Structure
Article
Solution NMR Investigation of the CD95/FADDHomotypic Death Domain Complex SuggestsLack of Engagement of the CD95 C TerminusDiego Esposito,1 Andrew Sankar,1 Nina Morgner,2 Carol V. Robinson,2 Katrin Rittinger,1 and Paul C. Driscoll1,*1Division of Molecular Structure, MRC-National Institute for Medical Research, Mill Hill, London NW7 1AA, UK2Department of Chemistry, Physical and Theoretical Chemistry Laboratory, University of Oxford, Oxford OX1 3QZ, UK
*Correspondence: [email protected] 10.1016/j.str.2010.08.006
SUMMARY
We have addressed complex formation between thedeath domain (DD) of the death receptor CD95 (Fas/APO-1) with the DD of immediate adaptor proteinFADD using nuclear magnetic resonance (NMR)spectroscopy, mass spectrometry, and size-exclu-sion chromatography with in-line light scattering.We find complexation to be independent of theC-terminal 12 residues of CD95 and insensitive tomutation of residues that engage in the high-orderclustering of CD95-DD molecules in a recentlyreported crystal structure obtained at pH 4. Differen-tial NMR linewidths indicate that the C-terminalregion of the CD95 chains remains in a disorderedstate and 13C-methyl TROSY data are consistentwith a lack of high degree of symmetry for the com-plex. The overall molecular mass of the complex isinconsistent with that in the crystal structure, andthe complex dissociates at pH 4. We discuss thesefindings using sequence analysis of CD95 orthologsand the effect of FADD mutations on the interactionwith CD95.
INTRODUCTION
Ligation of death receptor (DR) CD95 (Fas/APO-1) by the
cognate CD95-ligand (CD95L/FasL) or specific classes of anti-
CD95 monoclonal antibodies (mAbs) can initiate apoptotic path-
ways that play a central role in the control of homeostasis and
lymphocyte populations in multicellular organisms (Krammer,
2000; Krueger et al., 2003; Nagata, 1997; Strasser et al., 2009).
The DR family is defined by the presence of a �90 amino acid
residue-long cytoplasmic domain, known as a death domain
(DD) (Ashkenazi and Dixit, 1998; Park et al., 2007a; Wilson
et al., 2009). The DD is the focus for the binding of adaptor
proteins during the signaling events that follow DR engagement,
illustrated most pertinently by CD95-DD mutations that lead
to type Ia autoimmune lymphoproliferative syndrome (ALPS)
(Bidere et al., 2006). CD95 is present on the plasma membrane
as a preassociated homo-oligomeric receptor (Papoff et al.,
1999; Siegel et al., 2000). CD95L (or agonistic mAb) administra-
1378 Structure 18, 1378–1390, October 13, 2010 ª2010 Elsevier Ltd
tion rapidly triggers reorganization of CD95multimers (Algeciras-
Schimnich et al., 2002; Kischkel et al., 1995; Medema et al.,
1997) and the recruitment to the receptor of the adaptor protein
FADD (Boldin et al., 1995; Chinnaiyan et al., 1995). FADD
comprises an N-terminal death effector domain (FADD-DED)
tethered to the C-terminal death domain (FADD-DD). Mutual
association of FADD with CD95 via their DDs leads to binding,
by interaction with FADD-DED, of the tandem DED domains of
the apical caspase-8 to form the death-inducing signaling
complex (DISC) (Kischkel et al., 1995;Medema et al., 1997). Cas-
pase-8 is then activated by transautoproteolysis, and the cata-
lytic subunits dissociate from the DISC to effect cleavage of
cytoplasmic apoptotic effector proteins (Krammer, 2000; Peter
et al., 2007; Thorburn, 2004; Wilson et al., 2009).
The homotypic CD95-DD/FADD-DD interaction is an essential
step for apoptotic CD95 signaling. Overexpression of the FADD-
DD on its own is sufficient to block DISC formation and cell death
indicating dominant-negative behavior (Newton et al., 1998,
2001), and the analysis of several CD95-DD mutations has
shown that the breakdown in signaling occurs as a result of
a failure to bind FADD (Huang et al., 1996; Martin et al., 1999).
Currently many aspects of the precise mechanism of DISC
assembly, architecture and evolution in time within the cell
remain unresolved. Experimental approaches to probe the
earliest events in CD95 signaling have adopted a number of
agonists including native soluble CD95L (sCD95L), affinity-
tagged variants of sCD95L that can be cross-linked by a
secondary reagent, membrane-bound CD95L (mCD95L), and
several anti-CD95 mAbs. A consensus is emerging that the
precise pattern of DISC behavior can vary from one agonist to
another (Beneteau et al., 2007; Chaigne-Delalande et al., 2009;
Holler et al., 2003; Huang et al., 1999; Legembre et al., 2003;
O’Reilly et al., 2009; Schneider et al., 1998). Early in the signaling
process CD95 is commonly found in microaggregates contain-
ing high molecular weight (�200 kD) SDS- and b-mercaptoetha-
nol-resistant CD95 oligomers (Algeciras-Schimnich et al., 2002;
Feig et al., 2007; Holler et al., 2003; Kischkel et al., 1995). These
and related observations (Lee et al., 2006; Siegel et al., 2004)
have led to the description of the CD95/FADD/caspase-8 DISC
as an apoptotic signaling ‘‘platform.’’
A fundamental question in CD95-signaling is how the receptor
can transit from the ‘‘resting’’ state, where it displays no interac-
tion with the other DISC components, to one in the presence of
agonist that binds FADD. In vitro DISC reconstitution has been
demonstrated using fusion proteins that effectively cross-link
All rights reserved
Figure 1. Structure of Hetero-Octamer Assembly of CD95 and FADD DDs in the Crystal Structure Protein Data Bank Code 3EZQ
(A and B) Orthogonal views of the 4+4 hetero-octameric assembly CD95 DD chains (blue) and FADD DD chains (green) representing 50% of the contents of the
asymmetric unit in 3EZQ (Scott et al., 2009).
(C) Superposition (over helices a1–a4) of a single CD95 DD chain (blue) from 3EZQ on the solution structure of isolated CD95-DD+CT (pink) obtained by NMR at
pH 4.0 (PDB code 1DDF) (Huang et al., 1996), highlighting the closed-to-open rearrangement of the C-terminal region to form stem and C-helices.
(D) Amino acid sequences and observed regular secondary structure elements for the CD95 DD in 3EZQ and 1DDF, and FADD DD in 3EZQ and the essentially
identical solution structure conformation (PDB code 13EY) (Berglund et al., 2000). Residues discussed in the text are depicted in color in the 3EZQ ribbon
diagrams and in the amino acid sequences: CD95 K263 (orange), mutation of which to Ala yields a more tractable DD construct (Huang et al., 1996); CD95
Q311 and L315 (magenta) involved in contacts between CD95-DD homodimers; and residues reported to be important for CD95:FADD interaction (red) including
sites of type Ia ALPS mutations in CD95 A257D and D260A (Martin et al., 1999; Vaishnaw et al., 1999) and FADD residues R113, R117, D123, K125, and R146
(Hill et al., 2004). Further depiction of interdomain contacts is provided in Figure S1.
Structure
NMR of CD95/FADD Death Domain Complex
CD95-DDs (Hughes et al., 2009; Roy et al., 2008). On the other
hand, it has been observed that the isolated DDs of CD95 and
FADD can bind each other directly to form a homotypic
protein-protein complex (Boldin et al., 1995; Ferguson et al.,
2007; Scott et al., 2009; Yang et al., 2005). Understanding the
three-dimensional structure of this complex has the potential
to inform the basis of DISC assembly, because it presumably
reflects a state of the intact CD95 receptor that is competent
to form interactions with the adaptor protein. In vitro, recombi-
nant CD95-DD tends to form soluble aggregates at neutral pH.
The nuclear magnetic resonance (NMR) structure of monomeric
wild-type CD95-DD was obtained at pH 4, and mutations of
certain exposed residues can mitigate against CD95-DD self-
association (Huang et al., 1996), although mostly with loss of
FADD-binding. We have previously reported that CD95-DD
tagged with a solubility enhancement tag (SET) allowed for
Structure 18, 1378–1
a preliminary NMR analysis of CD95-DD interactions with
FADD-DD (Ferguson et al., 2007). More recently, the crystal
structure was reported of the complex of wild-type CD95-DD
with FADD-DD prepared from mixed Escherichia coli lysates,
with crystal growth achieved at pH 4.0 (Scott et al., 2009). The
asymmetric unit contains two ‘‘4+4’’ hetero-octameric DD
assemblies (Figure 1), within which the CD95-DD chain is
dramatically reconfigured compared to the structure in solution
(Huang et al., 1996). The C-terminal part of the chain comprising
the canonical helices a5 and a6 rearranges to form an elongated
‘‘stem’’-helix in a a-a hairpin interaction with a ‘‘C’’-helix
formed from the extremeC-terminal residues that are disordered
in the free CD95-DD (Huang et al., 1996). The rearrangement
exposes the hydrophobic core of CD95-DD, and sequesters
FADD-DD with a normal DD conformation (Figure 1). The overall
complex comprises a tetramer of the CD95-DD:FADD-DD
390, October 13, 2010 ª2010 Elsevier Ltd All rights reserved 1379
Figure 2. GST Pull-Down Assay of CD95-DD:FADD-DD Interactions
Both the intact intracellular domain of CD95 (GST-CD95-IC, lane 2) and the
death domain lacking the extreme C-terminal tail of CD95 (GST-CD95-DD,
lane 4) bind to FADD-DD. The interaction is abolished on mutation of either
CD95 A257D (corresponding to a type Ia ALPS lesion, lane 6) or FADD
R117E (lane 9), as before (Ferguson et al., 2007; Hill et al., 2004; Martin
et al., 1999). Controls indicate that GST alone (lane 8) or the glutathione-modi-
fied beads (lane 10) do not bind FADD-DD. See also Figure S2.
