Abstract
We have previously reported that the CD14+ monocytic subpopulation of human PBMC induces programmed cell death (apoptosis) in cocultured endothelial cells (EC) when stimulated by bacterial endotoxin (LPS). Apoptosis is mediated by two routes, first via transmembrane TNF-α (mTNF) expressed on PBMC and, in addition, by TNF-independent soluble factors that trigger apoptosis in EC. Neutralizing anti-TNF mAb completely blocked coculture-mediated apoptosis, despite the fact that there should have been additional soluble cell death factors. This led to the hypothesis that a reverse signal is transmitted from the TNF receptor on EC to monocytes (MO) via mTNF that prevents the production of soluble apoptotic factors. Here we have tested this hypothesis. The results support the idea of a bidirectional cross-talk between MO and EC. Peripheral blood MO, MO-derived macrophages (MΦ), or the monocytic cell line Mono Mac 6 were preincubated with human microvascular EC that constitutively express TNF receptor type I (TNF-R1) and subsequently stimulated with LPS. Cell-free supernatants of these preparations no longer induced EC apoptosis. The preincubation of MO/MΦ with TNF-reactive agents, such as mAb and soluble receptors, also blocked the production of death factors, providing further evidence for reverse signaling via mTNF. Finally, we show that reverse signaling through mTNF mediated LPS resistance in MO/MΦ as indicated by the down-regulation of LPS-induced soluble TNF and IL-6 as well as IL-1 and IL-10.
Endothelial cells (EC)3 are increasingly being recognized as integral components of the immune system. In contrast, leukocyte-endothelial interactions are pivotal steps in mediating inflammatory responses. In addition to the activation-induced transendothelial migration of blood leukocytes, EC programmed cell death (apoptosis) may indirectly contribute to an increased efflux of immune effector cells because of the leakiness of the vessel wall (1). We have previously identified the transmembrane form of TNF (mTNF) to be crucially involved in EC apoptosis mediated by either endotoxin (LPS) in combination with ionizing radiation (2) or by LPS-activated monocytes (MO) (3). In addition, cell-free supernatants (SN) of LPS-triggered MO also were cytotoxic for EC independent of soluble TNF. Surprisingly, we could completely block coculture-mediated EC death with a neutralizing anti-TNF mAb, despite the fact that there should be an additional apoptosis-inducing capacity coming from these cell-free SN. This observation prompted us to examine whether the mTNF-TNF receptor interaction may function bidirectionally in the sense that mTNF may not only induce apoptosis, but also elicit a negative feedback signal that suppresses the production of soluble death factors by monocytic cells.
Such a reverse signaling of membrane-bound members of the TNF ligand (L) superfamily has been reported for the CD30L (4), CD40L (5), CD95L (FasL) (6), and CD137L (7). In the present report, we provide evidence, to our knowledge for the first time on the basis of functional studies, for a reverse signaling by TNF itself that ultimately leads to the desensitization of MO against bacterial endotoxin.
These data argue for an active role of the vascular endothelium in regulating cytotoxic responses by inducing anergy in immune effector cells via TNF receptor-mTNF bidirectional signaling processes.
