Autoimmunity and immunodeficiency associated with monoallelic LIG4 mutations via haploinsufficiency

Abstract

Background: Biallelic mutations in LIG4 encoding DNA-ligase 4 cause a rare immunodeficiency syndrome manifesting as infant-onset life-threatening and/or opportunistic infections, skeletal malformations, radiosensitivity and neoplasia. LIG4 is pivotal during DNA repair and during V(D)J recombination as it performs the final DNA-break sealing step.

Objectives: This study explored whether monoallelic LIG4 missense mutations may underlie immunodeficiency and autoimmunity with autosomal dominant inheritance.

Methods: Extensive flow-cytometric immune-phenotyping was performed. Rare variants of immune system genes were analyzed by whole exome sequencing. DNA repair functionality and T-cell-intrinsic DNA damage tolerance was tested with an ensemble of in vitro and in silico tools. Antigen-receptor diversity and autoimmune features were characterized by high-throughput sequencing and autoantibody arrays. Reconstitution of wild-type versus mutant LIG4 were performed in LIG4 knockout Jurkat T cells, and DNA damage tolerance was subsequently assessed.

Results: A novel heterozygous LIG4 loss-of-function mutation (p.R580Q), associated with a dominantly inherited familial immune-dysregulation consisting of autoimmune cytopenias, and in the index patient with lymphoproliferation, agammaglobulinemia, and adaptive immune cell infiltration into nonlymphoid organs. Immunophenotyping revealed reduced naive CD4⁺ T cells and low TCR-Vα7.2⁺ T cells, while T-/B-cell receptor repertoires showed only mild alterations. Cohort screening identified 2 other nonrelated patients with the monoallelic LIG4 mutation p.A842D recapitulating clinical and immune-phenotypic dysregulations observed in the index family and displaying T-cell-intrinsic DNA damage intolerance. Reconstitution experiments and molecular dynamics simulations categorize both missense mutations as loss-of-function and haploinsufficient.

Conclusions: This study provides evidence that certain monoallelic LIG4 mutations may cause human immune dysregulation via haploinsufficiency.

Keywords: DNA damage–autoimmunity; DNA ligase 4; autosomal dominant; haploinsufficiency; immunodeficiency; inborn errors of immunity; primary immunodeficiency.

FIG 1.

MultipleautoimmunemanifestationsandreductionofnaiveTcellsintheperipheralbloodofP1andher father. (A) Clinical manifestations in the index patient P1, thrombocyte counts, and hemoglobin levels; gray background depicts reference range. (B) P1’s kidney biopsy during interstitial nephritis. Immunohistochemistry staining with anti-CD20 and anti-CD4. (C) Pulmonary tissue gated computer tomography scan of P1 during the pneumonitis episode and (D) after steroid treatment. (E) Lung biopsy specimens during the pneumonitis episode and stained with anti-CD20 and anti-CD3. (F) Cranial magnetic resonance imaging, showing parotid gland swelling (white arrowheads). (G) Peripheral blood T-cell subsets with naive (CD27⁺CD45RO⁻), effector memory (EM; CD27⁻CD45RO⁺) and central memory (CM; CD27⁺CD45RO⁺), and (H) quantification. (I) Cell Trace violet (CTV) dilution after 5 days of in vitro stimulation. (J) Enumeration of T cells bearing the TCR Vα7.2 segment by flow-cytometry. The number indicates the frequency within the CD3⁺ T-cell population. (K) Comparison of the TCR Vα7.2⁺ T-cell frequency in P1 and her father with patients affected by combined immunodeficiency (CID), primary antibody deficiency (PAD), autoinflammation (Autoinflamm) or to HDs. (K) Non-parametric Kruskal-Wallis test with Dunn’s correction. **P < .01. ENT, Ears, nose, and throat; ITP, immune thrombocytopenia; IVIG, intravenous immunoglobulin; VZV, varicella-zoster virus.

FIG 2.

