Recombinant mouse GRP78 was prepared as described in our previous report (11). Briefly, a plasmid encoding the full length of mouse GRP78 was transformed into Escherichia coli BL21 to generate glutathione-S-transferase (GST)-GRP78. Fusion protein was purified using Pierce^® Glutathione Spin Columns (16105; Thermo Scientific, Waltham, MA, USA). GRP78 was obtained by thrombin cleavage and identified by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting. Protein concentration was detected using a Bicinchoninic Acid Protein Assay kit (Beyotime, China). Endotoxins were removed by a Pierce High-capacity Endotoxin Removal Resin (88274; Thermo Scientific), and the final endotoxin concentration of protein samples was <10 EU/mg. Negative control (NC) extracts from empty vector-transformed E. coli BL21 were prepared in the same way.

C57BL/6 mice (HFK Bioscience, Beijing, China), C57BL/6 background TLR4 knockout (KO) mice (kindly donated by Professor Timothy R. Billiar, University of Pittsburgh, Pittsburgh, PA, USA), and CD14KO mice (The Jackson Laboratory, Bar Harbor, ME, USA) were bred in a specific pathogen-free facility and female mice were utilized at 6–8 weeks of age.

Bone marrow-derived dendritic cells from mice were generated as described previously (21). Cells were harvested after 7 days of culture with granulocyte-macrophage colony-stimulating factor (GM-CSF; 10 ng/mL; PeproTech, Rocky Hill, NJ, USA) and IL-4 (10 ng/mL; PeproTech). Bone-marrow-derived macrophages (BMDMs) were harvested after 7 days of culture with macrophage CSF (M-CSF; 50 ng/mL; PeproTech). HEK293T and DC2.4 cell lines were cultured in RPMI 1640 medium containing 10% fetal bovine serum.

Lipopolysaccharides (0111:B4) was obtained from Sigma-Aldrich (St. Louis, MO, USA). Antibodies specific for GST (sc-459) and β-actin (sc-47778) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Antibodies specific for GRP78 (ab32618), CD14 (ab182032), and TLR4 (ab22048) were obtained from Abcam (Cambridge, UK). Antibodies specific for interferon regulatory transcription factor (IRF)3 (4302) and IRF3 phosphorylated at Ser396(4947) were obtained from Cell Signaling Technology (Danvers, MA, USA). F(ab′)2-goat anti-mouse Alexa Fluor^®555 (AF555) was purchased from Thermo Fisher Scientific. Anti-TLR4-PE was obtained from PharMingen (Becton Dickinson, Franklin Lakes, NJ, USA). Anti-LAMP1-PE, anti-CD14-PECy7, and permeabilization buffer were obtained from eBioscience (San Diego, CA, USA).

Mouse CD14 cDNA was synthesized by Brookline Scientific (Shanghai, China) and cloned into the pcDNA3.1 vector (Invitrogen, Carlsbad, CA, USA). For transient transfection of plasmids into 293T cells, jetPRIME reagents (Polyplus-transfection^®, Illkirch-Graffenstaden, France) were used according to the manufacturer’s recommendations.

The labeling and purifying of Alexa Fluor^® 488-conjugated GRP78 (AF488-GRP78) was performed using a Protein Labeling kit (A10235; Molecular Probes, Eugene, OR, USA) according to the manufacturer’s recommendations, and AF488-BSA was used as an NC.

For flow cytometry (FCM) and immunofluorescence analyses, after blockade with 10% normal goat serum at 37°C for 30 min, cells were incubated with protein or antibodies at an appropriate temperature and time. Then, cells were detected by a flow cytometer (LSR II; BD Biosciences, Franklin Lakes, NJ, USA) or stained with Hoechst 33342 for cell nuclei before images were examined using a confocal laser scanning microscope (CLSM) (FV1000; Olympus, Tokyo, Japan). Images were analyzed by aFV10-ASW 2.1 Viewer (Olympus). The percentage of mean fluorescence intensity (MFI) for surface TLR4 was calculated using the following equation: % of surface TLR4 (MFI)=(GRP78 treatment−isotype)/(NC−isotype)

Membrane protein lysates from DC2.4 cells were extracted using a Membrane Protein Extraction kit (89842; Thermo Scientific). The cDNA encoding mouse GRP78 and fragments of GRP78 were cloned into the pGEX-4T3 vector. Recombinant fusion proteins of GST-GRP78, GST fragments of GRP78, and GST control (20 µg) were incubated with 500 µL DC2.4 cell membrane protein lysates (1 × 10⁶ cells) for 1 h on ice, followed by the addition of 50 µL glutathione agarose beads (16105; Thermo Scientific) for incubation overnight at 4°C with constant mixing. Protein complexes were washed extensively with buffers. The precipitates were analyzed by 10% SDS-PAGE and blotted using antibody specific for GST, GRP78, TLR4, and CD14.