Materials and Methods
Cell culture and reagents
The human dermal microvascular EC line CDC/EU.-HMEC-1 (further referred to as EC) was kindly provided by the Centers for Disease Control and Prevention (Atlanta, GA) and has been established as previously described (8). EC (passages 25–35) were cultured in MCDB131 medium, supplemented with 15% FCS, 1 μg/ml hydrocortisone (Sigma, Deisenhofen, Germany), 10 ng/ml epidermal growth factor (Collaborative Biochemical Products, Bedford, MA), and antibiotics. PBMC were derived from heparinized (Novo Nordisk, Mainz, Germany) blood of healthy human blood donors according to a standard protocol using Ficoll hypaque (Pharmacia, Freiburg, Germany) density gradient centrifugation and subjected to elutriation. Preparation of MO and MO-derived macrophages (MΦ) were performed as described previously (9). The monocytic cell line Mono Mac 6 was kindly provided by Dr. H.W.L. Ziegler-Heitbrock (University of Munich, Munich, Germany) (10). All experiments presented have been performed with all three types of monocytic cells. Therefore, we collectively named them MO/MΦ. All cell culture reagents have been purchased by Life Technologies (Karlsruhe, Germany) unless stated otherwise. Cytokines, endotoxin, and (monoclonal) Abs were obtained from the following sources or described in the following references: human recombinant TNF-α (Knoll AG, Ludwigshafen, Germany), LPS (serotype 026:B6 from Escherichia coli; Sigma), ELISA kit for IL-6 (PharMingen, Hamburg, Germany), anti-TNF-α mAbs MAK195 and MAK199 (Knoll AG) (11), and T1, as well as anti-TNF receptor TR60 mAb H398 (12), anti-TNF receptor TR80 Ab fragments (M80Fab), and mAb 80 M2 (13). For inhibition/stimulation studies, these Abs were used at a concentration of 20 μg/ml. The optimal Ab dose was determined by dose-response experiments (not shown).
Apoptosis assays
A total of 1 × 105/plate EC were seeded in 35-mm petri dishes (Nunc, Wiesbaden, Germany) and cocultured with SN of differently treated MO/MΦ for 48 h. MO/MΦ were either left untreated or incubated in the presence of LPS (10 ng/ml) for 4 h, washed, and subsequently incubated another 4 h before SN were collected. To monitor EC apoptosis, the cells were fixed with methanol/acetone (1:1) for 2 min, washed once in PBS, and stained with 4,6-diamidino-2-phenylindole (DAPI) (0.5 μg/ml; Sigma) dissolved in 20% glycerin/PBS. Samples were mounted and subjected to microscopic analysis. Nuclear condensation as revealed by DAPI staining in the absence of trypan blue uptake is considered characteristic of apoptosis as opposed to necrosis (14, 15). The quantitative analysis included counting the number of apoptotic relative to all identifiable cells from at least 10 microscopic fields with an average of 70 cells per field.
An alternative method for detecting apoptosis in human EC was performed as previously described (16) using flow cytometry. Apoptotic cells were identified by a characteristic side scatter image distinct from that of nonapoptotic cells. All results presented in this report were reproduced with this specific method. For the sake of clarity, these results are not displayed here.
ELISA
TNF-α in the SN of MO/MΦ was determined by the ELISA sandwich technique. Briefly, 96-well plates were coated with the anti-TNF-α mAb MAK199, subsequently (hyper)incubated with the relevant cell culture SN or titrated TNF-α standards, respectively. Development was performed with a biotinylated anti-TNF-α mAb (MAK195) and a peroxidase conjugate according to a standard protocol. The ELISA for the detection of IL-6, IL-1, and IL-10 in the SN of MO/MΦ was performed exactly according to the manufacturer’s instructions.
Statistical analysis
The significance of differences between experimental values was assessed by means of the Student’s t test. Analysis of the differences between EC apoptosis inducing vs noninducing SN of MO/MΦ revealed values of p < 0.001 in all cases. Statistical analysis of the ELISA results are, where applicable, presented in the legends to the tables.
Results
Preincubation of MO/MΦ with EC abrogates the production of LPS-induced EC death factors
All experiments presented below have been performed with elutriated MO, MΦ, and the monocytic cell line Mono Mac 6. For simplicity, these different cells are collectively named MO/MΦ throughout the manuscript. As the experiments involved a series of incubation and washing steps, the experimental design is illustrated in panel a of each figure for clarity.
