Endosomal Localization of TLR8 Confers Distinctive Proteolytic Processing on Human Myeloid Cells

Discrimination between self and nonself by the innate immune system is crucial for swift elimination of infectious microbes, as well as protection against autoimmune disorders (1, 2). The compartmentalized pattern recognition receptors, including TLR3, TLR7, TLR8, and TLR9, participate in the recognition of extracellular microbial nucleic acids and transmission of innate immune signaling (3). The nucleic acid–sensing TLRs localize to the endosomal compartments (4, 5), which prevents them from responding to self nucleic acids in steady-states. The endoplasmic reticulum (ER)-resident multispan transmembrane protein UNC93B1 is indispensable for intracellular localization and signaling of these TLRs (6–10). UNC93B1 associates with TLR3, TLR7, TLR8, and TLR9 through transmembrane domains in the ER and promotes intracellular trafficking of those TLRs from the ER to the Golgi. However, the destination of each TLR is regulated by distinct determinants within TLRs (10–13).

TLR7, TLR8, and TLR9 form a subfamily of proteins that shares structural features (14, 15). Their ectodomains (ECDs) consist of 26 leucine-rich repeats (LRRs) with a large insertion loop between LRR14 and LRR15 and N- and C-terminal flanking region, LRRNT and LRRCT (16). Ligand binding to TLR–ECD induces receptor dimerization, allowing access of adaptor molecule MyD88 to the cytoplasmic Toll–IL-1R (TIR) domains (17). TLR7 and TLR8 recognize ssRNA and synthetic imidazoquinoline derivatives (18–21), whereas TLR9 recognizes CpG-containing DNA (22, 23). Accumulating evidence indicates that TLR7 and TLR9 undergo proteolytic processing in the endolysosomes of macrophages and plasmacytoid dendritic cells (pDCs) to generate functional receptors (24–29). The pH-dependent endosomal cathepsins, as well as a cysteine lysosomal protease asparaginyl endopeptidase (AEP), participate in mouse (m)/human (h)TLR9 and mTLR7 cleavage at the loop region, which is necessary for nucleic acid sensing. Hipp et al. (30) also demonstrated that hTLR7 processing is mediated with furin-like proprotein convertase. Although the truncated receptors appear to be signaling competent (26, 29), the N-terminal fragment contributes to full activation of the receptors via association with the C-terminal half (31, 32).

In humans, TLR8 is expressed in myeloid cells, including monocytes, neutrophils, macrophages, and myeloid dendritic cells (DCs), and in regulatory T cells (33–36). TLR8 recognizes viral GU-rich ssRNA and endogenous RNA, such as microRNAs within exosomes, leading to the production of proinflammatory cytokines but not type I IFNs (37, 38). Recent structural analysis of hTLR8 ECD and its ligand complex showed that TLR8 is cleaved as well, and both the N- and C-terminal halves are engaged in ligand recognition (39). However, whether the proteolytic cleavage of endogenous TLR8 actually occurs in human primary cells and how TLR8 undergoes processing remain obscure. In this study, we investigated the cleavage of TLR8 and its requirement for ligand recognition in human primary cells, including monocytes and monocyte-derived macrophages.

