LAG-3 Inhibitory Receptor Expression Identifies Immunosuppressive Natural Regulatory Plasma Cells

Summary

B lymphocytes can suppress immunity through interleukin (IL)-10 production in infectious, autoimmune, and malignant diseases. Here, we have identified a natural plasma cell subset that distinctively expresses the inhibitory receptor LAG-3 and mediates this function in vivo. These plasma cells also express the inhibitory receptors CD200, PD-L1, and PD-L2. They develop from various B cell subsets in a B cell receptor (BCR)-dependent manner independently of microbiota in naive mice. After challenge they upregulate IL-10 expression via a Toll-like receptor-driven mechanism within hours and without proliferating. This function is associated with a unique transcriptome and epigenome, including the lowest amount of DNA methylation at the Il10 locus compared to other B cell subsets. Their augmented accumulation in naive mutant mice with increased BCR signaling correlates with the inhibition of memory T cell formation and vaccine efficacy after challenge. These natural regulatory plasma cells may be of broad relevance for disease intervention.

Graphical Abstract

Highlights

• •LAG-3 expression identifies natural regulatory plasma cells
• •LAG-3⁺CD138^hi plasma cells express IL-10 within hours of stimulation
• •LAG-3⁺CD138^hi plasma cells have a unique epigenome poised to express IL-10
• •LAG-3⁺CD138^hi plasma cells develop via an antigen-specific mechanism

Plasma cells secrete antibodies and play a key role in host defense against infection. Lino et al. identify a novel subset of natural regulatory plasma cells characterized by the expression of LAG-3 that develops at steady state independently of microbiota, and respond to innate stimulation by producing immunosuppressive IL-10.

Introduction

The immune system protects the host from infectious diseases and helps to remove its damaged components. Its activity is controlled by stimulatory and inhibitory forces. Inhibitory pathways are important to prevent its spontaneous hyperactivation and to regulate established immune reactions.

B lymphocytes have emerged as important players in the negative regulation of immunity via the production of interleukin-10 (IL-10) and IL-35 (Fillatreau et al., 2002, Shen and Fillatreau, 2015). This B cell function has noticeable effects in autoimmune, allergic, infectious, and malignant diseases (Shen and Fillatreau, 2015). The phenomenology of B cell-mediated suppression is well established, yet the B cells mediating these activities in vivo remain incompletely defined. Their identification has been approached by isolating known B cell subsets and assessing their capacity to produce IL-10 in vitro and to suppress immunity in recipient mice upon adoptive transfer. B cells expressing high amounts of CD1d, including marginal zone B cells and transitional T2-like B cells, as well as B1a cells and Tim-1^hi B cells, have the capacity to suppress immunity in an IL-10-dependent manner in such assays (Blair et al., 2009, Yanaba et al., 2008, Yang et al., 2012). However, only a fraction of the cells within these subsets express IL-10 after activation in vitro, even when strong pharmacological agents such as ionomycin and phorbol myristate acetate (PMA) are used, and the phenotype of the B cells actually producing IL-10 in vivo in the recipient is not defined after transfer. Other studies have used IL-10 reporter mice to identify IL-10-producing cells ex vivo without re-stimulation. These have revealed CD138^hi plasmocytes (plasmablasts and plasma cells) as the major source of B cell-derived IL-10 in vivo in autoimmune, infectious, and malignant diseases (Matsumoto et al., 2014, Neves et al., 2010, Shalapour et al., 2015, Shen et al., 2014, Teichmann et al., 2012). A hypothesis reconciling these findings could be that B cell-mediated regulation is an inducible function acquired by B cells such as CD1d^hi B cells upon activation and differentiation into IL-10- or IL-35-producing plasmocytes.

Here, we addressed whether IL-10-producing plasmocytes defined a separate subset using a model of infection by the bacterium Salmonella Typhimurium. In this model, B cell-derived IL-10 is produced exclusively by plasmocytes that emerge before day 1 post-infection (p.i.) and leads to a rapid modulation of immunity to Salmonella (Neves et al., 2010). We reasoned that the rapidity of this response would facilitate the identification of the precursors of immunosuppressive IL-10-producing plasmocytes without having to recourse to adoptive transfer protocols susceptible to creating non-physiological cellular responses.

Results

LAG-3 Identifies IL-10-Expressing Plasma Cells in Infected Mice

Salmonella infection results in the rapid appearance of IL-10⁺CD138^hi cells. To assess whether these cells defined a particular subset, we compared their transcriptome to the one of IL-10⁻CD138^hi cells from Il10eGFP mice on day 1 p.i. This yielded 22 genes coding for transmembrane receptors overexpressed in IL-10⁺ cells (Figure 1A), including lymphocyte-activation gene 3 (Lag3; CD223) that was previously identified as a marker for IL-10-producing T regulatory type (Tr1) cells (Gagliani et al., 2013).

Figure 1.

LAG-3 Identifies IL-10-Producing Plasma Cells in Infected Mice

Mice were infected i.v. with Salmonella (SL7207, 10⁷ CFU), and plasmocytes characterized in spleen on day 1 p.i.

(A) mRNA amounts for receptors overexpressed in IL-10⁺ compared with IL-10⁻ plasmocytes from Il10eGFP mice. Microarrays in triplicate. Expression amounts normalized using GCRMA are shown (Log2 transformed). Lag3 is expressed 9.4-fold higher in Il10eGFP⁺CD138^hi than Il10eGFP⁻CD138^hi cells.

(B) Flow cytometry plots showing IL-10 and LAG-3 expression in CD138^hi cells (left); frequency of LAG-3⁺ in IL-10⁺CD138^hi cells (middle); expression of IL-10 in B cells (CD19⁺CD138⁻) and LAG-3⁺CD138^hi cells (right). Representative of six experiments.

(C) Il10 mRNA expression in isolated subsets from C57BL/6 mice. Pool of two experiments.

(D) Transmission electron microscopy images of plasmocytes from Il10eGFP mice. Scale bar is 2 μM.

(E) Frequencies of antibody-secreting cells (ASCs) in IL-10⁺CD138^hi and IL-10⁻CD138^hi cells from Il10eGFP mice by ELISPOT. Pool of three experiments.

(F) Flow cytometry plots of LAG-3 and Ki67 expression in CD138^hi cells from C57BL/6 mice. Representative of four experiments.

(G) Flow cytometry plot of LAG-3 and BLIMP-1 in BLIMP-1⁺CD138^hi cells (left), and amounts of BLIMP-1 (MFI) in B cells and plasmocytes in Prdm1eGFP mice (n = 4) (right). Representative of two experiments.

(H) Surface molecules expression on LAG-3⁺CD138^hi (red), LAG-3⁻CD138^hi (blue), and CD19⁺CD138⁻ B cells (gray) from C57BL/6 mice. Representative of two experiments.

Data show mean ± SEM (^nsp > 0.05, ^∗∗p < 0.01). See also Figure S1.

LAG-3 protein was found on most IL-10⁺CD138^hi cells on day 1 p.i., and IL-10 was expressed uniformly in LAG-3⁺CD138^hi cells (Figure 1B). LAG-3 was not detected on CD19⁺CD138⁻ cells (Figure S1A). This staining was specific because it was not observed on Lag3^−/− plasma cells (Figure S1B). Consistently, Il10 mRNA was predominantly expressed in LAG-3⁺CD138^hi cells compared to other B cell subsets in infected C57BL/6 mice (Figure 1C).

IL-10⁺LAG-3⁺CD138^hi cells displayed typical plasma cell features including a plasmacytoid morphology (Figure 1D), the spontaneous secretion of antibodies (Figure 1E), a non-proliferative state (Figure 1F), and an elevated expression of BLIMP-1 (Figure 1G). LAG-3⁺CD138^hi cells also differed from LAG-3⁻CD138^hi cells in their higher expression of CD1d and CD200, as well as their lower expression of B220, MHC-II, CD43, CD71, and Fas (Figure 1H).

We conclude that LAG-3⁺CD138^hi plasma cells define the main population of IL-10-expressing B cells in infected mice.

LAG-3⁺CD138^hi Plasma Cells Develop Independently of Microbe-Derived Signals in Naive Mice

The non-proliferating status of IL-10⁺LAG-3⁺CD138^hi cells was unexpected because B cell differentiation into plasma cell normally requires cell proliferation over several days. This led us to hypothesize that LAG-3⁺CD138^hi cells were already present in naive mice. Indeed, LAG-3⁺CD138^hi cells were detected in the spleen, bone marrow (BM), and mesenteric lymph nodes (mLN) of naive mice (Figures 2A and S2A). They had the key attributes of plasma cells such as high BLIMP-1 expression (Figures 2B and S2B), a plasmacytoïd morphology (Figure 2C), and the spontaneous secretion of antibodies (Figure 2D). In contrast to proliferating LAG-3⁻CD138^hi plasmablasts, LAG-3⁺CD138^hi cells were non-proliferative and produced mostly IgM (Figures 2D and 2E). These features of LAG-3⁺CD138^hi cells therefore did not change between day 0 and day 1 p.i., except for the induction of IL-10 expression after challenge (Figure 2F). LAG-3⁺CD138^hi cells also differed from LAG-3⁻CD138^hi cells in their higher expression of CD1d, CD69, CD81, CD200, and CD273 (PD-L2), as well as their lower expression of B220, MHC-II, CD43, CD71, FAS, and CXCR3 in naive mice (Figure 2G). PD-L1 was highly expressed by both plasmocyte subsets. LAG-3⁺CD138^hi cells expressed surface IgM, CD79α, and CD79β (Figure 2G). LAG-3 expression was not observed on B cells (Figure S2C).

Figure 2.

LAG-3⁺CD138^hi Plasma Cells Are Present in Naive Mice

Analyses performed in spleen (except when indicated) of naive mice.

(A) Flow cytometry plots of LAG-3 on CD138^+/hi cells (top) and frequencies in spleen (n = 23), BM (n = 19), mLN (n = 10), subcutaneous LN (sLN) (n = 10), and PeC (n = 10) (bottom) of C57BL/6 mice.

(B) Flow cytometry plots of LAG-3 and BLIMP-1 in BLIMP-1⁺CD138^hi cells (left), and amounts of BLIMP-1 (MFI) in B cells and plasmocytes in Prdm1eGFP mice (n = 4) (right). Representative of three experiments.

(C) Electron microscopy images of plasmocytes from C57BL/6 mice. Scale bar is 2 μM.

(D) ASCs in plasmocytes from C57BL/6 mice by ELISPOT. Pool of two experiments.

(E) Flow cytometry plot of LAG-3 and Ki67 in CD138^hi cells from C57BL/6 mice. Representative of four experiments.

(F) Flow cytometry plot of LAG-3 and IL-10 in CD138^hi cells. Representative of six experiments.