To investigate the hypothesis of a bidrectional cross-talk between EC and MO/MΦ, MO/MΦ were first seeded for 30 min on an EC monolayer and subsequently stimulated with LPS (10 ng/ml) for 4 h. MO/MΦ were then washed three times to remove all reagents and incubated for another 4 h. Then, SN of these cells were collected and investigated for their ability to induce EC programmed cell death (Fig. 1⇓a). In addition, EC were pretreated with a TNF receptor type I neutralizing mAb (anti-TNF-R1) for 30 min. Fig. 1⇓b shows that SN derived from LPS-treated MO/MΦ strongly induced apoptosis in EC (bar 2). In contrast, MO/MΦ that had been preexposed to EC no longer secreted EC death factors (bar 3). This suppression could be abrogated by pretreatment of EC with anti-TNF-R1 (bar 4). Supernatants from untreated MO/MΦ did not induce EC programmed cell death, which confirms previously published observations (3). These experiments indicate that the cross-talk is mediated by the interaction of mTNF expressed by activated MO/MΦ and TNF receptors that are constitutively expressed by EC.
Experimental design (a) and results (b) of EC apoptotis mediated by SN of LPS-activated MO/MΦ, prevention by preincubation with EC, and dependence on TNF receptor TNF-R1. Bar 1, Untreated control EC. Bar 2, Cell-free SN of LPS-treated MO/MΦ induce apoptosis in EC. Bar 3, MO/MΦ were exposed to EC and subsequently stimulated with LPS. Bar 4, MO/MΦ were seeded on EC preincubated with anti-TNF-R1. Quantitative microscopic analysis of DAPI-stained EC after 48 h of culture. Results are given in percentage apoptotic EC (% apoptotic cells) ± SD (n = 10 microscopic fields with an average of 70 cells per field). This is one representative of five independent experiments. Results are considered significant at p < 0.001 of apoptosis inducing vs noninducing SN.
Another clue to an active contribution of the TNF receptor on EC to down-regulate the cytotoxic activity in SN from LPS-stimulated MO/MΦ was provided by the observation that preincubation of EC with soluble TNF completely blocked the negative signal transmitted by EC (data not shown), obviously due to the competition in engaging the receptor binding sites.
Anti-TNF monoclonal Fab inhibit the appearance of apoptosis-inducing capacity in the SN of LPS-stimulated MO/MΦ
These data clearly indicated that EC are able to generate a negative regulatory signal toward MO/MΦ that is TNF receptor dependent and prevents the expression of EC death factors. To confirm the direct involvement of MO/MΦ mTNF, monocytic cells were preincubated with anti-TNF mAb Fab for 30 min to engage mTNF before LPS challenge (Fig. 2⇓a). It is important to note that Fab have been used to prevent binding of the Abs to the monocytic Fc receptors. As depicted in Fig. 2⇓b, both the (Fab′)2 as well as the Fab of the TNF-specific mAbs MAK195 and T1 prevented the production of apoptosis-inducing SN in MO/MΦ (bars 5 and 6). The completeness of this blockade strongly supports a central role for mTNF in the initiation of reverse signaling. The fact that also the monovalent Fab could inhibit the production of cytotoxic factors further suggests that oligomerization of mTNF may not be necessary to elicit this response.
Preincubation of MO/MΦ with anti-TNF Fab prevents the LPS-induced release of EC death factors. Bar 5, Preincubation of MO/MΦ with anti-TNF (Fab′)2. Bar 6, Preincubation of MO/MΦ with anti-TNF Fab. For experimental details, see legend to Fig. 1⇑. Results are considered significant at p < 0.001.
As a control, irrelevant control Abs did not have any effect on either mTNF or the MO/MΦ response to LPS (data not shown).
Soluble TNF receptors induce reverse signaling in MO/MΦ
To conclusively prove that reverse signaling proceeds via mTNF, we used soluble TNF receptor (TNF-R1, TNF-R2) fusion constructs, where the respective extracellular domain is linked to the Fc part of a human IgG1 Ab (Fig. 3⇓a). It is important to note that monocytic cells lack a receptor for IgG1 Fc. MO/MΦ were pretreated for 30 min with either soluble TNF-R1-Fc or TNF-R2-Fc. Both constructs were able to suppress the apoptosis-inducing capacity in SN of LPS-stimulated MO/MΦ (Fig. 3⇓b, bars 7 and 8).