HEK293 cells were maintained in DMEM low glucose (Invitrogen) supplemented with 10% heat-inactivated FCS (BioSource International) and antibiotics. HEK293FT cells were maintained in DMEM high glucose supplemented with 0.1 mM nonessential amino acids, 10% heat-inactivated FCS, and antibiotics. RAW264.7 cells and THP-1 cells were maintained in RPMI 1640 (Invitrogen) supplemented with 10% heat-inactivated FCS, 55 μM 2-ME (for THP-1 cells), and antibiotics. Human monocytes and B cells were isolated from PBMCs obtained from healthy individuals with a magnetic cell sorting system using anti-CD14–coated and anti-CD19–coated MicroBeads (Miltenyi Biotec, Gladbach, Germany), respectively. Purity was checked routinely by FACS and was >95%. Monocyte-derived macrophages were differentiated from CD14⁺ monocytes by culturing with 20 ng/ml recombinant hGM-CSF (PeproTech) for 6 d. Anti-FLAG M2 mAb, anti-FLAG polyclonal Ab (pAb), brefeldin A, z-FA-FMK, DC1 (3,3′-(5-Indolyl methylene)bis(4-hydroxycoumarin)), and LPS (Escherichia coli 0111:B4) were purchased from Sigma-Aldrich. In addition, the following Abs were used in this study: PE mIgG1, PE anti-human CD80 mAb, and PE anti-human CD19 mAb (all from eBioscience); FITC mIgG2b, PE anti-human CD14 mAb, and FITC anti-human CD68 mAb (all from BioLegend); Alexa Fluor– or HRP-conjugated secondary Abs (all from Invitrogen); anti-early endosome Ag 1 rabbit mAb (Cell Signaling Technology Japan); anti-GM130 mAb (BD Transduction Laboratories); anti-calnexin pAb (Stressgen; Victoria, BC, Canada); anti-p115 pAb, anti-mannose 6 phosphate receptor (MPR) pAb, anti-MPR mAb, and anti-calnexin mAb (all from Abcam, Cambridge, U.K.), anti–Lamp-1 mAb and anti–tubulin-α mAb (BioLegend); and anti-hTLR8 rabbit mAb (CST Japan). Affinity-purified rabbit pAb against hTLR8 cytoplasmic region was generated by MBL. LysoTracker Red was purchased from Invitrogen. CL075 was from InvivoGen. ssRNA40 (5′-GCCCGUCUGUUGUGUGACUC-3′, a 20-mer phosphorothioate protected ssRNA oligonucleotide) and biotinylated ssRNA40 were synthesized by Hokkaido System Science (Sapporo, Japan).

cDNAs for hTLR7 and hTLR8 were cloned in our laboratory by RT-PCR from the mRNA of monocyte-derived macrophages and were ligated into the cloning site of the expression vector, pEF-BOS, which was provided by Dr. S. Nagata (Kyoto University). The FLAG tag was inserted into the C terminus of pEF-BOS expression vectors for hTLR7 and hTLR8. The C-terminal FLAG-tagged TLR8 mutant lacking the flexible loop between LRR14 and LRR15 (TLR8 lacking the flexible loop between LRR14 and LRR15 [TLR8Δloop]) and the mutant lacking LRR1–14 and the flexible loop (TLR8 C-terminal fragment [TLR8-C]) were generated by PCR with KOD-Plus DNA polymerase (TOYOBO) using specific primers (forward primer: 5′-TATGGAAAAGCCTTAGATTTAAGCC-3′, reverse primer: 5′-ATAACTCTGCCGGGTATCTTTTACC-3′ for TLR8Δloop; and forward primer: 5′-TATGGAAAAGCCTTAGATTTAAGCC-3′, reverse primer: 5′-TTTGCCCACCGTTTGGGGAACTTCC-3′ for TLR8-C), as described (11). The C-terminal FLAG-tagged TLR8 mutant, R467A/R470A/R472A/R473A, was generated by site-directed mutagenesis using specific primers (forward primer: 5′-GCAGCCTCAACAGATTTTGAGTTTGACCC-3′, reverse primer: 5′-TTTCGCGATATGAGCTTGAAAAGAGGAACTATTTGC-3′). The TLR8 N-terminal fragment (LRR-NT+LRR1–14) (TLR8-N) was generated by PCR using specific primers (forward primer: 5′-CTCGAGCCACCATGAAGGAGTCATCTTTGC-3′ and reverse primer: 5′-AAAGCGGCCGCTTAATAACTCTGCCGGGTATC-3′). pEFBOS/hTLR8-FLAG-IRES-Puromycin and pEFBOS/hTLR8Δloop-FLAG-IRES-Puromycin were made in our laboratory and used for stable expression of hTLR8 and hTLR8Δloop, respectively, in RAW294.7 cells. Plasmids for human UNC93B1 (pMD2/UNC93B1) and hTLR9 (pBluescript II/TLR9) were provided by Dr. K. Miyake (The University of Tokyo) and Dr. S. Akira (Osaka University), respectively. The HA tag and FLAG tag were inserted into the C terminus of the pEF-BOS expression vector for human UNC93B1 and hTLR9, respectively.

