Molekularbiologie, Freie Universitat Berlin, Universitatsklinikum. Rudolf Virchow, und Max-Delbrick-Centrum fur Molekulare Medizin,. 13122 Berlin and lInstitut ...
The EMBO Journal vol.14 no.6 pp.1156-1165, 1995
Tumor necrosis factor (TNF)-ax activates c-raf-1 kinase via the p55 TNF receptor engaging neutral sphingomyelinase Claus Belka, Katja Wiegmann1, Dieter Adam', Rudi Holland, Maria Neuloh, Friedheim Herrmann, Martin Kronkel and Marion A.Brach2 Abteilung fur Medizinische Onkologie und Angewandte Molekularbiologie, Freie Universitat Berlin, Universitatsklinikum Rudolf Virchow, und Max-Delbrick-Centrum fur Molekulare Medizin, 13122 Berlin and lInstitut fur Medizinische Mikrobiologie und Hygiene, Technische Universitat Munchen, 81675 Germany 2Corresponding author Communicated by P.A.Bauerle
TNF-oa mediates proliferation, functional activation and apoptotic death of cells depending upon its concentration and target cell type. The signaling pathways used by TNF-a to mount these responses are, at present, not completely understood. We report here that TNF-a promotes dose- and time-dependent phosphorylation and activation of the c-raf-1 kinase engaging the type I p55 TNF receptor (TNF-R). c-raf-kinase activation was duplicated by an agonistic monoclonal antibody directed against the p55 TNF-R. Moreover, ectopic expression of the human p55 TNF-R in murine pre-B 70Z/3 cells was sufficient to confer c-raf-1-kinase activation by human TNF-a. By inhibiting intracellular activation of acidic sphingomyelinase (SMase) and by using deleted forms of the type I TNF-R it was shown that the neutral, but not the acidic SMase, participated in TNF-a-mediated phoshorylation and activation of the c-raf kinase. TNF-a-induced transcriptional activation of a heterologous promoter construct harboring the AP-1 binding site was also mediated by the type I p55 TNF-R. In this case the initiation of transcription required the same cytoplasmic domain as that responsible for activation of c-raf-1 kinase and was liberated in the presence of a dominant negative mutant of c-raf-1. Keywords: p55 type I TNF-receptor/raf-kinase/signal transduction/TNF-a
Introduction Tumor necrosis factor-a (TNF-a) is a 17 kDa polypeptide primarily produced by activated mononuclear phagocytes. TNF-a exerts a wide range of biological responses depending on the cell type being targeted (Schutze et al., 1992a). In vivo TNF-a mediates endotoxic shock symptoms by stimulating acute phase protein production in the liver (Tracey et al., 1986) and participates in degenerative joint diseases and various inflammatory states (Dayer et al., 1985). TNF-a inhibits growth of various tumor cells while also providing mitogenic signals for
1156
hematopoietic cells and fibroblasts (Brach et al., 1993a). The mitogenic response induced by TNF-a in early hematopoietic cells has been ascribed to its capacity to induce release of secondary cytokines, to transmodulate cytokine receptors and to induce activation of early response genes such as c-jun (Elbaz et al., 1991a,b; Brach et al., 1992a,b). Moreover, TNF-a has been shown to enhance binding and transcriptional activation capacity of several transcription factors including NF-KB, NF-jun or AP-1 (Osbom et al.; 1987; Brach et al., 1992c; 1993a). TNF-a exerts its effects through two distinct surface binding sites, a type I receptor (p55-60 kDa) and a type II receptor (p75-80 kDa) (Tartaglia and Goeddel, 1992). These two receptors bind TNF-a with high affinity but differ in their intracellular domains and thus mediate distinct cellular responses (Tartaglia and Goeddel, 1992). Some studies have indicated that the type II receptor is linked to the mitogenic response (Tartaglia et al., 1991); while the type I receptor is thought to confer cytotoxic signals and to mediate cell adhesion (Mackay et al., 1993; Tartaglia et al., 1993a). Both receptor types have been implicated in TNF-a-induced apoptosis (Higuchi and Aggarwal, 1993; Tartaglia et al., 1993a). Other reports indicate that the presence of the type I receptor is sufficient to transmit the TNF-a signal (Wiegmann et al., 1992; Yanaga and Watson, 1992; Slowik et al., 1993) whereas the type II receptor may stimulate the association of TNF-a with the type I receptor and thereby reduce the TNF-a concentration required for activity (Tartaglia et al., 1993b). Signaling pathways inititated by TNF-a include the activation of PKC, phospholipase (PL) A2 phosphatidylcholine (PC)-phospholipase C (PLC) and sphingomyelinase (SMase) (Godfrey et al., 1987; Schutze et al., 1990; Kim et al., 1991; Matthias et al., 1991; Schutze et al., 1992b; Wiegmann et al., 1994). TNF-a has also been shown to induce serine/threonine or tyrosine phosphorylation of various cellular proteins (Guy et al., 1991). Signaling events mounted by TNF-a which result in proliferative responses are, however, still enigmatic. We therefore sought to determine whether TNF-a was capable of activating c-raf-1, a serine/threonine kinase, which is known to participate in stimulatory growth signals exerted by several other mitogenic cytokines (Kanakura et al., 1991; Brennscheidt et al., 1994).
