TLR8 gain-of-function syndrome

Toll-like receptors (TLRs) are type I transmembrane pattern recognition receptors that are a critical part of innate immunity against infection or injury (1). TLRs sense pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs), and in response promote proinflammatory gene expression and cytokine production (1). There are 10 human TLRs which can be grouped by those localized on the cell surface (TLR1, TLR2, TLR4, TLR5, TLR6, TLR10) and those in intracellular compartments such as the endosome or endoplasmic reticulum (TLR3, TLR7, TLR8, TLR9) (1, 2). TLRs function as homo- or heterodimers, with each monomer containing a horseshoe-shaped ectodomain of multiple leucine-rich repeats (LRRs), a transmembrane domain, and a cytoplasmic Toll/IL-1 receptor (TIR) domain (1, 2).

Defects in genes encoding several TLRs have been linked to monogenic inborn errors of immunity (IEIs) (3, 4). Loss-of-function (LOF) variants in nucleic acid–sensing TLRs can cause pathogen-specific susceptibility. TLR3 deficiency was first identified as causing susceptibility to herpes simplex encephalitis (5) and subsequently to severe influenza pneumonia (6). More recently, LOF variants in both TLR3 and TLR7 have been linked to severe COVID-19 (7, 8, 9). Biallelic LOF variants in TLR4 were described with a variable phenotype associated with impaired sensing of bacterial-derived lipopolysaccharides (10).

We originally reported 6 patients with gain-of-function (GOF) variants in TLR8 associated with neutropenia, lymphoproliferation, and bone marrow (BM) failure in 2021 (11). Since that time, 13 additional cases of TLR8 GOF have been identified (Table 1, summary of published TLR8 GOF patients) (11, 12, 13, 14, 15, 16, 17, 18). Recently, TLR7 GOF was identified as an IEI associated with susceptibility to monogenic systemic lupus erythematosus, with a phenotypic spectrum broadened beyond autoimmunity, with some hematologic abnormalities overlapping with TLR8 GOF syndrome including anemia, thrombocytopenia, and lymphopenia (19, 20, 21, 22, 23, 24, 25, 26, 27). Multiple other IEIs are caused by defects upstream and downstream of the TLR signaling pathway, for example, in genes encoding MyD88, IRAK4, IKBα, NEMO, and UNC93B1, highlighting the nonredundant importance of TLR signaling in human immunity (3, 4, 28, 29, 30, 31, 32, 33).

TLR8 has historically been understudied relative to other TLRs, in part due to the murine ortholog partially lacking single-stranded RNA (ssRNA)–sensing capacity as the human protein (34, 35, 36, 37, 38). However, the discovery of GOF variants in TLR8 leading to a monogenic IEI has transformed the understanding of TLR8 as a critical regulator of myeloid immune homeostasis (11). TLR8 is an endosomal receptor that is highly expressed in myeloid cells and largely absent in B or T lymphocytes (34, 39, 40, 41, 42, 43, 44). TLR8 functions as a homodimer and recognizes degradation fragments of ssRNA derived from pathogens including viruses, bacteria, and protozoa, as well as endogenous self-RNA (45). Structurally, each TLR8 monomer contains 26 LRRs that mediate ligand recognition with a z-loop inserted between LRR14 and LRR15 (residues 433–481) that must be cleaved for dimer formation (Fig. 1 A) (46). At baseline, TLR8 exists as a preformed homodimer in an inactive open conformation (Fig. 1 B) (47). Upon ligand binding, conformational changes bring the two C-terminal TIR domains closer in proximity, enabling downstream signaling through MyD88 (46). Here, we review current knowledge of the molecular, cellular, and clinical features of a spectrum of disease associated with TLR8 GOF and highlight unresolved questions important for understanding the mechanisms of disease.

TLR8 shares structural and mechanistic similarity to TLR7, also an endosomal DAMP and PAMP sensor that recognizes RNA degradation products (37, 50, 51, 52, 53), and the genes encoding TLR7 and TLR8 are adjacent to each other on the short arm of the X chromosome (54, 55, 56). Structurally, TLR7 and TLR8 each have three sites critical for binding agonists and antagonists (Fig. 1 C). Both share a conserved Site 1, which binds single nucleosides (TLR7: guanosine or 2′,3′-cGMP; TLR8: uridine) (57, 58). Site 2 differs in structure and motif preference; however, both TLR7 and TLR8 Site 2 regions bind to short uridine-rich motif (e.g., UUU, UG) oligonucleotides (46, 49, 59). Site 2 binding is auxiliary, enhancing binding affinity of Site 1 rather than being responsible for activation alone (59, 60). Recently, a third site was discovered as an antagonist binding region by the Gantier lab (61). They demonstrated that binding of 3-mer oligos of 2′-O-methyl (2′-OMe)-guanosine RNA with a 5′-guanosine within this site (TLR7: P435, F506, F507; TLR8: P432, F494, F495) stabilizes the inactive open dimer conformation of both proteins (27, 61). In mammalian cells, there are at least 31 ribosomal RNA (rRNA) sites that contain 2′-OMe guanosines (62); thus, this pool of fragmented rRNA may serve as a mechanism to dampen TLR7 and TLR8 signaling at baseline and prevent autoinflammation (61). Notably, binding at this antagonistic site prevents Site 1 and 2 binding, and vice versa (61). Thus, this finding established a competitive model in which agonists compete with antagonists bound to TLR7 and TLR8 at basal levels for control of the conformational state, subsequently switching signaling capacity of the receptors on and off (61). Thus, activation of both TLR7 and TLR8 likely depends upon the displacement of endogenous 2′-OMe guanosine from the antagonist binding site following substantial accumulation of agonistic RNA degradation products in the endosome, a protective mechanism allowing for sensing of pathogens while reducing autoinflammation.

