TLR4 endocytosis and endosomal TLR4 signaling are distinct and independent outcomes of TLR4 activation

Abstract

Toll-like receptor 4 (TLR4) signaling at the plasma membrane and in endosomes results in distinct contributions to inflammation and host defence. Current understanding indicates that endocytosis of cell surface-activated TLR4 is required to enable subsequent signaling from endosomes. Contrary to this prevailing model, our data show that endosomal TLR4 signaling is not reliant on cell surface-expressed TLR4 or ligand-induced TLR4 endocytosis. Moreover, previously recognized requirements for the accessory molecule CD14 in TLR4 endocytosis and endosomal signaling are likely attributable to CD14 binding as well as trafficking and transferring lipopolysaccharide (LPS) to TLR4 at different subcellular localizations. TLR4 endocytosis requires the TLR4 intracellular signaling domain, contributions by phospholipase C gamma 2, spleen tyrosine kinase, E1/E2 ubiquitination enzymes, but not canonical TLR signaling adaptors and cascades. Thus, our study identifies independently operating TLR4 signaling modes that control TLR4 endocytosis, pro-inflammatory cell surface-derived, as well as endosomal TLR4 signaling. This revised understanding of how TLR4 functions within cells might be harnessed to selectively amplify or restrict TLR4 activation for the development of adjuvants, vaccines and therapeutics.

Keywords: Endosome; LPS; Macrophage; Signaling; TLR4.

Figure 1. Expression and endocytosis of surface TLR4 are not required for LPS-induced TLR4 endosomal signaling.

(A, B) TLR4 endocytosis was assessed in WT BMM treated for 60 min with lipid raft inhibitors (A) filipin (5 μM) or (B) MβCD (10 mM), followed by stimulation with LPS (100 EU/mL) for 120 min, or left unstimulated (CTRL). (C, D) Impact of lipid raft inhibitors filipin or MβCD on LPS-induced (C) Ifnb1 and (D) Il1b mRNA expression in WT BMM stimulated with LPS (100 EU/mL) for the indicated times. (E) Immunoblots of TLR4-V5 in RAW^TLR4ko cells stably reconstituted with mTLR4^WT, mTLR4^N524/572Q or mTLR4^N572Q. Black arrow head indicates low-glycosylated intracellular TLR4, white arrow head indicates highly-glycosylated cell surface-expressed TLR4. (F) Flow cytometric analysis of TLR4 surface expression in RAW^TLR4ko cells stably expressing mTLR4^WT, mTLR4^N524/572Q or mTLR4^N572Q. (G) LPS-induced Ifnb1 expression assessed by qRT-PCR in RAW^TLR4ko cells stably expressing mTLR4^WT, mTLR4^N524/572Q or mTLR4^N572Q. (H) Analysis of V5-tagged TLR4 via immunoblot in RAW^TLR4ko cells expressing SpyTag-mTLR4^WT-V5 incubated with SpyCatcher protein, followed by stimulation with LPS (100 EU/mL), or left unstimulated, across indicated time points. Purple arrow head indicates SpyCatcher-labelled highly-glycosylated surface TLR4; black and white arrow heads indicate unlabelled intracellular and surface TLR4, respectively. Data information: Flow cytometry histograms and immunoblot images depict 1 representative of n = 3–5 biological replicates generated in independent experiments. Bar plots are mean ± SEM of n = 3–5 biological replicates generated in independent experiments indicated as data points. Unpaired two-tailed t test utilized in (A, B, G); ordinary two-way ANOVA with Dunnett’s multiple comparison test utilized in (C, D). Ordinary one-way ANOVA with Dunnett’s multiple comparison test utilized in (H). *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001, ns = not significant. Source data are available online for this figure.

Figure 2. Endocytosis of CD14 correlates with TLR4-mediated Ifnb1 expression.