The physical interaction between GRP78 and CD14 was determined by the protein–protein interactive precipitation assay. The cDNA encoding mouse CD14 was cloned into the pOptiVEC-hIgG1-Fc vector. Recombinant CD14-Fc and Fc control proteins were purified by protein G-agarose (sc-2002; Santa Cruz Biotechnology) from culture supernatants of HEK293T cells transfected transiently with relative plasmids, followed by incubation with GRP78 at 4°C overnight. Precipitates were analyzed by 10% SDS-PAGE and blotted using antibody specific for hIgG1-Fc, GRP78, and CD14.

Concentrations of IL-6, tumor necrosis factor (TNF)-α, and interferon (IFN)-β in culture supernatants were determined by ELISA kits from BioLegend (San Diego, CA, USA) following the manufacturer’s recommendations.

Cell total RNA was extracted with TriZol^®Reagent (Invitrogen) and cDNA was generated using a HiFiScript^® cDNA Synthesis kit (CW Biotech, Beijing, China). Quantitative real-time polymerase chain reaction (qRT-PCR) analyses were carried out using an SYBR Green Real-time PCR kit (Toyobo, Osaka, Japan) in a LightCycler^® (Bio-Rad Laboratories, Hercules, CA, USA). Data were normalized to glyceraldehyde 3-phosphate dehydrogenase, and fold changes were analyzed using the formula: 2^{−Δ Δ Ct}.

Mean values were compared using the Student’s t-test (two groups) or one-way ANOVA (three or more groups). Results are the mean and standard deviation (SD). All ELISA, qPCR, and fluorescence-activated cell sorting (FACS) experiments were performed thrice, and one representative result was presented. The immunofluorescence experiments and protein blots shown are representative data from at least three independent experiments.

To confirm the immunomodulatory properties of GRP78, the production of inflammatory cytokines in GRP78-treated BMDCs was measured. LPS stimulated BMDCs to release large quantities of pro-inflammatory cytokines such as TNF-α, IL-6, and IFN-β. However, such promotion could be attenuated by simultaneous supplementation of GRP78 in a concentration-dependent manner (Figure 1A). Even as little as 1 µg/mL of GRP78 could reduce IFN-β production significantly (P < 0.01). GRP78 (40 µg/mL) led to nearly complete abolishment of the expression of those cytokines. qRT-PCR results corroborated the data given earlier (Figure 1B). To exclude the possibility that GRP78 impaired LPS-induced cytokine expression by inhibiting the interaction between LPS and its receptor, instead of simultaneous treatment with LPS and GRP78, cells were treated with GRP78 first, and then rinsed extensively with culture medium (wash) or not (not wash) and, finally, stimulated with LPS. Expression of TNF-α and IFN-β in both groups was relatively lower than those in cells stimulated only by LPS (Figure 1C). A similar anti-inflammatory effect of GRP78 was observed in LPS-stimulated BMDMs (Figure 1D).

Modulatory effects of extracellular HSP70 members on innate immunity have been heavily debated, as neither the immunomodulatory receptors responsible for binding nor the intracellular signaling events have been established definitively (22, 23). To identify the functional mode of GRP78, we investigated whether GRP78 is bound by cell surface receptors or transported into cells.

Flow cytometry analyses showed that DC2.4 cells and BMDCs were stained by AF488-GRP78 but not AF488-BSA (Figures 2A,C). CLSM revealed that, at 4°C (to block vesicular trafficking), 90–95% of AF488-GRP78-stained cells emitted green fluorescence, whereas cells treated with AF488-BSA were not stained. The fluorescence of AF488 was localized only at the plasma membrane (Figures 2B,D). Moreover, when such staining was conducted at 37°C, fluorescence did not become “grainy” and, thereafter, no part of the grains were clustered (Figure 2E).

Collectively, these results suggested that it was through binding with cell-surface structures, not by being endocytosed, did GRP78 exert its modulatory effects.

TLR4 is a crucial “switch” for mediating LPS-induced signal transduction (24). In addition, TLR4-dependent pathways ultimately lead to the transcription of genes encoding various pro-inflammatory mediators (e.g., TNF, IL-6, and IL-1β) (18, 25, 26). After confirming that GRP78 could bind with unknown cell surface structures on DCs, we examined whether GRP78 could interact with TLR4 and, if so, its role in TLR4-triggered responses.

Dendritic cells were stained with GRP78 (green) for 30 min and then stained with TLR4 (red). Interestingly, as DCs were stained with GRP78 green fluorescence at the plasma membrane, TLR4 red fluorescence was localized beneath plasma, whereas in AF488-BSA-stained control cells, it was distributed evenly at the plasma membrane. Orange-to-yellow overlap fluorescence could not be observed in merged images (Figure 3A), so GRP78 and TLR4 were not co-localized. Furthermore, in GST precipitation assays, GST-GRP78 failed to precipitate TLR4 (Figure 3B). Therefore, GRP78 probably interacted not with TLR4, but with other partners.

Although GRP78 did not interact directly with TLR4, it did have roles in TLR4-triggered responses, because GRP78 showed dose-dependent inhibition to TNF-α in LPS-induced wild-type (WT) BMDCs (Figures 1 and 3C). Approximately 40 µg/mL GRP78 treatment nearly abrogated TNF-α production, which was in close proximity with no response in TLR4 KO BMDCs (Figure 3C). Hence, how did GRP78 attenuate TLR4-mediated responses?

