Membrane Protein Secretases
Mechanisms controlling the shedding of transmembrane molecules
A. Merlos-Suarez and 1. Arribas'
Laboratori de Recerca Oncologica, Servei d'Oncologia MPdica, Hospital General Universitari Val1 d'Hebron,
Psg. Val1 d'Hebron I 19- 129, Barcelona 08035, Spain
Although the number of cell-surface transmembrane molecules known to release their extracellular domain to the cell media has grown
greatly during past years, the protease(s)
involved have remained elusive until recently.
T h e analysis of ectodomain shedding of proteins
as diverse as membrane-anchored growth
factors, such as pro-transforming growth factor a
(proTGF-a) and pro-tumour necrosis factor a
(proTNF-ol), growth factor receptors, ectoenzymes, cell adhesion molecules and the /j amyloid precursor protein (/jAPP) has revealed
several common features. T h e shedding of most
proteins occurs at a fixed distance from the
plasma membrane, can be activated via protein
kinase C (PKC) and can be prevented by
hydroxamic acid-based metalloprotease inhibitors
initially developed to block the shedding of
proTNF-x [l]. T h e structural diversity of proteins that undergo protein ectodomain shedding
initially suggested the existence of several proteases involved, each endowed with a restricted
specificity. However, Chinese hamster ovary
(CHO) cell mutants selected for a lack of
proTGF-x shedding (M1 and M2 cells) showed a
general defect in protein ectodomain shedding,
indicating the existence of at least a common
component necessary for the shedding of multiple transmembrane proteins [2,3].
T h e first protease shown to be involved in a
shedding event acts on proTNF-z and belongs to
the family of metalloprotease disintegrins (also
known as the ADAM family), a family of modular
transmembrane Znzf-dependent metalloproteases that contain a disintegrin domain. Metalloprotease disintegrins are involved in sperm-egg
fusion, muscle fusion in C2C12 myoblasts and
the regulation of neural cell fate at different
stages of neurogenesis in Ilrosophila (metalloprotease disintegrins are reviewed in [4,5]). T h e
substrates of most members of the metalloproAbbreviations used: TGF-a, transforming growth
factor r ; TNF-a, tumour necrosis factor a; /{APP, /{
amyloid precursor protein; PKC, protein kinase C;
TFP, trifluoperazine; 'I'ACE, TNF-a-converting
enzyme; CHO, Chinese hamster ovary; €IA, haemagglutinin.
' T o whom correspondence should be addressed.
243
tease disintegrin family are unknown because
most metalloprotease disintegrins have been isolated on the basis of sequence homology. So far,
only TNF-a-converting enzyme (TACE) has
been convincingly shown to act on proTNF-a
[6,7]. In addition, the proteolytic activity of Kuzbanian (also known as ADAM 10) has been
shown to be necessary for the processing of
Notch, the receptor for the Delta ligand [8];
however, it is not clear whether Kuzbanian acts
directly on Notch [8a]. Although the finding that
the first protease found to be involved in a shedding event (TACE) belongs to the numerous
metalloprotease disintegrin family seemed to
confirm the notion of the participation of several
metalloproteases in protein ectodomain shedding, it has been recently shown that TACE has
a central role because it is responsible for the
shedding of several unrelated proteins [%I.
T h e regulation of protein ectodomain shedding remains largely uncharacterized. Although
PKC is a modulator of the shedding of most
proteins tested so far, the molecular basis of
PKC activation is unknown, except for that of
PAPP. Activation of endogenous PKC seems to
augment the formation from the trans-Golgi network of secretory vesicles containing /jAPP,
increasing the availability of /jMP at the cell surface [9], where ectodomain shedding takes place
[10,11]. This mechanism of modulation of protein ectodomain shedding seems to be a particularity of /jAPP because PKC frequently activates
ectodomain shedding of proteins resident at the
cell surface, such as proTGF-a, proTNF-a or
L-selectin. It has been recently shown that calmodulin binds directly to the cytoplasmic
domain of L-selectin and has a role in the shedding of its ectodomain, because calmodulin
inhibitors such as trifluoperazine (TFP) and
calmidazolium
induce
the
shedding of
L-selectin. However, it is not known whether
PKC mediates the activation of L-selectin shedding via calmodulin.