Preserved BCR and TCR repertoires. (A) High-throughput sequencing of the TCR loci. Complementarity-determining region 3 (CDR3) length distribution. (B) Shannon’s (H) entropy index; gray shadow for HD values. (C) Simpson clonality index. (D) Individual V gene segment usage. (E) Heatmaps displaying VJ gene pairing; box indicates most distal gene pairing. (F) Surface expression of the BCR light chains. (G) IGH locus cartoon for the constant region (adapted from Bashford-Rogers et al). IGH high-throughput RNA-sequencing for the determination of B-cell maturation status and constant region gene usage. (H) IgA and IgG subclass utilization. Box-plot indicates age-matched HDs values. (I) V family and (J) J gene segment usage. Box-plot indicates values of age-matched HDs. (K) Average of somatic hypermutations. The black line indicates the model fitting the somatic hypermutations increase by age; gray lines indicate the 95% CI. (L) Antigen selection was quantified by the computation of the mean R/S ratio. The black line indicates the model fitting and the R/S increase by age; gray lines indicate the 95% CI. (D) Differential expression analysis empirical Bayes method. (F) Mann-Whitney test with post hoc correction; the HDs’ SD was added to the value of P1. TRAV, T cell receptor alpha variable gene.

FIG 3.

Novel missense variant within the catalytic core of LIG4. (A) Sanger sequencing of c.A1739G in bulk T-cell–derived DNA, the resulting amino acid change at p.R580Q is indicated. (B) Multiple LIG4 protein sequence alignment; p.580 position is highlighted. (C) Molecular representation in ribbons of the human LIG4 catalytic core bound to a DNA duplex. The WT Arg580 is shown as stick (arrow). The corresponding β sheet 18 is indicated. The mutated amino acid resides in the catalytic oligonucleotide/oligosaccharide-fold domain (OBD; blue). Numbers indicate the amino acid position in NP_001091738. DNA binding domain (DBD) in green; nucleotidyltransferase (NTD) in orange. (D) Qualitative PCR was used to measure LIG4 mRNA levels in PBMCs of the 2 patients and healthy controls including the mother. The relative quantity (RQ) was normalized to multiple housekeeping genes and to the mean of the HDs. (E) The LIG4 protein levels were quantified by separating PHA T-cell blast-cell lysates by SDS-PAGE electrophoresis and probed with rabbit-anti LIG4. Right side normalization of LIG4 protein levels to β-actin levels. (D) Nonparametric Mann-Whitney rank test. ns, Not significant.

FIG 4.

LIG4 R580Q reduces DNA-ligation activity and weakens DNA binding. (A) Normalization of recombinant WT or R580Q LIG4 proteins. (B) 42mer nicked DNA duplex. Multiple turnover ligations for WT versus R580Q LIG4 with (C) increasing unadenylated 42mer concentrations and (D) time. Product separation on a Tris/Borate/EDTA (TBE)-urea polyacrylamide gel. (E) Molecular OBD representation, the Arg580 represented as stick (arrows: nearby DNA-backbone phosphorous atoms). (F) Computed LIG4 binding energy (BE) between the WT versus R580Q LIG4 and adenylated-DNA complex. Twelve independent trajectories, each >500 nanoseconds. (G) Residues with BE difference >20 kJ/mol between WT and R580Q. (H) Dihedral χ1 angle time series and (I) distribution focused on residue 580. (J) WT LIG4 and (K) R580Q LIG4 (stick) with the adenylated nicked-DNA as ball and stick; third and fourth phosphate group of DNA backbone (arrows). (L) Minimal distance between the residue side chain (SC) and DNA backbone phosphate groups. The phosphate group numbering is indicated. (M) Temporal fraction, during which residue 580 SC and the DNA backbone phosphate were <4 Å. (N) Bottom: Identification of likely DNA-interacting residues (distance to DNA <3 Å). Middle: Human LIG4 missense mutations (Table I). Top: Missense mutations with potential DNA binding. Mann-Whitney testing (F) with multiple comparison correction (L); (G) 2-way ANOVA with Šídàk correction.

FIG 5.

