Inhibition of transmembrane TNF-α shedding by a specific antibody protects against septic shock

Transmembrane TNF-α (tmTNF-α) and secretory TNF-α (sTNF-α) display opposite effects in septic shock. Reducing tmTNF-α shedding can offset the detrimental effects of sTNF-α and increase the beneficial effect of tmTNF-α. We previously developed a monoclonal antibody that is specific for tmTNF-α and does not cross-react with sTNF-α. In this study, we show that this antibody can specifically suppress tmTNF-α shedding by competing with a TNF-α converting enzyme that cleaves the tmTNF-α ectodomain to release sTNF-α. This tmTNF-α antibody significantly inhibited LPS-induced secretion of interleukin (IL)-1β, IL-6, interferon-β, and nitric oxide by monocytes/macrophages, and protected mice from septic shock induced by lipopolysaccharide (LPS) or cecal ligation and puncture, while reducing the bacterial load. The mechanism associated with the protective effect of this tmTNF-α antibody involved promotion of LPS-induced toll-like receptor 4 (TLR4) internalization and degradation by recruiting Triad3A to TLR4. Moreover, the tmTNF-α antibody inhibited LPS-induced activation of nuclear factor-κB and interferon regulatory factor 3 pathways by upregulating expression of A20 and monocyte chemotactic protein-induced protein 1. Similarly, treatment of macrophages with exogenous tmTNF-α suppressed LPS/TLR4 signaling and release of proinflammatory cytokines, indicating that increased levels of tmTNF-α promoted by the antibody contributed to its inhibitory effect. Thus, use of this tmTNF-α antibody for specific suppression of tmTNF-α shedding may be a promising strategy to treat septic shock.


Introduction
Tumor necrosis factor-α (TNF-α) is first synthesized as transmembrane TNF-α (tmTNF-α), which is then cleaved by the TNF-α converting enzyme (TACE) to release secretory TNF-α (sTNF-α) 1,2 . sTNF-α is widely recognized as a prototypic inflammatory cytokine that plays a pivotal role in the pathogenesis of early endotoxin shock [3][4][5] . In Gram-negative sepsis, lipopolysaccharide (LPS) binds toll-like receptor 4 (TLR4) and activates NF-κB to produce TNF-α through a MyD88-dependent signaling pathway 6 . sTNF-α induces fever, hypotension, multiple organ dysfunction and death in mice, similar to those evoked by LPS. In addition, sTNF-α promotes neutrophilmediated tissue injury and amplifies inflammatory cascades by activating macrophages and other types of cells to secrete other proinflammatory cytokines. Although neutralization of TNF-α by monoclonal antibodies can mitigate shock and increase survival in LPS-induced experimental septic shock models 4 , targeted TNF-α therapies have not shown benefits in clinical trials and can even lower patient survival rates by interfering with antiinfection defenses 3,4,7 .
Interestingly, our previous study revealed that, compared with a transient elevation in the levels of serum sTNF-α at 90 min after injection of bacterial into rats, tmTNF-α expression on peritoneal macrophages and liver tissue increased gradually, to peak at 4.5 h after injection, and then declined and stabilized at relatively higher levels up to 24 h after induction of endotoxin shock 8 . This finding indicates a role of tmTNF-α in sepsis. tmTNF-α is a type II transmembrane molecule that binds to TNF receptor (TNFR) to mediate signal transduction to target cells (forward signaling) and itself acts as a receptor that transduces signals in tmTNF-α-bearing cells from insideto-outside (reverse signaling) 9,10 . In contrast to the pathogenic effects of sTNF-α in sepsis, we and others demonstrated that tmTNF-α functions as an antiinflammatory factor through forward and reverse signaling. tmTNF-α downregulates LPS-or sTNF-α-induced release of proinflammatory cytokines by reverse signaling in monocytes and macrophages 11,12 . tmTNF-α also inhibits NF-κB activation and decreases IL-6 and MCP-1 production by forward signaling in adipocytes 13 . Moreover, transgenic mice expressing uncleavable tmTNF-α are resistant to LPS and are fully protected from endotoxic shock 14 . These data imply that tmTNF-α, unlike sTNF-α, is beneficial in controlling sepsis and septic shock.
