Likewise, processed MMP-2 was observed in the conditioned medium of K-562, NCI-H460 and Hep G2 cells, and their incubation with the metallo-protease inhibitors did not result in the prev
Trang 1activation of pro-matrix metalloprotease-2 by proprotein convertases
Bon-Hun Koo, Hee-Hyun Kim, Michael Y Park, Ok-Hee Jeon and Doo-Sik Kim
Department of Biochemistry, College of Life Science and Biotechnology, Yonsei University, Seoul, South Korea
Introduction
Matrix metalloproteases (MMPs) constitute a family
of 23 zinc-dependent endopeptidases They are
involved in many biological processes and diseases,
and catalyze the proteolysis of extracellular and
non-extracellular matrix molecules [1] For example,
MMP-2 is associated closely with organ growth,
endometrial cycling, wound healing, bone remodeling, tumor invasion and metastasis [2] Its functions are executed through the degradation of components of the basement membrane, including type IV collagen, fibronectin, elastin, laminin, aggrecan and fibrillin [1,3]
Keywords
furin; membrane type-1 matrix
metalloprotease (MT1-MMP); pro-matrix
metalloprotease-2 (pro-MMP-2); pro-MMP-2
activation; proprotein convertases
Correspondence
B.-H Koo, Department of Biochemistry,
College of Life Science and Biotechnology,
Yonsei University, 134 Sinchon-Dong
Seodaemun-Gu, Seoul 120-749, South Korea
Fax: 82 2 312 6027
Tel: 82 2 313 2878
E-mail: k4119@yonsei.ac.kr
D.-S Kim, Department of Biochemistry,
College of Life Science and Biotechnology,
Yonsei University, 134 Sinchon-Dong
Seodaemun-Gu, Seoul 120-749, South Korea
Fax: 82 2 312 6027
Tel: 82 2 2123 2700
E-mail: dskim@yonsei.ac.kr
(Received 14 July 2009, revised 26 August
2009, accepted 28 August 2009)
doi:10.1111/j.1742-4658.2009.07335.x
Matrix metalloprotease-2 is implicated in many biological processes and degrades extracellular and non-extracellular matrix molecules Matrix metalloprotease-2 maintains a latent state through a cysteine–zinc ion pairing which, when disrupted, results in full enzyme activation This pairing can be disrupted by a conformational change or cleavage within the propeptide The best known activation mechanism for pro-matrix metalloprotease-2 occurs via cleavage of the propeptide by membrane type-1 matrix metalloprotease However, significant residual activation of pro-matrix metalloprotease-2 is seen in membrane type-1 matrix meta-lloprotease knockout mice and in fibroblasts treated with metameta-lloprotease inhibitors These findings indicate the presence of a membrane type-1 matrix metalloprotease-independent activation mechanism for pro-matrix metalloprotease-2 in vivo, which prompted us to explore an alternative activation mechanism for pro-matrix metalloprotese-2 In this study, we demonstrate membrane type-1 matrix metalloprotease-independent propep-tide processing of matrix metalloprotease-2 in HEK293F and various tumor cell lines, and show that proprotein convertases can mediate the processing intracellularly as well as extracellularly Furthermore, processed matrix metalloprotease-2 exhibits enzymatic activity that is enhanced by intermolecular autolytic cleavage Thus, our experimental data, taken together with the broad expression of proprotein convertases, suggest that the proprotein convertase-mediated processing may be a general activation mechanism for pro-matrix metalloprotease-2 in vivo
Abbreviations
Con A, concanavalin A; dec-RVKR-cmk, dec-Arg-Val-Lys-Arg-chloromethyl ketone; FITC, fluorescein isothiocyanate; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; MMP, matrix metalloprotease; MT1-MMP, membrane type-1 matrix metalloprotease; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide; PC, proprotein convertase; pro-MMP-2, pro-matrix metalloprotease-2; TGN, trans-Golgi network; TIMP, tissue inhibitor of metalloprotease.
