apo2l/trail and the death receptor 5 agonist antibody amg 655 cooperate to promote receptor...
TRANSCRIPT
Cancer Cell
Article
Apo2L/TRAIL and the Death Receptor 5 AgonistAntibody AMG 655 Cooperate to Promote ReceptorClustering and Antitumor ActivityJonathan D. Graves,1,11 Jennifer J. Kordich,1,12 Tzu-Hsuan Huang,2,14 Julia Piasecki,3 Tammy L. Bush,2 Timothy Sullivan,2
Ian N. Foltz,4 Wesley Chang,5 Heather Douangpanya,1,13 Thu Dang,2 JasonW. O’Neill,6 Rommel Mallari,7 Xiaoning Zhao,7
Daniel G. Branstetter,8 John M. Rossi,9 Alexander M. Long,10 Xin Huang,10 and Pamela M. Holland2,*1Department of Oncology Research, Amgen Inc., Seattle, WA 98119, USA2Therapeutic Innovation Unit, Amgen Inc., Cambridge, MA 02142, USA3Therapeutic Innovation Unit, Amgen Inc., Seattle, WA 98119, USA4Department of Biologic Discovery, Amgen British Columbia, Burnaby, BC V5A 1V7, Canada5Department of Clinical Immunology, Amgen Inc., South San Francisco, CA 94080, USA6Department of Biologic Optimization, Amgen Inc., Seattle, WA 98119, USA7Department of Molecular Structure and Characterization, Amgen, South San Francisco, CA, 94080, USA8Department of Pathology, Amgen Inc., Seattle, WA 98119, USA9Department of Molecular Sciences and Computational Biology, Amgen Inc., Thousand Oaks, CA 91320, USA10Department of Molecular Structure and Characterization, Amgen Inc., Cambridge, MA 02142, USA11Present address: Belmont, MA 02478, USA12Present address: Van Andel Institute, Grand Rapids, MI 49503, USA13Present address: Seattle Genetics, Bothell, WA 98021, USA14Present address: AbbVie, North Chicago, IL 60064, USA
*Correspondence: [email protected]
http://dx.doi.org/10.1016/j.ccr.2014.04.028
SUMMARY
Death receptor agonist therapies have exhibited limited clinical benefit to date. Investigations into whyApo2L/TRAIL and AMG 655 preclinical data were not predictive of clinical response revealed that coadmin-istration of Apo2L/TRAIL with AMG 655 leads to increased antitumor activity in vitro and in vivo. The combi-nation of Apo2L/TRAIL and AMG 655 results in enhanced signaling and can sensitize Apo2L/TRAIL-resistantcells. Structure determination of the Apo2L/TRAIL-DR5-AMG 655 ternary complex illustrates how higherorder clustering of DR5 is achieved when both agents are combined. Enhanced agonism generated bycombining Apo2L/TRAIL and AMG 655 provides insight into the limited efficacy observed in previous clinicaltrials and suggests testable hypotheses to reconsider death receptor agonism as a therapeutic strategy.
INTRODUCTION
Evasion of apoptosis by tumor cells is key to the pathogenesis
and progression of cancer, and the activation of death receptors
has been considered an attractive therapeutic strategy to pro-
mote apoptosis of tumor cells (Ashkenazi et al., 2008; Wiezorek
et al., 2010). Apo2 ligand/tumor necrosis factor (TNF)-related
apoptosis-inducing ligand (Apo2L/TRAIL) is a member of the
Significance
Clinical studies with death receptor agonists have failed to advaered that the DR5 agonist antibody AMG 655 and Apo2L/TRAin vivo, which is associated with enhanced DISC assembly andare sensitized by AMG 655. The structure of the Apo2L/TRAILclustering is achieved in the presence of both agents. These fideath receptor agonists may have demonstrated limited clinpathway activation using Apo2L/TRAIL and AMG 655 in comb
TNF superfamily that induces apoptosis through engagement
of death receptors. Early observations that Apo2L/TRAIL prefer-
entially kills tumor cells over normal cells highlighted its potential
as a candidate therapeutic in cancer and hastened the develop-
ment of monoclonal antibodies selectively targeting death re-
ceptor 4 (DR4) and death receptor 5 (DR5) (Pitti et al., 1996;Wiley
et al., 1995; Ashkenazi et al., 1999). Binding of Apo2L/TRAIL or
agonist antibodies to their cognate death receptors initiates a
nce beyond phase 2 because of limited efficacy. We discov-IL can cooperate to promote tumor cell killing in vitro andapoptotic signaling. Cells resistant to soluble Apo2L/TRAIL-AMG 655-DR5 complex highlights how enhanced receptorndings provide functional and mechanistic insight into whyical activity and provide a strategy to reconsider extrinsicination as a cancer therapeutic.
Cancer Cell 26, 177–189, August 11, 2014 ª2014 Elsevier Inc. 177
Cancer Cell
TRAIL and AMG 655 Promote Higher Order Clustering
caspase-mediated apoptosis signaling cascade, resulting in cell
death (Pan et al., 1997; Pitti et al., 1996; Sheridan et al., 1997;
Walczak et al., 1997). Although binding of death receptors is pre-
requisite for signal propagation, the level of death receptor
expression in general is a poor predictor of Apo2L/TRAIL sensi-
tivity (Clodi et al., 2000; Kontny et al., 2001; Lincz et al., 2001;
Lippa et al., 2007; Wagner et al., 2007). Not all cancer cells are
sensitive to apoptosis induction via death receptor activation,
and multiple resistance mechanisms have been reported (Yang
et al., 2010). The heterogeneity of resistancemechanisms under-
scores the need to better understand the biological determinants
of responsiveness and how they may be enhanced.
The requirement for receptor aggregation and clustering is a
characteristic shared by many TNF receptor family members
that is essential for efficient signal transduction and may be
one determinant of response (Ashkenazi, 2002). In this regard,
many death receptor agonist antibodies require crosslinking
for optimal activity in vitro and in vivo (Chuntharapai et al.,
2001; Kaplan-Lefko et al., 2010; Wilson et al., 2011). For
example, AMG 655 (conatumumab), a fully human antibody
that selectively targets DR5, has no activity in the absence of
crosslinking, suggesting that receptor oligomerization induced
by crosslinking is required for optimal signal transduction (Ka-
plan-Lefko et al., 2010). Several investigational therapeutics
that target DR4 and/or DR5, including recombinant human
Apo2L/TRAIL (dulanermin), AMG 655, and others, have been
evaluated in clinical studies. However, findings from randomized
phase 2 studies failed to demonstrate significant clinical benefit
(Herbst et al., 2010a; Herbst et al., 2010b; Kindler et al., 2012;
Paz-Ares et al., 2013; Soria et al., 2011; Soria et al., 2010).
Whether these agents underperformed because of insufficient
pathway activation as a result of suboptimal agonism or because
of a lack of integrity of the Apo2L/TRAIL-death receptor pathway
in human tumors remains unclear. In the present study, we
sought to understand why preclinical antitumor activity of AMG
655 and Apo2L/TRAIL was not predictive of clinical efficacy.
RESULTS
Apo2L/TRAIL and AMG 655 Cooperate In Vitro andIn VivoTo elucidate the relationship between Apo2L/TRAIL and AMG
655 for binding and agonism of DR5, we used a cell-based
viability assay using WM35 cells. WM35 cells express surface
DR5 but undetectable levels of DR4, ensuring DR5-specific
signaling with Apo2L/TRAIL (Figures S1A and S1B available
online). Because AMG 655 binds to DR5 but does not signal
without additional crosslinking (Kaplan-Lefko et al., 2010), pre-
treatment of WM35 cells with AMG 655 followed by treatment
with Apo2L/TRAIL would allow the functional relationship be-
tween the two agents for DR5 binding and agonism to be deter-
mined. Although AMG 655 in the absence of protein G had no
effect, either AMG 655 plus protein G (which provides maximal
crosslinking) or Apo2L/TRAIL induced dose-dependent loss of
viability in WM35 cells (Figure 1A). Unexpectedly, pretreatment
with AMG 655 (in the absence of protein G) markedly enhanced
killing induced by Apo2L/TRAIL. This effect was more potent
than leucine zipper (LZ)-Apo2L/TRAIL, a nonoptimized form of
the ligand known to be a potent agonist (Lawrence et al., 2001;
178 Cancer Cell 26, 177–189, August 11, 2014 ª2014 Elsevier Inc.
Walczak et al., 1999). Because the order of addition did not
significantly influence their cooperative relationship (data not
shown), Apo2L/TRAIL and AMG 655 were added simultaneously
in subsequent experiments.
We next tested a panel of monoclonal antibodies to determine
whether DR5 agonist antibodies other than AMG 655 could
cooperate with Apo2L/TRAIL. Treatment of WM35 cells with
M413, previously shown to compete with LZ-Apo2L/TRAIL for
binding to DR5 (Griffith et al., 1999), or M411 blocked Apo2L/
TRAIL-mediated killing (Figure 1B). Despite a demonstrated abil-
ity to bind DR5, other antibodies either had no effect or were able
to cooperate with Apo2L/TRAIL, suggesting that only a subset of
anti-DR5 antibodies can cooperate with Apo2L/TRAIL.
