PU.1^low fetal progenitors generate activated and memory B and T cells through middle age

In contrast to mice with a germ-line deletion of Spi1 that die in utero or soon after birth (Scott et al, 1994; McKercher et al, 1996), UREΔ/Δ mice have no overt pathology despite the fact that lymphocyte development from the adult stem and progenitor cells is blocked because of an 80% reduction in expression of Spi1 (Rosenbauer et al, 2006; Montecino-Rodriguez et al, 2016, 2018). Further evidence of this developmental block was provided by analysis of early B- and T-cell progenitors in the bone marrow (BM) and thymus of UREΔ/Δ mice (Figs S1 and S2).

We performed FACS analysis of B-cell progenitors in the marrow based on the scheme developed by Hardy et al (1991) (Fig S1A). This analysis indicated that few CD45R(B220)⁺CD43⁺CD19⁻CD24⁻Ly51⁻ Fraction A pre-pro-B cells were present in 4-wk–old UREΔ/Δ bone marrow. CD45R(B220)⁺CD43⁺CD19⁺CD24⁺Ly51^−/+ Fraction B and C-C′ pro-B cells and CD45R(B220)⁺CD43⁻ pre-B cells were present in the marrow from 4-wk–old UREΔ/Δ mice, but their frequency was lower compared with WT animals. The fact that some B-cell progenitors were present in UREΔ/Δ mice 4 wk after birth is consistent with the known survival of some transient fetal derived stem and progenitor cells through young adulthood (Montecino-Rodriguez et al, 2006; Bowie et al, 2007). However, by middle-age B-cell development was non-existent in UREΔ/Δ mice because no pre-pro-B, pro-B, or pre-B cells were detected (Fig S1B and C).

A similar pattern was observed for T-cell development (Fig S2). Cell counts revealed that a major loss of thymocytes was already evident in 4–6-wk–old UREΔ/Δ mice, and almost no cells were present in 11-mo–old animals (Fig S2A). Based on the FACS gating strategy shown in Fig S2B, we observed that the frequency of lineage negative (Lin)⁻CD117⁺CD44⁺CD25⁻ early T-lineage progenitors (ETPs) in the thymus from 4-wk–old UREΔ/Δ was eightfold lower than in WT mice, and by middle age ETPs were not detected (Fig S2C). The sharp decline in the number of thymocytes in UREΔ/Δ mice by 4 wk of age and the severe loss of T-cell progenitors is distinct from the pattern of decreased thymopoiesis that typifies thymic involution. For example, we detected ETPs in 50-wk–old WT mice, consistent with a previous report from our laboratory (Min et al, 2004), but they as well as Lin⁻CD44⁺CD25⁺ double negative (DN) 2 and Lin⁻CD44⁻CD25⁺ DN3 populations were totally absent from UREΔ/Δ mice by 24 wk of age (Fig S2C).

Nevertheless, the normal health span of UREΔ/Δ mice suggested that immune cells generated from transient PU.1^low fetal progenitors must have provided some degree of immune protection. We initially performed a global analysis and quantified B and T cells in the spleen (SPL) of middle-aged (22–46 wk old) UREΔ/Δ mice to assess this possibility. The analysis of B-cell subpopulations in that organ using gating strategies shown in Fig S3A and B revealed that CD93⁻ CD23⁺ CD21^low surface (s)IgM⁺ naïve follicular (FO) B cells were present, but their frequency and total number were significantly lower than in middle-aged WT mice (Fig 1A and B). Instead, most of the cells in the SPL of UREΔ/Δ mice included plasma cells/plasmablasts (PC/PBs), class-switched cytoplasmic (c) IgA⁺ cells, and CD80⁺ activated/memory B cells; the latter two populations were present at WT levels (Figs 1A and B and S3C).

The level at which CD44 (Baaten et al, 2010; Schumann et al, 2015) and CD62L (Anderson et al, 2003) are expressed has been used to identify CD62L^hiCD44^hi central memory (T_CM), CD62L^lowCD44^hi effector memory (T_EM) (Benichou et al, 2017), CD62^lowCD44^low pre-effector memory (T_pEM) (Nakajima et al, 2021), and CD62L^hiCD44^−/low naïve (T_N) CD4⁺ and CD8⁺ T cells. We quantified these T-cell subsets using gating strategies shown in Fig S4A in the SPL of middle-aged UREΔ/Δ mice and showed that the total number of CD4⁺ and CD8⁺ T_N cells was significantly lower than in similarly aged WT mice. However, CD4⁺ and/or CD8⁺ T_CM, T_EM, and T_pEM cells were present, often at WT levels (Figs 1C and D and S4B).

