IL-33 Is a Cell-Intrinsic Regulator of Fitness during Early B Cell Development Free

This article is featured in In This Issue, p.1407

Interleukin-33 is an IL-1 family member protein that is a key mediator of innate and adaptive immune responses. IL-33 has been extensively studied as an extracellular signal that binds to a heterodimeric receptor complex composed of IL1RL1, also known as ST2, and the shared IL-1 receptor accessory protein (IL-1RAcP) (1–4). Primary targets of IL-33 include group 2 innate lymphoid cells (ILC2), mast cells, macrophages, dendritic cells, and CD4⁺ Th2 cells with resultant type 2 inflammatory responses (3, 5). Additionally, a subset of CD4⁺ T regulatory cells expresses ST2 and expands in response to IL-33, demonstrating that extracellular IL-33 can both promote and attenuate inflammation (5).

Major sources of IL-33 include epithelial cells in the lung, skin, gastrointestinal tract, and reproductive tract (2, 6, 7). IL-33 is also highly expressed in endothelial cells in adipose tissue, liver, and secondary lymphoid organs, as well as in fibroblastic reticular cells within lymph nodes (7–12). During acute inflammation and in adipose tissue, macrophages may also be a source of IL-33 (6, 13). However, IL-33 expression has largely been observed and studied in nonhematopoietic cells.

IL-33 has also been hypothesized to have an intracellular role as a transcriptional regulator. IL-33 is a two-domain protein. The C-terminal domain (aa 112–270 in humans) contains IL-1 family member homology and is responsible for mediating the extracellular, ST2-dependent effects of IL-33 (2). In contrast, the N-terminal domain (aa 1-111 in humans) targets IL-33 to the nucleus (9, 14), has a chromatin-binding motif (15), and exhibits potent transcriptional repressor capacity in an artificial tethered gene reporter assay (14, 15). Multiple studies have attempted to define gene targets of IL-33 regulation. Several small, focused investigations have suggested single or a few putative targets including IL6 and RELA (NF-κB p65) (16–19). Yet, in two large studies, intracellular IL-33 had no influence on the global transcriptome or proteome of cultured human esophageal epithelial cells or HUVECs, respectively (20, 21). Notably, all of these studies were conducted in vitro. Thus, an intracellular regulatory role for endogenous IL-33 and its in vivo relevance has yet to be conclusively established.

B cell development in the bone marrow is characterized by somatic recombination leading to formation of the BCR and rapid expansion of the B cell lineage. This process proceeds in a stepwise fashion, with committed development starting at the progenitor B (pro-B) cell stage, where the BCR undergoes H chain recombination. Hyperproliferation marks the transition to the large precursor B (LPB) cell stage, wherein the developing B cells rapidly expand to increase the number of cellular templates for BCR L chain rearrangement. Cessation of proliferation and initiation of L chain rearrangement denotes the transition to the small precursor B (SPB) stage. Following L chain rearrangement, developing B cells proceed to the immature B cell stage, where they undergo clonal selection and functional maturation prior to exiting the bone marrow. Although substantial effort and progress have been made to define the molecular cues that guide B cell development, the full complement of cell exogenous and endogenous regulators of B cell development remains undetermined.

In this study, we identified IL-33 expression during the early stages of B cell development in the bone marrow of both mice and humans. IL-33 expression restricted the capacity of B cells to repopulate the hematopoietic compartment in vivo through a cell-intrinsic, ST2-independent mechanism. Collectively, these data reveal an intracellular role for IL-33 in regulating early B cell development.

Animal experiments were performed with age-matched 8–14-wk-old male and female mice unless otherwise noted. Wild-type (WT) BALB/c, WT BALB/c CD45.1 [CByJ.SJL(B6)-Ptprc^a/J; stock no. 006584], Il33^GFP/GFP [also referred to as Il33^fl/fl; B6(129S4)-Il33^tm1.1Bryc/J; stock no. 030619], CD19-Cre [B6.129P2(C)-Cd19^tm1(cre)Cgn/J, stock no. 006785], and WT C57BL/6 CD45.1 (B6.SJL-Ptprc^aPepc^b/BoyJ; stock no. 002014) mice were obtained from The Jackson Laboratory. Il33^cit/cit mice (also referred to as Il33^−/− mice as the citrine reporter construct replaces endogenous Il33) and St2^−/− mice were graciously provided by Dr. A. McKenzie. Il33^Cit/Cit and St2^−/− mice were on a BALB/c background. Il33^GFP/GFP (Il33^fl/fl) and CD19-Cre mice were on a C57BL/6 background. In indicated experiments, Il33^Cit/Cit mice were bred with WT BALB/c to generate Il33^Cit/+ hemizygous reporter mice with an intact copy of Il33. Il33^fl/fl mice were bred with CD19-Cre mice to generate mice with a CD19-specific reduction of Il33 expression (Il33^fl/fl Cd19^Cre/+). WT BALB/c and WT BALB/c CD45.1 mice were bred to generate WT BALB/c mice with coexpression of CD45.1 and CD45.2 (referred to as WT BALB/c CD45.1/CD45.2). Animals were maintained in specific pathogen-free housing. All animals were used in compliance with the revised 1996 Guide for the Care and Use of Laboratory Animals prepared by the Committee on Care and Use of Laboratory Animals of the Institute of Laboratory Animal Resources, National Research Council. All animal experiments were approved by the Vanderbilt Institutional Animal Care and Use Committee.