Structure
NMR of CD95/FADD Death Domain Complex
heterodimers where contacts are only made between the CD95
stem/C-helix regions (Scott et al., 2009) (see Figure S1 available
online).
The crystal structure is notable in that it provides a dramatic
example of protein structure plasticity and can rationalize at least
a subset of the structure-activity data available for CD95-FADD
interactions. It is important to establish whether the model
for the formation of the core DISC is more widely applicable to
homologous death receptor interactions with FADD (Ashkenazi,
2008; Bodmer et al., 2000; Kischkel et al., 2000; Sprick et al.,
2002). We previously reported that high molecular weight com-
plexes of FADD-DD with a Gb1-CD95-DD construct lacking
the 12 C-terminal residues (Ferguson et al., 2007). In this report
we describe further biophysical measurements of soluble CD95-
DD:FADD-DD complexes under near physiological conditions.
We address the dependence of complex formation on the
C-terminal tail region of CD95, the sample pH and mutations in
CD95-DD and FADD-DD, and discuss the results with reference
to a sequence analysis of emergent CD95 orthologs. We con-
clude that in solution the complex formed between CD95 and
FADD DDs has characteristics significantly different from that
predicted by the current crystal structure.
RESULTS
Complexation of the DDs of CD95 and FADDThere are conflicting reports concerning the ability of FADD-DD
alone to interact with CD95 (Boldin et al., 1995; Ferguson et al.,
2007; Muppidi et al., 2006; Sandu et al., 2006; Yang et al.,
2005). In our hands, both the intact intracellular domain of
CD95 (CD95-IC, residues 194–335) and the CD95 DD (residues
218–323) tagged to glutathione S-transferase (GST) are able to
pull down recombinant FADD-DD (residues 89–192) (Figure 2).
The interaction is abrogated by introducing either the substitu-
tion A257D in CD95-DD or R117E in FADD-DD. CD95 A257D
corresponds to a type Ia ALPS lesion for which the globular DD
structure is maintained (Martin et al., 1999). Similarly FADD-DD
R117E is a stably folded variant of FADD-DD (Figure S2) with
reduced CD95 binding activity (Hill et al., 2004). The CD95-IC
protein pulls down FADD-DD at least as efficiently as the shorter
CD95-DD. Therefore, in this experiment, the extra CD95-IC resi-
dues outside the globular DD do not interfere in the interaction
with FADD-DD (Chinnaiyan et al., 1995). These results demon-
strate a direct interaction between CD95-DD and FADD-DD
that can be abrogated by point mutations on the surface of either
protein.
Wild-type CD95-DD has an intrinsic tendency to self-asso-
ciate at physiological pH (Huang et al., 1996; Scott et al.,
2009). The fusion of CD95-DD to the C terminus of a SET such
as GST or the B1 domain from streptococcal immunoglobulin-
binding protein G (Gb1) (Zhou et al., 2001) can significantly
enhance its tractability (Ferguson et al., 2007; Huth et al.,
1997). For the present analysis we sought to avoid the use of
an appended SET by exploiting the observation that the K263A
mutant of CD95-DD displays a markedly reduced tendency to
self-aggregation while retaining FADD binding (Ferguson et al.,
2007; Huang et al., 1996). It was previously reported that
a CD95 construct lacking the C-terminal 15 residues makes
a stronger interaction with FADD than the full length receptor,
1380 Structure 18, 1378–1390, October 13, 2010 ª2010 Elsevier Ltd
leading to the declaration of an inhibitory role for this region
(Chinnaiyan et al., 1995). However, in the recently reported
crystal structure of the complex of copurified CD95 and FADD
DDs this C-terminal region seems to carry out a critical role in
supporting the ‘‘bridge’’ between two ‘‘opened’’ CD95 chains
(Scott et al., 2009) (Figure S1). To probe the role of the C-terminal
residues of CD95 in the interaction with FADD-DD using solution
NMR spectroscopy, we created CD95(K263A) DD constructs
that either omitted (CD95-DD) or included (CD95-DD+CT) the
extreme C-terminal tail (CT) region.
Size-Exclusion Chromatography-Multiangle LightScattering AnalysisWe analyzed the complexes formed by CD95-DD and CD95-
DD+CT with FADD-DD by analytical size-exclusion chromatog-
raphy (SEC) coupled to multiangle light scattering analysis
(SEC-MALS) (Figure 3). Samples were run over a concentration
range corresponding to 2–10 mg/ml total protein. We obtained
an essentially symmetric chromatographic light scattering profile
consistent with a monodisperse particle with elution volume
12.4 ml and 12.2 ml, respectively, in the range of 4 to 10 mg/ml.
The elution volume was slightly larger at 2 mg/ml, suggesting
some limited dissociation of the complex at the lowest concen-
tration tested. The absence of any significant readout at elution
volumes >13 ml indicates the absence of any substantial popu-
lation of intermediate heteromeric species, suggesting that the
complex formation displays a considerable degree of coopera-
tivity. When fractions from semipreparative gel filtration experi-
ments are analyzed by SDS-PAGE, the eluted proteins appear
to be present in approximately equal amounts (Figure S3) con-
sistent with amino acid analysis of the excised gel bands
All rights reserved
Figure 3. SEC-MALS Analysis of Isolated
CD95-DD:FADD-DD Complexes
Both CD95-DD (A) and CD95-DD+CT (B) form
high molecular weight assemblies with FADD-DD
that elute at consistent volumes over a range of
concentration. The predicted particle molecular
masses from the MALS analysis are: 105, 105,
105, 103, and 100 kD for CD95-DD:FADD-DD
and 111, 110, 109, 108, and 104 kD for CD95-
DD+CT at 10, 8, 6, 4, and 2 mg/ml total protein
concentration, respectively. See also Figure S3.
Structure
NMR of CD95/FADD Death Domain Complex
(CD95-DD:FADD-DD ratio �1.2; data not shown). The SEC-
MALS data clearly indicate that under chromatographic condi-
tions the two DDs form a complex with a molecular mass of
�105 kD for CD95-DD with FADD-DD and �110 kD for CD95-
DD+CT with FADD-DD. Given the predicted molecular masses
of the individual domains (CD95-DD, 12.0 kD; CD95-DD+CT,
13.4 kD; FADD-DD, 11.9 kD) these results are consistent with
the formation of a particle that comprises at least nine polypep-
tides. We note that for the DD constructs that we have employed
(that lack the flexible C-terminal tail and 6*His affinity tag present
in the FADD construct used for the crystal structure [Scott et al.,
2009]) the SEC-MALS-derived masses are substantially larger
than would be predicted for a 4:4 complex. Critically, these
data clearly show that FADD-DD and CD95-DD form a complex
with high molecular mass independent of the presence of the CT
region of CD95.
Mass SpectrometryPurified CD95-FADD DD complexes were subjected to capillary
nanoflow electrospray ionization-mass spectrometry (nanoESI-
MS). Under conditions designed to maintain noncovalent inter-
actions, this methodology provides a snapshot of the overall
molecular mass and subunit composition of macromolecular
assemblies (Hernandez and Robinson, 2007). The mass spectra
obtained show signals for monomeric FADD-DD and CD95-DD
or CD95-DD+CT species in the low mass-to-charge (m/z)
region, together with higher m/z ion-series attributable to the
complexes. The clearly dominant signals represent complex
species with an overall mass of 120.1 kD and 127.1 kD respec-
tively, matching a 10-chain, 5:5 stoichiometry based on the
internal mass references for the monomeric species (Figure 4).
The spectra also indicated the presence of minor components
with a higher mass, correspondingmainly to 5:6 and 6:5 species.
The 5:5 ion series persisted as the dominant signal over a range
of protein concentration from 0.6 to 12mg/ml, indicating specific
complex formation. Importantly, no ion series was observed
consistent with a stable 4:4 complex in solution. Thus FADD-
DD and CD95-DD or CD95-DD+CT form complexes in solution
that are robust to nanoESI-MS analysis, with predicted molec-
ular mass consistent with a 5:5 domain composition. Again,
the ability to form the complex was independent of the presence
of the CD95 CT region.
Structure 18, 1378–1
NMR Titration of 15N-Labeled CD95-DD Constructswith FADD-DDUsing a protocol based on that reported previously (McAlister
et al., 1996), we separately titrated 15N-labeled CD95-DD and
CD95-DD+CT with unlabeled FADD-DD at constant concentra-
tion of the CD95 protein (Figures 5A and 5B). In each case, as
the FADD-DD concentration is increased, the intensities of the
majority of CD95 cross-peaks in the 2D 15N-HSQC spectrum
monotonically decrease to a level below the detection limit,
leaving a small number of signals derived from regions of the
CD95 backbone or Asn, Gln, or Arg side chains that remain
highly flexible in the complex. We obtained sequence-specific
resonance assignments relevant to the conditions employed
in the titration experiments, permitting unambiguous identifica-
tion that the persistent CD95 cross-peaks belong to the extreme
N- and C-terminal regions of the protein, e.g., for CD95-
DD+CT backbone amide cross-peaks from E218-A221 at the
N-terminus together with S322–V335 and the side chain NH2
cross-peaks of N324, N326, N329, and Q332 and the R328
N3H cross-peak from the C terminus (Figure 5B). 15N-relaxation
measurements (Kay et al., 1989) performed on the isolated
CD95-DD+CT protein confirm the dynamic flexibility on the
ns-to-ps timescale of the backbone for these N- and C-terminal
residues before complexation (data not shown). The lack of any
detectable FADD-dependent chemical shift perturbation for
these peaks coupled with the maintenance of their limited 1HN
chemical shift dispersion in the spectrum of the complex indicate
that CD95 residues E218–A221 and S322–V335 remain in
a highly disordered state in the complex. A similar pattern of
results was obtained with the shorter 15N-CD95-DD construct;
cross-peaks remaining at the end of the titration derive from
the disordered N- (E218–A221) and C-terminal (S322–E323)
chain segments, respectively (Figure 5B).