PU.1^low fetal progenitors generate an early, transient wave of naive B and T cells

The above results showing that only low numbers of naïve FO B and CD4⁺ and CD8⁺ T_N cells were present in the SPL of middle-aged UREΔ/Δ mice are consistent with the fact that PU.1^low progenitors are not maintained long-term after birth and that lymphoid development from adult stem and progenitor cells is blocked in that strain (Figs S1 and S2). To determine if PU.1^low fetal progenitors had the potential to generate naïve T and B cells during early life, lymphoid tissues from young (4–20 wk old) UREΔ/Δ mice were examined.

The number of cells in the SPL of young UREΔ/Δ mice was significantly higher than in middle-aged mice, although the total cell number was lower than in WT animals regardless of age (Fig 2A). As in middle-aged mice, the number of CD93⁻ CD23⁺ CD21^low sIgM⁺ FO B cells was significantly lower than in WT mice (Fig 2B and C). However, the number of FO B cells was significantly higher (P = 0.0002) in young compared with middle-aged UREΔ/Δ mice (compare Figs 1B and 2C). PC/PBs, class-switched cIgA⁺ cells, and CD80⁺ activated/memory B cells were also present in the SPL of young UREΔ/Δ mice (Fig 2B and C).

We also detected naïve CD4⁺ and CD8⁺ T cells in the SPL of young UREΔ/Δ mice (Fig 2D and E). The number of CD4⁺ T_N cells (P < 0.001) and CD8⁺ T_N cells (P < 0.0001) was significantly higher in young compared with middle-aged animals (compare Figs 1D and 2E).

Taken together, these results indicated that PU.1^low fetal progenitors produced naïve B and T cells that colonized young UREΔ/Δ SPL. However, their production was not sustained, as PU.1^low fetal progenitors were present only transiently, and by middle age numbers of naïve B and T cells had waned. We predicted that this would have a major impact on the maintenance of mesenteric lymph nodes (MLNs) and Peyer’s patches (PPs). Their initial organogenesis requires lymphoid tissue inducer and initiator cells, but once established, naïve B and T cells assume an important role in maintaining the differentiation and survival of mesenchymal and stromal cells that form their infrastructure (Randall et al, 2008).

MLNs were present in young animals, although, as in the SPL, the number of cells present was significantly lower than in WT mice (Fig 3A) and included FO and cIgA⁺ B cells and PC/PBs; the latter cells were present at a significantly lower level than in WT controls (Fig 3B and C). Low numbers of CD4⁺ and CD8⁺ T_N, T_CM, T_EM, and T_pEM cells were also detected in MLNs from young UREΔ/Δ mice but at significantly lower levels than in WT controls (Fig 3D and E). However, MLNs were difficult to identify in middle-aged UREΔ/Δ mice and were largely acellular (Fig 3A), and the limited number of cells that could be obtained precluded analysis.

PPs were more severely impacted than MLNs. Only a few PPs were identified in young UREΔ/Δ mice, and none were found in middle-aged animals, even by microscopic examination of the small intestine. Cell counts of PPs from young UREΔ/Δ mice indicated that they contained few cells compared with young WT tissues (Fig 3F). The limited analyses that could be performed indicated that the PPs from young UREΔ/Δ mice included sIgM⁺ and cIgA⁺ B cells and PB/PCs (Fig 3G and H) as well as CD4⁺ and CD8⁺ T, T_EM, and T_pEM cells (Fig 3I and J).

PU.1^low fetal progenitors generate innate B cells that are maintained through middle age

We next assessed whether transient PU.1^low fetal progenitors generated innate-like effectors. In view of our previous results indicating that they had marginal zone (MZ), B-1a, and B-1b B-cell potential (Montecino-Rodriguez et al, 2016), we focused on those populations.

B-1 B cells, and B-1a B cells in particular, can be generated from fetal progenitors (Kantor & Herzenberg, 1993; Montecino-Rodriguez & Dorshkind, 2012; Kobayashi et al, 2023). However, whether fetal derived B-1 B cells make a contribution to the adult B-1 B-cell compartment has been challenged by the identification of cells with B-1 potential in adult bone marrow (Berland & Wortis, 2002; Düber et al, 2009; Esplin et al, 2009), and a recent study concluding that postnatal rather than fetal lymphopoiesis is the predominant source of B-1a B cells in adult mice (Vergani et al, 2022). B-1a and B-1b B cells primarily localize to serous cavities such as the peritoneal cavity (PerC), and we used expression of CD5, CD23, and CD21 to identify them in that tissue (Fig 4A).

