Inhibition of proinflammatory and innate immune signaling pathways by a cytomegalovirus RIP1-interacting protein.

Inhibition of proinflammatory and innate immune
signaling pathways by a cytomegalovirus
RIP1-interacting protein
Claudia Mack*, Albert Sickmann†, David Lembo‡, and Wolfram Brune*§
*Division of Viral Infections, Robert Koch Institute, Nordufer 20, 13353 Berlin, Germany; †Rudolf Virchow Center for Experimental Biomedicine, University of
Wu¨rzburg, Versbacher Strasse 9, 97078 Wu¨rzburg, Germany; and ‡Department of Clinical and Biological Sciences, University of Turin, Regione Gonzole 10,
10043 Orbassano, Turin, Italy
Communicated by Thomas E. Shenk, Princeton University, Princeton, NJ, January 7, 2008 (received for review December 10, 2007)
TNF� is an important cytokine in antimicrobial immunity and
inflammation. The receptor-interacting protein RIP1 is an essential
component of the TNF receptor 1 signaling pathway that mediates
the activation of NF-�B, MAPKs, and programmed cell death. It also
transduces signals derived from Toll-like receptors and intracellular
sensors of DNA damage and double-stranded RNA. Here, we show
that the murine CMV M45 protein binds to RIP1 and inhibits
TNF�-induced activation of NF-�B, p38 MAPK, and caspase-inde-
pendent cell death. M45 also inhibited NF-�B activation upon
stimulation of Toll-like receptor 3 and ubiquitination of RIP1, which
is required for NF-�B activation. Hence, M45 functions as a viral
inhibitor of RIP1-mediated signaling. The results presented here
reveal a mechanism of viral immune subversion and demonstrate
how a viral protein can simultaneously block proinflammatory and
innate immune signaling pathways by interacting with a central
mediator molecule.
apoptosis � necrosis � herpesvirus � ribonucleotide reductase
Antiviral innate immune responses are triggered by receptorsand sensors that recognize pathogen-, damage-, or stress-
associated molecular patterns (1). These receptors can be lo-
cated at endosomal or cell surface membranes, as is the case for
the Toll-like receptors (TLRs). Other receptors, such as the
double-stranded RNA (dsRNA)-activated helicases RIG-I and
Mda5, sense the presence of potentially dangerous molecular
patterns inside of the cell. The receptors initiate specific signal-
ing cascades that lead to the activation of transcription factors,
(such as NF-�B and IFN regulatory factors) and MAPKs, or the
initiation of programmed cell death (PCD). Very similar re-
sponses are triggered by proinflammatory cytokines such as
TNF�, which also play important roles in controlling viral
infections.
The receptor-interacting protein RIP1 (also called RIP) is
located at the intersection of several signaling pathways [sup-
porting information (SI) Fig. 7]. It integrates signals from
membrane-bound receptors and intracellular stress sensors (re-
viewed in refs. 2 and 3). RIP1 has been investigated extensively
because of its crucial role in the TNF receptor (TNFR)1
signaling pathway (4). Stimulation with TNF� initially induces
the recruitment of RIP1, the TNFR-associated factor (TRAF)2,
and the TNFR-associated death domain to the plasma mem-
brane (5, 6). The subsequent ubiquitination of RIP1 by TRAF2
(7, 8) is required for activation of I�B kinase and NF-�B (9).
RIP1 also activates the MAPKs p38 and ERK and participates
in the activation of JNK (10, 11). In addition, RIP1 mediates
NF-�B activation upon stimulation of TLR3 and TLR4 via the
TIR domain-containing adaptor-inducing INF-� (TRIF) (12,
13). RIP1 also transmits signals derived from DNA-damage
sensors (14) and from intracellular sensors of dsRNA (15, 16).
TNFR1 and other death receptors can trigger apoptosis by
inducing the formation of a complex containing the Fas-
associated death domain and procaspase-8, in which the latter is
activated autocatalytically (5). However, when caspase-
dependent cell death is blocked, an alternative, caspase-
independent pathway to PCD can be activated, and this pathway
depends on RIP1 (17, 18). The various downstream effects of
RIP1, particularly the induction of cell death and activation of
proinflammatory genes through NF-�B, can obviously be det-
rimental to viral replication and spread, and, therefore, they are
potential targets for viral countermeasures.
CMVs, prototypes of the �-herpesviruses, are known to
interfere with many innate defense pathways. In only a few cases,
the responsible viral gene products have been identified. For
instance, the murine CMV (MCMV) M27 protein and the
human CMV (HCMV) IE1 protein block the induction of
IFN-responsive effector genes by interacting with STAT1 and/or
STAT2 (19, 20). Both viruses inhibit apoptosis upon stimulation
of death receptors by a viral inhibitor of caspase-8 activation
(vICA), encoded by the viral genes M36 and UL36, respectively
(21, 22). Death receptor- as well as stress-induced PCD is also
inhibited by the mitochondrial inhibitor of apoptosis encoded by
the HCMVUL37x1 and the MCMVm38.5 gene (23, 24) (SI Fig.
7). It has also been shown that TNF�-induced NF-�B activation
is blocked by CMVs (25–27), but the molecular mechanisms and
the viral genes involved have remained unknown.
In this study, we show that the MCMV protein M45 binds to
RIP1 and inhibits TNF�-induced activation of NF-�B and p38
MAPK, as well as the induction of caspase-independent cell
death. Moreover, M45 also blocks the RIP1-mediated activation
of NF-�B in response to TLR3 stimulation. These results show
that M45 functions as a viral inhibitor of RIP1-mediated
signaling.
Results
Inhibition of TNF�-Induced Cell Death by M45. In previous work, we
have shown that M45 is required for blocking PCD induced by
MCMV infection itself and that endothelial cells and macro-
phages are particularly sensitive to virus-induced cell death (28).