NMR Titration of 15N-Labeled FADD-DDwith CD95-DD ConstructsWhen 15N-FADD-DD was titrated with unlabeled CD95-DD+CT
virtually all of the FADD-DD cross-peaks dramatically diminish
in intensity (Figures 5C and 5D). At the highest concentration
of CD95-DD+CT only a few cross-peaks from the N-terminal
and C-terminal regions remain; the end-point spectrum
comprises the cross-peaks for the backbone NH of M92 and
390, October 13, 2010 ª2010 Elsevier Ltd All rights reserved 1381
Figure 4. Nanoflow ESI-MS Analysis of Isolated CD95-DD:FADD-DD
Complexes
CD95-DD+CT (A) and CD95-DD (B) complexes with FADD-DD, exchanged
into ammonium acetate buffer, yield ESI-MS spectra with a dominant ion
series in each case (blue symbols) with overall molecular mass 127,152 ± 22 D
and 120,131 ± 40 D, respectively. The corresponding predicted masses for
a 5:5 heterodecamer, based on the experimentally determined masses of
the protomer components, are 127,140 D and 120,100 D. The spectra suggest
the presence of small amounts of particles with 6:5 (green symbols) and 5:6
(red symbols) FADD:CD95 stoichiometry.
Structure
NMR of CD95/FADD Death Domain Complex
the side chain NH2 protons of N102 (N terminus), and the back-
bone NHs of residues R189–A192, side chain NH2 protons of
Q187 and N188, and the R189 N3H (C terminus). Essentially
identical behavior was observed when a solution of 15N-labeled
FADD-DD was titrated with unlabeled CD95-DD (data not
shown).
Ile, Leu, and Val 13CH3-Methyl Transverse RelaxationOptimized Spectroscopy AnalysisWe analyzed the complexation of CD95 and FADD DD proteins
within the paradigm of [1H,13C]-methyl group labeling for Ile,
Leu, and Val (ILV) side chainswithin a uniformly deuterated back-
ground (Goto et al., 1999), exploiting ILV-13C-methyl transverse
relaxation optimized spectroscopy (13CH3-TROSY) (Tugarinov
et al., 2003). This approach allows for the recording of slowly
relaxing methyl group NMR signals from ILV residues within
multiprotein assemblies (Ruschak and Kay, 2010). We sepa-
rately prepared ILV-13CH3-labeled perdeuterated samples of
1382 Structure 18, 1378–1390, October 13, 2010 ª2010 Elsevier Ltd
CD95-DD+CT and FADD-DD; the corresponding 13CH3-TROSY
spectra are presented in Figure 6. In each case, the cross-peaks
are well resolved as expected for a small globular protein.
Addition of a perdeuterated sample of the DD binding partner
dramatically changes the appearance of the spectrum. When2H-labeled FADD-DD is added to ILV-13CH3-CD95-DD+CT the
intensity of all the cross-peaks is significantly diminished consis-
tent with complex formation (Figure 6A). However, the effects of
complex formation differ markedly for two different sets of cross-
peaks within the spectrum. First, a small number cross-peaks
retains a normal ‘‘monolithic’’ line shape, with relatively high
peak intensity, and unperturbed chemical shifts relative to those
for the free protein. These peaks can be ‘‘selected’’ by plotting
the spectrum at a relatively high base contour level as in Figure 6.
Despite equivalent data processing parameters, these peaks are
marginally broader than in the free-state spectrum. The assign-
ments of these lesser-perturbed peaks correspond to the set
of ILV residues from the flexible N- (V220, I222) and C-terminal
(I318, I331, L334, V335) tail regions of CD95-DD+CT. Plotting
the spectrum at a lower contour level reveals the signals for the
remaining ILV-13CH3 groups. For each such cross-peak in the
free-state spectrum, there appears to have been a dispersion
of the intensity into a local cluster of broader, distinct but often
overlapping features. These effects are most clearly identified
for the more highly resolved peaks in the spectrum of the free
protein (Figures 7A and 7B). For example, the I314 d1CH3
cross-peak is clearly converted into a cluster of at least four
cross-peaks in the spectrum of the complex. The 1H upfield-
shifted dCH3 group cross-peak of L229 appears to be dispersed
into at least four components with differing intensities (one
feature is at the level of the noise). A similar pattern of effects is
observedwhen 2H-labeled CD95-DD+CT is added to ILV-13CH3-
FADD-DD. Thus only a very small number of the FADD-DD13C-methyl cross-peaks retains a monolithic appearance with
unperturbed chemical shifts and relatively high signal intensity
in comparison to the free protein (Figure 6B). These cross-peaks
(L97, L180, L186, and the LeuVal-dipeptide from the vector-
encodedN-terminal affinity tag) arise from residues in the flexible
N- and C-terminal regions of the FADD-DD construct (Berglund
et al., 2000). The other cross-peaks in the spectrum exhibit
the behavior witnessed for the second set of CD95-DD+CT
cross-peaks, namely dispersion into apparently local clusters
of much weaker features (e.g., I126 d1CH3 in Figure 7C and
the 1H-upfield L119 dCH3 in Figure 7D). In a similar experiment
in which ILV-13CH3-FADD-DDwas titrated with CD95-DD essen-
tially the same outcome was obtained (data not shown).
The 13CH3-TROSY results clearly indicate that on complex
formation by FADD-DD with either CD95-DD or CD95-DD+CT,
a particle is formed that has disordered N- and C-terminal tail
regions of both proteins in concordancewith the results obtained
with the 15N-labeling. The dispersal into clusters of weak peaks
for the methyl groups residing in the globular parts of the two
DDs suggests that different FADD-DD and CD95-DD/CD95-
DD+CT protomers occupy nonequivalent sites within the com-
plex, suggesting a degree of structural asymmetry. Alternatively,
the cross-peak clusters correspond to a range of distinct com-
plexes with variant mass and subunit composition. For each
resolved cross-peak cluster, one component exhibits 1H and13C chemical shifts essentially identical to that of the free protein
All rights reserved
Figure 5. 2D 15N,1H-NMR of the Interaction
between CD95-DD and FADD-DD
Detail of the 2D 15N,1H-HSQC NMR spectra of15N-labeled CD95-DD+CT in the absence (A) and
presence (B) of excess unlabeled FADD-DD,
and 15N-labeled FADD-DD in the absence (C)
and presence (D) of excess unlabeled CD95-
DD+CT. The resonance assignments for cross-
peaks visible in the spectra of the respective
complexes are indicated (B and D). In (B) the
R328 3NH cross-peak is aliased from outside the15N spectral window; similarly for the R189 3NH
cross-peak in (D). Although the majority of CD95-
DD+CT cross-peaks are absent in (B) the back-
bone and side chain cross-peaks for the flexible
N- and C-terminal tail regions are clearly visible.
Likewise in (D) only the cross-peaks for the N-
and C-terminal tail regions of FADD-DD are
retained. The inset in (B) shows a section of a
similar spectrum of 15N-labeled CD95-DD in the
presence of excess FADD-DD. See also Figure S4.
Structure
NMR of CD95/FADD Death Domain Complex
(Figure 7), although with distinctly broader 1H linewidth. These
peaks can be attributed either to a small population of free
protein chains in exchange with the oligomeric complex, or
to a subset of sites in the complex that exhibit unperturbed
chemical shifts relative to the free protein. The relative intensity
of this component varies from peak cluster to peak cluster but
is invariant to addition of excess partner protein indicating
saturation of binding. Under equivalent sample conditions in
the 15N-based experiments, we failed to detect signals from
free protein chains. We therefore attribute these cross-peaks
most likely to protein chains in the bound state that exhibit similar
to free state chemical shifts. Moreover, because multiple cross-
peaks can be resolved the exchange of DD protomers between
the nonequivalent sites must be slow on the timescale repre-
sented by the relatively small chemical shift differences between
the components in each cluster.
Effect of C-Helix Mutations in CD95-DD+CTIn the 3EZQ crystal structure, the hetero-octamer of CD95-DD
and FADD-DD domains includes pairwise reciprocal contacts
between CD95-DD dimers comprised mainly of a cluster of
four Q311 side chains (one from each of the polypeptides) sand-
wiched between two pairs of L315 side chains (Figure 1; Fig-
ure S1). We constructed Q311A and L315A mutants of CD95-
Structure 18, 1378–1390, October 13, 2010 ª
DD+CT. Both of these variants yield15N-HSQCNMR spectra indicating reten-
tion of a stable DD fold (Figures 8A and
8B), consistent with the side chains of
these two residues being highly exposed
in the structure of isolated CD95 DD
(Huang et al., 1996). For each mutant,
we observed complex formation in the
presence of FADD-DD evidenced by the
characteristic loss of cross-peak intensity
for the dispersed signals from the glob-
ular part of CD95-DD, and retention of
the cross-peaks from the flexible terminal
regions, exactly as for unmodified CD95-DD+CT (Figures 8A
and 8B).