Both CD5⁺CD23⁻CD21^−/low sIgM⁺ B-1a and CD5⁻CD23⁻CD21^−/low sIgM⁺ B-1b B cells were present in the PerC of young and middle-aged UREΔ/Δ mice, and the total number of B-1 B cells as well as B-1a and B-1b subsets was similar and/or elevated, often significantly so, when compared with age matched WT animals (Fig 4B–D). B-1 B cells were also present in SPL of young UREΔ/Δ mice (Fig 4E). However, in contrast to what was observed in the PerC, by middle age their number was significantly reduced compared with middle-aged WT mice (Fig 4F).

We previously showed that CD93⁻ CD23⁻ CD21^hi sIgM⁺ marginal zone (MZ) B cells (Fig 4E) were present in the SPL of young adult UREΔ/Δ mice (Montecino-Rodriguez et al, 2016) but did not determine if they were maintained long-term. We found that the total number of MZ B cells was similar in the SPL of UREΔ/Δ and WT mice regardless of age (Fig 4G). In fact, the frequency of MZ B cells was higher in UREΔ/Δ compared with WT mice regardless of age, which we attribute to the low number of naïve B-2 B cells present in UREΔ/Δ SPL (Fig 1B).

Transient PU.1^low fetal progenitors generated B and T cells that colonize the intestines

The intestines are colonized in utero by diverse populations of B and T cells (Spencer et al, 1986; Golby et al, 2002; Schreurs et al, 2019; Stras et al, 2019), including memory CD4⁺ T cells (Li et al, 2019), but their origin has not been defined. We thus considered the possibility that transient PU.1^low fetal progenitors were a source.

The analysis of young UREΔ/Δ mice indicated that cells harvested from the lamina propria of the small intestine (SI) included sIgM⁺ and cIgA⁺ B cells and PC/PBs that were present at WT levels (Fig 5A and B). WT levels of CD4⁺ and CD8⁺ T cells were also present and included CD4⁺ and CD8⁺ T_EM and T_pEM cells (Fig 5C and D). Similarly, WT levels of sIgM⁺ and cIgA⁺ B cells and PC/PBs (Fig 5E and F) and CD4⁺ and CD8⁺ T cells, including T_EM and T_pEM cells (Fig 5G and H), were present in the UREΔ/Δ large intestine (LI). A population of fetal-derived CD8⁺ T cells with innate-like functions has been described and is distinguished by their CD44^high CD122^high phenotype (Smith et al, 2018). These cells may be present in the SI and LI of young UREΔ/Δ mice because some of the CD8⁺ cells in these tissues expressed high levels of CD44 (data not shown).

Analysis of the intestines from middle-aged UREΔ/Δ mice indicated that these B- and T-cell populations were maintained, but often at significantly lower levels than in WT mice. For example, sIgM⁺ B cells and PC/PBs cells were detected in the UREΔ/Δ SI, but their total number was significantly lower than in WT mice. However, the number of cIgA⁺ cells was equivalent between UREΔ/Δ and WT mice (Fig 6A and B). Nevertheless, the number of cIgA⁺ cells declined between young and middle age in UREΔ/Δ mice (compare Figs 5B and 6B). The total number of CD4⁺ and CD8⁺ T cells was also significantly lower in the SI from middle-aged UREΔ/Δ compared with WT mice, although T_EM and T_pEM cells were still present (Fig 6C and D).

The number of sIgM⁺ B cells was also significantly lower in the LI of middle-aged UREΔ/Δ compared with WT mice, but WT levels of PC/PBs and cIgA⁺ cells were present (Fig 6E and F). CD4⁺ and CD8⁺ T cells as well as T_EM and T_pEM cells were also found in the LI of middle-aged UREΔ/Δ mice, albeit at lower levels compared with WT mice, and CD4⁺ and CD8⁺ T_EM and T_pEM cells were present (Fig 6G and H).

Although PC/PBs were present in the intestines, they were not uniformly distributed in all tissues of UREΔ/Δ mice. In this regard, the bone marrow is a reservoir of plasma cells (Tellier et al, 2024), but no PCs were present in that tissue from UREΔ/Δ mice (Fig 6I). That observation suggests that the presence of PC/PBs in the SI and LI was not an artifact because of a lack of competition from adult plasma cells. Instead, the data indicate that the PB/PCs derived from UREΔ/Δ stem/progenitor cells have distinct tissue-homing properties.