Trang 2Because of its potential for tissue destruction,
MMP-2 activity is regulated at multiple points, such
as gene expression, compartmentalization, zymogen
activation and enzyme inactivation by extracellular
inhibitors [e.g tissue inhibitors of metalloproteases
(TIMPs)] [1] Like other MMPs, pro-MMP-2
main-tains a latent state via an interaction between a
thiol group of a propeptide cysteine residue and the
catalytic zinc ion in the active site [4] Disruption of
this cysteine–zinc ion pairing, such as through
confor-mational changes [5] or proteolysis within the
propep-tide [e.g by plasmin, thrombin or membrane
type-MMPs (MT-MMPs)] [6–12], is required for the
activation of the latent enzyme The most studied
acti-vation mechanism for pro-MMP-2 is cleavage of the
propeptide by MT1-MMP, which requires cooperative
activity between MT1-MMP and TIMP-2 [7,13–15]
However, residual activation of pro-MMP-2 is
observed in MT1-MMP knockout mice, even though
the activation is reduced significantly [16,17] Thus,
these data suggest that an MT1-MMP-independent
activation mechanism for pro-MMP-2 may also exist
in vivo
Proprotein convertases (PCs), a family of Ca2+
-dependent serine proteases of which furin is the most
ubiquitous, have a major role in molecular maturation
[18–20] Most PCs reside within the trans-Golgi
net-work (TGN), but some are present at the cell surface
via a transmembrane domain (e.g furin) [21,22] or the
extracellular matrix (e.g PACE4 and PC5A) [23] Like
other PCs, furin cleaves its substrates immediately
downstream of the consensus sequence
Arg-Xaa-Arg⁄ Lys-Arg (where Xaa is any amino acid) in the
TGN [18,19,24] or at the cell surface [25] The
signifi-cance of furin in many biological processes is
attri-buted to its widespread expression [26] and the
developmental lethality of furin knockout mice [27]
Furthermore, elevated expression of PCs is frequently
observed in various human cancers and tumor cell
lines, implicating the importance of PCs in tumor
progression [28]
The presence of activated MMP-2 in MT1-MMP
knockout mice prompted us to explore an
MT1-MMP-independent activation mechanism for
pro-MMP-2 In this study, we demonstrate
MT1-MMP-independent propeptide processing of MMP-2 in
HEK293F and various tumor cell lines, where PCs
could mediate this processing Furthermore,
PC-pro-cessed MMP-2 showed enzymatic activity, which was
enhanced following intermolecular autolytic cleavage
Thus, these results strongly suggest a potential role
of PCs in pro-MMP-2 activation in PC-expressing
cells
Results
MT-MMP-independent processing of pro-MMP-2 Previous studies have shown that MT1-MMP is a major activator of pro-MMP-2 [29–31] Furthermore, TIMP-2 and integrin avb3 have been shown to play important roles in MT1-MMP-mediated pro-MMP-2 activation [13–15,32] Thus, prior to investigating the role of MT1-MMP in pro-MMP-2 activation, expres-sion of MT1-MMP, TIMP-2 and integrin avb3 was characterized in COS-1, HCT116, HEK293F, MCF-7, MDAH 2774, K-562, NCI-H460 and Hep G2 cells MT1-MMP protein was undetectable by immunoblot-ting with anti-MT1-MMP rabbit polyclonal IgG in all cell types (data not shown), whereas semi-quantitative RT-PCR detected its mRNA in HCT116, K-562, NCI-H460 and Hep G2 cells (Fig 1A) Because other MT-MMPs have also been demonstrated to process pro-MMP-2 [8–11], their expression was also examined
in these cells Although protein expression was not determined, different levels of MT-MMP mRNA were observed in the cells (Fig 1A) Immunoblotting with mouse anti-TIMP-2 IgG2a showed that COS-1 cells possessed the highest levels of TIMP-2, whereas the expression of this inhibitor was absent or much lower
in HCT116, HEK293F, MCF-7, MDAH 2774, K-562, NCI-H460 and Hep G2 cells (Fig 1B) Variable cell surface expression of integrin avb3 was assessed by flow cytometric analysis (Fig 1C), although its expres-sion was undetectable in HCT116, K-562 and Hep G2 cells (data not shown) These data demonstrate that different cell types have different cellular components for MT-MMP-mediated pro-MMP-2 activation
