Abstract
TRAIL (also called Apo2L) belongs to the tumor necrosis factor family, activates rapid apoptosis in tumor cells, and binds to the death-signaling receptor DR4. Two additional TRAIL receptors were identified. The receptor designated death receptor 5 (DR5) contained a cytoplasmic death domain and induced apoptosis much like DR4. The receptor designated decoy receptor 1 (DcR1) displayed properties of a glycophospholipid-anchored cell surface protein. DcR1 acted as a decoy receptor that inhibited TRAIL signaling. Thus, a cell surface mechanism exists for the regulation of cellular responsiveness to pro-apoptotic stimuli.
Apoptosis (programmed cell death) is crucial for the development and homeostasis of metazoans (1). The cell death program has three essential types of elements: activators, inhibitors, and effectors; inCaenorhabditis elegans, these components are encoded, respectively, by the ced-4, ced-9, andced-3 genes. The CD95 ligand (CD95L) and tumor necrosis factor (TNF) are important extracellular activators of apoptosis in the mammalian immune system (2). The cognate receptors for these cytokines, CD95 (also called Fas or Apo1) and TNFR1, contain cytoplasmic “death domains” that activate the cell's apoptotic machinery through interaction with the death domains of the adapter proteins FADD (also called MORT1) (3, 4) and TRADD (5). Upon activation by ligand, CD95 recruits FADD directly, whereas TNFR1 binds FADD indirectly, through TRADD. FADD in turn activates the ced-3–related protease MACHα/FLICE (caspase 8), thereby initiating a series of caspase-dependent events that lead to cell death (6, 7).
The cytokine TRAIL, also called Apo2L (8, 9), is structurally related to CD95L and TNF; TRAIL activates rapid apoptosis in tumor cell lines, acting independently of CD95, TNFR1, or FADD (9, 10). A receptor for TRAIL, designated DR4, belongs to the TNFR gene superfamily, contains a cytoplasmic death domain, and activates apoptosis independently of FADD (11). DR4 exhibits several mRNA transcripts that are expressed in multiple human tissues, including peripheral blood leukocytes (PBLs) and spleen (11).
On the basis of an expressed sequence tag (EST) that showed homology to death domains (12), we isolated human cDNAs encoding an undescribed member of the TNFR family, which we designated death receptor 5 (DR5) (Fig. 1A). The predicted DR5 precursor is a 411–amino acid type I transmembrane protein. DR5 shows more sequence identity to DR4 (55%) than to other apoptosis-linked receptors, namely, DR3 (also called Apo3, WSL-1, or TRAMP) (13-16) (29%), TNFR1 (19%), or CD95 (17%). DR5 and DR4 each contain two extracellular cysteine-rich domains (CRDs) (Fig. 1A), whereas other mammalian TNFR family members have three or more CRDs (17). DR5 contains a cytoplasmic death domain that shows substantially more identity to the death domain of DR4 (64%) than to the death domain of DR3 (29%), TNFR1 (30%), or CD95 (19%).
Primary structure and mRNA expression of DR5 and DcR1. The nucleotide sequences have been deposited with GenBank (accession numbers AF012535 and AF012536, respectively). (A) The deduced sequences of human DR5 and DcR1 are aligned with human DR4. Also included are the death domains of DR3, TNFR1, and CD95. Shown are predicted cysteine-rich domains (CRD1, 2), transmembrane domains or hydrophobic COOH-terminus (underlined), N-linked glycosylation sites (black boxes), and sequence pseudo-repeats (brackets). Signal cleavage sites (arrows) were determined by protein NH2-terminal sequencing. Amino acid abbreviations are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. (B) Expression of DR5 and DcR1 mRNA was analyzed by hybridization of human tissue or tumor cell line poly(A) RNA blots (Clontech), with probes based on full-length DR5 or DcR1 cDNA.
Using a signal sequence trap approach and extracellular domain (ECD) homology (18, 19), we isolated an additional TNFR family member, which we named decoy receptor 1 (DcR1) (Fig. 1A). The DcR1 precursor is 259 amino acids long. DcR1 has a hydrophobic NH2-terminal sequence, followed by two CRDs. Downstream of the CRDs are five nearly identical tandem repeats, each 15 amino acids long; these repeats are followed by a hydrophobic COOH-terminus without an apparent cytoplasmic tail (Fig. 1A). This latter feature, together with the presence of a pair of small amino acids (Ala223and Ala224) just upstream of the hydrophobic COOH-terminus, suggests that DcR1 may be processed into a glycosyl-phosphatidylinositol (GPI)–anchored cell surface protein (20). The ECD of DcR1 is most closely related to those of DR4 (60% identity) and DR5 (50% identity) and contains five potential N-linked glycosylation sites (Fig. 1A).
We investigated the mRNA expression of DR5 and DcR1 in human tissues and tumor cell lines (Fig. 1B). We detected a single DR5 mRNA transcript and several DcR1 transcripts in multiple tissues; the ∼1.5-kb DcR1 transcript corresponded in size to the cloned DcR1 cDNA. DR5 expression was relatively high in fetal liver and lung, and in adult PBL, ovary, spleen, liver, and lung. DcR1 expression was highest in PBL, spleen, lung, and placenta. Most of the tumor cell lines expressed DR5, but showed little or no expression of DcR1.