In the present study, we also observed that fibroblasts infected
with an M45 deletion mutant (�M45) were highly sensitive to
TNF�-induced cell death but resistant to staurosporine (STS)-
induced apoptosis (Fig. 1a). Similarly, �M45-infected cells were
sensitive to Fas-mediated but resistant to actinomycin D-induced
PCD (data not shown). By contrast, cells infected with wild-type
(wt) MCMV or an M45 revertant virus (RM45) were protected
from these cell death-inducing stimuli (Fig. 1a). The finding that
Author contributions: C.M. and W.B. designed research; C.M. and A.S. performed research;
D.L. contributed new reagents/analytic tools; C.M., A.S., and W.B. analyzed data; and C.M.,
D.L., and W.B. wrote the paper.
The authors declare no conflict of interest.
§To whom correspondence should be addressed. E-mail: BruneW@rki.de.
This article contains supporting information online at www.pnas.org/cgi/content/full/
0800168105/DC1.
© 2008 by The National Academy of Sciences of the USA
3094–3099 � PNAS � February 26, 2008 � vol. 105 � no. 8 www.pnas.org�cgi�doi�10.1073�pnas.0800168105

M45 is required to inhibit TNFR1- or Fas-induced cell death was
surprising, because MCMV M36/vICA should inhibit caspase-8
activation upon death receptor stimulation (22). Fibroblasts
infected with an M36 deletion mutant (�M36) were, indeed,
sensitive to TNF�-induced apoptosis, and this was blocked by the
caspase inhibitor z-VAD-fmk (Fig. 1b). However, TNF�-
induced cell death of �M45-infected cells was not blocked by
z-VAD-fmk (Fig. 1b), indicating that M45 is necessary to inhibit
TNF�-induced caspase-independent cell death.
M45 Interacts with RIP1. To determine the molecular mechanism
of action of M45, we tried to identify cellular and viral interac-
tion partners of M45 by affinity purification. A recombinant
MCMV was constructed that expresses an M45 protein with a
tobacco etch virus (TEV) protease cleavage site and an HA tag
at its C terminus (MCMV M45–TEV–HA). Lysates of cells
infected with M45–TEV–HA or wt MCMV were applied to
anti-HA affinity purification columns. M45 and proteins bound
to it were specifically eluted by cleavage with TEV protease.
Possible interaction partners were identified by mass spectrom-
etry. By this approach, the TNFR-interacting protein RIP1 was
identified as an interaction partner of M45 (SI Fig. 8). The
M45–RIP1 interaction was confirmed by coimmunoprecipita-
tion by using virus-infected cells (Fig. 2 a and b), as well as cells
transfected with an M45 expression plasmid (Fig. 2c). This
showed that no other MCMV proteins are required for the
binding of M45 to RIP1.
Upon TNF� stimulation, RIP1 and TRAF2 are recruited to
TNFR1 to form a signaling complex (5). Even though TRAF2
and TNFR1 were not found in our affinity purification screen,
we wanted to rule out the possibility that M45 binds to RIP1 only
indirectly via TRAF2 or TNFR1. To do this, TNFR1- and
TRAF2/5-deficient fibroblasts were infected with MCMV and
used for coimmunoprecipitation experiments. As shown in Fig.
2d, RIP1 was coprecipitated with M45 in the absence of TNFR1
or TRAF2 and -5, indicating that M45 binding to RIP1 does not
require TNFR1 or TRAF2/5.
M45 Inhibits TNF�-Induced NF-�B Activation. Because one of the
main functions of RIP1 is signal transduction from TNFR1 to
a
lysates IP
α
AH
α
14
m
α
54
M
α
AH
75
kDa
wt
MCMV
A
H54
M
tw AH54
M
WB:
αRIP1
α
galF
α
83p
α
1PI
R
α
1PIR
M45HA kco
m
IP
b
100
kDa
WB:
αHA
rotcev
54
M
341
m
IP: αHA, WB: αRIP1
lysates, αRIP1
c
tw AH54
M
tw AH54
M
tw AH54
M
tw AH54
M
lysates, WB: αRIP1
wt 3T3 rip1-/-traf2/5-/- tnfr1-/-
d
lysates, αHA
rotcev
54
M
341
m
IP: αHA, WB: αRIP1
Fig. 2. M45 binds to RIP1. (a) Lysates of cells infected with wt MCMV or an
HA-tagged virus (M45HA) were used for immunoprecipitation with the indi-
cated antibodies. RIP1 was detected by Western blot as a protein coprecipi-
tating with M45. (b) HA-tagged M45 was coprecipitated with RIP1 in lysates of
cells infected with MCMV–M45HA. (c) NIH 3T3 cells were transfected with
plasmids expressing HA-tagged M45 or m143 (a different MCMV protein),
respectively. RIP1 was coprecipitated with M45 but not with m143. (d) 3T3
fibroblasts derived from wt and knockout mice were infected with wt or
M45HA-expressing MCMV. RIP1 was coprecipitated with M45HA in all but
rip1�/� knockout cells.
37
kDa
150
37
a
TNFα
IκBα
–
+
–
+
–
+
–
+
M45
actin
mock ∆M45 RM45wt
b
150
kDa
75
37
rip1
-/- MSCVrip1
vector GFP M45vector GFP M45
–
+
–
+
–
+
–
+
–
+
–
+
rip1
-/- MSCV
RIP1
M45
IκBα
TNFα
c
fo
noitcudni
dlof
ytivitca
esareficul 5
10
15
20
25
30
vector M45 A20 m41 GFP
Fig. 3. M45 inhibits TNF�-induced NF-�B activation. (a) NIH 3T3 cells were
infected with wt or mutant MCMV for 24 h and treated with TNF� for 15 min.