To determine the prevalence of cooperativity between Apo2L/
TRAIL and AMG 655, we evaluated panels of breast, colon, and
lung cancer cell lines. We found that a high proportion of cells
that were sensitive to either single agent demonstrated
enhanced sensitivity to the combination (Figure 1C; Figure S1C).
All cells expressed both DR4 and DR5, and there was no strong
correlation with sensitivity and receptor expression levels (Fig-
ure S1B). We also tested whether Apo2L/TRAIL and AMG 655
would cooperate in vivo. In mice bearing SW403 xenograft
tumors, only the combination of Apo2L/TRAIL and AMG 655
resulted in significant antitumor activity (Figure 1D), demon-
strating that Apo2L/TRAIL and AMG 655 show synergistic
activity in vivo as well as in vitro. In addition, we tested Apo2L/
TRAIL and AMG 655 cooperativity in an H460 lung adenocarci-
noma xenograft model, which has been shown to be partially
sensitive to Apo2L/TRAIL (Jin et al., 2004). To confirm that the
in vivo cooperativity of AMG 655 and Apo2L/TRAIL was inde-
pendent of Fcg receptor (FcgR) interactions, we included an
aglycosylated variant of AMG 655 (N297A-AMG 655) that is
unable to bind FcgRs and therefore lacks intrinsic activity. Muta-
tion of the invariant glycan at N297 in the antibody Fc domain
is known to dramatically reduce the binding affinity to Fc recep-
tors (Jung et al., 2011). When we compared the efficacy of AMG
655 or N297A-AMG 655 in the presence or absence of Apo2L/
TRAIL, we observed partial tumor growth inhibition with
Apo2L/TRAIL and no antitumor activity with either AMG 655 or
the N297A -AMG 655 variant (Figure S1D). However, tumor
growth inhibition in both combination treatment groups was
significantly greater than Apo2L/TRAIL, indicating that both
AMG 655 and N297A-AMG 655 can cooperate with Apo2L/
TRAIL in vivo and cooperativity is independent of FcgR-medi-
ated crosslinking.
During the profiling of cell lines in vitro, we noted several exam-
ples of strong cooperativity in cells that were resistant to Apo2L/
TRAIL but responsive to AMG655when crosslinked by protein G
(Figure 2A). The ability of AMG 655 (plus protein G) to induce
apoptosis suggests that the death receptor signaling pathway
is intact but that soluble Apo2L/TRAIL is not an optimal agonist
in this subgroup of cells. To test this hypothesis we cocultured
293T cells or 293T cells expressing the membrane form of
Apo2L/TRAIL (Figure 2B) with cell lines that were resistant to
soluble Apo2L/TRAIL but sensitive to Apo2L/TRAIL plus AMG
655. Although there was no effect in response to soluble
Apo2L/TRAIL (Figure 2C), we observed a dose-dependent in-
crease in caspase-3/7 activity in the presence of 293T-TRAIL
cells (Figure 2D; Figure S2). These findings indicate that
Figure 1. Apo2L/TRAIL and AMG 655 Cooperate in Multiple Cell Lines In Vitro and In Vivo
(A) WM35 cells were treated with indicated agents for 24 hr and assayed for cell viability. For combinations, protein G and AMG 655 were at 1 mg/ml.
(B) WM35 cells were treated with Apo2L/TRAIL and 1 mg/ml of the indicated antibodies for 24 hr and assayed for cell viability.
(C) Viability of breast (left) or lung (right) tumor lines 24 hr after treatment with Apo2L/TRAIL + hIgG1 (1 mg/ml), Apo2L/TRAIL, AMG 655 plus protein G (1 mg/ml), or
Apo2L/TRAIL plus AMG 655 (1 mg/ml).
(D) Mice bearing SW403 tumors were treated with Apo2L/TRAIL (60 mg/kg, five times per week, i.p.), AMG 655 (100 mg/mouse, twice per week, i.p.), or the
combination for two cycles (n = 10/group). *For Apo2L/TRAIL plus AMG 655, p < 0.0001 versus PBS by Dunnett’s test.
Data are presented as mean ± SEM for all panels. See also Figure S1.
Cancer Cell
TRAIL and AMG 655 Promote Higher Order Clustering
membrane-associated Apo2L/TRAIL was more effective at
inducing DR5 crosslinking and clustering than the soluble form
and suggest that the coadministration of Apo2L/TRAIL with
AMG 655 may functionally mimic a membrane-bound form of
Apo2L/TRAIL, providing not only a quantitative but also a quali-
tative change in how the signal is perceived.
Apo2L/TRAIL and AMG 655 Promote ReceptorOligomerization and Higher Order ClusteringTo address the mechanistic basis of cooperativity between
Apo2L/TRAIL and AMG 655, we first tested whether AMG 655
must be bivalent in order to synergize with Apo2L/TRAIL.
Although AMG 655 or a Fab02 fragment could enhance
Apo2L/TRAIL-mediated apoptosis, a Fab fragment of AMG
655 could not (Figure 3A). These data support the hypothesis
that the mechanism of cooperativity between AMG 655 and
Apo2L/TRAIL likely involves enhanced receptor crosslinking
and oligomerization. We then asked whether Apo2L/TRAIL
and AMG 655 could be coimmunoprecipitated. After treatment
of Colo205 cells with Apo2L/TRAIL, AMG 655, or the combina-
tion, lysates were immunoprecipitated with protein G (to pull
down AMG 655) and immunoblotted for the presence of
Apo2L/TRAIL and DR5. Significantly more DR5 coimmunopre-
cipitated with AMG 655 in the presence of Apo2L/TRAIL, sug-
gesting the formation of a ternary complex on Colo205 cells
(Figure 3B; Figures S3A and S3B). We also evaluated whether
the coadministration of Apo2L/TRAIL and AMG 655 resulted
in enhanced formation of a death-inducing signaling complex
(DISC), a critical step in signal initiation by oligomerized death
receptors. AMG 655 augmented the formation of a complex
Cancer Cell 26, 177–189, August 11, 2014 ª2014 Elsevier Inc. 179
Figure 2. Cells Resistant to Soluble Apo2L/TRAIL Are Sensitive to Membrane-Bound Apo2L/TRAIL
(A) SkLu-1, SW403, and SW1573 cells were treated with the indicated agents for 24 hr and assayed for cell viability. For combinations, hIgG1, protein G, and AMG
655 were at 1 mg/ml. Data are presented as ±SEM.
(B) Staining of 293T cells or 293T cells transfected with control vector or a vector expressing Apo2L/TRAIL (293T-TRAIL) was performed with anti-TRAIL antibody
(M180), and cells were analyzed using flow cytometry.
(C) SkLu-1, SW403, and SW1573 cells (23 104 cells/well) were treated with soluble Apo2L/TRAIL as indicated for 4 hr and assayed for caspase-3/7 activity. Data
are presented as ±SEM.
(D) Coculturing of 293T or 293T-TRAIL cells was performed with SkLu-1, SW403, or SW1573 cells at the indicated ratio for 4 hr, and cells were assayed for
caspase-3/7 activity.
Data are shown as mean ± SEM and are representative of at least three independent experiments. See also Figure S2.
Cancer Cell
TRAIL and AMG 655 Promote Higher Order Clustering
between FADD, caspase-8, and DR5 upon stimulation of either
Colo205 or SkLu-1 cells with Apo2L/TRAIL (Figure 3C; Fig-
ure S3C). Further support for this hypothesis was provided by
confocal microscopy, showing that only coadministration of
Apo2L/TRAIL and AMG 655 resulted in the formation of large
clusters containing AMG 655 and cleaved caspase-8 (Fig-
ure 3D), reinforcing the conclusion that these enhanced clusters
represent functional signaling complexes. Taken together,
these data indicate that the cooperative induction of apoptosis
by Apo2L/TRAIL and AMG 655 is due to increased DISC forma-
tion as a consequence of enhanced DR5 crosslinking. To
ensure that Apo2L/TRAIL and AMG 655 were incorporated
into a ternary complex on cells, we performed a gel filtration
analysis. WM35 cells were treated with AMG 655, Apo2L/
TRAIL, or the combination, and lysates were separated by
size-exclusion chromatography and immunoblotted for the
presence of DR5. Figure 3E shows that DR5 elutes in the higher
molecular weight fractions only when Apo2L/TRAIL and AMG
655 are coadministered, indicating the formation of a higher
order complex.
180 Cancer Cell 26, 177–189, August 11, 2014 ª2014 Elsevier Inc.
To better understand the molecular interactions between
Apo2L/TRAIL, DR5, and AMG 655, we determined the crystal
structure of the ternary complex of Apo2L/TRAIL, the DR5 ecto-
domain, and the Fab fragment of AMG 655 at 3.3 A resolution
(Table 1). The Apo2L/TRAIL-DR5-AMG 655-Fab complex is
composed of an Apo2L/TRAIL homotrimer with three DR5
monomers bound diagonally along the crevices between two
neighboring Apo2L/TRAIL molecules and three AMG 655 Fabs
bound to the three DR5 monomers individually (Figure 4A).
