TLR9 stability and signaling are regulated by phosphorylation and cell stress

Short abstract

Multiple cell stressors including infection and autophagy regulate TLR9 phosphorylation, stability, and signaling.

Abstract

Innate sensing of pathogens elicits protective immune responses through pattern recognition receptors, including Toll‐like receptors. Although signaling by Toll‐like receptors is regulated at multiple steps, including localization, trafficking, proteolytic cleavage, and phosphorylation, the significance of post‐translational modifications and cellular stress response on Toll‐like receptor stability and signaling is still largely unknown. In the present study, we investigated the role of cytoplasmic tyrosine motifs in Toll‐like receptor‐9 stability, proteolytic cleavage, and signaling. We demonstrated that tyrosine phosphorylation is essential for mouse Toll‐like receptor‐9 protein stability and signaling. Upon inhibition of tyrosine kinases with piceatannol, Toll‐like receptor‐9 tyrosine phosphorylation induced by CpG deoxyribonucleic acid was inhibited, which correlated with decreased signaling. Furthermore, inhibition of Src kinases with 1‐tert‐Butyl‐3‐(4‐chlorophenyl)‐1H‐pyrazolo[3,4‐d]pyrimidin‐4‐amine also inhibited response to CpG deoxyribonucleic acid. Toll‐like receptor‐9 protein stability was also sensitive to autophagy, the cellular stress response pathway, and infection by a deoxyribonucleic acid virus. Whereas autophagy induced by rapamycin or low serum levels caused a preferential loss of the mature p80 proteolytic cleavage product, infection with herpes simplex virus‐1 and induction of cell stress with tunicamycin caused preferential loss of full‐length Toll‐like receptor‐9, which is localized to the endoplasmic reticulum. Our data reveal new information about the stability and signaling of Toll‐like receptor‐9 and suggest that immune evasion mechanisms may involve targeted loss of innate sensing receptors.

Abbreviations

3‐MA

= 3‐methyladenine

Atg5

= autophagy‐5

BMDM

= bone marrow–derived macrophage

Btk

= Bruton's tyrosine kinase

ER

= endoplasmic reticulum

HA

= hemagglutinin

HSV‐1(F)

= HSV‐1 F strain

PRR

= pattern recognition receptor

Syk

= spleen tyrosine kinase

Introduction

TLR9 is the receptor for DNA and activates multiple intracellular signaling cascades from different intracellular compartments leading to both proinflammatory cytokine production and type I IFN production [1, 2–3]. Recent evidence suggests that tyrosine phosphorylation and other post‐translational modifications of TLRs are critical for regulation of proinflammatory cytokine production [2, 4, 5, 6–7]. For example, we have recently demonstrated that a single tyrosine residue in the cytoplasmic tail is necessary for TLR9‐dependent TNF‐α, but not IFN, production [2, 8].

TLR9 is synthesized in the ER, where it is in large part stored before exposure to CpG DNA [4, 9, 10]. Small amounts of TLR9 continuously survey the endosomal system [1], which is dependent on UNC93B1 [11, 12–13] and the ER chaperone gp96 [14, 15, 16, 17–18]. Upon entering acidic endosomes, TLR9 is proteolytically cleaved to a mature form [19, 20–21] that has been proposed to be the active form of TLR9, and proteolytic cleavage to generate this form is critical for TLR9 function. However, more recent studies have shown that an N‐terminal fragment that is removed during cleavage is either absolutely necessary for signaling by the mature form [22] or negatively regulates signaling [23]. Furthermore, uncleaved TLR9 engineered to be surface expressed is hyperactive and can induce autoimmunity [24, 25]. We have described an alternative cleavage event that negatively regulates TLR9 signaling [26]. Thus, cleavage of TLR9 is complex and does not, at least in some contexts, correlate with signaling.

Very little is known about what regulates TLR9's stability and proteolytic cleavage. We showed that in the absence of the ER chaperone gp96, TLR9 is retained in the ER and that gp96 normally remains associated with TLR9 as it traffics to the endosomal compartment to assist in protein stability [14]. Multiple other chaperones have also been implicated in TLR9 stability [3, 27, 28], and several cathepsins regulate proteolytic cleavage [19, 20–21, 29]. However, the regulator of these processes is poorly understood. We investigated the role of TLR9 cytoplasmic tyrosine motifs and the role of activating or inhibiting cell stress pathways on TLR9 proteolytic cleavage and signaling.