M45-expressing viruses inhibited the degradation of I�B�. (b) rip1�/� knock-
out fibroblasts were stably transduced with a RIP1-expressing or an empty
MSCV retrovirus. Cells were also transduced with retroviral vectors expressing
M45 or GFP (as control). TNF�-induced I�B� degradation was blocked in all
RIP1-deficient cells and in RIP1-positive cells expressing M45. (c) HEK 293 cells
were transfected with an NF-�B-dependent luciferase reporter plasmid and
different expression plasmids. Luciferase activity was measured 12 h after
TNF� stimulation and is shown as fold induction compared with transfected
cells without TNF�. Luciferase induction was inhibited in cells expressing M45
or the cellular RIP1 inhibitor A20 but not in cells transfected with control
plasmids expressing irrelevant proteins (m41 or GFP) or empty pcDNA3 vector.
20
40
60
80
100 STS TNFα+CHX
ytilibaiv
%
20
40
60
80 DMSO z-VAD-fmk
kco
m
tw 54
M
R ∆
54
M ∆
63
M
kco
m
tw 54
M
R ∆
54
M ∆
63
M
kco
m
tw
∆
54
M
54
M
R
kco
m
tw
∆
54
M
54
M
R
a b
Fig. 1. M45 is required to block TNF�-induced caspase-independent cell
death. (a) 10.1 fibroblasts were infected with MCMVs as indicated and treated
6 h later with STS or TNF� plus CHX. Viability was measured 20 h later relative
to cells treated with solvent only (for STS) or CHX only (for TNF�). Results that
were similar to those with TNF�plus CHX were obtained with TNF�alone, with
the exception of mock-infected cells, which did not die from TNF� alone (data
not shown). (b) Cells were infected and treated with TNF� plus CHX and the
caspase inhibitor z-VAD-fmk or DMSO. Cells infected with �M45 or �M36
were sensitive to TNF�-induced cell death, which was blocked by z-VAD-fmk
only in the case of the �M36 virus.
Mack et al. PNAS � February 26, 2008 � vol. 105 � no. 8 � 3095
M
IC
RO
BI
O
LO
G
Y

NF-�B activation, we analyzed whether M45 interferes with this
pathway. Ubiquitination of RIP1 leads to the recruitment and
activation of the I�B kinase (IKK) complex, which phosphory-
lates I�B�, the inhibitor of NF-�B (9). The subsequent degra-
dation of I�B� leads to the release of NF-�B. We found that
TNF�-induced I�B� degradation was blocked in cells infected
with wt MCMV or the RM45 revertant virus but not in cells
infected with the �M45 virus (Fig. 3a).
To test whetherM45 by itself can inhibit I�B� degradation, we
expressed M45 or GFP in fibroblasts. RIP1-deficient cells and
derivatives of these cells, in which the rip1 gene was reintro-
duced, were used to confirm that this process is RIP1-dependent.
Fig. 3b shows that M45 inhibited I�B� degradation in RIP1-
expressing fibroblasts.
Although the analysis of I�B� degradation is an established
assay for NF-�B activation, we used an independent test system
to confirm the results. An NF-�B-dependent luciferase reporter
plasmid was transfected together withM45-expressing or control
plasmids into HEK 293 cells. Upon stimulation with TNF�,
luciferase expression was induced in cells transfected with
control plasmids but was blocked in cells expressing M45 or the
cellular RIP1 inhibitor A20 (Fig. 3c) (29).
M45 Inhibits Caspase-Independent Cell Death. Stimulation of
TNFR1 can activate caspase-8 and induce apoptosis (5) or lead
to caspase-independent PCD mediated by RIP1 (17, 18). The
latter pathway is activated when caspase-8-dependent apoptosis
is blocked and has, therefore, been termed a ‘‘backup’’ pathway
to cell death. During MCMV infection, caspase-8 activation is
blocked by M36. Hence the virus needs M45 to inhibit TNF�-
induced caspase-independent cell death (Fig. 1). Caspase-
independent PCD can be induced by treatment of susceptible
cells with TNF� and caspase inhibitors (17). Thus we tested
whether M45 by itself can inhibit the caspase-independent cell
death pathway. Fig. 4a shows that SVEC4–10 endothelial cells
died rapidly upon TNF� stimulation when the caspase-8-
dependent pathway was blocked by a pan-caspase (z-VAD-fmk)
or a caspase-8-specific inhibitor (z-IETD-fmk). By contrast,
M45-expressing SVEC4–10 cells were protected. Similar results
were obtained with L929 fibrosarcoma cells (Fig. 4b), which have
been used by other investigators for the analysis of RIP1-
dependent PCD (17) and 10.1 fibroblasts (data not shown). We
observed that different cell lines vary in their sensitivity to this
cell death pathway, with SVEC4–10 cells being highly sensitive
and NIH 3T3 cells rather insensitive.
RIP1 is essential for TNF� signaling to p38 MAPK (10). To
find out whether M45 inhibits activation of p38, cells were
infected with wt and mutant MCMVs, and p38 phosphorylation
was assayed at different times after TNF� addition. TNF�
stimulation increased the amount of phosphorylated p38 in
mock-infected cells and cells infected with �M45 but not in cells
infected with the wt MCMV or RM45, both of which express
M45 (Fig. 4c). This shows that M45 inhibits p38 activation in
infected cells. It is noteworthy that MCMV-infected cells showed
a slightly increased baseline level of phospho-p38 (i.e., before
TNF� stimulation) as compared with mock-infected cells.
The N-Terminal Domain of M45 Is Dispensable for Inhibition of RIP1.
