Recombinant Thrombomodulin Inhibits Lipopolysaccharide-Induced Inflammatory Response by Blocking the Functions of CD14

CD14, a GPI-anchored membrane protein, is a receptor for LPS (1). It transfers LPS to the TLR4/myeloid differentiation factor-2 (MD-2) complex and elicits downstream signaling pathways, including the MAPK, NF-κB, and IFN regulatory factor 3 (IRF3) pathways, resulting in the production of proinflammatory cytokines and type I IFNs (1–4). Recent studies reported that CD14 is involved in the activation of TLR2, TLR3, TLR7, and TLR9 because CD14-deficient macrophages and dendritic cells display reduced inflammatory responses to the specific ligands of these TLRs (5–7). Soluble forms of CD14 also were detected in circulating blood and facilitate LPS-induced inflammatory response in endothelial and epithelial cells that do not express the membrane form of CD14 (8). These results suggest that CD14 is a critical pattern-recognition receptor in the innate immunity against a broad spectrum of ligands (9, 10).

Originally, thrombomodulin (TM) was identified as an anticoagulant factor that activates protein C (11), but recent reports suggest that TM is involved in biological processes in addition to hemostasis, including cell–cell adhesion, epithelial–mesenchymal transition, and inflammation (12–14). TM consists of a C-type lectin-like domain (domain 1), a domain with six epidermal growth factor (EGF)-like structures (domain 2), a serine/threonine-rich domain (domain 3 [D3]), a transmembrane domain (domain 4), and a cytoplasmic domain (domain 5) (15). Soluble forms of TM are also reportedly high in the plasma of septic patients (16). Because TM is a natural anticoagulant protein, recombinant human soluble TM protein (ART-123) effectively reduces disseminated intravascular coagulation (17). Furthermore, recombinant TM lectin-like domain (rTMD1) suppresses LPS-induced inflammation by binding directly to LPS and high-mobility group box 1 protein (18, 19). Although the anti-inflammatory activity of activated protein C has been demonstrated (20), the pulmonary immune responses to respiratory pathogens and LPS in mice with strongly reduced protein C activation (TM^pro/pro mice) are not different from those in wild-type mice (21), suggesting that TM can modulate the host inflammatory response through a protein C–independent mechanism.

The expression of TM in monocytes and macrophages has been observed, but the function of monocytic TM remains unclear. Our previous work demonstrated that monocytic TM recognizes LPS through its lectin-like domain and interacts with the CD14/TLR4/MD-2 complex. Myeloid-specific TM-deficient mice (LysMcre/TM^flox/flox mice) display reduced inflammatory response to Klebsiella pneumoniae infection and cecal ligation and puncture, indicating that monocytic TM is involved in LPS- and Gram-negative bacteria–induced inflammation (14).

We also showed that rTMD1, but not the recombinant EGF-like domain plus serine/threonine-rich domain of TM (rTMD23), binds to LPS via the Lewis Y Ag (18). Because rTMD23 has six EGF-like domains, it acts as a novel angiogenic factor by facilitating endothelial cell proliferation and migration (22). rTMD23 also suppresses atherosclerosis in apolipoprotein E–deficient mice by binding to thrombin and inhibiting thrombin-induced endothelial cell activation (23). However, the function of rTMD23 in the LPS-induced inflammatory response remains unclear. In the current study, we investigated the function of rTMD23 in LPS-induced inflammation.