The lymphoid progeny of transient PU.1^low fetal progenitors may contribute to gut homeostasis

Taken together, the above results demonstrated transient PU.1^low fetal progenitors were the source of various activated and memory B and T cells that colonized the intestines and were maintained through at least middle age. We next determined if these cells, in the absence of input of B and T cells generated from adult stem and progenitor cells, had the potential to maintain intestinal homeostasis. 16S rRNA sequencing of fecal samples from young and middle-aged WT and UREΔ/Δ mice housed in the same colony and fed identical diets was performed.

Subtle differences in the gut microbiome were observed between middle-aged UREΔ/Δ and WT mice, as shown in the relative abundance (Fig 7A) and the Shannon index (Fig 7B) plots. The primary difference between UREΔ/Δ and WT mice was the level at which some bacterial genera were present (Fig 7A). The relatively normal gut microbiome in UREΔ/Δ mice is likely because of the fact that class-switched IgA⁺ B cells were present at almost WT levels in the Si and Li of middle-aged UREΔ/Δ mice (Fig 6B and F). The observation that young and middle-aged UREΔ/Δ and WT mice had similar concentrations of secreted IgA in their fecal samples (Fig 7C) is consistent with this conclusion.

Study design

This study was designed to determine the degree to which B and T cells generated in fetal waves of lymphocyte development colonized secondary lymphoid tissues, were retained after birth, and contributed to immunity in the adult. We used the PU.1 deficient Sfpi1tm1.3Dgt/J (UREΔ/Δ) mice because previous reports showed that fetal lymphopoiesis is intact, but adult lymphocyte development is blocked in this strain (Rosenbauer et al, 2006; Montecino-Rodriguez et al, 2016, 2018). Animals were not subjected to experimental interventions except for genotyping by tail clipping during the 1st wk of life. Both male and female mice were included in this study. Our analyses revealed no sex-related effects on the survival of fetal derived B and T cells. Mice included in this study were individually processed and genotyped at the time of sacrifice to confirm their PU1 hypomorphic or WT genotype.

The conclusions in this report are based on analysis of groups of at least eight animals per age tested. This number was determined using the G*Power statistical program (Faul et al, 2007, 2009), which indicated that detection of significant differences of the means within 1.5–2 SD with 95% confidence and an alpha value of 0.05 (probability of type 1 error) for very small hematopoietic populations as determined by flow cytometry would require groups of 8 mice. The results were analyzed for statistical significance using the t test included in Prism GraphPad software (Version 10.1.1). Individual data points corresponding to each animal are plotted in our figures. PU.1 hypomorphic and WT control mice ranging in age from 4 to 46 wk of age were used. Subsequent statistical analysis of all the data revealed that PU.1 hypomorphic animals between 4–20 wk of age had a similar immune cell profile that differed from 22- to 46-wk–old animals. Based on this, the data are presented based on whether the mice were young (4–22 wk old) or middle aged (22–46 wk old). The investigators who performed the experiments were not blinded.

Mice

Breeding pairs of Sfpi1tm1.3Dgt/J (UREΔ/Δ) mice were originally obtained from the Jackson Laboratory and were backcrossed with C57BL/6 (B6) mice for at least seven generations in the UCLA Division of Laboratory Animal Medicine. Although the original publication that described the generation of UREΔ/Δ mice indicated that they developed various leukemias (Rosenbauer et al, 2006), this was only observed in ∼5% of mice in our colony; animals with signs of disease were excluded from analysis. Age matched UREΔ/Δ heterozygotes and C57BL/6 mice were used interchangeably as WT controls based on our previous studies which had shown that the immune system of these animals did not differ significantly (Montecino-Rodriguez et al, 2016). All protocols were approved by the UCLA Institutional Animal Care and Use Committee. Genotyping of UREΔ/Δ mice was performed by amplification of cDNA samples extracted from neonatal tail clips using the Qiagene DNeasy Blood and tissue kit and the Standard PCR Assay—Sfpi1<tm1.3Dgt> protocol # 22510 as recommended on the Jackson Laboratory website.