M45 shows a high sequence similarity to the large subunit (R1)
of ribonucleotide reductases (RNR) within its C-terminal part
(28). However, a number of amino acid residues essential for
RNR activity are not conserved in M45, and the protein is
catalytically inactive (30). M45 also contains a large N-terminal
domain of unknown function. To test whether parts of the
protein are dispensable for its function, we constructed trunca-
tion mutants of M45 (Fig. 5a). Surprisingly, almost the entire
N-terminal domain could be deleted without losing the ability of
the protein to interact with RIP1 (Fig. 5b) and inhibit TNF�-
induced NF-�B activation (Fig. 5c). By contrast, a relatively
moderate truncation at the C terminus abrogated both, binding
to RIP1 and the ability to block NF-�B activation (Fig. 5 b and
c). Thus, the R1 homology domain, but not the N-terminal
tw ∆
54
M
54
M
R
kco
m
tw ∆
54
M
54
M
R
kco
m
tw ∆
54
M
54
M
R
kco
m
tw ∆
54
M
54
M
R
kco
m
0 min 5 min 12 min 20 minTNFα
P-p38
p38
c
a b
20
40
60
80
100
TNFα+
DMSO
TNFα+
zVAD
TNFα+
zIETD
ytilibaiv
%
kco
m
54
M
PF
G
rotcev
kco
m
54
M
PF
G
rotcev
kco
m
54
M
PF
G
rotcev
TNFα+
DMSO
TNFα+
zVAD
20
40
60
80
100
54
M
PF
G
54
M
PF
G
Fig. 4. M45 inhibits caspase-independent cell death and activation of p38
MAPK. (a) SVEC4–10 endothelial cells were transduced with retroviral vectors
expressing M45, GFP, or nothing. Cells were treated with TNF� (without CHX)
and z-VAD-fmk, z-IETD-fmk, or DMSO. Viability was measured after 24 h. M45
blocked the induction of caspase-independent cell death. (b) Similar results
were obtained with L929 fibrosarcoma cells. (c) Fibroblasts were infected with
wt or mutant MCMV and treated with TNF� for the indicated periods of time.
Lower amounts of phosphorylated p38 MAPK (P-p38) were detected in cells
infected with M45-expressing viruses (wt and RM45).
b c
M45 (1174 aa)
Nt1 ∆1-54
Nt2 ∆1-280
Nt3 ∆1-350
Ct ∆977-1174
R1 homology domain
a
54
M
2t
N
1t
N
3t
N
t
C cev
75
kDa
IP: αHA, WB: αRIP1
lysates, WB: αHA
75
kDa
100 5
10
15
20
vec M45 A20 Nt1 Nt2 Nt3 Ct
fo
noitcudni
dlof
ytivitca
esareficul
Fig. 5. Large parts of the N-terminal, but not of the C-terminal, domain of
M45 are dispensable for interaction with RIP1 and activation of NF-�B. (a)
Schematic representation of M45 and truncation mutants. The unique N-
terminal domain is shown as an open box, and the C-terminal HA tag is shown
in black. (b) NIH 3T3 cells were transfected with plasmids expressing full-
length or truncated M45, and HA-tagged proteins were immunoprecipitated.
RIP1 was coprecipitated with Nt1, -2, and -3 but not with Ct. (c) Luciferase
activity was measured in transfected HEK 293 cells as described in the legend
of Fig. 3c. The N-terminal, but not the C-terminal, truncation mutants inhib-
ited TNF�-induced NF-�B activation.
3096 � www.pnas.org�cgi�doi�10.1073�pnas.0800168105 Mack et al.

domain, is essential for the function of M45 as inhibitor of
RIP1-mediated signaling.
M45 Inhibits TLR3-Mediated NF-�B Activation. Recent studies have
demonstrated that TLR3 and TLR4 can activate NF-�B by
signaling through the adaptor proteins TRIF and RIP1 (12, 13).
Because TLR4 (but not TLR3) can activate NF-�B also inde-
pendently of RIP1 via MyD88 (13), we tested whether M45 was
also able to inhibit TLR3-induced NF-�B activation. Macro-
phages, which naturally express TLR3, are notoriously difficult
to transfect. Therefore, we transfected HEK 293 cells with a
TLR3 expression plasmid and an NF-�B-dependent luciferase
reporter plasmid and stimulated the cells with poly(I:C) as
performed by others previously (31). In 293 cells, NF-�B acti-
vation upon poly(I:C) addition depends on TLR3, because these
cells do not respond to poly(I:C) in the absence of TLR3 (ref. 31
and unpublished results). As shown in Fig. 6a, TLR3 stimulation
induced luciferase expression in cells cotransfected with control
plasmids expressing an inactive truncation mutant of M45 or an
irrelevant MCMV protein. By contrast, M45 and the cellular
RIP1 inhibitor A20 inhibited luciferase induction upon the
addition of poly(I:C) (Fig. 6a), indicating that M45 also inhibits
RIP1-mediated TLR3 signaling.
M45 Inhibits Ubiquitination of RIP1. Upon TNF� stimulation, RIP1
is ubiquitinated and activated by TRAF2 (7, 8, 32). Lysine 63
(K63)-linked ubiquitination of RIP1 is required for recruitment
of IKK�/NEMO and activation of I�B kinase, whereas K48-
linked ubiquitination causes proteasomal degradation of RIP1
(3, 9, 33). Overexpression of RIP1 leads to its ubiquitination and
activation of NF-�B, and this process is inhibited by the cellular
protein A20. A20 counteracts RIP1 activation by removing
K63-linked and attaching K48-linked ubiquitin molecules (33).
M45 shows no obvious sequence similarity to A20 or other
deubiquitinating enzymes, but we speculated that the strong
binding of M45 to RIP1 could block its ubiquitination. To test
this hypothesis, we overexpressed Myc-tagged RIP1 (MycRIP)
in HEK 293 cells by transient transfection. An A20 expression
vector, an empty vector plasmid, or increasing amounts of a
plasmid expressing M45, as well as an expression vector for
HA-tagged ubiquitin (HA–Ub), were cotransfected. As shown in
Fig. 6 b and c, A20 and M45 inhibited the ubiquitination of
MycRIP. HEK 293 cells were also transfected with similar sets
of plasmids, but an NF-�B-dependent luciferase reporter plas-
mid was used instead of HA–Ub. Consistent with the results in
Fig. 6b, MycRIP overexpression activated NF-�B, and A20, as
well as M45, markedly decreased its activation (Fig. 6d). These
results show that M45 inhibits ubiquitination of RIP1.