HEK293 cells (3 × 10⁴ cells/well), cultured in 96-well plates, were transfected with the indicated plasmid together with the reporter plasmid and an internal control vector, phRL-TK (Promega), using FuGENE HD (Roche). The reporter plasmid containing the ELAM-1 promoter was constructed in our laboratory. Twenty-four hours after transfection, cells were stimulated with CL075 and ssRNA40 complexed to DOTAP (Roche). The cells were collected 12 h after stimulation and lysed. Firefly and Renilla luciferase activities were determined using a dual-luciferase reporter assay kit (Promega). The firefly luciferase activity was normalized to the Renilla luciferase activity and expressed as the fold induction relative to the activity in unstimulated vector-transfected cells. All assays were performed in triplicate.

Small interfering RNA (siRNA) duplexes (hTLR8: #s27921; negative control: #AM4635) were obtained from Ambion-Applied Biosystems. Human monocytes (5 × 10⁵/ml) were cultured in 24-well plates with 20 ng/ml hGM-CSF. At day 4, cells were transfected with 30 pmol control or TLR8 siRNA using Lipofectamine RNAiMAX (Invitrogen). Forty-eight hours after transfection, cells were washed once and stimulated with medium or DOTAP alone, 2.5 μg/ml CL075, and ssRNA40 complexed to DOTAP for 3 h. Cells were collected by centrifugation at 1500 rpm for 3 min, and total RNA was extracted using TRIzol reagent (Invitrogen). Knockdown of hTLR8 was confirmed 48 h after siRNA transfection by quantitative PCR using specific primers (Supplemental Table I) and Western blotting with anti–TLR8-N mAb. For knockdown of TLR8 in THP-1 cells, cells were transfected with the Amaxa Cell Line Nucleofector kit V (Lonza) and 30 pmol control or TLR8 siRNA, according to the manufacturer’s instructions. Nucleofection was performed twice every 24 h. Forty-eight hours postnucleofection, cells were treated with 10 ng/ml IFN-γ for 15 h and stimulated with the indicated TLR8 ligands. Experiments were repeated three times for confirmation of the results.

Total RNA was extracted using TRIzol reagent and reverse transcribed using the high-capacity cDNA Reverse Transcription kit (Applied Biosystems) and random primers, according to the manufacturer’s instructions. Quantitative PCR was performed using the indicated primers (Supplemental Table I) and the StepOne Real-Time PCR System (Applied Biosystems).

Monocyte-derived macrophages (5 × 10⁵ /ml) were pretreated with DC1 (20 μM) or DMSO for 4 h and then were stimulated with CL075 (2.5 μg/ml), ssRNA40 complexed to DOTAP (2.5 μg/ml), or LPS (1 μg/ml) or left untreated in the presence of inhibitors for another 24 h. To examine stepwise processing of TLR8-N, monocytes were treated with 10 μM z-FA-FMK in the presence of 20 ng/ml recombinant hGM-CSF for 24 or 48 h and then stimulated with indicated ligands for 24 h. IL-12p40 in culture supernatants was measured by ELISA (R&D Systems).

Monocytes and monocyte-derived macrophages that were left untreated or stimulated with 2.5 μg/ml CL075 for 24 h were incubated with FcR Blocking Reagent, human (Miltenyi Biotec) in FACS buffer (PBS containing 5% FCS) for 5 min at 4°C and then incubated with PE mIgG1 or PE anti-human CD80 mAb (1:200) for 30 min at 4°C in the dark. For CD68 staining, cells were fixed and permeabilized by incubating with Fixation/Permeabilization Solution (BD Bioscience) for 20 min at 4°C. Cells were washed twice with 1× BD Perm/Wash buffer and incubated with FITC mIgG2b or FITC anti-human CD68 mAb (1:200) for 30 min at 4°C in the presence of mIgG2b (1:100). After washing twice with FACS buffer (for CD80 staining) or 1× BD Perm/Wash buffer (for CD68 staining), cells were analyzed on a FACSCalibur (BD Bioscience).