Harvest and preparation of tissues

Bone marrow, SPL, MLNs, and PerC-cell suspensions were harvested and processed as previously described (Min et al, 2006; Montecino-Rodriguez et al, 2006). Cells were isolated from SI and LI lamina propria using a protocol adapted from Sheridan and Lefrançois (2012). Whole mouse gut was harvested by cutting it at the pylorus and at the anus and cleaned from mesenteric fat. The tissue was then separated into SI and LI, which included the cecum, and placed in cold Ca⁺⁺Mg⁺⁺ free PBS supplemented at 1% with heat inactivated (hi) FCS. PPs were excised from SI using a microscope before proceeding with the subsequent steps. SI and LI were opened longitudinally, most digestive debris was gently scrapped off using the edge of a glass slide, and any remaining debris was removed by gently shaking of the tissues in PBS supplemented with cold Ca⁺⁺Mg⁺⁺ free PBS 1% hi FCS. These intestinal tissues were then cut in ∼1 cm fragments and incubated with a large volume of phenol red and bicarbonate free RPMI 1460 (Gibco) supplemented with 10% hi FCS, 15 mM Hepes (Gibco) 5 mM EDTA (Sigma-Aldrich) and 1 mM DTT (Thermo Fisher Scientific) for 30 min at 37°C in shaker set at 200 rpm to remove epithelial cells and mucus. The SI and LI fragments were then sedimented by gravity and the RPMI/EDTA/DTT supernatant was discarded, rinsed with large volumes of cold Ca⁺⁺Mg⁺⁺ free PBS, drained and minced in a solution at 200 U/ml of collagenase type 1A (Sigma-Aldrich) in phenol red and bicarbonate free RPMI 1460 10% hi FCS and 15 mM Hepes and digested for 30 min at 37°C in shaker set at 200 rpm. The SI and LI slurries were run through a syringe fitted with an 18 G needle, filtered using a 70 μm nylon mesh. The released cells were then washed with cold Ca⁺⁺Mg⁺⁺ free PBS and centrifuged at 400g for 5 min at RT. Lymphoid cells were then enriched by use of a discontinuous 40%:70% Percoll gradient.

PPs were incubated in RPMI/EDTA/DTT for 15 min at 37°C to remove epithelial cells and mucus. They were then sedimented by gravity, rinsed with cold Ca⁺⁺Mg⁺⁺ free PBS and digested for 15 min at 37°C in a solution at 200 U/ml of collagenase type 1A (Sigma-Aldrich) in phenol red and bicarbonate free RPMI 1460 10% hi FCS and 15 mM Hepes. PPs slurries were run through a syringe fitted with an 18 G needle, filtered using a 70 μm nylon mesh and washed with cold Ca⁺⁺Mg⁺⁺ free PBS and centrifuged at 400g for 5 min at RT.

Red blood cells in cell suspensions were lysed using 1X ACK lysis buffer on ice for 3 min. The cells were then counted with a hemocytometer and cell viability determined by eosin dye exclusion.

Flow cytometry

All staining procedures were performed in cold Ca⁺⁺Mg⁺⁺ free PBS. Samples were first incubated with a CD16/32 antibody to block non-specific binding of antibodies to cells via Fc receptors. All antibodies used are listed in Table S1. For surface staining, cells were incubated on ice for 30 min with the appropriate dilution of antibodies. Unbound antibodies were washed from cells with cold Ca⁺⁺Mg⁺⁺ free PBS. To be able to exclude dead cells during analyses of samples for intracellular staining, cells were first pre-incubated with the Fixable Viability Dye eFluor 780 (Thermo Fisher Scientific), as recommended by the manufacturer protocol. Subsequently, the cells were stained for surface antigens and fixed at 4°C for 20 min with 0.5% paraformaldehyde in Ca⁺⁺Mg⁺⁺ free PBS. The cells were then permeabilized for 15 min with cold 1x BD Perm/Wash Buffer in the presence of a CD16/32 blocking antibody and then stained for anti-murine IgA at 4°C overnight. Unbound antibodies were washed from cells with 1x BD Perm/Wash and cells were resuspended in an appropriate volume of cold Ca⁺⁺Mg⁺⁺ free PBS for flow. Stained samples were run on a LSRII (BD Biosciences) located in the Broad Stem Cell Research Center (University of California at Los Angeles). Frequencies of cell populations were determined using the FlowJo Software v10 samples. To compare the cellular composition of similar tissues samples across multiple ages, we analyzed them using the unsupervised clustering and nonlinear dimensionality reduction Uniform Manifold Approximation and Projection (UMAP) plugin developed by McInnes et al (2020 Preprint) available in the FlowJo Software v10 and the samples were processed as recommended (Ujas et al, 2023) before running the analyses.

