Src family kinase tyrosine phosphorylates Toll-like receptor 4 to dissociate MyD88 and Mal/Tirap, suppressing LPS-induced inflammatory responses

2.1. SFK-Src R388A and D386N mutant constructs

SFK-Src R388A/Y527F (Src-R/A) and SFK-Src D386N/Y527F (Src-D/N) encoding plasmid constructs were kindly provided by Dr. Philip A. Cole (The Johns Hopkins University, Baltimore) (21). They were subcloned into pcDNA3.1/Hygro (-) vector (Thermo Fisher Scientific, Grand Island, NY) and stably transfected in HEK293 cell or SW480 cells (ATCC, Manassas, VA).

2.2. SFK-Lyn constructs

Wild type murine Lyn cDNA was PCR amplified with primers F1 and R1 and subcloned into Not1 and BamH1 sites of p3×Flag-CMV14 expression vector (Sigma-Aldrich, St. Louis, MO) to obtain 3×Flag-tagged Lyn-WT. The tyrosine at 508 residue was replaced with phenylalanine by PCR with the specific primer set (F2 and R2), and EcoR1/BamH1 fragment of the PCR product replaced the C-terminal region of 3×Flag-Lyn-WT to give rise to 3×Flag-Lyn (Y508F). Although the F2 primer contains a mutation at Tyr397 residue, subcloning of EcoR1/BamH1 fragment results in only an Y508F mutation because the Tyr397 residue is outside of the subcloning sites.

Furthermore, the PCR product with the primer F2 and R2 containing an Y397F mutation was also used as a mega-reverse primer with the forward primer (F1) for PCR. The Xba1/EcoR1 fragment of the PCR product was inserted into the same restriction sites in 3×Flag-Lyn-WT to produce the 3×Flag-Lyn (Y397F) construct that is a kinase-dead mutant. Then, the Xba1/EcoR1 fragment from 3×Flag-Lyn (Y397F) replaced the region in 3×Flag-Lyn (Y508F) to prepare 3×Flag-Lyn (Y397/508F) which is in a kinase-dead open conformation. We confirmed the integrity of the complete cDNA sequence by sequencing all the constructs, and the expression of Lyn constructs was verified by western blotting. All PCR primers used for generating Lyn constructs are the followings: F1, AAA GCG GCC GCA CCC ATG GGA TGT ATT AAA TCA AAA AGG; R1, G ATG AAA GGA TCC CGG TTG CTG CTG ATA CTG CCC; F2, C GAA GAT AAC GAG TTC ACA GCA AGG GAA GG (bold nucleotides indicate a mutated codon of Y397F); R2, G ATG AAT GGA TCC CGG TTG CTG CTG AAA CTG CCC-3″ (bold nucleotides stand for a mutated codon of Y508F);

2.3. Other cDNA expression constructs

We previously described HA-tagged murine TLR4 wild type and P712H mutant cDNA expression constructs and the constitutively active HA-ΔTLR4-WT and dominant negative HA-ΔTLR4(P712H) lacking the extracellular LRR domain (22). We used TLR5-HA (23) and Tram-Flag (24) encoding constructs previously. The NFκB-Luciferase reporter construct and β-galactosidase expression plasmid (HSP70-β-gal) as an internal control were previously described (25).

2.4. Cell culture

Human colonic epithelial cells (NCM460) were cultivated in M3D media from ‘INCELL Corp’ (San Antonio, TX), as previously described (26). Human embryonic kidney (HEK293) and murine macrophages RAW264.7 cells were cultivated as described previously at 37°C in 5 % CO₂ air environment (22). Human colon epithelial cells SW480 (ATCC^® CCL-228™) were cultivated in Leibovitz's L-15 Medium in ATCC (Manassas, VA) supplemented with fetal bovine serum (FBS) to a final concentration of 10% and Penicillin (50 units/mL)-Streptomycin (50 μg/mL).

2.5. Cell transfection

Cells were transfected with the appropriate plasmid DNA using SuperFect transfection reagent from Qiagen (Valencia, CA). About 18 hours after transfection, cell lysates were prepared in lysis buffer, followed by western blot analysis. To generate stably transfected cells, transfected cells were plated in the proper medium containing selecting antibiotics (zeocin or hygromycin or neomycin) to obtain stably transfected cell pools.

