THE JOURNAL OF BIOLOGICAL CHEMISTRY
© 2002 by The American Society for Biochemistry and Molecular Biology, Inc.
Vol. 277, No. 50, Issue of December 13, pp. 48696 –48707, 2002
Printed in U.S.A.
Utilization of a Novel Recombinant Myoglobin Fusion Protein
Expression System to Characterize the Tissue Inhibitor of
Metalloproteinase (TIMP)-4 and TIMP-2 C-terminal Domain
and Tails by Mutagenesis
THE IMPORTANCE OF ACIDIC RESIDUES IN BINDING THE MMP-2 HEMOPEXIN C DOMAIN*
Received for publication, September 6, 2002, and in revised form, October 7, 2002
Published, JBC Papers in Press, October 8, 2002, DOI 10.1074/jbc.M209177200
Heidi S.-T. Kai‡, Georgina S. Butler‡, Charlotte J. Morrison‡, Angela E. King‡,
Gayle R. Pelman‡§, and Christopher M. Overall‡§¶
From the ‡Canadian Institute of Health Research Group in Matrix Dynamics and the Department of Oral Biological
and Medical Sciences, Faculty of Dentistry, and the §Department of Biochemistry and Molecular Biology,
Faculty of Medicine, University of British Columbia, Vancouver, British Columbia V6T 1Z3, Canada
Tissue inhibitor of metalloproteinase (TIMP)-4 binds
pro-matrix metalloproteinase (MMP)-2 and efficiently
inhibits MT1-MMP, but unlike TIMP-2 neither forms a
trimolecular complex nor supports pro-MMP-2 activation. To investigate the structural and functional differences between these two TIMPs, the C-terminal domains
(C-TIMP-4 and C-TIMP-2) were expressed independently from their N domains and mutations were introduced into the C-terminal tails. Myoglobin was used as a
novel recombinant fusion protein partner because spectroscopic measurement of the heme Soret absorbance at
408 nm readily enabled calculation of the molar equivalent of the red-colored recombinant protein, even in
complex protein mixtures. Both C-TIMP-4 and C-TIMP-2
bound pro-MMP-2 and blocked concanavalin A-induced
cellular activation of the enzyme. Measurement of kon
rates revealed that the inhibition of MMP-2 by TIMP-4 is
preceded by a C domain docking interaction, but in
contrast to TIMP-2, this is not enhanced by a C-terminal
tail interaction and so occurs at a slower rate. Indeed,
the binding stability of C-TIMP-4 was unaltered by deletion of the C-terminal tail, but replacement with the
tail of TIMP-2 increased its affinity for pro-MMP-2 by
⬃2-fold, as did substitution with the TIMP-2 C-terminal
tail acidic residues in the tail of C-TIMP-4 (V193E/
Q194D). Conversely, substitution of the C-terminal tail
of C-TIMP-2 with that of TIMP-4 reduced pro-MMP-2
binding by ⬃75%, as did reduction of its acidic character
by mutation to the corresponding TIMP-4 amino acid
residues (E192V/D193Q). Together, this shows the importance of Glu192 and Asp193 in TIMP-2 binding to proMMP-2; the lack of these acidic residues in the TIMP-4
C-terminal tail, which reduces the stability of complex
formation with the MMP-2 hemopexin C domain, probably precludes TIMP-4 from supporting the activation of
pro-MMP-2.
* This study was supported by grants from the National Cancer
Institute of Canada and the Canadian Institutes of Health Research.
The costs of publication of this article were defrayed in part by the
payment of page charges. This article must therefore be hereby marked
“advertisement” in accordance with 18 U.S.C. Section 1734 solely to
indicate this fact.
¶ Supported by a Canada Research Chair in Metalloproteinase Biology. To whom correspondence should be addressed: University of British Columbia, 2199 Wesbrook Mall, Vancouver, British Columbia V6T
1Z3, Canada. Tel.: 604-822-2958; Fax: 604-822-3562; E-mail: chris.
[email protected]; Web site: www.clip.ubc.ca.
Matrix metalloproteinases (MMPs)1 are a family of important processing enzymes that can cleave and regulate the activity of an expanding degradome of bioactive molecules (1–5)
as well as degrade extracellular matrix proteins in pathology
(6). Gelatinase A (MMP-2) has been implicated in numerous
biological processes, including the activation of cytokines such
as tumor necrosis factor-␣ (7), transforming growth factor-␤1
(8), and interleukin-1␤ (9). MMP-2 has also been shown to have
anti-inflammatory actions by converting monocyte chemokine
agonists to antagonists (10, 11) and causes loss of protection of
CD4⫹ cells from human immunodeficiency virus-1 infection by
processing stromal cell-derived factor-1␣ (12).2 MMP-2 cleavage of type IV collagen, a major component of basement membranes, is important for tumor cell metastasis and angiogenesis
(14 –16). In view of these diverse and biologically important
functions, it is not surprising that MMPs are under tight regulatory control, both at the transcriptional and post-transcriptional levels (17–19). Post-translational regulation is also pivotally important in regulating proteolytic activity in the
pericellular and extracellular compartments and involves zymogen activation and inhibition by four endogenous inhibitors,
the tissue inhibitor of metalloproteinases (TIMPs) (2). In contrast to the majority of MMPs, MMP-2 is constitutively expressed by a large number of cell types, indicating its importance during normal cellular functions, and is frequently
overexpressed in metastatic tumors or reactive stroma (3, 20).
Hence, activation of the MMP-2 zymogen is a critical control
point in regulating its activity (21).
Pro-MMP-2 can be activated at the cell surface by membrane
type 1 (MT1)-MMP (22–26) and by MT2-MMP (27). The MT1MMP pathway requires TIMP-2 to tether pro-MMP-2 to MT1MMP (21, 23, 24, 26). The inhibitory N-terminal domain of
TIMP-2 binds the MT1-MMP catalytic domain, inhibiting its
activity, and the TIMP-2 C-terminal noninhibitory domain
docks with the pro-MMP-2 hemopexin C domain, generating a
ternary complex (23, 26). Pro-MMP-2 is then activated on the
cell surface after forming a quaternary activation complex with
a free MT1-MMP molecule, a process particularly sensitive to
1
The abbreviations used are: MMP, matrix metalloproteinase; MTMMP, membrane-type matrix metalloproteinase; TIMP, tissue inhibitor of metalloproteinase; BSA, bovine serum albumin; FAB, fluorimetry
assay buffer.
2
K. Zhang, G. A. McQuibban, C. Silva, G. S. Butler, J. B. Johnston,
J. Holden, I. Clark-Lewis, C. M. Overall, and C. Power, submitted for
publication.
48696
This paper is available on line at http://www.jbc.org
Role of TIMP-4 and -2 C-terminal Tails in Binding pro-MMP-2
TIMP-2 and TIMP-4 levels (24, 26, 27), clustering (28 –30), and
MT1-MMP interaction with native collagen (31). In contrast,
the TIMP-2-independent activation of pro-MMP-2 by MT2MMP is only inhibited by TIMP-2 (27). Although yeast twohybrid analysis revealed that the isolated C domain of TIMP-2
stably interacts with the hemopexin C domain of MMP-2 (10,
28), this interaction has not been measured biochemically, nor
has the relative importance of the molecular determinants
critical for interaction with the hemopexin C domain been
characterized in the absence of the N-domain, due in part to the
difficulty in expressing small recombinant protein domains.
TIMP-4 also binds to pro-MMP-2 at the same site as TIMP-2
(26, 32), but it is unable to support pro-MMP-2 activation or to
compete with TIMP-2 in the binding of pro-MMP-2 (26). However, TIMP-4 can inhibit the activation of pro-MMP-2 by inhibiting free MT1-MMP (26, 33) and MT2-MMP (27). Similarly,
TIMP-3 has been shown to bind pro-MMP-2 and can inhibit
MT-MMPs (34), but its ability to support activation by MT1MMP is not reported. Hence, the role of TIMP-4 in binding
pro-MMP-2 remains unknown, and the structural or sequence
constraints that preclude the formation of functional trimolecular complexes remain to be characterized.
TIMP-2, TIMP-3, and TIMP-4 have nine additional C-terminal amino acid residues, termed the C-terminal tail, compared
with TIMP-1 (Table I). Notably, TIMP-1 is the only member of
the TIMP family that cannot interact with pro-MMP-2 through
its C domain (35, 36) or inhibit MT-MMPs through its inhibitory N-domain (24). Mutagenic analysis has identified key
amino acids in the A-B loop of the N-domain of TIMP-2, which
are absent in TIMP-1, accounting for its lack of MT1-MMPinhibitory properties (37). The TIMP-2 C-terminal tail was
proposed to be critical for binding both active and pro-MMP-2
at the same site on the hemopexin C domain (38, 39). However,
this docking site was mapped by mutagenesis to lie at the
junction of modules III and IV on the lower rim of the hemopexin C domain (40). Hence, this indicated that the binding
interaction to pro-MMP-2 differs from the inhibitory complex
formed between TIMP-2 and active MMPs in which TIMPs
adopt an elongated wedge morphology (41).