(G) Expression of indicated molecules on cells from C57BL/6 mice. Representative of 3–4 experiments.

(H) Numbers of plasmocytes in SPF (n = 8) and GF (n = 8) C3H/HeOuJ mice.

(I) Numbers and frequencies of plasmocytes in C57BL/6 mice of indicated ages (n = 6/age, pool of 2 experiments).

Data show mean ± SEM (^nsp > 0.05, ^∗p < 0.05, ^∗∗p < 0.01; ^∗∗∗p < 0.001). See also Figure S2.

Plasma cells are normally induced upon foreign antigen stimulation. LAG-3⁺CD138^hi cells, however, were present in similar numbers in germ-free and specific-pathogen-free mice (Figure 2H). These cells are therefore generated via an endogenous response. Their number increased markedly in aging mice (Figures 2I and S2D).

We conclude that, in naive mice, LAG-3⁺CD138^hi cells are a subset of natural plasma cells mainly producing IgM.

LAG-3⁺CD138^hi Cells Rapidly Upregulate IL-10 after Infection

IL-10 expression in CD138^hi cells increased already at 3 hr p.i. compared to naive mice (p = 0.0009; unpaired t test) (Figure 3A). IL-10 was expressed predominantly in LAG-3⁺CD138^hi cells during the first 24 hr p.i. (Figure 3A). IL-10 is thus programmed for a rapid expression in LAG-3⁺CD138^hi cells, whose numbers were similar at day 0 and day 1 p.i. (Figure 3B). As expected, IL-10 and LAG-3 were not detected in CD19⁺CD138⁻ B cells (Figure S3A).

Figure 3.

LAG-3⁺CD138^hi Cells Upregulate IL-10 Expression upon Infection

Analyses performed in spleen p.i. (SL7207, 10⁷ CFU) or with naive mice.

(A) Kinetics of IL-10 expression in CD138^hi cells (left), frequency of LAG-3⁺ in IL-10⁺CD138^hi cells (middle), numbers of LAG-3⁺IL-10⁺CD138^hi (LAG-3⁺) and LAG-3⁻IL-10⁺CD138^hi (LAG-3⁻) cells (right) p.i. in Il10eGFP mice. Pool of two experiments (n = 6/time point).

(B) Numbers of LAG-3⁺CD138^hi cells in C57BL/6 mice. Pool of four experiments (n = 11/time point).

(C) Isolated cells from naive C57BL/6 mice were stimulated in vitro for 18 hr, and IL-10 measured in supernatants. LAG-3⁺CD138^hi and LAG-3⁻CD138^hi cells indicated as LAG-3⁺ and LAG-3⁻, respectively. Pool of five experiments.

(D) Numbers of CD138^hi and LAG-3⁻CD138^hi cells (left), LAG-3⁺IL-10⁺CD138^hi cells (middle), frequency of LAG-3⁺ (red line) or IL-10⁺ (green line) cells in CD138^hi cells (right) p.i. in Il10eGFP mice. Pool of 4 experiments; at least 12 mice per group.

(E) Numbers of IL-10⁺CD138^hi and IL-10⁻CD138^hi cells at day 3 p.i. in B-Il10eGFP and B-Lag3^−/−Il10eGFP chimeric mice.

(F) Il10eGFP mice were vaccinated and challenged 90 days later (vaccinated), along with age-matched naive Il10eGFP mice (challenged). Frequencies of LAG-3⁺ in CD138^hi cells (left), and of IL-10⁺ in CD138^hi cells (right). Pool of two experiments; at least six mice per group.

(G) Il10eGFP mice treated as in (F) and analyzed at day 1 and 2 post-re-challenge. Flow cytometry plots show LAG-3 and CD138 on IL-10⁺CD138^hi cells (right, day 1) with quantifications (left). Pool of two experiments; at least six mice per group.

(H) Il10eGFP mice were treated as in (F). Frequency of ASC in indicated plasmocytes on day 1 post-re-challenge by ELISPOT. Pool of two experiments; at least six mice per group.

Data shown are mean ± SEM (^nsp > 0.05, ^∗p < 0.05, ^∗∗p < 0.01; ^∗∗∗p < 0.001). See also Figure S3.

The immediate upregulation of IL-10 in LAG-3⁺CD138^hi cells led us to compare the capacity of CD138^hi cells and other B cell subsets to produce IL-10 in vitro. CD138^hi plasmocytes secreted markedly more IL-10 than CD1d^hiCD19⁺ B cells (Figure 3C), and their IL-10 production derived mostly from LAG-3⁺CD138^hi cells (Figure 3C).

We then examined IL-10 expression at later time points p.i. during the extra-follicular response. The number of CD138^hi cells increased progressively p.i., and they represented 10% of all splenocytes on day 8 p.i. (Figures 3D and S3B). This increase reflected the accumulation of LAG-3⁻CD138^hi cells, while IL-10⁺LAG-3⁺CD138^hi cells did not expand greatly after day 1 p.i. (Figure 3D). IL-10 expression was associated with LAG-3⁺CD138^hi cells at all time points (Figure S3C). The frequencies of LAG-3⁺ and IL-10⁺ cells within CD138^hi cells sharply decreased from day 1 p.i. onward (Figure 3D). These results emphasize the unique response dynamics of IL-10⁺LAG-3⁺CD138^hi cells. Because LAG-3 has been implicated in the regulation of humoral immunity (Butler et al., 2011), we asked whether LAG-3 expressed on CD138^hi cells modulated the plasmocyte response p.i. To this end, we crossed Lag3^−/− mice on an Il10eGFP background and generated mixed bone marrow chimera in which only B cells lacked Lag3 but could express Il10eGFP. A LAG-3 deficiency in plasma cells did not alter the number of IL-10⁺CD138^hi cells p.i., yet led to an increased accumulation of IL-10⁻CD138^hi cells (Figure 3E). Thus, LAG-3⁺CD138^hi cells control the expansion of LAG-3⁻CD138^hi cells in a LAG-3-dependent manner.

An important property of the adaptive immune system is immunological memory. To assess whether LAG-3⁺CD138^hi cells acquire features of memory B cells, we compared their response in vaccinated and naive mice p.i. As expected, vaccinated mice better controlled Salmonella than naive mice (Figure S3D). IL-10 and LAG-3 expression in CD138^hi cells were comparable in the two groups p.i. (Figures 3F and 3G). Thus, LAG-3⁺CD138^hi cells did not display the typical sign of an amplified response upon rechallenge. IL-10⁺CD138^hi cells isolated from vaccinated mice on day 1 p.i. produced mostly IgM (Figure 3H), indicating that they had not undergone isotype switching away from IgM, unlike what is observed in classical memory B cell responses.

These data show that LAG-3⁺CD138^hi cells can rapidly upregulate IL-10 expression without dividing, and inhibit the expansion of LAG-3⁻CD138^hi cells p.i.

LAG-3⁺CD138^hi Cells Have a Unique Plasma Cell Epigenome

We next examined the hypothesis that LAG-3⁺CD138^hi cells were a distinct subset of plasma cells epigenetically poised to rapidly produce IL-10. To this end, we compared the genome-wide DNA-methylomes of LAG-3⁺CD138^hi cells, LAG-3⁻CD138^hi cells, and peripheral B cell subsets from naive mice. In addition, we included LAG-3⁺CD138^hi, LAG-3⁻CD138^hi, IL-10⁺CD138^hi, and IL-10⁻CD138^hi cells isolated at day 1 p.i.

Genome-wide DNA-methylation profiling was performed by reduced representation bisulfite sequencing (RRBS) covering on average 3–4 million CpG sites per sample. An unsupervised principal component analysis (PCA) placed LAG-3⁺CD138^hi and IL-10⁺CD138^hi cells at a distinctive position relative to LAG-3⁻CD138^hi and IL-10⁻CD138^hi cells, as well as to other B cell subsets, underlying their unique methylome profile. The position of LAG3⁺CD138^hi cells taken at day 0 and day 1 p.i. was similar and showed a striking co-localization with IL-10⁺CD138^hi cells (Figure 4A), underlining the intimate relationship between these subsets. The various groups of CD138^hi cells localized at the same position on PC1, which separated B cells from plasma cells. In keeping with the described hypomethylation of plasma cells’ genome (Kulis et al., 2015), the various CD138^hi subsets analyzed here displayed the lowest degree of genome-wide methylation (Figure S4A). Thus, LAG-3⁺CD138^hi and IL-10⁺CD138^hi cells have a plasma cell epigenetic profile but with unique features. A pairwise comparison of LAG-3⁺CD138^hi and LAG-3⁻CD138^hi cells from naive mice identified 469 differentially methylated regions (DMRs) showing a genome-wide distribution (Figure S4B). In a hierarchical analysis, these DMRs separated CD138^hi cells and B1 cells away from the other B cell subsets (Figure 4B), with LAG-3⁺CD138^hi cells showing the highest similarity to B1a cells (Figures 4B and S4C). A particular cluster of DMRs was found at the Il10 locus. IL-10⁺CD138^hi cells, LAG-3⁺CD138^hi cells, and B1a cells not only exhibited the lowest overall degree of DNA methylation at the Il10 locus (Figure 4C) but also shared distinct patterns of hypomethylated regions around the Il10 gene. Many of these hypomethylated sites overlapped with B cell-specific DNase hypersensitive sites, indicating an epigenetic “preprogrammed” open chromatin state of Il10 in IL-10⁺CD138^hi, LAG-3⁺CD138^hi, and B1a cells (Figures 4D and S4D). Of note, IL-10⁺CD138^hi cells, LAG-3⁺CD138^hi cells, and B1a cells also showed a preferential epigenetic relationship at the 469 DMR (Figure S4E).

Figure 4.

Molecular Characterization of LAG-3⁺CD138^hi Cells

(A) Unsupervised PCA of genome-wide DNA methylation data of cells from C57BL/6 mice.

(B) Methylation of the CpG found in the 469 DMR distinguishing LAG-3⁺CD138^hi cells and LAG-3⁻CD138^hi cells.

(C) DNA methylation of Il10 locus (mean ± SEM).

(D) Methylation for covered CpG in the Il10 locus. Coverage weighted average methylation of three replicates is represented. The positions of selected CpG are indicated by vertical black bars, with the Il10 gene depicted. ENCODE DNase I data are in red for splenic CD43⁻ B cells. PhastCons Vert30 conservation scores are in blue.

(E) mRNA expression for the 3,631 DEGs that distinguish LAG-3⁺CD138^hi and LAG-3⁻CD138^hi cells on day 0 and 1 p.i. (see Figure S4F).

(F) mRNA expression and local DNA methylation for genes both differentially expressed and methylated between LAG-3⁺CD138^hi and LAG-3⁻CD138^hi cells from naive mice.

