Fas Preassociation Required for Apoptosis Signaling and Dominant Inhibition by Pathogenic Mutations

See allHide authors and affiliations

Science  30 Jun 2000:
Vol. 288, Issue 5475, pp. 2354-2357
DOI: 10.1126/science.288.5475.2354


Heterozygous mutations encoding abnormal forms of the death receptor Fas dominantly interfere with Fas-induced lymphocyte apoptosis in human autoimmune lymphoproliferative syndrome. This effect, rather than depending on ligand-induced receptor oligomerization, was found to stem from ligand- independent interaction of wild-type and mutant Fas receptors through a specific region in the extracellular domain. Preassociated Fas complexes were found in living cells by means of fluorescence resonance energy transfer between variants of green fluorescent protein. These results show that formation of preassociated receptor complexes is necessary for Fas signaling and dominant interference in human disease.

Fas (CD95 or APO-1) is a cell surface receptor that transduces apoptotic signals critical for immune homeostasis and tolerance (1–3). The Fas protein is a 317–amino acid type 1 transmembrane glycoprotein with three extracellular cysteine-rich domains (CRDs) that are characteristic of the tumor necrosis factor receptor (TNFR) superfamily. Both Fas and Fas ligand (FasL) are predicted to form trimers, with CRD2 and CRD3 forming the major contact surfaces for FasL (4, 5). The Fas cytoplasmic portion contains a death domain that rapidly recruits the adaptor molecule FADD (Fas-associated death domain protein) and the caspase-8 proenzyme after binding of FasL or agonistic antibodies, leading to caspase activation and apoptosis (6–10).

Patients with autoimmune lymphoproliferative syndrome (ALPS) type 1A have heterozygous germ line mutations in the APT-1 Fas gene. Their lymphocytes are resistant to Fas-induced apoptosis, and transfection of the mutant allele causes dominant interference with apoptosis induced through Fas (11–16). This was thought to result from ligand-mediated crosslinking of wild-type and defective Fas chains into mixed trimer complexes. However, a mutation that causes an extracellular domain deletion of most of CRD2 (ALPS Pt 2, deletion of amino acids 52 to 96) as a result of altered RNA splicing shows no binding to agonistic antibodies or FasL, but still dominantly interferes with Fas-induced apoptosis almost as efficiently as does a death domain mutant [ALPS Pt 6, Ala241 → Asp (A241D)] (Fig. 1A) (13, 17). Control experiments showed equal cell surface expression of the wild-type and mutant Fas molecules (18). Thus, dominant interference cannot be explained by the conventional model of signaling by FasL-induced oligomerization of receptor monomers because, in this scheme, the Pt 2 mutant Fas molecule would not become part of a mixed receptor complex. We therefore tested for ligand-independent interactions between Pt 2 Fas and wild-type Fas. Both full-length and Pt 2 Fas coprecipitated with a Fas 1–210:GFP chimera in which green fluorescent protein (GFP) replaces the death domain (Fig. 1C). This interaction was specific, because the TNFR family receptors TNFR2/p80 and HveA did not interact with Fas (1).

Figure 1

(A) Surface expression and binding characteristics of the indicated wild-type and mutant Fas molecules transfected into 293T HEK cells. The bold line shows specific staining and the thin line indicates background staining of mock-transfected cells. AU-1 or HA staining was performed to confirm surface expression of each protein. Staining of functional receptors with FasL was performed with a preparation of FasL trimerized via a modified leucine zipper (FasL-LZ) and LZ mAb. Numbers indicate the percentage of positively staining cells in the indicated gates. Cell loss of transfected murine BW cells was calculated as described (13, 22). (B) Schematic of the Fas protein. Numbering is based on (21). TM, transmembrane domain. (C) Self-association of Fas molecules. 293T cells cotransfected with the indicated constructs were lysed and immunoprecipitated, and Western blots (WB) were probed as described (29). The open circle indicates the immunoglobulin heavy chain in the immunoprecipitates (IP). (D) Dominant interference depends on the PLAD. Jurkat cells were transfected with the indicated Fas expression vectors. Percentages represent apoptotic GFP+ cells staining positive for Annexin-V PE (Pharmingen) due to surface exposure of phosphatidylserine after treatment with Fas mAb (15). Results are representative of three independent transfections. (E) Induction and inhibition of apoptosis by the non–ligand-binding R86S Fas mutant. Murine BW cells were transfected with the indicated expression vectors, and apoptosis was induced with Fas mAb (solid bars) and FasL (hatched bars) as described (13, 22).

We have found that TNFR superfamily members share a self-association domain in CRD1, termed the “pre-ligand assembly domain” (PLAD) (Fig. 1B) (19). To test whether Fas contains a functional PLAD, we constructed hemagglutinin (HA)-tagged NH2-terminal Fas truncations (20). Deleting the first subdomain in CRD1 (amino acids 1 to 42) (21) substantially reduced ligand binding but did not prevent binding of the Fas monoclonal antibody (mAb) APO-1. Deleting the entire CRD1 (amino acids 1 to 66) abrogated binding of both FasL and Fas mAb (Fig. 1A). Both truncations eliminated coprecipitation with a differentially tagged Fas molecule and abrogated apoptosis signaling; this result indicates that the NH2-terminus of Fas, including CRD1, functions as a PLAD (Fig. 1, C and D) (19,22). Fas mutants from ALPS patients with truncated or mutated death domains are potent dominant-negative inhibitors of normal Fas function. However, if the PLAD was removed from Fas molecules lacking the death domain (Fas 1–210) or harboring an ALPS death domain point mutation [ALPS Pt 26, Asp244 → Val (D244V)], dominant interference was lost (Fig. 1D).