Bone marrow aspirates from healthy human volunteers were collected and processed for flow cytometric analysis of IL-33 in accordance with the Vanderbilt Institutional Review Board (no. 160060, 081012, and 160420). Briefly, bone marrow aspirates were collected in acid citrate dextrose tubes. RBC lysis was performed using room-temperature ammonium/chloride/potassium buffer, and washed samples were frozen in liquid nitrogen. All processing occurred within 24 h of sample collection; samples were thawed immediately before flow cytometric analysis. Donors averaged 34.4 (±11.0) y of age and were 75% men/25% women and 91.7% white/8.3% Hispanic. For gene expression analysis in human B cell malignancies, we analyzed 45 publicly available microarray gene expression-profiling datasets for chronic lymphocytic leukemia (CLL), six different B cell lymphomas (Burkitt, diffuse large B cell lymphoma [DLBCL], follicular, mantle cell, marginal zone, and Hodgkin), and normal B cells (germinal center centroblasts and centrocytes, memory B cells, and CD19⁺ B cells). We collected the data from gene expression omnibus (GEO) (22) or the authors’ Web site (23). The GEO identifications for CLL data are the following: GSE26725, GSE50006, GSE16455, GSE26526, GSE21029, GSE69034, GSE49896, GSE39671, GSE9250, and GSE22762, and the datasets for the six lymphoma types and normal samples are previously described (24). The data were generated on the Affymetrix Human Genome U133 Plus 2.0 platform following the standard Affymetrix protocol. The data were separated into batches, normalized using the robust multiarray average algorithm (25) in the Affymetrix package for R (26), and batch effects corrected using ComBat (27). Probe sets were averaged to obtain a single-expression intensity measure per gene per array if multiple probe sets corresponded to the same gene identification. Expression values from replicate samples were averaged. All analyses were carried out in R, version 3.3.2. Differential expression was measured by unpaired, two-tailed t test with Welch correction using GraphPad Prism v.7.

Murine bone marrow was prepared by flushing the tibia and femur of euthanized mice with RPMI 1640 and performing RBC lysis followed by straining the cells through a 70-μm filter to remove debris. Cells were incubated with anti-CD16/CD32 for 15 min at 4C to block Fc-mediated binding of Abs, followed by staining with the indicated Abs to discriminate populations of interest (Supplemental Table I). Live/Dead discrimination was performed by staining with propidium iodide, DAPI, 7-aminoactinomycin D (7-AAD), or Ghost Dye UV450 (for fixed samples) prior to analysis or fixation. Spleens were initially homogenized through a 70-μm filter and processed otherwise identical to bone marrow. For BrdU analysis, animals were injected i.p. with 1 mg of BrdU at 1 h prior to harvest and processed via manufacturer instructions. For intracellular IL-33 detection, human bone marrow was stained for surface markers and fixed for 60 min with the Foxp3/Transcription Factor Staining Kit (catalog [Cat.] no. TNB-0607-KIT; Tonbo Biosciences). Samples were stained at 4°C for 30 min with anti-human IL-33 or isotype control Ab in permeabilization buffer. Samples were evaluated on a BD LSR II and analyzed with FlowJo software. For FACS isolation, samples were magnetically enriched for CD19⁺ cells (Cat. no. 130-121-301; Miltenyi Biotec), surface stained as described for analytical flow cytometry, and separated on a BD FACSAria III.

RNA was isolated from FACS-purified cells using the QIAGEN RNeasy Micro Kit (Cat. no. 74004) per the manufacturer protocol with on-column DNase I digestion. cDNA was generated using the Invitrogen SuperScript IV First-Strand Synthesis System (Cat. no. 18091050) per the manufacturer protocol using Oligo(dT) priming. Quantitative PCR was performed using FAM–minor groove binder TaqMan primer/probes from Applied Biosystems/Thermo Fisher Scientific: Il33 exon 2-3 (Mm00505399_m1), Il33 exon 3-4 (Mm01195783_m1), Il33 exon 6-7 (Mm00505403_m1), and Gapdh (Mm99999915_g1). Reactions were carried out with the TaqMan Gene Expression Master Mix (Cat. no. 4369016) on an Applied Biosystems QuantStudio12K Flex Real-Time PCR machine. Samples were analyzed using Applied Biosystems QuantStudio 12K Flex Software v.1.2.2. Relative gene expression was calculated using the ΔΔ cycle threshold (ΔΔCt) protocol with respect to the indicated reference gene.