T o investigate whether calmodulin has a
role in the shedding of other transmembrane
molecules, we have analysed the effect of calmodulin inhibitors on the shedding of proTGFa. As shown in Figure 1(A), T F P promotes, in a
I999
Biochemical Society Transactions
244
dose-dependent manner, the typical loss of
proTGF-a immunoreactivity at the cell surface
that follows the proteolytic release of the ectodomain of proTGF-a. Half-maximal activation of
proTGF-a shedding is observed with approx.
30 pM TFP, a concentration comparable with
that required to inhibit the shedding of
L-selectin
[ 121. Time-course experiments
showed that the kinetics of proTGF-a ectodomain shedding induced by T F P (tli2z
15 min) is
slightly slower than that induced by the PKC
activator PMA ( t l , 2 z 4 min) (Figure lB, and
results not shown). Because shedding-defective
CHO cell mutants (M2 cells), initially isolated
for a lack of proTGF-a shedding, are affected in
an unidentified component necessary for the
shedding of a variety of molecules including
L-selectin, we also tested the effect of T F P in
M2 cells. As shown in Figure 1(B), T F P showed
little or no effect on proTGF-r shedding in M2
cells, indicating that the component affected in
M2 is unrelated to calmodulin.
T o characterize more fully the effect of the
calmodulin inhibitor T F P on the shedding of
proTGF-x, the effect of the hydroxamic acidbased metalloprotease inhibitor TAPI on the
shedding promoted by T F P was analysed. TAPI
is known to block the shedding of the ectodomain of a variety of proteins, including proTGF-
a and L-selectin.
T h e data in Figure 1(C)
indicate that, as with the activation produced via
PKC, the loss of cell-surface proTGF-a induced
by T F P is inhibited by simultaneous treatment
with TAPI. Similar results were obtained when
the effect of another calmodulin inhibitor, calmidazolium, was analysed under the same conditions (results not shown). T h e results presented
here show that calmodulin inhibitors known to
induce the shedding of L-selectin also induce
the proteolytic shedding of proTGF-a, indicating
that calmodulin is a possible modulator of the
general shedding machinery in vivo. However,
calmodulin probably does not have a role in the
mechanism that activates the shedding of
proTGF-a via PKC, because it has been recently
found that proTGF-a devoid of a cytoplasmic
tail, and therefore incapable of interacting with
calmodulin, is shed from the cell surface with
the same kinetics as wild-type molecules (J.
Ureiia, J. Baselga and J. Arribas, unpublished
work).
T h e availability of cell mutants defective in
the shedding of several molecules, and the
identification of TACE, the protease responsible
for the shedding of proTNF-r, have provided
useful tools for gaining insights into the mechanisms that regulate the shedding of transmembrane proteins. T o characterize the defect of M2
Figure I
Effect of calmodulin inhibitors on the shedding of proTGF-z
(A) CHO cells stably transfected with proTGF-r tagged in the ectodomain with the haemagglutinin (HA) epitope were incubated, as
indicated, with different concentrations of TFP in Dulbecco's modified Eagle's medium at 37°C for I 5 min. Then cells were shifted to 4"C,
immunostained with anti-HA, FITC-coupled secondary antibodies and analysed by flow cytometry as described previously [2]. The results
are expressed as percentages relative to mean fluorescence of untreated cells and are averages of duplicate determinations. ( 6 ) Wild-type
or shedding-defective mutant CHO cells (M2 cells) stably transfected with proHA/TGF-r were treated with 40 IIM TFP for variable periods
as indicated. Then cells were treated and analysed as described for (A). (C). Wild-type C H O cells stably transfected with proHA/TGF-r
were treated with I / t M PMA. 40 pM TFP or 50 pM TAPI. as indicated, in Dulbecco's modified Eagle's medium at 37°C for 20 min. Then
cells were treated and analysed as described for (A).
B
A
C
-M-
100
80
60
40
20
-
0
0
T
5
10
15
20
Time (min.)
Volume
27
WlldTyl:
25
-
+PMA +TFP +PMA +TFP
+TAP1
Treatment
Membrane Protein Secretases
cells more fully, we have recently analysed the
shedding of proTNF-a in these cells and found
that they lack PMA-induced proTNF-a shedding, indicating that, as expected (given the generality of the defect of M2 cells [3]), the
component mutated is necessary for the shedding of proTNF-cr [13]. This mutant component
is likely to be different from TACE because the
defect in M2 cells is not rescued by transfection
with TACE (Figure 2) [13]. T h e existence of an
unidentified component necessary for the shedding of several molecules, including proTNF-a,
and different from TACE, has been further substantiated by using somatic cell fusions between
TACE-null cells (TACE-'-) and M2 cells.