MATERIALS AND METHODS

Reagents and plasmids

The following reagents were used: CpG oligodeoxynucleotides (5′‐TCGTCGTTTCGTCGTTTTGTCGTT‐3′, Eurofins MWG Operon, Huntsville, AL, USA); Pam3CSK4 (InvivoGen, San Diego, CA, USA); ultrapure LPS 0111:B4 (Sigma‐Aldrich, St. Louis, MO, USA); a TNF‐α ELISA kit (Biolegend, San Diego, CA, USA); piceatannol and tunicamycin (Calbiochem, San Diego, CA, USA); 3‐MA, PP2, and rapamycin, cell‐counting kit‐8 (Sigma Aldrich); and the QuikChange site‐directed mutagenesis kit (Stratagene, La Jolla, CA, USA). The following antibodies were used: hemagglutinin tag (HA; Applied Biological Materials, Inc., Richmond BC, Canada, and Roche Diagnostics, Indianapolis, IN, USA), phosphotyrosine (4G10; EMD Millipore, Billerica, MA, USA), total p38 (Cell Signaling Technology, Danvers, MA, USA), GFP (Thermo Scientific‐Life Technologies, Grand Island, NY, USA), tubulin (eBioscience, San Diego, CA, USA), LC3 (Novus Biologicals, Littleton, CO, USA), and horseradish peroxidase‐labeled secondary antibodies (Southern Biotech, Huntsville, AL, USA).

Cell culture and virus production

RAW264.7 macrophages (American Type Culture Collection, Manassas, VA, USA) were maintained in DMEM) with 10% (v/v) heat‐inactivated low endotoxin FCS, 2 mM l‐glutamine, 10 mM HEPES, and 1 mM sodium pyruvate (complete DMEM) with the addition of 100 U/ml penicillin and 100 mg/ml streptomycin. TLR9^−/− macrophages (BEI Resources, Manassas, VA, USA) were cultured in complete DMEM with the addition of 10 μg/ml ciprofloxacin. All cell lines were cultured at 37°C with 5% CO₂ and routinely tested negative for mycoplasma by PCR. For the generation of BMDMs, femurs and tibias were collected from 8‐ to 10‐wk‐old mice. Bone marrow was flushed from the bones with cold DMEM supplemented with 20% L‐929 cell‐conditioned medium, 10% (v/v) heat‐inactivated FCS, 2 mM l‐glutamine, 10 mM HEPES, 1 mM sodium pyruvate, 100 U/ml penicillin, and 100 mg/ml streptomycin. Bone marrow cells were cultured in 10 cm Petri dishes (10 ml volume) at 37°C in 5% CO₂ for 7 d. At d 3 and 6, fresh medium was added to the cultured cells. Animal studies were approved by Cornell's Institutional Animal Care and Use Committee (IACUC) and Cornell is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). HSV‐1(F) was propagated in Vero cells. After infection with virus, the cells were washed 3 times with 1× PBS before subsequent analysis.

Generation of tyrosine mutants

Site‐directed mutagenesis was used to introduce changes at individual tyrosines, as previously described [2, 8].

Retroviral transduction of macrophages

Retroviral supernatants were generated with Lipofectamine 2000 (Thermo Scientific‐Invitrogen, Carlsbad, CA, USA) transfected ØNX‐Ampho cells. TLR9^−/− macrophages and RAW264.7 cells were spin transduced with retroviral supernatants plus polybrene (8 µg/ml final concentration), and cultured at 37°C for 44–48 h before stimulation.

Supernatant analysis

TNF‐α in the supernatants was measured by ELISA according to the manufacturer's recommendations (Biolegend). The Griess assay system was used to quantify nitrite as a surrogate of NO production [30].

Immunoblot analysis and immunoprecipitation

Immunoblot analysis was performed as described in several publications [2, 9, 26, 31]. In brief, stimulated cells were washed with ice‐cold HBSS, and lysed for direct immunoblot analysis with 1× SDS‐PAGE reduced sample buffer [62.5 mM Tris (pH 6.8), 12.5% glycerol, 1% SDS, 0.005% bromophenol blue, and 1.7% 2‐ME), or for immunoprecipitation with lysis buffer [50 mM Tris‐Cl (pH 7.4), 150 mM NaCl, 10% (w/v) glycerol, 1 mM EDTA, and protease inhibitors]. Lysates were incubated at 95°C for 5 min before they were resolved by 10% SDS‐PAGE. Proteins were transferred to nitrocellulose membranes and immunoblotted with the indicated antibodies. The membranes were incubated with a SuperSignal West Pico chemiluminescent substrate (Thermo Scientific, Waltham, MA, USA) and exposed to x‐ray film. Films were scanned, and images were assembled in Photoshop (Adobe, San Diego, CA, USA). For immunoprecipitation experiments, total protein was determined in clarified lysates using the BCA protein assay (Bio‐Rad, Hercules, CA, USA), and 5 μg of indicated antibody was used for immunoprecipitation.

Generation of Atg5‐deficient RAW264.7 cells

RAW264.7 cells were infected with retroviral particles encoding shRNA to Atg5. After 48 h, 2 μg/ml puromycin was added, and the cells were cultured for an additional week. Single‐cell selection in 96‐well plates was set up, and the cells were grown to near confluence. Individual clones were expanded and tested for Atg5 knockdown by immunoblot analysis. A clonal cell line was used for the described studies.

