Human peripheral blood was obtained from healthy donors and patient. Spleen and lymph node (LN) samples were obtained from patients undergoing splenectomy and LN biopsy. All material was used after written informed consent and in accordance with the approvals 78/2001 and 251/13 of the University Medical Center ethics committee.

Formalin-fixed lymphoid tissues embedded in paraffin were cut into 3 µm slices and stained after deparaffinization and antigen retrieval with different antibodies as listed in supplementary methods. After applying horse radish peroxidase-coupled secondary antibodies, a brown chromogen reaction developed in the presence of DAB as substrate (Dako Autostainer Link^®, Dako). Images were taken with a Zeiss Axioobserver using ZEN software. Colors were adjusted with Adobe Photoshop.

The 23-year-old male patient had bronchiolitis during infancy and suffered from recurrent upper respiratory tract infections and chronic sinusitis during the last 4 years before the diagnosis of CVID. He was also treated for hypothyroidism the last 3 years before diagnosis. One year before the diagnosis of CVID, splenomegaly, generalized lymphadenopathy in the mediastinum and retroperitoneum, pancytopenia, and severe hypogammaglobulinemia were discovered in the course of gastroenteritis. The patient did not respond to vaccination with pneumococcal polysaccharides. His bone marrow was hypercellular and a stomach biopsy indicated lymphocytic infiltrates. Severe splenomegaly was treated with splenectomy; LN biopsies did not reveal malignancies. The family members are healthy with normal serum immunoglobulin levels except for the Hashimoto’s thyroiditis of the mother.

Genomic DNA was extracted from peripheral blood using the QIAamp DNA Blood Mini Kit (Qiagen, UK). All exons and exon-intron boundaries of CTLA4 and JAK3 genes were amplified using primers shown in Tables 1 and 2. PCR products were purified (Qiagen) and sequenced using an ABI Prism 310 genetic analyzer (Applied Biosystems, Foster City, CA, USA) and a BigDye Terminator DNA sequencing kit (Applied Biosystems).

Cells were analyzed by flow cytometry as described before (14, 15). FOXP3 expression was tested using either the FOXP3 staining kit (eBioscience) following the manufacturer’s instructions, or Lyse/Fix-buffer (BD, 20 min, 37°C) and Methanol (90%, 1 h, −20°C) followed by staining for intracellular antigens. Analyses were performed using a Canto II or LSRFortessa (BD) flow cytometer and Diva (BD) and FlowJo software (TreeStar).

PBMC and EBV lines were isolated and cultivated and tested for mycoplasma contamination as described before (14). Jurkat and HEK293T/17 cells were obtained from ATCC (Manassas, VA, USA). The acute megakaryoblastic leukemia line M07e was received from DSMZ (Braunschweig, Germany). The cell culture medium for MO7e was supplemented with 10 ng/ml GM-CSF (R&D Systems). Dose-dependent proliferation of transduced MO7e cells (5 × 10⁴ cells/ml) was monitored by flow cytometry using 50–300 U/ml IL-2 (Novartis) or 10 ng/ml GM-CSF.

Activation of T cells was performed with 0.3–1 × 10⁶ PBMCs by adding anti-CD3 (1 µg/ml, OKT-3, Novartis, Basel, Switzerland) and anti-CD28 (1 µg/ml, 15E8, Novartis) in complete IMDM.

Suppression assays with Jurkat T cells were performed by mixing 5 × 10⁴ naïve CD4 T cells enriched from PBMCs using the EasySep human (naïve) CD4⁺ T Cell isolation kit (Stemcell Technologies, Vancouver, BC, Canada), with 1 × 10⁴ irradiated (50 Gy) EBV cells, increasing numbers of transduced, YFP sorted, and irradiated (50 Gy) Jurkat cells. The CD4 T cells were activated with anti-CD3 (1 µg/ml), cultured in complete IMDM and pulsed with 1 μCi/well of ³H-Thymidine (Perkin Elmer, Norwalk, CT, USA) for 16 h. The incorporation of ³H-Thymidine was measured using a β-plate counter (Wallac).

T_FH/B cell co-culture assays were carried out with PBMCs sorted by FACS into CD19⁺CD27⁻IgA⁻IgG- (naïve) B cells and CD4⁺CD8⁻CD45RA⁻CXCR5⁺ T_FH cells. The purity of the sorted B and T cell subsets was >95%. 1 × 10⁴ T_FH cells/well were co-cultured with allogeneic B cells (1:1) and activated with endotoxin-reduced Staphylococcal enterotoxin B (SEB, 1 µg/ml, Toxin Technology, Sarasota, FL, USA) in complete RPMI 1640 medium complemented with 2-mercaptoethanol and in 384 well plates. After 7 days, supernatants were collected and levels of secreted cytokines were assessed using Meso Scale Discovery (Meso Scale Diagnostics, LLC, MD, USA) following the manufacturer’s instructions and an MSD Sector Imager 6000 (MSD).