The C-terminal tails of TIMP-4 and TIMP-2 differ by four
residues, two of which (Glu192 and Asp193) are acidic in
TIMP-2, making this tail more anionic than that of TIMP-4
(Table I). The negatively charged tail is proposed to form salt
bridges with cationic clusters located on the hemopexin C domain of MMP-2 (40). We hypothesized that the absence of the
two anionic residues from the C-terminal tail of TIMP-4 reduces the binding affinity of TIMP-4 for the hemopexin C
domain of MMP-2 compared with TIMP-2, rendering it incapable of participating in pro-MMP-2 activation. After our present
mutagenesis studies were initiated to test this hypothesis, the
very recent report of the three-dimensional structure of
TIMP-2 bound to pro-MMP-2 (42) confirmed the importance of
the seven basic residues in the hemopexin C domain that we
had previously identified by mutagenesis as being important in
the interaction with TIMP-2 (40). Further, the structure revealed the importance of hydrophobic interactions, hydrogen
bond formation and several salt bridges in the C domain of
TIMP-2 that do not involve the C-terminal tail in forming the
large ⬃250-nm2 interface between these two complexed proteins (42). However, the C-terminal two residues of the tail
were not resolved, so any potential salt bridge formation involving Asp193 could not be determined. Notably, the C-terminal tail binds the rim of hemopexin module III, starting at the
interface with module IV (42), at an extended binding site that
had also been previously identified by mutagenesis to be separate from the main binding site (40). This site of interaction of
48697
the C-terminal tail on module III is intriguing; it is spatially
and structurally distinct from the main interaction site of the C
domain of TIMP-2 on the rim of hemopexin module IV (42), but
its relatively small surface area belies the importance ascribed
to the C-terminal tail previously in kinetic (38, 39) and mutagenesis studies (28, 38, 40). Hence, unresolved from the
structural studies is the relative importance played by the
different contact elements of the binding surfaces to the total
binding interaction which may be targeted by new anti-MMP
drugs (18).
To investigate the interactions and differences in the C domains of TIMP-4 and TIMP-2 (C-TIMP-4 and C-TIMP-2) in the
binding and activation of pro-MMP-2, C-TIMP-4 and C-TIMP-2
were expressed independently to avoid any effects of the inhibitory N-domains. To facilitate this, we developed a novel expression system that utilizes the red-colored protein myoglobin
as a fusion partner. Myoglobin has spectrometric properties
that readily enable accurate protein quantitation throughout
protein purification, even in complex mixtures. Differences between C-TIMP-4 and C-TIMP-2 binding of pro-MMP-2 were
explored by mutagenesis. We report that the acidic residues
Glu192 and Asp193 in the C-terminal tail of TIMP-2 are necessary for a stable interaction between TIMP-2 and pro-MMP-2;
their absence from the C-terminal tail of TIMP-4 appears to
preclude formation of a stable trimolecular complex with proMMP-2 and MT1-MMP, so TIMP-4 is unable to promote the
activation of pro-MMP-2.
EXPERIMENTAL PROCEDURES
Materials—TIMP-2-free human pro-MMP-2 and the catalytically inactive mutant pro-MMP-2 E375A (43) were expressed in Timp2⫺/⫺
embryonic fibroblasts and purified as previously described (12, 26).
MMP-2 hemopexin C domain was expressed in Escherichia coli and
purified as described before (44). Human TIMP-2 and TIMP-4 were
expressed in Chinese hamster ovary and baby hamster kidney cells,
respectively, and purified (26). Affinity-purified polyclonal antibodies
used in Western blotting included a rabbit polyclonal antibody raised to
horse heart myoglobin (␣-Mb) and two antipeptide antibodies raised to
the C-terminal tails of human TIMP-4 and TIMP-2, designated ␣-CT4Tail and ␣-CT2-Tail, respectively (26, 28). An affinity-purified rabbit
polyclonal antibody raised to the His6 tag (␣-His6) was used in enzymelinked immunosorbent assays (45). The quenched fluorescence general
MMP substrate ((7-methoxycoumarin-4-yl)acetyl-Pro-Leu-Gly-Leu-[3(2,4-dinitrophenyl)-L-2,3-diaminoproprionyl]-Ala-Arg-NH2) (46) was
supplied by Dr. C. G. Knight (University of Cambridge).
Construction of a Myoglobin Fusion Protein Expression Vector—The
E. coli expression construct, pGYMC-Mb, which expresses high
amounts of recombinant horse heart myoglobin in the bacterial cytosol
(47), was engineered to add at the C terminus of myoglobin a flexible
hinge (Gly-Gly) and linker amino acid residues encoded by a multiple
cloning site. Mutagenesis was performed (47, 48) using the oligonucleotide 5⬘-PGCCTGCAGTCATTAGCTAGC AGGCCT GCGGCCGCCCTGGAAACCCAG-3⬘ to insert NheI, StuI, and NotI restriction sites
(underlined) 3⬘ to the coding sequence for horse heart myoglobin to
make the pGYMC-MbMCS vector. The mutagenesis reaction was confirmed by DNA sequencing. Recombinant horse heart myoglobin (47)
and myoglobin with the C-terminal linker extension was expressed to
validate the new construct.
Cloning of C-TIMP-4 and C-TIMP-2—The coding regions of human
C-TIMP-4 and human C-TIMP-2, beginning with Gly128 and Glu127,
respectively, which lie between the conserved disulfide bonds at the
junction of the TIMP N and C domains, and ending with a stop codon,
were amplified from human TIMP-4 (kindly provided by Dr. Y. E. Shi,
Albert Einstein College of Medicine, New York) (49) and TIMP-2 cDNAs
(generously provided by Prof. D. Edwards, University of East Anglia),
respectively. The following primers were used to add 5⬘ NheI and 3⬘
HindIII (C-TIMP-4) or 3⬘ PstI (C-TIMP-2) restriction sites (underlined):
5⬘C-TIMP-4 (5⬘- CGGGGGGCTAGCGGCTGCCAAATCACC-3⬘); 5⬘C-TIMP-2 (5⬘-CGGGGGGCTAGCGAGTGCAAGATCACGC-3⬘); 3⬘C-TIMP-4 (5⬘-GTCAAGCTTCTAGGGCTGAACGATGTC-3⬘); 3⬘C-TIMP-2
(5⬘- TGCCTGCAGTTATGGGTCCTCGATGTC-3⬘). PCR products were
gel-purified and digested with the appropriate restriction enzymes before ligation; C-TIMP-4 was cloned directly into pGYMC-MbMCS,
48698
Role of TIMP-4 and -2 C-terminal Tails in Binding pro-MMP-2
TABLE I
Alignment of the C-terminal domains of human TIMP-1 to -4
The amino acid sequences shown are from the start of the C domain, the seventh cysteine residue within the TIMPs (Cys127 in TIMP-1; Cys128
in TIMP-2; Cys122 in TIMP-3; Cys129 in TIMP-4), and are displayed using the single letter code. The location of the C-terminal tail is as indicated.
TIMP-1 lacks the C-terminal tail that is present in the other three TIMPs. Conserved residues are indicated as dots except in the consensus, where
dots indicate differing residues.
whereas C-TIMP-2 was cloned into PCRScript (Stratagene) prior to
restriction digestion and cloning into pGYMC-MbMCS. Both clones
were fully sequenced.
Mutagenesis—Residues that were targeted by mutagenesis in the
C-terminal tail sequences of TIMP-4 and TIMP-2 are shown in Table I.
The following sites were mutated in C-TIMP-4 (Fig. 1): K187Stop
(CT4⌬T) to delete the C-terminal tail; V193E/Q194D (CT4⫹q⫺) to swap
in the homologous acidic residues of the tail of TIMP-2; and K187Q/
V190L/V193E/Q194D (CT43 T2) to change the tail of C-TIMP-4 to that
of C-TIMP-2. Mutations made in C-TIMP-2 were Q186Stop (CT2⌬T), to
delete the C-terminal tail (Fig. 1); E192V/D193Q (CT2⌬q⫺) to replace
the acidic residues with the homologous residues of TIMP-4; and
Q186K/L189V/E192V/D193Q (CT23 T4) to change the tail of C-TIMP-2
to that of C-TIMP-4. Amino acid numbering commences at the NH2terminal Cys1 residue of TIMP-4 and TIMP-2. Mutations were made
using the QuickChange威 site-directed mutagenesis kit (Stratagene)
using the oligonucleotides and templates presented in Table II, and all
mutant cDNAs were fully sequenced.
Recombinant Protein Preparation—C-TIMP recombinant proteins
were expressed in the E. coli strain BL21 DE3 Gold. Cultures (4 ⫻ 700
ml) were incubated at 37 °C for 24 h before harvesting. The cells were
lysed, and the inclusion bodies were solubilized in 6 M guanidine-HCl
according to the protocol of Steffensen et al. (45). Hemin (Sigma) was
added to the solubilized protein at 5 mg/ml prior to buffer exchange and
refolding in refolding buffer (55 mM Na2CO3, 45 mM NaHCO2, pH 10.0,
0.02% (w/v) NaN3) and slow step dialysis into 10 mM Tris-HCl, pH 8.0.
To prevent degradation of the fusion proteins, benzamidine-Sepharose
resin (5 ml) (Amersham Biosciences) equilibrated in 10 mM Tris-HCl,
pH 8.0, was added to the soluble refolded protein and incubated overnight at 4 °C to bind proteases. The resin was removed from the protein
solution, and NaCl and benzamidine were added to the supernatant to
0.5 M and 10 mM, respectively. The Ni2⫹-binding properties of myoglobin were utilized to purify the recombinant fusion proteins using a
Ni2⫹-charged chelating Sepharose column (Amersham Biosciences)
with a total column bed volume (Vt) of 30 ml, equilibrated in 10 mM
Tris-HCl, pH 8.0, 0.5 M NaCl, 10 mM benzamidine. The C-TIMPs were
eluted with a 0 –300 mM imidazole gradient in 10 mM Tris-HCl, pH 7.0,
1 mM benzamidine. The C-TIMP-containing eluate was dialyzed into 10
mM Tris-HCl, pH 7.0, 1 mM benzamidine prior to chromatography on
CM Sepharose (Amersham Biosciences; Vt ⫽ 5 ml) equilibrated in the
same buffer. Elution was accomplished with a 0 –500 mM NaCl gradient
in 10 mM Tris-HCl, pH 8.0, 1 mM benzamidine. The fractions containing
pure recombinant protein were pooled and dialyzed into 50 mM TrisHCl, pH 8.0, 0.15 M NaCl and then snap frozen in liquid nitrogen for
storage at ⫺70 °C. Protein concentrations were determined by analysis
of the Soret absorbance at 408 nm using a standard curve of known
amounts of recombinant horse heart myoglobin (47). The molar equivalent of C-TIMP was quantified throughout purification, even in impure
fractions, on the basis of the amounts of heme in the fusion proteins (1:1
molar ratio) calculated from the Soret peak of the recombinant myoglobin fusion protein samples.