Effect of Low pHThe 3EZQ crystal structure was reportedly obtained under
conditions of pH �4. We exchanged the buffer of a sample of
the 15N-CD95-DD+CT:FADD-DD complex prepared at pH 6.2
to pH 4.0 and rerecorded the NMR spectrum (Figure 8D). Under
these conditions the majority of the backbone NH cross-peaks
are clearly visible with characteristics consistent with unbound15N-CD95-DD+CT. This result suggests that the complex formed
in solution is dissociated at the lower pH.
Sequence Conservation of the Extreme C-TerminalRegion among CD95 OrthologsGiven the evidence from the crystal structure of the scope for
dramatic remodeling of the C-terminal region of CD95, we con-
structed an in-depth sequence analysis (Figure 9) of putative
CD95 species orthologs to examine the degree of conservation
of the residues involved. Sequences annotated as CD95 were
derived from a variety of genomic database sources and an
alignment constructed by optimization of the sequence similarity
over the globular part (a1–a6) of the canonical DD structure.
In Figure 9, only the part of the alignment focusing on the
2010 Elsevier Ltd All rights reserved 1383
Figure 6. ILV-13CH3-TROSY NMR of CD95-
DD:FADD-DD Complexes
Overview of the Ile- (top) and Leu/Val- (bottom)
regions of the 13CH3-TROSY spectrum of (A)
ILV-13CH3-,2H-labeled CD95-DD+CT and (B)
ILV-13CH3-,2H-labeled FADD-DD in the absence
(black contours) and presence (red) of 2H-labeled
FADD-DD and 2H-labeled CD95-DD+CT, respec-
tively. The spectrum of each complex is character-
ized by massive intensity loss for the majority of
cross-peaks. The residue-specific assignments
of the signals that retain significant intensity, cor-
responding to methyl groups within the N- and
C-terminal regions of the respective polypeptide
chain, are indicated; in (B) asterisks indicate
cross-peaks arising from the vector-encoded
Leu-Val dipeptide in the affinity tag (Berglund
et al., 2000). Dashed-line boxes indicate the
regions highlighted at lower contour level in
Figure 7.
Structure
NMR of CD95/FADD Death Domain Complex
C-terminal region, starting at the completely conserved Trp
residue in the globular DD, is presented. Inspection of the align-
ment as a whole (not shown) indicates that the sequence diver-
gence among known CD95 DDs is relatively limited, consistent
1384 Structure 18, 1378–1390, October 13, 2010 ª2010 Elsevier Ltd All rights reserved
with the predicted orthology. Neverthe-
less it is apparent that sequence con-
servation is significantly higher among
residues in helices a1–a6 than in the
adjoining regions,evidenced inFigure9by
the more definitive consensus sequence
for a5–a6 than for the extreme C ter-
minus. Most notable in this respect is that several species (cat,
dog, frog, and zebrafish) lack the extreme C-terminal region,
with polypeptides terminated at the position corresponding to
S320 in human CD95.
Figure 7. Effect of Complex Formation on
Selected, Well-Resolved ILV-13CH3 Methyl
Cross-PeaksSelected regions, corresponding to boxed areas
in Figure 6, of the 13CH3-TROSY NMR spectra of
ILV-13CH3-,2H-labeled CD95-DD+CT (A and B)
and ILV-13CH3-,2H-labeled FADD-DD (C and D)
in the absence (black contours) and presence
(red) of 2H-labeled FADD-DD and 2H-labeled
CD95-DD+CT, respectively. The residue-specific
assignments of the cross-peaks in the free
CD95-DD+CT and FADD-DD spectra are indi-
cated. The base contour level used for the spectra
of the complexes is substantially lower than in
Figure 6, and close to the noise floor for the
L229 cross-peak (B) that, although well-resolved,
is relatively weak in both free- and bound-states.
Figure 8. Effect of CD95 Mutations Q311A
and L315A and Low pH on Complex Forma-
tion with FADD-DD
Two-dimensional 15N,1H-HSQC NMR spectra of15N-labeled CD95-DD+CT Q311A (A) and L315A
(B) mutants of in the absence (back) and presence
(red) of excess unlabeled FADD-DD. Comparison
of these data with Figure 5 show that neither
mutation impacts complex formation, evidenced
by loss of all signals bar those in the flexible ter-
minal regions, exactly as for unmodified CD95-
DD+CT (Figure 5B). See also Figure S1. (C) 2D15N,1H-HSQC NMR spectrum of 15N-labeled
CD95-DD+CT in the presence of excess FADD-
DD at pH 6.2 (Figure 5B) and (D) the correspond-
ing spectrum of the sample after buffer exchange
to pH 4.0. The increase in the number and inten-
sity of cross-peaks in (D), with dispersion essen-
tially identical to that of a sample of isolated
CD95-DD+CT (not shown), indicates dissociation
of the CD95-DD+CT protein from the high molec-
ular weight complex at the lower pH.
Structure
NMR of CD95/FADD Death Domain Complex
DISCUSSION
We have described the analysis of the complex species formed
between the DDs of CD95 and FADD using heteronuclear NMR
spectroscopy, coupled with SEC-MALS and MS methods.
We cast the following discussion of our observations in terms of
the formation of a high molecular weight complex Cn:Fm from its
constituent protomers ‘‘C’’ (CD95-DD or CD95-DD+CT) and ‘‘F’’
(FADD-DD); n and m are integers. The results of our attempt to
characterize the composition of the Cn:Fm object using SEC-
MALS (Figure 3) suggest a composition that is at least nine protein
chains for both FADD-DD:CD95-DD and FADD-DD:CD95-DD
+CT complexes. In combination with the SDS-PAGE and amino
acid analysis data indicating an approximately equal number
of FADD and CD95 polypeptides in the complex we conclude
that the particles detected in the SEC-MALS experiment corre-
spond to either C4:F5 or C5:F4. When the same samples are sub-
jected to native phase nanoESI-MS, the clearly dominant ion-
series observed for both FADD-DD/CD95-DD and FADD-DD/
CD95-DD+CT combinations corresponds to a C5:F5 particle
(Figure 4). It is unclear why there is not a closer match between
the molecular weights reported by SEC-MALS and ESI-MS. In
the mass spectra the appearance of single monomers under the
soft instrument conditions used to prevent collision-induced
Structure 18, 1378–1390, October 13, 2010 ª
dissociation of the complexes suggests
the potential for dynamic rearrangement
and subunit exchange. Therefore we
must conclude that under the nonequilib-
rium conditions of the SEC-MALS experi-
ment one of the subunits of the 5:5 com-
plex is dynamically dissociated to leave
5:4 species. Critically, we did not ever
detect the presence of a C4:F4 species
that would correspond to the hetero-
octamer of the crystal structure.
We adopted asymmetric isotope label-
ing and heteronuclear NMR spectros-
copy to probe the nature of the complexes formed between
FADD and CD95 DD proteins. Both 15N-labeling with standard1H,15N-HSQC spectroscopy and ILV-13CH3-labeling with 13CH3-
TROSYNMR reveal that signals from the flexible terminal regions
of the proteins retain a relatively narrow line width and unper-
turbed chemical shifts, consistent with these regions retaining
their flexible, disordered characteristics in the complex. On the
other hand, the cross-peaks from the more globular parts of
the DDs are apparently broadened beyond detection (for the15N-labeling) or are dispersed into multiple cross-peaks with
slightly perturbed chemical shifts (ILV-13CH3-labeling). The
cross-peaks from the terminal regions appear to be in formal
fast exchange between the free and complex states, such that
only a very limited contribution to the cross-peak line widths
arises from the slower overall tumbling. The identification of
the exchange regime for the other signals is less straightforward
in the case of the 15N-labeled samples where the chemical shifts
of the cross-peaks are not recovered even at the highest
concentration of the binding partner. The overall outcome of
the experiments, where the titration of one (unlabeled) protein
with another 15N-labeled protein leads to intensity reduction
rather than gradual increase in the residual signal linewidth or
perturbation of the chemical shift, is consistent with slow
exchange (Figure 5; Figure S4). The spectrum changes because
2010 Elsevier Ltd All rights reserved 1385
Figure 9. Sequence Comparison of the
C Terminus of CD95 Orthologs
Sequence alignment of selected CD95 orthologs,
focused on the C-terminal region of the chain
including (top) the canonical a5–a6 helices of the
isolated CD95 DD in solution, PDB code 1DDF
(Huang et al., 1996), and the stem-C-helix regions
in the crystal structure of the CD95-DD:FADD-DD
complex PDB code 3EZQ (Scott et al., 2009). For
1DDF red symbols indicate residues buried in
the core of the death domain. For 3EZQ symbols
indicate residues involved in the intramolecular
stem-C-helix side chain contacts (green), inter-
molecular interactions between protomers in
the major CD95-DD homodimer contact (blue),
and side chain interactions describing the CD95-
DD dimer-dimer interface within hetero-octamer
(orange) (Figure 1; Figure S1). The alignment was
manually optimized using the sequences corre-
sponding to the entire globular DD (not shown).