Discussion
Numerous virus recognition systems and cytokine signaling
pathways activate a limited number of effector systems. The
cellular kinase RIP1 is at the converging point of several
pathways (2, 3). Its central role should make it an ideal target for
inhibition by a virus, because several signaling pathways could be
blocked at the same time (SI Fig. 7). In this study, we demon-
strate that the MCMV M45 protein binds to RIP1, inhibits its
activation by ubiquitination, and blocks the TNF�-induced ac-
tivation of NF-�B, p38 MAPK, and caspase-independent PCD.
We also show that M45 inhibits TLR3-induced NF-�B activation.
Previously, we have reported that M45 is required to inhibit
PCD induced by the virus itself (28). The molecular mechanism
of action of M45 described here offers an explanation for the
ability of M45 to block cell death induced by both external and
internal stimuli. RIP1 transduces signals from intracellular sen-
sors of DNA damage (14) and dsRNA (15, 16), and it is known
that CMV infection can trigger dsRNA-activated pathways (34)
and induce a DNA-damage response (35). Hence, virus repli-
cation could induce cell death via internal sensors that signal
through RIP1. The apparent differences in sensitivity to RIP1-
mediated cell death described here correlate with our previous
observation that SVEC4–10 endothelial cells are much more
sensitive to cell death induced by infection with an M45 mutant
virus than NIH 3T3 fibroblasts (28). A recent report has
suggested that the sensitivity of cells to caspase-independent cell
death may depend on the level of RIP1 expression and the ability
of cells to secrete TNF� (36). Like murine cells, human cells also
appear to differ in their sensitivity to caspase-independent PCD
(17, 18).
TLR3 and TLR4 are pattern-recognition receptors activated
by dsRNA or LPS, respectively. The role of TLR3 in MCMV
infection is under debate (37, 38), but the fact that RNA has been
75
100
kDa
150
c
b
setasyl
IP:
αMyc
WB:
αHA
75
100
kDa
MycRIP
HA-Ub
F-A20
M45
RIP1
Flag
M45
20
40
60
80
100
120
vector A20 M45HA M45
fo
noitcudni
dlof
ytivitca
esareficul
150
+ + + + +
–
+
+ + + + + +
–
+++
– – – – – –
– – + ++ +++ – –
WB:
αM45
+
αFlag
IP: Myc
d
fo
noitcudni
dlof
ytivitca
esareficul 2
4
6
8
10
12
14
M45 Ct A20m41
a
Fig. 6. M45 inhibits TLR3-induced NF-�B activation and ubiquitination of
RIP1. (a) HEK 293 cells were transfected with an NF-�B-dependent luciferase
reporter plasmid and expression plasmids for TLR3 and the indicated proteins.
Luciferase activity was measured 6 h after poly(I:C) addition and is shown as a
fold induction as compared with transfected cells without poly(I:C). Induction
of luciferase activity was inhibited by M45 and A20 but not by an irrelevant
MCMV protein (m41) or a truncated M45 (Ct). (b) (Upper) Cells were trans-
fected with plasmids as indicated. The total amount of plasmid for each
transfection was normalized with empty pcDNA3 vector. MycRIP was precip-
itated with an anti-Myc antibody, and polyubiquitinated RIP1 was detected as
a high-molecular-weight ladder by using an anti-HA antibody (lane 2). Ubiq-
uitination was reduced in cells expressing A20 or M45. (Lower) The same blot
after stripping and incubation with anti-M45 and anti-Flag antibodies. Com-
paring the two blots reveals that the anti-HA antibody also detects coprecipi-
tated M45 and A20, indicating that A20 and M45 are also ubiquitinated. (c)
Control Western blots show MycRIP, Flag-A20, and M45 in the cell lysates used
for immunoprecipitation. (d) Overexpression of MycRIP in HEK 293 cells
activates NF-�B, as measured with a luciferase reporter assay. NF-�B activation
is inhibited by M45 and A20.
Mack et al. PNAS � February 26, 2008 � vol. 105 � no. 8 � 3097
M
IC
RO
BI
O
LO
G
Y

detected in CMV particles (39) and that dsRNA-binding pro-
teins are incorporated into the tegument (40) suggests that
dsRNA may stimulate TLR3 or cytoplasmic dsRNA sensors.
M45 is also part of the viral tegument (30) and could inhibit
RIP1-mediated signaling immediately if delivered in sufficient
amounts to the infected cell. Whether TLR4 plays a role during
MCMV infection is questionable, because a previous study has
shown that TLR2, but not TLR4, is stimulated by HCMV (41).
Considering the central role of RIP1 in different pathways, one
would expect that an M45 mutant MCMV should be severely
attenuated in its natural host. In a previous study, we have shown
that M45 mutants are completely avirulent in SCID mice, even
though these mice are highly susceptible to MCMV infection (30).
Unfortunately, it is not possible to test whether RIP1 deficiency can
reverse the attenuated phenotype of an M45-deficient virus, be-
cause rip1 knockout mice die within the first 3 days of life (4).
Stimulation of death receptors can induce apoptosis by activation
of caspase-8 (5). To inhibit this pathway, many viruses, including
CMVs, �-herpesviruses, and poxviruses, express caspase-8 inhibi-
tors (21, 22, 42, 43). Our results show that the mere inhibition of
caspase-8 can render infected cells sensitive to TNF�-induced
caspase-independent PCD and that an additional inhibitor is re-
quired to block this backup pathway to cell death. Hence, it is likely
that other viruses that block caspase-8 also inhibit this RIP1-
dependent pathway, possibly in a similar way like M45.
The ability of M45 to inhibit both NF-�B activation and
caspase-independent cell death may seem paradoxical, because
NF-�B can induce the expression of antiapoptotic proteins (5).
However, a recent study has shown that caspase-independent
PCD is not affected by NF-�B activation (44), indicating that the
function of M45 is not as conflicting as it appears.