Electrophoresis—The recombinant proteins were electrophoresed on
15% polyacrylamide gels and visualized by Coomassie Brilliant Blue
R-250 or silver nitrate staining and by Western blotting using the ␣-Mb,
␣-CT4-Tail, or ␣-CT2-Tail antibodies and enhanced chemiluminescence. Samples analyzed by zymography were electrophoresed on 10%
polyacrylamide gels copolymerized with 320 ␮g/ml or 1 mg/ml gelatin
(Bio-Rad) and developed as previously described (50).
Mass Spectrometry and Absorbance Spectrometry—To measure the
masses of the recombinant proteins, electrospray ionization time of
flight mass spectrometry was performed (11). To confirm the absence of
apomyoglobin in the recombinant protein preparations, spectrophotometric analysis was performed to measure the Soret peak at 408 – 409
nm. Excess hemin was added, and the wavelength analysis was repeated. The absence of changes in the amplitude of the Soret peak
verified that all of the myoglobin was fully reconstituted with heme and
so the Soret absorbance could be used to quantitate protein quantities.
Hemopexin C Domain Binding Assays—The binding interaction between full-length TIMPs or C-TIMPs and MMP-2 hemopexin C domain
was measured by a microwell plate enzyme-linked immunosorbant
assay (32, 40). TIMPs and C-TIMPs (0.2 ␮g/well) were immobilized in
wells, and the MMP-2 hemopexin C domain was serially diluted in
Tris-buffered saline (50 mM Tris-HCl, pH 8.0, 0.15 M NaCl). Binding of
MMP-2 hemopexin C domain to TIMPs or C-TIMP domains was quantified using a rabbit polyclonal ␣-His6 antibody, which was detected
using a goat anti-rabbit IgG antibody conjugated with alkaline phosphatase, developed with p-nitrophenyl phosphate substrate, and analyzed on a microplate reader at 405 nm. Curve fitting of the amount of
protein bound was performed using SigmaPlot (45).
Binding to Active MMP-2—Binding of C-TIMPs to the hemopexin C
domain of active TIMP-2-free MMP-2 was measured by competition
with TIMPs; pro-MMP-2 was activated with 2 mM 4-aminophenylmercuric acetate and active site-titrated against a standard preparation of
TIMP-1 (kindly provided by Prof. G. Murphy, University of East Anglia,
Norwich, UK) in fluorimetry assay buffer (FAB; 100 mM Tris-HCl, pH
7.5, 10 mM CaCl2, 100 mM NaCl, 0.05% (v/v) Brij-35). TIMP-4 and
TIMP-2 were prepared and active site-titrated as described before (26).
The association rate constant (kon) was measured for TIMP-4 (380 pM)
Role of TIMP-4 and -2 C-terminal Tails in Binding pro-MMP-2
48699
TABLE II
Templates and mutagenic oligonucleotide primers used to generate C-TIMP mutant constructs
Mutant construct
CT4⌬T
CT4⫹q
MbCT4
⫺
MbCT4⫹q⫺
CT23T4
MbCT2⌬q⫺
CT2⌬T
MbCT2
CT2⌬q
a
MbCT4
CT43T2
⫺
Primera
Template
MbCT2
5⬘-GCCACCTGCCTCTCAGGTAGGAGTTTGTTGACATC-3⬘
5⬘-GATGTCAACAAACTCCTACCTGAGAGGCAGGTGGC-3⬘
5⬘-GAAGGAGTTTGTTGACATCGAGGACCCCTAGAAGCTTGGCACTG-3⬘
5⬘-CAGTGCCAAGCTTCTAGGGGTCCTCGATGTCAACAAACTCCTTC-3⬘
5⬘-CCTGCCTCTCAGGCAGGAGTTTCTTGACATCGAGG-3⬘
5⬘-CCTCGATGTCAAGAAACTCCTGCCTGAGAGGCAGG-3⬘
5⬘-CGCCCCCCAAGAAGGAGTTGTCGACATCGTG-3⬘
5⬘-CACGATGTCGACAAACTCCTTCTTGGGGGGCG-3⬘
5⬘-GCGCGGCGCCCCCCAAGTAGGAGTTTCTCGACATC-3⬘
5⬘-GATGTCGAGAAACTCCTACTTGGGGGGCGCCGCGC-3⬘
5⬘-GGAGTTTCTCGACATCGTGCAGCCATAACTGCAGGCATG-3⬘
5⬘-CATGCCTGCAGTTATGGCTGCACGATGTCGAGAAACTCC-3⬘
The mutated nucleotides are shown in boldface and italics.
or TIMP-2 (380 pM) with MMP-2 (15 pM) at 25 °C in FAB using 1 ␮M
quenched fluorescence substrate as detailed previously (26). Increasing
concentrations of C-TIMP-4 or CT4⌬T were added to TIMP-4 prior to
incubation with active MMP-2. Likewise, C-TIMP-2 or CT2⌬T was
premixed with TIMP-2 before assay.
Binding to Pro-MMP-2—Apparent binding affinities of pro-MMP-2 to
the C-TIMPs were determined by an enzyme capture assay with the
bound MMP-2 being measured by the cleavage of gelatin in the linear
range of zymograms and quantitated by densitometry. TIMP-4,
TIMP-2, C-TIMP proteins, or the control proteins, recombinant myoglobin and ovalbumin (Sigma), were immobilized onto 96-well high protein
binding fluorimetry plates (Dynex Microfluor威 2) at a concentration of
0.2 ␮g/well, prior to blocking with 1% bovine serum albumin (BSA).
Following incubation with TIMP-2-free pro-MMP-2 in FAB, the unbound pro-MMP-2 was removed by thorough washing with Tris-buffered saline, 0.5% Tween 20. The bound pro-MMP-2 was then eluted in
nonreducing SDS-PAGE sample buffer and analyzed by zymography.
Where saturable binding was achieved, binding affinities of pro-MMP-2
were determined by measuring enzymic activity against protein
amounts within the sensitivity limits of the assay. For weak binding
mutants, the order of relative binding affinities of the different C-TIMP
proteins was determined from amounts of enzyme captured in the
linear response range of the assay.
Chemical Cross-linking—TIMP-4 and TIMP-2 were incubated with
MMP-2 hemopexin C domain at 6:1, 4:1, 3:1, 2:1, and 1:0 molar ratios of
TIMP to MMP-2 hemopexin C domain for 1 h at room temperature and
cross-linked with 0.5% glutaraldehyde as previously described (10). The
TIMP-4䡠MMP-2 complex was identified by Western blotting using the
␣-CT4-Tail antibody. C-TIMP-4, C-TIMP-2, CT4⌬T, CT2⌬T, and myoglobin were also incubated with or without pro-MMP-2 at 2.5:1 molar
ratios and cross-linked as above. The reactions were analyzed by SDSPAGE, silver staining, and zymography.
Velocity Sedimentation—Velocity sedimentation was performed using a protocol modified from Loewen and Molday (51). Samples consisted of 4.7 ␮g of TIMP-4, 4.7 ␮g of TIMP-2, 4.9 ␮g of C-TIMP-4, or 5.0
␮g of C-TIMP-2 incubated for 7 h at 4 °C alone or with 2 mol equivalents
(8.5 ␮g) of the catalytically inactive mutant pro-MMP-2 E375A. Marker
proteins (1 ␮g of myoglobin (18.8 kDa), carbonic anhydrase (29 kDa),
ovalbumin (45 kDa), BSA (66 kDa), and phosphorylase B (97 kDa)) were
added to the samples, which were then applied to a 5–20% (w/v) sucrose
gradient in 50 mM Tris-HCl, pH 8.0, 0.15 M NaCl, 0.05% Brij-35. Due to
similarity in molecular mass, carbonic anhydrase was omitted from
samples containing C-TIMP, and BSA was omitted from samples containing pro-MMP-2. The samples were centrifuged at 50,000 rpm for
16 h at 4 °C in a Beckman TLS-55 rotor. Fractions were collected as
previously described (51), analyzed by SDS-PAGE, and quantitated by
densitometry.
Cell Assays—Early passage human gingival fibroblasts were seeded
into 96-microwell plates at 1 ⫻ 104 cells/well in Dulbecco’s modified
Eagle’s medium (Invitrogen) containing 10% Cosmic calf serum (Hyclone, Inc.). After 24 h, the cells were washed twice with phosphatebuffered saline (8 mM Na2HPO4, 1.5 mM KH2PO4, pH 7.4, 137 mM NaCl,
2.7 mM KCl), and the cultures were synchronized by incubation in
serum-free Dulbecco’s modified Eagle’s medium for 24 h before adding
up to 4.5 ␮M C-TIMP-4, C-TIMP-2, or myoglobin (as a control) in
serum-free Dulbecco’s modified Eagle’s medium containing 20 ␮g/ml
concanavalin A to induce the cellular activation of MMP-2 (50). Control
cells were incubated in the absence of C-TIMP proteins plus or minus
concanavalin A. Cells were incubated overnight at 37°C, and then the
culture supernatants were harvested and analyzed by zymography.