Dashes indicate gaps introduced to optimize the
alignment. The avian sequences extend signifi-
cantly beyond position 335 (indicated by ..). Very
highly conserved residues and strongly con-
served residues have a dark gray and a light
gray background respectively. Bottom: Con-
sensus sequence. Essentially invariable residues
are shown in bold uppercase, strongly conserved
residues in upper case; lower case letters indicate
residues that are relatively well conserved in the
extreme C-terminal region.
Structure
NMR of CD95/FADD Death Domain Complex
the signal intensity corresponding to the ‘‘free’’-form of the 15N-
labeled protein (e.g., C) is depleted by the presence of the unla-
beled partner protein F as a result of complex formation—the
separate C resonances from the Cn:Fm complex being too broad
for detection as a result of slow rotational diffusion, or exchange
broadening among multiple resonances for the complex.
Strategies for ILV-13CH3-group labeling (Goto et al., 1999) and
the exploitation of the relaxation-cancelling effects obtained in13CH3-TROSY NMR (Ollerenshaw et al., 2003; Tugarinov et al.,
2003) provide a useful means to probe the structural and
dynamic characteristics of high molecular mass protein assem-
blies (Ruschak and Kay, 2010). The overall outcome of this
approach applied to the complex formed between the DDs
of FADD and CD95 yields a picture broadly consistent with
that obtained by 15N-HSQC spectroscopy. Thus, the ILV-13CH3
cross-peaks from the disordered N- and C-terminal tail regions
of the two constituent domains in the complex retain a relatively
narrow intense signal with unperturbed chemical shifts, whereas
the cross-peaks from the ordered globular regions are massively
diminished in intensity (Figure 6). In contrast to the case for 15N-
labeling, the second group of cross-peaks is revealed at low
contour level: each signal appears to have been dispersed
into a cluster of often partially overlapping features (Figure 7).
The simplest interpretation of these results is that: (1) the N-
and C-terminal regions of the two distinct protein chains retain
a high degree of disorder in the complex; and (2) the globular
regions of the domains can occupy one of a number of microen-
vironments within the complex, each with a distinct set of influ-
ences on the methyl group chemical shifts. The latter might
also arise if the sample contains multiple complex species
each with a different overall subunit composition. However the
1386 Structure 18, 1378–1390, October 13, 2010 ª2010 Elsevier Ltd
pattern of cross-peak intensities is not consistent with the ESI-
MS and SEC-MALS analysis that indicate the predominance of
a single major species. On the other hand, there is no straightfor-
ward method to accurately characterize the compositional
dispersion of the sample at the relatively elevated NMR concen-
tration. An alternative explanation is that the FADD-DD:CD95-
DD complex possesses a low degree of symmetry and the local
environment of the equivalent NMR spins in different chains
of the complex can appear subtly different. It is interesting to
note that the reported 3.2 A resolution crystal structure of the
hetero-dodecameric �157 kD PIDDosome core complex com-
prising the DDs from the proteins PIDD and RAIDD, that pos-
sesses a stoichiometry of 7:5 RAIDD-DD:PIDD-DD, is a particle
with intrinsic low symmetry (Park et al., 2007b). Thus, although
the individual PIDDosome DDs are in quasi-equivalent environ-
ments, there are eight unique asymmetric DD-DD interfaces
and none of the individual PIDD-DDs (or RAIDD-DDs) resides
in an identical arrangement of contacts to any other protomer.
Moreover, the recently reported structure of the 14-chain
Myddosome assembly displays a similar nonexact equivalence
of the contacts between component DDs (Lin et al., 2010).
It seems possible that the NMR data for the FADD-DD:CD95-
DD complex could be reporting a similar lack of equivalence
for the constituent domains. Preliminary 13CH3-TROSY data
obtained in our laboratory for the PIDDosome core complex
exhibit highly similar characteristics to those observed for
FADD-DD:CD95-DD+CT (data not shown). We are not aware
that such a degree of complexity has been reported in 13CH3-
TROSY spectra before. Increased multiplicity (doubling) of Ile-
d1CH3 signals in slow exchange has been observed in the
ligand-induced T / R allosteric equilibrium for the regulatory
All rights reserved
Structure
NMR of CD95/FADD Death Domain Complex
chains of the heterododecameric aspartate transcarbamoylase
(ATCase) (Velyvis et al., 2007) and, recently, the observation of
nonequivalent methionine methyl group signals within an other-
wise symmetric homoheptameric system was reported (Religa
et al., 2010). However, in neither case was the outcome indica-
tive of a loss of overall symmetry of the major globular part of the
protein structure. Indeed in the ATCase case, the spectra imply
that symmetry of the overall structure is maintained in the T- > R
transition (Velyvis et al., 2007), and in the latter case the ‘‘split’’
signals derive from a flexible appendage that can exchange
position either side of an ‘annulus’ of limited dimension (Religa
et al., 2010). A complete accounting for the complexity of13C-ILV-methyl TROSY spectra for the FADD-DD:CD95-DD+CT
complex warrants further investigation. Nevertheless it is safe
to conclude that the spectra are not consistent with a monodis-
perse particle with high symmetry.
Relationship to the Crystal Structureof CD95-DD:FADD-DDIt is important to discuss our observations in the context of
the recently reported 2.7 A resolution crystal structure of a
FADD:CD95 DD-DD complex, Protein Data Bank code 3EZQ
(Scott et al., 2009). Apparently critical to the formation of the
contacts between the opened CD95-DD subunits within the
hetero-octameric assembly is the formation of an intrachain
a-helical hairpin loop that brings together hydrophobic side
chains from the stem-helix (L306, I310, I313, I314) with residues
from a new C-helix (F327, I331, L334) derived from the extreme
CT region of CD95 (Figure 1; Figure S1); this is a region that is
clearly disordered in the isolated CD95 DD (Huang et al.,
1996). Along with H-bond contacts involving the terminal side
chain amino group of K299, several of these residues combine
with others (notably L303, A307, I318) to make reciprocal pair-
wise interactions between the CD95-DD stem/C-helix hairpins,
forming an antiparallel four-helix bundle bridge as the CD95-
DD dimer interface (Figure S1). We note that the crystal
comprises a FADD-DD construct with a 23-residue C-terminal
tail region (including a His6-affinity purification tag) that is not
resolved in the electron density and a CD95-DD construct
appended by a nonnative C-terminal Leu336-Glu337 dipeptide
that makes multiple interchain contacts with the FADD-DD
partner and the neighboring CD95-DD (Figure S1). The predicted
molecular mass of the overall eight-chain assembly in the crystal
is 108 kD; the ordered region lacking the FADD tail peptide would
have a mass �98 kD.
The combination of our previous results (Ferguson et al.,
2007) together with data presented here unambiguously demon-
strate that complex formation occurs in vitro for combinations
of CD95 and FADD DD constructs both in the presence and
absence of the twelve C-terminal CD95 residues. Critically our
NMR data demonstrate that residues S322–V335 retain a high
degree of flexibility in the complex formed by CD95-DD+CT
(Figure 5B and Figure 6A), indicating that this region does not
participate in the formation of the stem/C-helix hairpin predicted
by the crystal structure. Moreover, we find an overall mass for the
complex that is substantially higher than would be consistent
with the hetero-octameric assembly in the crystal (Figure 3 and
Figure 4). Also we observed that high molecular weight complex
formation persists in solution despite Ala-substitution of residues
Structure 18, 1378–1
Q311 and L315 (Figures 8A and 8B), side chains that appear to
be important for crystal contacts between CD95-DD homo-
dimers (Figure 1; Figure S1). The result for the L315A mutant is
consistent with a previous report that this mutation does not
impact CD95 association with FADD (Huang et al., 1996).
It is difficult to rationalize the lack of concordance between
the crystal structure and the solution NMR data with respect to
the role of the C-terminal tail region of CD95. We note that the
crystallization buffer pH (reported as pH 4) is significantly
different from physiological conditions. Curiously, in our hands,
we find that lowering the pH of an NMR sample of the FADD-
DD:CD95-DD+CT complex leads to dissociation (Figures 8C
and 8D). Moreover, the crystal structure indicates that the
CD95 DD has the potential to dimerize via the conformational
transition from a closed to an open form. We have attempted
to find evidence that the CD95-DD+CT protein has a low popu-
lation of a significantly different conformation to the closed form
in solution by recording 15N relaxation dispersion experiments
(Palmer, 2004) at both pH 7 and pH 4, but we found no evidence
for chemical exchange (data not shown). In addition, it seems
unlikely that the structure of CD95-DD+CT is opened in our
NMR experiments on the complex, because we do not detect
ordering of the CT-region corresponding to the C-helix, and
the chemical shift changes of the ILV-13CH3-group cross-peaks
appear rather small (Figure 7).
If the remarkable remodeling of the CD95-DD structure in
the 3EZQ crystal structure properly describes the platform for
DISC formation then it would be reasonable to expect that the
structural features of the open conformation be conserved
across closely homologous receptors. The sequence align-
ment of putative CD95 orthologs (Figure 9) suggests that the
C-terminal tail region of CD95 is not conserved to the extent
that one could confidently predict that the formation of the
stem-C-helix hairpin is a common feature of CD95 signaling.
Some species appear to lack the C-terminal 15 residues that
would constitute the C-helix, and others have insertions, dele-
tions or substitutions compared to human CD95 that are not
straightforwardly incorporated in a model of intra- and intermo-
lecular interactions indicated by the crystal structure (Figure 1;
Figure S1).
