Broad-spectrum antibodies against self-antigens and cytokines in RAG deficiency

Abstract

Patients with mutations of the recombination-activating genes (RAG) present with diverse clinical phenotypes, including severe combined immune deficiency (SCID), autoimmunity, and inflammation. However, the incidence and extent of immune dysregulation in RAG-dependent immunodeficiency have not been studied in detail. Here, we have demonstrated that patients with hypomorphic RAG mutations, especially those with delayed-onset combined immune deficiency and granulomatous/autoimmune manifestations (CID-G/AI), produce a broad spectrum of autoantibodies. Neutralizing anti-IFN-α or anti-IFN-ω antibodies were present at detectable levels in patients with CID-G/AI who had a history of severe viral infections. As this autoantibody profile is not observed in a wide range of other primary immunodeficiencies, we hypothesized that recurrent or chronic viral infections may precipitate or aggravate immune dysregulation in RAG-deficient hosts. We repeatedly challenged Rag1S723C/S723C mice, which serve as a model of leaky SCID, with agonists of the virus-recognizing receptors TLR3/MDA5, TLR7/-8, and TLR9 and found that this treatment elicits autoantibody production. Altogether, our data demonstrate that immune dysregulation is an integral aspect of RAG-associated immunodeficiency and indicate that environmental triggers may modulate the phenotypic expression of autoimmune manifestations.

Figure 4. Autoantibody reactivity in Rag1^S723C/S723C (mut/mut) mice after stimulation with TLR3/MDA5, TLR7/-8, and TLR9 agonists.

(A) Detection of IgG autoantibodies by protein microarray. Six- to eight-week-old mut/mut and WT/WT mice received a weekly i.p. injection of either PBS, low-dose poly(I:C), R848, or CpG. The presence of IgG antibodies against self-antigens in plasma samples was determined at day 0 and week 12 of treatment. Plasma pooled from lupus-prone NZM/MRL mice served as a positive control. Scale bars represent the MFI fold increase (blue to yellow) of autoantibody reactivity as compared with the mean + 2 SDs of the MFI in samples from WT/WT mice. (B) Frequency of autoantibodies in mut/mut mice after TLR/MDA5 stimulation. Multireactive samples were defined as containing autoantibodies against greater than or equal to 20% of 76 self-antigens. Wilcoxon test with Holm’s correction was used to compare results in WT/WT and mut/mut mice at baseline (week 0) and for each of the treatments and approached significance at week 0 (P = 0.0101). When comparing each of the 4 treatments with week 0 separately in WT/WT and in mut/mut mice, only CpG in WT/WT mice was significantly different (P = 0.0036). (C) Validation of dsDNA antibodies by ELISA. Plasma samples were diluted 200-fold. Wilcoxon test with Holm’s correction was used to compare results in WT/WT and mut/mut mice at baseline (week 0) and for each of the treatments as well as to compare each of the 4 treatments with week 0 separately in WT/WT and mut/mut mice. Results from 3 separate experiments were pooled.

Figure 3. Neutralizing activity of anti-cytokine antibodies.

(A) Neutralizing activity of anti–IFN-α, anti–IFN-γ, and anti–IFN-ω antibodies. Normal PBMCs were incubated in the presence of plasma from a healthy control (left panel) or from a RAG-deficient patient (right panel), and left unstimulated or stimulated with either IFN-α, IFN-γ, or IFN-ω. STAT1 phosphorylation was measured by flow cytometry. The neutralizing effect of plasma from patient CID-12 on IFN-α (red), and IFN-ω (green), but not on IFN-γ (blue), is shown as a representative example. (B) Neutralizing activity of anti–IL-12p70 antibodies. Normal lymphoblasts were incubated in the presence of plasma from a healthy control (left panel) or from RAG-deficient patient CID-1 as a representative example (right panel) and left unstimulated or stimulated with IL-12p70. STAT4 phosphorylation was measured by flow cytometry. (C) Neutralizing activity of anti–IL-22 autoantibodies. A549 adenocarcinomic human alveolar basal epithelial cells were incubated in the presence of plasma from a healthy control (left panel) or from RAG-deficient patient CID-1 as a representative example (right panel) and left unstimulated or stimulated with IL-22. STAT3 phosphorylation was measured by flow cytometry. (D) Neutralizing activity of anti–TNF-α antibodies. Jurkat 3T8 cells were incubated in the presence of plasma from a healthy control (left panel) or from RAG-deficient patient CID-12 as a representative example (right panel) and left unstimulated or stimulated with TNF-α overnight. Thy1 surface expression was measured by flow cytometry.