RESULTS
Analysis of C-TIMP Fusion Proteins—E. coli producing the
myoglobin fusion proteins exhibited a rust red color. The recombinant proteins were located in inclusion bodies, which
were solubilized under denaturing conditions and required refolding in the presence of heme to reconstitute the myoglobin
fusion partner. The C-TIMPs and mutants (see Fig. 1) were
purified to homogeneity as revealed by single protein bands
that electrophoresed at ⬃28 kDa (Figs. 2 and 3). No disulfide
cross-linked multimers were present, as shown by nonreducing
SDS-PAGE analysis (Fig. 2). Electrospray ionization time of
flight mass spectrometry analysis confirmed the fidelity of gene
expression and protein translation and revealed that all CTIMP recombinant proteins were present as both N-terminal
methionine-processed and unprocessed forms (Table III). The
identities of the C-TIMP recombinant proteins were confirmed
by Western blotting with the ␣-Mb, ␣-CT4-Tail, and ␣-CT2-Tail
antibodies (Fig. 3, A and B). All C-TIMPs could be tracked
during purification due to the red color of the myoglobin fusion
partner (Fig. 4A). When analyzed by scanning spectrophotometry, the myoglobin C-TIMPs exhibited a characteristic Soret
absorbance at 408 nm upon refolding, indicative of a protein
complexed with heme (47). The addition of 1 mol equivalent of
heme to reconstituted and refolded myoglobin C-TIMPs did not
significantly affect the Soret absorbance but increased the absorbance at 380 nm where free heme absorbs (Fig. 4B). This
demonstrated that the myoglobin fusion partner was completely reconstituted in the holo form after heme addition during refolding. Protein that had been refolded in the absence of
heme did not display a Soret peak and was a light straw color
(data not shown). Therefore, the Soret absorbance could be
used for quantitation of the recombinant proteins, even in
complex mixtures, using the calculated extinction coefficient of
149,415 M⫺1 cm⫺1.
Characterization of C-TIMP-4 and C-TIMP-2 Tail Mutant
Proteins and Antibodies—To analyze the role of the nine residues of TIMP-4 and TIMP-2 that comprise the C-terminal tail
in binding pro-MMP-2, several mutations were made in both
recombinant C-TIMP-4 and C-TIMP-2. Silver staining and
Western blot analysis under reducing (Fig. 3, A and B) and
nonreducing (data not shown) conditions confirmed the presence of the tail swap and deletion mutations in the expressed
proteins. The ␣-CT4-Tail antibody recognized TIMP-4,
C-TIMP-4, and CT23 T4, whereas the ␣-CT2-Tail antibody
recognized TIMP-2, C-TIMP-2, CT43 T2, and CT4⫹q⫺. The
absence of recognition of C-TIMP-4 and CT2⌬q⫺ by the ␣-CT2Tail antibody, but its ability to recognize CT4⫹q⫺ indicated
that the specificity of this antibody is dominated by the acidic
residues Glu192 and Asp193. The absence of recognition for
CT4⫹q⫺ by the ␣-CT4-Tail antibody indicated that the epitope
for this antibody includes the residues Val193 and Gln194 at the
48700
Role of TIMP-4 and -2 C-terminal Tails in Binding pro-MMP-2
FIG. 3. SDS-PAGE and Western blot analysis of the C-TIMP
mutant proteins. The C-TIMP recombinant proteins, TIMP-2,
TIMP-4, and horse heart myoglobin (Mb) (100 ng) were reduced with 65
mM dithiothreitol, electrophoresed on 15% polyacrylamide gels, and
analyzed by silver staining (upper panel) and Western blotting (lower
three panels) with antibodies raised against peptides corresponding to
the C-terminal tails of TIMP-2 (␣-CT2-Tail) or TIMP-4 (␣-CT4-Tail) or
an antibody raised against recombinant horse heart myoglobin (␣-Mb).
The positions of molecular mass markers are indicated. A, C-TIMP-2
(CT2) variant proteins compared with myoglobin, full-length TIMP-2,
and TIMP-4. B, C-TIMP-4 (CT4) variant proteins compared with myoglobin and full-length TIMP-2 and TIMP-4.
FIG. 1. Schematic representation of the C-TIMP domain recombinant proteins. The hatched section represents horse heart myoglobin (Mb). The white sections represent the C domain of TIMP-2. The
black sections represent the C domain of TIMP-4. The amino acid
sequences shown represent the C-terminal tail and the mutated residues introduced.
FIG. 2. SDS-PAGE analysis of C-TIMP-2 and C-TIMP-4.
C-TIMP-2 (CT2), C-TIMP-4 (CT4), TIMP-2, TIMP-4, and horse heart
myoglobin (Mb) (100 ng) were electrophoresed on a 15% polyacrylamide
gel with or without reduction with dithiothreitol (DTT) and silverstained. The positions of migration of molecular mass markers are as
indicated.
homologous positions of the acidic residues in TIMP-2. However, because the ␣-CT4-Tail antibody also did not recognize
CT2⌬q⫺, which contains Val193 and Gln194, this suggested that
the epitope for the ␣-CT4-Tail antibody includes flanking residues of the C-terminal tail of TIMP-4.
C-TIMP-4 and C-TIMP-2 Bind to the MMP-2 Hemopexin C
Domain—Binding of the C domain of TIMP-4 and TIMP-2 in
the absence of the N-domain to the MMP-2 hemopexin C domain has not previously been shown. In a solid phase binding
assay, both C-TIMP-4 and C-TIMP-2 bound the MMP-2 hemopexin C domain to saturation (Fig. 5). TIMP-4 and
C-TIMP-4 exhibited similar binding affinities, whereas
C-TIMP-2 exhibited reduced binding to the MMP-2 hemopexin
C domain by an order of magnitude when compared with fulllength TIMP-2.
To confirm that the C-TIMPs could bind the MMP-2 hemopexin C domain in the context of the full-length proenzyme,
we examined these interactions in solution by glutaraldehyde
cross-linking, because solid phase assays may denature proteins coated on plastic. C-TIMP-2 (Fig. 6A) or C-TIMP-4
(Fig. 6B) formed complexes with pro-MMP-2 as revealed by
C-TIMP/pro-MMP-2 heterodimeric bands corresponding to a
combined mass of the enzyme and recombinant protein (⬃98
kDa). Zymography identified MMP-2 in the heterodimer bands,
which were shifted to a higher apparent molecular weight as
compared with the samples incubated in the absence of glutaraldehyde or MMP-2 incubated with buffer alone. The fuzziness of the MMP-2 bands in the samples containing the crosslinker is probably due to multiple coupling and intramolecular
cross-linkage, which reduced the activity of the samples. When
myoglobin was incubated with pro-MMP-2 prior to cross-linking as a control (Fig. 6C), no heterodimeric bands were generated, and there was no change in the apparent molecular
weight of MMP-2, confirming the specificity of the interaction
observed with the myoglobin C-TIMP fusion proteins. As positive controls, full-length TIMP-2 (Fig. 6D) or TIMP-4 (Fig. 6E,
upper panel) incubated with MMP-2 hemopexin C domain were
glutaraldehyde cross-linked, resulting in TIMP/MMP-2 hemopexin C domain heterodimer bands. Increasing amounts of
hemopexin C domain added to the TIMP proteins before
cross-linking resulted in a decrease in TIMP monomer and an
increase in TIMP/hemopexin C domain heterodimers. The
identity of the TIMP-4/MMP-2 hemopexin C domain band
was confirmed by Western blotting with ␣-CT4-Tail antibody
(Fig. 6E, lower panel).
To examine the interaction of the C-TIMP domain with proMMP-2 in solution in the absence of cross-linker, velocity sedimentation was performed. Like TIMP-2 (Fig. 7A) and TIMP-4
(Fig. 7B), the peak elution fractions of C-TIMP-2 (Fig. 7C) and
C-TIMP-4 (Fig. 7D) shifted, corresponding to an increased molecular weight, when incubated with the catalytically inactive
mutant pro-MMP-2 E375A, indicating complex formation. This
inactive mutant was employed to avoid autoactivation and
autodegradation of pro-MMP-2 over the long incubation and
centrifugation time employed in this method.
C-TIMP-4 and C-TIMP-2 Inhibit MMP-2 Activation—The
binding assays showed that the C domain of TIMP-4 and
Role of TIMP-4 and -2 C-terminal Tails in Binding pro-MMP-2
48701
TABLE III
Molecular mass determination of recombinant C-TIMP fusion proteins and myoglobin by electrospray ionization
time of flight mass spectrometry
⌬ Mass from Predicted2
Measured mass
Protein
⫹ N-terminal methlonine
⫺ N-terminal methionine
a
⫹ N-terminal methionine
Da
Recombinant Mb
CT4
CT4⌬T
CT4⫹q⫺
CT43T2
CT23T4
CT2
CT2⌬T
CT2⌬q⫺
17,086
25,658
24,598
25,672
25,688
25,502
25,534
24,448
25,520
⫺ N-terminal methionine
Da
16,954
25,526
24,472
25,542
25,556
25,372
25,400
24,312
25,388
⫹2
⫹4
⫹1
⫹1
⫹3
⫹4
⫹5
⫹7
⫹8
⫹2
⫹4
⫹6
⫹3
⫹3
⫹6
⫹3
⫹2
⫹8
a
All the proteins registered m/z peaks corresponding to protein with and without the N-terminal methionine indicating heterogeneous
processing.
b
Predicted mass was calculated from the amino acid sequence minus 2 Da per disulfide bond.