RAW264.7 cells stably expressing hTLR8-FLAG or TLR8Δloop-FLAG cultured in 10-cm dishes were lysed in 1% Nonidet P-40 lysis buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 10 mM EDTA, 5 mM Na₃VO₄, 30 mM NaF, 2 mM PMSF, and a protease inhibitor mixture) for 10 min at 4°C. Lysates were clarified by centrifugation at 15,000 rpm for 15 min, precleared with protein G–Sepharose (GE Healthcare, Buckinghamshire, U.K.), and incubated with anti-FLAG mAb. The immunoprecipitates were recovered by incubation with protein G–Sepharose for 1 h at 4°C, washed three times with 1% Nonidet P-40 lysis-washing buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 10 mM EDTA), and resuspended in denaturing buffer. Samples were analyzed by SDS-PAGE (7.5% gel) under reducing conditions, followed by immunoblotting with anti-FLAG pAb and anti–TLR8-N mAb.

Monocyte-derived macrophages (5.0 × 10⁵) were lysed in 150 μl lysis buffer. After centrifugation, the supernatants were aliquoted (50 μl each) and incubated with buffer alone, 1 μl Endoglycosidase H (Roche), or 2 μl N-glycosidase F (Roche) for 30 min at 37°C. Samples were mixed with denaturing buffer and analyzed by SDS-PAGE under reducing conditions, followed by immunoblotting with anti–TLR8-N mAb.

Monocyte-derived macrophages (2.5 × 10⁵ /sample) were lysed in lysis buffer, as described above. After centrifugation at 15,000 rpm for 15 min, supernatants were incubated with 2.5 μg ssRNA, biotinylated ssRNA, or 0.145 μg biotin for 1 h at 4°C. Streptavidin–Sepharose suspended in 1% BSA washing buffer was added to the reaction mixtures and incubated for 1 h at 4°C. After centrifugation, streptavidin beads were washed three times with washing buffer and resuspended in denaturing buffer for 5 min at 95°C. Samples were analyzed by SDS-PAGE under reducing conditions, followed by immunoblotting with anti–TLR8-N mAb.

HEK293 cells (2.0 × 10⁵ cells/well) were plated onto poly-l-lysine–coated micro coverglasses (BD Biosciences) in a 24-well plate. The following day, cells were transfected with the indicated plasmids using FuGENE HD. Twenty-four hours after transfection, cells were washed twice with PBS, fixed with 4% paraformaldehyde for 15 min, and permeabilized with PBS containing 100 μg/ml digitonin and 1% BSA. In the case of human macrophages, fixed cells (2.5 × 10⁴ cells/well) were permeabilized with PBS containing 0.1% Triton X-100 and 2% BSA for 15 min. Fixed cells were blocked in PBS containing 1% BSA and labeled with the indicated primary Abs (2–10 μg/ml) for 60 min at room temperature. Alexa Fluor–conjugated secondary Abs (1:400) were used to visualize the staining of the primary Abs. After mounting with ProLong Gold with DAPI (Molecular Probes), cells were visualized at ×63 magnification with an LSM510 META microscope (Zeiss, Jena, Germany).

To examine the requirement of proteolytic processing in ligand recognition and signaling by hTLR8, C-terminal FLAG-tagged wild-type TLR8 and truncated mutant forms were provided. TLR8Δloop lacks the insertion loop between LRR14 and LRR15. TLR8-C represents a deletion mutant lacking LRR1–14 and the insertion loop. These were transiently expressed in HEK293 cells and stimulated with a synthetic small molecule (CL075) and ssRNA40 complexed to DOTAP. Wild-type TLR8 was expressed at the expected molecular mass (∼150 kDa) and activated NF-κB in response to CL075, but not to ssRNA40, whereas TLR8Δloop failed to respond to both ligands (Fig. 1A, 1B). Cleavage products of TLR8 were undetectable in HEK293 cell lysates in either type of TLR8 expression (Fig. 1B). In addition, TLR8-C did not activate NF-κB in response to CL075 and ssRNA40 (Fig. 1A, 1B).

