM cells. We generated heterotypic immunity in these mice to allow for the tracking of PD-L2⁺CD19⁺ cells with TdTomato ( Figure 9E ). Live imaging of CD19-Cre PZTD lungs with Sp-experience revealed that the PD-L2⁺ memory population was indeed dispersed in the alveoli of experienced mice ( Figure 9F ). These cells were found to be largely stationary in the resting experienced lungs of PZTD mice, but they exhibited probing behaviors recognizable as cellular extensions and retractions throughout the 30-minute observation period ( Video 1-2 ). Sp3 rechallenge did not appear to induce changes in parenchymal B_RM cell motility after 24 hours ( Figure 9G ; Video 3-4 ). Altogether, the combination of morphometry and live lung imaging reveals that, while B_RM cells aggregate around the bronchovascular bundles of Sp-experienced lungs, the alveolar septa also contain a great number of these cells, which diffusely distribute independently of lymphocyte aggregations throughout the parenchyma of Sp-experienced lungs.

All animal protocols used in this study were approved by the Boston University Institutional Animal Care and Use Committee (IACUC). C57BL/6J (B6) mice were purchased from Jackson Laboratory (Bar Harbor, Maine). CD19-Cre+/- PZTD mice were generated by crossing CD19-Cre mice with homozygous PD-L2-ZsGreen-tdTomato-diphtheria toxin knock-in and inducible knockout (PZTD) background as previously described (64); mice were maintained in a specific pathogen-free facility with ad libitum food and water. Experiments were initiated when mice were 6-12 weeks of age and included both male and female mice. For i.t. and i.n. instillations, mice were first anesthetized with a mixture of ketamine (50mg/kg) and xylazine (5mg/kg) via i.p. injection prior to instillation. Mice were euthanized by isoflurane overdose. Bilateral thoracotomy was performed to ensure euthanasia before organ collections.

Bacteria were grown on blood agar plates (Trypticase Soy Agar (TSAII) with 5% Sheep Blood, BD Biosciences) for 14-16 hours at 37°C with 5% CO2. Mice were infected with Streptococcus pneumoniae serotype 19F (Sp19F EF3030) suspended in sterile saline (1 x 10⁶ bacterial colony forming units (CFUs) in 50 μL) given i.t. into the left lobe. i.t. infections were accomplished by cannulation of the trachea with a 24-gauge angiocatheter (BD 381112) For Streptococcus pneumoniae serotype 3 (Sp3 6303, ATCC) challenge infections, 0.5 x 10⁶ CFU suspended in sterile saline were administered to each mouse i.t.

Tissues were collected after euthanasia by isoflurane overdose. For BALF collection, the trachea was cannulated with an 18-gauge angiocatheter (BD 381444), and the lungs were lavaged with PBS and centrifuged. Blood was collected from the inferior vena cava into heparin-coated syringes and centrifuged for plasma collection. For lung tissue protein, harvested lungs were homogenized in a bullet blender (Next Advance) in a protease inhibitor solution (Roche 5892791001).

To distinguish IV circulating from EV lung tissue leukocytes, mice anesthetized with ketamine (50mg/kg) and xylazine (5 mg/kg) were injected i.v. with 2 μg of the indicated fluorescent anti-CD45.2 antibody 3 minutes before euthanasia. Single-cell suspensions were prepared by mincing isolated left lung lobes with sterile razors followed by enzymatic digest with type 2 collagenase (Worthington Biochemical LS004176) and DNase I (Sigma-Aldrich DN25) as previously described (12).

All single-cell suspensions were stained with specified fluorescent antibodies ( Supplementary Table 1 ) and 7-AAD (BD Biosciences) was used to determine cell viability other than in EdU studies. For EdU studies, the Click-iT AlexaFluor 488 EdU flow cytometry assay kit (Invitrogen C10420) was used per kit instructions, and Zombie UV (BioLegend) dye was used to determine viability. Stained cells were assessed on LSR-II (BD Biosciences) or Cytek Aurora (Cytek Biosciences). All cytometry data were analyzed using FlowJo v10 software (BD Biosciences) and expert gating was used to identify B cell subsets. OMIQ cloud analysis platform (OMIQ, Dotmatics) was used for the computational processing of the data. Briefly, live single EV B cell data exported from FlowJo were concatenated, asinh transformed (cofactor 6000) and selected fluorescence dataset features (B220, IgD, IgM, GL7, CD44, CD62L, CD43, CD11a, CD69, CD73, CXCR5, CD38, PDL2) were clustered via the Phenograph algorithm (kNN = 40, distance metric = Euclidean). The same features were projected into 2D mapping space using opt-SNE dimensionality reduction with principal component analysis-guided map initialization (perplexity = 30) to aid cluster visualization. Phenograph algorithm has identified 22 clusters in the data; 21 clusters with >0.5% frequency were used for downstream analysis.