To test the MT-MMP-independent processing of pro-MMP-2, cells were incubated with metalloprotease inhibitors (e.g GM6001 and TIMP-2), and the condi-tioned medium was analyzed by zymography TIMP-2
is a strong inhibitor of MT1-MMP at high concentra-tions [33], although it activates pro-MMP-2 with MT1-MMP at low concentrations [34] Thus, we used a high concentration of TIMP-2 (5 lgÆmL–1) to completely inhibit the enzymatic activities of MT1-MMP and other TIMP-2-sensitive MMPs The data showed that the processed MMP-2 was clearly seen in the condi-tioned medium of HEK293F, MDAH 2774 and
MCF-7 cells (Fig 2A) However, processing was not affected
by incubation with the inhibitors in these cells (Fig 2A) Likewise, processed MMP-2 was observed
in the conditioned medium of K-562, NCI-H460 and Hep G2 cells, and their incubation with the metallo-protease inhibitors did not result in the prevention of processing even though pro-MMP-2 accumulated
Trang 3slightly in the conditioned medium of K-562 cells
trea-ted with the inhibitors (Fig 2B) As a control, the
con-ditioned medium of concanavalin A (Con A)-treated
HT-1080 cells showed that MT1-MMP-mediated
pro-cessing of pro-MMP-2 was completely inhibited in the
presence of the metalloprotease inhibitors (Fig 2C)
Unlike HT-1080 cells, endogenous MMP-2 expression
in all the cell lines used was low, so that lytic bands
could hardly be seen in the zymograms without
con-centrating the conditioned medium (Fig 2A and not
shown in Fig 2B)
MT-MMPs cleave pro-MMP-2 at the Asn66–Leu
[7–10] or Asn109–Tyr peptide bond [11] Therefore, to
further investigate the role of MT-MMP in
pro-MMP-2 activation, pro-MMP-pro-MMP-2 mutants incapable of
cleav-age by MT1-MMP and⁄ or autolysis were generated
(N66I⁄ L67V, N109I ⁄ Y110F and N66I ⁄ L67V ⁄ N109I ⁄
Y110F) (Fig 3) In agreement with the data using
metalloprotease inhibitors, transient transfection of
these mutants into HEK293F, MCF-7 and
MDAH 2774 cells did not prevent cleavage, as
pro-cessing of the mutants persisted (Fig 4A) By contrast,
processing was increased significantly in the cells
expressing the pro-MMP-2 N66I⁄ L67V and
N66I⁄ L67V ⁄ N109I ⁄ Y110F mutants by unknown mechanisms (Fig 4A) As a control, COS-1 cells expressing the mutants with MT1-MMP showed that the pro-MMP-2 N66I⁄ L67V and N66I ⁄ L67V ⁄ N109I ⁄ Y110F mutants were not cleaved by MT1-MMP, and that the pro-MMP-2 N109I⁄ Y110F mutant was not processed autocatalytically following MT1-MMP cleavage (Fig 4B) Overall, these results suggest the presence of an MT-MMP-independent activation mechanism for pro-MMP-2 in some cell types
PCs mediate propeptide processing of MMP-2
To further explore the MT-MMP-independent activa-tion mechanism of pro-MMP-2, cells were incubated with various inhibitors targeted towards serine (e.g aprotinin, chymostatin, and leupeptin) and aspartyl (e.g pepstatin) proteases, and the conditioned medium was analyzed by zymography As shown in Fig 5A, this processing was not affected by effective concentrations
of these protease inhibitors in HEK293F, MCF-7 and MDAH 2774 cells These data suggest that pro-MMP-2 processing may be mediated by proteases other than serine and aspartyl proteases in these cells
A
C
B
Fig 1 Cellular expression of MT-MMPs,
TIMP-2 and integrin avb3 (A) Analysis of
MT-MMP mRNA in various cell lines The
data represent semi-quantitative RT-PCR
analysis of MT-MMP mRNA in HCT116,
HEK293F, MCF-7, MDAH 2774, K-562,
NCI-H460 and Hep G2 cells PCR products
(25 cycles) were resolved by agarose gel
electrophoresis and visualized by ethidium
bromide staining Arrowheads indicate PCR
products of MT-MMPs or GAPDH (B)
Wes-tern blotting of conditioned medium with
mouse anti-TIMP-2 IgG2a The conditioned
medium from the cells on 12-well plates
was concentrated prior to SDS-PAGE Arrow
indicates TIMP-2 protein (C) Flow
cytometric analysis for cell surface
expression of integrin avb3 The percentage
changes in fluorescence intensity by the
presence of integrin avb3 are shown.
A sample lacking primary antibody was used
as a control (n = 3 representative
experi-ments).