UNC93B1 physically associates with hTLR8 and regulates intracellular trafficking and signaling of TLR8 (10). When UNC93B1 was coexpressed with wild-type TLR8 in HEK293 cells, CL075-induced TLR8-mediated NF-κB activation was greatly increased concomitantly with the appearance of the N- and C-terminal halves of TLR8, but no response to ssRNA40 was observed (Fig. 1C, 1D). The molecular mass of the C-terminal half of TLR8 was almost the same as that of TLR8-C. In contrast, TLR8Δloop that remained uncleaved could not respond to CL075 or ssRNA40 even though it was coexpressed with UNC93B1 (Fig. 1C, 1D). Again, the truncated C-terminal half of TLR8 was unable to act as an RNA-sensing receptor. These results suggest that the loop region between LRR14 and LRR15 is crucial for ligand-induced TLR8-mediated signaling, as well as proteolytic processing of TLR8. UNC93B1 promoted proteolytic cleavage of TLR8 by facilitating ER exit of TLR8. In HEK293 cells transiently expressing TLR8, small amounts of cleaved TLR8 molecules appear to participate in the recognition of CL075 (Fig. 1A). Although the reason why ssRNA40 could not activate TLR8 in HEK293 cells ectopically expressing UNC93B1 is unclear, one possible interpretation is that oligomerization of multiple TLR8 molecules is required for ssRNA40-induced NF-κB activation. Notably, overexpressed wild-type and mutant TLR8, especially TLR8-C, activate NF-κB in a ligand-independent manner only when coexpressed with UNC93B1 (Fig. 1C, Supplemental Fig. 1), suggesting that transient overexpression and trafficking may allow TLR8–TIR domains to access each other, leading to activation of downstream signaling.

THP-1 cells expressed TLR8 mRNA, which was upregulated by stimulation with IFN-γ (Fig. 2A) (40). To examine the structure of endogenous TLR8, we generated a pAb that recognizes the hTLR8 C-terminal peptides. The anti–TLR8-C pAb specifically recognized the full-length and C-terminal half of TLR8 but not N-terminal half of TLR8, hTLR7, or hTLR9 (Supplemental Fig. 2). In the IFN-γ–treated THP-1 cells, both full-length and C-terminal half of TLR8 proteins were detected by Western blotting with anti–TLR8-C Ab, indicating that endogenous TLR8 undergoes proteolytic processing (Fig. 2B). IFN-γ−treated THP-1 cells induced IL-12p40 mRNA expression in response to CL075, as well as ssRNA40, which was abolished by TLR8 knockdown (Fig. 2C).

We investigated structural features of endogenous TLR8 in human primary cells, including monocytes and monocyte-derived macrophages. CD14⁺ monocytes were successfully differentiated into macrophages after GM-CSF treatment for 6 d, in which the macrophage marker CD68 was greatly induced and the costimulatory molecule CD80 was upregulated by stimulation with CL075 (Supplemental Fig. 3). Immunoblotting with anti–TLR8-N and anti–TLR8-C Abs under reducing or nonreducing conditions clearly showed that TLR8 underwent proteolytic processing, and cleaved forms of TLR8 were predominant in monocytes and monocyte-derived macrophages (Fig. 3A). The TLR8 N-terminal halves consisted of two bands with a molecular mass ∼100 and ∼90 kDa. The C-terminal half was detected as a single band with a molecular mass ∼90 kDa (Fig. 3A). None of the bands corresponding to TLR8 was observed in B cells, confirming the specificity of the TLR8 Abs used.

Upon stimulation with CL075 or ssRNA40 complexed to DOTAP, monocyte-derived macrophages produced a high level of IL-12p40 and some IL-6 and TNF-α, but their expression varied among different donors/individuals (Fig. 3B). Human monocyte-derived macrophages expressed low levels of TLR7; therefore, we examined whether the response to CL075 and ssRNA40 depended on TLR8 by knockdown analysis in macrophages. CL075- and ssRNA40-induced IL-12p40 mRNA expression was significantly reduced when TLR8 expression was knocked down at both the mRNA and protein levels (Fig. 3C). Notably, the expression and processing of TLR8 were mostly unaltered during differentiation from CD14⁺ monocytes to macrophages, with the exception of day-1 monocytes after GM-CSF treatment: full-length TLR8 disappeared, and the lower band of TLR8-N was detected primarily (Fig. 4A). The response of TLR8 to ssRNA40 was unchanged during differentiation (Fig. 4B). Absence of full-length TLR8 at day 1 upon differentiation of monocytes with GM-CSF was observed consistently, irrelevant of the donor. These results suggest that the cleaved form of TLR8 is a functional receptor in human primary cells.