Methods for the preparation of the TM domains were described previously (18, 22, 24). Briefly, pPICZα-A and pCR3-EK vectors (Invitrogen) were used to express and secrete recombinant TM domains from Pichia pastoris and HEK293 mammalian protein expression systems. The amino acid contents of TM domains are as follows: rTMD1 (Ala¹-Ala¹⁵⁵), rTMD12 (Ala¹-Cys⁴⁶²), rTMD123 (Ala¹-Ser⁴⁹⁷), rTMD23 (Ala²²⁴-Ser⁴⁹⁷), recombinant TM EGF-like structure (rTMEGF) 2 (rTMEGF2-6D3) (Ala²⁶⁶-Ser⁴⁹⁷), rTMEGF3-6D3 (Asp³⁰⁷-Ser⁴⁹⁷), rTMEGF4-6D3 (Pro³⁴⁷-Ser⁴⁹⁷), rTMEGF5-6D3 (Cys³⁸⁶-Ser⁴⁹⁷), rTMEGF6D3 (Asp⁴²³-Ser⁴⁹⁷), and rTMD2 (Ala²²⁴-Cys⁴⁶²). To prepare mutant rTMD23, which cannot activate protein C, we mutated the thrombin-binding site (I424A; rTMD23^I424A) using a QuikChange Site-Directed Mutagenesis Kit (Stratagene) (25). All recombinant TM proteins contained His-tag for purification and c-Myc tag for detection.

C57BL/6 mice (8−12 wk old) were injected i.p. with 4% thioglycollate. After 4 d, peritoneal macrophages were obtained from the mice, as previously described (14). For stimulating the mouse peritoneal macrophages, Pichia-expressed rTMD23 (prTMD23) or mammalian-expressed rTMD23 (mrTMD23) was incubated with macrophages for 30 min at 37°C before LPS (Escherichia coli, O111:B4; Sigma-Aldrich) stimulation. prTMD23 and mrTMD23 were inactivated by heat in the presence of 2-ME. Similar procedures were conducted using Pichia-expressed rTMD23^I424A (prTMD23^I424A), mammalian-expressed rTMD1 (mrTMD1), mammalian-expressed recombinant TMD12 (mrTMD12), mammalian-expressed recombinant TMD123 (mrTMD123), mammalian-expressed recombinant TMEGF2-6D3 (mrTMEGF2-6D3), mammalian-expressed recombinant TMEGF3-6D3 (mrTMEGF3-6D3), mammalian-expressed recombinant TMEGF4-6D3 (mrTMEGF4-6D3), mammalian-expressed recombinant TMEGF5-6D3 (mrTMEGF5-6D3), mammalian-expressed recombinant TMEGF6D3 (mrTMEGF6D3), and mammalian-expressed recombinant TMD2 (mrTMD2). After 24 h of LPS stimulation, culture media were harvested, and mouse TNF-α and IL-6 levels were determined using ELISA kits (R&D Systems). In some experiments, mrTMD23 and LPS were incubated simultaneously with macrophages for 24 h at 37°C.

For the stimulation of HUVECs (Invitrogen), recombinant human CD14 (rCD14; R&D Systems) was incubated with prTMD23 or prTMD23^I424A for 30 min at 37°C. LPS, CpG-oligodeoxynucleotide (ODN) (InvivoGen), rCD14/prTMD23, and rCD14/prTMD23^I424A were incubated with HUVECs for 24 h at 37°C. rCD14 was also incubated with mrTMD1, mrTMD12, or mrTMD123 for 30 min at 37°C. rCD14 and prTMD23 were inactivated by heat in the presence of 2-ME. The culture media were harvested for estimating human IL-6 levels with an ELISA kit (R&D Systems).

For analysis of cytokine production in CD14-knockdown macrophages, mouse peritoneal macrophages were transduced with recombinant lentiviruses containing a CD14-specific short hairpin RNA (targeting sequence: 5′- TATCAAGTCTCTGTCCTTAAA-3′) or a luciferase-specific short hairpin RNA (targeting sequence: 5′-TCACAGAATCGTCGTATGCAG-3′) and incubated with Polybrene (8 μg/ml) overnight at 37°C. All lentiviral vectors, as well as the recombinant lentivirus production system, were obtained from the National RNAi Core Facility (Academia Sinica, Taipei, Taiwan), and the preparation of recombinant lentiviruses was performed as previously described (14). The cells were incubated or not with prTMD23 for 30 min at 37°C. Following 24 h of LPS stimulation, the culture media were harvested, and the mouse TNF-α and IL-6 levels were determined using ELISA kits (R&D Systems).