Soluble TNF receptor fusion constructs TNF-R1-Fc and TNF-R2-Fc induce reverse signaling via mTNF and prevent the LPS-induced release of EC death factors. Bar 7, Preincubation of MO/MΦ with anti-TNF-R1-Fc fusion protein. Bar 8, Preincubation of MO/MΦ with anti-TNF-R2-Fc fusion protein. For experimental details see legend to Fig. 1⇑. Results are considered significant at p < 0.001.
As a control, irrelevant Fc receptor fusion proteins were tested and found to be incapable of affecting the LPS-induced production of death factors toxic for EC (data not shown).
These data strongly suggest a role of MO/MΦ-derived mTNF in regulation of cytotoxic factor production. However, they do not fully exclude that these effects are due to a paracrine or even autocrine interaction between mTNF and TNF receptors on the surface of MO/MΦ. To test this possibility, MO/MΦ were preincubated with receptor-neutralizing TNF-R1- and TNF-R2-specific Abs before SN were prepared as described for Figs. 1–3⇑⇑⇑. The production of LPS-induced death factors could not be prevented by this treatment, further supporting the concept of reverse signaling through mTNF (Fig. 4⇓, a and b).
Preincubation of MO/MΦ with TNF receptor-neutralizing Abs does not influence reverse signaling through mTNF, no autocrine signal loop in monocytic cells. Bar 5, Preincubation of MO/MΦ with anti-TNF (Fab′)2. Bar 9, Preincubation of MO/MΦ with anti-TNF-R Abs before stimulation of mTNF. For experimental details see legend to Fig. 1⇑. Results are considered significant at p < 0.001.
In addition, mTNF itself, given in the form of Chinese hamster ovary cell transfectants bearing a noncleavable mutant of mTNF (reviewed in Ref. 17), could not down-regulate EC apoptosis (not shown). This mutant was obtained by the deletion of the proteolytic cleavage site in wild-type mTNF for the production of soluble TNF. Taken together, any means of engaging mTNF on MO/MΦ, including genuine TNF-receptors, soluble TNF receptor constructs, Abs, or Ab fragments, prevented the expression of LPS-induced cytotoxic factors, suggesting a reverse signaling pathway.
Reverse signaling through mTNF inhibits release of soluble TNF induced by LPS
To further elucidate the nature of the reverse signaling by mTNF, we investigated LPS responses of MO/MΦ other than the production of EC death factors. One key response mediator of LPS challenge is the production of soluble TNF that is rapidly induced after LPS addition. The presence of soluble TNF in SN, prepared as described for Fig. 2⇑, was measured by ELISA. Briefly, elutriated MO and Teflon-grown MΦ were either left untreated or incubated with LPS for 6 h in the presence or absence of a previous preincubation with the anti-TNF mAb 195F. Pretreatment with anti-TNF alone served as a control. In both cases, anti-TNF had vigorously been washed before collection of the SN for ELISA analysis.
Table I⇓ demonstrates that stimulation of mTNF by the mAb 195F led to an almost complete down-regulation of soluble TNF production after LPS challenge (p < 0.001) in MO. In MΦ, TNF levels after LPS were found to be 8735 pg/ml (±2530). Preincubation of MΦ with anti-TNF down-regulated the release of soluble TNF to 1183 pg/ml (±1480; p = 0.01); reverse signaling alone (anti-TNF) did not influence TNF release.
Reverse signaling through mTNF leads to down-regulation of LPS-induced release of soluble TNFa
The same response pattern for the reverse signaling effect of mTNF could be observed for 2, 12, and 24 h post-LPS challenge (not shown). In summary, these data suggest that reverse signaling via mTNF causes MO/MΦ to become resistant to LPS-induced inflammatory responses.