Nuclear and cytoplasmic protein fractions from FACS-purified cells were isolated using the Thermo Fisher Scientific NE-PER Nuclear and Cytoplasmic Extraction Reagents (Cat. no. 78833) per the manufacturer protocol. Total protein was assessed for each sample using the Pierce BCA Protein Assay Kit (Cat. no. 23225). Equivalent protein (2.5 μg) per sample was loaded and separated on a 4–20% gradient polyacrylamide gel. Gels were rinsed, wet transferred to a nitrocellulose membrane, and blocked with Odyssey Blocking Buffer in TBS (Cat. no. 927-50000, LI-COR Biosciences). Blots were incubated overnight at 4°C while shaking with the indicated primary Abs: anti–IL-33 (Cat. no. AF3626; R&D), anti–Lamin B1 (Cat. no. ab16048; Abcam), or anti-Tubulin (Cat. no. 3873; Cell Signaling Technology). Blots were rinsed and incubated for 1 h at room temperature while shaking with the appropriate secondary Ab: anti-mouse IgG 680LT (Cat. no. 68022; LI-COR), anti-rabbit IgG 680LT (Cat. no. 68023; LI-COR), or anti-goat IgG 800 CW (Cat. no. 32214; LI-COR). Blots were rinsed and imaged on an Odyssey Classic Infrared Imaging System. Image capture and densitometry were performed using Image Studio v.3.1 software.

Pro-B and LPB cells were MACS enriched and FACS purified from the bone marrow of WT and Il33^−/− mice. Cells were sorted into cold PBS, pelleted, and lysed with Buffer RLT (QIAGEN). RNA was isolated using the QIAGEN RNA Mini Kit (Cat. no. 74104) with on-column DNase I digestion following the manufacturer instructions. RNA sequencing (RNA-seq) libraries were prepared using 20–50 ng of total RNA and Illumina TruSeq Stranded mRNA Kit (Cat no.: 20020595; Illumina) per manufacturer’s instructions, with mRNA enriched via poly(A) selection using oligo(DT) beads. The RNA was then thermally fragmented and converted to cDNA, adenylated for adapter ligation, and PCR amplified. The libraries were sequenced using the HiSeq 3000 with 75 bp end reads with a depth of ∼30 million reads per sample. Real-Time Analysis (version 2.7.6; Illumina) was used for base calling, and analysis was completed using MultiQC v1.2. On paired FASTQ files, transcript abundance was estimated, and these read counts were imported into edgeR for differential expression analysis (28). In edgeR, data were normalized based on negative binomial distribution. The differential expression of genes between Il33^−/− and WT samples was assessed by estimating an exact test p value. Venn diagram of differential gene expression was generated with Meta-Chart (https://www.meta-chart.com/venn#/display). RNA-seq data have been deposited in the GEO database under the access code GSE132123 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE132123).

We conducted gene set enrichment analysis (GSEA) (29) using the Kyoto Encyclopedia of Genes and Genomes (KEGG) suite (30) and the Hallmark gene set (31) embedded in GSEA to identify pathways that were differentially expressed in Il33^−/− samples compared with WT. GSEA was run on preranked list of the genes obtained from the edgeR analyses. Ranking was based on fold-change induction in Il33-null cells compared with control cells. Mouse genes were converted to human gene symbols before conducting GSEA preranked analysis. GSEA results were assessed as being statistically significant by permutation of 1000 samples.

WT BALB/c CD45.1/CD45.2 mice or WT C57BL/6 CD45.1 mice were lethally irradiated with 9 or 12 Gy, respectively, from a Cs-137 source. Recipient mice were transplanted with a 1:1 mixture of whole bone marrow (10 million total cells) from the indicated strains via retroorbital injection. Mice were maintained on antibiotic water (0.025% trimethoprim/0.125% sulfamethoxazole) for 2 wk following transplant. Mice were rested for 5–8 wk to allow for reconstitution prior to analysis. For WT BALB/c CD45.1/CD45.2 mice, residual cells from the recipient mice were gated as CD45.1⁺CD45.2⁺ by flow cytometry and excluded from analysis.

SPB cells were MACS enriched and FACS purified from the bone marrow of female 8-wk-old naive adult WT, Il33^−/−, and St2^−/− mice. Genomic DNA was isolated using the QIAGEN DNeasy Blood & Tissue Kit (Cat. no. 69504) and 0.5 μg of genomic DNA per sample was used for sequencing. Sequencing of the BCR hypervariable CDR3 was performed by the multiplex PCR-based immunoSEQ Assay (Adaptive Biotechnologies). BCR repertoire analyses were performed using the immunoSEQ Analyzer 3.0 (Adaptive Biotechnologies).

Data were managed and analyzed using GraphPad Prism version 7. When possible, data were pooled from multiple experiments. Mann–Whitney U test or one-way ANOVA with Tukey posttest were used to determine statistical significance, as appropriate and as indicated.