(G) Transcription regulators expressed at higher (top) or lower (bottom) amounts on both day 0 and 1 p.i. in LAG-3⁺CD138^hi compared to LAG-3⁻CD138^hi cells.

(H) Expression of transcription regulators differentially expressed between LAG-3⁺CD138^hi cells and LAG-3⁻CD138^hi cells (see G) with predicted binding motif in either the Il10, Lag3, Cd200, or Cd273 loci.

Row z-scores for expression are based on log₁₀(rpkm+1), and for DNA methylation on DNA methylation frequency (%).

Samples were from naive mice (A, C, D, E, G, H) except where indicated. See also Figure S4.

In summary, LAG-3⁺CD138^hi cells are the epigenetically primed precursors of IL-10⁺LAG-3⁺CD138^hi cells and show a distinct plasma cell-specific epigenome with the Il10 locus primed for expression and features reminiscent of B1a cells.

Transcriptional Properties of LAG-3⁺CD138^hi Cells

We next determined the transcriptomes of LAG-3⁺CD138^hi and LAG-3⁻CD138^hi cells on day 0 and day 1 p.i by mRNaseq. These analyses identified 648 differentially expressed genes (DEGs) (pval ≤ 0.01; FDR ≤ 0.05) on day 0 and 2,280 on day 1 p.i. (Figure S4F). A hierarchical clustering with all DEGs confirmed the relative proximity of LAG-3⁺CD138^hi cells from naive mice and day 1 p.i. compared to LAG-3⁻CD138^hi cells (Figure 4E). The top DEGs (Tables S1 and S2) included genes coding for IL-10 and the surface proteins LAG-3, CD200, and CXCR3, whose differential expression was also reflected in protein amount (Figures 1 and 2). A Gene Set Enrichment Analysis (MSigDB) of the 648 DEGs in naive mice highlighted cell cycle regulation as the major pathway discriminating the LAG3⁻ and LAG3⁺ subsets (Figure S4G). 23 of the 648 DEGs were also differentially methylated between LAG-3⁺CD138^hi and LAG-3⁻CD138^hi cells, including Il10, Bcl2, Zfp36l1, Irf8, and other B cell regulators (Figure 4F). This argues for an epigenetically primed control of their expression.

We finally focused on transcription regulators differentially expressed in higher (or lower) amount on both day 0 and 1 p.i. in LAG-3⁺CD138^hi cells as compared to LAG-3⁻CD138^hi cells. We identified 22 factors (including transcription factors and co-factors, gene expression regulators, and general DNA binding proteins) with higher expression in LAG-3⁺CD138^hi cells and 8 factors with higher expression in LAG-3⁻CD138^hi cells (Figure 4G). LAG-3⁺CD138^hi cells showed a higher expression of Klf4, Bhlhe40, and Bmyc that are involved in the inhibition of proliferation, and of Hes1 that indicates higher Notch signaling activity. They also displayed a higher expression of Fos, Junb, Egr1, Egr2, Cebpb, Irf8, and Satb1 (Figures 4G and 4H), i.e., genes coding for factors for which we found predicted binding sites in the Il10, Lag3, Cd200, and Cd273 (PD-L2) loci (Figure 4H). FOS, JUN-B, and EGR-2 drive IL-10 expression in other immune cells (Iwasaki et al., 2013, Wang et al., 2005, Yoshida et al., 2012). EGR2 was also important in CD4⁺ T cells for LAG-3 expression, BLIMP-1-mediated induction of IL-10, and their regulatory activity (Iwasaki et al., 2013). The reduced expression of Foxm1 in LAG-3⁺CD138^hi cells (Figures 4G and 4H) might be linked to two effects: their lack of proliferation (Chen et al., 2013) and the control of Il10 transcription since Foxm1 binding sites were predicted in this locus (Figure 4H).

We conclude that LAG-3⁺CD138^hi cells exhibit a distinct transcriptome, partially linked to epigenetic changes, keeping them in a quiescent state primed for IL-10 expression. This molecular program could explain why only this subset of plasmocytes expresses IL-10 and other immune checkpoint receptor molecules.

BCR Repertoire and Developmental Origin of LAG-3⁺CD138^hi Cells

The differentiation of B cells into plasma cells requires the engagement of their BCR by antigen. Our observation that LAG-3⁺CD138^hi cells develop independently of the microbiota prompted us to examine whether they express a particular BCR repertoire.

Single-cell analysis of their IgH and Igk revealed that LAG-3⁺CD138^hi cells and IL-10⁺CD138^hi cells displayed a lower utilization of IgH VH1-81 and a higher usage of VH7, VH10, and VH11 segments compared to LAG-3⁻CD138^hi cells and IL-10⁻CD138^hi cells (Figures 5A and S5A). The majority of VH11⁺ IgH were associated with a Vk14.126⁺ (formerly Vk9) Igk (Figure 5B), forming a combination typical of B1a cells (Hardy et al., 2004). Flow cytometry analyses confirmed the presence of a VH11⁺Vk14.126⁺ BCRs on splenic LAG-3⁺CD138^hi cells and IL-10⁺CD138^hi cells but not on LAG-3⁻CD138^hi cells or IL-10⁻CD138^hi cells (Figures 5C and S5B). Of note, the BM from naive mice contained IL-10⁺CD138^hi cells (Figure 5D) enriched in VH11⁺Vk14.126⁺ cells (Figure 5E). Thus, some LAG-3⁺CD138^hi and IL-10⁺CD138^hi cells carry a VH11⁺Vk14.126⁺ BCRs typical of B1a cells (Figures S5C and S5D). However, the other VH segment classically associated with B1a cells, namely VH12, was not found in spleen LAG-3⁺CD138^hi cells, and only on about 1% of BM IL-10⁺CD138^hi cells (Figure 5F). The VH11⁺ IgH from LAG-3⁺CD138^hi and IL-10⁺CD138^hi cells had few mutations as well as N-additions, and a large proportion contained the CDR3 characteristic of the anti-phosphatidylcholine (PtC) response (Figures S5E and S5F). Accordingly, nearly all VH11⁺Vk14.126⁺ cells displayed reactivity against PtC (Figure 5G), and about 2% of splenic LAG-3⁺CD138^hi cells as well as 26% of BM IL-10⁺CD138^hi cells in naive mice reacted toward this particular antigen (Figure 5H). In BM, more than 80% of PtC-reactive CD138^hi cells expressed IL-10 (Figure S5G). Taken together, these data strongly suggest that VH11⁺Vk14.126⁺LAG-3⁺CD138^hi spleen and BM cells derive from B1a cells. Some LAG-3⁺CD138^hi cells expressing other BCR might also derive from B1a cells because LAG-3⁺CD138^hi and IL-10⁺CD138^hi cells had overall fewer IgH somatic mutations than LAG-3⁻CD138^hi and IL-10⁻CD138^hi cells (Figure 5I). This correlated with a lower AID fate-mapping marking of LAG-3⁺CD138^hi cells (Figure 5J). Furthermore, a higher proportion of IgH from LAG-3⁺CD138^hi and IL-10⁺CD138^hi lacked both somatic mutation and N-addition as compared to LAG-3⁻CD138^hi and IL-10⁻CD138^hi cells (Figure 5K). Nonetheless, some LAG-3⁺CD138^hi and IL-10⁺CD138^hi cells expressed IgH sequences with mutations and N-additions (Figure 5L), suggesting an additional origin for LAG-3⁺CD138^hi cells.

Figure 5.

BCR Repertoire of LAG-3⁺ and IL-10⁺ Plasmocytes

(A) Frequency of Igh containing VH1-81 (left) or VH11 (middle) in spleen LAG-3⁺CD138^hi (n = 253), LAG-3⁻CD138^hi (n = 150), Il10eGFP⁺CD138^hi (n = 181), and Il10eGFP⁻CD138^hi (n = 121) cells on day 0 and day 1 p.i. (SL7207; 10⁷ CFU). Frequency of LAG-3⁺ in VH11⁺CD138^hi and VH11⁻CD138^hi spleen cells of naive mice (n = 12) (right).

(B) Frequency of Vk14.126⁺ cells in VH11⁺ cells for spleen and BM cells described in Figure S5A.

(C) Frequency of VH11⁺Vk14.126⁺ cells by flow cytometry. Pool of two experiments (6–7 mice/group/time point).

(D) Flow cytometry plot of LAG-3 versus Il10eGFP in CD138^hi BM cells at day 1, and quantifications. Pool of four experiments (n = 12–17/time point).

(E) Flow cytometry plot and frequency of VH11⁺Vk14.126⁺ cells in IL-10⁺CD138^hi cells in BM of Il10eGFP mice. Pool of two experiments (n = 6–7/time point).

(F) Frequency of VH11⁺Vk14.126⁺ and VH12⁺ cells in B1a cells from indicated tissues (left), spleen LAG-3⁺ and LAG-3⁻ (middle), and BM IL-10⁺ and IL-10⁻ (right) plasmocytes from naive mice. Pool of two experiments (n = 7).

(G) Frequency of PtC-reactive cells in VH11⁺Vk14.126⁺ cells for indicated cells from naive mice. Pool of two experiments (n = 7).

(H) Frequency of PtC-reactive cells in indicated plasmocytes from naive mice. Pool of four experiments (n = 12).

(I) Mutations in IgH sequences (naive BM Il10eGFP⁺CD138^hi, n = 165; Il10eGFP⁻CD138^hi, n = 151; naive spleen LAG-3⁺CD138^hi, n = 231; LAG-3⁻CD138^hi, n = 129; day 1 spleen Il10eGFP⁺CD138^hi, n = 155; Il10eGFP⁻CD138^hi, n = 108) (left). Number of somatic mutations per sequence for mutated IgH (LAG-3⁺CD138^hi cells, n = 31; LAG-3⁻CD138^hi cells, n = 19) (right).

(J) Aicda-cre-ERT2-ROSA-STOP-RFP mice were treated with tamoxifen and analyzed 15 days after the last treatment. Frequency of RFP⁺ cells in LAG-3⁺CD138^hi and LAG-3⁻CD138^hi spleen cells. Pool of five experiments (n = 14).

(K) Frequency of IgH sequences (as in I) having no mutation and no N-nucleotide addition.

(L) Frequency of IgH sequences (as in I) with mutation and N-nucleotide addition.

(M) Frequency of LAG-3⁺ in spleen CD138^hi cells 3 weeks after transfer of indicated cell fractions into Rag2^−/− mice. Pool of three experiments (n = 6–7/group).

(N) Frequency of eYFP⁻ cells in indicated cells of naive Cd21-cre-ROSA-STOP-eYFP mice. Pool of three experiments (n = 6).

Groups were compared using two-tailed unpaired t test with Welch’s correction in case of inequal variances (C–I) or using Mann-Whitney test (M). Data are mean ± SEM (^nsp > 0.05, ^∗p < 0.05, ^∗∗p < 0.01; ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001). See also Figure S5.