In addition to the sequence-structure correlates relating to
interactions between CD95 chains in the crystal structure, there
are additional aspects relevant to our own observations con-
cerning the interaction betweenCD95-DD and FADD-DD, partic-
ularly relating to the effects of mutations in the FADD-DD chain.
For example, in line with a similar report (Hill et al., 2004), we
found that mutation of FADD-DD R117E abrogates binding to
wild-type GST-CD95-DD in a pull-down assay (Figure 2). More-
over when 15N-labeled FADD-DD(R117E) (Figure S2) was titrated
with wild-type Gb1-tagged CD95-DD (Ferguson et al., 2007) the
interaction was strongly attenuated compared to wild-type
FADD-DD (data not shown). Other researchers have also found
that mutation of R117 in both human and mouse FADD has
significant impact on the CD95 interaction (Bang et al., 2000;
Hill et al., 2004; Hughes et al., 2009; Imtiyaz et al., 2005; Jeong
et al., 1999). In the 3EZQ structure the R117 side chain is
directed away from helix a2 on the surface of FADD-DD that is
almost diametrically opposed to the intermolecular interface
with the open CD95-DD (Figure 1). It was also reported that
390, October 13, 2010 ª2010 Elsevier Ltd All rights reserved 1387
Structure
NMR of CD95/FADD Death Domain Complex
mutation of FADD-DD residues R113E, D123R, K125A, and
R146A significantly impacted the ability of FADD to bind CD95-
DD in a GST-pull-down assay (Hill et al., 2004), although none
of these FADD-DD residues is close to the FADD-DD:CD95-DD
interface in the crystal structure (Figure 1). We have previously
reported that CD95(D260A), corresponding to a site of mutation
in type Ia ALPS (Martin et al., 1999; Vaishnaw et al., 1999), fails to
interact with FADD-DD as assessed by NMR (Ferguson et al.,
2007) consistent with earlier reports for the behavior of this
mutant (Huang et al., 1996). We obtained a similar result with
CD95-DD(A257D) that represents another well-characterized
ALPS mutant that fails to bind FADD-DD (Martin et al., 1999)
(data not shown). In the 3EZQ structure A257 and D260 are
directed away from the surface of the CD95-DD a3 helix and
a2-a3 linker respectively, well outside the FADD-DD contact
zone.
Concluding RemarksCollectively, the combination of our SEC-MALS, ESI-MS, and
NMR investigations of the complex formed between FADD-DD
and CD95-DD proteins coupled with the evidence from muta-
genesis studies, particularly for the impact of mutations in
FADD-DD, lack consistency with the picture of the interaction
between these two proteins that has been provided so far
by X-ray crystallography. Critically, our data show that the
C-terminal tail region of CD95 is dispensable for complex forma-
tion between the two DDs in solution under near physiological
conditions. The interpretation of the SEC-MALS data in terms
of a cooperatively formed 5:4 complex is compatible with the13CH3-TROSY evidence for slow exchange and nonequivalent
chemical shifts. The indication from the mass spectrometry
data of a possible 5:5 composition is reminiscent of the PIDDo-
some core complex that has, in part, the juxtaposition of a ring
of five PIDD-DDs against a ring of five RAIDD-DDs where each
of the chains occupies a unique site (Park et al., 2007b). Such
asymmetry for the FADD-DD:CD95-DD complex would be con-
sistent with the appearance of cross-peak clusters in the 13CH3-
TROSY spectra. From the biological perspective, it is trans-
parent that more work is required to properly understand at
a structural level the earliest events in the mechanism of DISC
formation and to resolve the lack of concordance engendered
by the current analysis with respect to the available high resolu-
tion data.
EXPERIMENTAL PROCEDURES
Plasmid Construction and Mutagenesis
We use the human CD95 numbering 1-335 corresponding to the intact
polypeptide including the membrane targeting signal sequence, consistent
with that used for the crystal structure 3EZQ (Scott et al., 2009). DNA encoding
CD95 residues 218–323 (CD95-DD) and 218–335 (CD95-DD+CT) was sepa-
rately inserted into a modified pET-32b vector that encodes an N-terminal
His6-affinity tag residues followed by a tobacco etch virus (TEV) protease-
cleavable streptococcal protein G immunoglobulin-binding B1 (Gb1) domain
(Zhou et al., 2001). After TEV-proteolysis the purified CD95 polypeptides
contain the nonnative N-terminal dipeptide Gly-Ser. N-terminal glutathione
S-transferase (GST)-fused CD95-DD and CD95-intracellular domain (IC;
residues 194–335), were obtained by subcloning the corresponding CD95
PCR products into a pGEX-2T bacterial expression vector. Point mutant
CD95 and FADD constructs were created using QuikChange site-directed
mutagenesis.
1388 Structure 18, 1378–1390, October 13, 2010 ª2010 Elsevier Ltd
Protein Expression and Purification
Expression and purification of FADD-DD (residues 93–192) was as described
(Berglund et al., 2000). CD95-DD and CD95-DD+CT proteins were obtained
by expression in E. coli Rosetta (DE3) cells at 22�C. The protein in the super-
natant was recovered by Ni-IMAC, TEV cleavage, and semipreparative SEC
with final elution into 20 mM MES pH 6.2, 100 mM NaCl, 1 mM EDTA, 3 mM
NaN3, and 0.5 mM TCEP (Buffer A).
Uniform 15N-isotope, 13C- and 2H-labeling were achieved using stan-
dard methods in M9 media. Uniform 2H-, Ile-d1[13CH3], Leu-d, Val-g
[13CH3,12CD3]-labeling of CD95 and FADD proteins for ILV 13CH3-TROSY
NMR was achieved in 99% D2O M9 medium containing [2H,12C]-glucose;
1 hr before induction, the medium was supplemented with 60 mg/l [meth-
yl-13CH3]-2H a-ketobutyric acid and 100 mg/l [dimethyl-13CH3,
12CD3]-2H
a-ketoisovaleric acid (Goto et al., 1999). Proteins were exchanged into Buffer
A in D2O before the NMR measurement.
GST Pull-Down
GST-tagged CD95-DD and CD95-IC constructs were expressed in Rosetta
(DE3) cells at 37�C. The cell lysate supernatant was incubated with gluta-
thione-Sepharose and the suspension washed twice with 50 mM HEPES,
pH 7.4, 50mMNaCl, 5mMEDTA, 10%glycerol, 0.1%Nonidet P-40. The slurry
was then mixed with FADD-DD and binding allowed to proceed for 2 hr at 4�C.After washing twicewith binding buffer, 10 ml resin was loaded on a SDS-PAGE
gel, which was run and stained with Coomassie Blue.
SEC-MALS
Analytical SEC-coupled MALS profiles were recorded at 16 angles using
aDAWN-HELEOS laser photometer (Wyatt Technology) and differential refrac-
tometer (Optilab rEX) equipped with a Peltier temperature-regulated flow
cell maintained at 25�C (Wyatt Technology). The protein concentration c of
the eluent was determined from the refractive index RI(n) change (dn/dc =
0.186); wavelength 658 nm. Samples (100 ml, flow rate 0.5 ml/min) were
applied to a Superdex 200 10/300 GL column (GE Healthcare) equilibrated
in Buffer A. Before injection on the column CD95:FADD complexes were iso-
lated by semipreparative S200 gel filtration and concentrated to 20 mg/ml in
Buffer A. The data were analyzed using ASTRA 5.1 as described (Ivins et al.,
2009).
Mass Spectrometry
NanoESI-MS experiments were carried out in the positive ionmode on aQStar
mass spectrometer (MDS Analytical Technologies) modified for optimized
performance at high masses (Chernushevich and Thomson, 2004; Sobott
et al., 2002) using nanoflow capillaries. The following experimental parameters
were used: ion spray voltage, 1.4 kV; declustering potential, 100 V; focusing
potential, 150 V; and collision energy, 20–40 V, MCP 2350 V. Data analysis
was done with Massign, a software package developed in house on LabVIEW
basis (Hernandez et al., 2009; Natan et al., 2009). Measurements were taken
for intact protein complexes prepared as for SEC-MALS analysis and
exchanged into 50mMammonium acetate using Biospin 6microspin columns
(Bio-Rad Laboratories). The results shown are representative of multiple inde-
pendent experiments.
NMR Spectroscopy
All spectra were acquired at 25�C on Varian INOVA or Bruker AVANCE spec-
trometers operating at 14.1 T, 16.5 T, and 18.8 T. Data were processed with
NMRPipe/NMRDraw (Delaglio et al., 1995) and analyzed with CCPN Analysis
(Vranken et al., 2005). Resonance assignments for CD95 protein constructs
were obtained using standard procedures. Titrations of a labeled protein
(0.5 mM) with its unlabeled partner (0–2.5 molar equivalents) were conducted
at constant concentration of the labeled component (McAlister et al., 1996).
Typical acquisition times for 2D WATERGATE-flipback 15N,1H-HSQC (Grze-
siek and Bax, 1993) data sets were �1 hr. Overall acquisition times for 13CH3-
TROSY (Tugarinov et al., 2003) experiments (18.8 T) were�5 hr, with relaxation
delay 1.3 s. To examine the effect of low pH on the 15N-CD95-DD+DT:FADD-
DD complex, the sample was exchanged into 20 mM sodium citrate, 1.0 mM
TCEP, pH 4.0.
All rights reserved
Structure
NMR of CD95/FADD Death Domain Complex
SUPPLEMENTAL INFORMATION
Supplemental Information includes four figures and can be found with this
article online at doi:10.1016/j.str.2010.08.006.