Trang 4A previous study has shown the furin processing
of pro-MMP-2 through cleavage of a consensus PC
cleavage site in the propeptide of MMP-2
(Arg98-Lys-Pro-Arg101) [35] We initially examined the enzymatic
activities of PCs in various cells using an enzymatic
activity assay The results showed PC activities in the
cell lysates from HEK293F, MCF-7, MDAH 2774,
K-562, NCI-H460 and Hep G2, whereas negligible
amounts of PC activity were detected in COS-1 and
HCT116 cells (Fig 5B) Next, we investigated whether
PCs play a crucial role in pro-MMP-2 processing by incubating cells with increasing concentrations of the membrane-permeable PC inhibitor dec-Arg-Val-Lys-Arg-chloromethyl ketone (dec-RVKR-cmk) As shown
in Fig 5C, D, pro-MMP-2 processing was reduced significantly in the PC inhibitor-treated cells, suggesting that PCs are major processing enzymes for pro-MMP-2
in these cell types The cells did not undergo apoptosis
at the doses of inhibitor used, which was confirmed
by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (MTT) assay (data not shown) Next, experi-ments were performed to identify which PCs associated with the constitutive secretory pathway mediated this processing When pro-MMP-2 was expressed in COS-1 and HCT116 cells, mainly the intact MMP-2 zymogen was detectable, whereas the processed form was negli-gible in the conditioned medium (Fig 5E) However, processed MMP-2 was seen in the conditioned medium
of cells co-expressing pro-MMP-2 and PCs (Fig 5E) Parallel experiments performed in HEK293F and MCF-7 cells showed endogenous processing of pro-MMP-2, which was enhanced following the transfection
of PCs (Fig 5E) The substitution of Arg101 with Ala decreased PC processing significantly, as well as the endogenous processing of pro-MMP-2 However, par-tially and fully processed MMP-2 was marginally detect-able in the conditioned medium of the cells expressing the mutant (Fig 5E) As such, PCs may be the major processing enzymes for pro-MMP-2 in these cells
PACE4- and PC5A-expressing cells exhibit intracel-lular and extracelintracel-lular processing of pro-MMP-2 Intracellular processing of pro-MMP-2 by furin has been shown previously in COS-1 cells [35] Despite
A
B
C
Fig 2 MT-MMP-independent processing of pro-MMP-2 in various
cell types (A) Effect of metalloprotease inhibitors on the propeptide
processing of MMP-2 in HEK293F, MDAH 2774 and MCF-7 cells.
Cells were treated with GM6001 (20 l M ) or TIMP-2 (5 lgÆmL –1 ) in
500 lL of 293 SFM-II medium on 24-well plates After 14 h of
incu-bation, MMP-2 was captured from 500 lL of the conditioned
med-ium using gelatin–Sepharose according to the manufacturer’s
recommendations (Amersham Biosciences) and eluted in 30 lL of
SDS sample buffer, and the eluted sample was analyzed by
zymog-raphy; 10 lL of the conditioned medium (pre-column sample,
unconcentrated) was loaded on the first lane Arrows indicate
pro- and processed MMP-2 (MMP-2) (B) Effect of metalloprotease
inhibitors on the processing of pro-MMP-2 in K-562, NCI-H460 and
Hep G2 cells Cells were treated as described above MMP-2 was
concentrated from 125 lL of the conditioned medium using
gela-tin–Sepharose and the eluted sample was used for zymographic
analysis Conditioned medium from HT-1080 cells was used as a
positive control for pro-MMP-2 (C) Validation of the
metallopro-tease inhibitors to prevent MT1-MMP-mediated processing of
pro-MMP-2 HT-1080 cells were treated with Con A (50 lgÆmL –1 ) in
the presence and absence of the metalloprotease inhibitors for
14 h (n = 3 representative experiments).
Fig 3 Structures of pro-MMP-2 mutants.
Trang 5this, the cellular localization of pro-MMP-2 processing
by other PCs, including PACE4 and PC5A, remains
unknown Therefore, whether PACE4- or
PC5A-medi-ated processing of pro-MMP-2 occurs intracellularly
was investigated using a variety of approaches First,
we examined the enzymatic activities of PCs in
condi-tioned medium and cell lysates using the enzymatic
activity assay The results showed PC activities in the
conditioned medium and cell lysates from COS-1 cells
expressing PCs (Fig 6A) We further verified whether
the activitie cell lysates with the PC inhibitor (1 s
shown reflected exclusively the presence of PCs by
treating the conditioned medium and 00 lm) The
enzymatic activities were almost completely inhibited
(more than 98%) in the PC inhibitor-treated samples
(data not shown), excluding the enzymatic activities of
other proteases in the conditioned medium and cell
lysates Next, pro-MMP-2-expressing COS-1 cells were