To evaluate the capacity of various B cell subsets to generate LAG-3⁺CD138^hi cells, we adoptively transferred B1a, B1b, and B2 cells from peritoneal cavity (PeC) as well as splenic B cells into Rag2^−/− mice (Figure 5M). The analysis of the recipient mice confirmed that several B cell subsets could give rise to LAG-3⁺CD138^hi cells, as observed also upon culture of B cell subsets in vitro (Figure S5H). Marginal zone B cells were not a non-redundant source of LAG-3⁺CD138^hi cells in vivo since mice that lacked marginal zone B cells due to a B cell-restricted deficiency in NOTCH2 (Saito et al., 2003) displayed normal frequency of LAG-3⁺CD138^hi cells (Figure S5I) despite their globally reduced numbers of splenic plasmocytes (data not shown). To further delineate the origin of LAG-3⁺CD138^hi cells in a physiological context, we next developed a fate mapping system using the fact that B1a cells poorly expressed CD21 compared to other B cell susbsets, and crossed Cd21-cre mice to ROSA-STOP-eYFP reporter mice. In Cd21-cre-ROSA-STOP-eYFP mice, most B1a cells were eYFP⁻ in PeC, spleen, and BM, which was even more apparent when focusing on the VH11⁺Vk14⁺ fraction of B1a cells (Figure 5N). In contrast, B1b and B2 cells from PeC as well as follicular and marginal zone B cells from spleen were mostly eYFP⁺ (Figure 5N). The LAG-3⁺CD138^hi cell subset contained an intermediate frequency of eYFP⁻ cells, underlining their heterogeneous origin. Importantly, the LAG-3⁺CD138^hi cells that carried a VH11⁺Vk14⁺ BCR were largely eYFP⁻, indicating that eYFP⁻ B1a cells did not switch-on eYFP expression during their differentiation into LAG-3⁺CD138^hi cells (Figure 5N).

We conclude that LAG-3⁺CD138^hi cells develop from several B cell subsets via a mechanism involving the BCR since they have a distinct BCR repertoire.

Roles of BCR and TLR in LAG-3⁺CD138^hi Plasma Cell Development and Function

We next examined how the BCR influenced the accumulation of LAG-3⁺CD138^hi cells in naive mice and their upregulation of IL-10 after challenge.

The Bruton tyrosine kinase (Btk) is essential for BCR signaling (Corneth et al., 2016). Btk^−/− mice showed a developmental defect in B1a cells but not in B1b cells and marginal zone B cells (Figure 6A). They also displayed reduced frequencies and numbers of LAG-3⁺CD138^hi cells (Figure 6A). We next assessed the effect of the BCR co-receptors CD19 and CD72 for the generation of LAG-3⁺CD138^hi cells. We also studied mice genetically deficient in proteins involved in antigen presentation (CD1d), innate immune signaling (MyD88 and TRIF), T cell:B cell interaction (CD40, T cell receptor for antigen), and inflammation (NOS2, IL-10) (Figure 6B). LAG-3⁺CD138^hi cells were present in lower amounts in mice deficient for CD19, which is a positive regulator of BCR signaling, and in higher numbers in mice lacking CD72, which is an inhibitory BCR co-receptor. In contrast, they were normally present in Myd88^−/−Trif^−/− mice, which lacked TLR signaling. Mice deficient in CD40 or αβTCR-expressing T cells also generated these cells normally. LAG-3⁺CD138^hi cells can therefore develop independently of canonical T cell:B cell interactions and are thus not strictly dependent on a T cell-dependent antigen. The finding that BCR but not TLR signaling contributed to the formation of LAG-3⁺CD138^hi cells led us to assess whether mice with a restricted BCR repertoire had an altered abundance of these plasma cells. Indeed, B1-8i mice carrying a fixed IgH chain on an Igk-deficient background displayed a 2-fold reduction in the frequency of LAG-3⁺CD138^hi cells compared to controls (Figure 6C). Of note, antibodies cloned from LAG-3⁺CD138^hi cells displayed little polyreactivity when tested as recombinant molecules in ELISA (Tiller et al., 2009), unlike those from BM IL-10⁻IgM⁺CD138^hi cells (Figure S6A). Considering that BCR signaling contributed to the development of LAG-3⁺CD138^hi cells at steady state, we next asked whether such cells could be generated upon immunization with 4-hydroxy-3-nitrophenylacetyl keyhole limpet hemocyanin (NP-KLH), NP-Ficoll, or NP-LPS i.e., classical T cell-dependent, T cell-independent type II, and T cell-independent type I antigens, respectively (Figure S6B). These immunizations induced NP-reactive LAG-3⁻CD138^hi cells, but not NP-reactive LAG-3⁺CD138^hi cells. We conclude that LAG-3⁺CD138^hi cells develop from B cells at steady state upon a particular form of BCR engagement that is not recapitulated by these immunizations.

Figure 6.

Molecules Implicated in the Homeostasis of LAG-3⁺CD138^hi Cells

(A) Flow cytometry plots and quantification of indicated cells in PeC (top) and spleen (bottom) of naive Btk^−/− mice and littermate controls. Pool of two experiments (n = 6).

(B) Numbers and frequencies of LAG-3⁺CD138^hi cells in CD138^hi plasmocytes in spleen of naive C57BL/6 (n = 40), Cd19^−/− (n = 5), Il10^−/− (n = 8), Tcrab^−/− (n = 7), Cd72^−/− (n = 10), Cd1d^−/− (n = 6), Myd88^−/− (n = 10), Myd88^−/−Trif^−/− (n = 12), Cd40^−/− (n = 6), Nos2^−/− (n = 10), and Fcgr2b^−/− (n = 5) mice.

(C) Frequencies of LAG-3⁺ in spleen CD138^hi cells of B1-8iIgk^−/− and WT mice. Pool of two experiments (n = 4–5).

(D) Frequency of IL-10 in spleen LAG-3⁺CD138^hi of Il10eGFP mice 24 hr after i.v. injection of indicated reagents. Pool of three experiments (n = 6).

Data show mean ± SEM (^nsp > 0.05, ^∗p < 0.05, ^∗∗p < 0.01; ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001). See also Figure S6.

We next asked whether BCR or other signals controlled the upregulation of IL-10 expression in LAG-3⁺CD138^hi cells after challenge. To this end, Il10eGFP mice were injected intravenously with anti-BCR, anti-CD40, or LPS, alone or in combination (Figure 6D). The administration of anti-BCR or anti-CD40 had no effect on IL-10 expression, yet LPS induced a strong upregulation of IL-10 in LAG-3⁺CD138^hi cells. These results demonstrated that IL-10 expression was induced in a polyclonal manner in LAG-3⁺CD138^hi cells upon TLR4 stimulation. This was in agreement with the fact that LAG-3⁺CD138^hi cells from naive mice, as well as of LAG-3⁺CD138^hi cells and IL-10⁺CD138^hi cells from day 1 p.i., had comparable frequencies of VH11⁺Vk14⁺ cells (Figure 5C), suggesting that IL-10 was induced in LAG-3⁺CD138^hi cells in a polyclonal manner.

We conclude that the development of LAG-3⁺CD138^hi cells is controlled by the BCR and independent of TLR signaling as well as T cell help, while their upregulation of IL-10 after challenge is determined by TLR signaling.

Elevated Abundance of LAG-3⁺CD138^hi Cells Correlates with Impaired Immunity against Salmonella

An ablation of IL-10 production by CD138^hi plasmocytes increased host defense to Salmonella (Neves et al., 2010). Here, we sought to determine how an increased abundance of LAG-3⁺CD138^hi cells affected host resistance to this infection using Cd72^−/− mice.

Cd72^−/− mice had an increased number of LAG-3⁺CD138^hi cells, while other B cell subsets did not display any major defect (Figure S7A). The increased accumulation of LAG-3⁺CD138^hi cells in Cd72^−/− mutants was confirmed with littermate mice (Figure S7A). After infection, Cd72^−/− mice displayed an increased accumulation of IL-10⁺CD138^hi and LAG-3⁺CD138^hi cells (Figures 7A and S7B), an impaired control of the infection (Figure S7C), and a reduced survival (Figure 7B) compared to controls. The CD72 deficiency had no effect in B cell-deficient mice, confirming that the observed differences were due to the lack of Cd72 in B cells (Figure S7D). To validate the inhibitory impact of IL-10 on host defense in Cd72^−/− mice, mice were treated intravenously with a combination of anti-IL-10 and anti-IL-10 receptor. This significantly improved the control of the bacteria (Figure 7C). Since LAG-3⁺CD138^hi cells are the major source of IL-10 after infection, this supports the notion that LAG-3⁺CD138^hi cells inhibit host defense through IL-10 provision.

Figure 7.

CD72 Deficiency Enhances Susceptibility to Salmonella

(A) Numbers of indicated spleen cells in Il10eGFP and Cd72^−/−Il10eGFP mice p.i. (SL7207, 10⁷ CFU). Pool of three experiments (n = 9–12/time point/group).

(B) Survival of Cd72^−/− (n = 16) and controls (n = 16) after infection (SL1344, 100 CFU). Pool of two experiments.

(C) CFU on day 3 p.i. (SL7207, 10⁷ CFU) of anti-IL10 plus anti-IL10R-treated and control Cd72^−/− mice. Pool of two experiments (n = 6–10/group).

(D) Mice were vaccinated (SL7207, 10⁶ CFU) and re-challenged on day 90 (SL7207, 10⁷ CFU). Numbers of cells in spleen after re-challenge. Pool of three experiments (n = 9/group/time point).

(E) Mice were vaccinated (SL7207, 10⁶ CFU), rechallenged on day 90 (SL7207, 10⁶ CFU), and analyzed 5 days later with (+) or without (−) rechallenge. Pool of four experiments (n = 15–18/group/time point).

(F) CFU on day 5 post-re-challenge for mice shown in (E) (left). Survival of Cd72^−/− (n = 17) and littermate control (n = 19) mice vaccinated (SL7207, 10⁶ CFU) and re-challenged 90 days later (SL1344, 100 CFU). Pool of two experiments (right).

(G) Cd72^−/− and WT mice were vaccinated (SL7207, 10⁶ CFU), and BM analyzed on day 90. Flow cytometry plot (left) and quantifications (right) for CD40L, TNF-α, and IFN-γ expression in Salmonella-reactive memory CD4⁺ T cells. Pool of three experiments (n = 12).

(H) Flow cytometry plots showing PD-1 and FOXP-3 in spleen CD4⁺ T cells of mice treated as in (E) and analyzed on day 0 and 5 post-rechallenge. Pool of three experiments (n = 12/group).