Mouse peritoneal macrophages were stimulated with LPS for 30 min (to analyze the phosphorylation of ERK1/2) or for 60 min (to analyze the degradation of IκB-α and the phosphorylation of IRF3) after pretreatment with prTMD23, mrTMD23, heat-inactivated prTMD23, or heat-inactivated mrTMD23 for 30 min at 37°C. Total cell lysates were harvested with cell lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 1% Nonidet P-40, and 1 mM PMSF [pH 8]), and the downstream effectors of LPS signaling were analyzed with Abs against IκB-α (Cell Signaling Technology), total IRF3 (Cell Signaling Technology), phospho-IRF3 (Cell Signaling Technology), phospho-ERK1/2 (Santa Cruz Biotechnology), total ERK1/2 (Santa Cruz Biotechnology), and α-tubulin (Santa Cruz Biotechnology). Relative signal intensity of proteins was analyzed using AlphaImager 2200 software.

To analyze the binding capabilities of prTMD23 and prTMD23^I424A for macrophages, we incubated prTMD23 or prTMD23^I424A with mouse peritoneal macrophages for 30 min at 4°C. The cells were stained with Alexa Fluor 488–conjugated mouse anti–c-Myc Ab (Cell Signaling Technology), and Alexa Fluor 488–conjugated mouse IgG was used as a negative control. For the LPS-binding assay, mouse peritoneal macrophages were incubated with prTMD23 or prTMD23^I424A for 30 min at 4°C, washed with PBS, and treated with FITC-LPS (E. coli, O111:B4; Sigma-Aldrich) for 30 min at 4°C. Following LPS binding, the prTMD23 and prTMD23^I424A molecules bound to macrophages were stained with an Alexa Fluor 488–conjugated mouse anti–c-Myc Ab (Cell Signaling Technology). The dead cells and cell debris were excluded by gating the forward and side scatter. The cells were analyzed by FACS (BD Biosciences), and the histogram and geometric mean fluorescence intensity were measured using WinMDI 2.9 software.

To examine the membrane form of CD14 on HUVECs and macrophages, we stained HUVECs and mouse peritoneal macrophages with FITC-labeled mouse anti-human CD14 Ab (BioLegend) and FITC-labeled rat anti-mouse CD14 Ab (eBioscience) for 30 min at 4°C. FITC-labeled mouse IgG and rat IgG were used as negative controls. Cells were analyzed by FACS (BD Biosciences), and the histogram was measured using WinMDI 2.9 software.

Mouse peritoneal macrophages were incubated with prTMD23 or prTMD23^I424A for 30 min at 37°C. The cells were washed and stimulated with LPS for 30 min at 37°C. Cell lysates were harvested and incubated with rabbit anti-mouse CD14 Ab (Abnova); rabbit IgG served as a negative control. The immune complexes were precipitated overnight using agarose A beads at 4°C. The beads were washed with PBS containing 0.05% Tween-20 and analyzed using Western blotting. For estimating the total levels of CD14 and TM, Abs against CD14 and β-actin (Santa Cruz Biotechnology) were used; the rabbit anti-mouse TM Ab was generated in our laboratory, as previously described (14).

rCD14 was incubated with prTMD23 or prTMD23^I424A in tubes containing 0.5% BSA/0.05% Tween-20/PBS for 30 min at 37°C. The mixtures were added to wells immobilized with mouse anti-human CD14 Ab (R&D Systems). For assessing the binding of CD14 and prTMD23 in vivo, C57BL/6 mice (8−12 wk old) were injected i.p. with prTMD23 and then injected with LPS i.p. (20 mg/kg) 30 min later. Mouse plasma was harvested 12 h after LPS stimulation and was added to wells immobilized with rat anti-mouse CD14 Ab (R&D Systems). Recombinant proteins were identified using biotinylated rabbit anti–c-Myc Ab (Cell Signaling Technology). The absorbance was analyzed at 450 nm after incubation with HRP-conjugated streptavidin and substrate development. To perform surface plasmon resonance analysis, mrTMD23 (10 μM) was diluted in HBS-P buffer (0.01 M HEPES, 0.15 M NaCl, 0.005% Surfactant P20 [pH 7.4]) and immobilized on NTA sensor chips. rCD14 (R&D Systems) that had been diluted in HBS-P buffer was passed over the immobilized NTA chips. NTA sensor chips and reagent kits were purchased from GE Healthcare, and the analyses were performed using a BIAcore 3000 instrument (GE Healthcare). The K_D values were evaluated using BIAevaluation software (GE Healthcare).