Unlike �- and �-herpesviruses, �-herpesviruses seem to have
abandoned the strategy of supplying enzymes required for the
biosynthesis of DNA precursors. Genes for a thymidine kinase,
a thymidylate synthase, and for the small RNR subunit are
absent, and those for the large RNR subunit and dUTPase
encode catalytically inactive proteins. The M45 gene became a
paradigm of the latter case. The ability of MCMV to induce the
cellular RNR allowed M45 to mutate and lose a direct involve-
ment in ribonucleotide reduction. M45 apparently maintained or
gained a second function that is indispensable for viral replica-
tion in certain cells and dissemination in vivo (28, 30). This study
reveals the molecular mechanism of the function of M45 and
demonstrates how a viral protein can simultaneously block
innate immune and proinflammatory signaling pathways by
interacting with a central mediator molecule.
Materials and Methods
Cells. NIH 3T3 (ATCC CRL-1658) and 10.1 cells are immortalized mouse embry-
onic fibroblasts. L929 (ATCC CCL-1) and SVEC4–10 (CRL-2181) are murine
fibrosarcoma and endothelial cell lines. 3T3-like fibroblasts derived from rip1,
tnfr1, and traf2/traf5 knockout mice (4, 32) were a gift from M. Kelliher
(University of Massachusetts, Boston, MA). Human embryonic kidney (HEK)
293 cells were purchased from Invitrogen.
Plasmids and Transfections. The following expression plasmids were used:
pCAGGS-FlagA20 (LMBP plasmid collection, University of Ghent), pFlagCMV1-
huTLR3 (Addgene), pRK5-MycRIP (a gift from Z. G. Liu, National Institutes of
Health, Bethesda, MD), pHA–Ub (provided by M. Nevels, University of Regens-
burg, Germany), pRSV-�Gal (Promega), pTranslucent NF-�B (Panomics),
pcDNA-CrmA (43), pcDNA-m143HA (34), and pcDNA-m41 (45). The pcDNA-
M45HA and pcDNA-M45 plasmids were obtained by inserting the PCR-
amplified M45 (GenBank accession no. DQ978788) via KpnI and XbaI into
pcDNA3 (Invitrogen). For the truncation mutant Nt1, nucleotides 162 to 559 of
M45 were amplified by PCR and inserted between the KpnI and BamHI sites of
pcDNA-M45HA. Nt2 and Nt3 were generated by digesting this plasmid with
KpnI and HindIII or EcoRI, respectively, blunting, and religation. For the Ct
truncation mutant, pcDNA-M45HA was digested with XhoI and XbaI, and a
synthetic linker encoding an HA tag was inserted. Transient transfections were
performed by calcium phosphate precipitation or with Polyfect (Qiagen)
according to the recommendations of the manufacturer.
Retroviral Transduction. The murine rip1 cDNA (IMAGE clone 5721177) was
inserted into pMSCVpuro (Clontech). The M45HA sequence was inserted into
the PmlI site of pRetroEBNA to generate pRetroM45. The pRetroEBNA and
pRetroGFP plasmids were obtained from Tom Shenk (Princeton University,
Princeton, NJ). Production of retroviruses by using Phoenix A cells and trans-
duction of target cells was performed as described (45).
CMVs and Infection. MCMV–GFP and the �M45 deletion mutant have been
described (28, 45). The �M36 mutant was constructed essentially as described
(22), with the exception that a zeocin resistance gene was used. For the RM45
revertant virus, the M45HA gene was used to replace the nonessential genes
m02 to m06 in �M45 by using the pReplacer plasmid as described (46). The
M45–TEV–HA mutant was generated by tagging of the M45 sequence at the
3� end with a TEV protease cleavage site and an HA tag as described (45).
Viruses were grown on fibroblasts according to standard procedures. Unless
stated otherwise, cells were infected at a multiplicity of infection (moi) of
three median tissue culture infective doses (TCID50) per cell.
Affinity Purification. Cells (8 � 107 10.1) were infected at an moi of one TCID50
per cell with MCMV M45–TEV–HA or control virus. After 48 h, cells were lysed
(50 mM Tris, pH 7.5/150 mM NaCl/2.5% Nonidet P-40, complete protease
inhibitor mixture; Roche) and centrifuged for 1 h at 20,000 � g. Supernatants
were loaded onto anti-HA 3F10 affinity columns (Roche). After washing (20
mM Tris, pH 7.5/0.1 M NaCl/0.1 M EDTA/0.05% Tween-20), M45 and associated
proteins were eluted by digestion with 100 units of AcTEV protease (Invitro-
gen) for 1 h at room temperature. Eluted proteins were concentrated and
separated by SDS/PAGE. Silver-stained bands were excised and analyzed by
mass spectrometry as described (47).
Western Blots and Immunoprecipitation.Monoclonal mouse antibodies against
RIP1 (Clone 38, BD Biosciences), Flag (M2, Sigma), Myc (4A6, Upstate), and
�-actin (Ac-74, Sigma) were purchased as indicated. Polyclonal rabbit anti-
bodies against RIP1, I�B�, and p38 MAPK were from Santa Cruz Biotechnol-
ogy; anti-phospho-p38 was from Cell Signaling; and anti-HA was from Sigma.
The anti-M45 antibody has been described (30). A mouse monoclonal anti-
body against MCMV m41 was generated in our laboratory by Maren Syta.
HRP-coupled secondary antibodies were obtained from Cell Signaling or
Dako. For immunoprecipitation, 6�106 cells were lysed in lysis buffer with 1%
Nonidet P-40. The proteins of interest were precipitated overnight with 2.5�g
of antibody and protein A Sepharose at 4°C. Precipitates were washed five
times, eluted with sample buffer, and separated by SDS/PAGE. For the analysis
of RIP1 ubiquitination, cells lysates were harvested 26 h after transfection by
using radioimmunoprecipitation assay buffer [20 mM Tris�HCl (pH 7.5), 300
mM NaCl, 1% sodium deoxycholate, 1% Triton X-100, 0.1% SDS] and homog-
enized with QIAshredder columns (Qiagen). MycRIP was precipitated with an
anti-Myc antibody and protein G Sepharose.