TACE-'- x M2 hybrids show wild-type shedding
of proTNF-a, confirming that the mutation in
M2 cells does not affect TACE. Furthermore,
although TACE-'- cells show defects in PAPP
shedding [ 13a], TACE-'- x M2 hybrids show
normal shedding of proTGF-a and PAPP [13].
Transport and maturation of TACE in wild-type
cells are indistinguishable from those in M2 cells
[13], arguing that the component mutated is not
involved in the regulation of the biosynthesis or
subcellular location of TACE. This unidentified
Figure 2
proTNF-cr shedding in wild-type and M2 mutant cells
transfected with TACE
Wild-type or shedding-defective mutant cells (M2) were transiently co-transfected with proTNF-r and TACE as described in
[13], then treated with or without PMA in Dulbecco's modified
Eagle's medium at 37°C for 20 rnin Then cells were shifted to
4°C. irnrnunostained with monoclonal antibodies against the ectodomain of proTNF-r, FITC-coupled secondary antibodies and
analysed by flow cytornetry The results are expressed as percentages of positive cells relative to that of untreated cells and are
the average of duplicate determinations
+
+
+
+
-
+
-
+
-+
+
WT
cells
+
+
Mutant
cells
proTNF-a
TACE
PMA
component seems to control only a subset of
metalloprotease disintegrins because the processing of Notch, which is dependent on the
metalloprotease disintegrin Kuzbanian, is normal
in M2 cells [13]. Collectively, the results presented here and elsewhere [13] indicate that the
activity of TACE is tightly controlled by a novel
component of the shedding system that controls
the activity of a subset of metalloprotease disintegrins. Because this component is not involved
in the regulation of the biosynthesis or subcellular localization of the substrates or shedding
machinery, it is tempting to speculate that the
novel component would mediate the recognition
of the substrate by certain metalloprotease disintegrins, such as TACE.
T h e regulation of the subcellular location of
components of the shedding machinery or its
substrates has been suggested as a likely mechanism for controlling the shedding of transmembrane molecules. However, only for PAPP has
this mechanism been proved [9]. Several years
ago the shedding of proTGF-a was found to be
dependent on the integrity of the most
C-terminal, cytoplasmic, residue: valine. Recently
we and others have shown that proTGF-a
C-terminal mutants are localized to the endoplasmic reticulum ([ 141, and J. Uretia, J. Baselga
and J. Arribas, unpublished work). T h e modest
percentage of proTGF-a C-terminal mutants
that reach the cell surface are shed with identical
kinetics to that of wild-type molecules (J. Ureiia,
J. Baselga and J. Arribas, unpublished work).
Therefore, retention in the endoplasmic reticulum provides an explanation for the defective
shedding of proTGF-a C-terminal mutants. Currently we are characterizing the mechanisms that
localize proTGF-a C-terminal mutants to the
endoplasmic reticulum. T h e identification of
proteins that recognize the C-terminus of
proTGF-cr will be useful for evaluating the
impact of the regulation of the proTGF-a subcellular location in the control of the shedding of
its ectodomain.
We thank Roy Black from Immunex Corporation for
reagents and continuous helpful discussions. This
work was supported by grants from the Spanish Comisidn Interministerial de Ciencia y Tecnologia (SAF970229), Fundaci6 La Marat6 de TV3 (036/97) and a
predoctoral fellowship from the Spanish ministry of
education to A.M.-S.
1 Ilooper, N. M., Karran, E. H. and Turner, A. J.
(1997) Biochem. J. 321, 265-279
I999
245
Biochemical Society Transactions
246
2 Arribas, J. and MassaguC, J. (1995) J. Cell Biol. 128,
433-441
3 Arribas, J., Coodly, L., Vollmer, P., Kishimoto,
T. K., Rosejohn, S. and MassaguC, J. (1996) J. Biol.
Chem. 271, 11376-11382
4 Blobel, C. P. (1997) Cell 90, 589-592
5 Wolfsberg, T. G., Straight, P., Gerena, R. L.,
Huovila, A.-P. J., Primakoff, P., Myles, D. G. and
White, J. M. (1995) Development 169, 378-383
6 Black, R. A., Rauch, C. T., Kozlosky, C. J.,
Peschon, J. J., Slack, J. L., Wolfson, M. F., Castner,
B. J., Stocking, K. L., Reddy, P., Srjnivasan, S. et
al. (1997) Nature (London) 385, 729-733
7 Moss, M. L., Jin, C. S.-L., Milla, M. E., Burkhart,
W., Carter, H. L., Chen, W.-J., Clay, W.-C.,
Didsbury, J. R., Hassler, D., Hoffman, C. R. et al.