2.6. Antibodies and materials

Anti-phospho-tyrosine antibody (4G10) conjugated with HRP was purchased from EMD Millipore (Billerica, MA). Antibodies of phospho-Src (Y416), Src, phospho-ERK1/2, phospho-JNK1/2, phospho-p105 (NFκB), and phospho-p65 (NFκB) were purchased from Cell Signaling Technology (Danvers, MA). Mal/Tirap and MyD88 antibody (HFL-296) were from Invitrogen and Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), respectively. HA antibody (12CA5) was from Roche Applied Science (Indianapolis, IN). Flag (M2) antibody and imidazole were from Sigma-Aldrich (St. Louis, MO). Ultrapure LPS from E. coli K12 and endotoxin-free water were purchased from InvivoGen (San Diego, CA).

2.7. Western blot (immunoblot) analysis

Cell lysates were prepared in lysis buffer [150 mM NaCl, 50 mM Tris-Cl (pH 8.0), 5 mM EDTA, 1% Nonidet P-40] containing a protease inhibitor cocktail (Roche, Indianapolis, IN) and a phosphatase inhibitor cocktail (Sigma-Aldrich, St. Louis, MO). The equal amount of total protein was subjected to SDS-PAGE analysis, and immunoblotting with the appropriate antibodies was performed as previously described (25). For immunoblotting with anti-phospho-tyrosine antibody (4G10), transferred PVDF blot was incubated with HRP-conjugated 4G10 antibody (1:2,500) in 3% IgG-free BSA-containing TBS mixed with Tween 20 (0.05%) for 1 hour with shaking. After washing three times, the antigen-antibody complexes were visualized by enhanced chemiluminescence (ECL).

2.8. Immunoprecipitation assay

Total cell lysates were prepared in lysis buffer as described above. Equal amounts of total protein were mixed with monoclonal anti-HA–agarose antibody and incubated at 4 °C for 4-5 hours. The precipitants were washed three times with lysis buffer and fractionated in SDS-PAGE, followed by the immunoblotting procedure described above.

2.9. Measuring IL-6 and IL-8 production

An enzyme-linked immunosorbent assay (ELISA) was performed to measure the level of the cytokine expression using the appropriate kits from R&D Systems (Minneapolis, MN), following the manufacturer's instructions. All assays were performed in triplicate, and data were shown as mean + SEM.

2.10. Luciferase reporter assays

RAW264.7 cells were plated in six-well plates (0.5 × 10⁶ cells/well) and co-transfected with NFκB-luciferase reporter and Src-R/A constructs, including a β-galactosidase expression plasmid (HSP70-β-gal) as an internal control, using SuperFect transfection reagent (Qiagen, Valencia, CA), according to the manufacturer's instructions. One day after transfection, cells were stimulated with LPS and imidazole for 6 hours, and the relative luciferase activity was determined by the normalization with β-galactosidase activity as previously described (25). The total amount of plasmid DNA was kept consistent by adding the empty vector for each transfection. All assays were performed in triplicate. Data were shown as mean + SEM.

3.1. Chemical rescue of SFK activity induces tyrosine phosphorylation of TLR4, but not TLR5

It has been suggested that a protein tyrosine kinase containing active-site mutations can be restored by the functional substitution of the missing side chain with a small molecule like imidazole. The small molecule may effectively stand in for the side chain of a mutated amino acid if it contains similar electronic or steric features (20,21). Arg388, found in SFK-Src, is a conserved residue in protein tyrosine kinases. Qiao et al. showed that when the Arg388 residue in SFK-Src was mutated to Ala (Src R388A), the catalytic activity could be effectively rescued by imidazole (21). In this way, the chemical rescue approach allows highly specific and precise temporal activation of SFKs in living cells (20,21). In studying the role of SFKs, the chemical rescue approach has a great advantage over gene silencing or gene knockout systems, in which surviving cells or genetically modified mice might have become adapted to a SFK gene-deficient environment, and consequently adopted a compensatory or redundancy mechanism using other SFK members to overcome the gene deficiency (19).

The tyrosine phosphorylation at the C-terminal regulatory tail of SFKs by C-terminal Src kinase (Csk) results in a closed conformation, leading to an inactive form of the enzyme (Fig. 1A). In contrast, dephosphorylation at the tyrosine residue leads to an open conformation, giving rise to a constitutively active SFK (Fig. 1B) (27,28). However, the presence of the C-terminal Y527F mutation in SFK-Src prevents inhibitory phosphorylation by Csk. When this mutation is present, the catalytic activity of the SFK-Src R388A/Y527F (Src-R/A) mutant can be rescued by imidazole treatment, while preventing inhibitory phosphorylation by Csk. On the other hand, the SFK-Src Asp386→Asn (D386N) mutant is a catalytically defective form of Src and is insensitive to imidazole (21). Thus, SFK-Src D386N/Y527F (Src-D/N) was used as control.