Intracellular cytokines expressed in activated T_FH cells from the co-culture assay were detected upon activating the T cells for 6 h with 100 ng/ml PMA (Sigma-Aldrich, St. Louis, MO, USA) and 750 ng/ml ionomycin (Sigma-Aldrich) in the presence of Brefeldin A (10 mg/ml, Sigma-Aldrich), which was added after 2 h.

CD19⁺CD27⁻IgA⁻IgG⁻ (naïve) B cells were differentiates in vitro into plasmablasts in U96 well plates with 10⁴/well in the presence of CD40L (1 µg/ml, Novartis), IL-4 (20 ng/ml, R&D), and IL-21 (20 ng/ml, Novartis) in complete RPMI 1640 medium supplemented with Glutathione (Sigma-Aldrich), MEM-non essential amino acids (Invitrogen), and insulin–transferrin-selenium-x supplement (Invitrogen). Cells and supernatant were collected after 6 days and tested for the expression of surface markers by flow cytometry and for the secretion of IgM and IgG by ELISA as described before (14–16).

EBV cells (5 × 10⁵) were incubated with abatacept (orenica, Bristol-Myers Squibb, New York, NY, USA) or recombinant CTLA-4-Ig (Novartis). CD28 on Jurkat cells (5 × 10⁵) was first blocked by treatment with anti-CD28 (CD28.2, BD) followed by incubation with recombinant CD80/CD86 (R&D). The cells were then stained with FITC-coupled anti-human IgG (eBioscience) and analyzed by flow cytometry.

Western blotting was performed as described before (15). Transduced MO7e cells were washed twice with FCS-free medium and incubated overnight in IMDM, 0.1% FCS. Then, 7 × 10⁵ cells were washed with ice-cold PBS, lysed with NuPAGE LDS sample buffer (5 × 10⁶ cells/ml), vortexed, and heated to 70°C for 10 min. 20 µl/sample were separated by 4–12% PAGE and transferred to PVDF membranes using the Trans-Blot turbo system (Bio-Rad, CA, USA).

Synthetic 738 bp DNA fragments encoding either wild-type or Y139C mutant CTLA-4 were obtained from Integrated DNA Technologies (Leuven, Belgium). The ssBMS-CTLA-4-Ig expression plasmid and the pcDNA3.1-based plasmids containing JAK3-WT or JAK3-A572V were obtained from C. Haan (17). CTLA-4-Y139C and JAK3-R840C were generated by using the QuikChange XL site-directed mutagenesis kit (Agilent Technologies) and following oligonucleotides: CTLA-4 (full-length)

f-5′-CATGTACCCACCGCCATGCTACCTGGGCATAG-3′ and

The lentiviral expression vectors pNL-CEF-eGFP and pNL-CEF-YFP/CFP were described before (14, 16). CTLA-4 coding regions were cloned into pNL-CEF-YFP/CFP by replacing a 702 bp fragment with the corresponding synthetic DNA fragments. An internal ribosomal entry site (IRES) was cloned from pIRES2-eGFP (Clontech Laboratories, Mountain View, CA, USA) into pNL-CEF-eGFP generating the pNL-CEF-IRES-eGFP backbone. JAK3 coding regions were then sub-cloned into pNL-CEF-IRES-eGFP. Enzymes used for cloning were from New England Biolabs (Ipswich, MA, USA), Roche (Basel, Switzerland) or from Fermentas (Vilnius, Lithuania). All constructs were verified by Sanger sequencing.

Lentiviral particles were generated using a standard protocol (14) upon transfection of HEK293T/17 cells with pCDNLBH* and pVSV-G, and the pNL-based constructs. Supernatants were collected 2 days after transfection, filtered (0.45 µm, Pall Corporation, NY, USA) and 1 ml were used to spin-infect 5 × 10⁵ Jurkat, MO7e cells, or anti-CD3 (1 µg/ml) and anti-CD28 (1 µg/ml) activated PBMCs or CD4⁺ T cells.