The sequence similarities between DR5, DcR1, and DR4 suggested that these receptors may interact with a common ligand. Epitope-tagged fusion proteins based on the ECD of DR5 or DcR1 (21) each coprecipitated with soluble TRAIL (22) (Fig.2A). Other cytotoxic TNF family members, namely, TNF, lymphotoxin-α, or CD95L, did not bind the DR5 or DcR1 ECDs (23). Thus, DR5 and DcR1 associate specifically with TRAIL.
(A) Interaction of DR5 and DcR1 with TRAIL. Supernatants from pRK5 vector-transfected 293 cells or from cells transfected by pRK5 encoding FLAG epitope–tagged DR5 or DcR1 ECD (5 ml) (21) were incubated with 1 μg of soluble, poly(His)-tagged TRAIL (22) for 30 min at 24°C. Complex formation with DR5 (top) or DcR1 (bottom) was tested by immunoprecipitation (IP) with anti-FLAG–conjugated (Sigma) or Ni-conjugated (Qiagen) agarose beads, followed by electrophoresis under reducing conditions and protein immunoblot (western blot, WB) with anti-TRAIL (34). (B) Effect of PI-PLC on the binding of TRAIL to DcR1-transfected cells. 293 cells were transiently transfected (35) by pRK5 vector or pRK5 encoding full-length DcR1. After 18 hours, the cells were put into suspension, treated with buffer (solid bars) or recombinant PI-PLC (1 μg/ml) (shaded bars) (24) for 2 hours at 37°C, and the binding of 125I-TRAIL (0.2 ng) to intact cells (106 per tube) was analyzed. Nonspecific binding was measured in the presence of 500-fold excess unlabeled TRAIL. Data are the means ± SEM of triplicate determinations.
To test whether DcR1 is GPI-linked, we analyzed the effect of recombinant phosphatidylinositol-specific phospholipase C (PI-PLC) (24) on the binding of TRAIL to intact DcR1-transfected cells (Fig. 2B). Transfection of human 293 cells by DcR1 led to an increase in the amount of specific TRAIL binding, consistent with interaction between DcR1 and TRAIL. DcR1 was not detected in the supernatants of DcR1-transfected cells (25), indicating that the protein is not secreted into the medium. Treatment by PI-PLC caused a marked reduction in TRAIL binding to cells (Fig. 2B), supporting the notion that DcR1 is GPI-anchored. This conclusion was substantiated by a 58% reduction in epitope tag–directed immunofluorescent staining of cells transfected with epitope-tagged DcR1 after PI-PLC treatment (26).
Because death domains function as oligomerization interfaces, overexpression of receptors that contain such domains leads to activation of signaling in the absence of ligand (2). To investigate whether DR5 can induce cell death, we transfected 293 or HeLa cells with a DR5 expression plasmid and assessed the level of apoptosis after 24 hours. DR5-transfected cells underwent apoptosis, as indicated by morphological changes, internucleosomal DNA fragmentation, and exposure of phosphatidylserine on the cell surface (Fig.3, A to C). The caspase inhibitors CrmA, DEVD-fmk, and z-VAD-fmk blocked apoptosis activation by DR5, indicating caspase involvement in this response. A dominant-negative form of the adapter FADD (FADD-DN), which blocks apoptosis induction by CD95, TNFR1, or DR3 (5, 13, 27) but not by TRAIL (10) or DR4 (11), did not inhibit apoptosis induction by DR5 (Fig. 3C), indicating that DR5 signals apoptosis independently of FADD. Consistent with previous work (10), TRAIL induced apoptosis in HeLa cells, which was blocked by immunoglobulin-fusion proteins (immunoadhesins) (28-30) based on the ECD of DR5, DcR1, or DR4, but not TNFR1 (Fig. 3D), thus confirming a specific interaction between TRAIL and DR5, DcR1, or DR4.
DR5 signaling. Human 293 cells (A and B) or HeLa cells (C) were transfected (35) by pRK5-based plasmids encoding DR5 or DR4, alone or together with plasmids encoding CrmA or FADD80 – 205 (FADD-DN). DEVD-fmk (Enzyme Systems) or z-VAD-fmk (Research Biochemicals) (200 μM) were added where indicated at the time of transfection. Apoptosis was assessed 24 hours later by morphology (A), DNA fragmentation (B), or FACS analysis of phosphatidylserine exposure (C) (14). Data in (C) are the means ± SEM of at least three experiments. (D) TRAIL (0.5 μg/ml) was preincubated (1 hour, 24°C) with immunoadhesins based on DR5 (○), DcR1 (▵), DR4 (•), or TNFR1 (▴) (28) and added to HeLa cells. Five hours later, the cells were analyzed for apoptosis by FACS. (E) 293 cells were transfected by pRK5 or pRK5 encoding DR5, DR4, or DR3 in the presence of z-VAD-fmk and analyzed 24 hours later for NF-κB activity (34). (F) HeLa, 293, or MCF7 cells were treated with TRAIL or TNF (30 min, 1 μg/ml) and analyzed for NF-κB activation.