Based on this background, we used the rescue approach of SFKs to investigate whether inducible temporal activation of SFKs would cause the tyrosine phosphorylation of TLR4. HEK293 cells were stably co-transfected with SFK-Src-R/A or SFK-Src-D/N in conjunction with TLR4-HA encoding construct. Our data showed that imidazole treatment clearly induced SFK-Src activation (evaluated by phospho-Src) in Src-R/A expressing cells, while Src-D/N expressing cells did not exhibit Src activation after imidazole treatment (Fig. 2A). In line with imidazole-rescued SFK activation, the immunoprecipitation results showed that imidazole treatment induced the tyrosine phosphorylation of TLR4 in Src-R/A cells, while TLR4 tyrosine phosphorylation was not observed in Src-D/N cells. Notably, the overexpression of SFK-Src-R/A or SFK-Src-D/N alone did not induce TLR4 tyrosine phosphorylation when imidazole was absent. We confirmed that the level of total Src (pan-Src) and TLR4 expression was similar between the samples. Therefore, these data indicate that temporal and specific SFK activation induces tyrosine phosphorylation in TLR4.

Similar to TLR4, TLR5 is a type-I transmembrane protein that recognizes bacterial flagellin and associates with MyD88 and Mal/Tirap to mediate inflammatory responses (26,29). We next tested whether rescuing SFK activity would also induce the tyrosine phosphorylation of TLR5. Since intestinal epithelial cells strongly express TLR5 and exhibit potent intracellular signaling following the binding of flagellin (24,29), human colonic epithelial cells (NCM460) were co-transfected with SFK-Src-R/A or SFK-Src-D/N with TLR5-HA constructs. In NCM460 cells transfected with SFK-Src-R/A, imidazole treatment induced Src activation, whereas SFK-Src-D/N expressing NCM460 cells did not exhibit SFK activation upon imidazole treatment (Fig. 2B, lane 1 to 6). However, the immunoprecipitation data showed that restored SFK activation did not induce the tyrosine phosphorylation of TLR5, while, as a positive control, restored SFK-Src activity in HEK293 cells still elicited TLR4 tyrosine phosphorylation (Fig. 2B, lane 7). It is worth noting that little difference was otherwise observed between imidazole-treated and vehicle-treated HEK293 or NCM460 cells. Moreover, when imidazole was absent, SFK-Src activation was not observed in HEK293 or NCM460 cells expressing Src-R/A or Src-D/N. Together, these data demonstrate that temporally restored SFK activation induces the tyrosine phosphorylation of TLR4, but not TLR5.

3.2. Rescued SFK catalytic activity inhibits LPS-induced inflammatory responses

Next, we examined a functional outcome of SFK activation in TLR4-mediated signaling pathways. Human colonic epithelial SW480 cells are responsive to LPS with functionally active endogenous TLR4 expression (30). Therefore, we stably transfected SW480 cells with the SFK-Src-R/A encoding construct to generate SW480-Src-R/A cells. SW480-Src-R/A cells were then treated with LPS in conjunction with imidazole or vehicle. Due to the preserved endogenous TLR4-mediated signaling, LPS stimulation was able to induce ERK1/2, JNK1/2, and NFκB (represented by phosphorylated p105 and p60) activation in SW480-R/A cells without imidazole treatment (Fig. 3A). As an immediate signaling event, the activation of SFK-Src was induced by LPS stimulation in a time-dependent manner in SW480-R/A cells without imidazole treatment. However, imidazole treatment enhanced LPS-induced SFK-Src activation. In addition, LPS-induced ERK1/2 activation was further increased by imidazole treatment. Because SFK activation is able to induce ERK1/2 activation in SW480 cells (31), these signaling events indicate that SFK activity was successfully rescued. Thus, we believe that the combination of rescued SFK activity and LPS stimulation led to potentiated SFK-Src and ERK/12 activation. However, imidazole treatment inhibited LPS-stimulated JNK1/2 and NFκB activation in SW480-Src-R/A cells. We confirmed that total Src expression was similar between the samples, and that Src-R/A expression itself induced no difference in MAPKs and NFκB activation in the absence of imidazole.