ACKNOWLEDGMENTS
The work in the P.C.D. group is supported by the MRC (file reference
U117574559). We are grateful to the staff of the MRC Biomedical NMR Centre
and to Acely Garza-Garcia for helpful discussions. N.M. and C.V.R. are sup-
ported by the European Union within the 7th Framework Program PROS-
PECTS.
Received: May 21, 2010
Revised: July 20, 2010
Accepted: August 13, 2010
Published: October 12, 2010
REFERENCES
Algeciras-Schimnich, A., Shen, L., Barnhart, B.C., Murmann, A.E., Burkhardt,
J.K., and Peter, M.E. (2002). Molecular ordering of the initial signaling events of
CD95. Mol. Cell. Biol. 22, 207–220.
Ashkenazi, A. (2008). Directing cancer cells to self-destruct with pro-apoptotic
receptor agonists. Nat. Rev. Drug Discov. 7, 1001–1012.
Ashkenazi, A., and Dixit, V.M. (1998). Death receptors: Signaling and modula-
tion. Science 281, 1305–1308.
Bang, S., Jeong, E.J., Kim, I.K., Jung, Y.K., and Kim, K.S. (2000). Fas- and
tumor necrosis factor-mediated apoptosis uses the same binding surface of
FADD to trigger signal transduction. A typical model for convergent signal
transduction. J. Biol. Chem. 275, 36217–36222.
Beneteau, M., Daburon, S., Moreau, J.F., Taupin, J.L., and Legembre, P.
(2007). Dominant-negative Fas mutation is reversed by down-expression of
c-FLIP. Cancer Res. 67, 108–115.
Berglund, H., Olerenshaw, D., Sankar, A., Federwisch, M., McDonald, N.Q.,
and Driscoll, P.C. (2000). The three-dimensional solution structure and
dynamic properties of the human FADD death domain. J. Mol. Biol. 302,
171–188.
Bidere, N., Su, H.C., and Lenardo, M.J. (2006). Genetic disorders of pro-
grammed cell death in the immune system. Annu. Rev. Immunol. 24, 321–352.
Bodmer, J.L., Holler, N., Reynard, S., Vinciguerra, P., Schneider, P., Juo, P.,
Blenis, J., and Tschopp, J. (2000). TRAIL receptor-2 signals apoptosis through
FADD and caspase-8. Nat. Cell Biol. 2, 241–243.
Boldin, M.P., Varfolomeev, E.E., Pancer, Z., Mett, I.L., Camonis, J.H., and
Wallach, D. (1995). A novel protein that interacts with the death domain of
Fas/APO1 contains a sequence motif related to the death domain. J. Biol.
Chem. 270, 7795–7798.
Chaigne-Delalande, B., Mahfouf, W., Daburon, S., Moreau, J.F., and Legem-
bre, P. (2009). CD95 engagement mediates actin-independent and -depen-
dent apoptotic signals. Cell Death Differ. 16, 1654–1664.
Chernushevich, I.V., and Thomson, B.A. (2004). Collisional cooling of large ions
in electrospray mass spectrometry. Anal. Chem. 76, 1754–1760.
Chinnaiyan, A.M., Orourke, K., Tewari, M., and Dixit, V.M. (1995). FADD,
a novel death domain-containing protein, interacts with the death domain of
Fas and initiates apoptosis. Cell 81, 505–512.
Delaglio, F., Grzesiek, S., Vuister, G.W., Zhu, G., Pfeifer, J., and Bax, A. (1995).
NMRPIPE—a multidimensional spectral processing system based on Unix
pipes. J. Biomol. NMR 6, 277–293.
Feig, C., Tchikov, V., Schutze, S., and Peter, M.E. (2007). Palmitoylation of
CD95 facilitates formation of SDS-stable receptor aggregates that initiate
apoptosis signaling. EMBO J. 26, 221–231.
Ferguson, B.J., Esposito, D., Jovanovic, J., Sankar, A., Driscoll, P.C., and
Mehmet, H. (2007). Biophysical and cell-based evidence for differential inter-
actions between the death domains of CD95/Fas and FADD. Cell Death Differ.
14, 1717–1719.
Structure 18, 1378–1
Goto, N.K., Gardner, K.H., Mueller, G.A., Willis, R.C., and Kay, L.E. (1999).
A robust and cost-effective method for the production of Val, Leu, Ile (delta 1)
methyl-protonated N-15-, C-13-, H-2-labeled proteins. J. Biomol. NMR 13,
369–374.
Grzesiek, S., and Bax, A. (1993). The importance of not saturating H2O in
protein NMR. Application to sensitivity enhancement and NOEmeasurements.
J. Am. Chem. Soc. 115, 12593–12594.
Hernandez, H., and Robinson, C.V. (2007). Determining the stoichiometry and
interactions of macromolecular assemblies from mass spectrometry. Nat.
Protoc. 2, 715–726.
Hernandez, H., Makarova, O.V., Makarov, E.M., Morgner, N., Muto, Y., Krum-
mel, D.P., and Robinson, C.V. (2009). Isoforms of U1-70k control subunit
dynamics in the human spliceosomal U1 snRNP. PLoS ONE 4, e7202.
Hill, J.M., Morisawa, G., Kim, T., Huang, T., Wei, Y., Wei, Y.F., and Werner,
M.H. (2004). Identification of an expanded binding surface on the FADD death
domain responsible for interaction with CD95/Fas. J. Biol. Chem. 279, 1474–
1481.
Holler, N., Tardivel, A., Kovacsovics-Bankowski, M., Hertig, S., Gaide, O.,
Martinon, F., Tinel, A., Deperthes, D., Calderara, S., Schulthess, T., et al.
(2003). Two adjacent trimeric Fas ligands are required for Fas signaling and
formation of a death-inducing signaling complex. Mol. Cell. Biol. 23, 1428–
1440.
Huang, B., Eberstadt, M., Olejniczak, E.T., Meadows, R.P., and Fesik, S.W.
(1996). NMR structure and mutagenesis of the Fas (APO-1/CD95) death
domain. Nature 384, 638–641.
Huang, D.C.S., Hahne, M., Schroeter, M., Frei, K., Fontana, A., Villunger, A.,
Newton, K., Tschopp, J., and Strasser, A. (1999). Activation of Fas by FasL
induces apoptosis by a mechanism that cannot be blocked by Bcl-2 or
Bcl-x(L). Proc. Natl. Acad. Sci. USA 96, 14871–14876.
Hughes, M.A., Harper, N., Butterworth, M., Cain, K., Cohen, G.M., and Mac-
Farlane, M. (2009). Reconstitution of the death-inducing signaling complex
reveals a substrate switch that determines CD95-mediated death or survival.
Mol. Cell 35, 265–279.
Huth, J.R., Bewley, C.A., Jackson, B.M., Hinnebusch, A.G., Clore, G.M., and
Gronenborn, A.M. (1997). Design of an expression system for detecting folded
protein domains and mapping macromolecular interactions by NMR. Protein
Sci. 6, 2359–2364.
Imtiyaz, H.Z., Zhang, Y.H., and Zhang, J.K. (2005). Structural requirements for
signal-induced target binding of FADD determined by functional reconstitution
of FADD deficiency. J. Biol. Chem. 280, 31360–31367.
Ivins, F.J., Montgomery, M.G., Smith, S.J.M., Morris-Davies, A.C., Taylor, I.A.,
and Rittinger, K. (2009). NEMO oligomerization and its ubiquitin-binding prop-
erties. Biochem. J. 421, 243–251.
Jeong, E.J., Bang, S., Lee, T.H., Park, Y.I., Sim,W.S., and Kim, K.S. (1999). The
solution structure of FADD death domain - Structural basis of death domain
interactions of Fas and FADD. J. Biol. Chem. 274, 16337–16342.
Kay, L.E., Torchia, D.A., and Bax, A. (1989). Backbone dynamics of proteins as
studied by N-15 inverse detected heteronuclear NMR-spectroscopy—appli-
cation to staphylococcal nuclease. Biochemistry 28, 8972–8979.
Kischkel, F.C., Hellbardt, S., Behrmann, I., Germer, M., Pawlita, M., Krammer,
P.H., and Peter, M.E. (1995). Cytotoxicity-dependent APO-1 (Fas/CD95)-
associated proteins form a death-inducing signaling complex (Dros. Inf.
Serv.C) with the receptor. EMBO J. 14, 5579–5588.
Kischkel, F.C., Lawrence, D.A., Chuntharapai, A., Schow, P., Kim, K.J., and
Ashkenazi, A. (2000). Apo2L/TRAIL-dependent recruitment of endogenous
FADD and caspase-8 to death receptors 4 and 5. Immunity 12, 611–620.
Krammer, P.H. (2000). CD95’s deadly mission in the immune system. Nature
407, 789–795.
Krueger, A., Fas, S.C., Baumann, S., and Krammer, P.H. (2003). The role of
CD95 in the regulation of peripheral T-cell apoptosis. Immunol. Rev. 193,
58–69.
Lee, K.H., Feig, C., Tchikov, V., Schickel, R., Hallas, C., Schutze, S., Peter,
M.E., and Chan, A.C. (2006). The role of receptor internalization in CD95
signaling. EMBO J. 25, 1009–1023.
390, October 13, 2010 ª2010 Elsevier Ltd All rights reserved 1389
Structure
NMR of CD95/FADD Death Domain Complex
Legembre, P., Beneteau, M., Daburon, S., Moreau, J.F., and Taupin, J.L.
