(Received for publication, September 10, 1996, and in revised form, December 20, 1996)
From the Respiratory and Critical Care Division, Departments of Medicine and Cell Biology and Physiology, Washington University School of Medicine at Barnes-Jewish Hospital, St. Louis, Missouri 63110 and the § Inflammatory Diseases Unit, Roche Bioscience, Palo Alto, California 94304-9819
Macrophage elastase (ME) was originally named
when metal-dependent elastolytic activity was detected in
conditioned media of murine macrophages. Subsequent cDNA cloning of
the mouse and human enzyme demonstrated that ME is a distinct member of
the matrix metalloproteinase family. To date, the catalytic parameters that describe the hydrolysis of elastin by ME have not been quantified and its activity against other matrix proteins have not been described. In this report, we have examined the action of purified recombinant human ME (rHME), produced in Escherichia coli, on elastin
and other extracellular matrix proteins. On a molar basis, rHME is approximately 30% as active as human leukocyte elastase in
solubilizing elastin. rHME also efficiently degrades
1-antitrypsin (
1-AT), the primary
physiological inhibitor of human leukocyte elastase. In addition, rHME
efficiently degrades fibronectin, laminin, entactin, type IV collagen,
chondroitan sulfate, and heparan sulfate. These results suggest that
HME may be required for macrophages to penetrate basement membranes and
remodel injured tissue during inflammation. Moreover, abnormal
expression of HME may contribute to destructive processes such as
pulmonary emphysema and vascular aneurysm formation. To further
understand the specificity of HME, the initial cleavage sites in
1-AT have been determined. In addition, the hydrolysis of a series of synthetic peptides with different P
1
residues has been determined. rHME can accept large and small amino
acids at the P
1 site, but has a preference for
leucine.
Macrophage elastase (MMP-12) shares many properties with other members of the matrix metalloproteinase (MMP)1 gene family, yet it is unique in several ways. Like other MMPs, metalloelastase requires zinc for catalytic activity, is inhibited by the tissue inhibitors of metalloproteinases (TIMPs), and has common structural domains with other MMPs (1, 2). Human macrophage elastase (HME) is most closely related to collagenase-1 (MMP-1) and stromelysin-1 (MMP-3), being 49% identical to each at the amino acid level (3). Moreover, the gene for macrophage elastase, composed of a common 10-exon, 9-intron structure, is on human chromosome 11q22.2/22.3 with at least six other MMPs (4). Despite these similarities, HME possesses certain distinct biological and biochemical properties. Expression appears to be largely restricted to tissue macrophages (4). Upon activation, it not only cleaves its 8-kDa N-terminal domain, but also has a unique propensity to autolytically release its 23-kDa C-terminal domain resulting in a mature active 22-kDa proteinase (3-5).
Macrophage elastase shares its elastolytic activity (6, 7) with only a few MMPs, including the gelatinases (MMP-2 and MMP-9) and, to a lesser extent, matrilysin (8, 9). However, despite this characteristic activity, the relative capacity of metalloelastase (human or mouse) to degrade elastin has never been quantified. In addition, the catalytic capacity of metalloelastase against other extracellular matrix components has never been described. We have recently generated mice that lack the capacity to produce macrophage elastase by gene-targeting. Macrophages from these mice not only lost 95% of their elastolytic capacity, but were also unable to penetrate a synthetic basement membrane (Matrigel) (10). The purpose of this study was to define the capacity of HME to degrade extracellular matrix components and characterize its substrate specificity. Knowledge of the catalytic properties of macrophage elastase will help define potential roles for this enzyme in biologic processes associated with macrophage activation and metalloelastase expression in vivo. Current evidence suggests that these processes may include pathologic conditions such as atherosclerosis, tumor invasion/angiogenesis, cerebrovascular disease, and pulmonary emphysema, which currently represent the four leading causes of death in the United States.
4-Aminophenylmercuric acetate and heparin-agarose
were obtained from Sigma. SP-Sepharose and Sephacryl S-200 were
obtained from Pharmacia (Uppsala, Sweden). All other chemicals were
reagent grade. Bovine ligament elastin and HLE were obtained from
Elastin Products (Owensville, MO). Human 1-AT was
obtained from Athens Research Products (Athens, GA). Fibronectin,
laminin, type IV collagen, chondroitan sulfate, and heparan sulfate
were obtained from Collaborative Research Products (Bedford, MA).