(A, B) Comparison of impact by endocytosis inhibitors and lipid raft disruptors (filipin, MβCD, EIPA, Pitstop-2, PCZ, dynasore) on LPS-induced TLR4 endocytosis, Ifnb1 expression, Il1b expression (100 EU/mL LPS) and CD14 endocytosis (10000 EU/mL LPS) in WT BMMs. Endocytosis and mRNA expression data in each independent experiment were expressed as a percentage of the parameter detected in cells treated with solvent control. (B) Percentage overlap of inhibitor perturbation profiles depicted in (A). (C, D) Impact of filipin treatment on (C) TLR4 endocytosis and (D) Ifnb1 and Il1b expression induced by 1Z105 (10 μM) in WT BMMs. (E, F) Comparison of impact by endocytosis inhibitors and lipid raft disruptors on 1Z105-induced TLR4 endocytosis, Ifnb1 expression and Il1b expression. Endocytosis and mRNA expression data in each independent experiment were expressed as a percentage of the parameter detected in cells treated with solvent control. (F) Percentage overlap of inhibitor perturbation profiles depicted in (E). Data information: Flow cytometry histograms depict 1 representative of n = 5 biological replicates generated in independent experiments. Bar plots are mean ± SEM of n = 3–5 biological replicates generated in independent experiments indicated as data points. Radar plots show means of n = 3–5 biological replicates generated in independent experiments. Unpaired two-tailed t test utilized in (C). Ordinary two-way ANOVA with Sidak’s multiple comparisons test utilized in (D). *P < 0.05; **P < 0.01; ****P < 0.0001, ns = not significant. Source data are available online for this figure.

Figure 3. TLR4 activity is required for ligand-induced TLR4 endocytosis.

Impact of TLR4 inhibitor TAK-242 on (A, B) TLR4 surface expression at indicated time points, and cytokine (C) mRNA expression (1.5 h/90 min) and (D) protein secretion (4 h) in WT BMMs stimulated with LPS (10, 100, 1000 EU/mL), Pam₃CSK₄ (10 ng/mL), or left unstimulated (CTRL). (E–G) Retroviral reconstitution of Tlr4^−/− BMMs with (E) mTLR4^WT, mTLR4^C745A, mTLR4^C745S or empty vector were treated with TAK-242 for 1 h prior to LPS stimulation (100 EU/mL). (F) TLR4 surface expression at 4 h post-stimulation; (G) IL-6 release at 24 h post-stimulation. (H, I) Impact of TAK-242 on (H) TLR4 surface expression (120 min) or (I) Ifnb1/Il1b expression (90 min) in WT BMM infected with E. coli (strains MG1655, EC958, CFT073) or L. monocytogenes strain 10403S (LM) (MOI 10). (J, K) Impact of TAK-242 on TLR4 endocytosis of WT and Cd14^−/− BMM in response to (J) CD14-independent TLR4 agonist 1Z105 (10 µM) or (K) CD14-dependent TLR4 agonist LPS (100 EU/mL). Data information: Flow cytometry histograms depict 1 representative of n = 3 biological replicates. Bar plots are mean ± SEM of n = 3–4 biological replicates generated in independent experiments indicated as data points. Ordinary two-way ANOVA with Dunnett’s multiple comparisons test utilized in (B, J, K). Unpaired two-tailed t test utilized in (C, D, H, I). Ordinary one-way ANOVA with Dunnett’s multiple comparisons test utilized in (F, G). *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001, ns = not significant. Source data are available online for this figure.

Figure 4. Presence and signaling capability of the TLR4 TIR domain are prerequisites for ligand-induced TLR4 endocytosis.

(A) Impact of TAK-242 treatment on TLR4 surface expression in C3H/HeN and C3H/HeJ BMM after LPS stimulation (indicated concentrations, 120 min). (B–D) Tlr4^−/− BMMs retrovirally reconstituted with (B) mTLR4^WT, mTLR4^ΔTIR, mTLR4^P712H or empty vector were stimulated with LPS (100 EU/mL) and (C) TLR4 surface expression (0, 120, 240 min) and (D) IL-6 and CXCL10 concentrations (24 h) assessed. Data information: Flow cytometry histograms depict 1 representative of n = 3 biological replicates. Bar plots are mean ± SEM of n = 3–4 biological replicates generated in independent experiments indicated as data points. Ordinary two-way ANOVA with Dunnett’s multiple comparisons test utilized in (A, C). Unpaired two-tailed t test utilized in (D). *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001, ns = not significant. Source data are available online for this figure.

Figure 5. TLR4 TIR domain-dependent endocytosis occurs independent of canonical TLR signaling pathways.