Structural analyses have localized these three functional sites to distinct areas of the receptors’ ectodomains (46, 59, 60, 61, 63) (Fig. 1 C). In TLR8, Site 1 consists of the LRR11-14 region of one monomer and LRR16-18 of the other monomer, where conserved residues coordinate nucleoside binding and are essential for receptor dimerization (46, 47, 59). Site 2 is formed at the dimer interface in the LRR10-13 region, where short uridine-rich oligonucleotides make contacts with residues from both monomers simultaneously (46, 49, 64). The antagonist binding site is principally defined by amino acid residues F494 and F495 in LRR15-16, with P432 in the z-loop contributing indirectly to the geometry of the antagonist pocket (16, 61). Of significance, all known TLR8 GOF-causing variants cluster at residues important for or adjacent to Site 1 in the dimer (p.A518T, p.G572D, p.G572V) or the antagonist binding site (p.P432L, p.P432Q, p.P432R, p.L433F, p.V434L, p.F494Y, p.F494L, with additional evidence for interaction of p.A518T as well) (Fig. 1 and Table 1) (11, 12, 46, 49, 61).

To date, 19 patients with 10 variants in TLR8 have been reported (Table 1) (11, 12, 13, 14, 15, 16, 17, 18). Proteins encoded by reported variants have been functionally characterized as GOF primarily through in vitro NF-κB reporter assays utilizing ectopic expression of TLR8 variants, phosphorylated NF-κB in peripheral blood mononuclear cell samples or patient induced pluripotent stem cell (iPSC)–derived myeloid cells, and/or quantification of basal and TLR8 stimulation–dependent cytokine production in primary patient samples or cell lines expressing variants of interest.

We originally identified 6 male patients who had a severe inflammatory phenotype with lymphoproliferation, neutropenia and other cytopenias, recurrent infections, and in many cases BM failure, all of whom had novel mosaic or germline GOF variants in TLR8 (11). Immunophenotyping identified expanded CD8⁺ T cells with a terminally differentiated effector memory T cell reexpressing CD45RA (TEMRA) phenotype, impaired B cell maturation and hypogammaglobulinemia with some patients having near-absent B cells, marrow infiltration by activated CD8⁺ T cells, and strongly elevated serum cytokines including IL-12, IL-18, and interferon (IFN)-γ, as well as soluble IL-2Rα (11). Most patients (5/6) had disease due to genetic mosaicism, with variant allele fractions (VAFs) between 8 and 26%. Of note, 4 patients had the same mosaic p.P432L variant. One patient who presented in infancy had a de novo germline variant (Table 1, #6). In 3 mosaic patients, VAFs were similar across hematopoietic lineages and between peripheral blood cells and skin fibroblasts, suggesting a postzygotic mutational event early in embryogenesis. One patient had mosaicism limited to the hematopoietic compartment, evidenced by the absence of the variant in a skin fibroblast line (11). The fact that variants with such low VAF could cause severe systemic disease leading to BM failure suggested that TLR8 GOF is a cell-extrinsic disease driven by a small fraction of cells.

Recently, we reported an additional 4 patients, for a total cohort of 10, and their treatment outcomes, including the first female with TLR8 GOF (with germline p.P432L) who had onset of disease at 2 wk of age (Table 1, #1–10) (12). 8 of the 10 patients had genetic variants at the same amino acid in the antagonist binding site, p.P432, with 3 of the 6 possible amino acid substitutions at that position identified in patients with TLR8 GOF thus far. Most recently, we identified a woman in her 70s with a 20-year history of refractory neutropenia, severe infection, CD8⁺ T cell infiltration in her BM, transient large granular lymphocytes (LGL) on a peripheral blood smear, and undetectable B cells with antibody deficiency associated (14). She has a mosaic variant at the amino acid position 494 in the antagonist binding site (p.F494Y, #13 in Table 1) leading to GOF, the same position as another patient from the original cohort (p.F494L, #3 in Table 1).

We have termed this severe spectrum of disease with neutropenia and BM involvement: inflammation, neutropenia, BM failure, and lymphoproliferation caused by TLR8 (INFLTR8) (11, 12).