Inhibition of tmTNF-α ectodomain shedding could be a valuable therapeutic strategy to prevent endotoxin shock not only by decreasing release of sTNF-α to attenuate its proinflammatory effects, but also by increasing tmTNF-α expression to enhance its benefits. Indeed, suppression or knockout of TACE, the enzyme that is mainly responsible for tmTNF-α shedding, protects animals from endotoxin shock 8,15,16 . However, TACE has about 76 substrates 17 , and inhibition of TACE may have side effects. We previously developed a tmTNF-α monoclonal antibody (mAb) that specifically recognizes the N-terminal fragment of tmTNF-α and dose not cross-react with sTNFα 18 . This antibody effectively kills tmTNF-α expressing breast cancer cells 18 and leukemia cells 19 by antibodydependent cell-mediated cytotoxicity and complementdependent cytotoxicity. In this study, we show that this tmTNF-α mAb can compete with TACE for binding to tmTNF-α and inhibit tmTNF-α ectodomain shedding to protect against endotoxin shock by facilitating LPSinduced TLR4 internalization and degradation, and actively suppressing TLR4 signaling pathways.
Results tmTNF-α antibody specifically inhibits ectodomain shedding of tmTNF-α by competing with TACE for binding to tmTNF-α As the epitope recognized by tmTNF-α mAb is closer to the TACE cleavage site, we hypothesized that tmTNF-α Ab may interfere with TACE binding to tmTNF-α and subsequently specifically inhibit tmTNF-α shedding. Since the epitope of human tmTNF-α does no share amino acid sequence homology with murine tmTNF-α, a polyclonal antibody (pAb) was ordered from GL Biochem Ltd (Shanghai, China) using the corresponding epitopecontaining peptide conjugated with keyhole limpet hemocyanin. As expected, both human tmTNF-α mAb and murine tmTNF-α pAb significantly increased LPSinduced tmTNF-α expression on the cell surface, but markedly decreased LPS-induced release of sTNF-α in culture supernatants of the murine macrophage cell line Raw264.7, murine peritoneal macrophages, and human monocytes (Fig. 1a-c). The human monocyte cell line THP-1 constitutively expressed high levels of tmTNF-α ( Supplementary Fig. 1A). Phorbol myristate acetate (PMA), used to differentiate THP-1 cells, is also an activator of TACE 20 and induced tmTNF-α shedding to decrease tmTNF-α expression levels and increase sTNF-α release ( Supplementary Fig. 1B, C). However, THP-1derived macrophages still expressed a relatively high level of tmTNF-α in the absence of LPS stimulation. Therefore, LPS-mediated tmTNF-α shedding induced lower levels of tmTNF-α compared with the control. Similarly, tmTNF-α mAb significantly blocked tmTNF-α shedding after LPS stimulation (Fig. 1d). These data indicate a possible role for antibodies in suppression of tmTNF-α processing.
As TNFR is a TACE substrate, a TACE inhibitor suppresses not only tmTNF-α cleavage but also release of soluble TNFR (sTNFR), which can buffer the effects of sTNF-α. Indeed, TAPI-1 inhibited LPS-induced sTNFR1 release into THP-1 supernatants, yet tmTNF-α mAb had no effect ( Supplementary Fig. 1D). This outcome was confirmed in a bioassay, in which Cas9-CRISPR was used to silence TNF-α expression in THP-1 cells (Supplementary Fig. 1E). Supernatants containing sTNFR obtained from LPS-stimulated TNF-α-KO THP-1 cells Fig. 1 tmTNF-α Ab inhibits the ectodomain shedding of tmTNF-α by blocking TACE binding to tmTNF-α. Raw264.7 (a), murine peritoneal macrophages (b), peripheral human monocytes (c) and THP-1-derived macrophages (d) were stimulated with 100 ng/ml LPS, combined with 2 μg/ ml murine tmTNF-α polyclonal antibody (pAb) or human tmTNF-α monoclonal Ab (mAB) for 4 h (a-c) or 1 h (d). The same amount of isotype antibody IgG or normal serum IgG served as a control. tmTNF-α on the cell surface was detected by flow cytometry and concentration of sTNF-α in supernatants were determined by ELISA. All quantitative data are presented as means ± SEM of at least three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001. e HEK 293T cells transiently cotransfected for 24 h to express human (h)TNF-α and ▵M-TACE were stimulated with tmTNF-α mAb for 4 h. pIRES2-EGFP (pIRES2) and pDsRed-monomer-N1 (pDsRed) served as empty vector controls. Western blot analysis for the levels of endogenous TACE, ectopically expressed ▵M-TACE and tmTNF-α in total protein and sTNF-α in supernatants. β-actin served as a loading control. f HEK 293T cells transiently transfected for 24 h to express hTNF-α were stimulated for 4 h with 100 ng/ml LPS, 60 ng/ml PMA, 2 μg/ml tmTNF-α mAb, or 10 µM TAPI-1. Western blot analysis for expression of 26 kDa tmTNF-α in total protein and 17 kDa sTNF-α in supernatants. g The −23~157 TNF-α-His on Ni-NTA-resin was incubated with tmTNF-α mAb or isotype IgG for 4 h, followed by incubation overnight with HEK 293T cell lysates. The resinbound protein complexes were analyzed by western blotting with antibodies against TACE and TNF-α. Western blot data are representative of three independent experiments significantly blocked sTNF-α-mediated cytotoxicity. The suppressive effect of LPS was not affected by supernatants from cells cotreated with tmTNF-α mAb, but was reversed by supernatants from cells cotreated with TAPI-1 ( Supplementary Fig. 1F). These data suggest that tmTNF-α mAb, unlike the TACE inhibitor, does not affect LPS-induced release of sTNFR and its buffering capacity.