Although there is extensive interaction between AMG 655 and
DR5, we did not observe any atomic contacts between AMG
655 and Apo2L/TRAIL. The AMG 655-DR5 interface covers an
area of approximately 1,249 A2, split evenly between AMG 655
(644 A2) and DR5 (605 A2). Residues of all six complemen-
tarity-determining regions (CDRs) of AMG 655 are involved in
the interaction with DR5 (Figure 4B). The AMG 655 binding site
on DR5 consists of 11 residues, forming a nearly contiguous
region in the cysteine-rich domain 2 (Figure 4C; Figure S4). The
structure of DR5 and Apo2L/TRAIL in the ternary complex super-
imposes well with the structure of the binary Apo2L/TRAIL-DR5
Figure 3. Apo2L/TRAIL and AMG 655 Promote
Receptor Oligomerization and Increased Signaling
(A) Colo 205 cells were treated with the indicated agents
and analyzed for cell viability after 6 hr. Fixed molar
amounts of AMG 655, AMG 655 Fab, or AMG 655 Fab02were added. Data are presented as ± SEM.
(B) Lysates from Colo205 cells treated for 30 min with
media (C), 1 mg/ml AMG 655 (655), 1 mg/ml Apo2L/TRAIL
(Apo), 1 mg/ml Apo2L/TRAIL plus 1 mg/ml human IgG (Apo
IgG), or 1 mg/ml Apo2L/TRAIL plus 1 mg/ml AMG 655 (Apo
655) were immunoprecipitated using protein G (to pull
down AMG 655) and immunoblotted using anti-hIgG (to
detect AMG 655), anti-DR5, or anti-TRAIL antibody.
(C) Lysates from Colo205 (left) or SkLu-1 (right)
cells treated for 4 hr with Apo2L/TRAIL (Apo) (Colo205
1.6 ng/ml, SkLu-1 1 mg/ml), 1 mg/ml AMG 655 (655), or the
combination (Cb) were immunoprecipitated using mouse
anti-FADD antibody (or IgG, far right lane) and immuno-
blotted using anti-DR5, anti-caspase-8 (C8), or anti-FADD
antibody.
(D) SkLu-1 cells treated with 1 mg/ml Apo2L/TRAIL (left),
1 mg/ml AMG 655 (center), or the combination (right) for
15 min were stained with DAPI (blue) or antibodies against
human IgG (red) to visualize AMG 655 or cleaved C8
(green). White triangles indicate the colocalization of AMG
655 and cleaved C8. The scale bar represents 4 mm.
(E) WM35 cells were treated for 15 min with 1 mg/ml AMG
655, 1 mg/ml Apo2L/TRAIL, 1 mg/ml Apo2L/TRAIL plus
1 mg/ml human IgG, or 1 mg/ml Apo2L/TRAIL plus 1 mg/ml
AMG 655. Total soluble protein from cell lysates was
separated by size-exclusion chromatography, and frac-
tions 2 to 11 from each treatment were immunoblotted for
the presence of DR5. (Top) UV absorption profile, which
was identical for all lysates. Numbers above the peaks
correspond to molecular weights from a standard.
Numbers 1 to 12 correspond to elution of indicated
fractions. Elution volume is noted on the x axis. mAU =
milli-absorbance units. For western blots, the first lane
corresponds to WM35 total lysate before fractionation.
See also Figure S3.
Cancer Cell
TRAIL and AMG 655 Promote Higher Order Clustering
Cancer Cell 26, 177–189, August 11, 2014 ª2014 Elsevier Inc. 181
Table 1. Apo2L/TRAIL-DR5-AMG 655 Structural Data Collection
and Refinement Statistics
Apo2L/TRAIL-DR5-AMG 655 Fab
Data Collection
Space group P6122
Cell dimensions
a, b, c (A) 152.01, 152.01, 613.16
a, b, g (�) 90, 90, 120
Resolution (A) 3.30
Rsym or Rmerge 0.166 (0.698)
I/sI 9.4 (2.25)
Completeness (%) 98.9 (99.9)
Redundancy 4.3 (4.4)
Refinement
Resolution (A) 3.30
No. of reflections 60,562
Rwork/Rfree 0.229/0.285
No. of atoms
Protein 15,933
Ligand/ion 1
Water
B-factors
Protein 75.39
Ligand/ion 57.08
Water
Rmsds
Bond lengths (A) 0.0099
Bond angles (�) 1.61
One crystal was used to obtain the structure. Values in parentheses are
for highest-resolution shell.
Cancer Cell
TRAIL and AMG 655 Promote Higher Order Clustering
complex (Hymowitz et al., 1999; Mongkolsapaya et al., 1999),
suggesting that binding of AMG 655 to DR5 does not alter the
conformation of either DR5 or Apo2L/TRAIL.
To infer the structure of full-length AMG 655, we superim-
posed the three Fab fragments in the ternary complex with Fab
fragments from previously determined full-length immunoglob-
ulin G1 (IgG1) antibody structures (Guddat et al., 1993; Saphire
et al., 2001). On the basis of the AMG 655 Fab conformation,
we predict that the Fc domain is pointing down toward the cell
surface (Figure 4D). Despite this orientation, our functional
studies support that the Fc domain of AMG 655 is available for
engagement with protein G and Fc receptors, and this may be
afforded by the inherent flexibility within the antibody hinge
region. We then modeled the positioning of three additional
ternary complexes, which are situated in the same plane as the
original ternary complex, parallel to the cell membrane (Fig-
ure 4E). Further addition of ternary complexes results in an array
of DR5 molecules on the cell surface in a hexagonal honeycomb
pattern, with three DR5 molecules in each node (Figure 4F). This
model illustrates how co-administration of trimeric Apo2L/TRAIL
and bivalent AMG 655 may promote superclustering of DR5,
thereby leading to efficient DISC assembly and enhanced
apoptotic signaling.
182 Cancer Cell 26, 177–189, August 11, 2014 ª2014 Elsevier Inc.
Apo2L/TRAIL and aMouseDR5Antibody AreEfficaciousat Tolerated DosesMice harbor only one death receptor for Apo2L/TRAIL, mouse
DR5 (mDR5)/muTRAILR2 (Takeda et al., 2007). Whereas AMG
655 does not bind to mDR5, MD5-1, a monoclonal antibody
that selectively recognizes mDR5, exhibits killing activity on
mouse tumor cells upon crosslinking in vitro (Takeda et al.,
2004). We reasoned that MD5-1 may have similar crosslinking
requirements to AMG 655 and therefore may also cooperate
with Apo2L/TRAIL. We treated murine tumor cell lines with
Apo2L/TRAIL and MD5-1 in vitro and found that the majority of
lines were efficiently killed only when Apo2L/TRAIL and MD5-1
were combined (Figure 5A). All of the lines expressed similar
levels of mDR5 (Figure S5A). We selected two cell lines, Renca
and LLC1, to test for Apo2L/TRAIL and MD5-1 cooperativity
in vivo. In both models, we observed marked antitumor activity
in the presence of Apo2L/TRAIL and MD5-1 but not with either
agent alone (Figure 5B; Figures S5B–S5D). Tumors were har-
vested for histological analysis 3 days after the start of treatment.
Upon macroscopic observation, tumors treated with Apo2L/
TRAIL and MD5-1 displayed a hemorrhagic appearance (Fig-
ure 5C). Histological evaluation as early as 8 hr after treatment
revealed that in contrast to either agent alone, tumor cells as
well as tumor-associated vasculature from mice treated with
Apo2L/TRAIL and MD5-1 were positive for activated caspase-
3, suggesting that both tumor cells and the surrounding vascula-
ture underwent apoptosis after treatment with Apo2L/TRAIL and
MD5-1 (Figure 5D; Figure S5E). Indeed, we observed strong
expression of mDR5 on both tumor cells and associated vascu-
lature but, interestingly, could not detect mDR5 expression on
vasculature in other tissues (Figure 5E; Figure S5F). Disruption
of tumor-associated vasculature by crosslinked Flag-Apo2L/
TRAIL, but not soluble Apo2L/TRAIL, was previously reported
(Wilson et al., 2012). Our results are consistent with previous
findings in that soluble Apo2L/TRAIL alone had no effect, but
oligomeric forms of Apo2L/TRAIL, either by Flag-mediated
crosslinking (Wilson et al., 2012) or by coadministration with
MD5-1, can induce antitumor activity by impacting tumor cells
directly, as well as tumor-associated vasculature where mDR5
is expressed.
We did not observe any evidence of toxicity in tissues exam-
ined from syngeneic tumor efficacy studies (Figure S5G). How-
ever, coadministration of Apo2L/TRAIL andMD5-1 but not either
agent alone to naive mice at high doses resulted in decreased
body weight, associated with diarrhea, dehydration, and scruffy
coat over the course of one cycle. Upon necropsy, gross obser-
vations were noted in the stomach and gastrointestinal (GI) tract
in the combination treatment group only. Histological analysis
indicatedminimal to severe apoptosis, degeneration, and necro-
sis in the epithelium of tissues that expressed the highest levels
of mDR5 only when Apo2L/TRAIL and MD5-1 were coadminis-
tered. This effect was observed predominantly in the stomach,
duodenum, cecum, colon, and gallbladder (Figures S5H and
S5I). There was evidence of minimal hepatocyte vacuolation in
the liver, but no mDR5 expression was detectable in liver by
immunohistochemistry (Figure S5I). Our findings from efficacy
studies in syngeneic models and high-dose tolerability studies
indicate that a therapeutic window can be achieved when
Apo2L/TRAIL and MD5-1 are coadministered in mice, and
Figure 4. Crystallographic Analysis of the
Apo2L/TRAIL-DR5-AMG 655 Fab Ternary
Complex
(A) Ribbon representation of the Apo2L/TRAIL-
DR5-AMG 655 Fab complex with view down the
3-fold axis. Apo2L/TRAIL (blue) with Zn atom (red)
binds three DR5 monomers (tan, space-filling
mode), and each DR5 monomer binds one AMG
655 Fab fragment (light chain, yellow; heavy chain,
green) individually.
