Immunodeficiency and bone marrow failure with mosaic and germline TLR8 gain of function

Abstract

Inborn errors of immunity (IEI) are a genetically heterogeneous group of disorders with a broad clinical spectrum. Identification of molecular and functional bases of these disorders is important for diagnosis, treatment, and an understanding of the human immune response. We identified 6 unrelated males with neutropenia, infections, lymphoproliferation, humoral immune defects, and in some cases bone marrow failure associated with 3 different variants in the X-linked gene TLR8, encoding the endosomal Toll-like receptor 8 (TLR8). Interestingly, 5 patients had somatic variants in TLR8 with <30% mosaicism, suggesting a dominant mechanism responsible for the clinical phenotype. Mosaicism was also detected in skin-derived fibroblasts in 3 patients, demonstrating that mutations were not limited to the hematopoietic compartment. All patients had refractory chronic neutropenia, and 3 patients underwent allogeneic hematopoietic cell transplantation. All variants conferred gain of function to TLR8 protein, and immune phenotyping demonstrated a proinflammatory phenotype with activated T cells and elevated serum cytokines associated with impaired B-cell maturation. Differentiation of myeloid cells from patient-derived induced pluripotent stem cells demonstrated increased responsiveness to TLR8. Together, these findings demonstrate that gain-of-function variants in TLR8 lead to a novel childhood-onset IEI with lymphoproliferation, neutropenia, infectious susceptibility, B- and T-cell defects, and in some cases, bone marrow failure. Somatic mosaicism is a prominent molecular mechanism of this new disease.

Graphical abstract

Figure 1.

Patients have mosaic and germline variants in TLR8 with normal expression of TLR8 protein and enhanced responsiveness to TLR8 stimulation. (A) Abdominal computed tomography scan from P1 showing marked splenomegaly. (B) H&E staining of bone marrow biopsy specimen (P3) showing hypocellularity for age with lymphohistiocytic aggregate (arrow). (C) Leder staining demonstrating myeloid hypoplasia (P3). (D) Family pedigrees of the 6 patients with variants in TLR8. Symbols with dots indicate mosaicism, and the solid black box indicates a germline variant in P6 (ND, not done). (E) ddPCR of patient DNA showing droplets with TLR8 variant p.P432L (upper left), WT TLR8 (lower right), both templates (upper right), or no TLR8 template (lower left) for P1 and the mother. Monos, monocytes; NK, natural killer; WB, whole blood, NT, no TLR8 template. (F) Percentage of droplets with variant or WT sequence in DNA or cDNA from whole blood or PBMCs, saliva, fibroblast lines, and/or sorted cell populations. (G) Intracellular TLR8 expression by flow cytometry in cells from age-matched healthy controls (solid red lines) and patients (P1, P3, and P5, solid blue lines). Similar expression of TLR8 was observed in patient monocytes and neutrophils. CD3⁺ T cells did not express TLR8. Red dashed line and blue dashed line indicates unstained control in healthy controls and patients respectively. (H) Expression of phosphorylated p65 (NF-κB) in monocytes (CD14⁺) from patients (P1 and P3) and healthy age-matched male controls stimulated with indicated doses of TLR8 agonist TL8-506. A small percentage (5% to 6%) of patient monocytes responded to the lower dose (100 ng/mL) of the stimulant. Healthy cells responded at the highest dose of TLR8 stimulation. There was no statistical difference between the patient cells and healthy cells with respect to their response at the highest dose of stimulation.

Figure 2.

Functional studies demonstrate gain of function of TLR8. (A) Analysis of residues altered by TLR8 variants show that positions P432 and F494 contact the symmetric positions in the other subunit of the active ligand-bound dimer. G572D in the ligand-binding state is predicted to enter into a hydrophobic pocket and interact with R375. (B-C) NF-κB reporter cells (HEK Blue Null1 cells) that do not express endogenous TLR8 were transfected with WT TLR8, patient TLR8 variants (encoding p.P432L, p.F494L, p.G572D), or a loss-of-function (LOF) TLR8 variant (encoding p.D543A) and stimulated with the indicated doses of the TLR8-specific agonist TL8-506 (B) or the TLR8/TLR7 agonist CLO75 (C) for 24 hours. Mosaic and germline TLR8 variants lead to gain of function in TLR8 activity as measured by NF-κB transcriptional activity. Data are represented as mean ± standard deviation of biological replicates and representative of 8 independent experiments (TL8-506) or 3 independent experiments (CLO75). **P ≤ .01, ***P ≤ .001 by 2-way ANOVA test.