Increasing tmTNF-α expression by tmTNF-α Ab suppresses LPS response of monocytes/macrophages and TNFR2 mediates the inhibitory effects of tmTNF-α Next, we tested the impact of tmTNF-α Ab on LPSinduced production of pro-and anti-inflammatory cytokines. Both tmTNF-α mAb and pAb significantly inhibited LPS-induced production of proinflammatory cytokines IL-1β ( Fig. 2a-d) and IL-6 ( Fig. 2e-h) at both the mRNA and protein level in murine and human macrophages and monocytes. However, tmTNF-α Abs had no effect on LPS-induced production of antiinflammatory cytokine IL-10 ( Fig. 2i-l). These data indicate that tmTNF-α Abs induce LPS resistance in monocytes/macrophages.
To test, whether the effect of tmTNF-α Ab is mediated through its action on TNF-α, we added the antibody to tmTNF-α-expressing or TNF-α-KO THP-1 cells stimulated with LPS for 12 h. We found that LPS induced release of IL-1β and IL-6 in both tmTNF-α-expressing and TNF-α-KO THP-1 cells, although IL-1β and IL-6 levels were decreased by TNF-α-KO. However, tmTNF mAb significantly suppressed LPS-induced production of these cytokines in tmTNF-α-expressing, but not in TNFα-KO THP-1 cells (Fig. 3a, b), indicating that tmTNF mAb had no direct effect on the LPS response itself, but instead induced LPS resistance through its action on TNF-α. As demonstrated above, tmTNF Abs increased tmTNF-α expression and decreased sTNF-α release by inhibition of tmTNF-α shedding. Although decreasing sTNF-α release attenuated its proinflammatory effects, increasing tmTNF-α levels induced by tmTNF-α Abs might contribute to the induction of LPS resistance. To test this possibility, we cocultured 4% paraformaldehydefixed NIH3T3 cells overexpressing murine tmTNF-α ( Fig.  3c) with Raw264.7 cells and bone marrow-derived macrophages (BMDM) from wild-type, TNFR1KO, and TNFR2KO mice. Similar to the effect of tmTNF-α Abs, direct addition of exogenous tmTNF-α to Raw264.7 cells significantly inhibited LPS-induced mRNA transcription of proinflammatory cytokines (Fig. 3d, e), but did not affect IL-10 transcription (Fig. 3f) in Raw264.7, indicating that tmTNF-α actively induces LPS resistance in macrophages. In addition, the inhibitory effect of tmTNF-α on LPS-induced production of IL-1β (Fig. 3g) and IL-6 ( Fig. 3h) could be totally blocked by TNFR2KO, but not by TNFR1KO in BMDM, although mRNA levels of these proinflammatory cytokines and IL-6 secretion were higher in TNFR1KO BMDM than those in wild-type and TNFR2KO BMDM.
We used another septic shock animal model, cecal ligation and puncture (CLP), in which sepsis is induced by a polymicrobial infection in the abdominal cavity that involves translocation of bacteria and toxins into the bloodstream, to confirm the protective activity of the tmTNF-α antibody. Treatment of mice with tmTNF-α pAb immediately or 5 h after the CLP operation also evidently increased survival from 40% to 66.7% and from 50% to 70% (Fig. 4f, g), respectively. Similarly, the antibody inhibited tmTNF-α shedding (Fig. 4h) and decreased IL-1β and IL-6 plasma levels ( Fig. 4i, j), indicating again that tmTNF-α pAb protected against septic shock. Since complete neutralization of TNF-α aggravates infection by interfering with defense mechanisms 4 , we evaluated whether tmTNF-α Ab affected host antibacterial defenses. Interestingly, tmTNF-α pAb markedly reduced bacterial load in blood and peritoneal lavage fluid (PLF) at 24 h after the CLP operation ( Fig. 4k, l), reserving the antiinfection effect of TNF-α.