We sought to characterize IL-33 expression in the bone marrow compartment. Using a BALB/c mouse model with a citrine reporter construct under the control of the endogenous IL-33 promoter (Il33^cit/+), we identified a population of IL-33–expressing cells in the bone marrow; of which, >95% were CD19⁺ B cells (Fig. 1A). Subset analysis in Il33^cit/+ mice demonstrated IL-33 expression during the pro-B cell stage that persisted through the LPB cell stage and decreased in SPB and immature B cells (Fig. 1B, 1C, gating strategy in Supplemental Fig. 1A). Per-cell expression of IL-33 was similar in pro-B and LPB cells, and higher in these early B cell populations than in SPB and immature B cells (Fig. 1D). Consistent with our flow cytometric analysis, FACS-purified pro-B and LPB cells expressed significantly higher levels of Il33 mRNA than SPB and immature B cells (Fig. 1E). Fractionation of FACS-purified B cells into nuclear and cytoplasmic protein, followed by immunoblotting demonstrated directly the presence of IL-33 protein and that this IL-33 protein was localized almost exclusively in the nucleus in pro-B and LPB cells (Fig. 1F, 1G).

We verified these results using an IL-33/GFP reporter mouse strain (Il33^GFP/GFP) independently developed on a C57BL/6 background. We identified IL-33 expression beginning in the pro-B cell stage that peaked as a percentage of the total population in the LPB cell stage and was only minimally detected in SPB and immature B cells (Fig. 1H, 1I). In addition, the mean fluorescence intensities (MFIs) of GFP in IL-33–expressing pro-B and LPB cells were higher than SPB and immature B cells (Fig. 1J).

Pre-pro-B cells are CD19⁻ but represent the stage of B cell development immediately preceding pro-B cells. Flow cytometric analysis of both Il33^cit/+ and Il33^GFP/GFP mice demonstrated no expression of IL-33 in pre-pro-B cells, indicating that IL-33 expression first occurs during the pro-B cell stage (Fig. 1K). Peripherally, IL-33 expression was not detected in splenic B cells or peritoneal B-1 B cells in adult naive mice (Supplemental Fig. 2A, 2B). IL-33 expression was rare, but detectable, in B cells within Peyer's patches, representing <1% of CD19⁺B220⁺ B cells (Supplemental Fig. 2C). Within this IL-33⁺ B cell population, a larger percentage were IgM⁻ representing, class-switched B cells (Supplemental Fig. 2C). Moreover, IL-33 expression among developing lymphocytes was unique to the B cell lineage, as developing T cells in the thymus universally lacked IL-33 expression (Supplemental Fig. 2D). Taken together, these data demonstrate that IL-33 is expressed within the B cell lineage primarily during the pro-B and LPB cell stages of development, and this expression pattern is independent of the mouse strain.

Given this expression pattern of IL-33 in pro-B and LPB cells and the known importance of IL-33 as an extracellular signal, we sought to determine whether developing B cells expressed ST2 and, therefore, had the capacity to respond to extracellular IL-33. ST2 expression was absent in all stages of B cell development in WT BALB/c mice, with respect to an isotype control Ab (Fig. 2A, 2B). In contrast, ST2 was readily detectable on bone marrow ILC2 progenitors, which are highly responsive to IL-33 (Fig. 2C, 2D) (32). Equivalent results were obtained using St2^−/− mice as a negative control (Fig. 2E–H). Collectively, these data demonstrate that developing B cells lack expression of the IL-33 receptor ST2.

To assess for functional differences between WT and Il33^−/− pro-B and LPB cells, we performed transcriptome analysis with mRNA sequencing. RNA-seq of FACS-purified WT and Il33^−/− pro-B and LPB cells revealed a distinct IL-33–dependent transcriptome in both pro-B and LPB cells. A total of 275 and 983 genes were differentially expressed with a false discovery rate (FDR) < 0.05 between WT and Il33^−/− pro-B and LPB cells, respectively (Fig. 3A). Of these, 179 and 281 genes were differentially expressed with at least a 1.5-fold change between WT and Il33^−/− pro-B and LPB cells, respectively. Among differentially expressed transcripts, 53 upregulated genes and 64 downregulated genes were shared between pro-B and LPB cells (Fig. 3A).

GSEA identified numerous candidate pathways within the Molecular Signatures Database Hallmark (31) and the KEGG gene set collections (30) that were modulated in early developing B cells by IL-33 (FDR < 0.05). Notably, we identified multiple pathways related to proliferation and cell death, which were of particular interest given that rapid cellular expansion and survival are cardinal features of early B cell development at the pro-B and LPB cell stages. Specifically, Il33^−/− pro-B and LPB cells were enriched for targets of E2F transcription factors and genes involved in DNA replication and the cell cycle, consistent with possible enhanced proliferation (Fig. 3B) (33). Il33^−/− pro-B and LPB cells also exhibited a downregulation of p53-dependent pathways, suggesting increased potential for cell survival (Fig. 3B). Collectively, these data suggest that IL-33 may attenuate fitness during early B cell development by decreasing proliferation or cell survival.