Augmented DNA-damage susceptibility in vitro. T cells derived from PBMCs were cultured for 2 days without stimulation. The phosphorylation of H2Ax (γH2Ax) and 53BP1 (p53BP1) were assessed by flow cytometry. (A) Quantification (mean of triplicates) and (B) representative flow cytometric plots of the γH2Ax⁺p53BP1⁺ population in bulk CD3⁺ T cells. (C) Kinetics of γH2Ax in CD45R0⁺CD4⁺ helper T cells after 10 Gy irradiation (IR). (D) Analysis of the nuclear γH2Ax⁺ fraction in memory CD45R0⁺ CD4⁺ T cells after in vitro treatment of PBMCs with bleomycin sulfate for 24 hours at indicated concentrations. (E) Cell death after 24 hours in vitro bleomycin sulfate exposure of CD4⁺ T cells (naive CD45R0⁻ and memory CD45R0⁺). (F) T-cell proliferation after IR. T cells were labeled with CTV, followed by IR and stimulation for 5 days in vitro with anti-CD3/anti-CD28 (aCD3/aCD28). Gray-shaded population indicates the maternal nonstimulated condition of T cells. (G) The relative proliferation index was computed for CD4⁺ and CD8⁺ T cells after different IR intensities; stimulation of cells as in (F). (A) Kruskal-Wallis test; (C, D, E, G) 2-way ANOVA with Šídàk correction. Single points represent mean values of duplicates or triplicates for the patients.

FIG 6.

A novel LIG4 A842D mutation substantiates linkage of monoallelic LIG4 mutations with DNA damage-induced T-cell death and immunodeficiency. (A) Sanger sequencing chromatogram of heterozygous LIG4 A842D mutation in P3 and P4. (B) Cross-species alignment of A842-proximal LIG4 residues. (C) LIG4-XRCC4 molecular complex highlighting residue 846-proximal area of BRCT2. Structural domains shown in black (BRCT1/BRCT2), blue (XRCC4-A), and red (XRCC4-B). Simulation snapshots in boxes for WT (top) and A842D (bottom) LIG4. Salt bridges shown as dashed lines when distances were mostly <5 Å during simulation. (D) Dead cell stain-positive frequencies (mean ± SD) in T cells following 24-hour bleomycin exposure in blood donors (n = 15, black); disease-controls (DCs; green); and P1 (R580Q), P3, and P4 (A842D). (E) Post hoc comparisons of 1-way ANOVA for bleomycin-treated groups. Representative data shown as mean of pooled triplicate/quadruplicate (P1), duplicate/triplicate (P3), or triplicate/quadruplicate (P4). (F) Flow-cytometric plots of TCRVα7.2⁺ T cells. (G) TCR Vα7.2⁺ T-cell frequencies of healthy controls (HCs; gray), DCs (green), and in patients with LIG4-mutation (LIG4mut; pink). (H) Two-dimensional plot of ex vivo TCRVα7.2⁺ versus in vitro 24-hour 50 μmol/L bleomycin-induced T-cell death. An empirical slope of 2 is appended. (I) One-way ANOVA of T-cell-functionality slope defined as (24-hour bleomycin-induced dead frequencies)/(TCRVα7.2-positive frequencies).

FIG 7.

LIG4 R580Q and A842D loss-of-function mutants manifest haploinsufficiency on reconstitution. (A) Verification of CRISPR-Cas9-mediated LIG4-KO in Jurkats (top). LIG4-expression impairment was verified by intra-cellular staining (bottom left) and Western blotting (bottom right). (B) Flow-cytometric plots of WT (left) versus LIG4-KO (right) Jurkat T-cells exposed to bleomycin (12 hours). (C) Dose-dependent (50 μmol/L) and time-dependent (12 hours) frequencies of Annexin V–positive apoptotic cell frequencies following bleomycin exposure. Performed in triplicate (0 μmol/L, 10 μmol/L) or quadruplicate (50 μmol/L) and compared by unpaired t-tests. (D) LIG4 functional reconstitution schematic via transient overexpression in LIG4-KO Jurkat T-cells. Cells were magnetofected via cationic polymers with a dual-promoter, LIG4/mCherry coexpressing vector (representative flow plot:bottom), then exposed to bleomycin and evaluated for Annexin V–positivity in mCherry(/LIG4)-positive/negative populations. A representative calculation is shown. (E) Comparison of postbleomycin survival rates in mCherry+ cells normalized against intrawell mCherry−fractions on WT versus mutant LIG4 transfection. Representative of 2 independent experiments performed in quadruplicate. Compared by unpaired t-tests. (F) Comparison of postbleomycin incubation survival rates in mCherry+ cells on WT and mutant LIG4 cotransfection at indicated ratios. Post hoc comparisons of 1-way ANOVA are shown. Pooled data of 2 independent experiments performed in triplicate/quadruplicate/control are shown (mean ± SEM). +BLM, Bleomycin-treated; Neg Ctrl, negative control; SSC-A, side scatter area; Unstim, unstimulated.