Recombinant CTLA-4-Ig was produced by growing suspension-adapted HKB11 cells (Human Embryonic Kidney cells fused with Burkitt’s lymphoma cells, Bayer) transfected with expression plasmids in serum-free culture medium M11V3 in a WAVE bioreactor. Cultures were supplemented with an equal volume of DM133 enriched with Yeastolate (2 g/l) (Irvine Scientific, Santa Ana, CA, USA) and incubated for 3–8 days.

Supernatants were filtered through a dead end filter (Millistak + Pod Disposable Depth Filter System Merck Millipore, Darmstadt, Germany), sterilized (0.22 µm stericup filter, Millipore), and concentrated (Cross flow filter Hemoflow, 10 kDa, Fresenius, Bad Homburg, Germany). Proteins were purified by MabSelect affinity chromatography (GE Heathcare, Chicago, IL, USA), eluted (50 mM Tris, 90 mM NaCl pH 3.2) and immediately neutralized to pH 7.3 by slow addition of 1 M Tris pH 10.

The modeling of the JAK3 JH2-JH1 homology domain was done using TYK2 (PDB 4OLI) (11) as template and Prime 4.2 software of Schrödinger Release 2015-4 (Schrödinger, New York, NY, USA). Images of the CTLA-4/CD80 structure [PDB 1I8L (18)] and the JAK3 model were generated with ICM (Molsoft LLC, San Diego, CA, USA).

Significant differences were analyzed by unpaired Student’s t-test, by Tukey’s or Holm–Sidak’s (one-way ANOVA) or Dunnett’s or Holm–Sidak’s (one-way ANOVA) multiple comparisons test using GraphPad Prism software (GraphPad). Values of P ≤ 0.05 were considered significant.

Analyzing the exomes of a primary antibody deficiency cohort of 37 individuals from 9 families, we identified in one family a novel heterozygous missense mutation (p.Y139C) in the CTLA4 gene (Figure 1A). The mutation is shared by the patient (A.II.1) and the father (A.I.1). While the patient is diagnosed with hypogammaglobulinemia, signs of autoimmunity and severe lymphoproliferation which required splenectomy, the father is asymptomatic and has normal IgG and IgA levels (Table 3). Since the p.Y139C CTLA-4 variant has not been documented before and because it is not shared by any of A.I.1 siblings, it most likely has arisen de novo in A.I.1. The mutation is localized in the evolutionary highly conserved ligand-binding motif (MYPPPY) of CTLA-4 and changes the second tyrosine to cysteine (Figure 1B). A very similar change (Y139A) introduced by site-directed mutagenesis was shown to abolish CTLA-4 binding to CD80 and CD86 (19, 20).

To address the question if the Y139C mutation changes the surface expression of CTLA-4, we constructed lentiviral expression vectors for wild-type (WT) and mutant (Y139Y) CTLA-4 expressing fusion proteins of CTLA-4 linked at the C-terminal end to CFP and YFP.

Co-expression of wild-type and mutant forms revealed that the mutant did not interfere with the expression of the wild-type form. However, comparing the expression of both fusion proteins, we found that the Y139C mutant was expressed at lower levels on the cell surface (Figure 1C) although total expression levels of CTLA-4-YFP and CTLA-4-Y139Y-YFP were very similar (Figure S1 in Supplementary Material).

In contrast to CTLA-4-WT, CTLA-4-Y139C did not bind recombinant soluble CD80 or CD86. Vice versa, recombinant CTLA-4-Y139C did not bind to CD80/CD86 expressed on an EBV immortalized B cell line (Figure 1D; Figure S2 in Supplementary Material). In an in vitro suppression assay, Jurkat T cells expressing the CTLA-4-Y139C variant did not suppress co-stimulation-dependent proliferation of naïve CD4⁺ T cells (Figure 1E). Analyzing the T_REG cells of the patient, we found that they did not suppress the proliferation of activated responder T cells in vitro, whereas the T_REG cells of the carrier A.I.1 still showed suppressor activity under the assay conditions (Figure S3 in Supplementary Material). We, therefore, conclude that CTLA-4-Y139C is a loss-of-function mutation with incomplete penetrance in vivo and similar to other previously reported mutations inactivating CTLA-4 (5–7).