Various negative regulatory mechanisms have evolved to attenuate the signaling of TLRs to sustain immunologic balance, including the reduction of TLR expression (18). It is believed that the endocytosis of plasma membrane-localized TLRs down-regulates their signaling functions after a microbial encounter for signal termination (27). To ascertain whether GRP78 could reduce cell surface expression of TLR4, the intensity of TLR4 on DCs before and after GRP78 treatment was assessed.

Confocal laser scanning microscopy (Figure 3D) showed that evenly distributed TLR4 fluorescence was reduced after GRP78 treatment. Using the loss of cell surface expression as a readout for TLR4 endocytosis (28), we observed that the MFI of cell surface TLR4 decreased dramatically as the GRP78 concentration increased (Figure 3E). This FACS-based assay was consistent with the anti-inflammatory functions of GRP78 (Figure 1). By counterstaining cells with an antibody against TLR4 and the late endosomal/lysosomal marker protein, LAMP-1, we observed a decrease in the expression of plasma membrane TLR4 and an increase in co-localization between intracellular TLR4 and LAPM-1 during 1 h of incubation with GRP78 (Figure 3F). We speculated that TLR4 was degraded in lysosomes after GRP78 stimulation.

Collectively, these data suggested that GRP78 could inhibit TLR4 signaling through the promotion of TLR4 endocytosis and reduction of its expression on the surface of DCs.

Although the GST precipitation assay did not support the notion of direct interaction of GRP78 with TLR4, GRP78 did reduce the cell surface expression of TLR4 to counteract LPS-induced production of pro-inflammatory cytokines. Given that CD14 is a crucial regulator of TLR4 endocytosis and is known to chaperone LPS molecules to TLR4 signaling complexes (28, 29), we tested whether GRP78 could interact with CD14 to regulate TLR4 endocytosis.

Unlike no co-localization of GRP78 and TLR4, double immunofluorescence staining showed that the pattern of staining of DC2.4 cells with CD14 red fluorescence reflected a high degree of co-localization with that of GRP78 green fluorescence (Figure 4A). Furthermore, GRP78 was found to stain the plasma membrane of 293T cells transfected with CD14, but failed to stain the 293T mock control (Figure 4B). Orange-to-yellow overlap fluorescence in merged images suggested the co-localization of GRP78 and CD14 on the plasma membrane. To ascertain whether GRP78 binds to CD14 directly, the GST precipitation assay was carried out again. CD14 was precipitated via binding to GST-GRP78, whereas TLR4 was not (Figure 4C). Recombinant fusion protein CD14-Fc was also used to determine the ability of CD14 to bind to GRP78 by the protein–protein interactive precipitation assay. Recombinant GRP78 was precipitated via binding to CD14-Fc, but control Fc was not precipitated (Figure 4D).

GRP78 contains two important domains: an adenosine triphosphate (ATP)-binding domain and a substrate-binding domain. The former is required for ATPase catalytic activity and the latter is responsible for binding to the substrate protein (1). To ascertain which domain was required for interaction with CD14, GST fusion proteins containing various fragments of GRP78 were used to carry out the GST precipitation assay. Surprisingly, none of the truncated GRP78 interacted with CD14, and only full-length GRP78 retained the ability to bind to CD14 (Figure 4E).

Collectively, this series of experiments revealed that GRP78 interacted with CD14 using its complete conformation.

As a TLR4 accessory protein, CD14 functions not only in ligand transport but also in receptor transport to endosomes (28, 29). To determine the effect of binding of CD14 by GRP78 on TLR4 transport, BMDCs from WT and CD14KO mice were monitored for cytokine production and TLR4 endocytosis.

PCR identification of the genomic DNA and FCM analyses confirmed the CD14 deficiency in CD14KO mice (Figure 5A). As expected, in the absence of CD14, the concentrations of LPS needed to activate TNF-α production increased by several orders of magnitude (Figure 5B) and AF488-GRP78 showed no significant difference in staining BMDCs compared with the bovine serum albumin control. However, the percentage and MFI of AF488-GRP78⁺ cells from WT mice were much higher than those from CD14KO mice (Figure 5C). In subsequent experiments in CD14KO BMDCs, no alterations were found in surface expression of TLR4 which was down-regulated previously by GRP78 in WT BMDCs in a concentration-dependent manner (Figure 5D). Furthermore, CD14KO BMDCs exhibited no changes in TNF-α production upon treatment with GRP78 and LPS (Figure 5E). However, inflammatory endocytosis of TLR4 mediated by CD14 can activate TIR-domain-containing adapter-inducing interferon-β (TRIF)-IRF3 signaling following LPS stimulation (28), we found GRP78 did not induce IRF3 phosphorylation and inhibited the activation of IRF3 triggered by LPS (Figure 5F), suggesting that GRP78 induced the non-inflammatory endocytosis of TLR4.

Collectively, these results suggested that the immunomodulatory property of GRP78 in LPS-induced TLR4 signaling was dependent upon CD14.