FIG. 4. Spectrophotometric properties of the myoglobin
C-TIMP fusion proteins. A, samples of the mutant protein CT4⫹q⫺
eluted from Ni2⫹-chelating Sepharose, representative of all C-TIMP
proteins, showing in the color figure that the recombinant C-TIMPs can
be visually tracked during purification due to the red coloring of the
horse heart myoglobin fusion protein. B, following heme reconstitution
and purification, spectrophotometric analysis of the C-TIMP-myoglobin
fusion protein (CT2⌬T, 1.32 nmol) revealed that the purified protein
exhibited a characteristic Soret absorbance at 408 nm, confirming the
presence of holomyoglobin (thin black line; Purified C-TIMP). The addition of a further mol equivalent of heme (1.32 nmol) did not significantly change the Soret absorbance, but it did increase the absorbance
around 380 nm, the absorption wavelength for free heme (thick colored
line; ⫹ 1.32 nmole heme). This indicates the presence of excess free
heme and that the protein was essentially all in the holo form with no
apomyoglobin fusion protein present. The analysis shown for CT2⌬T
was typical for all recombinant proteins.
TIMP-2 can bind to the MMP-2 hemopexin C domain in the
absence of the TIMP N-domain, and this was confirmed using
MMP-2 in the full-length zymogen form. To investigate
whether this binding interaction affected pro-MMP-2 activation, C-TIMP-4 or C-TIMP-2 were added to concanavalin Astimulated human fibroblasts. Above 1 ␮M, both C-TIMP-4 and
C-TIMP-2 inhibited the cellular activation of pro-MMP-2
(Fig. 8), whereas the control protein, myoglobin, had no effect.
This suggests that at high concentrations, the C-TIMP proteins
could compete for the low levels (⬍100 nM, quantified by Western analysis) of endogenous TIMP-2 binding to pro-MMP-2 and
so prevent formation of the ternary activation complex.
The Effect of the C-TIMP Tail on MMP-2 Binding—C-
FIG. 5. Binding of C-TIMP-4 and C-TIMP-2 to MMP-2 hemopexin C domain. Full-length TIMP-4 and C-TIMP-4 (A) or fulllength TIMP-2 and C-TIMP-2 (B) were coated onto a 96-well enzymelinked immunosorbent assay plate, blocked with 1% BSA, and overlaid
with a serial dilution of MMP-2 hemopexin C domain. Bound MMP-2
hemopexin C domain was quantified using an affinity-purified rabbit
polyclonal ␣-His6 antibody, which was detected using a goat anti-rabbit
IgG antibody conjugated with alkaline phosphatase and developed with
p-nitrophenyl phosphate substrate.
TIMP-4 without the tail, CT4⌬T (Fig. 9A), and C-TIMP-2 without the tail, CT2⌬T (Fig. 9B), were found to complex with
pro-MMP-2 by chemical cross-linking. Binding was then quantitated using a competition assay as follows. Increasing concentrations of C-TIMP proteins were added with a constant
amount of TIMP-4 or TIMP-2 to active MMP-2, and the rate of
association (kon) was measured. Increasing the concentration of
C-TIMP-4 from 0 to 157 nM reduced the kon of TIMP-4 for
MMP-2 from 4.51 ⫻ 106 M⫺1 s⫺1 to 1.85 ⫻ 106 M⫺1 s⫺1 (Fig.
10A). Similarly, the kon for TIMP-2 with MMP-2 was reduced
from 4.97 ⫻ 106 M⫺1 s⫺1 to 1.33 ⫻ 106 M⫺1 s⫺1 at 120 nM
C-TIMP-2 (Fig. 10B). The steady state rate (vs) also increased
with increasing concentrations of C-TIMPs (data not shown).
The C-TIMP proteins alone did not inhibit MMP-2 proteolytic
48702
Role of TIMP-4 and -2 C-terminal Tails in Binding pro-MMP-2
FIG. 6. Demonstration of C-TIMP/pro-MMP-2 interactions in solution using glutaraldehyde cross-linking. C-TIMP-2 (CT2) (A),
C-TIMP-4 (CT4) (B), or horse heart myoglobin (Mb) (C) were incubated with pro-MMP-2 (0.25 mol equivalents) for 1 h prior to the addition of
glutaraldehyde (GA) to 0.5%. After 30 min, the reaction was stopped with an equal volume of 2⫻ SDS-PAGE sample buffer and analyzed on
SDS-polyacrylamide gels (10%) stained with silver nitrate. A and B, pro-MMP-2 incubated with C-TIMP-2 or C-TIMP-4 formed pro-MMP-2/CTIMP heterodimer bands that were identified as containing MMP-2 in the zymogram. C, pro-MMP-2 incubated with myoglobin in the presence of
glutaraldehyde did not form heterodimer bands as analyzed by either silver-stained gels or zymograms. The myoglobin monomer band was
electrophoresed off the gel. Small amounts of C-TIMP-4, C-TIMP-2, and myoglobin homodimers were sometimes detected by silver staining after
extended incubation times at these concentrations. Full-length TIMP-2 (D) and TIMP-4 (E) were incubated with MMP-2 hemopexin C domain
(HexCD) and cross-linked as described. The samples were analyzed on 15% polyacrylamide gels. For MMP-2 hemopexin C domain, ⫹ denotes 0.25
mol equivalent, ⫹⫹ denotes 0.5 mol equivalent, and ⫹⫹⫹ denotes 1 mol equivalent versus TIMP-2 and TIMP-4. The positions of migration of
molecular mass markers (M) are indicated.
activity. This indicates that C-TIMP-4 and C-TIMP-2 bind to
the hemopexin C domain of active MMP-2, preventing the C
domain docking interaction of full-length TIMP-4 and TIMP-2,
respectively. Hence, C-TIMP binding retards the interaction of
full-length TIMP to the catalytic domain, with similar effects
observed for CT4⌬T and CT2⌬T.
In the enzyme capture assay at identical pro-MMP-2 concentrations, TIMP-4 captured ⬃80% less enzyme than TIMP-2
(Fig. 11A), consistent with the Kd values for TIMP-4 and
TIMP-2 (32, 40). Deletion of the C-terminal tail of C-TIMP-2
greatly reduced binding to pro-MMP-2 in this assay, with
CT2⌬T having an apparent Kd ⬎⬃10⫺6 M (Fig. 11, B and C). In
Role of TIMP-4 and -2 C-terminal Tails in Binding pro-MMP-2
FIG. 7. Demonstration of C-TIMP domain/pro-MMP-2 interactions using velocity sedimentation. The catalytically inactive mutant pro-MMP-2 E375A (pE375A) (0.12 nmol) was incubated alone or
with 0.22 nmol of TIMP-2 (A), TIMP-4 (B), C-TIMP-2 (C), or C-TIMP-4
(D), or TIMPs and C-TIMPs were incubated in the absence of pE375A.
Samples plus markers were then layered on a 5–20% sucrose gradient
and centrifuged for 16 h at 4 °C. The protein layers were collected by
needle perforation and then analyzed by SDS-PAGE on 10% polyacrylamide gels and silver staining. Densitometric analysis was performed
on the protein bands in the elution fractions and plotted with arbitrary
units. The arrows indicate the peak elution fraction of co-incubated
protein markers, which also confirms the specificity of the interaction
between the TIMP and MMP-2 proteins.
FIG. 8. Recombinant C-TIMP-4 and C-TIMP-2 block MMP-2 activation by concanavalin A-stimulated fibroblasts. Human gingival fibroblasts were cultured for 24 h in serum-free medium containing
4.5 ␮M C-TIMP-2 (CT2), C-TIMP-4 (CT4), or horse heart myoglobin
(Mb) (n ⫽ 4 for each treatment) in the presence or absence of 20 ␮g/ml
concanavalin A as indicated. Controls were cultured in the absence of
recombinant protein. The cell culture supernatants were analyzed by
gelatin zymography (10% polyacrylamide gels). The proenzyme (pro),
activation intermediate (intermediate), and active (active) MMP-2 enzyme bands are indicated.
contrast, deletion of the C-terminal tail of C-TIMP-4 had only a
minor effect on its binding affinity (Fig. 11C). Exchanging the
tails of C-TIMP-4 and C-TIMP-2 showed the importance of the
C-terminal tail sequence; CT43 T2 resulted in increased binding of pro-MMP-2 to levels similar to that found for C-TIMP-2
under identical conditions (Fig. 11, B and C). In contrast,
replacing the tail of C-TIMP-2 with that of TIMP-4 (CT23 T4)
markedly reduced the amount of pro-MMP-2 captured (apparent Kd of ⬎⬃10⫺6 M) to similar levels seen for the CT2⌬T
protein described above. Hence, the C-terminal tail of TIMP-4
plays a less important role in binding to the MMP-2 hemopexin
C domain docking site than the TIMP-2 C-terminal tail.
