Cre-mediated excision of Il1r2 exon 3 results in loss of IL-1R2 protein. IL-1R2 is the first identified example of a decoy receptor and is proposed to reduce inflammation by inhibiting IL-1 signaling. However, proof of mechanism of action in vivo is lacking. Thus, we generated mice deficient in IL-1R2 by CMV-Cre–mediated removal of exon 3 from Il1r2 (Figure 1A), followed by selective crossing to remove CMV-Cre. These mice lacked exon 3 in the genome by PCR (Figure 1B), and sequencing confirmed the correct recombination that left only a LoxP site in the genome (data not shown). Notably, ELISA showed an approximately 50% reduction in soluble IL-1R2 in the sera of Il1r2+/– mice compared with that of Il1r2+/+ (WT) mice and no detectable serum IL-1R2 in Il1r2–/– knockout (R2–/–) mice (Figure 1C). In addition, flow cytometry of whole blood showed a measurable loss of IL-1R2 on neutrophils from R2–/– mice compared with those from WT mice (Figure 1D), but no difference on monocytes/macrophages (Figure 1E), which express lower levels of surface IL-1R2. R2–/– mice were viable and displayed no gross morphological or phenotypic difference (data not shown). These data show that deletion of Il1r2 exon 3 results in loss of IL-1R2 protein.

Cre-mediated excision of Il1r2 exon 3 results in loss of IL-1R2 protein. (A) Targeting strategy for generation of Il1r2 floxed mice. SHA, short homology arm; LHA, long homology arm. The asterisks indicate the position of genotyping primers. (B) Separated PCR products showing CMV-Cre–mediated removal of approximately 1.4 kb of Il1r2 containing exon 3. Values shown are in base pairs (BP). (C) ELISA data for serum IL-1R2 in female mice with IL-1R2 genotypes as indicated. (D and E) Flow cytometric analysis for IL-1R2 and lineage markers on whole blood from IL-1R2+/+ (WT) and IL-1R2–/– (R2–/–) mice, showing IL-1R2 level on the surface of neutrophils (D) and monocyte/macrophages (E). Data are shown as the mean ± SEM of n = 6 (C) and are representative of n = 4 (D and E). ****P ≤ 0.0001, t test.

IL-1R2–deficient mice show a prolonged inflammatory response due to uncleared IL-1. To investigate whether deletion of Il1r2 affects IL-1–driven processes in vivo, we intravenously injected IL-1α and collected sera over time to measure induced cytokine level. No significant difference in the level of IL-6 between groups was found after 2 hours (Figure 2A). However, after 4 hours sera IL-6 levels were higher in R2–/– mice and nearly undetectable in WT mice (Figure 2A). Indeed, comparing the ratio of IL-6 at 4 hours to that at 2 hours revealed a substantial prolonging of the inflammatory reaction in R2–/– mice (Figure 2B), suggesting IL-1–driven inflammation does not resolve as swiftly without IL-1R2. To test if this was due to a lack of IL-1 scavenging by the high concentration of IL-1R2 in serum (16), we directly measured the serum level of the injected human IL-1α over time. This showed markedly increased levels of IL-1α at both time points in R2–/– mice compared with those in WT mice (Figure 2C), supporting reduced clearance of IL-1α from the circulation. Using a sterile peritonitis model, we examined the effect of IL-1R2 loss without the high level of IL-1R2 in the circulation, which showed increased neutrophil recruitment in response to IL-1α in R2–/– mice (Figure 2D). Together, these data show that IL-1R2 limits inflammation and scavenges free IL-1 and, thus, that R2–/– mice have increased IL-1 signaling.