Since the original discovery in 2021, a spectrum of TLR8 GOF has begun to emerge (Table 1, Fig. 2) (11, 12, 13, 14, 15, 16, 17, 18), with additional clinical phenotypes including hemophagocytic lymphohistiocytosis (HLH), autoimmunity, and relatively isolated anemia. Commonalities among most patients include cytopenias, infections, general symptoms of inflammation including fever and oral ulcers, and antibody deficiency with impaired B cell maturation (Fig. 2).

In general, severe disease is indicated by refractory neutropenia and BM involvement including infiltration of T cells, associated with significant mortality necessitating allogeneic hematopoietic cell transplant (HCT) in most patients. Consistent with this, Qu et al. identified another male patient with a germline de novo variant at a residue critical for interacting with Site 1 within the dimer, p.G572D, a variant previously identified and functionally validated in our original cohort (11, 13). The patient presented with infections, refractory neutropenia, and lymphoid hyperplasia with T cell infiltrate in the BM, anemia, and what was described as autoimmune myelofibrosis of the BM, similar to other patients with INFLTR8 phenotype.

Barber et al. expanded the phenotypic spectrum by identifying a young male with HLH associated with a germline variant in the antagonist binding site (p.L433F, #11 in Table 1) (18). While distinct from prior patients, he had refractory neutropenia and T cell infiltration on BM biopsy characteristic of other patients with INFLTR8. Hemophagocytosis, without meeting criteria for HLH, was observed in one other patient in the liver (Table 1, #1); however, this is the first report of an HLH phenotype due to TLR8 GOF. The mother of this patient (Table 1, #19) also carried the same variant and had a history of autoimmune hemolytic anemia (AIHA) and prolonged B cell lymphopenia after rituximab therapy (with the maternal grandmother of unknown genotype having thrombocytopenia). This suggests the possibility that in the germline state, this variant may have differing penetrance, potentially due to expression on the X chromosome.

Distinct from the INFLTR8 phenotype, Fejtkova et al. reported identical twins with AIHA, autoinflammation, and autoimmunity including arthritis, and central nervous system (CNS) vasculitis with a maternally inherited p.G572V variant (Table 1, #14 and #15) (15). Unlike in patients with INFLTR8, neither child was neutropenic, though they did have mildly reduced CD19⁺ B cell frequencies. Their mother, who carried the variant in the heterozygous state, had a history of autoimmunity with arthritis and antiphospholipid antibody syndrome without cytopenias. Both patients responded to immunosuppressive therapy, with one brother (Twin B) requiring HCT for severe CNS vasculitis. Despite the variant having heightened sensitivity to TLR8 stimulation in a cell line model, primary cells had increased degradation of the corresponding TLR8 protein (15). Further, the p.G572V variant increased both TLR8 and TLR7 signaling. Of interest, a full-body deletion of Tlr8 in mice showed overactive Tlr7 in dendritic cells, suggesting that interactions between TLR7 and TLR8 regulate their activity (65). Thus, this unique disease phenotype in patients with the p.G572V variant with prominent autoimmunity may be driven predominantly by enhanced TLR7 activity rather than TLR8 hyperactivation. Indeed, these patients had increased levels of multiple type I IFNs more typical of TLR7 GOF than interferonopathies (15, 24, 66). This is in contrast to other patients with germline variants at the same amino acid position, p.G572D, who had more severe disease (Table 1), suggesting that there may be variable phenotypes associated with different missense variants at this amino acid position (11, 13).

Liang et al. reported a male infant presented at d24 of life with anemia with impaired erythropoiesis on BM biopsy, with a family history of hematologic malignancy and autoimmunity (Table 1, #18) (17). Whole-genome sequencing identified a p.V434L variant in the child and their unaffected mother, as well as a maternal grandmother who had pure red cell aplasia. The investigators suggested that there are intrinsic effects of TLR8 GOF on erythroid precursors leading to anemia in this case. The child’s anemia was improved with corticosteroids, but they subsequently died at a young age of unreported cause (17). Overall, this case suggests that some patients may have a more “mild” phenotype with relatively isolated anemia, although the early mortality and uncertain outcome of this patient make this challenging to know.

Finally, Skenteris et al. reported two male siblings with a maternally inherited p.A518T variant who presented with anemia, thrombocytopenia, and leukopenia without neutropenia; with recurrent infections, hypogammaglobulinemia, and splenomegaly in the first year of life (Table 1, #16–17) (16). Their mother was unaffected but had some mild changes in her B cell phenotype. The siblings were diagnosed with vasculitis, described as cerebral and polyarteritis nodosa, and received multiple therapies with remission for both brothers on anti-TNF therapy. Similar to the siblings with the p.G572V variant (15), they identified that the p.A518T variant led to lower abundance of the TLR8 protein with increased protein turnover, but increased NF-κB activation following stimulation of primary cells with a TLR8 agonist (16).