Groups were compared using two-tailed unpaired t test (^nsp > 0.05, ^∗p < 0.05, ^∗∗p < 0.01; ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001). Survival curves were compared using Wilcoxon test, ^∗∗p < 0.01, ^∗∗∗∗p < 0.0001. Data show mean ± SEM. See also Figure S7.

Finally, we investigated how CD72 deficiency affected the host response to secondary infection. Il10eGFP and Cd72^−/−Il10eGFP mice were vaccinated and later challenged with Salmonella. Mice lacking Cd72 displayed a higher accumulation of IL-10⁺CD138^hi cells and LAG-3⁺CD138^hi cells upon re-challenge, while LAG-3⁻CD138^hi and total CD138^hi cells were present in comparable numbers in the two groups of mice (Figures 7D, 7E, and S7E). Consistently, CD72 deficiency led to higher bacterial load and reduced survival (Figure 7F), compromising vaccine efficacy. Memory T helper 1 (Th1) cells play a major role in vaccine-induced protection against Salmonella. In vaccinated Cd72^−/− mice, numbers of memory Salmonella-reactive Th1 CD4⁺ T cells in the BM were reduced (Figures 7G and S7F), and accumulation of exhausted PD-1⁺CD4⁺ T cells, PD-1⁺CD8⁺ T cells, and regulatory Foxp3⁺CD4⁺ T cells were increased (Figures 7H and S7G).

These data demonstrate that in the absence of CD72, a higher accumulation of LAG-3⁺CD138^hi plasma cells is associated with impaired immunity to Salmonella.

Discussion

This study demonstrates the existence of natural regulatory plasma cells, which are the major source of IL-10 early after infection with Salmonella.

An unexpected finding was that the cells expressing IL-10 were not reactive plasmablasts but instead resident plasma cells. Upon activation, B cells successively generated antibody-secreting plasmablasts that proliferated and then plasma cells that did not divide and expressed higher amounts of the transcription factor BLIMP-1. LAG-3⁺CD138^hi cells were non-proliferating and expressed higher amounts of BLIMP-1 than LAG-3⁻CD138^hi cells, thus qualifying as bone fide plasma cells. Their non-dividing status correlated with the lack of expression of factors necessary for cell division including the transferrin receptor CD71 and the transcriptional regulator Foxm1, as well as the increased expression of the cell proliferation inhibitors Egr2, Zfp36l1, Pdcd4, Klf4, and Bhlhe40. KLF4 is absent in plasmablasts generated in vitro but is expressed in long-lived BM plasma cells (Schoenhals et al., 2016). LAG-3⁺CD138^hi cells thus display a profile of mature plasma cells, while retaining the capacity to respond to external stimuli, in particular TLR, via the production of IL-10.

LAG-3⁺CD138^hi plasma cells expressed several immune checkpoint receptors implicated in the negative regulation of immunity such as LAG-3, PD-L1, PD-L2, and CD200. LAG-3 is a co-inhibitory receptor related to CD4 and implicated in the down-modulation of T cell immunity. The co-expression of several inhibitory molecules is important because these molecules usually act synergistically, so that their combined blockade achieves greater effect than the neutralization of single molecules. Blockade of LAG-3 and PD-L1 augments immunity toward Plasmodium falciparum by amplifying T cell activation and antibody production (Butler et al., 2011). Here, we found that LAG-3 on plasma cells did not influence their IL-10 expression, but inhibited the accumulation of IL-10⁻CD138^hi cells, which were LAG-3⁻, early after infection. The suppressive function of LAG-3⁺CD138^hi cells therefore extends beyond IL-10 production, suggesting a more global role in immune regulation.

LAG-3⁺CD138^hi cells developed from several B cell subsets in a BCR-dependent manner. Their preserved accumulation in germ-free mice suggests that they are selected on self-antigens. Our study has not identified the antigens involved in the generation of LAG-3⁺CD138^hi cells. Since these cells increased with age, which has also been associated with an accumulation of damaged cells (Baker et al., 2016), it is tempting to speculate that LAG-3⁺CD138^hi cells react toward antigens released by damaged cells. This hypothesis is consistent with the recognition by natural IgM of senescent cells (Frescas et al., 2017) and by natural antibodies binding PtC of apoptotic cells (Shaw et al., 2000). LAG-3⁺CD138^hi cells may thus provide a feedback mechanism, which senses the number of damaged cells and proportionally downregulates the activation of immunity to limit additional and potentially excessive immunopathology when an immune stimulus arises. Mice lacking the BCR inhibitory co-receptor CD72 had an increased abundance of LAG-3⁺CD138^hi cells and an impaired capacity to control Salmonella infection. The induction of IL-10 in LAG-3⁺CD138^hi cells was primarily controlled by TLR. We cannot, however, exclude that BCR signaling played a role in IL-10 production, even if it was not the major switch for IL-10 expression. For instance, it might contribute to the transcription of Il10 in LAG-3⁺CD138^hi cells at steady state in naive mice and synergize with TLR to upregulate IL-10 expression after challenge. It is indeed possible that these cells express BCR that recognize endogenous antigens already available at steady state and persisting during challenge.

In conclusion, natural regulatory plasma cells can rapidly provide a first layer of B cell-mediated immune regulation in response to TLR signals. They might subsequently be complemented by other subsets of suppressive B cells and plasma cells induced in an antigen-specific manner.