We next evaluated Il33 expression in WT pro-B and LPB cells in relation to known transcriptional regulators of B cell development. Il33 expression in pro-B and LPB cells was less than hallmark early B cell transcriptional regulators, including Pax5, Tcf3 (encoding E2A), and Ebf1, but was at the low end of expression of endogenous early B cell regulatory proteins, which include Gfi1, Spi1 (encoding PU.1), Bcl6, and Gfi1b (Fig. 3C). In contrast, key T cell and NK cell transcription factors were not detected (Fig. 3C). Il1rl1 encodes for the IL-33 receptor ST2 and was undetectable in both pro-B and LPB cells (Fig. 3D), consistent with our flow cytometric analysis of ST2 expression on developing B cells (Fig. 2). This is similar to other non–B cell receptors and surface proteins, including Cd4, Cd8, and Ncr1, which were not detected, and in contrast to known pro-B and LPB cell receptors and surface proteins, including Cd24a, Cd19, and Il7r, that were highly expressed (Fig. 3D).

We sought to compare Il33 expression between early-developing B cells and stromal cells, many of which are known to express and release IL-33 extracellularly for binding to ST2 and immune modulation. We compared Il33 expression in the ultra-low-input RNA-seq dataset (GSE127267) from the Immunological Genome Project among B cell subsets and stromal cells (34). Within B cell subsets, Il33 expression was detected in Hardy fraction B/C cells within the bone marrow, which represent pro-B cells, and are absent in mature splenic B cell populations, consistent with our expression analysis (Fig. 3E). In contrast, Il33-expresing stromal cells including fibroblastic reticular cells and lymphatic endothelial cells of subcutaneous lymph node expressed Il33 at 40–100 fold higher levels than pro-B cells (Fig. 3E). Overall, expression of Il33 in early developing B cells is markedly less than stromal cells and more aligned with established intrinsic regulators of developing B cells.

Given the identification of prosurvival gene signatures in Il33^−/− pro-B and LPB cells, we assessed during in vivo B cell development for differences in B cell fitness (i.e., the ability of B cells to survive and mature). Moreover, because developing B cells lacked ST2 expression, we evaluated whether any IL-33–dependent differences in fitness were due to a cell-intrinsic or cell-extrinsic mechanism. In competitive reconstitution experiments with mixed bone marrow chimeric mice, WT and Il33^−/− pre-pro-B cells were identified in approximately equal quantities (Fig. 4A–C). Alternatively, highly stringent gating for pre-pro-B cells (B220⁺ IgM⁻ CD19⁻ CD93⁺ Ly-6D⁺ PDCA⁻) resulted in a similar frequency as our initial pre-pro-B cell analysis (Supplemental Fig. 1B). However, Il33^−/− pro-B cells accumulated to a significantly higher frequency than WT pro-B cells, with a ratio of ∼2:1 Il33^−/− to WT (Fig. 4B, 4C). This enhanced fitness of the Il33^−/− B cell lineage compared with WT B cells persisted throughout development, with a nearly 3:1 ratio of Il33^−/− to WT cells at the immature B cell stage (Fig. 4B, 4C). Moreover, this increased fitness during B cell development led to alterations in peripheral B cell repopulation, with increased frequency of Il33^−/− B cells in the spleen compared with WT (Fig. 4D). These frequency differences translated into significant increases in the total number of Il33^−/− relative to WT B cells in the bone marrow and spleen downstream of IL-33 expression (Fig. 4E, 4F).

To further control for exogenous, IL-33 signaling via ST2, we evaluated developing B cell frequencies in WT:St2^−/− mixed bone marrow chimeric mice. St2^−/− pre-pro-B cells were recovered at a modestly, but statistically significantly, reduced frequency compared with WT pre-pro-B (Fig. 4G), possibly related to differences at early hematopoietic stem and progenitor cell stages in which ST2 is expressed (35, 36). Across the span of B cell development, WT and St2^−/− B cell frequencies normalized to a 1:1 ratio (Fig. 4G). In comparison, Il33^−/− B cells increased in frequency significantly more across the arc of IL-33 expression from the pre-pro-B cell to LPB stage compared with St2^−/− B cells within their respective bone marrow chimeras, consistent with a B cell–intrinsic role for IL-33 in regulating cellular fitness (Fig. 4H). Of note, the total number of developing B cells by subset in intact and naive adult WT, Il33^−/−, and St2^−/− mice was comparable, indicating that additional regulatory mechanisms restrict pathologic overexpansion of the B cell compartment in healthy adult Il33^−/− mice (data not shown).