tmTNF-α Ab facilitates LPS-induced TLR4 internalization and degradation
Since LPS induces internalization of TLR4 21 , we tested cell surface levels of TLR4 by flow cytometry. In THP-1derived macrophages, tmTNF-α mAb markedly promoted a LPS-induced decline in TLR4 levels on the cell surface within 60 min of stimulation ( Fig. 5a) but did not affect TLR production at 1 h after stimulation (Fig. 5b), indicating that tmTNF-α mAb had an enhancing effect on LPS-induced TLR4 internalization. However, both tmTNF-α mAb and pAb evidently decreased total TLR4 expression at 12 h after LPS stimulation in THP-1-derived and Raw264.7 macrophages, respectively, but LPS or tmTNF-α Abs alone had no effect (Fig. 5c). We next used real-time PCR to test whether tmTNF-α Ab inhibited TLR4 production. Neither LPS nor tmTNF-α mAb Fig. 2 tmTNF-α Ab decreases LPS-induced production of proinflammatory cytokines. THP-1-derived macrophages (a, e, i), primary human monocytes (b, f, j), Raw264.7 (c, g, k), and murine peritoneal macrophages (d, h, l) were stimulated with 100 ng/ml LPS, combined with 2 μg/ml human tmTNF-α mAb or murine tmTNF-α pAb. The same amount of isotype antibody IgG or normal serum IgG served as a control. The stimulation time was 4 h for detection of mRNA levels by real-time PCR and 10 h for detection of cytokine concentrations by ELISA. The production of IL-1β (a-d), IL-6 (e-h), and IL-10 (i-l) was detected at mRNA (upper panels) and protein (lower panels) levels. All data are presented as means ± SEM of at least three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001 affected TLR4 mRNA transcription in THP-1-derived macrophages, but LPS suppressed TLR4 transcription in Raw264.7 macrophages, which was not affected by tmTNF-α pAb (Fig. 5d). These results suggested that tmTNF-α Ab had no effect on TLR4 gene expression. Instead, the tmTNF-α Ab might affect TLR4 levels by promoting TLR4 degradation. Indeed, tmTNF-α mAb did induce TLR4 degradation in the presence of LPS, although LPS itself did not induce TLR4 degradation after a 24 h treatment of THP-1-derived macrophages with cycloheximide ( Fig. 5e). This effect was completely blocked by the proteasome inhibitor MG132 at 12 h after LPS stimulation (Fig. 5f). Triad3A is known to interact with TLR4, promoting its ubiquitination and degradation 22,23 . Although neither LPS nor tmTNF-α Abs affected Triad3A expression in THP-1-derived and Raw 264.7 macrophages (Fig.  5g), IP/western blot analysis revealed that LPS induced trace amounts of Triad3A to be recruited to TLR4, whereas tmTNF-α mAb promoted the recruitment of substantial amounts of Triad3A to TLR4 at 1 h and 2 h after LPS stimulation (Fig. 5h), indicating that this antibody facilitated Triad3A-dependent TLR4 degradation. These data suggested that tmTNF-α Ab promotes LPSinduced TLR4 internalization at an early stage and degradation at a later stage to induce LPS resistance.
As tmTNF-α Ab increased the tmTNF-α expression levels, we tested the ability of tmTNF-α to promote TLR4 internalization and degradation. Because THP-1 is a human cell line, we stably transfected HEK 293T cells with full-length of human TNF-α and confirmed the high level of tmTNF-α expression on the cell surface (>75%, Supplementary Fig. 2A). Exogenous human tmTNF-α on fixed 293T cells or murine TNF-α on fixed NIH 3T3 cells was added to THP-1-derived or Raw 264.7 macrophages, respectively, at a ratio of 10:1. As expected, tmTNF-α significantly increased LPS-induced TLR4 internalization within 60 min ( Supplementary Fig. 2B) and also induced TLR4 degradation 12 h after LPS stimulation (Supplementary Fig. 2C, D), suggesting that increased amounts of tmTNF-α mediated the effect of the Ab on TLR4.

Discussion
Here we demonstrated that tmTNF-α Ab inhibited tmTNF-α shedding by competing with TACE for binding to tmTNF-α. This antibody induced LPS resistance of monocytes/macrophages and protected mice against LPSand CLP-induced septic shock by suppressing TLR4 signaling pathways.
Previously, we developed a tmTNF-α antibody that effectively kills tmTNF-α expressing tumor cells in vitro and in vivo 18,19 . In this study, we explored the novel function of this antibody, which involves specific blockage of ectodomain shedding of tmTNF-α by competing with TACE for the substrate binding. This antibody selectively inhibited sTNF-α release to reduce its detrimental effects while increasing tmTNF-α expression to exert its beneficial effect of inducing LPS resistance in vitro and in vivo.