Results TNF-c induces phosphorylation of c-raf-1 kinase The first set of experiments examined whether exposure of human myelogenous leukemia cells, monocytic leukemia cells, fibroblasts or primary blood monocytes to recombinant human (rh) TNF-a was associated with tyrosine phosphorylation of the c-raf- 1 kinase. Cells were
TNF-a activates c-raf-1 kinase
serum- and factor-deprived for 24 h and then exposed to TNF-ax for 15 min. Cell lysates were immunoprecipitated with an antiserum to c-raf-1 followed by immunoblotting with an anti-phosphotyrosine antibody. An increase in tyrosine phosphorylation of c-raf-1 of -3- to 4-fold in response to TNF-a in GF-D8 cells and -2-fold in U 937 cells was observed by laser densitometric analysis of autoradiograph (Figure 1). Comparable findings were also obtained with other cell sources such as lung fibroblasts (FH 109) which have previously been shown to proliferate in response to TNF-a (Brach et al., 1993a), and with primary human blood monocytes which are activated by TNF-a (data not shown).
Time- and dose-dependent activation of c-raf- 1 kinase by TNF-a Phosphorylation of c-raf-1 was associated with enhanced kinase activity as determined by in vitro kinase assays using recombinant histone HI and recombinant kinaseinactivated MEK (Force et al., 1994) as substrates. C-raf-1 kinase activity increased in GF-D8 cells within 5 min of exposure to TNF-a, peaked after 15 min and declined to starting levels within 60 min (Figure 2A and B). Comparable time-kinetics were also observed in U 937 cells, FH 109 lung fibroblasts and in primary blood monocytes (Figure 2C-E, respectively). Again, the capacity of TNF-a to induce c-raf-1 kinase activity was more pronounced in GF-D8 cells than in U 937 cells (Figure 2B cf. 2C). Moreover, TNF-a stimulated c-raf kinase activity in a dose-dependent manner in all cell types investigated at concentrations ranging from 1 to 10 U/ml. Whilst at higher TNF-cx concentrations dose-dependency appeared to be lost (Figure 2A-E, right panel).
TNF-a fails to promote AP-1 transcriptional activation in cells transfected with a transdominant negative mutant of c-raf-1 Two transdominant mutants of c-raf-1 kinase were used in order to confirm that the effects observed were mediated by the c-raf- kinase and to exclude the possibility that other proteins, which may have co-immunoprecipitated with the c-raf- 1 protein, were participating in this response (Bruder et al., 1992). One mutant (Raf-C4) carries one cysteine finger only and has been depleted of the raf-1 kinase domain (Bruder et al., 1992). This mutant prevents the activation of the oncogene-responsive element in the polyoma enhancer in response to serum, phorbol ester TPA and ras (Bruder et al., 1992). The other mutant (RafC4pm 17) harbors a single amino acid substitution (C-*S) within the last cysteine of the cysteine finger motif of raf. Thereby, this dominant negative mutant decreases c-raf- 1 activity by only -30% (Bruder et al., 1992). FH 109 cells respond to TNF-a with enhanced c-raf- 1 kinase activation which is prevented in cells transfected with the Raf-C4 mutant and reduced in cells transfected with the RafC4pml7 mutant (Figure 3A). These mutants had also been co-transfected into FH 109 cells with a heterologous promoter construct carrying the AP- 1 binding site as an enhancer as previously described.(Brach et al., 1993b). FH 109 cells transiently transfected with pAP1-TKGH respond to TNF-a with -5- to 6-fold enhanced reporter gene activity (Figure 3B). When co-transfected with Raf-C4, enhanced reporter gene
GF-D8 c-Raf-1
_
-
GF-D8
U 937
-,
i
*
-
U 937 97 _- 69
PP
46 - 32
TNF-cx
-
+
-
ct-c-Raf-1
+
-T+
-
+
c^-pTyr
Fig. 1. TNF-a induces tyrosine phosphorylation of c-raf in GF-D8 and U 937 leukemia cells. Cells were serum- and factor-deprived for 24 h and then exposed (or not) to TNF-ax (100 U/ml) for 15 min. Cell lysates were immunoprecipitated with an antiserum to c-raf-1 (antiSP63); immunoprecipitated proteins were separated by gel electrophoresis, transferred to Hybond C+ membrane and immunoblotted to c-raf-1 antibody (left panel) or an antiphosphotyrosine antibody (right panel).