(A–C) Impact of adaptor protein deficiency on TAK-242-sensitive TLR4 endocytosis. WT and (A) Myd88^-/- and Trif^−/−, (B) Mal^−/− or (C) Bcap^−/− BMM were treated with TAK-242 for 60 min prior to LPS stimulation (100 EU/mL) for 120 min. (D, E) Impact of inhibitors of canonical TLR signaling pathway components on TLR4 endocytosis. WT BMM pre-treated for 60 min with DMSO, TAK-242, IKK-16 (10 μM), SB203580 (10 μM), MRT67307 (5 μM), U0126 (10 μM) or SP600125 (10 μM) followed by stimulation with (D) LPS (100 EU/mL) or (E) 1Z105 (10 µM) over a time-course. (F) Impact of NF-κB pathway inhibitors IKK-16 or SC-514 on TLR4 endocytosis. WT BMM were pre-treated for 60 min with DMSO, TAK-242, IKK-16 or SC-514 (20 μM), followed by stimulation with LPS (100 EU/mL) for 120 min. Data information: Flow cytometry histograms depict 1 representative of n = 3 biological replicates. Bar plots are mean ± SEM of n = 3 biological replicates generated in independent experiments indicated as data points. Ordinary two-way ANOVA with Dunnett’s multiple comparisons test utilized in (A–E). Ordinary one-way ANOVA with Dunnett’s multiple comparisons test utilized in (F). *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001, ns = not significant. Source data are available online for this figure.

Figure 6. A ubiquitination-linked cellular process, independent of canonical TLR signaling, governs TLR4 endocytosis.

(A, B) Impact of ubiquitination enzyme inhibitors on TLR4 endocytosis. WT BMM were treated for 60 min with TAK-242, E1 inhibitor PYR-41 (25 μM), E2 inhibitors BAY 11-7082 (10 μM), NSC697923 (10 μM), or proteasome inhibitor MG132 (25 μM) followed by stimulation with (A) LPS (100 EU/mL) or (B) 1Z105 (10 µM) over a time-course. (C, D) Impact of NEDDylation and SUMOylation inhibitors on TLR4 endocytosis. WT BMM were treated for 60 min with TAK-242, MLN7243 (1 μM), MLN4924 (2.5 μM) or TAK-981 (2 μM) followed by stimulation with (C) LPS (100 EU/mL) or (D) 1Z105 (10 µM) over a time-course. Data information: Flow cytometry histograms depict 1 representative of n = 3 biological replicates. Bar plots are mean ± SEM of n = 3 biological replicates generated in independent experiments indicated as data points. Ordinary two-way ANOVA with Dunnett’s multiple comparisons test utilized in (A–D). *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001, ns = not significant. Source data are available online for this figure.

Figure 7. PLCγ2 promotes ligand-induced TLR4 endocytosis and type I IFN expression.

(A) LPS- and 1Z105-induced Ifnb1 and Il1b expression (90 min) in the presence of filipin (5 µM) and/or U73122 (10 µM). (B, C) Immunoblot analysis of impact of TAK-242 (1 μM) on PLCγ2 phosphorylation at tyrosine 1217 in WT BMM upon stimulation with (B) LPS (100 EU/mL) or (C) 1Z105 (10 µM) for the indicated times. TBK1 phosphorylation was assessed as control for LPS stimulation and TAK-242 inhibitory activity. Data information: Bar plots are mean ± SEM of n = 3–4 biological replicates generated in independent experiments indicated as data points. Western blot images are represent 1 of n = 4 biological replicates performed in independent experiments. Ordinary two-way ANOVA with Dunnett’s multiple comparisons test utilized in (A). Ordinary one-way ANOVA with Dunnett’s multiple comparisons test utilized in (B, C). *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001, ns = not significant. Source data are available online for this figure.

Figure 8. Model of TLR4 signaling modes and outcomes.