For the analysis of I�B� degradation, infected or transduced cells were
stimulated for 15 min with 20 ng/ml recombinant murine TNF� (Promokine).
For p38 MAPK activation, cells were stimulated with 7.5 ng/ml TNF�. Proteins
were harvested with boiling SDS/PAGE sample buffer, sonicated, and analyzed
by Western blot.
Luciferase Assay. HEK 293 cells were transfected with pTranslucent–NF-�B
luciferase reporter vector, pRSV-�Gal control vector, pFlagCMV1-huTLR3 (for
analysis of TLR3 signaling), or pRK5-MycRIP (for RIP overexpression) and
expression vectors for the proteins to be analyzed. After 24 h, cells were
stimulated with 20 ng/ml TNF� for 12 h or with 100 �g/ml poly(I:C) for 6 h.
Luciferase and �-galactosidase activities were analyzed with a Dual-Light
reporter assay kit (Applied Biosystems). Experiments were done in triplicate,
and results are shown as fold induction of luciferase activity in induced
samples in comparison to noninduced samples with standard deviation.
Cell Viability Assay. Cell death was induced with 150 nM STS or 20 ng/ml TNF�
with or without 250 ng/ml cycloheximide (CHX). MCMV-infected cells were
treated 6 h after infection at an moi of 3 TCID50 per cell. Where indicated, TNF�
was added to the cells together with z-VAD-fmk, z-IETD-fmk (MBL Interna-
tional), or DMSO. Cell viability was measured by using a 3-(4,5-dimethyl-2-
thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide assay as described (46). The
results are shown as percentages of viability in comparison with control cells
with SD of four to eight parallel samples.
3098 � www.pnas.org�cgi�doi�10.1073�pnas.0800168105 Mack et al.

ACKNOWLEDGMENTS. We thank C. Hasselberg-Christoph and C. Winkler for
technical assistance; M. Kelliher for providing knockout cells; and A. Loewen-
dorf, C. Benedict, I. Mohr, and S. Voigt for a critical reading of the manuscript.
This work was supported by Deutsche Forschungsgemeinschaft Grant SFB421,
Project B14 (to W.B.) and by grants from Ministero dell’Istruzione,
dell’Universita´ e della Ricerca Grants ex 60% and PRIN 2005 (to D.L.).
1. Kawai T, Akira S (2006) Innate immune recognition of viral infection. Nat Immunol
7:131–137.
2. Meylan E, Tschopp J (2005) The RIP kinases: Crucial integrators of cellular stress. Trends
Biochem Sci 30:151–159.
3. Festjens N, Vanden Berghe T, Cornelis S, Vandenabeele P (2007) RIP1, a kinase on the
crossroads of a cell’s decision to live or die. Cell Death Differ 14:400–410.
4. Kelliher MA,et al. (1998) The death domain kinase RIP mediates the TNF-induced NF-�B
signal. Immunity 8:297–303.
5. Micheau O, Tschopp J (2003) Induction of TNF receptor I-mediated apoptosis via two
sequential signaling complexes. Cell 114:181–190.
6. Devin A, et al. (2000) The distinct roles of TRAF2 and RIP in IKK activation by TNF-R1:
TRAF2 recruits IKK to TNF-R1 while RIP mediates IKK activation. Immunity 12:419–429.
7. Legler DF, Micheau O, Doucey MA, Tschopp J, Bron C (2003) Recruitment of TNF
receptor 1 to lipid rafts is essential for TNF�-mediated NF-�B activation. Immunity
18:655–664.
8. Lee TH, Shank J, Cusson N, Kelliher MA (2004) The kinase activity of Rip1 is not required
for tumor necrosis factor-�-induced I�B kinase or p38 MAP kinase activation or for the
ubiquitination of Rip1 by Traf2. J Biol Chem 279:33185–33191.
9. Ea CK, Deng L, Xia ZP, Pineda G, Chen ZJ (2006) Activation of IKK by TNF� requires
site-specific ubiquitination of RIP1 and polyubiquitin binding by NEMO. Mol Cell
22:245–257.
10. Lee TH, et al. (2003) The death domain kinase RIP1 is essential for tumor necrosis factor
� signaling to p38 mitogen-activated protein kinase. Mol Cell Biol 23:8377–8385.
11. Devin A, Lin Y, Liu ZG (2003) The role of the death-domain kinase RIP in tumour-
necrosis-factor-induced activation of mitogen-activated protein kinases. EMBO Rep
4:623–627.
12. Meylan E, et al. (2004) RIP1 is an essential mediator of Toll-like receptor 3-induced
NF-�B activation. Nat Immunol 5:503–507.
13. Cusson-Hermance N, Khurana S, Lee TH, Fitzgerald KA, Kelliher MA (2005) Rip1
mediates the Trif-dependent toll-like receptor 3- and 4-induced NF-�B activation but
does not contribute to interferon regulatory factor 3 activation. J Biol Chem
280:36560–36566.
14. Janssens S, Tinel A, Lippens S, Tschopp J (2005) PIDD mediates NF-�B activation in
response to DNA damage. Cell 123:1079–1092.
15. Balachandran S, Thomas E, Barber GN (2004) A FADD-dependent innate immune
mechanism in mammalian cells. Nature 432:401–405.
16. Kawai T, et al. (2005) IPS-1, an adaptor triggering RIG-I- and Mda5-mediated type I
interferon induction. Nat Immunol 6:981–988.
17. Vercammen D, et al. (1998) Inhibition of caspases increases the sensitivity of L929 cells
to necrosis mediated by tumor necrosis factor. J Exp Med 187:1477–1485.
18. Holler N, et al. (2000) Fas triggers an alternative, caspase-8-independent cell death
pathway by using the kinase RIP as effector molecule. Nat Immunol 1:489–495.
19. Paulus C, Krauss S, Nevels M (2006) A human cytomegalovirus antagonist of type I
IFN-dependent signal transducer and activator of transcription signaling. Proc Natl
Acad Sci USA 103:3840–3845.