Half of identified TLR8 GOF patients have disease due to mosaic variants in TLR8 with VAFs ranging from 7 to 26%, including one female (Table 1), (11, 12, 14). Mosaicism has been detected not only in peripheral blood but is also present at similar frequencies in skin-derived fibroblasts and saliva in some patients (11, 12). This suggests that somatic mutations in TLR8 can arise early in embryonic development and that TLR8 GOF variants may not confer a selective survival advantage or disadvantage to cells of the hematopoietic compartment, including immune cells. Mosaicism additionally demonstrates that a small population of GOF cells exerts a dominant effect on the immune system disproportionate to their numerical frequency, indicating a cell-extrinsic rather than cell-intrinsic mechanism of disease.

While most patients with this X-linked disease are male, three female patients with TLR8 GOF have been identified, including one with early-onset germline disease, further demonstrating the dominant nature of TLR8 GOF (12, 14, 18). Interestingly, a prior study demonstrated that TLR8 may escape X chromosome inactivation (XCI) in human monocytes, suggesting that in females both WT and GOF TLR8 protein may be co-expressed in the same myeloid cell (56). XCI studies in an affected infant female with a germline variant (Table 1, #10) showed no skewing (12), while an older woman with a mosaic variant (Table 1, #13) had significant skewing of 95% (14), although it is unknown whether that was due to the TLR8 variant or age-acquired changes (67, 68). The final female patient (Table 1, #19) had a milder phenotype than her son who required HCT, and thus clinically may be characterized as having a phenotype of anemia with autoimmunity (18).

Of note, there are two other families with maternal transmission where females do not have a phenotype clearly related to TLR8 GOF (Table 1, #14–15, #16–17), both at positions important for Site 1 (p.G572V and p.A518T). One mother had no reported symptoms, while the other had rheumatologic diagnoses including arthritis and antiphospholipid antibody syndrome, though of uncertain relationship to her TLR8 variant. In contrast, the three affected females all had variants at the antagonist binding site. This suggests that penetrance of the clinical phenotype in females may be related to the affected region.

It is uncertain whether the predominance of males identified thus far is due to a difference in expression of the GOF allele based on X chromosome dosage. One possible explanation is that the majority of patients identified had mosaic disease identified by exome. It is technically more challenging to identify a female with similar mosaicism by exome due to differences in VAF than a male; e.g., a male with 7% of cells harboring a GOF allele will have a VAF of 7%, while a female with the same 7% cells positive for GOF would have a VAF of 3.5% due to two X chromosomes per cell. The limit of detection of mosaic variants by exome is ∼10% depending on read depth; thus, approaches including deep targeted sequencing may be important for identifying patients, including females (69). It is also possible that a WT-GOF heterodimer formation could attenuate net GOF signaling output, leading to a milder phenotype in some females, though this has not been directly demonstrated in patient cells.

The TLR8 protein is expressed by myeloid cells including neutrophils, monocytes, macrophages, and myeloid dendritic cells, without significant expression in other cell types including adaptive lymphocytes (34, 39, 40, 42, 43, 44, 70). Despite this, immunophenotyping has demonstrated the most profound effects on T and B cells, indicating that the hyperinflammation in TLR8 GOF myeloid cells has far-reaching effects (11, 12, 16). Consistent with this, patients have elevated circulating proinflammatory cytokines produced by both myeloid cells and T cells, suggesting possible myeloid-derived paracrine mediators of lymphocyte reprogramming and hyperinflammation (Table 1). As additional patients are identified with varying phenotypes of clinical disease, it may be possible to correlate genotype with immunophenotype to help identify clinical course and potential treatment strategies.

The immunologic signature of TLR8 GOF is dominated by a chronically activated and terminally differentiated T cell compartment, defects in B cell maturation, and refractory neutropenia (11, 12, 13, 14, 15, 16, 18). Common findings in the peripheral blood include an inverted CD4:CD8 ratio, a shift in the CD45RA:CD45RO ratio of CD4⁺ T cells away from naive cells, expansion of TEMRAs, and increased double-negative T cells expressing TCRαβ and/or TCRγδ (11, 12, 14, 15, 16, 18). Additional immunophenotyping in one patient also identified an increased percentage of CD4⁺ T helper 1 cells and elevated PD-1 expression on both CD4⁺ and CD8⁺ memory T cells, consistent with chronic antigen- or cytokine-driven stimulation rather than a discrete clonal event (12).

Increased TEMRAs in TLR8 GOF may directly contribute to neutropenia. TEMRAs are characterized by a CD45RA⁺CCR7⁻CD27⁻CD28⁻ surface phenotype, high expression of cytotoxic granule proteins including granzyme A/B and perforin, and the ability to potently target cell killing (71, 72). Whether T cell cytotoxicity is a primary driver of neutropenia or a secondary consequence of the broader inflammatory environment remains unresolved.