To verify these results, we generated mice with a B cell–specific reduction of Il33. Il33^fl/fl mice were crossed with mice expressing Cre recombinase under the control of the endogenous CD19 promoter/enhancer (Il33^fl/fl CD19^Cre/+ mice, referred to as Il33^fl/fl CD19-Cre), leading to a 50–60% reduction in whole-Il33 transcript in sorted pro-B and LPB cell populations as assessed by quantitative PCR of both N-terminal and C-terminal sequences (Fig. 4I). Similar expression levels of Il33 were found in the lungs and skin of Il33^fl/fl and Il33^fl/fl CD19-Cre mice, confirming B cell lineage specificity of our Il33 knockdown (Fig. 4I). In WT:Il33^fl/fl CD19-Cre chimeric mice, Il33^fl/fl CD19-Cre B cells were significantly more fit than their WT counterparts as both a percentage of each B cell subset, and as measured by total B cell numbers starting at the LPB cell stage (Fig. 4J–L), supporting our WT:Il33^−/− chimera data using a separate, distinct mouse model. Of note, the delayed fitness advantage to the LPB cell stage in Il33^fl/fl CD19-Cre chimeras compared with Il33^−/− chimeras may reflect strain-specific differences (C57BL/6 versus BALB/c), partial (50–60% in Il33^fl/fl CD19-Cre) versus complete (Il33^−/−) abrogation of IL-33, or progressive Cre-mediated recombination in CD19-Cre mice from pro-B to LPB cells (37, 38). In contrast, an approximately even ratio of WT and Il33^fl/fl cells was detected at all stages of B cell development from pre-pro-B to immature B cells in WT: Il33^fl/fl chimeric mice, confirming that IL-33 attenuation is required for this fitness advantage (Fig. 4M). Viewed another way, from the pro-B to LPB stage, the frequency of Il33^fl/fl CD19-Cre B cells increased significantly more than the frequency of Il33^fl/fl controls within their respective chimeras (Fig. 4N).

We next assessed for differences in proliferation and cell death that might contribute to a difference in B cell frequency. After reconstitution, WT:Il33^−/− chimeras were subsequently treated with 1.25-Gy radiation to induce a hematopoietic burst (39, 40). In regard to proliferation, no difference in BrdU⁺ B cells was observed at any stage of B cell development (Fig. 4O). At the LPB cell stage, there was a decrease in the frequency of Il33^−/− dead cells (Annexin V⁺ 7-AAD⁺) as compared with WT dead cells, consistent with an overall increase in fitness of the Il33^−/− B cell lineage (Fig. 4P). No difference in the frequency of either proliferating or dead cells was observed at the pro-B cell stage, in which the dichotomy in fitness between WT and Il33^−/− B cells was initiated. This may reflect insufficient sensitivity of our present assays to capture a difference in proliferation or cell death in the pro-B stage. However, the reproducible difference in cell death captured at the LPB cell stage suggests that enhanced fitness in the Il33^−/− B cell lineage may, at least in part, be due to resistance to cell death. Collectively, these bone marrow chimeric experiments demonstrate that IL-33 is functioning as a negative regulator of fitness via a cell-intrinsic, ST2-independent mechanism during early B cell development.

Given that IL-33 expression spans the arc of IgH rearrangement, we considered whether IL-33 altered the diversity of the BCR repertoire. We assessed the IgH diversity by multiplex PCR and sequencing of FACS-sorted SPB cells. We chose to characterize the SPB cell stage, as it was immediately downstream of peak IL-33 expression and occurs prior to receptor editing, which takes place during the immature B cell stage and is a mechanism of central tolerance that eliminates autoreactive B cells leading to modification of the initially encoded BCR repertoire (41).

Epitope specificity is encoded, in part, by the CDR3 region of the BCR (42). Accordingly, we first assessed average CDR3 length between WT and Il33^−/− SPB cells. The overall CDR3 length distribution was consistent between WT and Il33^−/− SPB cells, with minor (<1%), but statistically significant, differences at lengths of 24, 48, 57, and 69 nt (Fig. 5A). The average CDR3 length assessed by Gaussian distribution analysis was not different between WT and Il33^−/− SPB cells (Fig. 5B). Further, BCR diversity is achieved by the somatic recombination of V, D, and J gene segments within the IgH locus of each developing B cell clone. Our sequencing approach allowed for assessment of IgH V-family usage, which was comparable between WT and Il33^−/− SPB cells (Fig. 5C). We next assessed the amino acid sequence composition of the recombined IgH. There was a small, but statistically significant, decrease in the fraction of IgH rearrangements in Il33^−/− SPB cells that encoded a productive IgH sequence (i.e., in-frame and without early-stop codons) (Fig. 5D).

To quantitate global diversity within the samples, we assessed for the following three parameters: richness, evenness, and clonality. Richness measures the number of unique IgH rearrangements within the sequencing pool and is displayed as a fraction of the total number of sequenced templates. There was no difference in richness between WT and Il33^−/− SPB cells (Fig. 5E). Evenness assesses for the distribution of the measured sequences within the sample. We measured evenness with Simpson calculation, in which values of 1 and 0 indicate maximally even or skewed distribution, respectively. There was no difference in evenness between WT and Il33^−/− SPB cells (Fig. 5F). Furthermore, values for evenness in WT and Il33^−/− SPB cells were closer to 1, indicating a comparatively even distribution of each sequence within the sample. Clonality is related to evenness and assesses for the dominance of single or a few clones within a sample. Highly monoclonal or oligoclonal samples have a clonality approximating 1, whereas infinitely polyclonal samples have a clonality of 0. There was no difference in productive clonality between WT and Il33^−/− SPB cells (Fig. 5G). Moreover, both WT and Il33^−/− SPB cells had productive clonality values approaching 0, indicating a lack of a single or few dominate clones. Collectively, these data demonstrate that IL-33 during early B cell development does not profoundly alter the BCR repertoire.