Reverse signaling via mTNF inhibits the LPS-induced release of IL-6
IL-6 is one of the primary acute-phase proteins that are released upon LPS stimulation. To answer the question whether LPS resistance or unresponsiveness induced by reverse signaling of mTNF also affects other inflammatory mediators, SN of MO/MΦ were analyzed for IL-6 production 2, 6, and 24 h after LPS stimulation using a standard ELISA technique. A representative experiment is shown in Table II⇓. Reverse signaling via mTNF partially down-regulated the LPS-induced IL-6 release by 50–80% at all time points tested. IL-6 release varied considerably among different blood donors, so that statistical analysis was not applicable. It remains to be elucidated why inhibition of LPS-induced IL-6 release was incomplete. One can speculate that LPS may also trigger intracellular signal transduction pathways independent of a control by mTNF reverse signaling.
Reverse signaling through mTNF leads to down-regulation of LPS-induced release of IL-6a
Reverse signaling via mTNF inhibits release of LPS-induced IL-1 and IL-10 release
To gain more information about the nature of the LPS resistance or anergy that is induced by reverse signaling through mTNF, we also tested the influence of mTNF stimulation on the LPS-induced release of IL-1 and IL-10. The monocytic cell line Mono Mac 6 (see Ref. 10) was treated as given, and SN of these cells were analyzed 6 h (IL-1) and 24 h (IL-10) posttreatment, respectively. Interestingly, an immediate LPS challenge could not fully be blocked by mTNF reverse signaling (Table III⇓, anti-TNF/LPS), whereas a 2-h delay between the mTNF stimulus and LPS challenge (Table III⇓, anti-TNF/2h/LPS) almost fully abrogated the LPS-induced release of both cytokines. It remains to be elucidated why mTNF reverse signaling influenced IL-1 and IL-10 secretion mediated by LPS in an apparently different way like soluble TNF and the EC death factors, but one can speculate that mTNF is able to transmit different downstream signals that take different amounts of time until they can block incoming LPS signals.
Reverse signaling through mTNF leads to down-regulation of LPS-induced release of IL-1 and IL-10a
In summary, our data indicate the existence of a bidirectional cross-talk between EC and cells from the monocytic system that balances the production of cytotoxic effector molecules and other proinflammatory mediators.
Discussion
Previous work from our group (3) had led to the hypothesis of a reverse signaling of mTNF. Here we prove a functional basis for the idea that stimulating mTNF on MO/MΦ by various means elicits a negative regulatory signal that induces monocytic cells to become resistant to inflammatory responses triggered by LPS. Three functional parameters strongly support the notion of this bidirectional cross-talk between EC and MO/MΦ. First, EC expressing TNF receptor type I (TNF-R1) prevent the expression of endothelial death factors that are released by monocytic cells upon LPS stimulation. Second, soluble TNF as a prime response mediator of LPS activation is completely down-regulated through the engagement of mTNF. Third, the release of IL-6, an LPS-triggered acute-phase response protein, is partially blocked by mTNF reverse signaling. Finally, a later mTNF signal can block IL-1 and IL-10 release mediated by LPS.
It is currently under investigation in our laboratory to delineate the scope and extent of the mTNF-induced LPS resistance. An interesting question is, for example, to assess the potential of mTNF to, for example, restore the IL-12 production in MO/MΦ that is suppressed by LPS (18).
Indirect effects of mTNF with the TNF receptors on MO/MΦ could be ruled out by a number of control experiments. Here, the probably most convincing evidence was that neutralizing TNF receptor Abs (Fig. 4⇑) failed to prevent the production of death factors by MO/MΦ in response to LPS stimulation. In addition, we assured that LPS resistance was not caused by shedding of the primary monocytic LPS receptor CD14. Flow cytometric analyses of LPS-treated MO/MΦ showed that CD14 expression was not affected by reverse signaling (data not shown).
The final molecular proof of principle for a reverse signaling by mTNF is certainly to use cells from TNF knockout mice (19). It is the subject of ongoing studies in our laboratory to transfect MO/MΦ from TNF knockout mice with either wild-type mTNF or a mutant lacking the cytoplasmic domain. If the latter mutant is no longer able to provide resistance to LPS-mediated responses, it must be due to intracellular signal transduction emanating from mTNF.