In addition, in contrast to the TACE inhibitor, the tmTNF-α Ab did not affect LPS-induced, TACEdependent TNFR1 shedding. This is helpful for neutralization of sTNF-α that was reduced by the antibody, thus further alleviating its detrimental effects. Moreover, binding of released sTNFR1 to tmTNF-α initiates reverse signaling, which extends the protective function of tmTNF-α from local to distant through conversion of its action from juxtacrine to retrocrine. This is beneficial for controlling sepsis and septic shock.
Our results revealed that the antibody induced LPS resistance of monocytes/macrophages in vitro, showing downregulated LPS-induced production of NO, IL-1β, IL-6 and IFN-β, and conferred protection against LPS and CLP-induced septic shock, manifested as reduced inflammation and increased survival. The benefit of the antibody can mainly be attributed to enhanced tmTNF-αmediated anti-inflammatory activity, which was further confirmed by our results showing that exogenous tmTNFα actively suppressed LPS-induced production of proinflammatory cytokines through TNFR2 in macrophages. Increasing evidence indicates that tmTNF-α exerts protective effect against bacterial infections, chronic inflammation, and autoimmunity diseases 24 . As a ligand, tmTNF-α attenuates the inflammatory processes caused by mycobacterial pleurisy in association with TNFR2 expression on myeloid cells 25 . Interactions between tmTNF-α and TNFR2 are important for the expansion and function of Treg cells 26 and for activation of myeloidderived suppressive cells to exert immune suppressive activities 27 . On the other hand, as a receptor, tmTNF-α induces LPS resistance to suppress proinflammatory cytokine production through MAPK/ERK and NF-κB pathways 13,28 , to control inflammation by induction of TGF-β expression through reverse signaling 29 .
Our data revealed that the Ab-mediated increases in tmTNF-α expression reduced the response of macrophages to LPS at the receptor and postreceptor levels. Upon LPS stimulation, the MyD88-dependent signaling pathway is first initiated at the plasma membrane and (see figure on previous page) Fig. 5 tmTNF-α Ab facilitates LPS-induced TLR4 internalization and degradation. THP-1-derived macrophages were stimulated with 100 ng/ml LPS and 2 μg/ml tmTNF-α mAb or isotype IgG for indicated time points. a TLR4 expression on the cell surface was evaluated by flow cytometry. Representative images of FCM on the left, and quantitative data on the right. b Western blot analysis of TLR4 expression after stimulation for 1 h. THP-1-derived and Raw264.7 macrophages were stimulated with 100 ng/ml LPS, combined with 2 μg/ml tmTNF-α mAb or tmTNF-α pAb, respectively. Isotype antibody IgG or normal serum IgG served as a control. c Representative images of western blot analysis of TLR4 expression at 12 h after stimulation (upper) and their quantitative data (lower). d Relative levels of TLR4 mRNA were assessed by real-time PCR 4 h after stimulation. e THP-1derived macrophages were stimulated with LPS and tmTNF-α mAb for indicated time points in the presence of 10 μg/ml cycloheximide. Representative images of western blot analysis of TLR4 expression (upper) and their quantitative data (lower). f THP-1-derived macrophages were treated for 4 h with 10 μM MG132 prior to the stimulation with LPS and tmTNF-α mAb for 12 h. Representative images of western blot analysis of TLR4 expression (upper) and their quantitative data (lower). g Western blot analysis of Triad3A expression in THP-1-derived or Raw264.7 macrophages stimulated with LPS and tmTNF-α mAb or pAb for 12 h, respectively. h Representative images of IP/western blot analysis of Triad3A recruited to TLR4 in THP-1-derived macrophages stimulated with LPS and tmTNF-α mAb for indicated time points. All quantitative data are presented as means ± SEM of at least three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001 subsequently activates NF-κB to drive production of proinflammatory mediators. Then, a TRIF-dependent signaling pathway is triggered after TLR internalization to activate IRF3 and induce type I interferon production. Eventually, TLR4 is ubiquitinated and degraded 21 . We found that increased tmTNF-α expression levels promoted by tmTNF-α Ab facilitated earlier LPS-induced TLR4 internalization and later TLR4 degradation. These phenomena were also observed upon addition of exogenous tmTNF-α to THP-1-derived macrophages exposed to LPS, indicating that tmTNF-α mediated the inhibitory action of the Ab toward TLR4. Notably, LPS alone neither induced TLR4 expression nor affected its degradation in THP-1-derived macrophages. The former result is in line with a report by Aerbajinai et al. 30 Furthermore, tmTNF-α Ab induced recruitment of the E3 ubiquitin-protein ligase Triad3A to TLR4 to promote TLR4 degradation through the ubiquitin-proteasomedependent pathway, as tmTNF-α Ab-induced TLR4 degradation could be completely blocked by a proteasome inhibitor. Our data thus revealed a novel mechanism for tmTNF-α-induced LPS resistance.