IL-1R2–deficient mice show a prolonged inflammatory response due to uncleared IL-1. (A) ELISA data for serum IL-6 at the indicated times after intravenous injection of IL-1α into IL-1R2+/+ (WT) and IL-1R2–/– (R2–/–) mice. (B) Ratio of serum IL-6 at 4 hours to that at 2 hours after intravenous injection of IL-1α. (C) ELISA data for serum IL-1α at the indicated times after intravenous injection of IL-1α. (D) Flow cytometry for the number of GR1+ cells recruited into the peritoneum 6 hours after i.p. injection of IL-1α. Data are shown as the mean ± SEM of n = 4 (A–C) and n ≥ 6 (D). *P ≤ 0.05, **P ≤ 0.02, ***P ≤ 0.005, t test.

T follicular regulatory cell–derived IL-1R2 restrains GCs and antibody production. As loss of IL-1R2 can increase availability of IL-1 (Figure 2), we examined processes known to be IL-1 dependent. GC reactions are driven by interaction between T follicular helper (Tfh) cells and B cells, with T follicular regulatory (Tfr) cells providing negative regulation (17). Tfr cells are suggested to dampen GC reactions by producing IL-1R2 that antagonizes IL-1 activation of Tfh cells (13), but if this occurs in vivo and any consequence of interfering with this regulation is unknown. We produced Il1r2fl/fl/Foxp3CreERT2+ mice to enable conditional deletion of IL-1R2 in Tfr cells. As Tfr cells expressed the highest level of IL-1R2 (Figure 3, A–C, and Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.174005DS1), while Tregs expressed IL-1R2 at a lower frequency and level (Figure 3, A–C, and Supplemental Figure 1) (in keeping with previous reports; ref. 13) and do not accumulate in the GC, this supports that the main target of Foxp3-directed IL-1R2 deletion in the GC is the Tfr cell. Without tamoxifen, splenocyte cDNA only generated the Il1r2fl/fl amplicon, but tamoxifen via diet lead to generation of the Il1r2–/– amplicon and a loss of the Il1r2fl/fl band (Supplemental Figure 2). Importantly, MACS sorting for Tregs/Tfr cells resulted in near-complete loss of the Il1r2fl/fl amplicon and generation of the Il1r2–/– band (Supplemental Figure 2). In addition, approximately 20% of Tfol cells (CD4/CXCR5/PD-1+) expressed IL-1R2, and tamoxifen treatment resulted in a large decrease in this signal (Figure 3D). Together, Tfr cells are the major Tfol cell expressing IL-1R2, and tamoxifen induces Foxp3-directed deletion of IL-1R2 in Tfr cells. Challenging this model with the classic sheep red blood cell (sRBC) immunization resulted in more Tfol cells (Figure 3E) but no change in the ratio of Tfh to Tfr cells (Figure 3F), expanded GCs (Figure 3G), as evidenced by increased GL7/B220+ B cells, and much higher titers of serum anti-sRBC IgG (Figure 3H) in mice lacking IL-1R2 in Tfr cells. GCs showed normal gross morphology between groups, with no increase in follicle size (Supplemental Figure 3A), but more GL7+ area per follicle was seen with Tfr IL-1R2 loss (Supplemental Figure 3B), in keeping with the flow cytometry data. In addition, Il1r2fl/fl/Foxp3CreERT2+ mice showed increased numbers of CD4+ T cells (Figure 3I) and altered ratios of naive and effector/central memory CD4+ and CD8+ T cells (Figure 3J) in the spleens; splenocytes stimulated ex vivo with PMA/ionomycin showed increased IFN-γ+ CD4+ and CD8+ T cells (Figure 3K), compared with mice lacking Foxp3CreERT2. However, splenic Treg level was not altered with IL-1R2 deletion (Figure 3L). Near identical results were found when OVA/Alum was used for immunization (Supplemental Figure 4), suggesting a general effect of IL-1R2 loss on GC responses. Most importantly, administering Anakinra (IL-1RA) to block both IL-1α and IL-1β reversed the increased responses seen in Il1r2fl/fl/Foxp3CreERT2+ mice (Figure 4), showing that the action of IL-1R2 loss is mediated via IL-1. All together, this shows that deficiency of IL-1R2 in Tfr cells augments GC reactions and antibody output after immunization and that this occurs due to increased IL-1 signaling.