activity following exposure to TNF-a is almost completely abolished. In contrast, FH 109 cells co-transfected with Raf-C4pm 17 display an -4-fold enhancement of reporter gene activity when exposed to TNF-a. Co-transfection with a control plasmid containing the P-galactosidase gene had no effect on TNF-a-inducible reporter gene activity.
The type I p55 TNF-R mediates activation of c-raf- 1 kinase Monoclonal antibodies targeting either the p55 type I receptor (Htr-9) or the p75 type II receptor (Utr- 1) (Brockhaus et al., 1990) were used to investigate the role of type I and type II TNF-R in c-raf-1 kinase activation (Figure 4). Upon exposure to an agonistic anti-p55 TNF-R antibody (Htr-9), GF-D8 cells displayed enhanced c-raf-1 phosphorylation and increased c-raf-1 kinase activity as determined in in vitro kinase assays using either histone HI or recombinant kinase-inactivated GST-MEK fusion protein as substrates. These effects were only marginally enhanced by TNF-a. Blockade of the type II receptor in GF-D8 cells using the Utr-1 antibody interfered with the ability of low-dose TNF-a (10 U/ml) to induce c-raf-1 phosphorylation or kinase activity but the response was preserved at high doses of TNF-a (50-100 U/ml) (Figure 4A). Comparable results were also obtained with FH 109 fibroblasts and primary blood monocytes (data not shown). In U 937 cells, high basal levels of c-raf- 1 kinase activity were significantly reduced in the presence of Utr- 1, which could be overcome in the presence of high concentrations of TNF-a (50-100 U/ml). The agonistic Htr-9 antibody targeting the p55 type TNF-R enhanced constitutive c-raf- 1 phosphorylation or c-raf- 1 kinase activity by 1.5-fold in the absence of TNF-a and by -2-fold in the presence of 10-100 U/ml TNF-a in U 937 cells (Figure 4B). These findings indicate that high constitutive c-raf- 1 kinase activity seen in serum-deprived U 937 cells may have resulted from endogenously synthesized TNF-a, as previously suggested (Ghibelli et al., 1994). More importantly, these findings indicate that phosphorylation and activation of the c-raf- 1 kinase mediated by TNF-a can be duplicated by the agonistic antibody to the p55 type I TNF-R and is not further enhanced by TNF-a. Blockade of the p75 type II TNF-R did not abolish the ability of high concentrations of TNF-a
1157
C.Belka et al
A
ALDC
TNF-c. (U/mI)
TNF-cz ( Minutes)
I0 p33MEK >
B
%.]
.f
601
30
15
5
69
_
-
69
-
46
-
46
-
32 21
-
32 21
-
T
.I
-il
-,
T
".
-,
4
,:.
zf
'..
-
I1
-
L "
,,
C -'
-1;1
I
.
.
< C
L
:t
,
,,
L
T.::l,,,
21. :i
3~~~~~~~~1 :.. -. ;.
I.-S.;
C3
r.