At the postreceptor level, we found that tmTNF-α Ab blocked MyD88-and TRIF-dependent signaling pathways. The tmTNF-α Ab reduced IκBα degradation and phosphorylation of p65, MAPK, and IRF3 in response to LPS, thereby decreasing expression of NF-κB-targeted genes, including iNOS and inflammatory cytokines, as well as expression of IRF3-targeted genes such as IFN-β. These results suggested that increased levels of tmTNF-α promoted by the antibody are responsible for the inhibitory effect on LPS/TLR4 signaling, which was supported by the evidence showing that exogenous tmTNF-α suppressed LPS-induced activation of NF-κB and MAPK through TNFR2. In mechanistic studies, we found that either tmTNF-α Ab or exogenous tmTNF-α upregulated LPS-induced mRNA levels of A20 and MCPIP1 in macrophages. A20, a ubiquitin-modifying enzyme that interferes with sTNF-α-mediated signaling to NF-κB 31 , and MCPIP1 deubiquitinate TRAF6 and negatively regulate NF-κB and JNK signaling to terminate the TLR4 signaling pathway and protect mice from LPS-induced septic shock 32,33 . Our previous study demonstrated that silencing of A20 expression abolishes the suppressive effect of tmTNF-α on NF-κB activation and subsequent production of proinflammatory adipokines 13 . Furthermore, MCPIP1 deubiquitinates TRAF3, which negatively regulates IFN-β expression 32 . These findings indicated that tmTNF-α actively blocks TLR4 signaling by upregulating expression of these negative regulators in addition to promoting TLR4 internalization and degradation (Fig. 8).
Despite the efficacy of anti-TNF drugs for treatment of rheumatoid arthritis, Crohn's disease and psoriasis, these drugs block the action of both tmTNF-α and sTNF-α and thus can increase the risk of infection, malignancy and development of secondary autoimmune diseases 34 . Our data demonstrated that tmTNF-α Ab significantly reduced bacterial load in the blood and PLF while exerting anti-inflammatory activity in septic shock. tmTNF-α alone is sufficient to retain a certain level of immunity against pathogens, including resolving infection with Leishmania major in cultured macrophages and in mice 35,36 and partial protection against acute infection by Myobacterium tuberculosis or Listeria monocytogenes [37][38][39] . tmTNF-α Ab selectively inhibited detrimental effects of sTNF-α while preserving tmTNF-α-mediated antiinfection and anti-inflammation activities. Thus, for treatment of septic shock and inflammatory diseases, tmTNF-α Ab could be superior to the full blockage of both forms of TNF-α.

Reagents and antibodies
LPS from Escherichia coli O111: B4, phorbol myristate acetate (PMA), and MG132 were purchased from Sigma-Aldrich (St. Louis, MO, USA). An antihuman tmTNF-α monoclonal antibody (mAb) was made by our lab 18 and is unable to cross-react to sTNF-α and murine tmTNF-α. A polyclonal antibody (pAb) specific to the corresponding epitope in murine tmTNF-α was ordered from GL Biochem Ltd (Shanghai, China) and used for murine macrophages and mice models. Isotype IgG (Santa Cruz Biotechnology, Dallas, TX, USA) and normal rabbit serum (see figure on previous page) Fig. 6 tmTNF-α Ab inhibits the MyD88-dependent TLR4 signaling pathway. THP-1-derived or Raw264.7 macrophages were stimulated with 100 ng/ml LPS, combined with 2 μg/ml tmTNF-α mAb or pAb, respectively. Isotype antibody IgG or normal serum IgG served as a control. a, d Representative western blot of three independent experiments for IκBα degradation and p65 phosphorylation in THP-1-derived (1 h after stimulation) or Raw264.7 macrophages (45 min after stimulation). Relative levels of iNOS mRNA were assessed by real-time PCR at 4 h (b, e), and NO production at 10 h after stimulation (c, f). All quantitative data are presented as means ± SEM of at least three independent experiments. Mice were intraperitoneally injected with 600 μg tmTNF-α pAb or normal serum IgG immediately after the CLP operation (n = 6 each group). Western blot analysis of IκBα degradation and p65 phosphorylation (g) and relative levels of iNOS mRNA in the liver (h), and plasma levels of NO (i) 24 h after the CLP operation. THP-1-derived or Raw264.7 macrophages were stimulated for 4 h with LPS and tmTNF-α mAb or pAb, respectively. Relative mRNA levels of A20 (j, l) and MCPIP1 (k, m) were assessed by real-time PCR. All quantitative data are presented as means ± SEM of at least three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001 IgG purified by protein G column (house made) served as controls. The other antibodies used for FACS, immunoprecipitation and Western blotting were listed in Supplementary Table 1.