Two TLR8-Ns with different molecular masses were detected in monocyte and macrophage lysates; thus, we examined the proteases involved in TLR8 cleavage in human primary cells using protease inhibitors. Treatment of macrophages with z-FA-FMK, a cysteine protease inhibitor that blocks cathepsin proteolytic activity, failed to reduce both TLR8 cleavage and response to TLR8 agonists, probably because the cleaved form of TLR8 was abundant in macrophages (data not shown). When monocytes were differentiated into macrophages with GM-CSF in the presence of z-FA-FMK, an ∼100-kDa upper band of TLR8-N accumulated from days 1 to 3, compared with DMSO-treated monocyte differentiation that mainly contained the lower band of TLR8-N (Fig. 5A, upper panels). This suggests that the upper band of TLR8-N is a premature form, and cysteine proteases, such as the cathepsin family, participate in further processing to generate the mature form of TLR8-N. Correlatively, when day-1 and day-2 monocytes were stimulated with CL075 or ssRNA40 complexed to DOTAP for 24 h, IL-12p40 production was decreased in z-FA-FMK–treated monocytes compared with DMSO-treated cells (Fig. 5A, lower panel). Considering that TLR7 was barely expressed in day- 1 to day-3 differentiated monocytes, involvement of TLR7 in CL075-induced IL-12p40 production is minimal, at best, under these experimental conditions. LPS-induced IL-12p40 production by these differentiated monocytes was unaltered in the presence of z-FA-FMK (data not shown).

A recent study demonstrated that furin-like proprotein convertase participates in processing of hTLR7 (30). We assessed the role of furin-like proprotein convertase in hTLR8 processing. Inhibition of furin-like proprotein convertase using DC1 reduced the response of macrophage TLR8 to CL075 and ssRNA40 but not to LPS (Fig. 5B). The cleaved TLR8-C and two TLR8-Ns, especially the upper band of TLR8-N, were reduced in DC1-treated macrophages compared with DMSO-treated cells (Fig. 5B, data not shown). Because a potential furin-like proprotein convertase–recognition site is present in the flexible loop between LRR14 and LRR15 of hTLR8, we made TLR8 mutant R467A/R470A/R472A/R473A in which the arginine residues in the potential furin-like proprotein convertase–recognition site in the flexible loop were substituted with alanine. This TLR8 mutant completely failed to undergo proteolytic processing when ectopically expressed in HEK293 cells, with or without UNC93B1, resulting in no activation of NF-κB in response to CL075 (Fig. 5C). These results indicate that furin-like proprotein convertase is indispensable for TLR8 cleavage at an initial step. Thus, both furin-like proprotein convertase and cathepsins are involved in stepwise processing of TLR8-N in human primary monocytes and macrophages.

To explore the association of the N- and C- terminal halves of TLR8, C-terminal FLAG-tagged wild-type and mutant hTLR8 were stably expressed in the mouse macrophage cell line RAW264.7. Wild-type TLR8, but not TLR8Δloop, underwent proteolytic processing in RAW cells, similar to an endogenous TLR8 in human macrophages (Fig. 6A). The N-terminal half of TLR8 coimmunoprecipitated with the C-terminal half of TLR8 under reducing and nonreducing conditions, suggesting that both the N- and C-terminal halves were noncovalently associated with each other in macrophages (Fig. 6B). We examined the interaction of ssRNA with TLR8 by pull-down assay using human macrophage lysates and biotinylated ssRNA40. After incubation of biotinylated ssRNA40 in the human macrophage lysates, ssRNA40–receptor complex was pulled down with avidin-Sepharose. Both the N- and C-terminal halves of TLR8 were pulled down, indicating that ssRNA40 bound to the cleaved/associated form of TLR8 (Fig. 6C).