Loss of IL-1R2 in Tfr cells increases the GC response after immunization. Flow cytometry for IL-1R2 on WT splenic T follicular helper (Tfh) cells, T follicular regulatory (Tfr) cells, Tregs, and CD4+ or CD8+ T cells, showing the percentage of population positive for IL-1R2 (A) and median fluorescent intensity (MFI) of the whole population (B) and only those positive for IL-1R2 (C). Presented values for WT mice have percentage and MFI seen in IL-1R2–/– mice subtracted. (D–L) Il1r2fl/fl ± Foxp3-Cre-ERT2 (Cre) littermate mice were all tamoxifen treated and immunized with sheep red blood cells (sRBC); spleens were immunophenotyped 8 days later. (D–G) Flow cytometry for IL-1R2 on splenic T follicular (Tfol) cells (D), Tfol cells (E), the ratio of Tfh cells to Tfr cells (F), and germinal center (GC) B cells (G) in the genotypes indicated. (H) Flow cytometry for binding of serum anti-sRBC IgG antibodies to sRBCs. (I and J) Flow cytometry for the splenic CD4+/CD8+ T cell ratio (I) and CD4+/CD8+ T cell subtype (J) in the genotypes indicated. TCM, central memory; TEM, effector memory. (K) Intracellular cytokine staining in splenic CD4+/CD8+ T cells activated with PMA/ionomycin. (L) Flow cytometry for splenic Tregs. Data are shown as the mean ± SEM; n = 3 individual mice (A–C) or n = 5 individual -Cre and 4 individual +Cre mice (D–L) representative of ≥3 repeats. ***P ≤ 0.001, ****P ≤ 0.0001, t test and ANOVA.

Anakinra blocks the heightened GC response in mice lacking Tfr IL-1R2. Il1r2fl/fl/Foxp3-Cre-ERT2 littermate mice were all tamoxifen treated, immunized with sheep red blood cells (sRBC), along with concurrent PBS or Anakinra (Ana) treatment until spleen immunophenotyping 8 days later. (A–C) Flow cytometry for IL-1R2 on splenic T follicular (Tfol) cells (A), Tfol cells (B), and germinal center (GC) B cells (C) in the genotypes indicated. (D) Flow cytometry for binding of serum anti-sRBC IgG antibodies to sRBCs. (E and F) Flow cytometry for splenic CD4+/CD8+ T cell ratio (E) and CD4+/CD8+ T cell subtype (F) in the genotypes indicated. TCM, central memory; TEM, effector memory. (G) Intracellular cytokine staining in splenic CD4+/CD8+ T cells activated with PMA/ionomycin. (H) Flow cytometry for splenic Tregs. Data are shown as the mean ± SEM; n = 3 individual mice (PBS) or n = 4 individual mice (+Ana). *P ≤ 0.05, **P ≤ 0.01, t test and ANOVA.

Tfr IL-1R2 deficiency does not increase GC reactions after a booster immunization. As IL-1R1 signaling can amplify GC responses, we investigated if loss of IL-1R2 on Tfr cells would further heighten GC reactions and antibody production after a second booster immunization of sRBCs. Interestingly, Il1r2fl/fl/Foxp3CreERT2+ mice still showed increased Tfol cells and CD4+ T cells, altered ratios of CD4+/CD8+ subtypes, and increased IFN-γ expression in CD4+/CD8+ T cells (Supplemental Figure 5, A–G), as seen before (Figure 3, D–L). However, the titer of serum anti-sRBC IgG was no different between Il1r2fl/fl/Foxp3CreERT2+ and Il1r2fl/fl mice (Supplemental Figure 5H), with levels in Il1r2fl/fl/Foxp3CreERT2+ mice no higher than those seen after a single immunization (Figure 3H) but anti-sRBC IgG levels in Il1r2fl/fl mice boosted to the same level as those witnessed in Il1r2fl/fl/Foxp3CreERT2+ mice (Supplemental Figure 5H). Interestingly, IL-1R2 on Tfol cells in boosted Il1r2fl/fl mice dropped to an equivalent level (~10%) (Supplemental Figure 5A) as that seen in mice lacking Tfr IL-1R2 (Figure 3D and Figure 4A), suggesting that a second immunization may intrinsically lower GC IL-1R2. Together, without IL-1R2 on Tfr cells, a higher GC output can be reached after a single immunization, while a subsequent booster generates similar responses with IL-1R2 present.