Animals and septic shock models C57BL/6 mice and BALB/c male mice (6-8 weeks) were purchased from Beijing HFK Bioscience Company (Beijing, China). TNFR1 or TNFR2 knockout (KO) mice on a BALB/c background were kindly gifted from Prof. Zhihai Qin (National Laboratory of Biomacromolecules, Institute of Biophysics Chinese Academy of Sciences, Beijing, China). Mice were housed on a 12-h light/12-h dark cycle and cared for in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. The study was approved by the Animal Care and Use Committee of Huazhong University of Science and Technology. Fig. 7 tmTNF-α Ab inhibits the TRIF-dependent TLR4 signaling pathway. THP-1-derived or Raw264.7 macrophages were stimulated with 100 ng/ml LPS, combined with 2 μg/ml tmTNF-α mAb or pAb, respectively. Isotype antibody IgG or normal serum IgG served as a control. a, d Representative western blot analysis of three independent experiments for IRF3 phosphorylation in THP-1-derived (1 h after stimulation) or Raw264.7 macrophages (45 min after stimulation). Relative levels of IFN-β mRNA were assessed by real-time PCR at 4 h (b, e), and IFN-β release was detected by ELISA at 10 h (c, f) after stimulation. All quantitative data are presented as means ± SEM of at least three independent experiments. Mice were intraperitoneally injected with 600 μg tmTNF-α pAb or normal serum IgG immediately after the CLP operation (n = 6 each group). Western blot analysis of IRF3 phosphorylation (g) and relative levels of IFN-β mRNA in the liver (h) and plasma levels of IFN-β (i) at 24 h after the CLP operation. *p < 0.05, **p < 0.01, ***p < 0.001 For LPS-induced septic shock, mice were challenged with intraperitoneal injection of 30 mg/kg LPS. Mice were observed for 72 h. For cecal ligation and puncture (CLP)induced septic shock, a 1 cm midline incision was performed after anaesthetizing mice with ketamine and xylazine. The cecum was exteriorized, ligated at half the distance between distal pole and base, perforated by a single through-and-through puncture with a 21 G needle and extruded a droplet of feces 40 . The sham group underwent the same procedures except CLP. 600 μg of tmTNF-α pAb or the same amount of serum IgG was singly injected intraperitoneally immediately or 5 h after the CLP operation or 30 min before LPS treatment.

Cell culture and preparation of PBMC, peritoneal macrophages, and BMDM
A human monocyte cell line THP-1 and a murine macrophage cell line Raw264.7 were obtained from American Type Culture Collection (ATCC) and cultured in RPMI-1640 or DMEM supplemented with 10% heatinactivated fetal calf serum (FCS; Sijiqing, Hangzhou, China), 100 U/mL penicillin and 100 μg/mL streptomycin at 37°C in 5% CO 2 . THP-1 cells were differentiated into macrophages by stimulation with 100 ng/ml PMA for 3 days. Peripheral human blood mononuclear cells (PBMC) from healthy donors were separated by Ficoll-Paque density gradient centrifugation, followed by Fig. 8 The mechanisms of tmTNF-α Ab-induced LPS resistance. tmTNF-α Ab suppresses ectodomain shedding of tmTNF-α, increasing tmTNF-α expression, and decreasing sTNF-α release and its detrimental effects (1). Increased tmTNF-α expression induced by the Ab results in LPS resistance via TNFR2 by promotion of LPS-induced TLR4 internalization (2) and degradation through recruiting Triad3A to TLR4 to induce its ubiquitination (3), and by upregulation of gene expression of A20 and MCPIP1 to suppress TLR4-mediated activation of MyD-88-and TRIF-dependent signaling pathways (4) adherence for 1 h in 10% FCS RPMI-1640 medium at 37°C. Peritoneal macrophages were isolated from mice by peritoneal lavage with cold sterile PBS.
Bone marrow-derived macrophages (BMDM) were isolated by flushing the femurs of BALB/c mice with PBS. Cells were cultured in DMEM with 10% heat-inactivated FCS and L929 conditioned medium at a ratio of 2:1. On day 4, the medium was exchanged with fresh medium of the same composition. After 7 days, >95% adherent cells were macrophages as evidenced by F4/80 expression (Becton, Dickinson and Company).