The Critical Role of Glu192 and Asp193 in the TIMP-2 Cterminal Tail—To define in molecular detail the important
elements in the TIMP-2 C domain interaction that strengthen
binding compared with TIMP-4, the role of Glu192 and Asp193 in
the C-terminal tail was examined. Mutating V193E and Q194D
in the C-TIMP-4 variant protein CT4⫹q⫺ (Fig. 1) increased
pro-MMP-2 binding (apparent Kd of ⬃2 ⫻ 10⫺9 M) and approached levels found for C-TIMP-2 in this assay (apparent Kd
of ⬃1 ⫻ 10⫺9 M) (Fig. 11B). The importance of Glu192 and
Asp193 in the C-terminal tail of TIMP-2 was confirmed by their
48703
replacement with the corresponding residues of the C-terminal
tail of TIMP-4 (CT2⌬q⫺). Like replacement of the entire Cterminal tail with that of TIMP-4, loss of these negative
charges reduced the amount of pro-MMP-2 captured to levels
similar to that found for C-TIMP-4 (Fig. 11B). Because assay
sensitivity set detection limits in these experiments, the rank
order of binding affinities of the variant proteins was established by comparison of the amount of pro-MMP-2 bound at the
same protein levels in the linear response range of the assay in
experiments performed under identical conditions (Fig. 11C).
Consistent results were obtained when the zymograms were
quantified at other values in the linear range of the assay. The
relative binding affinities of the C-TIMP variant proteins were
in accordance with the apparent Kd values that could be determined. Hence, these mutagenesis studies demonstrated that
the C-TIMP domains can still bind to pro-MMP-2 independent
of the C-terminal tail. However, unlike TIMP-4, where the
C-terminal tail plays a lesser role in this interaction, in TIMP-2
the tail greatly enhances binding stability. In particular, our
studies revealed the pivotal importance of the unique acidic
residues at positions 192 and 193 in stabilizing the TIMP-2/
pro-MMP-2 interaction.
DISCUSSION
The different hemopexin C domain and TIMP interactions
that occur on forming the noninhibitory complexes of TIMP-4
and TIMP-2 with pro-MMP-2 compared with those made in the
binding and inhibition of active MMP-2 render interpretation
of TIMP binding studies complex. The MMP-2, TIMP-2, and
MT1-MMP ternary complex has been extensively studied by
many groups (21). The catalytic domain of pro-MMP-2 alone is
unable to bind to full-length TIMP-2 (24), and since N-TIMP-2
inhibits MT1-MMP and MMP-2 in the absence of the TIMP-2 C
domain (37–39) but does not support MMP-2 activation, it was
concluded that the TIMP-2 C domain binds to the hemopexin C
domain of pro-MMP-2. These deletion mutant studies were
supported by MMP-2 hemopexin C domain binding studies (28,
34 –36), mapping the TIMP-2 docking site on the MMP-2 hemopexin C domain by mutagenesis (40), yeast two-hybrid analysis (10, 28), and very recent crystallographic studies (42).
TIMP-4 was also found to bind the TIMP-2 docking site of
pro-MMP-2 (32), but when so bound cannot also bind the catalytic domain of MT1-MMP, despite its potent inhibitory properties for this MMP (26) and so cannot form a ternary complex.
The TIMP C domains had not previously been expressed and
characterized in the absence of the N-domain, which would
simplify the analysis of TIMP binding interactions; nor had the
differences between TIMP-4 and TIMP-2 been explored by mutagenesis to define the crucial molecular determinants that
form the pro-MMP-2 binding site and that determine binding
affinity. Therefore, to dissect the differences in the domain
interactions that occur in the activation and inhibition of
MMP-2, the C domains of TIMP-4 and TIMP-2 were expressed
with an N-terminal horse heart myoglobin fusion protein in
E. coli. Due to myoglobin’s red color, which results from the
bound heme cofactor, the recombinant fusion proteins could be
visually tracked and quantified spectrophotometrically by
measuring the Soret absorbance at 408 nm. This is a decided
advantage over fusion partners such as ␤-galactosidase, which
cannot be monitored in this manner. We found that C-TIMP-4
and C-TIMP-2 bind to the MMP-2 hemopexin C domain as an
isolated domain or in the context of full-length pro-MMP-2.
However, C-TIMP-2 binds less avidly than full-length TIMP-2.
The lack of the N-terminal domain may partially destabilize
the C-TIMP domain, leading to a weaker interaction with the
MMP-2 hemopexin C domain. It is unlikely that myoglobin
sterically hinders this interaction, since C-TIMP-4 binds pro-
48704
Role of TIMP-4 and -2 C-terminal Tails in Binding pro-MMP-2
FIG. 9. Glutaraldehyde cross-linking analysis of pro-MMP-2 binding by the C-terminal tail deletion mutants of C-TIMP-4 and
C-TIMP-2. CT4⌬T and pro-MMP-2 (A) or CT2⌬T and pro-MMP-2 (B) were incubated for 1 h and cross-linked upon the addition of 0.5%
glutaraldehyde (GA). Complexes were analyzed by 10% polyacrylamide gel electrophoresis and silver staining. The position of molecular mass
markers (M) is indicated.
FIG. 10. Competition of C-TIMPs with TIMPs for binding to the
hemopexin C domain of MMP-2. The association rate constant (kon)
for the interaction of 380 pM TIMP-4 (A) or TIMP-2 (B) with MMP-2 (15
pM) was measured at 25 °C as described under “Experimental Procedures.” Increasing amounts of C-TIMP-4 (closed circles) or CT4⌬T (open
circles) (A), or C-TIMP-2 (closed circles) or CT2⌬T (open circles) (B) were
mixed with TIMP-4 or TIMP-2, respectively, prior to the addition to
MMP-2 and measurement of kon.
MMP-2 with only slightly less affinity than TIMP-4. Hence, the
present studies using TIMP C domain proteins biochemically
confirm that this domain is necessary and sufficient for the
noninhibitory TIMP-4 and TIMP-2 interaction with proMMP-2 on the hemopexin C domain.
TIMP binding to the hemopexin C domain of pro-MMP-2 is
distinct from the more dynamic interactions that occur during
binding and inhibition of active MMP-2. Following mapping of
the noninhibitory TIMP-2 docking site on the hemopexin C
domain of pro-MMP-2 by mutagenesis studies, it was evident
that topographically this site cannot be used for inhibitory
complex formation with active MMP-2 (40). This prompted
alternate explanations involving two structurally and functionally distinct binding sites for TIMP-2 on the hemopexin C
domain of pro- and active MMP-2, termed the docking and
stabilization sites, respectively (5, 26, 28, 40), to account for the
critical observations of Willenbrock et al. (38, 39) that the
C-terminal tail of TIMP-2 enhances inhibition of active MMP-2.
The TIMP stabilization site on the hemopexin C domain increases the affinity of TIMP binding to the active site cleft (24,
26, 34, 36, 52), forming stabilizing contacts with all TIMPs
when bound in the elongated wedge orientation (41). Homologous stabilization sites are predicted on all MMP hemopexin C
domains for all inhibitory TIMP interactions (5). Our present
kon data shows that C-TIMP-4 and C-TIMP-2 slowed the inhibition of MMP-2 by full-length TIMP-4 or TIMP-2, respectively.
Since the C-TIMP proteins are noninhibitory as expected, this
suggests that binding of the C-TIMP proteins to the MMP-2
hemopexin C domain prevents full-length TIMP-4 or TIMP-2
binding at the hemopexin C domain docking site. This initial
docking interaction has only been characterized for TIMP-2
inhibition of active MMP-2 where TIMP-2 was demonstrated to
utilize its anionic C-terminal tail (38). Competition of
C-TIMP-4 with TIMP-4 suggests that a C domain docking
interaction also initiates and enhances the rate of inhibition of
active MMP-2 by TIMP-4, although, unlike TIMP-2, this is not
driven by a C-terminal tail interaction and consequently occurs
at a slower rate.
Although a structure has not been reported for any fulllength TIMP in an inhibitory complex with a full-length active
MMP, modeling predictions suggest that the stabilization site
lies on the rim of the hemopexin C domain at the junction of
modules I and II that are proximal to the active site (5). Hence,
in the inhibitory interaction, the TIMP-2 C domain cannot bind
in the same orientation that occurs at the docking site mapped
by mutagenesis and crystallography to the rim of the hemopexin C domain on the opposite side of the molecule along
the edge of modules III and IV (40, 42). Therefore, how does
binding at the docking site enhance inhibition where contact
with the catalytic domain cannot occur? We have suggested
that, by initial tethering of TIMP-2 to the enzyme, the docking
interaction enhances the rate of inhibition by increasing the
probability of productive inhibitory complex formation that
subsequently occurs (5, 28, 40). Inhibition, therefore, requires
disengagement from the docking site and rebinding at the
stabilization site to adopt the inhibitory elongated wedge topology. The three-dimensional structure of the TIMP-2䡠proMMP-2 complex adds support to this proposal; the orientation
of TIMP-2 at the docking site precludes simultaneous interac-
Role of TIMP-4 and -2 C-terminal Tails in Binding pro-MMP-2
FIG. 11. Zymographic analysis of pro-MMP-2 enzyme capture
assay. TIMP-2, TIMP-4, C-TIMP mutant recombinant proteins, and
horse heart myoglobin (Mb) (0.5 ␮g) were coated onto a high proteinbinding 96-well plate and blocked with 1% BSA. A serial dilution of
pro-MMP-2, as indicated, was incubated with each coating protein for
2 h. The captured pro-MMP-2 was analyzed by gelatin zymography on
10% polyacrylamide gels. A, pro-MMP-2 captured by TIMP-2, TIMP-4,
and myoglobin controls. B, pro-MMP-2 captured by C-TIMPs. C, relative binding affinities of C-TIMP variant domains. For all mutant
proteins, the quantities of bound enzyme were quantified in the linear
response range of the assay (250 ng) by densitometric analysis and
expressed as band density normalized to the band density of MMP-2
captured by C-TIMP-2. For domains to which pro-MMP-2 could be
bound to saturation, apparent Kd values were determined from curve
fits.
tion with the catalytic domain (42). The residual association
rates of about 1.8 ⫻ 106 M⫺1 s⫺1 for TIMP-4 and 1 ⫻ 106 M⫺1 s⫺1
for TIMP-2 at higher concentrations of C-TIMPs in the competition experiments could reflect inhibition of MMP-2 by the
full-length TIMPs binding directly to the catalytic domain in
the elongated wedge topography. In other words, TIMP-4 and
TIMP-2 may bind the hemopexin C domain stabilization site
without prior binding to the docking site, the majority of which
would be blocked by the excess C-TIMP present, so inhibition
occurs at a slower rate resembling that for TIMP-1 (26). These
data imply that in the inhibition of active MMP-2 two molecules of TIMP-2 or TIMP-4 may be simultaneously bound: one
at the stabilization site and one at the docking site, as must
occur for inhibition of active MMP-2 in the ternary complex.
