IRAK-4- and MyD88-dependent pathways are essential for the removal of developing autoreactive B cells in humans

Abstract

Most autoreactive B cells are normally counterselected during early B cell development. To determine whether Toll-like receptors (TLRs) regulate the removal of autoreactive B lymphocytes, we tested the reactivity of recombinant antibodies from single B cells isolated from patients deficient for interleukin-1 receptor-associated kinase 4 (IRAK-4), myeloid differentiation factor 88 (MyD88), and UNC-93B. Indeed, all TLRs except TLR3 require IRAK-4 and MyD88 to signal, and UNC-93B-deficient cells are unresponsive to TLR3, TLR7, TLR8, and TLR9. All patients suffered from defective central and peripheral B cell tolerance checkpoints, resulting in the accumulation of large numbers of autoreactive mature naive B cells in their blood. Hence, TLR7, TLR8, and TLR9 may prevent the recruitment of developing autoreactive B cells in healthy donors. Paradoxically, IRAK-4-, MyD88-, and UNC-93B-deficient patients did not display autoreactive antibodies in their serum or develop autoimmune diseases, suggesting that IRAK-4, MyD88, and UNC-93B pathway blockade may thwart autoimmunity in humans.

Figure 1. Human B cell activation by TLR7 and TLR9 requires UNC-93B

(A) Human new emigrant and mature naive B cells do not express IL-1R family members. Histograms represent means ± s.e.m. of the expression of studied genes (IL1R1, IL18R1, IL1RL1, TLR1, TLR2, TLR3, TLR4, TLR5, TLR6, TLR7, TLR8, TLR9, TLR10, IRAK4, MYD88, and UNC93B1) in human new emigrant (black bars, n=8) or mature naive (grey bars, n=4) B cells assessed by gene expression profiling using the Affimetrix Human Genome U133 Plus 2.0 Array. (B) UNC93B1 gene is expressed in human DCs and to a lesser extent in B cells and not in T cells. IL1R, TLR7, TLR9, TLR10, IRAK4, MYD88, UNC93B1, and CD79B gene expression was analyzed by RT-PCR in CD19⁺CD10⁺IgM⁺CD27⁻ new emigrant, CD19⁺CD10⁻IgM⁺CD27⁻ mature naïve, CD19⁺CD10⁻CD27⁺ total memory B cells; TCRαβ⁺CD4⁺CD25⁻ T cells; CD19⁻CD11c⁺HLA-DR⁺ myeloid DC enriched cells; and CD19⁻CD11c⁻HLA-DR⁺ plasmacytoid DC enriched cells from healthy controls. (C) Human B cells responses to TLR7 and TLR9 agonists depend on UNC-93B expression. Dot plots show CD80 and CD86 expression on naive CD19⁺CD27⁻ B cells from a healthy control, two UNC-93B-deficient patients and one IRAK-4-deficient patient that were either left unstimulated (NS) or stimulated with F(ab’)₂ anti-IgM, CpG (TLR9 agonist) or Loxoribine (TLR7 agonist) for 48 hours.

Figure 2. Defective central B cell tolerance checkpoint in IRAK-4-, MyD88- and UNC-93B-deficient patients

(A) Antibodies from new emigrant B cells from controls and patients were tested by ELISA for reactivity against single-stranded DNA (ssDNA), double-stranded DNA (dsDNA), insulin and lipopolysaccharide (LPS). Antibodies are considered polyreactive when they recognize at least 2 and usually all 4 antigens. Dotted lines show ED38-positive control (Wardemann et al., 2003). Horizontal lines show cut-off OD₄₀₅ for positive reactivity. For each individual, the frequency of polyreactive and non polyreactive clones is summarized in pie charts with the number of antibodies tested indicated in the centers. The frequencies of polyreactive (B) and HEp-2-reactive (C) new emigrant B cells are compared between controls and patients, and statistically significant differences are indicated. Each diamond represents an individual, and the average is shown with a bar.

Figure 3. ANA expressing B cell removal requires IRAK-4 and MyD88 but not UNC-93B

(A) The frequency of anti-nuclear new emigrant B cells in IRAK-4 and MyD88-deficient patients is higher than in controls and UNC-93B-deficient patients. Each diamond represents an individual, and the average is shown with a bar. (B) The frequencies of chromatin non-reactive and reactive clones in IRAK-4- and MyD88-deficient patients are higher than in controls and UNC-93B-deficient patients. Histograms represent means ± s.e.m. of frequencies of chromatin non-reactive (open bars) or chromatin reactive (black bars) antibodies. Statistically significant differences between patients and controls are indicated. (C) ANAs from IRAK-4- and MyD88-deficient new emigrant B cells show various patterns of anti-nuclear staining. White arrows indicate dividing cells that demonstrate negative (top row) or positive (bottom row) staining for the condensed chromatin material. (D) Increased anti-kinetoplast antibodies in new emigrant B cells from IRAK-4- (black bars) and MyD88-deficient (dark grey bars) compared to controls (open bars) and UNC-93B-deficient patients (light grey bars). Histograms represent means ± s.e.m. of proportion of kinetoplast reactive clones. Statistically significant differences between patients and controls are indicated. (E) IRAK-4- and MyD88-deficient clones express ANAs that can recognize the kinetoplast of C. Luciliae. White and red arrows indicate positively stained kinetoplasts and nucleus of C. Luciliae respectively. (F) Kinetoplast reactive ANA-expressing B cells display the highest frequency of positively charged residues in their heavy chain CDR3s. New emigrant B cells from IRAK-4 and MyD88-deficient patients were grouped into non-autoreactive, autoreactive-non ANA (HEp-2-reactive and/or polyreactive), kinetoplast non-reactive ANA and kinetoplast reactive ANA clones. Pie charts show the proportion of heavy chain CDR3s with 0, 1, 2 and 3 or more positive charges. Absolute numbers of clones analyzed in each B cell fraction is indicated in the center. The average number of positively charged residues per CDR3 is indicated below each fraction. The numbers of positively charged residues in immunoglobulin heavy chain CDR3s were calculated excluding the arginin or lysin found at position 94 in most germ line VHs.