T
_ 2 E]15's-.9'~~~El
MEr ,I
-v; e-
.-4
D
I-
-
7-
-t-
E
-"
0
t
;5
±
"'F-at v[-T'
si:-a-K~~~~~~~~~~~~~~~. e
±~~~~~~~~~~~~~~~~~~~~~1
Fig. 2. TNF-a induces activation of the c-raf- kinase in a time- and dose-dependent fashion. Cells (GF-D8 myelod leukemia cells, U 937 monocytic leukemia cells, FH 109 embryonic lung fibroblasts and primary blood monocytes) were serum- and factor-deprived for 24 h followed by exposure to TNF-a (100 U/ml) for the time points indicated (left panel) or to TNF-a (0-100 U/ml) for 15 min (right panel). Cell lysates were immunoprecipitated with an anti-c-raf-t antiserum (anti-SP63) and immunoprecipitates were subjected to in vitro kinase assays using recombinant histone Hl or recombinant kinase-inactivated GST-MEK fusion protein as substrates as detailed in Materials and methods. Reaction products were
separated by SDS-gel electrophoresis. Phosphorylation of histone H)
(histone HI) or GST-MEK fusion protein (MEK) was analyzed by
quantitation of photostimulated luminescence. In order to control for equal amounts of c-raf-
protein subjected to kinase assays protein loading was
confirmed by densitometric analysis of Coomassie Blue-stained gels (data not shown). (A) Representative blot obtained with GF-D8 cells employing
recombinant kinase-inactivated GST-MEK fusion protein as a substrate. (B-E) The fold increase of signal intensity versus background is expressed as mean ± SD of two independent experiments employing recombinant histone H)
or recombinant inactivated MEK as substrates obtained from
GF-D8 cells (B), U 937 cells (C), FH-109 cells (D) and primary blood monocytes (E). Signals obtained from medium-treated cells were arbitrarily set as '1'. C-raf-1 was also immunoprecipitated with two other antibodies followed by in vitro kinase assays which gave similar results (data not
shown).
1158
TNF-a activates c-raf-1 kinase
A
activation in the presence of Utr- 1 may be explained by interference of Utr- 1 with surface binding of TNF-a (Brockhaus et al., 1990).
8c
0 -I.-
LL
:3 z -0 ..c .C: 0 0 0-
75 LL
:3
6-
T
4-
@
Ectopic expression of the type I receptor is sufficient to mediate c-raf-1 activation A murine pre-B cell line (70Z/3) expressing an ectopic human type I TNF-R was employed (70Z/3TR55) to
m
investigate further the notion that the p55 type I TNF-R
c-raf-nI
activation of the TNF-a-mediated to human had previously been shown not to respond70Z/3
\is involved in \ \ _
kinase (Wiegmann et al., 1992). Wild-type
(D
O
0.
5
TNF-a (Wiegmann et al., 1992). In line with these findings Ewe demonstrated that wild-type 70Z/3 cells failed to a display enhanced tyrosine phosphorylation of c-raf- 1 upon TNF-oa exposure (Figure 5A). In contrast, 70Z/3TR55 cells which had been transfected with the human p55 type I TNF receptor responded to TNF-a with enhanced tyrosine phosphorylation of the c-raf-1 protein (Figure 5A). In vitro kinase assays using histone HI or recombinant kinase-inactivated GST-MEK fusion protein as substrates showed that phosphorylation of c-raf-1 was associated with enhanced c-raf-1 kinase activity (Figure SB). TNF-ainduced c-raf- 1 kinase activity was dose dependent in IT' concentrations up to 10 U/ml, but lost its dose-dependency at higher concentrations (Figure 5B). Activation of c-raf- 1 occurred within 5-10 min following exposure to TNF-a, to 30 min and declined thereafter to peaked after
B c
oa 7Z
6-I I I
4-
._
7' o _ Q
15
.
nI
a
I
,
_ T _ starting levels (data not shown). __~~~~~~~~I C-raf-1 activation by TNF-a does not require PKC
.
.
v
cr
aD
z
+
+
I Y
y
+
CL
()
+
aDQ -
(
--sA! -,::I 10
-'"
p33 Histone Hi >
w?;.
TNF-cu -
TR55
-.
-_
U-
Ia
i,..
+
+
+
TR55-393
TR55-345
TR55-205
+
69
-
46
-
32
...