Figure 3.

Immunological findings in TLR8 gain-of-function patients. (A) Representative flow plots of T and B cells immunophenotyping showing a skewed ratio of CD4⁺ T cells expressing CD45RA or CD45RO (RA:RO ratio) and reduced class-switch memory B cells in a representative patient. (B) Quantitation of CD4:CD8 ratio, CD45RA:CD45RO ratio, percentage of effector memory T cells re-expressing CD45RA (CD45RA⁺ CCR7⁻ effector memory T cells [TEMRAs]), and percentage of memory B cells and class-switch memory B cells relative to age-matched healthy male controls. In contrast to healthy controls, patients had a significantly reduced percentage of class-switch memory B cells. Young healthy male controls (<6 years) are indicated in red and young adult healthy male controls in green. Horizontal line indicates the median; a Mann-Whitney rank sum test was used for statistical evaluation. Findings that are statistically significant are denoted by an asterisk (*P ≤ .05).

Figure 4.

Elevated serum cytokines in patients. Patients have increased levels of IL-18, IL-18 binding protein (BP), IFN-γ, sCD25, BAFF, and IL12/23p40. No statistical difference was observed between the levels of Fas ligand in the patient and healthy group; however, 3 patients had elevated levels of Fas ligand (>200 pg/mL). Data are represented as mean ± standard deviation; a Mann-Whitney rank sum test was used for statistical evaluation. Multiple samples collected >6 months apart were assayed for P1 and P3. Young healthy male controls (<6 years) are indicated in red and young adult healthy controls in green. L, ligand; ns, not significant; S, soluble.

Figure 5.

Myeloid differentiated cells derived from patient iPSCs and gene-edited CD34s with the TLR8 variants demonstrated increased responsiveness to TLR8 stimulation. (A) Cytospin analysis of cells derived from neutrophil differentiation of WT or p.P432L iPSC clones from patient (P2) and WT or p.F494L iPSC clones from patient (P3) and their upregulation of phosphorylated p65 (NF-κB) in response to stimulation with the indicated doses of the TLR8-ligand TL8-506. (B) H&E stain of cytospin from cells derived from neutrophil differentiation of WT and gene-edited (p.P432L) healthy donor human CD34⁺ HPCs and their upregulation of phosphorylated p65 (NF-κB) in response to the TL8-506. Cells in flow plots are gated on CD45⁺CD66b⁺ neutrophils. Data are representative of 2 independent experiments. (C) Cytospin analysis of cells derived from macrophage differentiation of WT or p.P432L induced pluripotent stem cell (iPSC) clones from patient (P2) and phosphorylated p65 (NF-κB) in response to stimulation with the indicated doses of the TLR8-ligand TL8-506. Cells are gated on CD45⁺CD14⁺ macrophages. (D-E) iPSC-derived macrophages with WT or p.P432L TLR8 from patient (P2) were cultured overnight with the indicated doses of TL8-506 (TLR8 agonist), CL264 (TLR7 agonist), or LPS (TLR4 agonist). (D) Cytokines were measured in the cell culture supernatant and demonstrate that macrophages with variant TLR8 produced significantly more IL-6 and TNF-α with low-dose TLR8 stimulation (TL8-506, 25ng/mL) compared with cells with WT TLR8. There was no difference in the cytokine response to TLR7 stimulation (CL264), and doses of CL264 <100 ng/mL did not result in cellular activation (data not shown). (E) WT and p.P432L macrophages had a similar response to the TLR4 ligand LPS with respect to production of TNF-α and IL-6. NS, no stimulation. Data are presented as mean ± SEM and includes data from 3 independent experiments, analyzed by 2-way ANOVA. Findings that are statistically significant are denoted by an asterisk (*P ≤ .05).