We used standard flow best practices to minimize variations between samples during acquisition: all samples were run on the same LSRII instrument with identical parameter settings to control for fluorescence intensity variation between runs. We also followed the recommended best practice for UMAP analyses: identical number of acquired cells from WT and UREΔ/Δ samples were concatenated by age and tissue before running the UMAP algorithm (FlowJo10). In this case, flow samples that included too low events or out of average range of fluorescence intensities for the parameters analyzed were excluded.

16S microbiome sequencing

Three fecal pellets per mouse were harvested from UREΔ/Δ or WT mice of different age and frozen at—20°C until sequencing. DNA extraction and sequencing from fecal samples were performed by Charles River. DNA was isolated from submitted samples using optimal sample-dependent column or magnetic purification kit. Recovery yield and DNA quality was determined by fluorometric analysis. DNA concentration was adjusted to specifications and amplified using broadly reactive 16s rRNA primers spanning the V3 and V4 regions. Resulting amplified PCR products were analyzed for quantity and correct product size then purified and amplified with primers containing unique sample nucleotide barcodes. PCR products quality and quantity were further analyzed by SYBR green qPCR. All samples were pooled and adjusted to a normalized concentration. The DNA library pool was denatured with sodium hydroxide, normalized to optimal loading concentration, and combined with PhiX control. Extended read lengths up to 2 × 300 bp was used for cluster generation and sequencing. Following the sequencing run, the sequence data were de-multiplexed based on the nucleotide barcode and compared with the One Codex Targeted Loci Database for taxonomy identification and subsequent alpha and beta diversity analysis. All raw sequences were analyzed by Charles River Laboratories bioinformaticians using the One Codex Targeted Loci Database (TLDB)1 that contains 250,000 curated gene records. To analyze samples against the Targeted Loci Database, every read was aligned with high sensitivity (using SNAP) to identify the best alignments to the database. Each read was assigned to the most specific taxonomic grouping that the data support. Abundance-based filtering to minimize the number of false positive assignments that may be introduced by sequencing error was then performed. As amplification using primers to the V3 and V4 regions may not allow reliable quantification of differences at the species level (Johnson et al, 2019), bacterial identification was limited to taxa at the genus level. The raw results of these analyses are provided in the Supplemental Material.

Fecal IgA quantification by sandwich ELISA

Three fecal pellets per mouse were harvested from the anal canal of UREΔ/Δ and WT mice and resuspended in 500 μl of sterile Ca⁺⁺ Mg⁺⁺ free PBS in 1.5 ml Eppendorf tubes. The samples were well mixed and incubated for 4–5 h at RT. The samples were then vortexed and fecal debris was sedimented by centrifugation at 20,000g for 10 min at 4°C. The supernatants were then transferred to new Eppendorf tubes and stored at −20°C until they were tested for IgA concentrations using a standard double sandwich ELISA. Briefly, wells of microtiter plates (NUNC Immunoplate Maxi Sorp, Denmark) were coated with 50 μl/well of goat anti anti-mouse IgA overnight at 4°C. The plates were washed with Mg⁺⁺ PBS 0.5% Tween 20 and Mg⁺⁺ PBS and subsequently blocked with Mg⁺⁺ PBS supplemented with 2% BSA (Thermo Fisher Scientific) for 2 h at 37°C. Duplicate samples of 50 μl/well of supernatant were distributed in wells and incubated for 1 h and 30 min at 37°C. The plates were washed with Mg⁺⁺ PBS 0.5% Tween 20 and Mg⁺⁺ PBS and bound IgA was detected using 50 μl/well appropriate dilutions of goat anti-mouse IgA alkaline phosphatase-conjugated antibody in Mg⁺⁺ PBS 0.5% Tween 20 0.5% BSA. After final washes, 100 μl of p-Nitrophenyl phosphate (PNPP, Sigma-Aldrich) at 0.1 mg/ml in 1x Diethaloamine buffer (Thermo Fisher Scientific) were added per well as substrate. Optical density was read after a 30 min incubation at 37°C using a plate reader fitted with 405 nm filter. IgA concentrations in the supernatants were determined using a standard curve obtained using serial dilutions of known concentrations of murine IgA. The monoclonal antibodies used to detect murine IgA in this assay are listed in Table S1.

Statistical analysis

Data are expressed as a mean ± SEM, as indicated in the figure legends. Differences between groups were tested using a two-tailed, unpaired t test. As noted in the study design, the number of animals processed was determined based on the G*Power statistical program (Faul et al, 2007, 2009). The data are presented so that information about each individual mouse processed can be viewed.