(2003). Cutting edge: SDS-stable Fas microaggregates: an early event of
Fas activation occurring with agonistic anti-Fas antibody but not with Fas
ligand. J. Immunol. 171, 5659–5662.
Lin, S.C., Lo, Y.C., and Wu, H. (2010). Helical assembly in the MyD88-IRAK4-
IRAK2 complex in TLR/IL-1R signalling. Nature 465, 885–890.
Martin, D.A., Zheng, L., Siegel, R.M., Huang, B., Fisher, G.H., Wang, J., Jack-
son, C.E., Puck, J.M., Dale, J., Straus, S.E., et al. (1999). Defective CD95/APO-
1/Fas signal complex formation in the human autoimmune lymphoproliferative
syndrome, type Ia. Proc. Natl. Acad. Sci. USA 96, 4552–4557.
McAlister, M.S., Mott, H.R., van der Merwe, P.A., Campbell, I.D., Davis, S.J.,
and Driscoll, P.C. (1996). NMR analysis of interacting soluble forms of the
cell-cell recognition molecules CD2 and CD48. Biochemistry 35, 5982–5991.
Medema, J.P., Scaffidi, C., Kischkel, F.C., Shevchenko, A., Mann, M., Kram-
mer, P.H., and Peter, M.E. (1997). FLICE is activated by association with the
CD95 death-inducing signaling complex (Dros. Inf. Serv.C). EMBO J. 16,
2794–2804.
Muppidi, J.R., Lobito, A.A., Ramaswamy,M., Yang, J.K.,Wang, L.,Wu, H., and
Siegel, R.M. (2006). Homotypic FADD interactions through a conserved
RXDLL motif are required for death receptor-induced apoptosis. Cell Death
Differ. 13, 1641–1650.
Nagata, S. (1997). Apoptosis by death factor. Cell 88, 355–365.
Natan, E., Hirschberg, D., Morgner, N., Robinson, C.V., and Fersht, A.R.
(2009). Ultraslow oligomerization equilibria of p53 and its implications. Proc.
Natl. Acad. Sci. USA 106, 14327–14332.
Newton, K., Harris, A.W., Bath, M.L., Smith, K.G., and Strasser, A. (1998).
A dominant interfering mutant of FADD/MORT1 enhances deletion of autor-
eactive thymocytes and inhibits proliferation of mature T lymphocytes.
EMBO J. 17, 706–718.
Newton, K., Kurts, C., Harris, A.W., and Strasser, A. (2001). Effects of a domi-
nant interfering mutant of FADD on signal transduction in activated T cells.
Curr. Biol. 11, 273–276.
Ollerenshaw, J.E., Tugarinov, V., and Kay, L.E. (2003). Methyl TROSY: expla-
nation and experimental verification. Magn. Reson. Chem. 41, 843–852.
O’Reilly, L.A., Tai, L., Lee, L., Kruse, E.A., Grabow, S., Fairlie, W.D., Haynes,
N.M., Tarlinton, D.M., Zhang, J.G., Belz, G.T., et al. (2009). Membrane-bound
Fas ligand only is essential for Fas-induced apoptosis. Nature 461, 659–663.
Palmer, A.G. (2004). NMR characterization of the dynamics of biomacromole-
cules. Chem. Rev. 104, 3623–3640.
Papoff, G., Hausler, P., Eramo, A., Pagano, M.G., Di Leve, G., Signore, A., and
Ruberti, G. (1999). Identification and characterization of a ligand-independent
oligomerization domain in the extracellular region of the CD95 death receptor.
J. Biol. Chem. 274, 38241–38250.
Park, H.H., Lo, Y.C., Lin, S.C., Wang, L., Yang, J.K., and Wu, H. (2007a). The
death domain superfamily in intracellular signaling of apoptosis and inflamma-
tion. Annu. Rev. Immunol. 25, 561–586.
Park, H.H., Logette, E., Raunser, S., Cuenin, S., Walz, T., Tschopp, J., andWu,
H. (2007b). Death domain assembly mechanism revealed by crystal structure
of the oligomeric PIDDosome core complex. Cell 128, 533–546.
Peter, M.E., Budd, R.C., Desbarats, J., Hedrick, S.M., Hueber, A.O., Newell,
M.K., Owen, L.B., Tschopp, J., Wajant, H., Wallach, D., et al. (2007).
The CD95 receptor: apoptosis revisited. Cell 129, 447–450.
Religa, T.L., Sprangers, R., and Kay, L.E. (2010). Dynamic regulation of
archaeal proteasome gate opening as studied by TROSY NMR. Science
328, 98–102.
Roy, A., Hong, J.H., Lee, J.H., Lee, Y.T., Lee, B.J., and Kim, K.S. (2008). In vitro
activation of procaspase-8 by forming the cytoplasmic component of the
death-inducing signaling complex (DISC). Mol. Cells 26, 165–170.
1390 Structure 18, 1378–1390, October 13, 2010 ª2010 Elsevier Ltd
Ruschak, A.M., and Kay, L.E. (2010). Methyl groups as probes of supra-molec-
ular structure, dynamics and function. J. Biomol. NMR 46, 75–87.
Sandu, C., Morisawa, G., Wegorzewska, I., Huang, T., Arechiga, A.F., Hill,
J.M., Kim, T., Walsh, C.M., and Werner, M.H. (2006). FADD self-association
is required for stable interaction with an activated death receptor. Cell Death
Differ. 13, 2052–2061.
Schneider, P., Holler, N., Bodmer, J.L., Hahne, M., Frei, K., Fontana, A., and
Tschopp, J. (1998). Conversion of membrane-bound Fas(CD95) ligand to its
soluble form is associated with downregulation of its proapoptotic activity
and loss of liver toxicity. J. Exp. Med. 187, 1205–1213.
Scott, F.L., Stec, B., Pop, C., Dobaczewska, M.K., Lee, J.J., Monosov, E.,
Robinson, H., Salvesen, G.S., Schwarzenbacher, R., and Riedl, S.J. (2009).
The Fas-FADD death domain complex structure unravels signalling by
receptor clustering. Nature 457, 1019–1022.
Siegel, R.M., Frederiksen, J.K., Zacharias, D.A., Chan, F.K.M., Johnson, M.,
Lynch, D., Tsien, R.Y., and Lenardo, M.J. (2000). Fas preassociation required
for apoptosis signaling and dominant inhibition by pathogenic mutations.
Science 288, 2354–2357.
Siegel, R.M., Muppidi, J.R., Sarker, M., Lobito, A., Jen, M., Martin, D., Straus,
S.E., and Lenardo, M.J. (2004). SPOTS: signaling protein oligomeric transduc-
tion structures are early mediators of death receptor-induced apoptosis at the
plasma membrane. J. Cell Biol. 167, 735–744.
Sobott, F., Hernandez, H., McCammon, M.G., Tito, M.A., and Robinson, C.V.
(2002). A tandem mass spectrometer for improved transmission and analysis
of large macromolecular assemblies. Anal. Chem. 74, 1402–1407.
Sprick, M.R., Rieser, E., Stahl, H., Grosse-Wilde, A., Weigand,M.A., andWalc-
zak, H. (2002). Caspase-10 is recruited to and activated at the native TRAIL
and CD95 death-inducing signalling complexes in a FADD-dependent manner
but can not functionally substitute caspase-8. EMBO J. 21, 4520–4530.
Strasser, A., Jost, P.J., and Nagata, S. (2009). The many roles of FAS receptor
signaling in the immune system. Immunity 30, 180–192.
Thorburn, A. (2004). Death receptor-induced cell killing. Cell. Signal. 16,
139–144.
Tugarinov, V., Hwang, P.M., Ollerenshaw, J.E., and Kay, L.E. (2003). Cross-
correlated relaxation enhanced H-1-C-13 NMR spectroscopy of methyl
groups in very high molecular weight proteins and protein complexes.
J. Am. Chem. Soc. 125, 10420–10428.
Vaishnaw, A.K., Orlinick, J.R., Chu, J.L., Krammer, P.H., Chao, M.V., and
Elkon, K.B. (1999). The molecular basis for apoptotic defects in patients with
CD95 (Fas/Apo-1) mutations. J. Clin. Invest. 103, 355–363.
Velyvis, A., Yang, Y.R., Schachman, H.K., and Kay, L.E. (2007). A solution NMR
study showing that active site ligands and nucleotides directly perturb the allo-
steric equilibrium in aspartate transcarbamoylase. Proc. Natl. Acad. Sci. USA
104, 8815–8820.
Vranken, W.F., Boucher, W., Stevens, T.J., Fogh, R.H., Pajon, A., Llinas, P.,
Ulrich, E.L., Markley, J.L., Ionides, J., and Laue, E.D. (2005). The CCPN data
model for NMR spectroscopy: development of a software pipeline. Proteins
59, 687–696.
Wilson, N.S., Dixit, V., and Ashkenazi, A. (2009). Death receptor signal trans-
ducers: nodes of coordination in immune signaling networks. Nat. Immunol.
10, 348–355.
Yang, J.K., Wang, L.W., Zheng, L.X., Wan, F.Y., Ahmed, M., Lenardo, M.J.,
and Wu, H. (2005). Crystal structure of MC159 reveals molecular mechanism
of DISC assembly and FLIP inhibition. Mol. Cell 20, 939–949.
Zhou, P., Lugovskoy, A.A., and Wagner, G. (2001). A solubility-enhancement
tag (SET) for NMR studies of poorly behaving proteins. J. Biomol. NMR 20,
11–14.
All rights reserved