STAR★Methods

Key Resources Table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
B220	BioLegend	Clone RA3-6B2; RRID: AB_2563491
CD1d	BD PharMingen/BioLegend	Clone 1B1; RRID: AB_2073521, AB_1236543
CD3	BioLegend	Clone 145-2C11; RRID: AB_893317
CD4	BioLegend	Clone GK1.5 or RM4-5; RRID: AB_893323, AB_10898318
CD5	BD PharMingen/BioLegend	Clone 53-7.3; RRID: AB_394559, AB_2563930
CD8	BioLegend	Clone 53-6.7; RRID: AB_893423, AB_2561389
CD11b	BioLegend	Clone M1/70; RRID: AB_755986, AB_830642
CD11c	BioLegend	Clone N418; RRID: AB_830649, AB_389306
CD19	BioLegend	Clone 6D5; RRID: AB_493734, AB_830707
CD21/35	BD PharMingen/BioLegend	Clone 7G6/7E9; RRID: AB_395070, AB_1953277
CD23	BD PharMingen/BioLegend	Clone B3B4; RRID: AB_394652, AB_2103038, AB_312829
CD24	BioLegend	Clone M1/69; RRID: AB_2563464
CD40L	Miltenyi Biotec	Clone MR1; RRID: AB_2661127
CD40	DRFZ	Clone FGK-45
CD43	BD PharMingen	Clone S7; RRID: AB_10895376
CD69	DRFZ	Clone H1.2F3
CD71	BioLegend	Clone RI7217; RRID: AB_313564
CD72	BD PharMingen	Clone K10.6; RRID: AB_393982
CD79α	BioLegend	Clone F11-172; RRID: AB_2075634
CD79β	BioLegend	Clone HM79-12; RRID: AB_1575061
CD81	BioLegend	Clone Eat-2; RRID: AB_313138
CD93	eBioscience	Clone AA4.1; RRID: AB_469466
CD138	BD PharMingen/BioLegend	Clone 281-2; RRID: AB_395000, AB_2565621
CD200	BioLegend	Clone OX-90; RRID: AB_10900996
CXCR3	BioLegend	Clone CXCR3-173; RRID: AB_1088994
Foxp3	eBioscience	Clone FJK-16 s; RRID: AB_465243, AB_465936
Fc receptor	DRFZ	Clone 2.4G2
FAS	BD PharMingen	Clone Jo2; RRID: AB_396768
GL-7	BD PharMingen	Clone GL7; RRID: AB_10894953
IgM	BioLegend	Clone RMM-1; RRID: AB_2650758
IgD	BioLegend	Clone 11-26c.2a; RRID: AB_2562887
IFN-γ	BD PharMingen/BioLegend	Clone XMG1.2; RRID: AB_2034014, AB_1595591
Ki-67	BioLegend	Clone 16A8; RRID: AB_2561929
LAG-3 (CD223)	eBioscience	Clone C9B7W; RRID: AB_494214, AB_2573428
MHC-II	DRFZ	Clone M5/114
PD-1	BioLegend	Clone 29F.1A12; RRID: AB_2251944
TCR-β	BD PharMingen	Clone H57-597; RRID: AB_10584335
TNF-α	BD PharMingen/BioLegend	Clone MP6-XT22; RRID: AB_2562918, AB_469508
PD-L1	BD PharMingen	Clone MIH5; RRID: AB_397018
PD-L2	BioLegend	Clone TY25; RRID: AB_2566345
VH11	Prof. Kyoko Hayakawa (Fox Chase Cancer Center, Philadelphia, USA)	Clone P18-3H7
Vk14 (Vk9)	Prof. Kyoko Hayakawa (Fox Chase Cancer Center, Philadelphia, USA)	Clone P18-13B5
VH12	Prof. Klaus Rajewsky (MDC Berlin)	Clone 5C5
IL-10	BioXCell	Clone JES5-2A5; RRID: AB_1107696
IL-10R	BioXCell	Clone 1B1.3A; RRID: AB_1107611
Ig (H+L)	Southern Biotechnology	Cat. #1010-01; RRID: AB_609680
IgM-AP	Southern Biotechnology	Cat. #1020-04; RRID: AB_619829
IgG-AP	Southern Biotechnology	Cat. #1030-04; RRID: AB_609689
IgG1-AP	Southern Biotechnology	Cat. #1070-04
IgG2b-AP	Southern Biotechnology	Cat. #1090-04; RRID: AB_619828
IgG2c-AP	Southern Biotechnology	Cat. #1079-04; RRID: AB_2692321
IgG3-AP	Southern Biotechnology	Cat. #1100-04
IgA-AP	Southern Biotechnology	Cat. #1040-04; RRID: AB_619826
Goat Anti-Human IgG-UNLB	Southern Biotech	Cat. #2040-01; RRID: AB_617099
Goat Anti-Human IgG-Biotin	Southern Biotech	Cat. #2040-08
Human IgG standard	Sigma-Aldrich	Cat. #I2511; RRID: AB_1163604
Streptavidin-Alexa 488	Thermo Fisher Scientific	Cat. #S32354
Streptavidin-PE	BD PharMingen	Cat. #554061
Streptavidin-APC	BD PharMingen	Cat. #554067
Streptavidin-APC-Cy7	BioLegend	Cat. #405208
Streptavidin-BV421	BioLegend	Cat. #405225
Streptavidin-BV605	BioLegend	Cat. #405229
Streptavidin-BV785	BioLegend	Cat. #405249
Streptavidin-PE/Cy7	BD PharMingen	Cat. #557598
Streptavidin-PerCP	BD PharMingen	Cat. #554064
Anti-PE	Miltenyi Biotec	Cat. #130-048-801; RRID: AB_244373
B-1a Cell Isolation Kit	Miltenyi Biotec	Cat. #130-097-413
CD19	Miltenyi Biotec	Cat. #130-052-201
CD43	Miltenyi Biotec	Cat. #130-049-801
CD11b	Miltenyi Biotec	Cat. #130-049-601
CD11c	Miltenyi Biotec	Cat. #130-052-001
CD4	Miltenyi Biotec	Cat. #130-049-201
CD8	Miltenyi Biotec	Cat. #130-049-401
CD90.2	Miltenyi Biotec	Cat. #130-049-101
Chemical Peptides and Recombinant Proteins
DAPI	Sigma	Cat. #D8417
Propidium Iodide	Sigma	Cat. #P4864
Glutaraldehyde	Electron Microscopy Sciences	Cat. #16220
Osmiumtetroxide	Polysciences	Cat. #0223D
Tannic acid	Polysciences	Cat. #04459
Uranyl acetate	Ted Pella inc	Cat. #19481
CutSmart Buffer	NEB	Cat. #B7204S
Trypsin 0.5% EDTA	Life Technologies	Cat. #25300054
2-Mercaptoethanol	Life Technologies	Cat. #31350010
Calcium chlorure dihydrated	Roth	Cat. #5239.2
TMB	Ebioscience	Cat. #00-4201-56
Tween20	Sigma-Aldrich	Cat. #P1379
Critical Commercial Assays
Gene Chip Mouse 430 2.0	Affymetrix GmbH	Cat. #900495
Mouse Cell Screening (PE) kit	BioLegend	Cat. #700003
Foxp3 Staining Buffer Set	Ebioscience	Cat. #00-5523-00
BD Cytofix/Cytoperm Plus Kit	BD PharMingen	Cat. #554715
One Step RT-PCR kit	QIAGEN	Cat. #210212
Reverse Transcription System	Promega	Cat. #A3500
LightCycler FastStart DNA MasterPLUS SYBR Green	Roche Diagnostik	Cat. #03515885001
Bio-Plex Pro Mouse Cytokine IL-10 Set, 1 × 96 well	Bio-Rad	Cat. #171-G5009M
Deposited Data
Gene array data	This paper	GEO: GSE103458
RNA-seq and DNA methylation data	This paper	ENA: PRJEB22138
Other
Expression vectors containing the human IGG1 constant regions	Prof. Dr. Hedda Wardemann, (DKFZ, Heidelberg)	N/A
Expression vectors containing the human IGK constant regions	Prof. Dr. Hedda Wardemann, (DKFZ, Heidelberg)	N/A
Expression vectors codings for Human monoclonal antibody mGO53	Prof. Dr. Hedda Wardemann, (DKFZ, Heidelberg)	N/A
Expression vectors codings for Human monoclonal antibody JB40	Prof. Dr. Hedda Wardemann, (DKFZ, Heidelberg)	N/A
Expression vectors codings for Human monoclonal antibody ED38	Prof. Dr. Hedda Wardemann, (DKFZ, Heidelberg)	N/A
PtC-containing Liposomes labeled with Texas Red	FormuMax	Cat. #F60103F-TR
Lipopolysaccharides from Escherichia coli 055:B5	Sigma	Cat. #L2637-25MG
Tamoxifen	Sigma	Cat. #T5648-1G
Sunflower seed oil from Helianthus annuus	Sigma	Cat. #S5007-250ML
NP-KLH	BioCat GmbH	Cat. # N-5060-25
NP-LPS	BioCat GmbH	Cat. # N-5065
NP-AECM-FICOLL	BioCat GmbH	Cat. # F-1420
AffiniPure Fab2 Fragment Goat Anti-Mouse IgM	Dianova GmbH	Cat. #115-006-075
MacConkey Agar	Becton Dickinson GmbH	Cat. #212123
RNeasy Mini Kit	QIAGEN	Cat. #74106
Recombinant murine IL-6 protein	R&D Systems GmbH	Cat. #406-ML-005
Recombinant murine IL-21 protein	R&D Systems GmbH	Cat. #594-ML-010
MultiScreenHTS IP Filter Plate	Millipore	Cat. #MSIPN4510
BCIP/NBT substrate	Thermo Fisher Scientific	Cat. #34042
Proteinase K	Sigma Aldrich	Cat. #P2308
Pefabloc SC	Sigma Aldrich	Cat. #76307
HaeIII	New England Biolabs	Cat. #R0108M
Klenow Fragment exo-	New England Biolabs	Cat. #M0212L
T4-Ligase	New England Biolabs	Cat. #M0202M
T4-Ligase	New England Biolabs	Cat. #M0202L
RNase H	New England Biolabs	Cat. #M0297L
DNA Polymerase I	New England Biolabs	Cat. #M0209L
NEBNext High-Fidelity 2X PCR Master Mix	New England Biolabs	Cat. #M0541L
TruSeq DNA PCR-Free LT Library Preparation Kit - Set A	Illumina	Cat. #FC-121-3001
Nextera DNA Sample Preparation Kit	Illumina	Cat. #FC-121-1030
EZ DNA Methylation-Gold Kit	Zymo research	Cat. #D5006
HotStarTaq DNA Polymerase	QIAGEN	Cat. #203207
MinElute PCR Purification Kit	QIAGEN	Cat. #28006
Agencourt AMPure XP	Beckman Coulter	Cat. #A63881
mRNA Capture Kit	Roche	Cat. #11787896001
M-MLV Reverse Transcriptase, RNase H Minus, Point Mutant	Promega	Cat. #M3683
Agilent High Sensitivity DNA Kit	Agilent	Cat. #5067-4626
Pfu polymerase	Promega	Cat. #M7745
Hot Taq polymerase	Promega	Cat. #M7805
dNTP Master Mix	Eurogentec	Cat. #NU-0010-100
AgeI–HF	NEB	Cat. #R3552L
SalI–HF	NEB	Cat. #R3138L
BsiWI	NEB	Cat. #R0553L
NEBuffer 1	NEB	Cat. #B7001S
Gel and PCR clean up kit	Macherey-Nagel	Cat. #740609.250
10X Buffer T4 ligase	Promega	Cat. #C126A
T4 ligase	Promega	Cat. #M180B
One Shot TOP10 Chemically Competent E.coli	Invitrogen	Cat. #C404006
Luria Broth	Dutscher	Cat. #777495
Luria Agar	Dutscher	Cat. #777494
S.O.C Medium	Life Technologies	Cat. #15544034
Ampicillin sodium salt	Sigma-Aldrich	Cat. #A9518
NucleoSpin Plasmid	Macherey-Nagel	Cat. #740588.250
NucleoSpin 96 Plasmid Transfection-grade	Macherey-Nagel	Cat. #740491.4
DMEM	Life Technologies	Cat. #61965026
FBS South America	BIOWEST	Cat. #S1810-500
Penicillin/Streptomycin sol	Life Technologies	Cat. #15140-122
Nutridoma-SP	Sigma-Aldrich	Cat. #11011375001
Double strand DNA	Sigma-Aldrich	Cat. #31149-10G-F
Insulin	Sigma-Aldrich	Cat. #I9278-5ML
Bovine Serum Albumin	Sigma-Aldrich	Cat. #A7906-100G
Extravidin, Peroxidase Conjugate	Sigma-Aldrich	Cat. #E2886-1ML

Contact for Reagent and Resource Sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Simon Fillatreau (simonfillatreau@googlemail.com). Certain materials are shared with academic and non-profit research organizations for research and educational purposes only under an MTA to be discussed in good faith with the recipient.

Experimental Model and Subject Details

Mice

C57BL/6, Il10eGFP (Neves et al., 2010), Btk^−/−, Cd72^−/−, Cd19^−/−, Cd1d^−/−, Nos2^−/−, Rag2^−/−, B1-8iIgk^−/−, Fcγr2b^−/−, Cd19-cre-Notch2^fl/fl, Il10^−/−, Lag3^−/− (Miyazaki et al., 1996, Workman et al., 2004), Tcrα/β^−/−, Cd40^−/−, Myd88^−/−, Myd88^−/−Trif^−/−, J_HT, Cd72^−/−J_HT, Prdm1eGFP, Cd72^−/−Il10eGFP, Lag3^−/−Il10eGFP, Cd21-cre-ROSA-STOP-eYFP, Aicda-cre-ERT2-ROSA-STOP-RFP, C3H/HeOuJ were bred under specific pathogen-free conditions. C3H/HeOuJ mice were also bred under germ-free conditions. Mouse strains used in infection experiments were NrampS. B-Lag3^−/− mice were obtained by reconstituting recipient mice with a mixture of bone marrow cells from B cell-deficient J_HT mice (80%) and Lag3^−/− mice (20%). Control B-WT chimera was obtained using a mixture of bone marrow cells from J_HT mice (80%) and C57BL/6 mice (20%). B-Lag3^−/−Il10eGFP and their B-Il10eGFP controls were obtained using the same strategy. All experiments were reviewed and approved by appropriate institutional review committees (LAGeSo Berlin), and were conducted according to French, and German legislations, in compliance with European community council directive 68/609/EEC guidelines. Mice were of C57BL/6 strain, 6-14 weeks old at start of experiments, unless otherwise stated, and of sex-matched male and female genders.

Method Details

Bacterial infection

Mice were infected with 1x10⁶ or 1x10⁷ CFU for Salmonella typhimurium SL7207 strain or 100 CFU for SL1344 strain intravenously (i.v.). For survival and vaccination plus rechallenge experiments mice were first vaccinated with 1x10⁶ CFU of SL7207 given i.v. and 90 days later were rechallenged with 100 CFU of SL1344 or 1x10⁶ or 10⁷ CFU of SL7207. During survival experiments, mice were daily checked and presented as percentage of live animals. The bacterial loads were determined by plating a series of dilutions of homogenized organs on MacConkey agar plates. Heat-killed S. typhimurium (HKST) was prepared by inactivation of the SL1344 strain in water bath at 70°C for 1 hour. Bacteria were suspended in PBS. Salmonella infection was performed in a blinded manner, and identities of the mice were revealed upon termination of the experiment. No randomization was used. Estimation of size groups was based on our previous experience with these disease models, without a priori determination via power calculation. For neutralization of IL-10 signaling, mice were treated i.v. with a combination of anti-IL-10 (JES5-2A5; 500 μg/mouse/injection; BioXCell) and anti-IL-10R (1B1.3A; 500 μg/mouse/injection; BioXCell) on days −3, −1, and +1 after infection.

Tamoxifen administration

Tamoxifen was resuspended in sunflower seed oil from Helianthus annuus to a final concentration of 20mg/ml. 10mg of tamoxifen was administered by gavage to Aicda-cre-ERT2-ROSA-STOP-RFP mice twice with 5 days interval between the first and second administration. Mice were analyzed 15 days after the last administration.