Cell viability test

RAW264.7 cells were pretreated with indicated drugs and then stimulated with TLR ligands. After stimulation, the cell‐counting kit‐8 (CCK8; Dojino Molecular Technologies, Gaithersburg, MD, USA) was used to check cell viability, according to the manufacturer's instructions.

RNA isolation and real‐time quantitative PCR

Total RNA was isolated and purified with TRIzol reagent (Life Technologies) according to the manufacturer's instructions. RNA samples were eluted in 40 μl diethylpyrocarbonate‐treated water (Omega Bio‐tek, Norcross, GA, USA) and stored at −80°C. RNA concentration and the A260/280 ratio of each sample were measured on an ultraviolet spectrophotometer (Q3000; Quawell Technology, San Jose, CA, USA). After incubation with DNase I (Thermo Scientific‐Invitrogen) to digest residual genomic DNA, cDNA was synthesized with Invitrogen reverse transcription reagents on a thermocycler (My Cycler; Bio‐Rad), according to manufacturer's instructions.

Quantitative real‐time PCR was performed in triplicate with Power SYBR Green master mix reagent (Thermo Scientific‐Applied Biosystems, Foster City, CA, USA) according to the manufacturer's instructions. Primers for murine TLR9 (forward, 5′‐CCTGGCTAATGGTGTGAAG‐3′; reverse, 5′‐CAAAGCAGTCCCAAGAGAG‐3′) and GAPDH (forward, 5′‐TCCCACTCTTCCACCTTC‐3′; reverse, 5′‐ACCACCCTGTTGCTGTA‐3′) were used at a final concentration of 15 μM/reaction, and cDNA was used at 20 ng/reaction. Reactions were run on a 7500 Fast Real‐Time PCR System (Thermo Scientific‐Applied Biosystems) in standard mode. Cycle parameters were: 1 cycle at 50°C for 2 min, followed by 1 cycle at 95°C for 10 min, then 45 cycles at 95°C for 15 s and 60°C for 1 min. Expression data were obtained in the form of threshold cycle (Ct) values, and relative gene expression was calculated with the ^−ΔΔCt method, with GAPDH used as the reference housekeeping gene.

Statistical analysis

Two way ANOVA with Tukey's post‐hoc test was performed using Prism software. Significance of P < .01 is indicated by an asterisk. Unmarked samples were not statistically different.

Online supplemental material

Supplemental Figure 1 demonstrates that expression of TLR9 in the absence of Syk is reduced, whereas expression of GFP is unaffected.

RESULTS

Proteolysis and tyrosine phosphorylation of TLR9 are important for signaling

We previously characterized a tyrosine‐based motif in the cytoplasmic tail of TLR9 that is important for appropriate intracellular localization and demonstrated that mutation of a single tyrosine residue at Y888 in this motif reduces proinflammatory cytokine production [2, 8]. Proteolytic processing of other nucleic acid–sensing TLRs has been shown to trigger proinflammatory responses [7, 19, 20, 32]. To assess whether tyrosines at residues other than Y888 of TLR9 influences proteolytic cleavage and signaling, we generated single tyrosine mutants (either to alanine or phenylalanine) of those tyrosines within YxxΦ motifs in the cytoplasmic domain of TLR9. As has been demonstrated [19, 20], expression of C‐terminally HA‐tagged mouse TLR9 in TLR9‐deficient macrophages resulted in expression of full‐length TLR9 and a proteolytic cleavage product of ∼80 kDa by immunoblot analysis of HA immunoprecipitates ( Fig. 1 ). Stimulation with CpG DNA ligand for 5 min did not change the proteolytic cleavage pattern for TLR9. Mutation of tyrosine 888 to either phenylalanine (F) or alanine (A) reduced, but did not eliminate, proteolytic cleavage of TLR9 ( Fig. 1), and the relative ratio of full‐length to mature fragment was higher in these mutants than in WT TLR9. Mutation of Tyr⁸⁷⁰ to F had no appreciable effect on TLR9 proteolytic cleavage; however, mutation of the same tyrosine to A abolished cleavage to the 80‐kDa fragment, suggesting that structural aspects around Y870 are important for TLR9 stability. Likewise, there was little or no change in proteolytic cleavage when Tyr⁹⁸⁰ was mutated to F, but cleavage was lost if 980 was mutated to A. Similar amounts of total lysate were used for immunoprecipitation, as shown by immunoblot analysis of the input lysate for total p38 (Fig. 1). Together these data suggest that specific tyrosine residues are critical for the appropriate expression and proteolytic processing of TLR9, although mutation to A appears to have a greater effect.

Figure 1.

Critical role of multiple TLR9 tyrosine residues for protein stability and cleavage. Immunoblot analysis of TLR9 proteolysis. TLR9‐deficient BMDMs were retrovirally transduced with wild‐type TLR9‐HA or the indicated TLR9 mutants. After overnight rest, cells were treated with medium or 1 μM CpG DNA for 5 min. The cells were lysed, and HA immunoprecipitates were analyzed by immunoblot (IB) with antibodies to HA. Input samples of total lysates were immunoblotted for p38. All samples were from the same gel, but are reordered for clarity. Location of image cuts is indicated with spaces. Data are representative of 3 independent experiments.