(B) Locations of DR5 and AMG 655 residues
involved in binding (red). CDRH1-H3 correspond to
heavy chain, and CDRL1-L3 correspond to light
chain.
(C) The AMG 655 binding site (orange) on DR5 (tan)
involves residues K98, Y99, Q101, R118, C119,
S121, E123, V124, E125, P128, and E140. The
Apo2L/TRAIL binding region is shown in purple.
(D) Inferred structure of full-length AMG 655 in
complex with Apo2L/TRAIL and DR5 by super-
position with previously determined structures of
IgG1 antibodies. The Fc domain of AMG 655
(orange) is predicted to point down toward the cell
surface.
(E) View of three Apo2L/TRAIL-AMG 655-DR5
ternary complexes down the 3-fold axis. The AMG
655 Fc domain points down into the page.
(F) Model for induction of higher order clustering by
multiple ternary complexes. The image in (A) is
represented in the black box. Red dots correspond
to Apo2L/TRAIL bound to three DR5 receptors.
Blue lines denote AMG 655 Fab fragments ex-
tending from DR5.
See also Figure S4.
Cancer Cell
TRAIL and AMG 655 Promote Higher Order Clustering
potential target tissues for toxicity are associated with higher
mDR5 expression. To understand whether DR5 expression
was conserved across species, we evaluated a subset of tissues
from mouse, human, and nonhuman primate, with emphasis on
those that exhibited high mDR5 expression in mouse. As in the
mouse, DR5 expression was more abundant in the GI tract
tissues of human and nonhuman primate compared with other
tissues examined (Figure S5J).
Apo2L/TRAIL and AMG 655 Are Efficacious againstPrimary Human Tumor CellsTo determine whether our findings might have clinical relevance,
we tested Apo2L/TRAIL plus AMG 655 on fresh primary tumor
disaggregates from 12 lung carcinoma patients (Table S1). We
initially evaluated the degree of Apo2L/TRAIL and receptor
expression on tumor infiltrating CD45+ cells and EpCAM+ tumor
cells. CD45+ cells expressed higher amounts of the decoy recep-
tors DcR1 and DcR2, whereas EpCAM+ cells expressed mostly
DR5 and DcR1 (Figure 6A). Tumor disaggregates were then
treated with AMG 655 or Apo2L/TRAIL alone or in combination,
and caspase-3 activity was measured after 18 hr (Figure 6B).
Cancer Cell 26, 177–18
There were varying degrees of sensitivity
among the lung tumor samples tested
and a trend toward increased responsive-
ness in the presence of both agents
versus either agent alone.
We also tested whether Apo2L/TRAIL plus AMG 655 might
have killing activity on nonmalignant primary cells. All of the
nonmalignant cells tested expressed DR5, and although some
activity was noted in hepatocytes and keratinocytes from
different donors, it was not consistent and did not approach
the degree of cooperativity observed on tumor cell lines (Fig-
ure 6C; Figure S6). No combinatorial effect was observed on
monocytes, renal epithelial cells, endothelial cells, or fibroblasts
from multiple donors.
DISCUSSION
The therapeutic potential of death receptor agonist antibodies
or soluble Apo2L/TRAIL has not yet been realized in a clinical
setting. For agonist antibodies, one possible reason may be
the crosslinking requirement for effective receptor engagement
and clustering. In vivo, crosslinking is mediated by FcgRs
expressed on the surface of immune cells. Both drozitumab
and AMG 655 are dependent on such crosslinking for activity
(Kaplan-Lefko et al., 2010; Wilson et al., 2011). Moreover,
in phase 2 trials conducted with AMG 655, a trend toward a
9, August 11, 2014 ª2014 Elsevier Inc. 183
Figure 5. Murine Tumor Cells Are Sensitive to Apo2L/TRAIL and MD5-1 In Vitro and In Vivo
(A) Murine tumor lines treated with Apo2L/TRAIL, Apo2L/TRAIL plus hamster IgG (1 mg/ml), MD5-1 or Apo2L/TRAIL + MD5-1 (1 mg/ml) were assayed for cell
viability after 48 hr. Data are shown as mean ± SEM and are representative of at least three independent experiments.
(B) Renca tumor-bearing mice were treated with Apo2L/TRAIL (5 mg/kg, five times per week), MD5-1 (1 mg/mouse, once per week), or the combination for two
cycles (n = 10/group). *For Apo2L/TRAIL plus MD5-1, p < 0.0001 versus PBS by Dunnett’s test. Three of ten tumors in the combination group had nomeasurable
tumor. Data are presented as mean ± SEM.
(C) Representative tumors from macroscopic analysis of PBS or Apo2L/TRAIL plus MD5-1 treated tumors 3 days after the start of treatment.
(D) Representative tissues from Renca tumor-bearing mice treated with PBS or Apo2L/TRAIL plus MD5-1, 8 hr after treatment, stained with antibody for active
caspase-3.
(E) Representative tissues stained with antibody for mDR5. For both (D) and (E), Apo2L/TRAIL and MD5-1 single-agent groups were indistinguishable from PBS-
treated mice (data not shown). Individual vessels are denoted by black arrows. The scale bars represent 100 mm (upper left panel) and 50 mm (other panels).
See also Figure S5.
Cancer Cell
TRAIL and AMG 655 Promote Higher Order Clustering
stronger treatment effect was observed in patients carrying
a high-affinity allele of the FcgRIIIa receptor, suggesting
that the high-affinity form of the receptor may be associated
with enhanced responsiveness (Pan et al., 2011). Therefore,
insufficient crosslinking could contribute to a diminished
antitumor response mediated by death receptor agonist
antibodies.
184 Cancer Cell 26, 177–189, August 11, 2014 ª2014 Elsevier Inc.
The limited clinical efficacy observed with Apo2L/TRAIL could
be attributed to its short half-life (Herbst et al., 2010a), but addi-
tional limitationsmay be associatedwith its soluble form. Several
soluble TNF family ligands show different activity profiles from
their corresponding membrane-associated forms. For example,
FASL has weak apoptosis-inducing activity as a soluble form
(O’Reilly et al., 2009; Schneider et al., 1998; Suda et al., 1997)
Figure 6. Activity of Apo2L/TRAIL and AMG 655 on Primary Tumor and Nonmalignant Cells
(A) Freshly resected lung tumor-derived cell suspensions were stained with antibody cocktails against Apo2L/TRAIL, DR4, DR5, DcR1, or DcR2 with either
EpCAM (top) or CD45 (bottom). The symbol on each graph represents an individual tumor sample. Black horizontal lines represent mean expression.
(B) A gating strategy to identify EpCAM+ events was used, and samples were treated with the indicated agents (IgG1, 1 mg/ml; AMG 655, 1 mg/ml; Apo2L, 2 mg/ml;
Apo2L lo, 0.13 ng/ml; Apo 2L med, 1.6 ng/ml; Apo 2L hi, 2 mg/ml) for 18 hr and analyzed for caspase-3/7 activity. Results are shown as percentage viable events
remaining in treated samples relative to vehicle. Black horizontal lines represent mean expression. For one sample (black squares), there was not enoughmaterial
to run Apo2L plus IgG1 control.
(C) Viability of primary human cells treated with Apo2L/TRAIL, AMG 655 plus protein G (1 mg/ml), or the combination (AMG 655, 1 mg/ml) for 48 hr (monocyte,
human umbilical vein endothelial cell, fibroblast, epithelial, keratinocyte) or 72 hr (hepatocyte).
Data are presented as mean ± SEM. See also Figure S6 and Table S1.
Cancer Cell
TRAIL and AMG 655 Promote Higher Order Clustering
but can cooperate with an agonistic anti-FAS receptor antibody
to kill target cells (Xiao et al., 2002). Others have also described
limited responses with soluble Apo2L/TRAIL compared with
membrane-associated or oligomerized forms (Berg et al.,
2007; Muhlenbeck et al., 2000; Wajant et al., 2001). This has
prompted the development of alternative death receptor ago-
nists engineered to promote enhanced receptor clustering.
Many approaches to either improve the stability of Apo2L/TRAIL
or overcome the limitations of bivalent antibodies have been
reported (de Bruyn et al., 2013; Gieffers et al., 2013; Huet
et al., 2012; Pavet et al., 2010; Pfizenmaier and Szymkowski,
2011; Swers et al., 2013).