Human B cell development proceeds in a stepwise manner analogous to mice. We considered whether IL-33 was similarly expressed in early developing B cells in humans. Bone marrow was collected from healthy human donors and analyzed by intracellular flow cytometry for IL-33 expression (gating strategy in Supplemental Fig. 1C). Compared with an isotype control Ab, IL-33 expression was detected in all stages of B cell development and in mature B cells (CD19⁺ CD10⁻), which include recirculating B cells (Fig. 6A). High IL-33 expression (IL-33^hi) was clustered in pro-B (>40%) and LPB (>65%) cells, whereas less than 20% of SPB, 10% of immature B, and 10% of mature B cells were IL-33^hi (Fig. 6A, 6B). Moreover, the per-cell expression of IL-33 as measured by the MFI of IL-33 staining in pro-B and LPB cells was significantly higher than SPB, immature, and mature B cells (Fig. 6C). Collectively, these data demonstrate that IL-33 is expressed during human B cell development, peaking at the pro-B and LPB stages, paralleling what we observed in mice.

Given that reduced or absent IL-33 expression increased developing B cell fitness in mice and that IL-33 expression was low, but detectable, in mature recirculating B cells in human bone marrow, we considered whether IL-33 expression was reduced in human B cell malignancies wherein enhanced cell fitness is a defining disease characteristic. First, IL33 mRNA was detectable in peripheral B cells from healthy donors and patients with B cell CLL (B-CLL). Compared with peripheral blood CD19⁺ B cells from healthy donors, expression of IL33 mRNA was significantly decreased in peripheral blood tumor cells from patients with B-CLL (Fig. 6D). In contrast, IL33 mRNA expression was increased in biopsies of all B cell lymphoma types that we evaluated, including follicular lymphoma, DLBCL, Burkitt lymphoma, mantle cell lymphoma, marginal zone lymphoma, and Hodgkin lymphoma (Supplemental Fig. 3). Collectively, these data suggest that IL-33 downregulation is favored in B-CLL, possibly secondary to an adaptive fitness advantage, but not B cell lymphomas.

In this study, our data establish an in vivo role of IL-33 during early B cell development that is both cell intrinsic and ST2 independent. IL-33 is a two-domain protein, with the C-terminal domain primarily involved in binding to ST2 on immune cells following release of IL-33 from epithelial and endothelial cells. The N-terminal domain has a nuclear localization sequence, a putative homeodomain-like, helix-turn-helix motif, and a chromatin-binding motif that interacts with histones H2A and H2B, suggesting a potential cell-intrinsic role for IL-33 (9, 14, 15). Early evidence supported a gene regulatory role for IL-33, with an IL-33/Gal4/DNA-binding domain fusion protein potently repressing GAL4-responsive luciferase expression in an in vitro gene reporter assay (14, 15). However, if such a gene regulatory role exists, endogenous targets have largely remained elusive to date. Identification of targets has focused on endothelial and epithelial cells given the high level of expression of IL-33 in these lineages. Knockdown of IL-33 in primary HUVECs via multiple small interfering RNAs did not alter the global proteome (21). Moreover, inducible overexpression of IL-33 in a human esophageal epithelial cell line lacking endogenous IL-33 and ST2 expression did not alter the transcriptome of these cells (20). Whereas these global, unbiased approaches have failed to identify a nuclear role for IL-33 in regulating gene expression, multiple smaller in vitro studies have posited targets including IL6, IL13, CCL5, and RELA (NF-κB p65) based on gene expression differences with overexpression or knockdown of IL-33 complemented by chromatin immunoprecipitation of IL-33 (16–19). Although intriguing, these studies were all conducted in vitro in cell lines with variable potential for endogenous ST2 expression, and whether such interactions exist in vivo and have any phenotypic or mechanistic significance remained to be determined.

Consistent with a unique cell-intrinsic role for IL-33, the magnitude of Il33 expression in developing B cells was 40–100-fold lower than in stromal cells and, instead, at the low end of expression of well-established, B cell–intrinsic regulators. Our present work does not definitively establish a nuclear versus cytoplasmic role for IL-33. However, expression of IL-33 was largely restricted to the nucleus in pro-B and LPB cells, indicating a high likelihood that the IL-33 is acting in the nucleus. Moreover, we determined a cell-intrinsic role of IL-33 in the context of cellular and environmental stress using competitive reconstitutions in mixed bone marrow chimeric mice. Prior studies on endothelial and epithelial cells have evaluated the role of IL-33 under optimal growth conditions or during catastrophic cellular insult (20, 21). Further investigations will be needed to clarify whether IL-33 functions in a cell-intrinsic manner in other cell types, including stromal cells, in the context of cellular and environmental stress. Precedent exists within the IL-1 family for cell-intrinsic functions associated with cellular fitness, with the N-terminal domain of IL-1α conferring nuclear translocation and promoting apoptosis in the context of malignancy but not in nontransformed cells (43, 44).