Mouse peritoneal macrophages were seeded onto coverslips and left overnight at 37°C. Cells were incubated with prTMD23 (10 μg/ml) for 5 min at 37°C, fixed with 3.7% formaldehyde/PBS for 10 min at 37°C, blocked with 10% FBS/PBS for 30 min at room temperature, incubated with primary Abs (mouse anti–c-Myc and rabbit anti-mouse CD14 Abs) for 60 min at room temperature, and stained with secondary Abs (Alexa Fluor 488–conjugated goat anti-mouse Ab and Alexa Fluor 546–conjugated goat anti-rabbit Ab; Invitrogen) for 60 min at room temperature. All images were acquired using an Olympus confocal microscope.

For measuring soluble CD14 and soluble TM concentrations in the plasma of mice after LPS stimulation, C57BL/6 mice (8−12 wk old) were injected i.p. with LPS (20 mg/kg; E. coli, O111:B4; Sigma-Aldrich), and the plasma concentrations of soluble CD14 and soluble TM were determined using ELISA kits (R&D Systems).

To observe the effect of recombinant TM domains on endotoxemia, C57BL/6 mice were injected i.p. with prTMD23, prTMD23^I424A, or prTMD2 and then injected i.p. with LPS (20 mg/kg; E. coli, O111:B4; Sigma-Aldrich) 30 min later. For the posttreatment administration of rTMD23 in mice, C57BL/6 mice were injected i.p. with LPS (20 mg/kg; E. coli, O111:B4; Sigma-Aldrich), and prTMD23 was administered by i.p. injection 30 min later. Mouse survival was monitored every 24 h. The sera were harvested and body temperature was measured 24 h post-LPS stimulation. IL-6 concentration was measured using a mouse IL-6 ELISA kit (R&D Systems). All animal experiments were approved by The Institutional Animal Care and Use Committee of National Cheng Kung University.

Statistical significance was analyzed using one-way ANOVA with a Bonferroni posttest and parametric unpaired t test. Differences between more than two groups were compared using two-way ANOVA with a Bonferroni posttest. Survival data were analyzed using a log-rank test. The p values < 0.05 were considered statistically significant.

Our earlier study showed that rTMD1 inhibits LPS-induced inflammation by binding to the Lewis Y Ag on LPS (18), but the function of rTMD23 in LPS-induced inflammation remains unclear. To investigate whether rTMD23 inhibits LPS-induced inflammation, we treated mouse peritoneal macrophages with mrTMD1, mrTMD12, mrTMD123, and mrTMD23. The results showed that mrTMD23 administration suppressed the LPS-induced inflammatory cytokine secretion in a manner similar to the suppression induced by administration of mrTMD1, mrTMD12, or mrTMD123 (Fig. 1A, 1B). Furthermore, both prTMD23 and mrTMD23 significantly suppressed TNF-α and IL-6 production in macrophages after LPS stimulation (Fig. 1C, 1D). The inhibitory effect of rTMD23 on LPS-induced inflammation was abrogated when prTMD23 and mrTMD23 were inactivated by heat (Fig. 1E, 1F). These data suggest that rTMD23 inhibits LPS-induced inflammation, and that mammalian- and Pichia-expressed rTMD23 similarly suppress LPS-induced inflammation.