Coadministration of Apo2L/TRAIL and AMG 655 potently kills
tumor cells without a requirement for additional crosslinking. We
propose that the additional receptor crosslinking provided by
AMG 655 changes the signal delivered by soluble Apo2L/TRAIL
into one that functionally resembles membrane-associated
Apo2L/TRAIL. The failure of soluble Apo2L/TRAIL to appropri-
ately engage receptors may have implications into its limited
clinical activity. Similarly, the efficacy of AMG 655 may have
Cancer Cell 26, 177–189, August 11, 2014 ª2014 Elsevier Inc. 185
Cancer Cell
TRAIL and AMG 655 Promote Higher Order Clustering
been limited by insufficient crosslinking by Fc receptor-bearing
cells in the tumor. Importantly, AMG 655 lacks activity in the
absence of an exogenous crosslinker such as protein G, and
protein G-mediated cross-linking in vitro is likely an overrepre-
sentation of the extent of crosslinking that occurs in vivo. Thus,
the dose response of single-agent AMG 655 (plus protein G) is
not reflective of the contribution of AMG 655 in the context of
soluble Apo2L/TRAIL, and, importantly, agonism by the combi-
nation is independent of the need for crosslinking (i.e., by Fc re-
ceptors in vivo).
Despite the enhanced killing activity of Apo2L/TRAIL and AMG
655 administered together, not all cell lines could be sensitized
by the combination. Current understanding of mechanisms of
resistance suggests thatmultiple factors contribute to a resistant
phenotype. These include receptor expression, the ability of
the receptor to initiate a signal, and the threshold sensitivity of
the cell to the initiating signal, which may be dependent on the
expression of other pro- and antiapoptotic factors. We have
observed that cells that are resistant to combination Apo2L/
TRAIL plus AMG 655 are also resistant to soluble Apo2L/TRAIL
and AMG 655 (plus protein G) individually. We believe this pop-
ulation likely includes cells that lack pathway integrity and for
which agonism is not a limiting factor. In contrast, cells that
have an intact pathway butmay be below the threshold for sensi-
tivity are potently sensitized by the combination.
Wepresent the structure of aTNF receptor in a ternary complex
with its ligand and an agonist antibody. The X-ray structure of the
Apo2L/TRAIL-DR5-AMG 655 Fab ternary complex provides a
unique opportunity for studying the molecular mechanism of
DR5 activation by Apo2L/TRAIL and AMG 655. Although only
the Fab fragment of AMG 655 was used for crystallization, we
could infer the structure of AMG 655 IgG from previously deter-
mined human IgG1 antibody structures. Our structural data pro-
vide the basis for amodel depicting how a ternary complex forms
andhow recruitment of additional ternary complexes can result in
enhanced DR5 clustering and signaling. This is consistent with
the model of DISC formation through clustering represented by
the FAS-FADD death domain complex structure (Scott et al.,
2009). Similar findings based on structure have been described
for the facilitation of TRAF6-mediated ubiquitination, where
TRAF6 dimerization induces higher order oligomerization (Yin
et al., 2009). Both this model and ours describe a proposed scaf-
fold for higher order oligomerization thatmay facilitate ubiquitina-
tion or proximity-induced caspase activation. It was recently
proposed that Apo2L/TRAIL-induced cell death is triggered via
ligand-induced DR5 dimerization to generate organized ligand-
receptor networks (Valley et al., 2012). In our model, the addition
of AMG 655 serves to stabilize Apo2L/TRAIL-DR5 complexes by
bridging between receptors, thereby promoting enhanced clus-
tering and signaling.
Our structural analysis also provides insight into the improved
killing activity observed both in vitro and in vivo. Importantly,
coadministration of Apo2L/TRAIL with the murine mDR5 anti-
body MD5-1 resulted in strong antitumor activity. The coopera-
tivity of Apo2L/TRAIL and MD5-1 provided an opportunity to
test the effects of enhanced signaling in a model in which safety
margins could be assessed. At high doses (200 times higher
MD5-1 and 12 times higher Apo2L/TRAIL than used in the effi-
cacy studies), we saw evidence of toxicity in areas where
186 Cancer Cell 26, 177–189, August 11, 2014 ª2014 Elsevier Inc.
mDR5 was most abundantly expressed, notably in tissues of
the GI tract. Analysis of DR5 expression in other species also
showed the highest expression in GI tract tissues. Collectively,
this supports the idea that despite the increased potency of
Apo2L/TRAIL and AMG 655, a therapeutic window may still
be achievable, and evaluation of combination agonists in a
nonhuman primate may be predictive of tolerability. As previ-
ously reported for an oligomerized form of Apo2L/TRAIL, we
also noted effects specifically on tumor vasculature, where
mDR5 was abundantly expressed. It is not clear whether
Apo2L/TRAIL and AMG 655 would exert similar effects, as
AMG 655 and MD5-1 may have unique properties. Moreover, lit-
tle is known about DR5 expression in human tumor-associated
vasculature. In previous reports, endothelial DR5 expression in
human lung tumor specimens was found to be rare (Wilson
et al., 2012).
Notably, we observed a trend toward increased sensitivity
to Apo2L/TRAIL and AMG 655 in primary human lung tumor
samples compared with either agent alone. Treatment with
AMG655 alone or Apo2L/TRAIL up to 2 mg/ml resulted inminimal
activity. In contrast, even small amounts of Apo2L/TRAIL
(<2 ng/ml) were sufficient to cooperate with AMG655 in sensitive
samples. Although there was variable sensitivity to the combina-
tion across the lung samples profiled, these findings highlight the
potential for improving the likelihood of response in a patient
setting by coadministration of Apo2L/TRAIL and AMG 655.
In summary, the quantitatively and qualitatively enhanced
agonism generated by combining Apo2L/TRAIL and AMG 655
provides insight into the limited activity in previous clinical trials.
We propose that soluble Apo2L/TRAIL and AMG 655 are subop-
timal agonists as single agents because of an inability to induce
sufficient receptor clustering and activation. In our model, solu-
ble Apo2L/TRAIL can efficiently induce the formation of active
receptor trimers but is less efficient at promoting their clustering,
thereby producing insufficient signaling. The bivalent nature of
AMG 655 means that it is unlikely to be an efficient inducer of re-
ceptor trimers. However, its contribution in the combination
setting may be to bridge receptors to promote clustering and
productive signaling. This model provides possible explanations
for the limited activity of both Apo2L/TRAIL and AMG 655 that
are unified by a common aspect of TNF receptor biology: the
need for receptor clustering. Furthermore, in the context of a
combination that is both potent and independent of exogenous
crosslinking, Apo2L/TRAIL and AMG 655 compensate for each
other’s liabilities, and our results support continued investigation
of this combination for cancer therapy.
EXPERIMENTAL PROCEDURES
Cells and Reagents
Tumor cell lines were obtained from established sources (American Type
Culture Collection, Deutsche Sammlung von Mikroorganismen und Zellkultu-
ren, and Sigma) and maintained according to suppliers’ recommendations.
AMG 655 was made as previously described (Kaplan-Lefko et al., 2010).
Mouse antihuman DR5 antibodies M410, M411, M412, M413, and M415
were made at Immunex/Amgen, and M413 has been previously described
(Griffith et al., 1999). LZ-hu Apo2L/TRAIL, huApo2L/TRAIL (114–281),
N297A-AMG 655, AMG 655 Fab, AMG 655 Fab02, DR5 extracellular domain,
and anti-mu DR5 clone MD5-1 were made at Amgen. Additional details are
available in Supplemental Experimental Procedures.
Cancer Cell
TRAIL and AMG 655 Promote Higher Order Clustering
Cell Viability and Caspase-3/7 Activity Assays
For tumor cell viability assays, 2 3 104 cells/well were plated in 96-well plates
and treated for 24 hr (human tumor lines) or 48 hr (mouse tumor lines) in dupli-
cate or triplicate with the indicated amounts of human IgG1 (1 mg/ml), Apo2L/
TRAIL, AMG 655, AMG 655 plus protein G (Pierce) (1 mg/ml), Apo2L/TRAIL plus
AMG 655 (1 mg/ml), MD5-1, Apo2L/TRAIL plus MD5-1 (1 mg/ml), or Apo2L/
TRAIL plus hamster IgG (1 mg/ml). As shown in Figure 3A, a fixed concentration
(68 nM) of AMG 655, AMG 655 Fab, AMG 655 Fab02, or control medium was
added, and cell viability was assayed after 6 hr using Cell Titer Glo (Promega)
or ATPLite (Perkin-Elmer) according to the manufacturer’s protocols. Nonma-
lignant cells were treated for 48 or 72 hr. For coculture experiments, indicated
amounts of 293T cells transected with control pCDNA3.1 vector or pAMG21-
hu TRAIL (T95-G281) (generated at Amgen) were mixed with 2 3 104 cells
(SkLu-1, SW1573, SW403) in 96-well plates. After 4 hr, caspase-3/7 activity
was measured using a Caspase-Glo assay (Promega) according to the manu-
facturer’s protocols. Additional details are available in Supplemental Experi-
mental Procedures.