The percentage of CD45RA⁺CCR7⁺ naïve T cells was reduced in the patient A.II.1 but not the healthy carrier A.I.1, while the proportion of CD45RA^–CCR7^– CD4⁺ effector memory T cells (T_EM cells) and of CD45RA⁺CCR7^– CD8⁺ effector memory T cells (T_EMRA cells) was markedly higher than in controls (Figure 2A). Similar changes in the T cell compartment have been reported previously for other CTLA-4 haploinsufficiency patients (6, 7), including an increased percentage of CD4⁺FOXP3⁺ T cells, which was higher in patient A.II.1 than in the carrier A.I.1 (Figure 2B). In contrast to the increase in CD4⁺FOXP3⁺ T cells frequency, the expression levels of intracellular FOXP3 and CTLA-4 in FOXP3⁺CD25⁺ T_REG cells were comparable among all family members and healthy donors (Figure 2C). Therefore, we conclude that CTLA-4 haploinsufficiency in A.I.1 and A.II.1 results—like the recently described CTLA-4-P137R mutation (21)—from a functional defect of CTLA-4 and not from impaired expression of CTLA-4 by T_REG cells.

Since the heterozygous CTLA4 mutation was found in the patient and in the healthy carrier, we assumed that the patient might have inherited another genetic defect with incomplete penetrance from his mother (A.I.2). This genetic defect would enhance the consequences of CTLA-4 haploinsufficiency and lead to an overt immune dysregulation syndrome. Therefore, we searched for rare genetic variants in the exome of patient A.II.1, which were inherited from the mother A.I.2, but not from the father A.I.1. A rare, heterozygous missense JAK3 mutation (p.R840C, rs200077579) fulfilled these criteria. Among Europeans, missense mutation has been found in 1/3335 individuals (22, 23). Since the R840 residue is located in the N-lobe of the JH1 domain (24) close to a loop of a beta-sheet motif, which covers the ATP binding site, the R840C mutation might change the conformation of the N-lobe and its ability to interact with the pseudokinase domain (Figures 3A,B). This would change the kinase function and the activity of JAK3. As a consequence, cells expressing R840C JAK3 would respond differently to ILs like IL-2, which bind to receptors containing the JAK3-associated cytokine receptor common γ-chain.

Using different readout systems, we, therefore, tested the activity of the R840C variant in response to IL-2. First, we transduced the IL-2 indicator cell line MO7e with lentiviral expression vectors encoding the wild-type form of JAK3, the constitutive-active, and lymphoma-associated variant JAK3-A572V (17), or the JAK3-R840C variant, and analyzed IL-2-dependent growth at increasing IL-2 concentrations (Figure 3C). Different from wild-type JAK3, the constitutive-active A572V JAK3 and, to a lesser extent, the JAK3-R840C mutant supported IL-2-independent growth as well as higher growth rates at low IL-2 concentrations (Figure 3C). Similar to MO7e cells, expression of the JAK3 variant R840C also enhanced the growth of lentivirally transduced primary human CD4⁺ T cells (Figure 3D). Next, we assessed the phosphorylation of STAT5, the direct target of JAK3 kinase activity in MO7e (Figure 3E) and primary human CD4⁺ T cells (Figure 3F) transduced either with wild-type JAK3 or the gain-of function mutants of JAK3. As expected, basal level and IL-2-induced STAT5 phosphorylation was higher in cells expressing the JAK3-R840C or the constitutive active A572V variant than in cells expressing WT JAK3 (Figures 3E,F). Thus, the R840C missense mutation results in a gain-of-function variant of JAK3 as it increases the kinase activity of the enzyme.

Compared to the constitutive-active and lymphoma associated JAK3-A572V variant, the R840C variant has lower intrinsic JAK3 kinase activity, which explains why it has not yet been reported in conjunction with T cell hyperplasia, and why the carrier A.I.2 does not show signs of hematological malignancies.

Our data revealed that the patient’s hypogammaglobulinemia is associated with CTLA-4 haploinsufficiency, increased JAK3 activity, and increased percentages of circulating CD4⁺ and CD8⁺ effector memory T cells. Since IgG and IgA-secreting long-lived plasma cells develop in germinal centers, we assumed that the immunological changes described above correlate with changes in the architecture of germinal centers. To this end, we analyzed paraffin-embedded tissue sections of spleen and LN samples of the patient using a set of markers defining T cell and B cell subsets. LN sections showed many follicles with very large, irregularly shaped BCL6-positive germinal centers (GC) containing many ICOS and PD-1 positive T_FH cells (Figure 4A). In spite of the GC hyperplasia, the typical polarization was clearly detected with distinct dark zones containing proliferating Ki67⁺ cells, and light zones containing IRF4⁺ GC B cells. Although the GCs contained IgG⁺ class-switched plasmablasts as well as BLIMP-1 expressing plasma cell precursors (Figure 4A), mature CD138⁺ plasma cells were not detected in GCs and found only in very small numbers in extrafollicular areas. Staining of B cells for IgD, IgM, and CD20 in spleen sections (Figure 4B) revealed a “follicle in the follicle-like” architecture in both LN and spleen. These structures also contained large ICOS⁺ and PD-1⁺ germinal centers with CD25⁺ cells, which include activated and T_REG cells. Therefore, the development of GC B cells into plasma cells seems to proceed up to the stage of plasmablasts within GCs but appears not to continue beyond this stage because CD138⁺ extrafollicular plasma cells were almost absent. Similar to the differentiation into plasma cells, the development of switched memory B cells was defective, as memory B cells were not detected in circulation (see below).