Subsequently, with the transient co-transfection of NFκB-luciferase reporter and Src-R/A constructs into murine macrophages RAW264.7 cells, we found that imidazole treatment substantially reduced LPS-induced NFκB reporter activity (Fig. 3B). Furthermore, production of the LPS-stimulated cytokines IL-6 and IL-8 was markedly reduced in imidazole-treated Src-R/A cells compared to vehicle-treated Src-R/A cells (Fig. 3C and D). Imidazole treatment itself did not influence NFκB reporter activity or cytokine production.

Together, our results demonstrate that restored SFK activation inhibits LPS-induced JNK1/2 and NFκB activation, while it enhances ERK1/2 activation. Due to its inhibitory effects on JNK1/2 and NFκB activation, SFK activation diminishes cytokine production in LPS-induced inflammatory responses.

3.3. Restored SFK activation disrupts the association of MyD88 and Mal/Tirap with TLR4, butTLR4-Tram association is preserved

In order to decipher the mechanism by which SFK activation can inhibit LPS-induced responses, HEK293-Src-R/A cells were treated with LPS in the presence or absence of imidazole, followed by an immunoprecipitation assay. In the absence of imidazole, LPS stimulation induced Src activation in a time-dependent manner in Src-R/A cells; however, the presence of imidazole substantially potentiated LPS-induced SFK-Src activation, indicating that Src activation was restored by the small molecule (Fig. 4A). As expected, in the absence of imidazole, LPS stimulation induced the association of endogenous MyD88 and Mal/Tirap adaptors with TLR4. However, the interaction of MyD88 or Mal/Tirap with TLR4 was substantially reduced by imidazole treatment. Notably, LPS-induced interaction of Tram with TLR4 was preserved in vehicle- or imidazole-treated Src-R/A cells (Fig. 4B). These data indicate that the induced SFK catalytic activity disrupted the association of MyD88 and Mal/Tirap with TLR4, whereas SFK activation did not alter LPS-induced TLR4-Tram interaction.

Together with the finding that rescued SFK activity induces the tyrosine phosphorylation of TLR4 (Fig. 2), our immunoprecipitation data suggest that TLR4 tyrosine phosphorylation by activated SFKs suppresses LPS-induced responses by dissociating MyD88 and Mal/Tirap from TLR4. Therefore, SFK activation should be regarded as a negative regulator in LPS-induced inflammatory responses.

3.4. The catalytic activity of SFKs is required to induce tyrosine phosphorylation of TLR4

We previously demonstrated that a murine TLR4 mutant with the deletion of a large part of the ectodomain (ΔTLR4) becomes a constitutively active TLR4 (22). In human TLR4 (hTLR4), the CD4-hTLR4 chimera renders hTLR4 constitutively active (32). Like the LPS-induced tyrosine phosphorylation in the full-length wild type of TLR4, the constitutively active TLR4 (both ΔTLR4 and CD4-hTLR4) is tyrosine phosphorylated when expressed in cells; however, this occurs even in the absence of LPS stimulation (5). The P712H mutation and equivalent P714A mutation in the cytoplasmic TIR domain of murine and human TLR4, respectively, not only inhibits TLR4 signaling (22,33), but also diminishes the tyrosine phosphorylation of TLR4 (5).

Among the nine SFK members, SFK-Lyn is highly expressed in macrophages/monocytes, DCs, and B cells (34). It was shown that SFK-Lyn is immediately activated by LPS stimulation and physically interacts with TLR4 in response to LPS (5). LPS induces autophosphorylation of SFK-Lyn in wild type macrophages, but not in macrophages with the TLR4 (P712H) mutation (9). In agreement with these findings, we observed that the constitutively active ΔTLR4-WT recruits SFK-Lyn in the absence of LPS stimulation, while this association is disrupted by the P712H mutation in TLR4 (Fig. 5A).

Subsequently, we investigated whether the catalytic activity of SFKs is responsible for inducing TLR4 tyrosine phosphorylation. We co-transfected HEK293 cells with a constitutively active TLR4 (HA-ΔTLR4-WT) together with a kinase-dead form of SFK-Lyn (Y397/508F). In contrast to the elevated tyrosine phosphorylation seen in the constitutively active ΔTLR4, expression of the kinase-dead SFK-Lyn (Y397/508F) practically eliminated the tyrosine phosphorylation of ΔTLR4, as revealed by the immunoprecipitation assay (Fig. 5B). However, tyrosine phosphorylation of ΔTLR4 was augmented by the expression of active SFK-Lyn (Y508F) (Fig. 5C, lane 3 and 4). Notably, the low levels of tyrosine phosphorylation observed in ΔTLR4 (P712H) were also increased by the expression of active SFK-Lyn (Y508F) (Fig. 5C, lane 5 and 6).