Cleavage of TLR9 in the hinge region (aa 441–470) has been described as a prerequisite for TLR9 signaling [19, 20]. To determine whether changes in proteolytic cleavage correlates with changes in TLR9 response to CpG DNA, we expressed each mutant in TLR9‐deficient macrophages and assayed for response to CpG DNA. As we observed in other work [2], mutation of Tyr⁸⁸⁸ to F on TLR9 reduced macrophage TNF‐α production in response to CpG DNA, and mutation of the same tyrosine to A eliminated CpG DNA response, probably because of compromises to the structure surrounding Tyr⁸⁸⁸ ( Fig. 2A ). Some of our mutants lend support for a model with obligate cleavage for TLR9 signaling. We observed that wild‐type and the Y870F mutant, which were similarly cleaved, supported similar strong TNF‐α production, and the complete lack of cleavage of Y870A and Y980A correlated with no detectable TNF‐α production. In contrast, both the Y980F and Y888F mutants were cleaved similar to wild‐type, yet their response to CpG DNA was much lower than wild‐type. In each case where we mutated the tyrosine to A, we saw a significant or complete reduction of cleavage and a lack of signaling suggesting that the structure around each tyrosine is important. Mutation of rigid amino acids W886 or L891 flanking the Y888NEL tyrosine motif reduced signaling, supporting our model that disrupting structure of the tyrosine motif interferes with signaling (data not shown). When we mutated all 3 tyrosines to either alanine or phenylalanine, we observed a higher background activity, but no CpG DNA responsiveness (Fig. 2C). However, the triple F mutant had very low expression (data not shown), which suggests that the tyrosines are necessary for TLR9 stability. The response to LPS, mediated by TLR4, was similar in macrophages expressing TLR9 or any of the mutants (Fig. 2B, D). Therefore, the levels of the cleaved product do not directly correlate with the level of TLR9 signaling. Combined with lack of response from the cleaved form of TLR9 [unpublished observation and (22)], we concluded that cleavage does not directly correlate with TLR9 signaling, supporting a model where proteolytic cleavage is not an obligate event before signaling. We also conclude that tyrosines play a fundamental role in regulating both cleavage and signaling.

Figure 2.

Multiple tyrosine residues regulate TLR9‐dependent TNF‐α production. TLR9‐deficient BMDMs were retrovirally reconstituted with the indicated TLR9 single (A, B) or triple (C, D) mutants with alanine (triple A) or phenylalanine (triple F) substitutions. After overnight culture, triplicate wells of cells were stimulated for 24 h with 1 μM CpG DNA (A, C) or 100 ng/ml LPS (B, D). TNF‐α production was determined by ELISA. Data are representative of 3 independent experiments. Error bars, sd; *P < .01.

Tyrosine kinase inhibitor blocks TLR9 phosphorylation and TNF‐α secretion

Tyrosines can be either structural or a site for posttranslational phosphorylation to regulate protein activity. We observed differences in cleavage and function when TLR9 tyrosines were mutated to either alanine or phenylalanine, which disrupts local protein structure and phosphorylation, respectively. Several tyrosine kinases have been implicated in TLR9 signaling, including Syk and Btk [4, 33, 34–35]. Thus, we asked whether acute inhibition of tyrosine kinase activity would inhibit TLR9 phosphorylation and function. Stimulation with CpG DNA or pretreatment with the broad tyrosine kinase inhibitor piceatannol for 3 h did not significantly affect cleavage of TLR9 ( Fig. 3A ). As we described, TLR9‐HA was phosphorylated in response to CpG DNA stimulation ( Figs. 3A , 2, 4 ). A 30 min pretreatment with piceatannol completely inhibited CpG DNA‐induced TLR9 phosphorylation (Fig. 3A). Pretreatment with piceatannol also inhibited CpG DNA‐induced NO production and TNF‐α secretion by macrophages (Fig. 3B; data not shown). Piceatannol (≤20 μM) had no effect on macrophage viability for 24 h when compared to DMSO (Fig. 3C). Furthermore, there was no inhibitory effect on Pam3Csk4‐induced TLR2 signaling, suggesting that the effect was specific for TLR9 (Fig. 3B). One target of piceatannol is the kinase Syk. To address the role of Syk in TLR9 expression and stability, we retrovirally transduced the TLR9‐HA construct into both wild‐type rat basophilic leukemia cells and those lacking Syk kinase. Syk kinase deficiency is lethal in mice precluding studies on Syk‐deficient BMDMs. We observed much lower levels of TLR9 expression in Syk‐deficient cells than in wild‐type cells. However, TLR9 exhibited a different cleavage pattern in both wild‐type and Syk‐deficient rat cells than previously described for human and mouse cells (Supplemental Fig. 1). Regardless, the reduced TLR9 expression in the absence of Syk was a specific effect on TLR9 protein, given that both wild‐type‐ and Syk‐deficient cells had similar expression of retrovirally expressed GFP protein (Supplemental Fig. 1). Therefore, our data support a model where Syk is necessary for optimal TLR9 protein expression and signaling.