Flow Cytometry
Antibodies used for flow cytometry were as follows: anti-TRAIL: M180; anti-
DR4: M271; anti-DR5: M413 (all at 10 mg/ml, all made at Immunex/Amgen)
with goat antimouse Fab-PE secondary (Jackson Labs); and anti-mDR5:
biotinylated MD5-1 (1.25 mg/ml; eBioscience). After washing, cells were resus-
pended in 10 mg/ml biotinylated secondary for 45 min at 4�C, washed again,
and incubated with 10 mg/ml streptavidin-PE (BD PharMingen) for 45 min at
4�C. For analysis of Apo2L/TRAIL and Apo2L/TRAIL receptors on human pri-
mary tumor cells, tumor-derived cell suspensions and control cell lines were
stained for 30 min at room temperature with antibody cocktails containing
EpCAM-APC (clone EBA-1; BD Biosciences), CD45 (PE-Cy7; clone HI30;
BD Biosciences), and antibodies against IgG1, IgG2b, Apo2L/TRAIL, DR4,
DR5, DcR1, or DcR2. Before flow cytometric analysis, samples were stained
with 5 mg/ml 7-AAD (Life Technologies) to exclude dead and apoptotic cells.
For each sample, 104 viable EpCAM+ events were acquired. Additional details
are available in Supplemental Experimental Procedures.
Gel Filtration Analysis
WM35 lysates (0.75 ml) prepared from 6 3 107 cells for each condition were
applied to a Superose 6 10/300 GL column (GE Healthcare) equilibrated in
25 mM Tris (pH 7.5), 150 mM NaCl, 10 mM 2-ME, 10% glycerol, and 0.1%
Tween 20, and 1 ml fractions were collected starting just prior to the column
void volume. The column was calibrated using gel filtration standards (BioRad
151-1901). Fractions were subsequently analyzed by immunoblotting for the
presence of DR5.
Protein Crystallization, Structure Determination, and Refinement
Crystals of the Apo2L/TRAIL-DR5-AMG 655 Fab ternary complex were ob-
tained in hanging drops with 0.1 M Tris (pH 8.0), 1.0 M LiCl, 0.2 M MnCl2,
and 10% polyethylene glycol 6000. These crystals belong to the space group
P6122 with unit cell parameters of a = b = 152 and c = 613 A. Paratone-N min-
eral oil was used as a cryoprotectant. For data collection, crystals were ori-
ented so that the longest axis (c) was perpendicular to the X-ray beam. Diffrac-
tion data were collected on beamline 21-ID-F at the Advanced Photon Source
and processed and scaled with HKL 2000. The structure of the ternary Apo2L/
TRAIL-DR5-AMG655 Fab complex was solved bymolecular replacement with
PHASER using the structures of the binary Apo2L/TRAIL-DR5 complex (Pro-
tein Data Bank [PDB] accession number 1D0G) and variable (VH and VL) and
constant (CH1 and CL) domains of an IgG1 antibody (PDB accession number
3B2U) with 89% sequence similarity to AMG 655 as the search models. Model
building was carried out with Coot, and refinement was done using Refmac in
CCP4. In the final structure, 99.1% of the residues are in the allowed regions of
the Ramachandran plot. Details on data processing and refinement statistics
are provided in Table 1. Additional details are available in Supplemental Exper-
imental Procedures.
Mouse Efficacy and Tolerability Studies
For all in vivo studies, mice were cared for in accordance to the Guide for the
Care and Use of Laboratory Animals, eighth edition. All research protocols
were approved by the Amgen Institutional Animal Care and Use Committee.
For xenograft models, mice were injected with 2 3 106 SW403 or 5 3 106
H460 tumor cells. Mice were randomized into groups (n = 10) and treated intra-
peritoneally (i.p.) with Apo2L/TRAIL (60 mg/kg, five times per week), AMG 655
(100 mg/mouse, twice per week), or the combination for two cycles. For synge-
neic models, mice were injected with 53 105 Renca or LLC1 cells. Mice were
randomized into groups (n = 10) and treated i.p. with Apo2L/TRAIL (5 mg/kg,
five times per week), MD5-1 (1 mg/mouse, once per week), or the combination
for two cycles. For the tolerability model, mice were treated with PBS, Apo2L/
TRAIL (60 mg/kg, once per day for 5 days), MD5-1 (200 mg, twice per week), or
the combination (n = 10/group). Mice were necropsied on day 5 of treatment.
Additional details are provided in Supplemental Experimental Procedures.
Immunoprecipitation and Western Blotting
SkLu-1 and Colo205 cells treated for 4 hr with control medium, AMG 655,
Apo2L/TRAIL, or Apo2L/TRAIL plus AMG 655 at the indicated amounts and
lysed. Seven microliters of anti-FADD (Sigma F8053), anti-caspase-8 (Cell
Signaling 9746), or IgG (Fisher BP268550) was added to each reaction, and
after 4 hr, 50 ml of 50% protein G Sepharose slurry was added and rotated
at 4�C overnight. For Figure 3B, 50 ml of 50% protein G Sepharose slurry
was added and rotated at 4�C overnight. Immunoprecipitates were analyzed
by immunoblotting with anti-TRAIL (Abcam ab9959), anti-hIgG (Abcam
ab109489), anti-FADD (Cell Signaling 2782), anti-DR5 (Cell Signaling 8074),
and anti-caspase-8 (Cell Signaling 9746) or antitubulin (Sigma T5168), using
the Odyssey image system (LI-COR Biosciences).
Immunofluorescence and Confocal Microscopy
SkLu-1 cells were treated with Apo2L/ TRAIL or AMG 655 alone or in combi-
nation as indicated for 15 min at 37�C. Cells were fixed, permeabilized, and
incubated overnight at 4�C with rabbit anticleaved caspase-8 (Cell Signaling
9496) andCy3-conjugated donkey antihuman IgG (4 mg/ml; Jackson Immunor-
esearch). After washes in Tris-buffered saline and Tween-20, AlexaFluor 488-
conjugated donkey antirabbit IgG (4 mg/ml; Life Technologies) was applied.
After final washes, cells were mounted in ProLong Gold antifade reagent
(Life Technologies) and imaged on a Zeiss LSM 510 confocal microscope.
Additional details are provided in Supplemental Experimental Procedures.
Immunohistochemistry
All nonhuman primates were cared for in accordance with the Guide for the
Care and Use of Laboratory Animals. Animals were socially housed in a facility
accredited by the Association for Assessment and Accreditation of Laboratory
Animal Care International. All research protocols were approved by the Sierra
Biomedical Institutional Animal Care and Use Committee. For human and
nonhuman primate DR5 staining, mouse antihuman DR5 (M413) (Amgen) at
5 mg/ml for 60min ormouse IgG1 isotype control (BD PharMingen 550878), fol-
lowed by biotinylated goat antimouse IgG (H&L) (Vector Labs, BA-9200) for
30 min, was used. For mDR5 staining, hamster antimouse DR5 (eBioscience
14-5883), or hamster IgG isotype control (eBioscience 14-4888) at 7.5 mg/ml
for 60 min, followed by biotinylated rabbit anti-hamster IgG (Southern
Biotech 6215-05) for 30 min, was used. For caspase-3, cleaved caspase-3
(Cell Signaling 9661) for 60 min, followed by goat antirabbit Envision (Dako)
for 30 min, DAB (Dako) for 5 min, and Mayer’s hematoxylin (Dako) for 30 s
was used. Additional details are provided in Supplemental Experimental
Procedures.
Primary Tumor Disaggregate Analysis
Non-small-cell lung tumor samples were collected under Institutional Review
Board approval for each institution, with appropriate informed consent, from
surgical resections or after biopsy. Freshly resected human tissue samples
were obtained from the following institutions: Asterand USA, Bio-Options,
MT Group, and Conversant Bio. In all cases, tissue obtained was surplus to
standard clinical practice, and patient identity and identifying data were
redacted from tissues and clinical data. Freshly resected lung tumors were di-
gested, and cells were suspended in RPMI. Aqua Viability Reagent (Life Tech-
nologies) was added to the cell suspension per the manufacturer’s protocol to
label dead cells. A fraction of cells were removed for live cell flow cytometry
to measure receptor surface expression. Tumors (106 cells) were incubated
for 18 hr with vehicle, 1 mg/ml human IgG1 (Life Technologies), AMG 655
(1 mg/ml), or Apo2L/TRAIL (0.13, 1.6, or 2 mg/ml) combined with vehicle,
Cancer Cell 26, 177–189, August 11, 2014 ª2014 Elsevier Inc. 187
Cancer Cell
TRAIL and AMG 655 Promote Higher Order Clustering
1 mg/ml human IgG1, or AMG 655 (1 mg/ml). Cells were fixed, permeabilized,
and stained with fluorochrome-conjugated antibodies specific for cleaved
caspase-3, EpCAM, and pan-cytokeratin. Samples were run on an LSR II using
a gating strategy to identify EpCAM+ and cytokeratin+ events and exclude
Aqua Viability dye-positive events (i.e., debris). Results were reported as the
percentage of viable events remaining in treated samples relative to vehicle.
Additional details are provided in Supplemental Experimental Procedures.
Statistical Analysis
Experiments were repeated three times unless otherwise indicated. Data were
presented as mean ± SEM. For in vivo tumor studies, statistical analysis
was performed using JMP7 software (SAS Institute). Repeated-measures
ANOVA was performed over time using Dunnett’s post hoc test to determine
the p value.