Previous results showed that Ile-424 of TM is a potential thrombin-binding site, and mutant TM protein, which Ile-424 of TM mutates to Ala (I424A), exhibits less activity in the activation of protein C (25). To test whether the activated protein C pathway is involved in the anti-inflammatory activity of rTMD23, we prepared a mutant rTMD23 (rTMD23^I424A). prTMD23^I424A had an m.w. similar to that of mrTMD23 and of prTMD23, as estimated following SDS-PAGE with Coomassie Brilliant Blue staining (Supplemental Fig. 1A). As shown in Supplemental Fig. 1B, prTMD23^I424A displayed lower protein C activation compared with wild-type prTMD23. In addition, prTMD23 and prTMD23^I424A showed a similar activity with regard to the inhibition of LPS-induced inflammatory cytokine secretion by macrophages (Fig. 1G, 1H). These data suggest that the anti-inflammatory activity of rTMD23 is independent of the activated protein C pathway.

We next tested whether rTMD23 inhibits the intracellular signaling pathways elicited by LPS in macrophages. The results showed that treatment with either prTMD23 or mrTMD23, but not with heat-inactivated prTMD23 or heat-inactivated mrTMD23, effectively suppressed the phosphorylation of ERK1/2 (Fig. 2A), the degradation of IκB-α (Fig. 2B), and the phosphorylation of IRF3 (Fig. 2C) in macrophages following LPS stimulation. The relative intensities of phospho-ERK1/2 (Fig. 2D), IκB-α (Fig. 2E), and phospho-IRF3 (Fig. 2F) were measured. The results supported the conclusion that rTMD23 suppresses LPS-induced downstream signaling pathways in macrophages.

Monocytic membrane-bound TM is required for the binding of LPS to macrophages through the formation of a complex with CD14/TLR4/MD-2, and TM deficiency in macrophages reduces LPS binding to the cell surface and suppresses LPS-induced inflammation (14). Therefore, it is possible that rTMD23 binds to CD14 and competes with the interaction between monocytic membrane-bound TM and CD14 in macrophages. Macrophages were incubated with prTMD23 and prTMD23^I424A, and the binding ability of FITC-LPS was measured with flow cytometry. The results showed that both prTMD23 and prTMD23^I424A bound to the macrophage surface (Fig. 3A, 3B) and that LPS binding to the macrophages was reduced significantly by treatment with prTMD23 and prTMD23^I424A (Fig. 3C–F).

We next investigated how rTMD23 affects LPS binding to the macrophage surface. We found that the interaction between endogenous membrane-bound TM and CD14 could be suppressed by treatment with prTMD23 or prTMD23^I424A (Fig. 3G). The total protein levels of CD14 and TM were unaltered after treatment with LPS, prTMD23, or prTMD23^I424A (Fig. 3H). These results strongly suggest that the interaction between monocytic membrane-bound TM and CD14 is crucial for LPS binding to macrophages; disruption of this interaction with rTMD23 treatment significantly reduced LPS binding to macrophages and suppressed LPS-induced inflammation.

We next investigated whether rTMD23 binds to CD14 and inhibits the inflammatory response involving CD14. The results, obtained with a solid-phase binding assay, showed that both prTMD23 and prTMD23^I424A interacted directly with rCD14 (Fig. 4A), and surface plasmon resonance analysis showed that the K_D value of the interaction between mrTMD23 and rCD14 was 1.27 × 10⁻⁷ M (Fig. 4B). To analyze whether rTMD23 specifically inhibits the functions of CD14, we used endothelial cells, which do not express the membrane form of CD14 (Fig. 4C). The results showed that LPS-induced IL-6 secretion by HUVECs was increased strongly in the presence of rCD14; however, heat-inactivated rCD14 did not promote LPS-induced IL-6 production (Fig. 4D). Additionally, treatment with prTMD23 or prTMD23^I424A markedly reduced the secretion of IL-6 by HUVECs after stimulation with LPS and rCD14 (Fig. 4D), whereas the heat-inactivated form of prTMD23 did not suppress IL-6 production by HUVECs following stimulation with LPS and rCD14 (Fig. 4E). We also used CpG-ODN, a TLR9 ligand, to stimulate HUVECs, because CD14 reportedly participates in TLR9 activation (7). As shown in Fig. 4F, the addition of rCD14 greatly enhanced IL-6 production elicited by CpG-ODN, and prTMD23 suppressed CpG-ODN–induced IL-6 secretion by HUVECs. mrTMD1, mrTMD12, and mrTMD123 also inhibited LPS- and rCD14-induced IL-6 production (Fig. 4G), which may be the result of the direct binding of the TM lectin-like domain and LPS, as we reported previously (18). Furthermore, CD14-knockdown macrophages displayed reduced LPS-induced production of TNF-α and IL-6 compared with controls, and prTMD23 treatment reduced LPS-induced IL-6 production in CD14-knockdown macrophages (Fig. 4H–J). These data suggest that rTMD23 interacts with CD14 and inhibits CD14-mediated inflammation.