co-cultured with cells expressing furin, PACE4, PC5A
or MT1-MMP, and zymographic analysis of the
condi-tioned medium was performed Although soluble furin
was detected in the conditioned medium (Fig 6A) and
furin cell surface expression was also confirmed by
flow cytometry and cell surface biotinylation (Fig 6C),
furin-expressing COS-1 cells did not show extracellular
processing of pro-MMP-2 (Fig 6B) Moreover, when
purified pro-MMP-2 was incubated with purified furin
under cell-free conditions, processing did not occur (data not shown) By contrast, the presence of PACE4, PC5A or MT1-MMP resulted in extracellular cleavage
of pro-MMP-2 (Fig 6B) Moreover, substitution of Arg101 with Ala abolished the extracellular processing
of pro-MMP-2 by PACE4 and PC5A completely (Fig 6B), suggesting the direct cleavage of pro-MMP-2
by these enzymes
To further compare PACE4 and PC5A processing
of pro-MMP-2 with that of furin, cell lysates from the co-transfected cells were analyzed by zymography, in which cells were pretreated with trypsin⁄ EDTA to remove MMP-2 from the cell surface (Fig 6E) Cell lysates from furin-expressing cells showed dramatically larger amounts of processed MMP-2 than did those from cells expressing PACE4 or PC5A (Fig 6D) Interestingly, processed MMP-2 was also seen in cell lysates from HEK293F and MCF-7 cells expressing pro-MMP-2 alone (Fig 6D)
To further verify the intracellular localization of PC-processed MMP-2, a FLAG-tagged pro-MMP-2 mutant, in which the FLAG epitope was inserted imme-diately downstream of Arg98-Lys-Pro-Arg101, was gen-erated In these experiments, PC-processed MMP-2 with
a free N-terminal FLAG tag was detected by western blotting with mouse anti-FLAG M1 IgG2b The anti-FLAG M1 IgG2b only recognizes proteins with a free
A
B
Fig 4 (A) Zymograms of conditioned
medium from cells expressing pro-MMP-2
(WT), pro-MMP-2 N66I⁄ L67V mutant
(N66I ⁄ L67V), pro-MMP-2 N109I ⁄ Y110F
mutant (N109I ⁄ Y110F) and pro-MMP-2
N66I⁄ L67V ⁄ N109I ⁄ Y110F mutant
(N66I ⁄ L67V ⁄ N109I ⁄ Y110F) The conditioned
medium was analyzed without
concentra-tion The bar graph shows the ratio of
processed MMP-2 to unprocessed
pro-MMP-2 Arrows indicate pro- and
processed MMP-2 (MMP-2) (B) Zymogram
of conditioned medium from COS-1 cells
expressing pro-MMP-2, pro-MMP-2
N66I⁄ L67V mutant, pro-MMP-2
N109I ⁄ Y110F mutant and pro-MMP-2
N66I⁄ L67V ⁄ N109I ⁄ Y110F mutant with or
without MT1-MMP Note that COS-1 cells
expressing pro-MMP-2 N66I ⁄ L67V and
N66I⁄ L67V ⁄ N109I ⁄ Y110F mutants show
prominent processing of the propeptide
without MT1-MMP expression Arrows
indicate pro-, intermediate and processed
MMP-2 (MMP-2) Data are the means and
standard deviations of n = 3 experiments.
*P < 0.05 versus wild-type.
Trang 6N-terminal FLAG tag These results demonstrated that
furin-processed MMP-2 was recognized specifically by
the anti-FLAG M1 antibody (Fig 7B), whereas
pro-MMP-2 and processed pro-MMP-2 were detected by the
anti-MMP-2 antibody (Fig 7A) Furthermore,
muta-tion of Arg101 to Ala resulted in the complete loss of
FLAG-tagged MMP-2, further supporting the
specific-ity of the anti-FLAG M1 antibody and validating this
experimental approach (Fig 7B) Immunofluorescence
of fixed and permeabilized COS-1 cells stained with
anti-MMP-2 antibody revealed MMP-2 localization to
be mostly intracellular (Fig 7C) Staining with
anti-FLAG M1 antibody also showed the intracellular
local-ization of PC-cleaved MMP-2, even though cells
expressing PACE4 or PC5A stained weakly (Fig 7C)
Cells expressing MMP-2 alone exhibited little staining
with the anti-FLAG M1 antibody (Fig 7C) Further-more, no signal was observed in cells expressing the pro-MMP-2-RKPA(101)-FLAG mutant with PCs, providing additional evidence for the specificity of the anti-FLAG M1 antibody (data not shown) Taken together, these results demonstrate that furin-expressing cells undergo exclusive intracellular processing of pro-MMP-2, whereas cells expressing PACE4 or PC5A exhibit both intracellular and extracellular processing
of pro-MMP-2
Pro101 regulates excessive PC processing of pro-MMP-2
Because Arg98-Lys-Pro-Arg101 in the propeptide of MMP-2 is a minimal recognition motif for PC
cleav-A
C
E
D
B
Fig 5 PC-dependent processing of pro-MMP-2 (A) Effect of various inhibitors
on the propeptide processing of MMP-2