48705
Since TIMP-4 is homologous to TIMP-2 and can bind to
pro-MMP-2 at the docking site (26, 32), it was proposed that
TIMP-4 might also support or modulate the activation of
pro-MMP-2 (32). We found here that high concentrations of
C-TIMP-4 and C-TIMP-2 block the cellular activation of proMMP-2. We interpret this to occur by C-TIMP binding proMMP-2 in solution and thereby preventing the formation of
trimolecular activation complexes on the cell surface. Recombinant hemopexin C domain of MMP-2 also inhibits the activation of pro-MMP-2, but in this case by competing for the
available TIMP-2 (23, 27, 28). These results also demonstrate
in a cellular context that the C domain alone of TIMP-4 and
TIMP-2 can bind pro-MMP-2. The kinetic competition data
confirm that in binding pro-MMP-2 the C-TIMP proteins
block TIMP-2 binding to the docking site. Although we previously reported that TIMP-4 does not compete with TIMP-2
for binding of pro-MMP-2 (26), TIMP-4 was not used at the
high concentrations that could be used for C-TIMP-4. Since
TIMP-4 levels in vivo are unlikely to reach those of the
C-TIMP proteins effective here (ⱖ1 ␮M), this is unlikely to
represent a physiological mechanism whereby TIMP-4 can
modulate MMP-2 activation.
Our C-TIMP data are consistent with previous reports using
TIMP-2 having a modified N terminus, which eliminates the
ability of the amino group of Cys1 to coordinate with the MMP
active site Zn2⫹ ion. This resulted in a molecule unable to bind
MT1-MMP and therefore unable to form trimolecular complexes or facilitate pro-MMP-2 activation (53, 54). It has also
been reported that TIMPs bind to cell surface receptors in a
noninhibitory manner, potentially via their C domains, to modify cell behavior such as growth and differentiation (13, 20). It
cannot be discounted that the C-TIMP proteins may have triggered such a pathway, leading to modulation of cell behavior
and reduced MMP-2 activation. Indeed, in ongoing studies, the
recombinant proteins generated here will prove extremely valuable for exploring the growth factor effects of TIMP-4 and
TIMP-2 in the absence of any MMP-inhibitory activity.
TIMP-4 down-regulates MMP-2 activation by inhibiting
MT1-MMP in a manner that does not support binding and
trimolecular complex formation with pro-MMP-2 (26, 33). Since
TIMP-4 and TIMP-2 have a similar inhibition constant (Ki) and
kon for MT1-MMP (26), this indicates that the difference between TIMP-4 and TIMP-2 in effecting MMP-2 activation resides within the TIMPC domain. Due to the similar binding
properties found for C-TIMP-4 and CT4⌬T, the C-terminal tail
of C-TIMP-4 does not appear to be as important as the TIMP-2
tail in its interaction with pro-MMP-2. This was also evident,
since CT23 T4 exhibited as marked a reduction in binding to
pro-MMP-2 as CT2⌬T. Conversely, the importance of the Cterminal tail of C-TIMP-2 in strengthening binding to proMMP-2 was confirmed by the increase in pro-MMP-2 captured
by CT43 T2 and by the reduction in pro-MMP-2 binding by
CT2⌬T.
In deletion experiments, Willenbrock et al. first revealed the
importance of the TIMP-2 C-terminal tail in binding to the
docking site of the MMP-2 hemopexin C domain of the active
enzyme (38, 39), but the exact residues involved in the interaction had not been defined. Several residues in the unique
cationic clusters of the MMP-2 hemopexin C domain have been
shown to be involved in TIMP-2 binding (40). Because the net
charge of the TIMP-2 C-terminal tail is ⫺4, compared with ⫺1
for TIMP-4, we hypothesized that the stronger negative character of the TIMP-2 C-terminal tail plays an important role in
promoting and stabilizing the interaction between TIMP-2 and
pro-MMP-2, thereby allowing TIMP-2 but not TIMP-4 to form
a stable trimolecular complex with pro-MMP-2 and MT1-MMP.
48706
Role of TIMP-4 and -2 C-terminal Tails in Binding pro-MMP-2
We found that the removal of the acidic residues Glu192 and
Asp193 from C-TIMP-2 reduced binding by ⬃60%. Conversely,
the addition of these residues to the tail of C-TIMP-4 resulted
in an increase in the amount of pro-MMP-2 bound, reflecting
the reduced apparent Kd of CT4⫹q⫺, which was similar to the
increased affinity found for CT43 T2. Together, these data
show that the additional anionic character of the C-terminal
tail of TIMP-2 is responsible for its tighter binding to proMMP-2 compared with TIMP-4 and that this largely accounts
for the deficiency in the ability of TIMP-4 to participate in the
formation of a functional ternary complex.
TIMP-3, like TIMP-4, can bind pro-MMP-2, although it has
not been shown to participate in the MT1-MMP-mediated activation of pro-MMP-2 (34). Notably, the C-terminal tail of
TIMP-3 has a net charge of 0 with a positively charged lysine
residue at the position homologous to Lys187 found in the Cterminal tail of TIMP-4 but not in TIMP-2. This charge may
further weaken binding to the positively charged MMP-2 hemopexin C domain docking site. Although the C-terminal tail
appears to be important in the interaction with the MMP-2
hemopexin C domain docking site, our data show that the
presence of acidic residues at positions 192 and 193 in TIMP-2
specifically leads to greatly added stability. Nonetheless, other
elements of the TIMP C domain also contact the docking site as
evidenced by the ability of CT4⌬T and CT2⌬T to be crosslinked with pro-MMP-2 and to compete for TIMP-4 and TIMP-2
binding to active MMP-2. This is consistent with the inability of
TIMP-2 C-terminal tail peptide analogues to compete or prevent TIMP-2 binding (28) and with the structure of the TIMP2䡠pro-MMP-2 complex (42). In particular, a hydrophobic pocket
centered around Phe621 on the hemopexin C domain that accommodates Met149 in the C domain of TIMP-2 is believed to be
a major contributor to binding stability (42). In TIMP-4, the
residue at position 149 is a threonine and not a methionine as
found in TIMP-2. Hence, an aliphatic hydroxyl that reduces the
hydrophobic character of this part of the TIMP-4 contact face
should also contribute to the lower binding affinity of TIMP-4
compared with TIMP-2 (42). Hydrophobic interactions are also
proposed to be important in stabilizing the interaction of the
C-terminal tail of TIMP-2, where Phe188 closely packs with
hemopexin C domain residues Tyr552, Phe559, Phe573, Ala580,
and Trp581. Other hydrophobic interactions and hydrogen
bonds were also revealed in the structure of the large interface
of the TIMP-2䡠pro-MMP-2 complex (42).
Although the structure confirmed that several salt bridges
were formed between TIMP-2 and the hemopexin C domain
(42) that involved the cationic residues identified previously by
mutagenesis (42) as having important roles in this interaction
(Lys547, Lys550, Arg561, Lys566, Lys568, Lys610, Lys617) (40),
structural information alone does not reveal the relative contribution to the total binding energy of the individual interactions that make up the large ⬃250-nm2 contact surface. Moreover, Asp193 in the C-terminal tail of TIMP-2 was not resolved
in the structure, so its contribution to binding could not be
ascertained. Our present experimental data reveal that a potential salt bridge involving Asp193 together with the salt
bridge formed between Glu192 of TIMP-2 and Lys550/Lys566 of
MMP-2 (42) is critical for the binding stability of TIMP-2. In
view of the multiple and diverse forces that contribute to the
overall binding energy, it is not obvious which are the important amino acid residues and forces that contribute most to the
unique ability of TIMP-2 to form ternary complexes. Accordingly, it was somewhat surprising that two negatively charged
residues alone of the C-TIMP-2 tail had such a strong influence
on the stability of the TIMP-2 interaction with pro-MMP-2 and
could exert a similar effect on C-TIMP-4 following their introduction into its C tail.
Overall, the present work demonstrates a clear and important role for the unique negatively charged residues Glu192 and
Asp193 in the anionic tail of TIMP-2 in binding pro-MMP-2.
Physically, this tail binding site is quite distant and runs
perpendicular to the main C domain contact surface (42) and
thus appears to “bracket” TIMP-2 in place on the hemopexin C
domain, potentially explaining the profound effects on binding
stability that the replacement of Glu192/Asp193 shows. We suggest that the lack of these charges in the C-terminal tail of
TIMP-4 is largely responsible for its weaker interaction with
pro-MMP-2 and therefore its inability to participate in the
activation of pro-MMP-2; not only is there an energy cost from
the loss of salt bridge partners, but there is an additional
energy penalty imposed by the introduction of nonpacking side
chains at the end of the TIMP-4 tail. Identification of this site
also suggests a new target to develop antagonists of TIMP-2
binding (18) that may block MMP-2 activation in disease. The
role of TIMP-4 binding the hemopexin C domain of pro-MMP-2
remains unclear, but it may be to recruit this inhibitor to the
vicinity of proteolytic activity that is initiated following displacement by TIMP-2 upon MMP-2 activation.