The levels of soluble TM and CD14 in blood circulation were increased markedly in mice after LPS challenge (Fig. 5A, 5B). In addition, an interaction between prTMD23 and soluble CD14 in mouse plasma was observed after LPS stimulation (Fig. 5C), and prTMD23 colocalized with the membrane form of CD14 in mouse macrophages (Fig. 5D). We next evaluated whether treatment with rTMD23 improves survival and inhibits inflammatory responses in mice with endotoxemia. As shown in Fig. 5E, treatment with prTMD23 or prTMD23^I424A increased mouse survival after LPS challenge. Consistently higher body temperature and lower serum IL-6 concentrations were observed in mice after treatment with prTMD23 and prTMD23^I424A compared with PBS (Fig. 5F, 5G). Furthermore, posttreatment administration of prTMD23 significantly increased mouse survival rate and suppressed the LPS-induced inflammatory response (Fig. 5H–J). These data indicate that rTMD23 reduces the inflammatory response to LPS in mice and significantly improves their survival rate.

To investigate which domain of rTMD23 is critical for its anti-inflammatory activity, we generated a series of recombinant truncated domains of rTMD23, including mrTMEGF2-6D3, mrTMEGF3-6D3, mrTMEGF4-6D3, mrTMEGF5-6D3, mrTMEGF6D3, and mrTMD2 (Fig. 6A, Supplemental Fig. 2). We found that all of the recombinant truncated domains, with the exception of mrTMD2, suppressed LPS-induced inflammation in both macrophages (Fig. 6B) and HUVECs (Fig. 6C). In addition, the survival rate of mice treated with prTMD2 was not different from that of mice treated with PBS injection after stimulation with LPS (Fig. 6D). Similar results were obtained for body temperature and serum IL-6 concentrations: mice treated with prTMD2 showed similar body temperature (Fig. 6E) and IL-6 levels (Fig. 6F) to those in the PBS group after LPS challenge. These results indicate that TM D3 is essential for the anti-inflammatory activity of rTMD23. The proposed models of rTMD23 in anti-inflammation are shown in Fig. 7.

Soluble TM is increased significantly in the plasma of septic patients (16); however, its functions in sepsis remain unclear. Our earlier work demonstrated that rTMD1 inhibits LPS-induced inflammation by binding to the Lewis Y Ag on LPS (18), but the function of rTMD23 in LPS-induced inflammation is largely unknown. In the current study, we demonstrated that rTMD23 significantly decreased the LPS-induced inflammatory response in macrophages by inhibiting the association between endogenous membrane-bound TM and CD14, resulting in reduced binding of LPS to macrophages. These results are consistent with our previous work showing that TM knockout in macrophages reduces LPS binding to macrophages and suppresses LPS-induced inflammation (14). Furthermore, rTMD23 suppressed LPS- and CpG-ODN–induced inflammation in HUVECs by interacting with soluble CD14, and it inhibited the functions of soluble CD14, suggesting that rTMD23 interacts with the membrane form of CD14, as well as binds to soluble CD14. In an endotoxemia model, rTMD23 exerted a protective effect against LPS-induced inflammation, because mice treated with rTMD23 displayed reduced inflammation and improved survival after LPS stimulation. The interaction between rTMD23 and CD14 also was observed in the plasma of mice challenged with LPS. Therefore, we propose that rTMD23 binds to the membrane and soluble forms of CD14 and specifically inhibits the CD14-mediated inflammatory response (Fig. 7).