Figure 4. Defective receptor editing induction in IRAK-4- and MyD88-deficient developing autoreactive B cells

(A) New emigrant B cells from IRAK-4- and MyD88-deficient patients have an increased proportion of autoreactive lambda clones compared to controls and UNC-93B-deficient patients. Histograms represent means ± s.e.m. of proportion of κ (black bars) and λ (open bars) clones that are autoreactive (polyreactive and HEp-2-reactive combined). (B) Autoreactive IRAK-4- and MyD88-deficient new emigrant B cells display a lambda repertoire suggesting defective receptor editing. Pie charts show the proportion of the different Jλ genes in non-autoreactive and autoreactive new emigrant B cells from healthy controls (top panel) or IRAK-4- and MyD88-deficient patients (bottom panel), with the number of analyzed sequences indicated in the centers. (C) Most ANA and all kinetoplast-reactive B cells from IRAK-4 and MyD88-deficient patients express kappa chains. Histograms represent means ± s.e.m. of proportion of κ (black bars) and λ (open bars) clones that are anti-nuclear (top panel) and anti-kinetoplast (bottom panel). (D) The ANA expressed by IRAK-4- and MyD88-deficient new emigrant B cells do not show any bias towards downstream Jκ3,4,5 usage compared to the non-ANA clones. Pie charts show the proportion of the different Jκ genes, with the number of analyzed sequences indicated in the centers. (E) The ANA expressed in IRAK-4-and MyD88-deficient new emigrant B cells do not show any bias towards upstream Vκ gene usage compared to the non-ANA clones. The Vκ locus is shown below clustered into 4 clans of Vκ gene segments. Histograms represent the proportion of genes of each Vκ group for ANA (black bars) and non-ANA (open bars) expressed by new emigrant B cells from IRAK-4- and MyD88-deficient patients.

Figure 5. Defective peripheral B cell tolerance checkpoint in IRAK-4-, MyD88- and UNC-93B-deficient patients

(A) Increased frequency of HEp-2-reactive antibodies in IRAK-4-, MyD88-, and UNC-93B-deficient mature naïve B cells. Antibodies from mature naïve B cells from one control (HD10), three IRAK-4-deficient patients, one MyD88-deficient, and two UNC-93B-deficient patients were tested in ELISA for reactivity with HEp-2 cell lysate. Dotted lines show ED38-positive control (Meffre et al., 2004; Wardemann et al., 2003). Horizontal lines show cut-off OD₄₀₅ for positive reactivity. For each individual, the frequency of HEp-2-reactive (in black) and non HEp-2-reactive (in white) clones is summarized in pie charts with the number of antibodies tested indicated in the centers. The frequency of (B) HEp-2-reactive and (C) polyreactive (tested against single-stranded DNA, double-stranded DNA, insulin and lipopolysaccharide) clones in mature naive B cells of IRAK-4-, MyD88- and UNC-93B-deficient patients is higher than in controls. Each diamond represents an individual, and the average is shown with a bar. Statistically significant differences between patients and controls are indicated.

Figure 6. Identification of ANA-expressing mature naïve B cells in IRAK-4-, MyD88-, and UNC-93B-deficient patients

(A) The frequency of anti-nuclear mature naive B cells in IRAK-4, MyD88-, and UNC-93B-deficient patients is higher than in controls. Each diamond represents an individual, and the average is shown with a bar. (B) Chromatin non-reactive and reactive and (C) kinetoplast reactive clones are enriched in the mature naïve B cell compartment of IRAK-4-, MyD88- and UNC-93B-deficient patients compared to controls. Histograms represent means ± s.e.m. of proportion of (B) chromatin non-reactive (open bars) and chromatin reactive (black bars) clones and (C) kinetoplast reactive clones.

Figure 7. Lack of serum ANAs in IRAK-4-, MyD88-, and UNC-93B-deficient patients

The presence of IgG ANAs was assessed by ELISA in the serum of healthy controls, IRAK-4-, MyD88-, and UNC-93B-deficient patients. Each diamond represents an individual. Dashed lines indicate threshold between negative, low positive and positive individuals for serum IgG ANAs.