Figure 3.

Piceatannol and PP2 inhibit TLR9 phosphorylation and signaling. (A) TLR9 deficient macrophages were retrovirally transduced with TLR9‐HA and stimulated with CpG DNA for 5 min, with or without 30 min piceatannol (Pic. 20 μM) pretreatment. Immunoprecipitation (IP) was performed with anti‐HA antibody, and precipitated immune complexes were analyzed by sequential immunoblot analysis (IB) with anti‐phosphotyrosine (4G10, top) and anti‐HA antibody (bottom). Note that phosphorylation specifically occurs on the p80 mature form of the protein. (B) RAW264.7 macrophages were pretreated for 3 h with 20 μM piceatannol or DMSO control and stimulated for 18 h with 1 μM CpG DNA or 1 μg/ml Pam3Csk4. TNF‐α production was determined by ELISA. Error bars, sem. (C) CCK8 assay for viability of RAW264.7 cells treated for 18 h with medium, indicated concentrations of piceatannol or DMSO corresponding to the concentration found in the 20 μM piceatannol‐treated cells. Error bars, sd. (D, E) RAW264.7 macrophages were pretreated with DMSO or the indicated concentrations of PP2 and stimulated with 3 μM CpG DNA or 1 μg/ml Pam3Csk4 for 18 h. TNF‐α production was determined from triplicate samples by ELISA. Error bars, sd. (F) CCK8 viability test for RAW264.7 cells treated as indicated for 18 h. Error bars, sd; *P < .01. Data are representative of 3 independent experiments, except (A), which is representative of 2 independent experiments.

Figure 4.

Induction of autophagy and infection with HSV cause loss of TLR9 protein. (A) TLR9‐deficient BMDMs were retrovirally transduced with empty vector or wild‐type TLR9‐HA. Forty‐eight hours after transduction, the cells were treated with DMSO or 0.5 μM rapamycin, 0.5 mM 3‐MA, or 20 μM piceatannol for an additional 6 h. Whole‐cell lysates were harvested and immunoblotted with antibodies to HA and tubulin. (B) RAW264.7 cells were treated with medium, DMSO, 5 μg/ml tunicamycin, 0.5 μM rapamycin, or 0.5 mM 3‐MA for 6 h. Whole‐cell lysates were immunoblotted for LC3 and tubulin. (C) As above, but after transduction, cells were left untreated (10% FCS), serum starved (1% FCS), or infected with HSV‐1 (MOI = 10) for 3 h, and lysates were subsequently immunoblotted with antibodies to HA or tubulin. (D) RAW264.7 cells were treated with the indicated concentrations of rapamycin or 3‐MA for 6 h. Whole‐cell lysates were harvested and immunoblotted with antibodies to endogenous TLR9 (IMG‐431) and tubulin. (E) RAW264.7 cells were pretreated with DMSO, 5 μg/ml tunicamycin, 0.5 mM 3‐MA, or 0.5 μM rapamycin and then stimulated with 1 μM CpG DNA. Supernatants collected at 3 and 12 h were assayed for TNF‐α by ELISA. *P < .01 comparing 12 h CpG DNA stimulated to media under each condition, or comparing 12 h CpG DNA between drug treatment groups as indicated with a line. Non‐indicated samples were not significantly different. (F) CCK8 viability test on the same cells in (E) at 12 h. *P < .01 compared to none. (G) TLR9‐deficient BMDMs were retrovirally transduced with wild‐type TLR9‐HA. Forty‐eight hours after transduction, the cells were stimulated with 1 μM CpG DNA or 100 ng/ml LPS or were infected with HSV‐1 (MOI = 10) for 3 h and lysates were immunoblotted for HA and total p38. (H) RAW264.7 cells were stimulated with DMSO, 5 μg/ml tunicamycin, 0.5 mM 3‐MA, or 0.5 μM rapamycin for 6 h. RNA was harvested and used in real‐time quantitative PCR reactions for GAPDH and TLR9. Data are expressed as relative gene expression normalized to medium‐treated cells. Data are representative of 2 (D, H) to 3 (all others) independent experiments.

We next tested the role of Src kinases in TLR9 signaling using the Src inhibitor PP2. Preincubation with PP2 diminished CpG DNA‐induced TNF‐α production in a dose‐dependent manner (Fig. 3D) and was not toxic to the macrophages (Fig. 3F), but unlike piceatannol, PP2 reduced TLR2 signaling (Fig. 3E). From these studies, we conclude that inhibiting tyrosine kinases abolishes TLR9 phosphorylation and proinflammatory responses. However, further studies are needed to determine which kinases directly phosphorylate TLR9.