B cell abnormalities, including near-absent B cells in some cases, and hypogammaglobulinemia, are seen in many patients (11, 12, 14, 18). B cell immunophenotyping has shown significantly reduced class-switched memory B cells relative to age-matched controls (11, 12, 14, 15, 16, 18). Additional immunophenotyping has identified reduced frequency of CD24^hiCD38^hi transitional B cells and CCR6⁺ naive B cells, with reduced CD21 surface expression on naive B cells in some patients (11, 16). Two patients showed a prominent CD21^lo/neg population (11, 16), which is associated with autoimmunity, common variable immunodeficiency, splenomegaly, and cytopenias (73, 74, 75). The combination of reduced transitional B cells and reduced memory B cells implies a defect in the marrow and in peripheral maturation of lymphocytes.

For patients with BM abnormalities (Table 1), findings have been variable but frequently characterized by marked myeloid hypoplasia with lymphoid infiltrates composed of activated CD8⁺ T cells (11, 12, 13, 14, 18). In 6 cases, patient BM displayed increased numbers of LGL cells, with 2 patients diagnosed with CD8⁺ T cell LGL (T-LGL) leukemia at a relatively young age (#4 and 9 here) (11, 12, 14). In addition, some patients reported to have milder phenotypes (p.L433F, p.V434L, and p.A518T, Table 1) have had familial history of anemia and BM malignancies, suggesting there may be variable penetrance or other modifying factors of clinical disease (16, 17, 18).

TLR8 signals through the transcription factors NF-κB and IRF5 (1, 76, 77, 78). These signaling transduction pathways can lead to proinflammatory cytokine production and type I IFN production, respectively (32, 77). Most patients with TLR8 GOF, however, have evidence of heightened NF-κB activity rather than an IRF5-driven IFN signature. In addition, as TLR8 is not significantly expressed in plasmacytoid dendritic cells (pDCs), the main producers of IFNs (40, 79, 80, 81), TLR8 GOF immunopathology may be due more to NF-κB–mediated inflammation in myeloid cells rather than pDC-driven IFN signatures. Interestingly, patients with germline variants reported as TLR8 GOF but with reduced levels of the TLR8 protein, p.G572V and p.A518T, had evidence of elevated type I IFNs or IFN-stimulated genes in the peripheral blood in addition to increased NF-κB phosphorylation upon stimulation of primary cells with TLR8 agonists or at baseline, suggesting there may be some different signaling mechanisms dominating (15, 16).

Serum cytokine profiles of patients with TLR8 GOF are generally consistent with excessive NF-κB–mediated signaling in monocytes and macrophages and provide a possible paracrine link between myeloid TLR8 hyperresponsiveness and lymphocyte phenotypes. Cytokines and inflammatory markers elevated in the serum of patients include IFN-γ, IL-1β, IL-2, soluble IL-2Rα, IL-6, IL-12, IL-18, soluble Fas ligand (sFasL), BAFF, and CXCL9 (11, 12, 14, 15, 16, 18). IL-12 and IL-18 are of particular mechanistic interest as they are potent myeloid-derived drivers of CD8⁺ T cell differentiation toward effector and terminally differentiated fates, with the potential to reprogram T cells independent of TCR engagement through cytokine-driven bystander activation (82, 83, 84, 85, 86). Further, IL-12 and IL-18 in combination stimulate CD8⁺ T cells to produce IFN-γ (81, 84), which is elevated in patients along with CXCL9, an IFN-γ–induced chemokine (87, 88), suggesting heightened T cell activation (89). Lastly, elevated BAFF likely reflects reduction in B cell maturation or survival (90).

The expression of the TLR8 protein is almost exclusively limited to myeloid lineage cells. Immunophenotyping of monocytes from patients has in some cases demonstrated small increases in overall frequency in the peripheral blood, with variable changes in nonclassical CD14^loCD16⁺ monocytes, but without consistent phenotypic changes in monocytes by flow cytometry (11, 15, 16, 17).

Identification of patients with genetic mosaicism has led to an interesting question: how do 7% of cells harboring a TLR8 GOF variant cause disease? While analysis of peripheral blood monocytes has not identified enrichment of the variant in these cells in cases of mosaicism (11), one possibility is that GOF myeloid progenitors are selectively expanded in a tissue-specific inflammatory environment they themselves help create; chronic NF-κB activation promotes myeloid cell survival and proliferation, potentially increasing the effective GOF fraction over time. Testing this will require careful analysis of myeloid cell populations at sites of inflammation over time, e.g., BM and lymphoid organs, rather than circulating cells. Another possibility is that myeloid cells occupy strategic environments, particularly in the BM, where even a small number of hyperactivated cells can disproportionately shape the local cytokine milieu encountered by other developing hematopoietic progenitors and trafficking lymphocytes.