TLR8 localizes to the ER and early endosome in primary human monocytes and HeLa cells transiently expressing hTLR8 (10). We examined the glycosylation of TLR8 protein in monocyte-derived macrophages with Endoglycosidase H, which hydrolyzes the high-mannose type N-glycans and N-glycosidase F that cleaves all types of asparagine-bound N-glycans. Deglycosylation analysis clearly showed that TLR8 was cleaved after passing through the Golgi, because most of the sugars on the cleaved TLR8-N were resistant to Endoglycosidase H and sensitive to N-glycosidase F (Fig. 7A). In contrast, N-linked sugars on full-length TLR8 were sensitive to Endoglycosidase H, indicating that an ∼150-kDa band corresponding to full-length TLR8 was derived from the ER (Fig. 7A). The full-length TLR8 proteins accumulated in macrophages by brefeldin A treatment that disrupted the Golgi pathway (Fig. 7B).

Coexpression of UNC93B1 with TLR8 facilitated intracellular trafficking of TLR8 in HEK293 cells, resulting in accumulation of cleaved TLR8 (Fig. 1D). Confocal immunofluorescence analysis demonstrated that colocalization of TLR8 with p115 (Golgi-resident protein) was increased by UNC93B1 coexpression (Fig. 8A). TLR8 colocalized with MPR (late endosome marker protein), but not LysoTracker (lysosome marker), in UNC93B1-expressing cells (Fig. 8A). In human monocyte-derived macrophages, TLR8 colocalized with early endosome Ag 1 (early endosome marker protein), MPR, and calnexin (ER protein) but not with Lamp-1 (lysosome marker protein) (Fig. 8B). Taken together, these results indicated that TLR8 exits the ER, passes through the Golgi, and is targeted to early/late endosomes where the processing occurs.

In the current study, we showed for the first time, to our knowledge, that proteolytic processing of TLR8 occurs in human primary cells, including monocytes and monocyte-derived macrophages, in a different manner from that of TLR7 and TLR9 cleavage. The insertion loop between LRR14 and LRR15 in TLR8 ECD is indispensable for cleavage and stepwise processing that occurs in the N-terminal fragment. Both furin-like proprotein convertase and cathepsins contribute to TLR8 cleavage in the early/late endosomes. Recent structural analysis of hTLR8 ECD–chemical ligand complex demonstrated that purified TLR8 ECD protein is cleaved at the loop region, and the N- and C-terminal halves remain associated, which contributes to ligand recognition and dimerization (39). Indeed, noncovalent association of the N-terminal half of TLR8 with the C-terminal half was detected by immunoprecipitation assay in RAW macrophages stably expressing hTLR8 (Fig. 6B). Furthermore, a pull-down assay revealed that ssRNA bound to the cleaved/associated form of TLR8 in human macrophages (Fig. 6C).

TLR8 belongs to the same subfamily as TLR7 and TLR9 (14, 15), but localization and signaling are quite different. We showed that hTLR8 localized to the ER and the early/late endosomes, but not to the lysosome, in monocytes and macrophages (Fig. 8B) (10), whereas mTLR7 and mTLR9 localized to the endolysosomes of macrophages and pDCs (8). The pH-dependent cathepsin family and AEP are involved in the stepwise processing of TLR7 and TLR9 at the C-terminal portion in mouse macrophages and DCs, although their contribution depends on cell type (24–29). In contrast, hTLR7 is processed at neutral pH, which is mediated with furin-like proprotein convertase (30). In the case of hTLR8, cathepsins mediate second-step processing of the N-terminal half, resulting in the generation of the mature form of the N-terminal half (Fig. 5A). Like hTLR7 processing, furin-like proprotein convertase is indispensable for TLR8 cleavage at an early step (Fig. 5B, 5C). Indeed, potential furin-like proprotein convertase–recognition sites are present in LRR14 and the flexible loop of hTLR8; one such site is located just before the N-terminus of the C-terminal ECD fragment (39). The TLR8 mutant R467A/R470A/R472A/R473A, in which arginine residues in the potential furin-like proprotein convertase–recognition site were substituted with alanine, was uncleaved when ectopically expressed in HEK293 cells and failed to transmit signals upon stimulation with CL075 (Fig. 5C). Although a putative asparagine cleavage site for AEP is found in the flexible loop of hTLR8 compared with that of mTLR7/9 (27), this site is located within the TLR8-C sequence, suggesting that AEP does not participate in TLR8 cleavage. Thus, proteases involved in processing of TLR7, TLR8, and TLR9 appear to be different and might depend on the localization of receptors and species- and cell type–specific protease distribution.