Our data clearly indicate that the kinase-dead mutant of SFK-Lyn eliminates tyrosine phosphorylation of TLR4, whereas the kinase-active SFK-Lyn enhances TLR4 tyrosine phosphorylation. Therefore, the catalytic activity of SFK is required to induce TLR4 tyrosine phosphorylation.

3.5. The open conformation of SFKs is important for TLR4 interaction, and SFK-active kinase activity promotes this interaction

The phosphorylation of either of two tyrosine residues, one found in the kinase domain and the other in the C-terminal tail region, exerts opposing regulatory effects on the activity of SFKs. A phosphorylated tyrosine at the C-terminal tail of SFK binds to its SH2 domain to lock the molecule into an inactive closed conformation, in which an essential motif for the catalytic activity is buried (Fig. 1A). Conversely, dephosphorylation at the C-terminal tyrosine residue leads to an open conformation, exposing the tyrosine residue within the kinase domain for phosphorylation in order to reconstitute fully active tyrosine kinase activity (Fig. 1B) (35,36).

These two tyrosine residues are conserved in all SFKs. In SFK-Lyn, Tyr508 at the C-terminal region is key to generating either the inactive closed or active open conformation. Thus, the Y508F mutation in SFK-Lyn prevents phosphorylation, instead forcing an open conformation. Subsequent autophosphorylation of Tyr397 in the activation loop of the kinase domain renders SFK-Lyn enzymatically active (37). Thus, the presence of both Y508F and Y397F mutations gives rise to a kinase-dead open conformation of SFK-Lyn.

Within this context, we next examined whether the catalytic activity of SFK-Lyn is required for interaction with TLR4. Compared to the interaction between ΔTLR4-WT and SFK-Lyn-WT (Fig. 5D, lane 7), ΔTLR4-WT:SFK-Lyn (Y508F) interaction was greatly enhanced (Fig. 5D, lane 4). Just like the disrupted interaction observed between ΔTLR4 (Y712H) and SFK-Lyn-WT (Fig. 5A, lane 5), the interaction between ΔTLR4 (Y712H) and SFK-Lyn (Y508F) was dramatically reduced compared to the strong ΔTLR4-WT: SFK-Lyn (Y508F) interaction (Fig. 5D, lane 4 and 5). Importantly, elevated tyrosine phosphorylation was identified in SFK-Lyn (Y508F), which exhibited a strong interaction with ΔTLR4-WT (Fig. 5D, lane 4), indicating that the key Tyr397 residue in the kinase domain of SFK-Lyn (Y508F) was phosphorylated. Together, these data suggest that the open conformation of SFK-Lyn has a greater capacity to interact with TLR4 than the wild type SFK-Lyn.

Next, we tested whether it is necessary for the SFK to be kinase-active in order to interact with TLR4. To investigate, we used an Y397F mutation in SFK-Lyn. Tyr397 in SFK-Lyn is a conserved residue within the kinase domain which must be phosphorylated for the enzyme to be active; therefore, this mutation rendered SFK-Lyn inactive. Moreover, an Y508F mutation causes SFK-Lyn to adopt an open conformation, which is normally followed by autophosphorylation at Tyr397 to confer enzymatic activity. Therefore, we used a SFK-Lyn (Y397/508F) double mutant which held an open conformation but was enzymatically inactive. We found that SFK-Lyn (Y397/508F) was still able to interact with ΔTLR4-WT (Fig. 6, lane 4). However, the degree of ΔTLR4-WT:SFK-Lyn (Y397/508F) interaction was dramatically reduced compared to the strong interaction between ΔTLR4-WT and SFK-Lyn (Y508F) (Fig. 6, lane 5). It is worth noting that, as expected, SFK-Lyn (Y397/508F) interacting with ΔTLR4-WT did not exhibit tyrosine phosphorylation, while SFK-Lyn (Y508F) strongly interacting with ΔTLR4-WT revealed strong tyrosine phosphorylation. These data indicate that the tyrosine phosphorylation detected in SFK-Lyn (Y508F) occurs at Tyr397 in the kinase domain.

Taken together, our data clearly indicate that the open conformation of SFK promotes the TLR4-SFK interaction, and that the kinase activity of SFK greatly potentiates the interaction between TLR4 and SFK-Lyn.