in HEK293F, MCF-7 and MDAH 2774 cells Cells were treated with aprotinin (20 lgÆmL –1 ), chymostatin (10 lgÆmL –1 ), leupeptin (20 lgÆmL –1 ) and pepstatin (2 l M ) After 8 h of incubation, the conditioned media were analyzed by zymography Arrows indicate pro- and processed MMP-2 (MMP-2) (B) Protease activities of PCs in cell lysates from the cells indicated (C) Inhibition of pro-MMP-2 processing by a PC inhibitor dec-RVKR-cmk in HEK293F, MCF-7 and MDAH 2774 cells Cells were incubated with 0–100 l M of the inhibitor for 8 h, and the conditioned medium was analyzed by zymography The bar graph shows the ratio
of processed MMP-2 to pro-MMP-2 in the conditioned medium of cells treated with
100 l M of the PC inhibitor Data are the means and standard deviations of n = 3 experiments *P < 0.05 versus untreated control (D) Inhibition of pro-MMP-2 processing by the PC inhibitor in K-562, NCI-H460 and Hep G2 cells Cells were incubated with 100 l M of the inhibitor Note that almost complete inhibition of
pro-MMP-2 processing is seen in the conditioned medium of the inhibitor-treated cells (E) Zymograms of conditioned medium from cells expressing pro-MMP-2 or pro-MMP-2 R101A mutant with furin, PACE4 or PC5A (n = 3 representative experiments).
Trang 7age [18,19,24], and Pro100 is highly conserved in the
MMP family, we investigated whether this amino acid
residue can play a role in regulating the PC cleavage
of pro-MMP-2 The conditioned medium of cells
expressing the pro-MMP-2 P100K mutant showed
highly increased propeptide processing by PCs
(PC5A > PACE4 > furin), whereas the PC-mediated
intracellular cleavage of the mutant was increased
slightly (Fig 8A) Interestingly, the co-culture of
COS-1 cells expressing individual PCs and cells expressing
the mutant showed more robust processing of the
propeptide by extracellular PACE4 and PC5A, but not
by extracellular furin (Fig 8B) These experimental data suggest that Pro100 plays a critical role in regu-lating excessive pro-MMP-2 processing, especially by extracellular PACE4 and PC5A
PC-cleaved MMP-2 is further processed to achieve full activation
Because PCs cleave pro-MMP-2 immediately upstream
of the Cys102 residue that interacts with the catalytic
A
B
D E
C
Fig 6 Cellular location of PC processing of
pro-MMP-2 (A) Protease activities of PCs in
conditioned medium and cell lysates from
COS-1 cells expressing PCs (B) Co-culture
of COS-1 cells expressing pro-MMP-2 or
pro-MMP-2 R101A mutant with cells
expressing PCs or MT1-MMP The
condi-tioned media were analyzed by zymography.
Note that processed MMP-2 is faintly seen
in the conditioned medium of cells
express-ing pro-MMP-2 alone Arrows indicate
pro-, intermediate and processed MMP-2
(MMP-2) (C) Cell surface localization of furin
in the transfected COS-1 cells Flow
cytom-etry and cell surface biotinylation
approaches were used to detect cell
surface furin Arrowhead indicates furin.
(D) Zymographic analysis of cell lysates
from COS-1, HEK293F and MCF-7 cells
co-expressing pro-MMP-2 and PCs Cells
were pretreated with 0.05%
tryp-sin ⁄ 0.53 m M EDTA for 30 min on ice to
remove cell surface proteins (E) Flow
cyto-metric analysis shows the cell surface
locali-zation of MMP-2 in COS-1, HEK293F and
MCF-7 transfected with pro-MMP-2 Cell
surface MMP-2 (dotted line) is absent in the
cells treated with trypsin ⁄ EDTA (T ⁄ E) (full
line) (n = 3 representative experiments).
Trang 8domain zinc ion, MMP-2 processed in this manner
may not possess catalytic activity Therefore, we
inves-tigated whether PC-cleaved MMP-2 exhibits enzymatic
activity or whether intermolecular autolytic cleavage at
the Asn109–Tyr peptide bond [36] is required for its
activity after initial processing The degradation of
collagen IV and a fluorescein-conjugated gelatin by
processed MMP-2 from the conditioned medium of HEK293F cells expressing pro-MMP-2, the pro-MMP-2 E404A mutant and the pro-MMP-2 N109I⁄ Y110F mutant in the presence and absence of furin was com-pared Because the catalytic glutamic acid residue within the pro-MMP-2 active site is replaced in the E404A mutant, a proteolytically inactive enzyme is
A
C
B
Fig 7 Intracellular localization of PC-cleaved MMP-2 Cell lysate was obtained from COS-1 cells expressing RKPR(101)-FLAG or pro-MMP-2-RKPA(101)-FLAG with or without furin Western blotting of cell lysate was performed using anti-MMP-2 (A) and anti-FLAG M1 (B) antibody Arrowheads indicate furin-processed MMP-2, and arrow indicates uncleaved pro-MMP-2 (C) Confocal microscope imaging of fixed and
permeabilized COS-1 cells stained with anti-MMP-2 (total MMP-2 staining) or anti-FLAG M1 (MMP-2 or FLAG M1 staining
is green) A negative control in which the primary antibody was omitted showed no signal (data not shown) Scale bar, 20 lm.