In freshly isolated human monocytes, cathepsins B and L are distributed in endosomes rather than lysosomes, and their sp. act. is observed in endosomal, but not in lysosomal, fractions (41). In addition, both cathepsins are expressed in HEK293 and THP-1 cells, whereas cathepsin S is expressed in THP-1 cells but not in HEK293 cells (K. Iwano, A. Watanabe, and M. Matsumoto, unpublished data). Considering the endosomal, but not lysosomal, localization of hTLR8, stepwise processing of hTLR8 by furin-like proprotein convertase and members of the cathepsin family, such as cathepsins B and L, might occur in early/late endosomes.

UNC93B1 mediates ER exit of TLR8, resulting in the accumulation of cleaved TLR8 in HEK293 cells (Fig. 1D). Notably, the TLR8-activating ability was different between CL075 and ssRNA40 in HEK293 cells transiently expressing TLR8 (Fig. 1). Structural analysis of unliganded and liganded TLR8 ECD revealed that cleaved/associated TLR8 dimerizes without ligand, and ligand binding to both N- and C-terminal halves induces structural reorganization of the TLR8 dimer (39). Mutagenesis analysis showed that interaction sites of TLR8 ECD with ssRNA appear to differ from those with chemical ligands. Oligomerization of TLR8 dimer might be required for ssRNA-induced signaling like dsRNA-induced TLR3-mediated signaling (42). In HEK293 cells, a small number of cleaved/associated TLR8 molecules is unable to induce signals to activate NF-κB in response to ssRNA40. In any case, our cellular analysis indicates that cleaved/associated TLR8 is responsible for recognition of both chemical ligands and ssRNA, and it induces innate immune responses in human primary cells.

Recent reports showed that dsRNA-sensing TLR3 undergoes cathepsin-mediated cleavage in a cell type–dependent manner (43–45). In addition, TLR3 recognizes incomplete stem structures formed in ssRNA (46). The nucleic acid–sensing TLRs respond to microbial nucleic acid, as well as to endogenous self nucleic acids, in a sequence-independent, but motif-dependent, manner. Hence, the cleaved/associated form of receptors might be beneficial for recognition of nucleic acids with different nucleotide sequences and structures (47).

The role of TLR8 in antiviral immunity in humans remains unknown. In the case of HIV infection, TLR8 signaling appears to benefit HIV replication (48), but another study demonstrated an anti-HIV function for TLR8 (49). TLR8 expressed in neutrophils mediates neutrophil extracellular trap formation in HIV-1 infection via recognition of viral nucleic acids, which is useful for HIV-1 elimination (50). In addition, TLR8-mediated IL-12p70 production by monocytes polarizes naive CD4⁺ T cells into Th1 cells that mediate cellular immunity (51). Thus, TLR8 triggers important antimicrobial signals in distinctive cells that express TLR8 but not other nucleic acid–sensing TLRs.

TLR7 and TLR9 are closely associated with autoimmune disorders because of their expression in B cells and pDCs (52, 53). The relationship between TLR8 and autoimmune disorders was suggested in a TLR8-knockout mouse study that showed a pivotal role for mTLR8 in the regulation of TLR7 expression and prevention of spontaneous autoimmunity (54). In addition, a recent study using hTLR8-transgenic mice clearly demonstrated the connection between TLR8 and autoimmune inflammation (55). TLR8 was shown to induce proinflammatory cytokine production in response to microRNA within exosomes from tumor cells (38). In view of the unique expression profile and signaling skewed toward NF-κB activation, TLR8 might be involved in the development of inflammatory disorders in a distinct manner. Identification of endogenous and exogenous TLR8 ligands and their recognition mechanisms are important for a full understanding of the role of TLR8 in innate immunity and protection against undesirable inflammation and autoimmune responses.

We thank our laboratory members for valuable discussions. We also thank M. Nakai, R. Takemura, K. Mugikura, Y. Takeda, A. Maruyama, and K. Takashima for preparing blood cells. We also thank Dr. K. Miyake, Dr. S. Akira, and Dr. S. Nagata for providing the plasmids.

The authors have no financial conflicts of interest.