Increasing evidence suggests that IL-33 sequestration in the nuclei of stromal cells may serve a protective function to restrict autoinflammation (3). Transgenic mice expressing an N-terminal deletion of Il33 under the control of the endogenous Il33 promoter accumulated high serum levels of IL-33 and developed fatal autoinflammation in the first 6 mo of life via an ST2-dependent mechanism (45). Similarly, skin overexpression of truncated IL-33 lacking the N-terminal domain, but not full-length IL-33 induced epidermal and dermal thickening, inflammation, and hyperkeratosis (46). Recent kinetic analysis of IL-33 release demonstrated that chromatin binding promotes a slower, phased extracellular release of IL-33 during necrosis, serving as an additional posttranslational mechanism of IL-33 regulation (20). These collective data provide evidence that, particularly among high IL-33–expressing cells, the nuclear localization of IL-33 may serve a host protective function. Our data suggest that that nuclear IL-33 may serve an additional function, depending on the cell type (i.e., B cells versus stromal cells) or the environmental conditions (i.e., stress versus homeostasis). These multiple roles of IL-33 support the high degree of sequence conservation across species (14) and may reconcile data regarding the potential of IL-33 to regulate gene expression (14, 15) versus the lack of identifiable targets or cell-intrinsic, IL-33–dependent phenotypes in high–IL-33–expressing cells (20, 21).

Signals for both increased proliferation (E2F targets, cell cycle, and DNA replication) and decreased cell death (p53 pathway) in Il33^−/− pro-B and LPB cells were present in our transcriptomic analysis. Using in vivo competitive reconstitution experiments with radiation-stimulated hematopoiesis, we consistently identified fewer dead cells among LPB cells of Il33^−/− origin and no difference in BrdU incorporation at any B cell developmental stage. One interpretation of these data are that the increase in fitness of Il33^−/− B cells is most substantially attributable to a decrease in cell death. Alternatively, the sensitivity or timing of our assays to measure proliferation and cell death may be insufficient to detect differences. Further work will be needed to assess the relative balance of proliferation and apoptosis in the context of IL-33 expression or deficiency.

Given the effect of IL-33 on modulating fitness during B cell development, we considered whether IL-33 expression levels are increased or decreased in B cell malignancies. Interestingly, B-CLL samples were associated with a decreased expression of IL33 mRNA compared with healthy controls. Although B-CLL is not considered to originate from pro-B or LPB cells, cancer cells frequently employ diverse prosurvival strategies, including those classically used during B cell development regardless of their initiating cell stage. Our data suggest this reduction in IL-33 expression in B-CLL samples from baseline may promote increased leukemic cell fitness. In contrast, all tested B cell lymphomas displayed significantly increased expression of IL33 mRNA. This dichotomy may be related to the differences in tumor microenvironment between B cells that form tumors (i.e., lymphomas) and those that do not (i.e., CLL) and the diverse roles of IL-33, including in the activation of T regulatory cells enhancing immunotolerance to the tumor. Such IL-33–induced immunotolerance has been observed in breast, colorectal, and certain lung cancers (47). Most notably, another group has similarly shown selection for increased IL33 expression in DLBCL, which, in mice, promoted T regulatory cell activation (48). Additionally, whereas a large proportion of B-CLL samples (79%) had a decrease in IL33 expression, there were samples that displayed an increase in IL33 mRNA compared with healthy controls, suggesting that additional factors, including tumor genetics and the tumor microenvironment, may influence the relative benefit of modulated IL33 expression. Collectively, the correlation in B-CLL with downregulated IL33 expression, unique among the tested B cell malignancies, suggests beneficial adaptation at the IL33 locus derived from mechanism(s) wherein decreased IL33 expression promotes CLL cell survival.

Although our data establish an intracellular role for IL-33 during early B cell development, it remains unclear whether this is due to the hypothesized role of IL-33 as a transcriptional regulator bound either directly to DNA or as part of a transcriptional regulatory complex, or whether IL-33 is acting via an alternative mechanism. Moreover, peak IL-33 expression in pro-B and LPB cells is well conserved across mouse strains and in humans, but the effectors that govern IL-33 expression during early B cell development remain unidentified. Further investigation will be needed to clarify these factors. Collectively, our work defines IL-33 as a B cell–intrinsic, ST2-independent modulator of fitness during in vivo B cell development.

We appreciate the critical insights of Rachel Brown on this manuscript. We are thankful to Dr. Paul Bryce, who provided us with Il33^fl/fl-GFP mice prior to deposition in The Jackson Laboratory repository, and to Dr. Andrew McKenzie, who provided us with Il33^cit/cit and St2^−/− mice. We appreciate the assistance of Kevin Weller, David Flaherty, Brittany Matlock, and Chris Warren in the Vanderbilt University Medical Center Flow Cytometry Shared Resource.

The authors have no financial conflicts of interest.