Induction of autophagy, ER stress, and infection cause loss of TLR9 protein

CpG DNA induces Syk‐dependent phosphorylation of Akt [4]. Inhibition of Syk by piceatannol suppresses mTOR activity and subsequently results in autophagy [36]. In B cells, after B‐cell receptor engagement, TLR9 is recruited to an autophagosome‐like compartment, which correlates with enhanced TLR9 signaling [37]. Therefore, we next asked whether direct induction of autophagy in macrophages has any effect on TLR9 protein expression or cleavage. Induction of autophagy with rapamycin reduced expression of both full‐length and cleaved TLR9 compared to DMSO‐treated control cells (Fig. 4A). In contrast, inhibition of autophagy by 3‐MA resulted in increased total TLR9 protein expression and the mature cleavage fragment (Fig. 4A). Rapamycin induced the expected change in migration of LC3, indicating that autophagy was induced (Fig. 4B). Serum starvation, which also induces autophagy, caused loss of TLR9 protein. When cells were serum starved at 1% serum compared to 10% serum, there was greater loss of the cleaved form with less loss of full‐length TLR9 (Fig. 4C). Endogenous full‐length TLR9 was also reduced by rapamycin treatment and increased by 3‐MA treatment (Fig. 4D). However, the antibody used to immunoblot endogenous TLR9 detects only the full‐length, not the mature, form of the protein. Treatment with rapamycin for 12 h dramatically reduced CpG DNA‐induced TNF‐α production (Fig. 4E), but also caused a significant reduction in viability, as determined by CCK8/MTT assay (Fig. 4F). However, rapamycin treatment for only 3 h still significantly reduced TNF‐α production (Fig. 4E). In contrast, treatment with 3‐MA for 12 h had little or no effect on viability, but did reduce TNF‐α production (Fig. 4E). The reduction in TLR9 expression was not related to an effect on mRNA expression, because there was no statistically significant difference in TLR9 mRNA in cells treated with DMSO, rapamycin, or 3‐MA (Fig. 4H). We concluded that induction of autophagy with rapamycin causes loss of TLR9 protein, which may occur by targeting the autophagosome for turnover after signaling.

We next asked whether infection with virus causes loss of TLR9 protein. We observed that within 3 h of HSV‐1 infection of RAW264.7 cells, TLR9 protein was lost, but the pattern of protein loss was different from that observed after induction of autophagy (Fig. 4B). HSV‐1 infection also caused a similar loss of TLR9 protein in primary mouse BMDMs (Fig. 4G). In both cases, HSV‐1 infection resulted in preferential loss of full‐length TLR9, sparing the cleaved form (Fig. 4B, G). In contrast, LPS stimulation had no effect on TLR9 expression or cleavage pattern (Fig. 4G). Although HSV‐1 infection likely has multiple different effects on the cells, our data are consistent with viral infection causing loss of TLR9 expression.

Autophagy is known to cross‐talk with TLR signaling pathways. In the absence of Atg5, MyD88 formed higher order structures that resulted in more TLR signaling [38]. However, disruption of Atg5 inhibited TLR7‐induced type I IFN, but not IL‐12 production in plasmacytoid dendritic cells [39]. We next asked whether disruption of Atg5 had an effect on TLR9 expression or cleavage. We generated stable TLR9‐deficient macrophages with little or no expression of Atg5 (Atg5‐KD9K; Fig. 5A ). When we expressed TLR9‐HA in these cells, full‐length TLR9 was similar in both wild‐type and Atg5‐KD9KO macrophages. However, expression of proteolytically cleaved TLR9 decreased in Atg5‐KD9KO macrophages compared with that in the wild‐type controls (Fig. 5B). In addition, a similarly generated stable RAW264.7 Atg5‐KD macrophage cell line produced more TNF‐α in response to CpG DNA stimulation (Fig. 5C), whereas LPS signaling was unaffected (Fig. 5D). HSV‐1 infection of Atg5‐KD9KO macrophages cells resulted in similar, or perhaps more robust, loss of TLR9 (Fig. 5B), which may explain the reduced IFN production from Atg5 KO plasmacytoid dendritic cells after HSV‐1 infection [39]. Thus, it remains possible that Atg5 regulates TLR signaling in a cell‐type–specific manner.

Figure 5.

Autophagy regulates TLR9 expression and signaling. (A) Whole‐cell lysates from wild‐type and stable Atg5‐knockdown TLR9‐deficient macrophages were immunoblotted for Atg5 and tubulin. (B) Wild‐type and stable Atg5 knockdown TLR9‐deficient macrophages were retrovirally transduced with empty vector or TLR9‐HA. After 48 h, cells were treated with medium, infected with HSV‐1 (MOI = 10) or stimulated with 1 μM CpG DNA for 3 h. Whole‐cell lysates were immunoblotted for HA and total p38. (C, D) Wild‐type and stable Atg5 knockdown RAW264.7 cells were stimulated with (C) 1 μM CpG DNA or (D) 100 ng/ml LPS for 12 h. TNF‐α production was determined from quadruplicate samples by ELISA. Data are representative of 3 independent experiments. *P < .01 Both media compared to CpG DNA were significant, but not indicated on the graph. Other sample comparisons were not significantly different.