Building on existing knowledge of TLR4 signaling, the data presented here revise current understanding of TLR4 signaling. TLR4 endocytosis is initiated upon TLR4 ligand-driven TLR4 activation. In case of the TLR4 ligand LPS, GPI-anchored membrane-bound CD14 (mCD14) (Zanoni et al, 2011), or soluble CD14 (sCD14) with lower efficiency (Tsukamoto et al, 2018), are required to transfer LPS to TLR4. In contrast, the synthetic agonist 1Z105 activates TLR4 independent of CD14 (Rajaiah et al, 2015). TLR4 endocytosis is dependent on TLR4 TIR domain activity (Figs. 3 and 4), but does not require TLR signaling adaptors MAL, MyD88, TRAM, TRIF (Rajaiah et al, ; Zanoni et al, 2011) (Fig. 5; Appendix Figs. S2 and S7), or BCAP (Fig. 5). Instead, E1 and E2 ubiquitination enzyme activity, independent of ubiquitination events that regulate canonical TLR4 signaling pathways, are required for TLR4 endocytosis (Fig. 6). PLCγ2 and SYK activity contributes to TLR4 endocytosis (Chiang et al, ; Zanoni et al, 2011) (Fig. 7; Appendix Fig. S10), and TLR4 activity regulates cellular PLCγ2 protein levels (Fig. EV3). Endocytosed TLR4 is directed towards degradation (Husebye et al, 2006) (Fig. 1). TLR4-MAL-MyD88 signaling activated at the cell surface (Latz et al, 2002) results in expression of pro-inflammatory cytokines (Kawai et al, 1999) among other cellular responses to exogenous and endogenous danger signals. CD14 is required for activation of TLR4-MAL-MyD88 signaling at low but not high LPS concentrations (Zanoni et al, 2011) with both mCD14 and sCD14 able to promote surface TLR4 activation (Frey et al, ; Wright et al, 1990). In contrast, the small molecule 1Z105 activates TLR4 independent of CD14 (Hayashi et al, 2014). TLR4 endocytosis negatively regulates TLR4-MAL-MyD88 signaling, thereby curbing pro-inflammatory outputs of TLR4 activation (Husebye et al, ; Latz et al, 2002). TLR4-MAL-MyD88 and TLR4-TRAM-TRIF signaling occur in parallel, and the latter is not reliant on TLR4 surface expression or TLR4 endocytosis (Figs. 1 and EV1). In the case of LPS, cellular uptake and trafficking requires mCD14 (Wright et al, 1990), but not sCD14 (Jacque et al, ; Tapping and Tobias, 1997), to deliver LPS to endosomes (Vasudevan et al, 2022). CD14 undergoes endocytosis with contributions by PLC and ubiquitination enzyme activity (Appendix Figs. S9 and S10), trafficking bound LPS into the cell (Kitchens and Munford, ; Kitchens et al, 1998) and transferring it to endosomal TLR4. This results in TLR4-TRAM-TRIF signaling and TBK1/IKKe activation that promotes type I IFN expression (Fig. 2). In addition, TRIF and RIPK1 promote prolonged NF-κB activation (Cusson-Hermance et al, ; Sato et al, 2003). As LPS is a cell-impermeable molecule (Guerville and Boudry, 2016), the requirement for mCD14 for TLR4-TRAM-TRIF signaling cannot be overridden by high LPS concentrations (Lloyd-Jones et al, 2008) or complexing with sCD14 (Saito et al, 2000). In contrast to LPS, 1Z105 does not require CD14 endocytosis to be delivered to endosomal TLR4. Nevertheless, perturbation of 1Z105-induced Ifnb1 expression by endocytosis inhibitors (Fig. 2) and concomitant induction of LPS- and 1Z105-induced Ifnb1 expression (Appendix Fig. S5) suggests that 1Z105 is actively trafficked to endosomal compartments where it activates TLR4-TRAM-TRIF signaling.

Figure EV1. Impact of dynasore and lipid raft inhibitors on TLR-mediated macrophage activation.