There is accumulating evidence in the literature for reverse signaling by members of the TNF/nerve growth factor family. Cross-linking of CD30L on freshly isolated neutrophils increases IL-8 production by these cells (4). The engagement of CD40L can influence Ig production in B lymphocytes, depending on the density of CD40L (5). In addition, reverse signaling through the FasL is required for an optimal proliferation and cytotoxic activity of T lymphocytes (6). Finally, CD137 (ILA/4-1BB) has been recognized as a novel and potent MO activation factor by signaling via CD137L (7).
To our knowledge, we show for the first time on a functional basis that also mTNF itself is able to transmit reverse signals. A first report on an active role for mTNF as a costimulatory molecule revealed that anti-TNF up-regulated anti-CD3-induced IFN-γ mRNA expression, whereas IL-4 mRNA expression was blocked (20). Work is in progress in our and in other laboratories to elucidate intracellular signal transduction mechanisms associated with reverse signaling of the TNF family. It is important to note in this context that also Fab induced reverse signaling of mTNF (Fig. 2⇑). This suggests that cross-linking of mTNF, in contrast to other TNF-like ligands, may not be mandatory for signal transduction. With regard to downstream signals, Watts et al. provide evidence that casein kinase I phosphorylates the cytoplasmic domain of mTNF at a consensus sequence that is conserved among members of the TNF family (21). In addition, a phosphorylation site of mTNF at serine residues could be identified (22). While these studies show that the cytoplasmic portion of mTNF is a target for signal transduction molecules, they do not yield information on the mechanism of signaling. However, as phosphorylation sites often serve as docking sites for adapter proteins and initiate the assembly of signaling complexes, it is worthwhile to investigate whether mTNF uses classical signal transduction pathways. With the help of activation-specific Abs and kinase assays, one can identify several components that might be involved in reverse signaling.
Along these lines, we have investigated whether the LPS resistance phenomenon could be explained by the mTNF reverse signaling interfering with NF-κB activation in response to LPS stimulation. But preliminary results from our group argue against an effect of mTNF on NF-κB activation, as determined by Western blot analysis of the phosphorylation and degradation of the NF-κB inhibitor I-κB (data not shown).
mTNF has been demonstrated to have some unique properties that are not shared with the soluble form of this cytokine. (2, 17, 23, 24). The observation that mTNF also can transmit reverse signals may have considerable implications for the therapeutic or prophylactic interference with TNF-triggered inflammation. A negative regulatory signal inducing LPS resistance could be of benefit for the control of undesired activation processes triggered by TNF. Success in the treatment of transplant patients with neutralizing TNF Abs (25) may also in part be explained by suppression of macrophage activation due to reverse signaling.
It will be interesting to investigate whether cells other than MO/MΦ are able to transmit similar reverse TNF signals, e.g., EC, which are known to constitutively express mTNF (2).
Acknowledgments
We thank Professor Manuela Baccarini for helpful discussions, Professor H.-W. Löms Ziegler-Heitbrock for providing the monocytic cell line Mono Mac 6, and Dr. Marina Kreutz for instructions in in vitro MΦ differentiation. The expert technical assistance of Silvia Haffner is gratefully acknowledged.
Footnotes
-
↵1 This work was supported by Grants No. Ei68/1-1 and 2-1-2 from the Deutsche Forschungsgemeinschaft and No. 97.042.1 from the Wilhelm-Sander-Stiftung.
-
↵2 Address correspondence and reprint requests to Dr. Günther Eissner, Department of Hematology and Oncology, University of Regensburg, Franz-Josef-Strauss-Allee 11, D-93053 Regensburg, Germany. E-mail address: guenther.eissner{at}klinik.uni-regensburg.de
-
↵3 Abbreviations used in this paper: EC, endothelial cell; mTNF, transmembrane TNF-α; MO, monocyte; MΦ, macrophage; SN, supernatant; L, ligand.
- Received December 20, 1999.
- Accepted April 3, 2000.
- Copyright © 2000 by The American Association of Immunologists
References
In this issue