Neutropenia is a hallmark of TLR8 GOF with BM involvement/INFLTR8 phenotype, but the mechanism(s) responsible remain unknown. Some clues may come from similarities to findings in patients with T-LGL leukemia. Overlapping findings between TLR8 GOF and T-LGL leukemia include expanded CD8⁺ T cells with a terminally differentiated phenotype (clonal in T-LGL leukemia and oligoclonal or nonclonal in TLR8 GOF), elevated cytokines including IFN-γ, which is associated with immune-mediated BM failure, prominent neutropenia, and in some patients pure red cell aplasia (91, 92, 93). In studies of T-LGL leukemia, proposed mechanisms of neutropenia include direct cytotoxic effects of the T-LGL cells on neutrophils, elevated circulating levels of sFasL, and autoimmune neutropenia (94, 95, 96), suggesting potential pathogenic mechanisms in TLR8 GOF.

The adaptive immune profiles, cytokines, and myeloid hyperresponsiveness in TLR8 GOF combined with similarities to T-LGL leukemia suggest potential models of neutropenia due to heightened TLR8 signaling through NF-κB in myeloid cells, production of proinflammatory cytokines leading to CD8⁺ T cell activation, and a positive feedback loop. We hypothesize that neutropenia in patients is due to depletion of neutrophils through Fas ligand–mediated apoptosis, direct cytotoxic effects of terminally differentiated CD8⁺ T cells on hematopoietic progenitors or neutrophils, and/or effects of the hyperinflammatory environment (e.g., IFN-γ) (Fig. 3).

While detecting germline variants is now relatively standard, somatic variants with low VAF may escape detection by standard clinical testing; conventional sequencing techniques including exome sequencing typically have read depths of 100–200×, while genome sequencing has much lower coverage of ∼20–30× at any particular nucleotide (97). Thus, identifying mosaicism requires careful review of sequencing data or alternative approaches including targeted deep sequencing, with several clinically available deep-sequencing tests becoming more widely available (69). Validation of the findings using an orthogonal approach such as droplet digital PCR from multiple tissue samples may also be required to confirm sequencing findings.

Once a potentially pathogenic variant is identified, functional testing to confirm its pathogenicity is necessary to begin delineating disease mechanisms. Most variants have undergone functional testing using an in vitro protocol in which ectopic expression of TLR8 variants in a cell line lacking endogenous TLR8 and expressing an NF-κB reporter (TLR8-null HEK-Blue Null1 cells) was used. By stimulating these cells with increasing doses of TLR8-specific agonists, such as TL8-506, activity can be compared with controls (11, 12). The GOF phenotype may not be obvious at lower or higher doses; thus, it is important to test a range of agonists. It is also possible that such testing may reveal a strength of signaling that correlates with disease. For example, we demonstrated that the p.V434L variant, associated with relatively isolated anemia, was only GOF with strong stimulation as compared to other variants associated with the more severe INFLTR8 phenotype (12).

Testing of primary patient cells is important for determining functional effects of TLR8 GOF in a relevant cell type. In patients with an INFLTR8 phenotype with mosaicism, functional studies demonstrated that a small percentage of monocytes (5–6%) had heightened phosphorylation of NF-κB in response to a low dose of TLR8 agonist, suggesting effects due to the presence of the variant in a subpopulation of cells (11). Disease due to genetic mosaicism presents challenges in assays of primary cells as it is not possible to delineate cells expressing WT versus GOF TLR8. To overcome this limitation, we utilized iPSCs derived from patients with mosaicism and differentiated cells into macrophages or neutrophils expressing either WT or GOF variant TLR8, thereby offering a mechanism to directly test the effects of variation in TLR8 only. In this model, cells demonstrated hyperphosphorylation of NF-κB upon stimulation with a TLR8 ligand, with corresponding elevated cytokine production (11, 12). The major advantage of this system is the ability to generate isogenic pairs—WT and GOF iPSC lines from the same individual—enabling mechanistic comparisons unconfounded by interindividual genetic variation.

Distinct from other variants, TLR8 GOF due to p.A518T and p.G572V associated with a phenotype of anemia and autoimmunity (Table 1) displayed GOF activity in primary cells and cell line systems, but at the same time led to reduced levels of the TLR8 protein (15, 16). Stimulation of primary monocytes from patients with both variants with TLR8 ligands demonstrated increased production of inflammatory cytokines compared with healthy controls (15, 16). Interestingly, stimulation with a TLR7 ligand in a patient with a p.G572V variant also showed heightened response to the TLR7-specific agonist imiquimod, suggesting an imbalance of TLR7/TLR8 signaling (15). The expression of variants in HEK cells followed by treatment with an inhibitor of new protein synthesis showed that both p.A518T and p.G572V protein variants degraded faster compared with WT (15, 16). Skenteris et al. further compared both variants and showed increased proteasomal degradation of the p.A518T, but not p.G572V, variant, as partially responsible for protein loss (16).