Acknowledgment—We appreciate the advice of Dr. C. J. Loewen with
the velocity sedimentation studies.
REFERENCES
1. McCawley, L. J., and Matrisian, L. M. (2001) Curr. Opin. Cell Biol. 13,
534 –540
2. Sternlicht, M. D., and Werb, Z. (2001) Annu. Rev. Cell Dev. Biol. 17, 463–516
3. Egeblad, M., and Werb, Z. (2002) Nat. Rev. Cancer 2, 161–174
4. López-Otı́n, C., and Overall, C. M. (2002) Nat. Rev. Mol. Cell. Biol. 3, 509 –519
5. Overall, C. M. (2002) Mol. Biotechnol. 22, 51– 86
6. Nagase, H., and Woessner, J. F., Jr. (1999) J. Biol. Chem. 274, 21491–21494
7. Gearing, A. J., Beckett, P., Christodoulou, M., Churchill, M., Clements, J. M.,
Davidson, A. H., Drummond, A. H., Galloway, W. A., and Gilbert, R. (1994)
Nature 370, 555–557
8. Yu, Q., and Stamenkovic, I. (2000) Genes Dev. 14, 163–176
9. Schonbeck, U., Mach, F., and Libby, P. (1998) J. Immunol. 161, 3340 –3346
10. McQuibban, G. A., Gong, J. H., Tam, E. M., McCulloch, C. A., Clark-Lewis, I.,
and Overall, C. M. (2000) Science 289, 1202–1206
11. McQuibban, G. A., Gong, J. H., Wong, J. P., Wallace, J. L., Clark-Lewis, I., and
Overall, C. M. (2002) Blood 100, 1160 –1167
12. McQuibban, G. A., Butler, G. S., Gong, J. H., Bendall, L., Power, C., ClarkLewis, I., and Overall, C. M. (2001) J. Biol. Chem. 276, 43503– 43508
13. Hayakawa, T., Yamashita, K., Ohuchi, E., and Shinagawa, A. (1994) J. Cell
Sci. 107, 2373–2379
14. Brooks, P. C., Silletti, S., von Schalscha, T. L., Friedlander, M., and Cheresh,
D. A. (1998) Cell 92, 391– 400
15. Kato, T., Kure, T., Chang, J. H., Gabison, E. E., Itoh, T., Itohara, S., and Azar,
D. T. (2001) FEBS Lett. 508, 187–190
16. Stetler-Stevenson, W. G. (2001) Surg. Oncol. Clin. N. Am. 10, 383–392
17. Overall, C. M., Wrana, J. L., and Sodek, J. (1991) J. Biol. Chem. 266,
14064 –14071
18. Overall, C. M., and López-Otı́n, C. (2002) Nat. Rev. Cancer 2, 657– 672
19. Vincenti, M. P. (2001) Methods Mol. Biol. 151, 121–148
20. Stetler-Stevenson, W. G., and Yu, A. E. (2001) Semin. Cancer Biol. 11, 143–152
21. Murphy, G., Stanton, H., Cowell, S., Butler, G., Knauper, V., Atkinson, S., and
Gavrilovic, J. (1999) APMIS 107, 38 – 44
22. Sato, H., Takino, T., Okada, Y., Cao, J., Shinagawa, A., Yamamoto, E., and
Seiki, M. (1994) Nature 370, 61– 65
23. Strongin, A. Y., Collier, I., Bannikov, G., Marmer, B. L., Grant, G. A., and
Goldberg, G. I. (1995) J. Biol. Chem. 270, 5331–5338
24. Butler, G. S., Butler, M. J., Atkinson, S. J., Will, H., Tamura, T., van Westrum,
S. S., Crabbe, T., Clements, J., d’Ortho, M. P., and Murphy, G. (1998)
J. Biol. Chem. 273, 871– 880
25. Zhou, Z., Apte, S. S., Soininen, R., Cao, R., Baaklini, G. Y., Rauser, R. W.,
Wang, J., Cao, Y., and Tryggvason, K. (2000) Proc. Natl. Acad. Sci. U. S. A.
97, 4052– 4057
26. Bigg, H. F., Morrison, C. J., Butler, G. S., Bogoyevitch, M. A., Wang, Z.,
Soloway, P. D., and Overall, C. M. (2001) Cancer Res. 61, 3610 –3618
27. Morrison, C. J., Butler, G. S., Bigg, H. F., Roberts, C. R., Soloway, P. D., and
Overall, C. M. (2001) J. Biol. Chem. 276, 47402– 47410
28. Overall, C. M., Tam, E., McQuibban, G. A., Morrison, C., Wallon, U. M., Bigg,
H. F., King, A. E., and Roberts, C. R. (2000) J. Biol. Chem. 275,
39497–39506
29. Lehti, K., Valtanen, H., Wickstrom, S., Lohi, J., and Keski-Oja, J. (2000)
J. Biol. Chem. 275, 15006 –15013
30. Itoh, Y., Takamura, A., Ito, N., Maru, Y., Sato, H., Suenaga, N., Aoki, T., and
Seiki, M. (2001) EMBO J. 20, 4782– 4793
31. Tam, E. M., Wu, Y. I., Butler, G. S., Stack, M. S., and Overall, C. M. (2002)
J. Biol. Chem. 277, 39005–39014
Role of TIMP-4 and -2 C-terminal Tails in Binding pro-MMP-2
32. Bigg, H. F., Shi, Y. E., Liu, Y. E., Steffensen, B., and Overall, C. M. (1997)
J. Biol. Chem. 272, 15496 –15500
33. Hernandez-Barrantes, S., Shimura, Y., Soloway, P. D., Sang, Q. A., and Fridman, R. (2001) Biochem. Biophys. Res. Commun. 281, 126 –130
34. Butler, G. S., Apte, S. S., Willenbrock, F., and Murphy, G. (1999) J. Biol. Chem.
274, 10846 –10851
35. Howard, E. W., and Banda, M. J. (1991) J. Biol. Chem. 266, 17972–17977
36. Murphy, G., Willenbrock, F., Ward, R. V., Cockett, M. I., Eaton, D., and
Docherty, A. J. (1992) Biochem. J. 283, 637– 641
37. Butler, G. S., Hutton, M., Wattam, B. A., Williamson, R. A., Knauper, V.,
Willenbrock, F., and Murphy, G. (1999) J. Biol. Chem. 274, 20391–20396
38. Willenbrock, F., Crabbe, T., Slocombe, P. M., Sutton, C. W., Docherty, A. J.,
Cockett, M. I., O’Shea, M., Brocklehurst, K., Phillips, I. R., and Murphy, G.
(1993) Biochemistry 32, 4330 – 4337
39. Willenbrock, F., and Murphy, G. (1994) Am. J. Respir. Crit. Care Med. 150,
(Suppl.) 165–170
40. Overall, C. M., King, A. E., Sam, D. K., Ong, A. D., Lau, T. T., Wallon, U. M.,
DeClerck, Y. A., and Atherstone, J. (1999) J. Biol. Chem. 274, 4421– 4429
41. Fernandez-Catalan, C., Bode, W., Huber, R., Turk, D., Calvete, J. J., Lichte, A.,
Tschesche, H., and Maskos, K. (1998) EMBO J. 17, 5238 –5248
42. Morgunova, E., Tuuttila, A., Bergmann, U., and Tryggvason, K. (2002) Proc.
Natl. Acad. Sci. U. S. A. 99, 7414 –7419
48707
43. Crabbe, T., Zucker, S., Cockett, M. I., Willenbrock, F., Tickle, S., O’Connell,
J. P., Scothern, J. M., Murphy, G., and Docherty, A. J. (1994) Biochemistry
33, 6684 – 6690
44. Wallon, U. M., and Overall, C. M. (1997) J. Biol. Chem. 272, 7473–7481
45. Steffensen, B., Wallon, U. M., and Overall, C. M. (1995) J. Biol. Chem. 270,
11555–11566
46. Knight, C. G., Willenbrock, F., and Murphy, G. (1992) FEBS Lett. 296,
263–266
47. Maurus, R., Overall, C. M., Bogumil, R., Luo, Y., Mauk, A. G., Smith, M., and
Brayer, G. D. (1997) Biochim. Biophys. Acta 1341, 1–13
48. Kunkel, T. A., Roberts, J. D., and Zakour, R. A. (1987) Methods Enzymol. 154,
367–382
49. Greene, J., Wang, M., Liu, Y. E., Raymond, L. A., Rosen, C., and Shi, Y. E.
(1996) J. Biol. Chem. 271, 30375–30380
50. Overall, C. M., and Sodek, J. (1990) J. Biol. Chem. 265, 21141–21151
51. Loewen, C. J., and Molday, R. S. (2000) J. Biol. Chem. 275, 5370 –5378
52. Olson, M. W., Gervasi, D. C., Mobashery, S., and Fridman, R. (1997) J. Biol.
Chem. 272, 29975–29983
53. Higashi, S., and Miyazaki, K. (1999) J. Biol. Chem. 274, 10497–10504
54. Wingfield, P. T., Sax, J. K., Stahl, S. J., Kaufman, J., Palmer, I., Chung, V.,
Corcoran, M. L., Kleiner, D. E., and Stetler-Stevenson, W. G. (1999) J. Biol.
Chem. 274, 21362–21368