ACCESSION NUMBER
The PDB code for this structure is 4N90.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Supplemental Experimental Procedures,
six figures, and one table and can be found with this article online at http://
dx.doi.org/10.1016/j.ccr.2014.04.028.
AUTHOR CONTRIBUTIONS
J.J.K., J.P., T.-H.H., and H.D. performed in vitro cell line work. R.M. and X.Z.
performed the colorectal cancer cell panel screen. T.S., T.L.B., and T.D.
performed mouse in vitro and in vivo studies. T.-H.H. performed biochemical
studies. W.C. performed confocal microscopy analyses. D.G.B. provided
pathology support. J.W.O. supported protein production. J.M.R. performed
tumor disaggregate studies. A.M.L. and X.H. performed structure studies.
J.D.G, P.M.H., and I.N.F. guided the project and contributed to experimental
design and data interpretation. P.M.H. and J.D.G. wrote the manuscript.
ACKNOWLEDGMENTS
We thank T. Polverino, B. Fanslow, R. Kendall, J. Hill, D. Saffran, and R. Loberg
for discussions; R. Kegel and B. Thangaraj for protein supply; A. Jones for cell
panel screening; K. Rohrbach for histology support, and J. Danao for receptor
expression analyses. All authors are current or former employees of Amgen
and have received stock or stock options from Amgen.
Received: January 2, 2014
Revised: February 28, 2014
Accepted: April 30, 2014
Published: July 17, 2014
REFERENCES
Ashkenazi, A. (2002). Targeting death and decoy receptors of the tumour-
necrosis factor superfamily. Nat. Rev. Cancer 2, 420–430.
Ashkenazi, A., Pai, R.C., Fong, S., Leung, S., Lawrence, D.A., Marsters, S.A.,
Blackie, C., Chang, L., McMurtrey, A.E., Hebert, A., et al. (1999). Safety and
antitumor activity of recombinant soluble Apo2 ligand. J. Clin. Invest. 104,
155–162.
Ashkenazi, A., Holland, P., and Eckhardt, S.G. (2008). Ligand-based targeting
of apoptosis in cancer: the potential of recombinant human apoptosis ligand 2/
Tumor necrosis factor-related apoptosis-inducing ligand (rhApo2L/TRAIL).
J. Clin. Oncol. 26, 3621–3630.
Berg, D., Lehne, M., Muller, N., Siegmund, D., Munkel, S., Sebald, W.,
Pfizenmaier, K., and Wajant, H. (2007). Enforced covalent trimerization
increases the activity of the TNF ligand family members TRAIL and CD95L.
Cell Death Differ. 14, 2021–2034.
Chuntharapai, A., Dodge, K., Grimmer, K., Schroeder, K., Marsters, S.A.,
Koeppen, H., Ashkenazi, A., and Kim, K.J. (2001). Isotype-dependent inhibi-
188 Cancer Cell 26, 177–189, August 11, 2014 ª2014 Elsevier Inc.
tion of tumor growth in vivo by monoclonal antibodies to death receptor 4.
J. Immunol. 166, 4891–4898.
Clodi, K., Wimmer, D., Li, Y., Goodwin, R., Jaeger, U., Mann, G., Gadner, H.,
and Younes, A. (2000). Expression of tumour necrosis factor (TNF)-related
apoptosis-inducing ligand (TRAIL) receptors and sensitivity to TRAIL-induced
apoptosis in primary B-cell acute lymphoblastic leukaemia cells. Br. J.
Haematol. 111, 580–586.
de Bruyn, M., Bremer, E., and Helfrich, W. (2013). Antibody-based fusion
proteins to target death receptors in cancer. Cancer Lett. 332, 175–183.
Gieffers, C., Kluge, M., Merz, C., Sykora, J., Thiemann, M., Schaal, R., Fischer,
C., Branschadel, M., Abhari, B.A., Hohenberger, P., et al. (2013). APG350
induces superior clustering of TRAIL receptors and shows therapeutic anti-
tumor efficacy independent of cross-linking via Fcg receptors. Mol. Cancer
Ther. 12, 2735–2747.
Griffith, T.S., Rauch, C.T., Smolak, P.J., Waugh, J.Y., Boiani, N., Lynch, D.H.,
Smith, C.A., Goodwin, R.G., and Kubin, M.Z. (1999). Functional analysis of
TRAIL receptors using monoclonal antibodies. J. Immunol. 162, 2597–2605.
Guddat, L.W., Herron, J.N., and Edmundson, A.B. (1993). Three-dimensional
structure of a human immunoglobulin with a hinge deletion. Proc. Natl.
Acad. Sci. U S A 90, 4271–4275.
Herbst, R.S., Eckhardt, S.G., Kurzrock, R., Ebbinghaus, S., O’Dwyer, P.J.,
Gordon, M.S., Novotny, W., Goldwasser, M.A., Tohnya, T.M., Lum, B.L.,
et al. (2010a). Phase I dose-escalation study of recombinant human Apo2L/
TRAIL, a dual proapoptotic receptor agonist, in patients with advanced
cancer. J. Clin. Oncol. 28, 2839–2846.
Herbst, R.S., Kurzrock, R., Hong, D.S., Valdivieso, M., Hsu, C.P., Goyal, L.,
Juan, G., Hwang, Y.C., Wong, S., Hill, J.S., et al. (2010b). A first-in-human
study of conatumumab in adult patients with advanced solid tumors. Clin.
Cancer Res. 16, 5883–5891.
Huet, H., Schuller, A., Li, J., Johnson, J., Dombrecht, B., Meerschaert, K.,
Cromie, K., Bilic, S., Li, S., Garner, A., et al. (2012). TAS266, a novel tetrameric
nanobody agonist targeting death receptor 5 (DR5), elicits superior antitumor
efficacy than conventional DR5-targeted approaches. Cancer Res. 78, 3853.
Hymowitz, S.G., Christinger, H.W., Fuh, G., Ultsch, M., O’Connell, M., Kelley,
R.F., Ashkenazi, A., and de Vos, A.M. (1999). Triggering cell death: the crystal
structure of Apo2L/TRAIL in a complex with death receptor 5. Mol. Cell 4,
563–571.
Jin, H., Yang, R., Fong, S., Totpal, K., Lawrence, D., Zheng, Z., Ross, J.,
Koeppen, H., Schwall, R., and Ashkenazi, A. (2004). Apo2 ligand/tumor necro-
sis factor-related apoptosis-inducing ligand cooperates with chemotherapy to
inhibit orthotopic lung tumor growth and improve survival. Cancer Res. 64,
4900–4905.
Jung, S.T., Kang, T.H., Kelton, W., and Georgiou, G. (2011). Bypassing glyco-
sylation: engineering aglycosylated full-length IgG antibodies for human ther-
apy. Curr. Opin. Biotechnol. 22, 858–867.
Kaplan-Lefko, P.J., Graves, J.D., Zoog, S.J., Pan, Y., Wall, J., Branstetter,
D.G., Moriguchi, J., Coxon, A., Huard, J.N., Xu, R., et al. (2010).
Conatumumab, a fully human agonist antibody to death receptor 5, induces
apoptosis via caspase activation in multiple tumor types. Cancer Biol. Ther.
9, 618–631.
Kindler, H.L., Richards, D.A., Garbo, L.E., Garon, E.B., Stephenson, J.J., Jr.,
Rocha-Lima, C.M., Safran, H., Chan, D., Kocs, D.M., Galimi, F., et al. (2012).
A randomized, placebo-controlled phase 2 study of ganitumab (AMG 479) or
conatumumab (AMG 655) in combination with gemcitabine in patients with
metastatic pancreatic cancer. Ann. Oncol. 23, 2834–2842.
Kontny, H.U., Hammerle, K., Klein, R., Shayan, P., Mackall, C.L., and
Niemeyer, C.M. (2001). Sensitivity of Ewing’s sarcoma to TRAIL-induced
apoptosis. Cell Death Differ. 8, 506–514.
Lawrence, D., Shahrokh, Z., Marsters, S., Achilles, K., Shih, D., Mounho, B.,
Hillan, K., Totpal, K., DeForge, L., Schow, P., et al. (2001). Differential hepato-
cyte toxicity of recombinant Apo2L/TRAIL versions. Nat. Med. 7, 383–385.
Lincz, L.F., Yeh, T.X., and Spencer, A. (2001). TRAIL-induced eradication of
primary tumour cells from multiple myeloma patient bone marrows is not
Cancer Cell
TRAIL and AMG 655 Promote Higher Order Clustering
related to TRAIL receptor expression or prior chemotherapy. Leukemia 15,
1650–1657.
Lippa, M.S., Strockbine, L.D., Le, T.T., Branstetter, D.G., Strathdee, C.A., and
Holland, P.M. (2007). Expression of anti-apoptotic factors modulates Apo2L/
TRAIL resistance in colon carcinoma cells. Apoptosis 12, 1465–1478.
Mongkolsapaya, J., Grimes, J.M., Chen, N., Xu, X.N., Stuart, D.I., Jones, E.Y.,
and Screaton, G.R. (1999). Structure of the TRAIL-DR5 complex revealsmech-
anisms conferring specificity in apoptotic initiation. Nat. Struct. Biol. 6, 1048–
1053.