7 days after respiratory infection with Sp19F, mice were treated on days 7, 9, 11, 13, and 15 with various interventions to abrogate processes occurring during repeat exposure to Sp19F administered on day 10, unless otherwise described. The S1P1R inhibitor FTY720 (Sigma SML0700) was delivered i.p. at doses of 1 mg/kg. CD4⁺ cell depletion was achieved using the anti-CD4 monoclonal antibody (clone GK1.5) or IgG2b isotype control (BioXcell) administered i.p. at a dose of 500 μg/100 μL and i.n. at 100 μg/50 μL i.n. CXCL13 was neutralized using the anti-CXCL13 antibody VX5/5378 (graciously provided by Vaccinex) or IgG2a (BioXcell) administered at a dose of 500 μg/100 μL i.p. and 200 μg/50 μL i.n. The CD40L-blocking antibody (clone MR1) or IgG isotype control (BioXcell) was delivered in i.p. instillations (250 μg/100 μL) and i.n. instillations (100 μg/50 μL) per each mouse to prevent CD40L-CD40 interactions. Mice were placed under light sedation with ketamine/xylazine for all i.n. instillations.

Pneumococcal-specific antibodies were measured in BALF using whole bacterial cell ELISAs. Nunc MaxiSorp plates (Thermo Fisher Scientific) were coated with 1x10⁷ CFU of an acapsular pneumococcal strain derived from a serotype 3 parent. The following HRP-conjugated detection antibodies were used: anti-total IgG (R&D Systems), anti-IgA (Thermo Fisher Scientific), and anti-IgM (Thermo Fisher Scientific). The optical density of each well was determined at 450 nm after the reaction was stopped with 2N H₂SO₄.

CXCL12 and CXCL13 protein levels were quantified in homogenized lung tissue using a mouse CXCL12 ELISA kit (R&D Systems MCX120) and a mouse CXCL13 ELISA kit (R&D Systems MCX130) per kit instructions.

Slides bearing 5 μm formalin-fixed paraffin-embedded tissue section blocks were deparaffinized in xylene and rehydrated in a graded ethanol series. Slides were exposed to heat-mediated antigen retrieval in a low pH citrate buffer (Vector Laboratories H-3300-250) followed by blocking, primary, and secondary antibody incubation in 5% donkey serum (Abcam ab138579).

For DAB staining, sections were exposed to 3% H₂O₂ after antigen retrieval. Slides were treated with the Vectastain ABC-HRP kit (Vector Laboratories PK-4000) followed by development with DAB substrate (Vector Laboratories SK-4100). Sections were counterstained with hematoxylin prior to dehydration and coverslip mounting. Rabbit anti-mouse CD19 antibody (Abcam) was used as the primary antibody and donkey anti-rabbit biotinylated antibody (Jackson Immunoresearch) was used as a secondary antibody. Brightfield images were captured on a Zeiss Axio Scan.Z1 slide scanner with Zeiss ZEN 3.1 Blue software and analyzed using QuPath v0.4.3. Bronchovascular tissue was manually annotated and lumenal areas were excluded using an optimized automated pixel classifier. CD19⁺ cells were enumerated in each tissue compartment using an optimized threshold for automated positive cell detection. Blinded analysis was performed on randomly selected sections.

For IF, rabbit anti-mouse CD19 (Abcam) and rat anti-mouse CD3 (Abcam) were used as primary antibodies and stained with Cy3 donkey anti-rabbit (Jackson ImmunoResearch) and AF647 donkey anti-rat secondary (Jackson ImmunoResearch) antibodies. Sections were treated with TrueVIEW autofluorescence quenching kit (Vector Laboratories SP-8400-15) prior to coverslip mounting with ProLong Gold Antifade mounting media with DAPI (Thermo Fisher P36931). IF images were captured on a Zeiss Axio Observer.Z1 inverted fluorescence microscope with Zeiss ZEN 3.3 Blue software.

Live ex vivo imaging of CD19-Cre PZTD mouse lungs was performed as previously described (65). Briefly, extracted mouse lungs were ventilated through a tracheal cannula and perfused through cannulas inserted into the pulmonary artery and the left atrium. A bolus of fluorescent Cascade Blue dextran (Invitrogen) was administered to stain pulmonary vasculature. The lung-heart bloc was placed into crystal ribcage, a synthetic transparent ribcage for microscopy under controlled ventilation and circulation. Ex vivo Lungs were imaged using an Olympus FV 3000 inverted laser-scanning confocal microscope and Fluoview acquisition software.

Statistical analyses were performed using GraphPad Prism 9.0. Each graph represents at least two independent experiments. Each individual point on a graph corresponds to an individual animal/sample. Summarized data are shown as means with standard errors. Mann-Whitney tests were used to assess comparisons of two groups, while three or more groups were compared using 2way ANOVA with a Holm-Sidak multiple comparisons test. In multiple comparisons testing, all possible comparisons were made unless otherwise indicated. Two-sided P values were calculated in all cases. Values of p < 0.05 were considered to indicate statistically significant differences.