Harvesting of exogenous murine and human tmTNF-α NIH3T3 cells or HEK 293T cells were stably transfected with a retrovirus vector pMCSV-CMV-P2A-puro containing murine wild-type TNF-α cDNA at EcoRI and BamHI site (Hanbio, Biotechnology Co., Ltd., Shanghai, China) or cotransfected with a lentiviral vector pTK642 containing human wild-type TNF-α cDNA at BamHI and XhoI site, a packaging vector psPAX2 and an envelope vector pMD2G (gifted form Prof. Tongcun Zhang, Wuhan University of Science and Technology, Wuhan, China), respectively. tmTNF-α overexpressing NIH3T3 cells or 293T cells as effector cells were fixed with 4% paraformaldehyde for 30 min at room temperature and were used as the source of exogenous murine or human tmTNF-α, respectively. Murine RAW264.7 and BMDM, and human THP-1-derived macrophages as target cells were coincubated with corresponding effector cells at an effector/target ratio of 10:1.

Western blot analysis
Total protein was extracted by lysis of cells with ice-cold NP-40 lysis buffer (Beyotime Biotechnology, Shanghai, China) or by the homogenization of liver tissue in RIPA lysis buffer (BOSTER, Wuhan, China) containing protease inhibitors 0.5 mM PMSF, 5 μg/ml aprotinin, and 5 μg/ml leupeptin, followed by incubation on ice for 30 min at 4°C . After centrifugation at 12,000 × g for 15 min at 4°C, total protein in the supernatant was collected. For soluble protein isolation, 500 μl of culture supernatant was mixed with 500 μl of methanol and 125 μl of chloroform, vortexed and centrifuged at 12,000 × g for 10 min at 4°C. The upper phase was removed and 500 μl of methanol was added and mixed. After centrifugation at 12,000 × g for 10 min at 4°C, the pellet was dissolved in loading buffer for western blotting 41 .

Immunoprecipitation (IP)
The total protein extracted after stimulation was pretreated with 1 μg of normal mouse IgG and 25 μl of Protein G PLUS-Agarose (Santa Cruz Biotechnology) at 4°C for 1 h to remove nonspecific binding, then followed by incubation overnight with a monoclonal antibody specific to TLR4 at 4°C. Subsequently, the immune complexes were incubated with Protein G PLUS-Agarose in rotation at 4°C for 4 h. The immunoprecipitated molecules were analyzed by western blotting.

Quantitative real-time PCR
Total RNA was extracted using the TRIzol reagent (Invitrogen) and reverse-transcribed to cDNA using HiFiScript cDNA Synthesis Kit (CoWinBiotech, Beijing, China) according to the manufacturer's instructions. The primers were synthesized by Tsingke (Wuhan, China) and their sequences were listed in Supplementary Table 2. Realtime-PCR amplification of cDNA was conducted in 20 μl UltraSYBR Mixture (with ROX) (Beijing CoWin Biotech, Beijing, China) using the CFX Connect Real-Time PCR Detection System (Bio-Rad). The reactions were performed in triplicate as follows: 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Results were analyzed using the 2 −▵▵Ct method and normalized to the corresponding level of GAPDH.

Flow cytometry
To detect cell surface expression of tmTNF-α and TLR4, cells were incubated at 4°C for 1 h with PE-conjugated antimurine TNF-α antibody or primary antibodies including anti-human TNF-α and anti-human TLR4 antibodies, followed by FITC-conjugated secondary antibodies. The fluorescence-stained cells were analyzed on a LSR II flow cytometer (Becton Dickinson, San Jose, CA, USA).
NO was detected by a commercial Nitric Oxide Assay Kit (Beyotime Biotechnology, Shanghai, China) according to the manufacturer's instructions.

Generation of Cas9-CRISPR-mediated TNF-α knockout in THP-1 cells
The Cas9-CRISPR system was used to generate TNF-α gene knockout (KO) in THP-1 cells and targeting sequence (TGAAAGCATGATCCGGGACG) was designed using the web-based tool CRISPR Design at crispr.mit.edu. A lentivirus plasmid GV393 containing TNF-α-targeted guide RNA, Cas9 and EGFP was ordered from Shanghai Genechem Co., Ltd. THP-1 cells were infected with virus (50 MOI). At 48 h after infection, GFPpositive cells were sorted by flow cytometry (Beckman MoFlo) to allow single-colony formation. After 14 days, individual colonies were picked and selected clones were expanded. The absence of TNF-α was confirmed by western blotting.

Bacterial load
Peripheral blood and peritoneal lavage fluid were aseptically collected at 24 h after the CLP operation. Samples were serially diluted with sterile saline and cultured overnight at 37°C on tryptic soy agar plates. The number of bacterial colonies was counted in a blind manner and expressed as colony forming units (CFU) per milliliter.

Statistical analysis
Statistical analysis was performed with GraphPad Prism V6 software using one-way ANOVA followed by post hoc Turkey's test. The survival curves were plotted using the Kaplan-Meier method and compared by the log-rank test. A value of P < 0.05 was considered statistically significant.