Trang 9generated Furthermore, the pro-MMP-2 N109I⁄
Y110F mutant showed complete loss of intermolecular
autolytic cleavage following MT1-MMP cleavage
(Fig 4B) Because of concerns about possible
unwanted structural effects of the processed MMP-2
during purification, these experiments were performed
using conditioned medium from cells expressing the
various constructs The concentrations of MMP-2 were
measured by ELISA Substrates were incubated with
an identical amount of MMP-2 and subsequently
ana-lyzed by SDS-PAGE or fluorometry Although the
substrates were degraded proteolytically by MMP-2
that was processed via endogenous routes,
furin-medi-ated cleavage conferred increased proteolytic activity
on MMP-2 (Fig 9) However, the pro-MMP-2
N109I⁄ Y110F mutant showed less proteolytic activity
than the wild-type, even though its enzymatic activity
was also increased by furin-mediated cleavage (Fig 9)
We also obtained similar results with the conditioned
medium from MCF-7 cells expressing the same
constructs (data not shown) However, furin-processed MMP-2 from the conditioned medium of co-transfected COS-1 cells showed much reduced levels of enzymatic activity (e.g 1⁄ 50th to 1 ⁄ 200th of the activity of processed MMP-2 from HEK293F and MCF-7 cells)
by unknown mechanisms (data not shown) Taken together, these data suggest that PC-cleaved MMP-2 is further processed for full enzymatic activity
Discussion
The role of MT1-MMP in pro-MMP-2 activation is well established [7,14], but residual activation of pro-MMP-2 can also be seen in MT1-MMP) ⁄ ) mouse fibroblasts cultured in collagen gel and lung extract [16,17] The residual activation observed might be caused by the presence of other proteases, including metalloproteases (e.g other MT-MMPs) [8–11] and
A
B
Fig 8 Regulation of excessive PC processing of pro-MMP-2 by
Pro100 residue (A) PC processing of pro-MMP-2 and pro-MMP-2
P101K mutant Conditioned medium (CM) and cell lysate from
co-transfected COS-1 cells were analyzed by zymography Data are
the means and standard deviations of n = 3 experiments.
*P < 0.05 versus wild-type (B) Co-culture of COS-1 cells
express-ing the pro-MMP-2 P100K mutant with PC-expressexpress-ing cells The
conditioned medium was analyzed by zymography (n = 3
represen-tative experiments).
A
B
Fig 9 Furin-processed MMP-2 gains its full activity by intermolec-ular autolytic cleavage Digestion assay of collagen IV (A) and a fluo-rescein-conjugated gelatin (B) Conditioned medium was obtained from HEK293F cells expressing pro-MMP-2, pro-MMP-2 E404A mutant and pro-MMP-2 N109I ⁄ Y110F mutant with or without furin The digested substrates were analyzed as described in Materials and methods Zymogram and western blot using anti-MMP-2 show equal amounts of MMP-2 used Arrows indicate pro- and processed MMP-2 (MMP-2) Data are the means and standard deviations of
n = 6 experiments ANOVA test indicates statistically significant dif-ferences in furin versus control and in the N109I ⁄ Y110F mutant versus the wild-type (P < 0.05).