Fig. 9. A region close to the transmembrane domain of the p55 TNF-R mediates c-raf-I tyrosine phosphorylation and raf kinase activation. (A) Schematic representation of the type I p55 TNF-R and the deleted forms which were stably transfected into 70Z/3 cells. (B) 70Z/3 transfectants were serum-deprived for 12 h followed by exposure to TNF-ax (100 U/ml) for 15 min. Cell lysates were immunoprecipitated with an antiserum to c-raf-1 (anti-SP63) and in vitro kinase assays were performed as detailed above. Reaction products were separated by SDS-gel electrophoresis and phosphorylation of histone HI, recombinant kinase-inactivated GST-MEK fusion protein, and of c-raf-1 is shown. Comparable results were obtained when recombinant kinase-inactivated GST-MEK fusion protein was used following immunoprecipitation of c-raf-1 with c-raf-I antiserum (anti-SP63) or two additional c-raf-1 antibodies (data not shown). In order to control for equal amounts of immunoprecipitated c-raf-1 protein subjected to kinase assays, the protein loading in Coomassie Blue-stained gels was quantified by densitometnc analysis. One representative blot is shown, two additional experiments using different antibodies to c-raf-I for immunoprecipitation and recombinant kinase-inactivated GST-MEK fusion protein as substrate gave comparable results (data not shown).
have stressed the importance of tyrosine phosphorylation of c-raf- 1 protein in eliciting c-raf- 1 kinase activity (Turner et al., 1991; Fabian et al., 1993). In addition, c-raf-1 has been shown in in vitro assays to bind to activated GTPbound ras (Vojtek et al., 1993). However, the functional implication of this interaction with respect to c-raf- 1 kinase activation is not fully understood. Recent data suggest that ras is required to anchor raf to the plasma membrane (Leevers et al., 1994) and thereby allows activation of raf- 1 kinase by membrane-associated molecules such as lipids. Several studies have suggested that lipid molecules may bind to the cysteine finger motif within the N-terminal regulatory region of the c-rafkinase and thereby contribute to its activation (Li et al., 1991; Bruder et al., 1992). Our results provide evidence that the intracellular activation of membrane-associated neutral SMase through the p55 type I receptor resulting in formation of ceramide at the plasma membrane has to be considered as a pathway leading to c-raf- 1 phosphorylation as has recently been hypothesized and activation (Kolesnick and Golde, 1994). Activation of the c-raf-1 kinase in response to TNF-a may involve the recently identified ceramide-activated kinase (Mathias et al., 1991; Liu et al., 1993), which promotes activation of MAP kinase (Raines et al., 1993). The ceramide-activated kinase
,3
z
.'O OCL0LD
6543210CO) 4
LO
_uI)
C')
e v
C'C)
Co)
C'f)
C
Nl-
U) 0
Co4 _
C'I)
NJ
NJ
rl.
0
0
rl-
r,.
Fig. 10. The region responsible for c-raf- 1 kinase activation within the type I p55 TNF-R is also required for TNF-az-induced activation of a heterologous promoter harboring the AP-1 binding site. 70Z/3 cells and their respective transfectants (see Figure 9A) were transiently transfected with pAP-lTKGH and pTKGH as detailed in Materials and methods, maintained in SCM for 24 h and then exposed (or not) to TNF-a (100 U/ml) for additional 24 h. Thereafter, hGH activity was assessed in cell-free culture supernatants. The increase of reporter gene activity detected in TNF-a-treated transfectants as compared to untreated transfectants is shown. Values are expressed as means SD of three independent experiments. 70Z/3 wild-type cells and all mutants transfected with the control construct pTKGH failed to respond to TNF-a with enhanced reporter gene activity (data not shown).
1163
C.Belka et al.
contains a putative consensus phosphorylation site within the c-raf-1 kinase (C.Belka and M.A.Brach, unpublished observations). There is, as yet, no evidence for the phoshorylation of this site. Finally, our data suggest that the activation of c-raf1 kinase is not sufficient to allow activation of the NF-KB transcription factor as has been suggested by others; (Finco and Baldwin, 1993; Li and Sedivy, 1993) Both the 70Z/3TR55-394 and 70Z/3TR55345 mutants, which still respond to TNF-x with c-raf-1 kinase activation, are clearly defective in mediating NFKB activation (Wiegmann et al., 1994). However, the enhanced transactivating capacity of the AP-1 transcription factor complex following TNF-a is mediated by the same cytoplasmic region which is responsible for c-raf1 kinase activation and appears to depend on the capacity of TNF-a to promote c-raf-1 activation.
Materials and methods Reagents Recombinant human TNF-a (Escherichia cli derived, purity >99%, endotoxin content as determined in LAL assays