IL-1R1 signaling is not essential for induction of the GC response after immunization. As the major action of IL-1R2 is to bind IL-1, increased GC responses with IL-1R2–deficient Tfr cells suggest that IL-1R1 signaling may normally be required for GC reactions. To investigate this, we immunized littermate Il1r1+/+ and Il1r1–/– mice with sRBC and again performed immunophenotyping of the spleens. Surprisingly, loss of IL-1R1 had very little effect on the GCs, with no difference in the number of Tfol cells (Figure 5A), Tfol IL-1R2 expression (Figure 5B), titers of serum anti-sRBC IgG (Figure 5C), T cell numbers (Figure 5D), subtype (Figure 5E), Tregs (Figure 5F), or activation status (Figure 5G) and only a marginal decrease in the number of GC B cells (Figure 5H). This shows that IL-1 is not obligatory for a productive GC reaction and suggests that IL-1R2 normally suppresses IL-1 in the GC, with the enhancing effects of IL-1 only apparent when IL-1R2 level is compromised.

IL-1R1 signaling is not essential for induction of the GC response after immunization. Il1r1–/– andIl1r1+/+ (WT) littermate mice were immunized with sheep red blood cells (sRBC) and spleens were immunophenotyped 8 days later. (A and B) Flow cytometry for IL-1R2 on splenic T follicular (Tfol) cells (A) and Tfol cells (B) in the genotypes indicated. (C) Flow cytometry for binding of serum anti-sRBC IgG antibodies to sRBCs. (D–F) Flow cytometry for splenic CD4+/CD8+ T cell ratio (D), CD4+/CD8+ T cell subtype (E), and Tregs (F) in the genotypes indicated. TCM, central memory; TEM, effector memory. (G) Intracellular cytokine staining in splenic CD4+/CD8+ T cells activated with PMA/ionomycin. (H) Flow cytometry for splenic germinal center (GC) B cells (H). n = 3 individual WT mice or n = 6 individual Il1r1–/– mice (A) and n = 6 individual WT mice or n = 9 individual Il1r1–/– mice (B–H). *P ≤ 0.05, t test and ANOVA.

GC reactions are only increased with temporal loss of IL-1R2 in adulthood. As global IL-1R2–/– mice show altered IL-1–driven responses (Figure 2) and total loss of the high levels of soluble IL-1R2 in sera (Figure 1C), we tested if they also showed different GC dynamics. Unexpectedly, IL-1R2–/– mice only had a weak phenotype, comprising increased CD4+ T cell numbers, altered CD8 subtype ratio, and increased IFN-γ+ CD8+ T cells (Supplemental Figure 6, A–C). However, IL-1R2–/– mice showed no changes to Treg, Tfol cell, or GC B cell number or anti-sRBC IgG titers (Supplemental Figure 6, D–G), compared with littermate IL-1R2+/+ mice. We reasoned that the anomaly between Tfr-specific and global IL-1R2 deletion could be caused by IL-1R2 somehow having a “dual function,” such as opposing effects within the GC versus systemic, or more likely that it was caused by the temporal deletion of IL-1R2 via Cre/Lox in adulthood that was important. To test the second possibility, we generated Il1r2fl/fl/Rosa26-CreERT2 mice to allow conditional IL-1R2 deletion upon tamoxifen treatment. Fascinatingly, globally deleting IL-1R2 in adulthood gave a near-identical phenotype (Figure 6) after sRBC immunization to that seen with IL-1R2 deletion in Tfr cells (Figure 3, D–L), with the only tangible difference being more IL-1R2 deletion in the Tfol cells (Figure 6A), likely due to deletion of IL-1R2 on Tfh and Tfr cells and a more efficient Cre. All together, this suggests a complicated system in which IL-1R2 normally restrains GC reactions but that absence of IL-1R2 from conception leads to some form of compensation that mitigates against its absence in adulthood. Indeed, similar functional compensation has been seen in other gene families (18), but the mechanism remains poorly understood.