Trang 10serine proteases (e.g urokinase⁄ plasmin, thrombin,
chymase, cathepsin G and trypsin-2) [6,12,37,38]
Thus, we examined MT1-MMP-independent
process-ing of pro-MMP-2 usprocess-ing various approaches, all of
which provided evidence for non-MT1-MMP
compo-nents within cells MT-MMPs have been shown to
process pro-MMP-2 through cleavage of pro-MMP-2
at the Asn66–Leu peptide bond by MT1-, MT2-,
MT3- and MT5-MMP, followed by intermolecular
autolytic cleavage at the Asn109–Tyr peptide bond [7–
10], or directly at the Asn109–Tyr peptide bond by
MT6-MMP [11] Moreover, various MT-MMPs are
expressed in the cell types analyzed in this study
How-ever, studies using metalloprotease inhibitors and
pro-MMP-2 mutants incapable of cleavage by MT-MMPs
excluded MT-MMPs as candidates for pro-MMP-2
processing in these cell lines A previous study has also
shown residual MMP-2 activation in fibroblasts treated
with a general metalloprotease inhibitor [16] These
data, taken together with ours, strongly suggest the
existence of non-MT-MMP components for
pro-MMP-2 processing in various cell types
In this study, we found that PCs such as furin, PACE4
and PC5A mediate the MT-MMP-independent
process-ing of MMP-2 Furin-mediated cleavage of
pro-MMP-2 immediately downstream of R(98)KPR(101)
has been reported previously [35], even though it has a
minimal PC recognition motif [18] Moreover,
thrombin-mediated activation of pro-MMP-2 has been shown to
occur in human endothelial cells at a predicted PC
recognition motif for cleavage [12] Because other
serine proteases, including factor Xa and plasmin, have
also been shown to activate pro-MMP-2 [6,39], it
would be interesting to investigate whether these
prote-ases can activate pro-MMP-2 by cleavage at the same
site If this is the case, the PC recognition motif would
be a general cleavage site for pro-MMP-2 activation
Although the conditioned medium of cells expressing
the PC-uncleavable mutant showed a significant
decrease in pro-MMP-2 processing, lower levels of
processed MMP-2 were also seen MT-MMP
expres-sion may contribute to the residual processing of the
mutant in these cells as the mutant can be processed
more fully following MT1-MMP overexpression PCs
are overexpressed in various cancers, including lung,
breast and skin, with a significant role in tumor
progression [40] Moreover, our studies using
metallo-protease and PC inhibitors showed PCs to be major
enzymes for pro-MMP-2 processing in various tumor
cell lines Thus, PC-mediated processing may be a
major mechanism for the activation of pro-MMP-2 in
PC-overexpressing cells (e.g tumor cell lines), whereas
PCs are likely to act alongside MT-MMPs for
pro-MMP-2 activation in cells expressing high levels of MT-MMPs
Furin is localized predominantly in the TGN, but some is present at the plasma membrane [21,22,25] This enzyme is also secreted by cells and may be func-tional in the extracellular space [41,42] However, despite numerous attempts, furin-mediated extracellu-lar processing of pro-MMP-2 could not be detected, leading to the conclusion that furin processing of pro-MMP-2 is exclusively intracellular This finding contra-dicts that of Cao et al [35], who reported that furin processing was both intracellular and extracellular, with intracellular processing as the predominant mech-anism Unlike furin, secreted PACE4 and PC5A are anchored to the cell surface by heparan sulfate proteo-glycan in the extracellular space despite their residence
in the TGN [23] Likewise, MMP-2 has an affinity for binding heparin via its hemopexin-like domain and anchors to heparan sulfate proteoglycans [43,44] Therefore, the binding property of heparan sulfate pro-teoglycans may lead to their co-localization to the cell surface, facilitating the extracellular processing of pro-MMP-2 However, the inability of extracellular furin
to bind heparan sulfate proteoglycans may prevent its co-localization with pro-MMP-2 at the cell surface and its extracellular processing of pro-MMP-2 It is also likely that, once exported from the cell, pro-MMP-2 undergoes a conformational change, thereby hamper-ing furin processhamper-ing Although cells expresshamper-ing PACE4
or PC5A exhibit both intracellular and extracellular processing of pro-MMP-2, intracellular processing
is likely to be predominant because these PCs are concentrated in the TGN to allow more efficient processing of pro-MMP-2
By contrast with previously published data [35], furin-processed MMP-2 showed enzymatic activity However, when furin-processed MMP-2 was obtained from the conditioned medium of COS-1 cells, its enzy-matic activity was much lower than that obtained from HEK293F and MCF-7 cells Several reasons can explain this observation TIMP-2 was highly expressed
in COS-1 cells Furthermore, TIMP-2 is a strong inhibitor of MMP-2 at high concentrations [45], although it activates pro-MMP-2 with MT1-MMP at low concentrations [34] Therefore, we propose that the lack of enzymatic activity is a result of the presence
of a high concentration of TIMP-2 in the conditioned medium of COS-1 cells This was also supported by our observation that autoproteolytic activation of MT1-MMP-processed MMP-2 was detected marginally
in COS-1 cells (Fig 4B) In this study, we found that furin-processed MMP-2 was further processed to its fully activated form, but the detailed mechanism