Together, testing of identified variants thus far suggests that demonstrating hyperresponsiveness to TLR8 signaling through cell line models, primary cells, and/or iPSC-derived myeloid cells can identify TLR8 GOF through measurement of NF-κB activity and/or cytokine production. At the same time, additional nuances including alterations of TLR8 protein stability and possibly TLR7 signaling may modify signaling pathways, and it will be interesting to determine how they may affect the clinical phenotypes of patients and potential treatment options.

There are no approved targeted therapies for TLR8 GOF. In the most severe cases, patients have tended to have an overall poor response with immunosuppressive treatment. Therapies that have not produced consistent clinical benefit in patients with INFLTR8 and neutropenia with BM involvement include G/GM-CSF, corticosteroids, azathioprine, thalidomide, mycophenolate mofetil, sirolimus, ruxolitinib, and rituximab (11, 12, 13, 17). Notably, neutrophil counts are refractory to nearly all therapies and neutropenia is a major driver of morbidity in INFLTR8 due to overwhelming opportunistic infections. Of the 13 patients with INFLTR8 phenotype, 4 have died from infection or BM failure (11, 12). Allogeneic HCT represents the only treatment demonstrated to produce durable remission, with six patients benefiting from transplantation. However, 2 of the 8 transplanted patients died from transplant-related complications, reflecting the severity of preexisting disease at the time of conditioning (12).

Most patients with TLR8 GOF have significant infection risk. Supportive care is a critical component of TLR8 GOF management given the combined immunodeficiency driven by neutropenia and hypogammaglobulinemia. IVIg replacement therapy is used to reduce infectious burden in patients with low immunoglobulin levels, and has been administered alongside other therapies (11, 12, 14). Prophylactic antimicrobial therapy is similarly important given the high rates of serious bacterial and fungal infections seen across patients (11, 12). Although growth factors (G-CSF and GM-CSF) have been used in attempts of improving neutrophil counts, it has not produced consistent or durable benefit in most patients, with the notable exception of an elderly female with mosaic disease who has stabilized with growth factor with prednisone and a JAK inhibitor (11, 12, 14).

Allogeneic HCT is currently the only intervention demonstrated to produce durable disease remission in patients with INFLTR8 phenotype (neutropenia and BM involvement) and has led to sustained correction of cytopenia and immune dysregulation in 5 patients (12). The progressive nature of BM failure in INFLTR8 argues for early referral before irreversible hematopoietic damage or infectious burden accumulates. While the optimal approach to HCT is uncertain, complications in patients thus far have included high rates of cytopenias after transplant and graft-versus host disease. The mosaic nature of the disease suggests that 100% donor chimerism is required for remission, particularly in the myeloid compartment, with the caveat that it is possible that residual tissue-resident myeloid cells expressing TLR8 GOF may contribute to posttransplant complications despite the appearance of complete chimerism in the periphery. In addition, our experience suggests avoiding high-dose busulfan due to concern for therapy-related toxicity (12). A final consideration is to use agents to reduce inflammation prior to transplant, including JAK inhibitors used in 2 patients and emapalumab (anti-IFN-γ monoclonal antibody) in another patient (12, 18).

The finding that hyperinflammatory cytokines are consistently elevated in patients with TLR8 GOF provides a strong rationale for cytokine-directed therapy. Two brothers with cytopenias, infections, and vasculitis associated with a p.A518T variant had a positive response to anti-TNF therapy in combination with IVIg (16). This suggests that in some patients, blocking a single cytokine can produce meaningful clinical benefit possibly related to severity of disease and clinical symptoms (16).

As noted above, emapalumab was successfully used in a patient with an HLH phenotype due to a germline p.L433F TLR8 variant (Table 1, #11) as a “bridge” to successful allogeneic HCT, reflecting the rationale for targeting elevated IFN-γ levels seen across TLR8 GOF patients (18). JAK inhibitors, which broadly suppress cytokine signaling downstream of multiple receptors, have been used in a subset of patients without consistent benefit (12), though this may reflect insufficient drug exposure, late initiation, or irreversibility of BM damage at the time of treatment. However, such agents may have an important role in the pretransplant phase to reduce inflammation prior to HCT. Of note, the elderly female we recently reported with a p.F494Y variant has demonstrated sustained clinical benefit from a combination that included JAK inhibition, suggesting that this approach may be effective in some patients, particularly those who are not transplant candidates (14).

CU-CPT9a is a small TLR8-specific antagonist that binds to the inactive TLR8 dimer (98). It has been shown to potently suppress WT TLR8 signaling but fails to inhibit p.P432L TLR8, the most common GOF variant (12). However, we did observe a reduction in inflammatory cytokine production in iPSC-derived macrophages harboring the p.P432L variant following TLR8 stimulation when pretreated with chloroquine, a drug that inhibits endosomal TLR signaling, or CPG-52364, a small-molecule inhibitor of TLR7/8 (12). These findings suggest that there may be mechanisms to inhibit TLR8 signaling in patients, although there was unclear benefit of hydroxychloroquine treatment in two patients (Table 1) (12, 15).