We observed preferential loss of the ER‐resident full‐length TLR9 upon HSV‐1 infection, and TLR9 primarily resides in the ER [10]; furthermore, induction of ER stress with tunicamycin treatment inhibits signaling through both surface and endosomal TLRs [40]. Thus, we next asked if direct induction of ER stress would cause preferential loss of ER‐resident TLR9. Tunicamycin treatment caused significant loss of full‐length TLR9 protein at 4 h, which was undetectable by 6 h ( Fig. 6 ). The cleaved form of TLR9 was not affected by tunicamycin treatment for up to 4 h, but by 6 h, the cleaved form of TLR9 was also slightly reduced. TLR9 cleavage occurs in several steps involving multiple enzymes [29]; therefore, it was interesting that a slightly slower migrating form of cleaved TLR9 was observed at these later time points. This finding may suggest that partial cleavage of the remaining full‐length TLR9 still occurs after tunicamycin treatment. It is important to note that tunicamycin treatment induced LC3 lipidation (Fig. 4B), which is consistent with reported data showing that tunicamycin induces autophagy via ER stress [41]. When macrophages were pretreated with tunicamycin for 6 h, there was a slight reduction in viability (Fig. 4F), but there was a dramatic inhibition of CpG DNA‐induced TNF‐α production (Fig. 4E). However, tunicamycin treatment did not have any effect on TLR9 mRNA expression (Fig. 4H). Altogether, our data demonstrate that multiple mechanisms that disturb cell homeostasis including autophagy, viral infection, and ER stress cause a loss of TLR9 protein. Because different TLR9 fragments are targeted by these different mechanisms, it would be interesting to define the proteases and pathways responsible for induced loss of TLR9.

Figure 6.

Induction of ER stress causes preferential loss of full‐length TLR9. TLR9‐deficient BMDMs were retrovirally transduced with empty vector or wild‐type TLR9‐HA. Forty‐eight hours after transduction, cells were pretreated with DMSO or 5 μg/ml tunicamycin for the indicated times. Whole‐cell lysates were immunoblotted with antibodies to HA and tubulin. Full‐length (FL TLR9) and p80 fragments of TLR9 are indicated. Data are representative of 3 independent experiments.

DISCUSSION

TLRs are innate immune receptors important for initial recognition of microbial structures, yet some TLRs respond to nucleic acids that can be present in both microbe and the host. Dysregulation of mechanisms controlling access to nucleic acid ligands and TLR signaling contribute to autoimmune disease. Thus, we need to understand how inflammation, stress, and infection regulate nucleic acid sensing TLRs. In this study, cell stress and DNA virus infection reduced TLR9 stability, and tyrosine kinases played an important role in TLR9 homeostasis.

TLR9 is phosphorylated upon engagement by CpG DNA [2, 4]. Syk is also phosphorylated upon CpG DNA stimulation, and inhibition of Syk kinase blocked both Syk and TLR9 phosphorylation [4]. However, inhibition of endosomal acidification, which inhibits TLR9‐mediated cytokine production and proteolytic cleavage [42], had no effect on Syk and TLR9 phosphorylation, suggesting that these events occurred very early in the signaling cascade. In our hands, acute inhibition of Syk also causes a loss of TLR9 protein. Because phosphorylation of TLR9 is important for TLR9 localization and trafficking [2], Syk‐dependent phosphorylation of TLR9 likely regulates both TLR9 stability and function. However, TLR9 phosphorylation is not necessary for dimer formation, given that treatment with PP2 or piceatannol for 6 h had no effect on preformed dimers (data not shown). Analysis of our tyrosine mutants showed that TLR9 cleavage does not directly correlate with signaling. For example, TLR9^Y888A and TLR9^Y888F had similar levels of proteolytic cleavage, albeit much lower than wild‐type, but TLR9^Y888A failed to respond to CpG DNA, whereas TLR9^Y888F exhibited partial response. Also, TLR9^Y980F was cleaved, yet its response to CpG DNA was low (Fig. 2). Although not consistent with a model that requires proteolytic cleavage for TLR9 to be functional [19, 20], our data are in line with those of other studies that have failed to show that the proteolytically cleaved p80 form of TLR9 is functional [22]. In addition, our data show that in the absence of Atg5, the level of p80 is reduced, yet TNF‐α production is increased (Fig. 5), again suggesting that proteolytic cleavage and signaling do not correlate. A recent crystal structure analysis showed that unprocessed TLR9 binds to either stimulatory or inhibitory DNA, but the proportion of TLR9 in dimeric form increases for processed TLR9 in association with stimulatory CpG DNA [43]. Thus, it remains unclear whether proteolytic cleavage occurs before or after CpG DNA binding and exactly what role it plays in regulating TLR9 signaling.