Immunization and treatment with agonists of BCR, CD40, TLR4

C57BL/6 mice were immunized i.p. with 200 μg NP-KLH precipitated in alum, NP-Ficoll, or NP-LPS. Mice were analyzed 7 days after immunization. In house conjugated APC or PE nitrophenyl hapten was used to detected NP specific B and plasmocytes.

Il10eGFP mice were injected i.v. with 100 μg anti-BCR (Fab2 anti-mouse IgM), 50 μg anti-CD40 (FGK-45), or 10 μg LPS (E. coli 055:B5), alone or in combination. Mice were analyzed 24h later.

Isolation of plasmocytes and B cells

For gene array, quantitative PCR, restimulation in vitro, ELISPOT assays and transmission electron microscopy analysis, plasmocytes and B cell subsets were obtained from C57BL/6 or Il10eGFP mice, naive or on day 1 p.i. with 10⁷ CFU Salmonella (SL7207) by a two steps process. First, CD138⁺ cells were enriched by autoMACS after incubation with anti-CD138-PE followed by anti-PE microbeads. CD138⁺ cells obtained after autoMACS were then stained with anti-CD19, CD138, LAG-3, CD22 and CD11b/CD11c/TCR-β/DAPI. The following populations were sorted by FACS Sorter Aria I and/or II from this fraction: CD138^hiCD22⁺LAG-3^-CD11b^-CD11c^-TCR^-β^-DAPI^-, CD138^hiCD22^-LAG-3^-CD11b^-CD11c^-TCR^-β^-DAPI^-, CD138^hiCD22^-LAG-3⁺CD11b^-CD11c^-TCR^-β^-DAPI^-, CD138^hiCD11b^-CD11c^-TCR^-β^-DAPI^-, CD138^hiIl10eGFP⁺CD11b^-CD11c^-TCR-β^-DAPI^- and CD138^hiIl10eGFP^-CD11b^-CD11c^-TCR-β^-DAPI^-, CD138^hiLAG-3⁺CD11b^-CD11c^-TCR-β^-DAPI^- and CD138^hiLAG-3^-CD11b^-CD11c^-TCR-β^-DAPI^-. The cells from the CD138-negative fraction after autoMACS were stained with anti-CD19, CD138, CD1d, CD21, CD23 and CD11b/CD11c/TCR-β/DAPI. CD19⁺CD138^-CD11b^-CD11c^-TCRβ^-DAPI^-; CD19⁺CD1d^hiCD11b^-CD11c^-TCR-β^-DAPI^- and CD19⁺CD1d^lowCD11b^-CD11c^-TCR-β^-DAPI^-; CD19⁺CD1d^lowCD21⁺CD23⁺CD11b^-CD11c^-TCR^-β^-DAPI^- (FO) and CD19⁺CD1d^highCD21^highCD23^low/-CD11b^-CD11c^-TCR^-β^-DAPI^- (MZ) were then isolated from this fraction by FACS Sorter Aria I and/or II. Total B cells were obtained by magnetic isolation using negative selection with anti-CD11b, CD11c, and CD43 microbeads. For PeC, cells were incubated with B-1a Cell Biotin-Antibody Cocktail (Miltenyli Biotec) for 5 minutes at 2-8°C and then with anti-Biotin Microbeads for an additional 10 minutes in the same conditions in order to deplete non-B cells, according to Manufacturer’s instructions. LD columns were used for removing the non-B cells. Unlabeled cells were collected representing the pre-enriched B1a cell fraction that contained all PeC B cells fractions.

For single cell sorting, single CD138^hiIl10eGFP⁺CD11b^-CD11c^-TCR-β^-DAPI^- or CD138^hiIl10eGFP^-CD11b^-CD11c^-TCR-β^-DAPI^-, CD138^hiLAG-3⁺CD11b^-CD11c^-TCR-β^-DAPI^-, CD138^hiLAG-3^-CD11b^-CD11c^-TCR-β^-DAPI^- plasmocytes were directly sorted into 96-well PCR plates supplemented 1X RT-PCR buffer. Plates were then covered with microseal B film and immediately frozen on dry ice before storage at −80°C for further experiments.

For epigenetic and RNaseq analysis plasmocyte and B cell populations were sorted by FACS Sorter Aria I and/or II from spleen and PeC of naive C57BL/6 mice using the gating strategy illustrated in Methods S1.

For adoptive transfer, 10⁶ of flow cytometry-sorted B1a, B1b, and B2 cells from PeC, as well as CD19⁺CD138^- splenic B cells were administered i.p. into Rag2^−/− mice, which were analyzed 3 weeks later.

B cell stimulation and cytokine production measurement

Plasmocytes and/or B cell subsets were stimulated at 5x10⁵ or 1x10⁵ cells per well in 96-well flat or round bottom plates, respectively, in complete RPMI or with LPS (1μg/ml), α-IgM (Fab) (Dianova GmbH, 5μg/ml), agonistic α-CD40 antibody (clone FGK-45, produced in house at 10μg/ml), and in combinations in the presence of IL-6 (20ng/ml; R&D Systems GmbH), IL-21 (20ng/ml; R&D Systems GmbH). For analysis of cytokine production, 150μl supernatant of activated cells was collected and transferred into new 96 well plates and kept at −20°C until further use. 50μl of supernatant was used to measure cytokine concentration using Bio-Plex kits (Bio-Rad) according to the manufacturer’s instruction.

Transmission Electron Microscopy

Sorted cells, fixed in 2.5% glutaraldehyde (EM grade), were sedimented in warm low melting agarose (2% in PBS) and left to set. Excised bits of agarose containing cell-groups were then postfixed in 0.5% osmium tetroxide, contrasted with tannic acid and 2% uranyl acetate, dehydrated in a graded ethanol series and embedded in epoxy resin. After polymerization, sections were cut at 60 nm and contrasted with lead citrate. Specimens were analyzed in a Leo 906E transmission electron microscope at 100KV (Zeiss, Oberkochen, DE) using a side mounted digital camera (Morada; SIS-Olympus Münster DE).

Gene array hybridization

cRNA were hybridized on Affymetrix MG 430 2.0 arrays using standard Affymetrix protocol after quality control with Agilent 2100 Bioanalyzer and quantification with NanoDrop ND-1000 spectrophotometer.

Library preparation for methylation analysis of plasmocytes and B cell subsets

Sorted cells were lysed and digested with proteinase K followed by addition of 1mM Pefabloc SC (Sigma-Aldrich) to inhibit protease activity. Cell lysates were digested overnight with 50U HaeIII (New England Biolabs) followed by inactivation at 80°C. A-tailing was performed using 5U Klenow exo^- (New England Biolabs), and Illumina TruSeq adaptors were ligated at 16°C overnight using 2000U T4-Ligase (New England Biolabs). DNA fragments were purified with 1.5x Agencourt AMPure XP (Beckman Coulter) and bisulfite treated using the EZ DNA Methylation-Gold Kit (Zymo research) following the manufacturer’s protocol. The libraries were amplified by 14-15 cycles of PCR using HotStarTaq DNA Polymerase (QIAGEN), and purified with 0.9x Agencourt AMPure XP (Beckman Coulter). The library quality was assessed using Agilent 2100 Bioanalyzer, and the quantity was measured using Qubit® dsDNA HS Assay Kit (Thermo Fisher Scientific). The Illumina HiSeq 2500 system was used for 100 bp single-end sequencing.

Library preparation for mRNA Seq

mRNA Capture Kit (Roche) was used to isolate mRNA with the modification that cell pellets (10⁵ cells) were lysed in 49μl lysis buffer, and 1μl of the biotin labeled oligo(dT)₂₀ working solution was added to the lysate. After immobilization and washing of mRNA in streptavidin coated tubes first strand synthesis was performed in 50μl using 400U M-MLV Reverse Transcriptase RNase H^- (Promega) at 37°C for 1h. For second strand synthesis 400U T4-Ligase, 5U RNase H, and 50U DNA Polymerase I (New England Biolabs) were used at 16°C for 2.5h. The double stranded cDNA was fragmented and tagged with sequencing adapters using the Nextera DNA Library Preparation Kit (Illumina) with the deviation that 0.2μl of the Tagment DNA Enzyme 1 were used in a 50μl reaction. After incubation at 55°C for 5 minutes the reaction was stopped and the DNA was released from the tubes by the addition of 250μl of Buffer PB (QIAGEN). The DNA was purified using MinElute PCR Purification Kit (QIAGEN) following the manufacturer’s protocol, and the elution step was performed with 11.5μl distilled water. The libraries were amplified for 12 cycles using NEBNext High-Fidelity 2X PCR Master Mix (New England Biolabs) and 0.2μM Index adapters (AATGATACGGCGACCACCGAGATCTACAC[i5]TCGTCGGCAGCGTC and CAAGCAGAAGACGGCATACGAGAT[i7]GTCTCGTGGGCTCGG). The libraries were purified with 0.8x Agencourt AMPure XP (Beckman Coulter), and sequenced using the Illumina HiSeq 2500 system with 1x100 bp single-end reads.

Analysis of mRNA expression by B cells and plasmocytes

Sorted or cultured B cells and plasmocytes were lysed in RLT buffer (QIAGEN) and kept at −80°C until use. Total RNA was extracted using RNeasy Mini Kit (QIAGEN) according to the manufacturer’s instruction. After DNase treatment, cDNA was synthesized from RNA with a Reverse Transcription System (Promega). Quantitative RT-PCR was performed on an MX3005P QPCR System (Stratagene), with LightCycler FastStart DNA Master SYBR Green I (Roche). Transcripts were quantified using β-actin as standard, and the following forward (FP) and reverse (RP) primers (MWG Biotech): β-actin FP: 5′-TGGAATCCTGTGGCATCCATGAAAC-3′, β-actin RP: 5′-TAAAACGCAGCTCAGTAACAGTCC-3′; IL-10 FP: 5′-AGCCGGGAAGACAATAAC TG-3′, IL-10 RP: 5′-CATTTCCGATAAGGCTTG G-3′.

Ig gene amplification from single cells

Single cells sorted in 96-well PCR plate were immediately frozen on dry ice and stored at −80°C until further use. To amplify Ig gene from single cells, a nested PCR approach was used. Reverse transcription (RT) and the first PCR step for Igμ (IgM) and Igκ variable gene transcripts were carried out in a one-step reaction using the QIAGEN OneStep RT-PCR kit according to the manufacturer’s instructions. The Ig gene primers were chosen based on Tiller et al. (2009). The second round of PCR was carried out with 2μl of a 100-fold dilution of the unpurified first PCR product at 94°C for 4 minutes followed by 50 cycles of 94°C for 30 s, 60°C (Igμ) or 48°C (Igκ) for 30 s, 72°C for 45 s, and final incubation at 72°C for 10 minutes. The second PCR products were analyzed on 2% agarose gels.