Circulating PD-1⁺CD4⁺CXCR5⁺ T cells were increased in frequency with a large proportion of cells expressing ICOS (Figure 5A). Moreover, CXCR5⁺PD-1⁺ T cells of A.II.1 expressed more PD-1 than the cells of all controls (Figure 5A). Most of these CXCR5⁺ T_FH cells also expressed TIGIT (Figure 5B), a “checkpoint” molecule expressed by circulating T_FH cells. This subset was reported to have an increased capacity in supporting B cell proliferation and class switch recombination (25). In the patient, all CXCR5⁺TIGIT⁺ T cells were expressing PD-1 (Figure 5B). A slight, but significant increase of these activation markers was also observed on CD4⁺ T cells of the non-symptomatic CTLA-4 Y139C carrier A.I.1 compared to controls (Figure 5C). Since the phenotypic analysis of another patient with CTLA-4 haploinsufficiency (p.Y89X) but wild-type-JAK3 identified in this study revealed normal expression levels of PD-1 and TIGIT in circulating CD4⁺T cells (Figure 5D), we tested if activation of the JAK3 pathway as found for the R840C mutation upregulates PD-1 and TIGIT expression in CD4⁺ T cells. To this end, we activated CD4⁺ T cells from healthy controls with IL-7 and IL-15 in the presence and absence of the JAK inhibitor tofacitinib. While activation with IL-7 and IL-15 lead to an increase in PD-1 and TIGIT surface expression, the upregulation was blocked by tofacitinib (Figure 5E). This suggests that the enhanced activity of the JAK3-R840C variant might contribute to the unusual PD-1⁺TIGIT⁺ T_FH cell phenotype of the patient.

Extrapolating from the observations made in CTLA-4 deficient mice (9), it would have been expected that the increased proportion of T_FH cells in patient A.II.1 enhances plasma cell development. However, consistent with the patient’s hypogammaglobulinemia, immunohistochemical analysis of LN sections revealed very few extrafollicular CD138⁺ plasma cells although IgG⁺ cells were detectable within the GCs (Figure 4A). Numbers of circulating B cells were normal but lacked a CD27⁺ memory B cell compartment (Figure 6A). Interestingly, B cell numbers increased and remained at relatively high counts after splenectomy without being enriched for CD10⁺CD38⁺ transitional B cells (Figures 6A,B). In addition, the compartment of circulating B cells contained mainly CD21^lo cells, which are likely to represent recently activated cells that respond poorly to new stimulating signals (Figure 6C) (26). An increase in CD21^lo B cells observed in HIV patients (27) is mainly attributed to the expansion of dysfunctional T-helper cells (28). To discriminate between intrinsic B cell defects impairing the differentiation into IgG switched memory B cells and plasma cells and disturbed T_FH cell function, we next tested in vitro differentiation of B cells and cytokine secretion by activated T cells.

The intrinsic B cell defect was excluded as CD27⁻IgG⁻IgA⁻ sorted (naïve) B cells of A.II.1 developed in vitro into class-switched IgG⁺ (Figure 6D) and antibody secreting cells (Figure 6E) similar to naïve B cells from healthy controls. Next, we tested T_FH cells function by monitoring cytokine production. The frequency of IL-21 and IL-4-expressing T_FH cells from A.II.1 and the amount of secreted IL-4, IL-6, and IL-10 were significantly lower, while other cytokines appeared normal compared to T_FH cells from HDs (Figure 6F). Thus, T_FH cells from A.II.1 are deficient in providing the key cytokines IL-21 and IL-4, which are crucial for the differentiation, expansion, class-switch recombination, and survival of antigen-selected B cells in the GC (29, 30). The molecular cause for the selectively impaired production of IL-21, IL-4, and IL-10 is not understood yet, but the high expression of PD-1, a negative regulator of T cell activation (31) might interfere with cytokine secretion and contribute to the immunodeficiency.