Other TLRs are also regulated by phosphorylation, including both surface and nucleic acid–sensing intracellularly localized TLRs [5, 6–7, 44, 45, 46–47]. For example, TLR3 dimerizes upon engagement with double‐stranded RNA, which results in Src‐dependent TLR3 phosphorylation [45]. Mutation of 2 tyrosine residues (Y759 and Y858) abrogates TLR3 signaling, indicating the importance of these tyrosines in TLR3 signaling [7, 32, 45, 47]. Our mutational analysis shows that mutation to phenylalanine of any of the 3 tyrosine residues located within YXXΦ motifs in the cytoplasmic tail of TLR9 (Fig. 1) has no effect on TLR9 stability and cleavage. However, mutation of either Y888 or Y980 dramatically reduces signaling. Interaction between Src kinase and TLR3 does not seem to occur through traditional SH2, and TLR adapters, such as TIR‐domain‐containing adapter‐inducing IFN‐β, may be necessary for the interaction; furthermore, other kinases such as Btk are necessary for TLR3 and ‐9 signaling [31, 34, 48, 49]. Our data demonstrate that the inhibitor PP2 reduces both TLR9 and ‐2 signaling pathways, yet piceatannol selectively inhibited TLR9, but not TLR2, signaling. PP2 is proposed to be an Src family kinase inhibitor, but it inhibits receptor‐interacting protein‐2, which is described to play a role in signaling through multiple TLRs, as well as NOD2. Piceatannol is more selective for Syk family kinases, suggesting that it plays an important role in TLR9 signaling. Our data showed a significant reduction in TLR9 expression upon 6 h treatment with piceatannol (Fig. 4), suggesting that Syk plays a role in TLR9 stability. Whether there is direct interaction between Syk or other kinases and the cytoplasmic tail of TLR9 and what residues are necessary for such a potential interaction requires further study.

Viruses employ many mechanisms to evade innate immune detection [50]. For example, HSV encodes infected cell polypeptide‐0, which inhibits TLR2 [51]. Infected cell polypeptide‐0 causes translocation of ubiquitin‐specific peptidase‐7 to the cytoplasm where it deubiquitinates TNFR‐associated factor‐6 and Iκ B kinase‐γ, which results in inhibition of NF‐κB‐dependent signaling [52]. EBV also interferes with TLR signaling through expression of the deubiquitinase BPLF1 [53]. Varicella zoster virus also targets NF‐κB, but inhibition requires the e3 ubiquitin ligase activity of open‐reading frame‐61, suggesting that this virus actually ubiquitinates targets to effect evasion [54]. Our study shows that infection by HSV and induction of the cell stress pathway with tunicamycin both result in preferential loss of the full‐length glycosylated, typically ER‐resident, form of TLR9. This result is in contrast to our observation that induction of autophagy with rapamycin or serum starvation preferentially causes loss of the cleaved mature form of TLR9, typically found in the endosomal compartment. We could not evaluate the role of ER stress in HSV‐induced TLR9 loss because tunicamycin has antiviral effects on HSV [55]. Viruses targeting the ER and TLR9 may also affect other critical TLR9‐chaperoning proteins such as gp96 and UNC93B1, which are necessary for full maturation and trafficking of TLR9 from the ER to the endosomal compartment. The result of this targeting may be reduced ability of the cell to detect infection; thus, virus‐induced loss of TLR9 protein may be a defense mechanism that viruses, such as HSV‐1, use to evade innate immune detection.

Together, our studies describe a new mechanism whereby tyrosines in the cytoplasmic tail of TLR9 regulate protein stability, and suggest that cell perturbances such as cell stress, autophagy, and viral infection also regulate TLR9 stability. However, viral infection–directed loss of TLR9 may also interfere with normal homeostatic mechanisms mediated by TLR9 in intestinal repair [56]. Future studies are needed to determine whether similar tyrosine‐based regulatory networks are involved in regulation of other nucleic acid‐sensing TLRs.

AUTHORSHIP

M.H. and C.A.L. conceived the project, designed the experiments, analyzed the data, and wrote the manuscript. M.H., E.G., J.C., and C.A.L. performed the experiments.

DISCLOSURES

The authors declare no conflicts of interest.

ACKNOWLEDGMENTS

This work was supported by U.S. National Institutes of Health (NIH) National Institute of Allergy and Infectious Diseases (NIAID) Grants R01AI076588, R01AI076588S1, and R03AI097671 (to C.A.L.). The following reagent was obtained through Biodefense and Emerging Infections (BEI) Resources, NIH NIAID: Macrophage Cell Line Derived from TLR9 Knockout Mice (NR‐9569).The authors thank J. Baines for providing the HSV‐1 virus for infections, C. Heyward for critically reading the manuscript, and R. Siraganian for permission to obtain the Syk‐BBL cells from D. Holowka and use them in our studies [57].

Supporting information

Supplementary material 1