Figure 2. Anti-cytokine antibodies in RAG-deficient patients.

(A) Heatmap of autoantibody reactivity. Plasma samples from 16 healthy controls (Ctr), 14 RAG-deficient patients (OS, n = 3; LS, n = 2; CID-G/AI, n = 8; idiopathic CD4⁺ TCL, n = 1), and 1 patient with APS-1 were tested for anti-cytokine antibodies. The complete array is shown in the left panel, and the area with the highest reactivity is magnified on the right. Antibodies against IFN-α, IFN-ω, and IL-12 (in red) were detected with high MFI in cluster 2, including RAG-deficient patients and APS-1 patients as a positive control. (B) Elevated levels of antibodies against TPO and IFN-α were detected in RAG-deficient patients as compared with healthy controls using SAM after 10,000 permutations of the data with an FDR of less than 0.00001. (C) Multiplex bead assay for anti-cytokine antibodies. Levels of antibodies targeting IFN-α, IFN-ω, IL-12p70, IFN-γ, IFN-β, TNF-α, and IL-22 in healthy controls (n = 15) and RAG-deficient patients (n = 23), grouped by phenotype: SCID, n = 3; LS, n = 3; OS, n = 5; CID-G/AI, n = 11; and TCL, n = 1. (D) Detection of anti–IFN-α-2A, –IFN-ω, and –IL-12p70 antibodies by ELISA. Plasma samples were assayed for IgG autoantibodies at a 200-fold dilution. RAG-deficient patients included those with SCID (n = 2); LS (n = 3); OS (n = 3); CID-G/AI (n = 9); and TCL (n = 1). In both C and D, floating bars indicate the range of values for each autoantibody in healthy controls (n = 15 in C; n = 6 in D).

Figure 1. Autoantibodies in RAG-deficient patients as detected by protein microarray.

(A) IgG autoantibodies in 19 healthy controls (HCs) and 22 patients with RAG mutations. RAG-deficient patients were divided into 2 groups according to the severity of the clinical phenotype. Group 1 included patients with T^–B^– SCID, OS, and LS. Group 2 included patients with delayed presentation and/or a milder phenotype: CID-G/AI and TCL. MFI was normalized to that of healthy controls (mean + 2 SDs = 1), generating the RAR. Healthy controls had a lower percentage of positivity than did patients with RAG mutations (P = 0.0000008), and group 1 had a lower percentage of positivity than did group 2 (P = 0.028) as determined by Wilcoxon test with Holm’s adjustment. Samples that reacted to at least 20% of the self-antigens were defined as multireactive. (B) RAR to 11 autoantigens for which significantly higher levels of autoantibodies were found in the plasma of 5 CID-G/AI patients as compared with levels in 19 healthy controls. *P < 0.0005, **P < 0.0001, and ***P < 0.0001 by Wilcoxon test with Holm’s adjustment. Empty boxes indicate the range of RAR in healthy controls, with the bar representing the mean value. Hemo, hemocyanin; MPO, myeloperoxidase; PCNA, proliferating cell nuclear antigen; PL-12, alanyl-tRNA synthetase; PG, proteoglycan; RPLP, ribosomal phosphoprotein 0; Ro/SSA, ribonucleoprotein/Sjögren’s syndrome antigen A, 52 kDa; Tg, thyroglobulin; U1-BB’, U1 small nuclear ribonucleoprotein BB’ 9; U1-C, U1 small nuclear ribonucleoprotein C.