(1997) Nature (London) 385, 733-736
8 Pan, D. and Rubin, G. M. (1979) Cell 90, 271 -280
8a Logeat, F., Bessia, C., Brou, C., LeBail, O.,
Jarriault, S., Seidah, N. G. and Israel, A. (1998)
Proc. Natl. Acad. Sci. U.S.A. 95, 8108-8112
8b Peschon, J. J., Slack, J. L., Reddy, P., Stocking,
K. L., Sunnarborg, S. W., Lee, D. C., Russell,
W. E., Castner, B. J., Johnson, R. S., Fitzner, J. N.
et al. (1 998) Science 282, 128 1 - 1284
9 Xu, H., Greengard, P. and Gandy, S. (1996) J. Biol.
Chem. 270,23243-23245
10 Sisodia, S. S. (1992) Proc. Natl. Acad. Sci. U.S.A.
89,6075-6079
1 1 Bosenberg, M. W., Pandiella, A. and MassaguC, J.
(1992) Cell 71, 1157-1165
12 Kahn, J., Walcheck, B., Migaki, G. I., Jutila, M. A.
and Kishimoto, T. K. (1998) Cell 92, 809-818
13 Merlos-Suirez, A., Fernindez-Larrea, J., Reddy, P.,
Baselga, J. and Arribas, J. (1998) J. Biol. Chem., in
the press
13a Buxbaum, J. D., Liu, K.-N., Luo, Y., Slack, J. L.,
Stocking, K. L., Peschon, J. J., Johnson, R. S.,
Castner, B. J., Cerretti, D. P. and Black, R. A.
(1998) J. Biol. Chem. 273,27765-27767
14 Briley, G. P., Hissong, M. A., Chiu, M. L. and Lee,
D. C. (1997) Mol. Biol. Cell 8, 1619-1631
Received 21 August 1998
Proteinase-activated receptors: a growing family of heptahelical receptors for
thrombin, trypsin and tryptase
0. Dery and N. W. Bunnett'
Departments of Surgery and Physiology, University of California San Francisco, 52 I Parnassus Avenue, San Francisco,
CA 94 143-0660, U.S.A.
Introduction
Although proteases usually function as degradative enzymes, certain enzymes have biological
activity that is mediated through a new and growing family of G-protein-coupled receptors
(GPCRs) (reviewed in [l]). T h e best example is
thrombin, which has activity distinct from its role
in the coagulation cascade. In 1991, Coughlin
and colleagues isolated a cDNA encoding a
GPCR that mediates many of thrombin's cellular
effects [Z]. Initially named the thrombin receptor, it was subsequently renamed proteinase-activated receptor 1 (PAR-1) after the discovery of
related receptors. Thrombin activated platelets
from PAR-1 knockout mice, but had no effect on
fibroblasts from these animals, proving that addiAbbreviations used: PAR, proteinase-activated receptor; GPCR, G-protein-coupled receptor; PKC, protein
kinase C; MAP, mitogen-activated protein; GRK,
G-protein-coupled receptor kinase; GFP, green fluorescent protein; h, human.
'To whom correspondence should be addressed.
Volume 27
tional thrombin receptors exist [3]. Indeed, PAR1 has no role in the activation of mouse platelets
by thrombin. A cDNA encoding PAR-3 was subsequently amplified by reverse transcriptasemediated PCR from rat platelet RNA with
degenerate primers based on conserved regions
of other GPCRs [4]. Although low concentrations
of thrombin had no effect on platelets from PAR3 knockout mice, high concentrations caused a
delay and diminished activation, suggesting the
existence of yet another thrombin receptor. A
gene bank search for PAR-related sequences
identified an expressed sequence tag similar to
other PARS; full-length cDNA was subsequently
obtained [5,6]. T h i s cDNA encoded a unique
thrombin receptor, designated PAR-4. T h u s
thrombin interacts with three related receptors:
PAR-1, PAR-3 and PAR-4. Pancreatic trypsin
and tryptase, a major secretory granule protease
of human mast cells, also possess biological
activity that might be receptor-mediated. In 1994
Sundelin and colleagues screened a genomic