(A) Flow cytometric analysis of TLR4 surface expression in WT BMM treated for 60 min with DMSO or dynasore (80 μM), followed by stimulation with LPS (100 EU/mL) or left unstimulated (CTRL) for 120 min. (B) qRT-PCR analysis of Ifnb1 and Il1b expression in WT BMM treated for 60 min with DMSO or dynasore (80 μM), followed by stimulation with LPS (100 EU/mL) for 90 min. (C) qRT-PCR analysis of Il1b, Ifnb1 and Il6 mRNA expression in WT BMM treated for 60 min with DMSO or dynasore (80 μM), followed by stimulation with Pam₃CSK₄ (10 ng/mL), poly I:C (10 μg/mL), flagellin (0.5 μg/mL) or CpG DNA (1 μM), or left unstimulated (CTRL), for 90 min. (D, E) Impact of TAK-242 (1 µM) or MRT67307 (5 µM) on LPS-induced (D) Ifnb1 and (E) Il1b expression (90 min) in WT BMM treated with filipin (5 µM), MβCD (10 mM), or DMSO as control. (F) Time-course of LPS-induced TBK1 phosphorylation in WT BMM treated with filipin, MβCD, dynasore or DMSO as solvent control. Data information: Flow cytometry histograms and immunoblot images depict 1 representative of n = 3 biological replicates generated in independent experiments. Bar plots are mean ± SEM of n = 3–4 biological replicates generated in independent experiments indicated as data points. Unpaired two-tailed t test performed in (A–C); ordinary two-way ANOVA with Dunnett’s multiple comparison test utilized in (D–F); *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001, ns = not significant. Source data are available online for this figure.

Figure EV2. Lysine residues in the TLR4 TIR domain were not required for LPS-induced TLR4 endocytosis.

(A–C) Tlr4^-/- BMMs retrovirally reconstituted with (A) mTLR4^WT, mTLR4^9KΔR, mTLR4^7KΔR, mTLR4^5KΔR or empty vector were stimulated with LPS (100 EU/mL); (B) TLR4 surface expression assessed at 120 and 240 min post-stimulation and (C) IL-6 and CXCL10 production assessed at 24 h post-stimulation. (D, E) RAW^TLR4ko cells were stably transfected with V5-tagged constructs of mTLR4^WT or mTLR4^9KΔR. (D) Total cellular expression of TLR4-V5 during LPS stimulation (100 EU/mL, 0, 120, 1240 min) was assessed via immunoblot. Phospho-TBK1 served as control for cellular activation; a-tubulin served as loading control. Black arrow head indicates low glycosylated intracellular TLR4, white arrow head indicates highly-glycosylated cell surface-expressed TLR4. (E) TLR4 surface expression was assessed in resting cells via flow cytometry. Data information: Immunoblot images depict 1 representative of n = 3 biological replicates generated in independent experiments. Flow cytometry histogram represents cells post sorting. Bar plots are mean ± SEM of n = 3–4 biological replicates generated in independent experiments indicated as data points. RM two-way ANOVA with Dunnett’s multiple comparisons test utilized in (B). Ordinary one-way ANOVA with Dunnett’s multiple comparisons test utilized in (C). *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001, ns = not significant. Source data are available online for this figure.

Figure EV3. PLCγ2 regulates TLR4 endocytosis.

(A) Impact of PLC inhibitor (U73122, 10 μM) on LPS- (100 EU/mL) and 1Z105-induced (10 µM) TLR4 endocytosis (120 min) in WT BMM. (B–D) CRISPR/Cas9-mediated knockdown of PLCγ2 expression in WT BMM (B) was verified by immunoblot analysis, and (C, D) shown to impair LPS-induced (100 EU/mL, 120 min) TLR4 endocytosis as assessed by flow cytometry. (E, F) Immunoblot analysis of the impact of TAK-242 (1 µM) or NSC694923 (10 µM) on total PLCγ2, phosphorylated NF-κB p65 and α-tubulin in WT BMM upon stimulation with (C) 1Z105 (10 µM) or (D) LPS (100 EU/mL) for the indicated times. (G) Comparison of PLCγ2 expression in unstimulated cells presented in (E, F). (H) Immunoblot analysis of PLCγ2 cellular expression in WT and Tlr4^−/− BMM treated with DMSO or TAK-242 (1 μM) for 120 min. Data information: Flow cytometry histograms and immunoblot images depict 1 representative of n = 3–4 biological replicates generated in independent experiments. Bar plots are mean ± SEM of n = 3–7 biological replicates generated in independent experiments indicated as data points. Unpaired two-tailed t test utilized in (A, D). Ordinary one-way ANOVA with Dunnett’s multiple comparisons test utilized in (E–G). *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001, ns = not significant. Source data are available online for this figure.