To further explore the requirement for ligand binding in receptor self-association, we tested the Fas point mutation Arg86→ Ser (R86S) that removes a critical CRD2 contact residue for FasL (5) and prevents FasL binding (Fig. 1A). The overall receptor structure was preserved, as indicated by staining with APO-1 and DX2 Fas mAbs (Fig. 1A) (18), and self-association with wild-type Fas still occurred (Fig. 1C). Even though it could not bind FasL, this mutant dominantly interfered with FasL-induced apoptosis through wild-type Fas (Fig. 1E). Apoptosis induced by Fas mAb in the same cells was unimpaired, which indicated that Fas was functionally expressed on the cell surface (Fig. 1E). Thus, receptor self-association is independent of ligand binding, yet critical for both normal function and dominant interference.

To quantitate Fas receptor self-association in living cells, we developed flow cytometric and microscopic assays based on fluorescence resonance energy transfer (FRET) between spectral variants of GFP. [See protocol at Science's STKE (;2000/38/pl1)]. When in close proximity (<100 Å), cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP) will exhibit FRET (23). Flow cytometry using CFP excitation of cells expressing both CFP and YFP Fas fusion proteins triggered strong fluorescence emission at the YFP wavelength attributable to FRET (Fig. 2A) (24). FRET was detected between Fas fusion proteins with or without the death domain, but not between Fas and other TNFR family members, such as TNFR1/p60, HveA, or DR4 (Fig. 2) (1, 18). Microscopic measurement of CFP dequenching after selectively photobleaching YFP yielded a FRET efficiency of 11%. With the death domain on both molecules, FRET efficiency rose to 23%, consistent with increased oligomerization via the death domain (25). Pt 2 Fas gave a FRET efficiency comparable to that of Fas 1–210, but there was reduced signal with Fas 43–210 and no detectable FRET with Fas 67–210 or the DR4 control (Fig. 2B). Hence, Fas molecules are in close proximity on the surface of living cells, and this proximity depends on the PLAD.

Figure 2

(A) Dot plots showing relationships between CFP, YFP, and FRET signals in the indicated cotransfectants. CFP and YFP fusion proteins were constructed, transfected into 293T cells, and analyzed on a FACSvantage cytometer (24). Numbers are the percentage of cells positive for CFP or YFP with FRET signal (top right quadrant). The bottom panel shows FRET from a construct in which CFP was covalently fused to YFP through a nine–amino acid peptide linker (CFP-YFP) (30). (B) FRET efficiency (E%) for the indicated CFP and YFP pairs, as determined by CFP dequenching after photobleaching of YFP on individual cells (five readings of four- to seven-cell regions) (24). The numbers represent the average E% and standard error for each plasmid pair.

To test whether native Fas receptors normally self-associate, we performed chemical crosslinking studies on H9 T lymphoma cells expressing endogenous human Fas (Fig. 3). Crosslinking shifted the apparent molecular size of Fas in deglycosylated cell lysates from 45 to 140 kD under nonreducing conditions (Fig. 3A). Densitometry suggested that 60% of the Fas chains were crosslinked. Unlike FasL- or Fas mAb–treated cells, Fas complexes from surface-crosslinked cells showed only partial recruitment of FADD and no recruitment of caspase-8, with no cleavage of the downstream caspase substrate poly(ADP-ribose)polymerase (PARP). Interestingly, crosslinking prevented the formation of active signaling complexes in response to subsequent treatment with agonistic mAb (Fig. 3B).

Figure 3

Preassociation of endogenous Fas receptor chains. H9 lymphoma cells (107) were treated for 30 min with 10 mM of the thiol-cleavable crosslinker 3,3′-dithiobis[sulfosuccinimidyl propionate] (DTSSP) and/or stimulated with the indicated reagents or were incubated in medium alone (No Tx). (A) Western blots of deglycosylated cell lysates run under reducing or nonreducing conditions were treated with and probed with Fas mAb. The arrow indicates the position of the major crosslinked species. Size markers (in kilodaltons) are at the right. (B) After treatment with the indicated reagents, cells were lysed, immunoprecipitated, and blotted for FADD and caspase-8 as described (15). The positions of the two isoforms of procaspase-8 (p54/52) and caspase-8 cleavage products after proteolysis of the p11 caspase subunit (p43/41) are shown with arrows. PARP cleavage was analyzed on lysates from cells cultured at 37°C for an additional 4 hours after the indicated treatments. The upper band is the 115-kD full-length PARP; the lower band is the 85-kD signature fragment produced by caspase cleavage during apoptosis. The results are representative of at least three independent experiments.

Our findings redefine how death signals are triggered through Fas and how mutations in ALPS dominantly interfere with normal Fas function. In a large number of ALPS patients evaluated at the National Institutes of Health (13–15, 17), we found that the PLAD was preserved in every example of a dominant-interfering mutation associated with disease, including mutations that create premature termination polypeptides encoding only the first 57 and 62 amino acids of the mature Fas protein. To cause dominant interference, mutant proteins must physically interact with wild-type proteins in a functional complex (26). Previously, dominant-negative receptor mutations associated with human diseases have been shown to interfere with normal receptor signaling by sequestering ligand, blocking intracellular signaling, or preventing transport of the wild-type chain to the cell surface (27). For Fas, dominant interference stems instead from PLAD-mediated association between wild-type and mutant receptors before ligand binding. PLAD interactions are essential for ligand binding and signaling and have been observed in the regulation of apoptosis by soluble alternatively spliced forms of Fas that all include this domain (28). PLAD-mediated dominant interference may also play a role in modulation of signaling by decoy receptors (2) and in the pathogenesis of diseases due to heterozygous genetic abnormalities in other members of the TNFR superfamily.

  • * To whom correspondence should be addressed. E-mail: lenardo{at}


View Abstract

Navigate This Article