The Ig gene PCR products from different plasmocyte subsets were sequenced (LGC) with the respective forward primers. Nucleotide sequences were then analyzed using IMGT/V-Quest to determine germline V, D, and J gene members, somatic hypermutation (counted for CDR1-FWR3 inclusively), CDR3 sequences, N-nucleotide addition.

Cloning of mouse Ig genes into expression vectors

After identifying V and J gene segment, second PCR reactions were repeated using 5μl of a 100-fold dilution of the unpurified first PCR product as template with specific V and J gene primers containing restriction sites to clone directly into expression vectors. PCR products were purified using Gel and PCR clean up kit before digestion with AgeI and SalI (Igh), AgeI and BsiWI (Igk). Digested Igh and Igk PCR products were purified from 0.9% agarose before ligation into expression vectors containing the human IGG1 and IGK constant regions, respectively. Ligation products were transformed into competent E.Coli via heat shock at 42°C, then E.Coli cells were plated on Ampicillin plates (100μg/ml) overnight at 37°C. Inserted genes were screened by PCR in Ampicillin resistant bacterial colonies.

Expression of recombinant immunoglobulin

Four single positively bacterial clones of each Igh and Igk gene were grown overnight at 37°C in 5ml of LB medium containing 100μg/ml Ampicillin. Plasmids were purified using NucleoSpin Plasmid - plasmid Miniprep kit (Macherey-Nagel), followed with digestion and analysis on 1% agarose. Inserted plasmids were then sequenced (GATC) to confirm identity with the second round PCR product and no mutation which probably is introduced by polymerase.

One day before transfection, 4x10⁶ HEK293 T cells were plated in T175 culture flasks in complete DMEM (DMEM + 10% (v/v) FCS + Penicillin (100U/ml)/Streptomycin (100 μg/ml) + 2-Mercaptoethanol (50 μM). The cell should be 80% confluency on the day of transfection. Transient transfections were performed with calcium-phosphate precipitation. In brief, equal amounts of IgH (7μg) and corresponding IgK chain vector were mixed with sterile water and 124μl of 2M CaCl₂. An equal volume of 1ml of 2X HBS was added drop-wise to the above mixture under slow vortexing and incubated at room temperature for 15min to allow formation of precipitates. The resulting solution was distributed to the culture flask. On the next day, cells were washed with 5ml of serum-free DMEM and 20ml of DMEM supplemented with 1% Nutridoma-SP, Penicillin (100U/ml)/Streptomycin (100 μg/ml) and 2-Mercaptoethanol (50 μM) was added. Supernatant were harvested at day 6 post transfection, cell debris was removed by centrifugation at 800 g for 10 minutes and culture supernatants were stored at −20°C.

Detection of antibody reactivity by ELISAs

IgG concentration in the supernatants were determined by ELISA. In brief, enhanced protein-binding ELISA plates were coated with 50μl of Goat Anti-Human IgG-UNLB diluted in carbonate buffer as capture antibody at 4°C overnight. Plates were washed 3 times with PBS. Unspecific binding was blocked by 200μl PBS/ 3%BSA for 1 hour at 37°C and washed again 4 times with PBS/0.1% Tween. Human IgG standard and culture supernatants were added, incubated for 2 hours at 37°C and washed again to remove unbound antibody with PBS/0.1% Tween. 100μl of Biotinylated Goat Anti-Human IgG Secondary Antibody was added, incubated for 1 hour at room temperature and washed again with PBS/0.1% Tween prior to adding ExtrAvidin^®−Peroxidase for 1 hour at room temperature. After washing, assays were developed using 100μl TMB solution and incubated at RT for color development. Assays were stopped by H₂SO₄ and ODs were measured at 450nm. Antibody concentration was determined by Softmax Pro software.

Antibodies with concentration higher than 5μg/ml were selected to test for polyreactivity with LPS, dsDNA and Insulin by ELISA. In brief, enhanced protein-binding ELISA plates were coated with 50μl of LPS (10μg/ml) or insulin (10μg/ml) or dsDNA (10μg/ml) in carbonate buffer and incubated at 4°C overnight. After washing, plates were incubated with 100μl nonpurified antibodies for 2 hours at 37°C and washed again. Nonpolyreactive (mGO53), low polyreactive (JB40), highly polyreactive (ED38) recombinant human monoclonal antibodies served as controls and were included on each plate. Unbound antibodies were removed by washing 4 times with PBS/0.1% Tween before detected using Biotinylated Goat Anti-Human IgG Secondary Antibody, followed with ExtrAvidin^®−Peroxidase for 1 hour at room temperature. The rest steps were performed as above mentioned.

Cell staining for flow cytometry

For surface staining, splenocytes, BM, PeC, and lymph node cells were first incubated with anti-Fc receptor antibody (clone 2.4G2) for 15 minutes to block unspecific bindings, and surface staining was then performed with antibodies conjugated directly with fluorochrome for 15 minutes. For biotinylated antibodies, cells were further incubated with Streptavidin conjugated to fluorochrome for 15 minutes. Dead cells were excluded by DAPI or PI.

For cytokine intracellular staining, cells were plated at 5x10⁶ cells per well in flat-bottom 48-well plates with 1ml complete RPMI 1640 containing HKST for 1 hour. Golgistop was added afterward, and cells were incubated for 5 more hours. Cells were harvested and stained for surface markers, including uncoupled pacific orange to exclude dead cells before cells were fixed. Intracellular staining was performed with Cytofix/Cytoperm kit (BD) as recommended.

Ki67 (BioLegend) and Foxp3 nuclear stainings were performed with Foxp3 staining kit (Ebioscience). Stained cells were acquired on FACS Symphony, Fortessa, Canto, LSR II and analyzed with FlowJo software.

ELISPOT Assay

Sorted day 0 CD138^hiLAG-3⁺ and CD138^hiLAG-3^- cells; day 1 CD138^hiIl10eGFP⁺ and CD138^hiIl10eGFP^- cells, and day 1 after re-challenge CD138^hiIl10eGFP⁺ and CD138^hi Il10eGFP^- cells were seeded at a starting number of 10⁴ cells per well, with seven successive three-folds serial dilutions, in MultiScreenHTS IP Filter Plate (Millipore) pre-coated with anti-mouse Ig(H+L) chain (5 μg/ml; Southern Biotechnology Associates, cat number 1010-01). After 3 h incubation, plates were washed, and incubated with alkaline phosphatase-conjugated anti-IgM or total IgG or IgG1 or IgG2b or IgG2c or IgG3 or IgA overnight at 4°C (Southern Biotechnology Associates). ELISPOT were then developed using BCIP/NBT substrate.

Quantification and Statistic Analysis

Data and statistical analyses were performed using GraphPad Prism (GraphPad Software, USA) and data were represented as mean ± SEM. Equality of variances between groups was assessed before analyses by t test. Groups were compared using two-tailed t test, or Wilcoxon test, or Mann-Whitney test. t test was modified using Welch’s correction in case of unequal variance (^nsp > 0.05, ^∗p < 0.05, ^∗∗p < 0.01; ^∗∗∗ p < 0.001, ^∗∗∗∗ p < 0.0001). No sample was excluded from analysis.

Gene array analysis

Normalization of microarrays was performed using gcrma R package (Wu et al., 2004). For each sample, we determined if a probe was present (i.e., expressed) or not using mas5calls function of affy R package (Gautier et al., 2004). The significantly differentially regulated genes were detected using limma R package (Ritchie et al., 2015) with p values adjusted according to Benjamini Hochberg procedure. In order to be selected in a comparison of two conditions, each Affy IDs had to fulfill the following criteria: (i) be present in at least two of the three arrays for at least one of the two conditions compared; (ii) to have a mean signal intensity higher than 50 in at least one of the two conditions; (iii) to show an adjusted (adj.) p value < 0.01 in the comparison of the two conditions. The genes differentially expressed between CD138^hiIl10eGFP⁺ and CD138^hiIl10eGFP^- plasmocytes (adj. p value < 0.01) were then selected and filtered using the gene ontology resource (www.geneontology.org) to focus on surface molecules. Heat-map of Figure 1A has been produced using pheatmap R package (pheatmap: Pretty Heatmaps. R package version 1.0.2. http://CRAN.R-project.org/package=pheatmap). If several probes were available for a given gene, their signal intensity has been averaged. For the hierarchical clustering, we used ward distance function and complete linkage function.

DNA methylation analysis

Trimming, alignment to the mm10 reference genome and DNA methylation value calling were performed as described before (Durek et al., 2016). Pairwise differential DNA methylation was calculated using the R package methylKit (Akalin et al., 2012) and DMR were defined as 1kb regions covering ≥ 3 CpG sites with ≥ 20% methylation difference (q-value ≤ 0.01) between the cell types. All analyzed CpG were filtered for coverage ≥ 10x and presence in all replicates. Average DNA methylation per cell type was calculated using coverage weighted average methylation of 3 replicates. DMR were annotated to genes and genomic location by using ChipSeek (Chen et al., 2014). Transcription factor binding sites were predicted using TRAP (Thomas-Chollier et al., 2011) and AliBaba2 (http://gene-regulation.com/pub/programs/alibaba2/index.html). The integrated Genome Viewer (IGV) genome browser was used to display data in Figure 4D and Figure S4D.

RNaseq analysis

Adaptor removal and trimming of low quality ends (phred score = 20) of FastQ format reads was performed using Trim Galore! (version 0.3.3) (http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). Alignment to the mm10 reference genome was done using 2-step STAR alignment (Dobin and Gingeras, 2015). PCR duplications were determined using MarkDuplicate (version 1.115) from Picard tools (http://broadinstitute.github.io/picard/) and quality control was performed using RNA-seQC (DeLuca et al., 2012). Subsequently, featureCounts (Liao et al., 2014) was used to count reads mapping to the genes from Gencode annotation (vM2) (https://www.encodeproject.org/files/gencode.vM2.annotation). Pairwise differential expression was analyzed using the R package edgeR (Robinson et al., 2010). A p value cutoff of 0.01 and an FDR cutoff of 0.05 was set to select differentially expressed genes. GO Term analysis were performed using GREAT (McLean et al., 2010).

Data and Software Availability

The Affymetrix gene array data have been deposited at NCBI GEO depository and are accessible with the accession number GSE103458, or using the link https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE103458

The RNaseq and DNA methylation data have been deposited at the European Nucleotide Archive (ENA), and will be accessible with the accession number PRJEB22138, or using the links.http://www.ebi.ac.uk/ena/data/view/PRJEB22138<https://deref-web-02.de/mail/client/dAlzMDs2lmY/dereferrer/?redirectUrl=http%3A%2F%2Fwww.ebi.ac.uk%2Fena%2Fdata%2Fview%2FPRJEB22138>

Published: July 10, 2018