Resistance to antagonism associated with many of the variants has significant therapeutic implications, which may mean that direct TLR8 antagonism will require variant-specific or alternative approaches. Oligonucleotide antagonists, such as 2′-OMe-guanosine RNA fragments that bind to the antagonist binding site, additionally highlight different variant resistance profiles. The Gantier lab demonstrated that the p.G572D TLR8 GOF variant was able to be suppressed with 2′-OMe–modified guanosine RNA fragments, yet the p.F494L variant was not, indicating again that the resistance landscape varies by variant, target site, and the antagonist (61). Several small-molecule dual TLR7/8 antagonists, for example, E6742, enpatoran, and afimetoran, are undergoing clinical trials for inflammatory disease and may soon reveal more immediate translational paths for TLR8 GOF (99, 100, 101). Variants that remain sensitive to antagonism, such as those affecting p.G572, may be amenable to this approach.

TLR8 signaling utilizes the MyD88-IRAK4 Myddosome (1, 76, 102). Inhibiting IRAK4 or other Myddosome-associated signal transducers therefore represents a therapeutic strategy that bypasses the problem of variant-specific antagonist resistance. Several IRAK4 inhibitors are currently in early clinical trials for inflammatory conditions including rheumatoid arthritis and myeloid malignancies, and preclinical data support their efficacy in suppressing TLR7/8-dependent and NF-κB–mediated cytokine production (103, 104, 105). Supporting this approach, IRAK4 inhibition was shown to suppress cytokine production in vitro in cells carrying activating UNC93B1 variants that exacerbate TLR7/8 activity (106). Their utility in TLR8 GOF has not yet been tested and may have significant off-target effects given that it would likely be required long term.

The identification of TLR8 GOF disease has repositioned a long-neglected receptor as an important mediator of myeloid immune function. INFLTR8 illustrates how a defect confined to the innate compartment, often to a minor mosaic population of myeloid cells, can remodel the entire immune landscape to drive CD8⁺ T cell reprogramming, neutropenia, and, at its most severe, BM failure. In the spectrum of TLR8-associated disease, the recurring theme across the molecular, cellular, and clinical evidence is that TLR8 GOF is fundamentally a cell-extrinsic, cytokine-driven disease in which a small number of hyperactivated myeloid cells dictate the phenotype and functional capacity of immune subsets that do not themselves express TLR8.

Identifying therapies for affected patients will require further mechanistic understanding, including clarifying how myeloid hyperactivation leads to shifts in immune cell frequency and phenotype that define TLR8 GOF and INFLTR8. Specifically, discerning the key inflammatory drivers leading to CD8⁺ T cell dysfunction, mechanisms of neutropenia and B cell lymphopenia, and triggers of TLR8 activation including significance of variants in the antagonist binding site will be important. Such questions will require model systems that capture the disease’s defining feature: the interaction between a population of GOF myeloid cells and the adaptive immune cells they reprogram. There are direct therapeutic implications, including identifying patients who should proceed to early transplant and who may benefit from “bridging” therapies including anticytokine agents.

Additional discovery of genotype-, site-, and potentially sex-specific effects of TLR8 GOF variants may help to predict disease severity and guide early intervention. A common polymorphism in TLR8, rs3764880 (p.Met1Val), leads to a reduction in the primary TLR8 transcript and a shift toward a longer isoform, with studies suggesting reduced inflammation (107, 108). Given that the minor allele frequency of this variant ranges from ∼20 to 80% (109, 110), it will be interesting to determine whether this polymorphism or other cis-regulatory elements modify the severity or penetrance of TLR8 GOF. The heterogeneity of TLR8 GOF variants’ response to antagonism and patients’ refractoriness against some therapies argues for variant- or patient-specific therapy matched to mechanism rather than a single standardized approach.

Finally, TLR8 GOF serves as a natural model for understanding innate immune dysregulation by which localized myeloid hyperactivation reshapes systemic immunity. As one of the growing number of genetic errors of immunity caused by a single innate sensor, TLR8 GOF offers a valuable framework to inform the broader field of TLR-driven inflammatory disease.

This work was supported by the St. Louis Children’s Hospital Foundation, the Jeffrey Modell Diagnostic and Research Center for Primary Immunodeficiencies at St. Louis Children’s Hospital and Washington University, the CDI Center for Pediatric Immunology at Washington University, and the National Institutes of Health (NIH) grant 5P01AI186771 (to M.A. Cooper). H.M. Gil is supported by NIH grant T32AR007279. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. This manuscript is the result of funding in whole or in part by the NIH. It is subject to the NIH Public Access Policy. Through acceptance of this federal funding, the NIH has been given a right to make this manuscript publicly available in PubMed Central upon the Official Date of Publication, as defined by the NIH.

Author contributions: Hwi M Gil: conceptualization, visualization, and writing—original draft, review, and editing. Megan A Cooper: conceptualization, supervision, and writing—review and editing.