Conditional global loss of IL-1R2 in adulthood increases the GC response after immunization. Il1r2fl/fl ± Rosa26-Cre-ERT2 (Cre) littermate mice were all tamoxifen treated and immunized with sheep red blood cells (sRBC); spleens were immunophenotyped 8 days later. (A–C) Flow cytometry for IL-1R2 on splenic T follicular (Tfol) cells (A), Tfol cells (B), and germinal center (GC) B cells (C) in the genotypes indicated. (D) Flow cytometry for binding of serum anti-sRBC IgG antibodies to sRBCs. (E and F) Flow cytometry for splenic CD4+/CD8+ T cell ratio (E) and CD4+/CD8+ T cell subtype (F) in the genotypes indicated. TCM, central memory; TEM, effector memory. (G) Intracellular cytokine staining in splenic CD4+/CD8+ T cells activated with PMA/ionomycin. (H) Flow cytometry for splenic Tregs. Data are shown as the mean ± SEM; n = 6 individual mice (both -Cre and +Cre). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001, t test and ANOVA.

Individuals with inflammatory bowel disease have lower serum IL-1R2. If IL-1R2 can control GC reactions and/or alter CD4+/CD8+ T cell function, lower levels of IL-1R2 could lead to autoimmune and/or inflammatory disease in humans. To initially investigate this, we collected samples from patients with UC, Crohn’s disease, and antineutrophil cytoplasmic antibody–associated (ANCA-associated) vasculitis (AAV) positive for myeloperoxidase (AAV-MPO) or proteinase 3 (AAV-PR3) — diseases with well-documented autoimmune and autoinflammatory components. Testing serum IL-1R2 level revealed substantially decreased IL-1R2 in patients with UC and Crohn’s disease but no difference in patients with either form of AAV (Figure 7A), compared with individuals acting as healthy controls. Lower serum IL-1R2 was witnessed in both male (Figure 7B) and female (Figure 7C) patients with UC and Crohn’s disease. Although there was a statistically significant age difference between patients with Crohn’s disease and patients with AAV-MPO and individuals acting as healthy controls (Figure 7D), sera IL-1R2 levels did not correlate to age in either healthy control (Figure 7E) or Crohn’s disease (Figure 7F) populations, suggesting that age was not a confounding factor for IL-1R2 level. Interestingly, GWAS have identified common IL1R2 variants as a key risk factor for UC (19) and Crohn’s disease (20). All together, these initial data reveal that lower levels of serum IL-1R2 are associated with human disease.

Individuals with inflammatory bowel disease have lower serum IL-1R2. (A–C) ELISA data for serum IL-1R2 levels in humans with ulcerative colitis (UC), Crohn’s disease (Crohns), ANCA-associated vasculitis MPO+ (AAV-MPO), and AAV proteinase 3+ (AAV-PR3) or individuals acting as healthy controls (Cont) displayed for both male and female patients (A), and male (B) or female (C) patients only. (D) Age of individuals with conditions as indicated. (E and F) Scatter plot of serum IL-1R2 versus age for individuals acting as healthy controls (E) and patients with Crohn’s disease (F) showing no correlation. Data are shown as the median ± SEM of n = 40 (Cont), 20 (UC), 19 (Crohns, AAV-MPO), 18 AAV-PR3). P values are stated in the figure (ANOVA and linear regression).