Muhlenbeck, F., Schneider, P., Bodmer, J.L., Schwenzer, R., Hauser, A.,
Schubert, G., Scheurich, P., Moosmayer, D., Tschopp, J., and Wajant, H.
(2000). The tumor necrosis factor-related apoptosis-inducing ligand receptors
TRAIL-R1 and TRAIL-R2 have distinct cross-linking requirements for initiation
of apoptosis and are non-redundant in JNK activation. J. Biol. Chem. 275,
32208–32213.
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.
Pan, G., O’Rourke, K., Chinnaiyan, A.M., Gentz, R., Ebner, R., Ni, J., and Dixit,
V.M. (1997). The receptor for the cytotoxic ligand TRAIL. Science 276,
111–113.
Pan, Y., Haddad, V., Sabin, T., Baker, N., Yong-Jiang, H., Galimi, F., Graves, J.,
Huang, C.P., and Cottrell, S. (2011). Predictive value of Fc gamma receptor IIIa
genotype in response to Conatumumab in three Phase II studies. J. Clin.
Oncol. 29, 3103.
Pavet, V., Beyrath, J., Pardin, C., Morizot, A., Lechner, M.C., Briand, J.P.,
Wendland, M., Maison, W., Fournel, S., Micheau, O., et al. (2010).
Multivalent DR5 peptides activate the TRAIL death pathway and exert tumor-
icidal activity. Cancer Res. 70, 1101–1110.
Paz-Ares, L., Balint, B., de Boer, R.H., van Meerbeeck, J.P., Wierzbicki, R., De
Souza, P., Galimi, F., Haddad, V., Sabin, T., Hei, Y.J., et al. (2013). A random-
ized phase 2 study of paclitaxel and carboplatin with or without conatumumab
for first-line treatment of advanced non-small-cell lung cancer. J. Thorac.
Oncol. 8, 329–337.
Pfizenmaier, K., and Szymkowski, D.E. (2011). Workshop Summary:
Introduction to Rational Design of New Means for Therapeutic Modulation of
Function of the TNF Family. Adv. Exp. Med. Biol. 691, 487–491.
Pitti, R.M., Marsters, S.A., Ruppert, S., Donahue, C.J., Moore, A., and
Ashkenazi, A. (1996). Induction of apoptosis by Apo-2 ligand, a new member
of the tumor necrosis factor cytokine family. J. Biol. Chem. 271, 12687–12690.
Saphire, E.O., Parren, P.W., Pantophlet, R., Zwick, M.B., Morris, G.M., Rudd,
P.M., Dwek, R.A., Stanfield, R.L., Burton, D.R., and Wilson, I.A. (2001). Crystal
structure of a neutralizing human IGG against HIV-1: a template for vaccine
design. Science 293, 1155–1159.
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 recep-
tor clustering. Nature 457, 1019–1022.
Sheridan, J.P., Marsters, S.A., Pitti, R.M., Gurney, A., Skubatch, M., Baldwin,
D., Ramakrishnan, L., Gray, C.L., Baker, K., Wood,W.I., et al. (1997). Control of
TRAIL-induced apoptosis by a family of signaling and decoy receptors.
Science 277, 818–821.
Soria, J.C., Smit, E., Khayat, D., Besse, B., Yang, X., Hsu, C.P., Reese, D.,
Wiezorek, J., and Blackhall, F. (2010). Phase 1b study of dulanermin (recombi-
nant human Apo2L/TRAIL) in combination with paclitaxel, carboplatin, and
bevacizumab in patients with advanced non-squamous non-small-cell lung
cancer. J. Clin. Oncol. 28, 1527–1533.
Soria, J.C., Mark, Z., Zatloukal, P., Szima, B., Albert, I., Juhasz, E., Pujol, J.L.,
Kozielski, J., Baker, N., Smethurst, D., et al. (2011). Randomized phase II study
of dulanermin in combination with paclitaxel, carboplatin, and bevacizumab in
advanced non-small-cell lung cancer. J. Clin. Oncol. 29, 4442–4451.
Suda, T., Hashimoto, H., Tanaka, M., Ochi, T., and Nagata, S. (1997).
Membrane Fas ligand kills human peripheral blood T lymphocytes, and soluble
Fas ligand blocks the killing. J. Exp. Med. 186, 2045–2050.
Swers, J.S., Grinberg, L., Wang, L., Feng, H., Lekstrom, K., Carrasco, R., Xiao,
Z., Inigo, I., Leow, C.C., Wu, H., et al. (2013). Multivalent scaffold proteins as
superagonists of TRAIL receptor 2-induced apoptosis. Mol. Cancer Ther. 12,
1235–1244.
Takeda, K., Yamaguchi, N., Akiba, H., Kojima, Y., Hayakawa, Y., Tanner, J.E.,
Sayers, T.J., Seki, N., Okumura, K., Yagita, H., and Smyth, M.J. (2004).
Induction of tumor-specific T cell immunity by anti-DR5 antibody therapy.
J. Exp. Med. 199, 437–448.
Takeda, K., Stagg, J., Yagita, H., Okumura, K., and Smyth, M.J. (2007).
Targeting death-inducing receptors in cancer therapy. Oncogene 26, 3745–
3757.
Valley, C.C., Lewis, A.K., Mudaliar, D.J., Perlmutter, J.D., Braun, A.R., Karim,
C.B., Thomas, D.D., Brody, J.R., and Sachs, J.N. (2012). Tumor necrosis
factor-related apoptosis-inducing ligand (TRAIL) induces death receptor 5
networks that are highly organized. J. Biol. Chem. 287, 21265–21278.
Wagner, K.W., Punnoose, E.A., Januario, T., Lawrence, D.A., Pitti, R.M.,
Lancaster, K., Lee, D., von Goetz, M., Yee, S.F., Totpal, K., et al. (2007).
Death-receptor O-glycosylation controls tumor-cell sensitivity to the proapo-
ptotic ligand Apo2L/TRAIL. Nat. Med. 13, 1070–1077.
Wajant, H., Moosmayer, D., Wuest, T., Bartke, T., Gerlach, E., Schonherr, U.,
Peters, N., Scheurich, P., and Pfizenmaier, K. (2001). Differential activation of
TRAIL-R1 and -2 by soluble and membrane TRAIL allows selective surface
antigen-directed activation of TRAIL-R2 by a soluble TRAIL derivative.
Oncogene 20, 4101–4106.
Walczak, H., Degli-Esposti, M.A., Johnson, R.S., Smolak, P.J., Waugh, J.Y.,
Boiani, N., Timour, M.S., Gerhart, M.J., Schooley, K.A., Smith, C.A., et al.
(1997). TRAIL-R2: a novel apoptosis-mediating receptor for TRAIL. EMBO J.
16, 5386–5397.
Walczak, H.,Miller, R.E., Ariail, K., Gliniak, B., Griffith, T.S., Kubin,M., Chin,W.,
Jones, J., Woodward, A., Le, T., et al. (1999). Tumoricidal activity of tumor
necrosis factor-related apoptosis-inducing ligand in vivo. Nat. Med. 5,
157–163.
Wiezorek, J., Holland, P., and Graves, J. (2010). Death receptor agonists as a
targeted therapy for cancer. Clin. Cancer Res. 16, 1701–1708.
Wiley, S.R., Schooley, K., Smolak, P.J., Din, W.S., Huang, C.P., Nicholl, J.K.,
Sutherland, G.R., Smith, T.D., Rauch, C., Smith, C.A., et al. (1995).
Identification and characterization of a new member of the TNF family that
induces apoptosis. Immunity 3, 673–682.
Wilson, N.S., Yang, B., Yang, A., Loeser, S., Marsters, S., Lawrence, D., Li, Y.,
Pitti, R., Totpal, K., Yee, S., et al. (2011). An Fcg receptor-dependent mecha-
nism drives antibody-mediated target-receptor signaling in cancer cells.
Cancer Cell 19, 101–113.
Wilson, N.S., Yang, A., Yang, B., Couto, S., Stern, H., Gogineni, A., Pitti, R.,
Marsters, S., Weimer, R.M., Singh, M., and Ashkenazi, A. (2012).
Proapoptotic activation of death receptor 5 on tumor endothelial cells disrupts
the vasculature and reduces tumor growth. Cancer Cell 22, 80–90.
Xiao, S., Jodo, S., Sung, S.S., Marshak-Rothstein, A., and Ju, S.T. (2002). A
novel signaling mechanism for soluble CD95 ligand. Synergy with anti-CD95
monoclonal antibodies for apoptosis and NF-kappaB nuclear translocation.
J. Biol. Chem. 277, 50907–50913.
Yang, A., Wilson, N.S., and Ashkenazi, A. (2010). Proapoptotic DR4 and DR5
signaling in cancer cells: toward clinical translation. Curr. Opin. Cell Biol. 22,
837–844.
Yin, Q., Lin, S.C., Lamothe, B., Lu, M., Lo, Y.C., Hura, G., Zheng, L., Rich, R.L.,
Campos, A.D., Myszka, D.G., et al. (2009). E2 interaction and dimerization in
the crystal structure of TRAF6. Nat. Struct. Mol. Biol. 16, 658–666.
Cancer Cell 26, 177–189, August 11, 2014 ª2014 Elsevier Inc. 189