20. Zimmermann A, et al. (2005) A cytomegaloviral protein reveals a dual role for STAT2
in IFN-� signaling and antiviral responses. J Exp Med 201:1543–1553.
21. Skaletskaya A, et al. (2001) A cytomegalovirus-encoded inhibitor of apoptosis that
suppresses caspase-8 activation. Proc Natl Acad Sci USA 98:7829–7834.
22. Me´nard C, et al. (2003) Role of murine cytomegalovirus US22 gene family members for
replication in macrophages. J Virol 77:5557–5570.
23. Goldmacher VS, et al. (1999) A cytomegalovirus-encoded mitochondria-localized in-
hibitor of apoptosis structurally unrelated to Bcl-2. Proc Natl Acad Sci USA 96:12536–
12541.
24. McCormick AL, Meiering CD, Smith GB, Mocarski ES (2005) Mitochondrial cell death
suppressors carried by human and murine cytomegalovirus confer resistance to pro-
teasome inhibitor-induced apoptosis. J Virol 79:12205–12217.
25. Popkin DL, Virgin HW (2003) Murine cytomegalovirus infection inhibits tumor necrosis
factor � responses in primary macrophages. J Virol 77:10125–10130.
26. Montag C, Wagner J, Gruska I, Hagemeier C (2006) Human cytomegalovirus blocks
tumor necrosis factor �- and interleukin-1�-mediated NF-�B signaling. J Virol
80:11686–11698.
27. Jarvis MA, et al. (2006) Human cytomegalovirus attenuates interleukin-1� and tumor
necrosis factor � proinflammatory signaling by inhibition of NF-�B activation. J Virol
80:5588–5598.
28. Brune W, Me´nard C, Heesemann J, Koszinowski UH (2001) A ribonucleotide reductase
homolog of cytomegalovirus and endothelial cell tropism. Science 291:303–305.
29. Heyninck K, et al. (1999) The zinc finger protein A20 inhibits TNF-induced NF-�B-
dependent gene expression by interfering with an RIP- or TRAF2-mediated transacti-
vation signal and directly binds to a novel NF-�B-inhibiting protein ABIN. J Cell Biol
145:1471–1482.
30. Lembo D, et al. (2004) The ribonucleotide reductase R1 homolog of murine cytomeg-
alovirus is not a functional enzyme subunit but is required for pathogenesis. J Virol
78:4278–4288.
31. Alexopoulou L, Holt AC, Medzhitov R, Flavell RA (2001) Recognition of double-
stranded RNA, activation of NF-�B by Toll-like receptor 3. Nature 413:732–738.
32. Tada K, et al. (2001) Critical roles of TRAF2 and TRAF5 in tumor necrosis factor-induced
NF-�B activation and protection from cell death. J Biol Chem 276:36530–36534.
33. Wertz IE, et al. (2004) De-ubiquitination and ubiquitin ligase domains of A20 down-
regulate NF-�B signalling. Nature 430:694–699.
34. Valchanova RS, Picard-Maureau M, Budt M, Brune W (2006) Murine cytomegalovirus
m142 and m143 are both required to block protein kinase R-mediated shutdown of
protein synthesis. J Virol 80:10181–10190.
35. Gaspar M, Shenk T (2006) Human cytomegalovirus inhibits a DNA damage response by
mislocalizing checkpoint proteins. Proc Natl Acad Sci USA 103:2821–2826.
36. Martinet W, De Meyer GR, Timmermans JP, Herman AG, Kockx MM (2006) Macro-
phages but not smooth muscle cells undergo benzyloxycarbonyl-Val-Ala-DL-Asp(O-
Methyl)-fluoromethylketone-induced nonapoptotic cell death depending on recep-
tor-interacting protein 1 expression: Implications for the stabilization of macrophage-
rich atherosclerotic plaques. J Pharmacol Exp Ther 317:1356–1364.
37. Tabeta K, et al. (2004) Toll-like receptors 9 and 3 as essential components of innate
immune defense against mouse cytomegalovirus infection. Proc Natl Acad Sci USA
101:3516–3521.
38. Edelmann KH, et al. (2004) Does Toll-like receptor 3 play a biological role in virus
infections? Virology 322:231–238.
39. Bresnahan WA, Shenk T (2000) A subset of viral transcripts packaged within human
cytomegalovirus particles. Science 288:2373–2376.
40. Romanowski MJ, Garrido-Guerrero E, Shenk T (1997) pIRS1 and pTRS1 are present in
human cytomegalovirus virions. J Virol 71:5703–5705.
41. Compton T, et al. (2003) Human cytomegalovirus activates inflammatory cytokine
responses via CD14 and Toll-like receptor 2. J Virol 77:4588–4596.
42. Thome M, et al. (1997) Viral FLICE-inhibitory proteins (FLIPs) prevent apoptosis induced
by death receptors. Nature 386:517–521.
43. Tewari M, Dixit VM (1995) Fas- and tumor necrosis factor-induced apoptosis is inhibited
by the poxvirus crmA gene product. J Biol Chem 270:3255–3260.
44. Vanden Berghe T, et al. (2006) Necrosis is associated with IL-6 production but apoptosis
is not. Cell Signal 18:328–335.
45. Brune W, Nevels M, Shenk T (2003) Murine cytomegalovirus m41 open reading frame
encodes a Golgi-localized antiapoptotic protein. J Virol 77:11633–11643.
46. Jurak I, Brune W (2006) Induction of apoptosis limits cytomegalovirus cross-species
infection. EMBO J 25:2634–2642.
47. Winkler C, Denker K, Wortelkamp S, Sickmann A (2007) Silver- and Coomassie-staining
protocols: Detection limits and compatibility with ESI MS. Electrophoresis 28:2095–
2099.
Mack et al. PNAS � February 26, 2008 � vol. 105 � no. 8 � 3099
M
IC
RO
BI
O
LO
G
Y