It was demonstrated that the lectin-like domain of TM can bind to LPS through the Lewis Y Ag and inhibits LPS-induced inflammation; however, rTMD23 did not interact with LPS (18). Therefore, the suppressive effects of rTMD1, rTMD12, and rTMD123 on the LPS- and CD14-induced production of inflammatory cytokines may be caused by the direct interaction between the lectin-like domain of TM and LPS, whereas the inhibitory activity of rTMD23 may occur through suppression of the function of CD14. The cofactor activity of TM in the activation of protein C is well characterized (26). Recent studies revealed that activated protein C regulates hemostasis, as well as plays a role in anti-inflammation through various pathways (20). Our data showed that rTMD23 with a mutated thrombin-binding site (rTMD23^I424A) lost most of its cofactor activity, but LPS-induced inflammation was still suppressed following treatment with rTMD23^I424A in vitro and in vivo. These data indicate that rTMD23 inhibits LPS-induced inflammation through an activated protein C–independent pathway. The TM–thrombin complex also converts thrombin-activatable fibrinolysis inhibitor (TAFI) to TAFIa. Although TAFIa participates in the inhibition of fibrinolysis, the anti-inflammatory character of TAFIa is reportedly linked to inactivation of C3a, C5a, bradykinin, and osteopontin (27). Because rTMD23^I424A is mutated at a thrombin-binding site (25), we can exclude the possibility that the anti-inflammatory activity of rTMD23 occurs through activation of TAFI.

Our data showed that the minimal fragment of rTMD23 that we produced in this study, rTMEGF6D3, significantly suppressed the LPS-induced inflammatory response, whereas rTMD2 did not suppress LPS-induced inflammation in vitro or in vivo, suggesting that TM D3 is required for the anti-inflammatory activity of rTMD23. The D3 of TM is a serine/threonine-rich domain that is crucial for thrombin binding and the activation of protein C (15). However, the anti-inflammatory activity of rTMD23 is independent of its cofactor activity and the activated protein C pathway, suggesting that the D3 of TM provides an activated protein C–independent mechanism for the anti-inflammatory activity of rTMD23.

CD14 is a well-studied LPS receptor for eliciting LPS-induced inflammation via activation of TLR4 downstream signaling pathways (28, 29). Recent studies reported that CD14 is involved in the activation of TLR2, TLR3, TLR7, and TLR9, because CD14-null mice and CD14-deficient cells display less inflammatory response to the specific ligands of these TLRs (5–7). We also observed that addition of rCD14 to endothelial cells, which do not express the membrane form of CD14 but do express TLR4 and TLR9 (8, 30, 31), markedly increased IL-6 secretion by HUVECs after LPS or CpG-ODN challenge. rTMD23 not only inhibited LPS-induced IL-6 secretion by HUVECs, it also suppressed CpG-ODN–induced IL-6 production, suggesting that the anti-inflammatory capability of rTMD23 occurs through specific suppression of CD14 functioning.

Accumulating evidence indicates that CD14 contributes to pathological phenomena in addition to sepsis, including liver fibrosis, metabolic syndrome, Alzheimer’s disease, and neuropathic pain (32–35). In an experimental cholestasis model, mice with CD14 deletion display reduced liver fibrosis resulting from a decrease in the production of TNF-α and TGF-β (32). Endotoxemia-initiated obesity and insulin resistance can be attenuated by knocking out CD14 in mice (33). CD14-knockout mice also exhibit reduced deposition of β-amyloid plaque in the brain as a result of changes in the brain inflammatory environment (34). In a neuropathic pain model (spinal nerve L5 transection), knockout of CD14 in mice suppresses mechanical allodynia and thermal hyperalgesia. Increased mechanical hypersensitivity in mice is observed after intrathecal injection of soluble CD14 (35). According to these reports, targeting CD14 may be a potential therapeutic strategy in CD14-related diseases, as described above, and rTMD23 may be useful in future treatments of these diseases because it can block the biological functions of CD14.

The authors have no financial conflicts of interest.