In addition to inducing apoptosis, TNFR1, CD95, and DR3 activate the transcription factor nuclear factor kappa B (NF-κB) (13-16, 31, 32), which controls expression of multiple immunomodulatory genes (33). Previous work suggested that DR4 is not linked to NF-κB, because transfection of DR4 in MCF7 cells did not lead to NF-κB activation (11). However, upon transfection into HeLa cells, DR5, DR4, and DR3 induced NF-κB activation (Fig. 3E). Antibody to the p65 subunit of NF-κB inhibited the mobility of the NF-κB probe, implicating p65 in the response to all three receptors. TRAIL also induced detectable NF-κB activation in HeLa and 293 cells, but not in MCF7 cells (Fig. 3F); TNF induced a more pronounced activation in each cell line. Thus, TRAIL activates NF-κB in a cell type–dependent manner, and both DR5 and DR4 can mediate this function. Dose-response analysis showed that TNF activates NF-κB at substantially lower concentrations than does TRAIL (25), suggesting distinct signaling mechanisms for NF-κB induction.
The absence of a cytoplasmic region in DcR1 suggested that this receptor is involved in modulation, rather than in actual transduction, of TRAIL signaling. We investigated the effect of DcR1 expression on cellular responsiveness to TRAIL. Ectopic expression of DcR1 reduced sensitivity to apoptosis induction by TRAIL in 293 cells (Fig. 4A), as well as in HeLa cells (26). Six of the eight tumor cell lines that expressed little or no DcR1—HL-60, HeLa, MOLT-4, Raji, SW40, and A549 (Fig.1B)—were sensitive to TRAIL-induced apoptosis (8-10, 26). In contrast, primary human umbilical vein endothelial cells (HUVEC), human microvascular endothelial cells (HUMEC), and PBLs, which expressed DcR1 (25) (Fig. 1B), were resistant to TRAIL (Fig. 4, C and D) (25). PI-PLC treatment of untransfected 293 cells sensitized these cells to apoptosis induction by TRAIL, but not by antibody to CD95 (anti-CD95) (Fig. 4B), consistent with removal of endogenous GPI-linked DcR1 from the cell surface. In addition, PI-PLC treatment of HUVEC or HUMEC sensitized these cells to TRAIL-induced apoptosis (Fig.4, C and D). Hence, DcR1 inhibits TRAIL function, and DcR1 expression correlates with resistance to TRAIL.
Inhibition of TRAIL function by DcR1. (A) 293 cells were transfected by pRK5 (open symbols) or pRK5 encoding DcR1 (solid symbols) plus pRK5 encoding GFP (36). After 18 hours, buffer (triangles) or TRAIL (0.5 μg/ml) (circles) was added, and GFP-positive cells were examined for apoptotic morphology under a fluorescence microscope (Leica) equipped with Hoffmann optics. (B) 293 cell monolayers were treated in the culture dish with buffer (open bars) or PI-PLC (1 μg/ml, 2 hours at 37oC) (solid bars), washed, incubated for 6 hours with buffer, TRAIL (0.1 μg/ml), or anti-CD95 (CH-11) (0.5 μg/ml plus 1 μg/ml cyclohexamide), and scored for apoptosis (37). HUVEC (Clonetics) (C) or HUMEC (Cell Systems) (D) were treated with buffer (triangles) or PI-PLC (circles) as in (B), washed, incubated for 6 hours with buffer (open symbols) or TRAIL (0.1 μg/ml) (solid symbols), and scored for apoptosis. Cyclohexamide (1 μg/ml) was added to all incubations to prevent resynthesis of DcR1. Data are the means ± SEM of triplicate determinations, each consisting of 100 to 200 cells.
The existence of multiple receptors for TRAIL suggests an unexpected complexity in the regulation of signaling by this cytokine. The two signaling receptors, DR4 and DR5, appear to be functionally redundant, and their expression patterns are not sufficiently different to suggest a distinct, tissue-specific involvement in TRAIL signaling. One possible explanation is that expression of DR4 and DR5 may differ at the level of individual cell types within tissues. The two receptors also may have additional, nonredundant signaling functions, perhaps mediated by regions outside the death domain.
TRAIL, DR4, and DR5 are expressed in multiple human tissues. The expression of a decoy receptor for TRAIL in normal tissues but not in many tumor cell lines suggests an explanation for the resistance of normal tissues and the broad sensitivity of tumor cell lines to TRAIL-induced apoptosis. Several TNFR superfamily members (for example, TNFR1 and TNFR2) are shed from the cell surface to form soluble inhibitors that neutralize their ligands at remote locations, for example, in the bloodstream (17). As a membrane-anchored protein, DcR1 can inhibit responsiveness to its ligand directly at the cell surface. Perhaps this mode of regulation represents a general mechanism that protects cells against the action of potent pro-apoptotic cytokines.