Entactin was a generous gift of Dr. Robert Senior and 92-kDa gelatinase
and interstitial collagenase were kindly provided by Dr. Howard Welgus, both of the Washington University School of Medicine, St. Louis, MO.
Full-length HME cDNA was ligated as an
NdeI/BamHI cassette into the pET 5b vector which
permitted translation in the proper reading frame beginning with the
HME initiation methionine. pET 5b alone (control plasmid) and pET
5b/HME were transformed into E. coli strain BL21(DE3)
(Novagen Inc., Madison, WI). Single colonies of E. coli
(/+ rHME) from Lennox Broth (LB)/agar plates with 20 µg/ml
ampicillin were grown to log phase in 1 liter of LB media (with
ampicillin) in a shaking incubator at 37 °C. To induce T7 RNA
polymerase and drive high level expression of rHME,
isopropyl-
-D-thiogalactopyranoside was added to a final
concentration of 0.4 mM. Cells were maintained in culture
for an additional 4 h. Cell pellets were resuspended in 10 ml of
50 mM Tris, pH 8.0, with 10 mM
CaCl2 and 150 mM NaCl and then lysed by
sonication. After centrifugation, the rHME was localized in the pellet.
It was solubilized with 40 ml of 8 M urea containing 50 mM Tris, pH 8.0, with 10 mM CaCl2
and 30 mM NaCl continuously stirred for 2 h at
4 °C. Following centrifugation, the supernatant containing soluble
rHME was collected.
The extracts from several 1-liter preparations were dialyzed against 3 M urea in the Tris-containing solution and the lysates were applied sequentially to SP-Sepharose ion exchange chromatography, high resolution Sephacryl S-200 column gel filtration (column 2.5 × 90 cm, bed volume 400 ml, flow rate 30 ml/h), and heparin-agarose affinity/ion exchange chromatography. Following these purification procedures, polyacrylamide gel electrophoresis (SDS-PAGE) and both Coomassie and silver staining demonstrated rHME without any visible contaminating proteins. The identity of this protein as HME was confirmed by Western blotting. N-terminal amino acid sequence analysis of the final product was determined by Edman degradation.
The bicinchoninic acid protein assay (Pierce Chemical Co.) and a TIMP inhibition assay were used to determine the concentration of rHME. The latter method involved preincubation of known concentrations of TIMP with fixed amounts of rHME followed by addition of Ac-Pro-Leu-Gly-S-Leu-Gly-OEt. Hydrolysis of this thiopeptolide substrate (Bachem Bioscience, King of Prussia, PA) by metalloproteinases was determined as described previously (11). The concentration of rHME was further confirmed by pulse liquid sequencing performed on an Applied Biosystems model 473A sequencer equipped with the model 610A analysis software. The initial yield of the protein was extrapolated from the repetitive yield calculations for the first 10 cycles of sequencing. The average repetitive yield during the run was 96.4%. Comparison of the TIMP inhibition assay and pulse liquid sequencing results demonstrated that greater than 90% of the purified rHME was catalytically active. Control plasmids subjected to the same purification scheme were catalytically inactive against all substrates tested.
The catalytic domain of matrilysin was expressed and purified to homogeneity in bacteria using the same techniques described for metalloelastase. Matrilysin expressed in bacteria had equal catalytic activity to matrilysin expressed in eukaryotic cells (12).
Degradation of Basement Membrane Components by MMPsTo qualitatively compare the degradative capacity of rHME to either matrilysin or 92-kDa gelatinase, the basement membrane components fibronectin, laminin, entactin, chondroitan sulfate, heparan sulfate, and type IV collagen were incubated with each enzyme and cleavage products resolved with polyacrylamide gel electrophoresis. Specifically, fibronectin and laminin (5 µg) were each incubated at 37 °C for 18 h with rHME or matrilysin in a final reaction mixture volume of 30 µl. Entactin was incubated under similar conditions for 2 h. Native and pepsinized type IV collagen (20 µg) were incubated with rHME or 92-kDa gelatinase at 25 °C for 18 h. Two different concentrations (46 and 460 nM for type IV collagen; 23 and 230 nM for all other substrates) of each enzyme were utilized. Chondroitan sulfate and heparan sulfate (10 µg) were incubated with rHME or matrilysin at 37 °C for 18 h and two different concentrations (0.23 and 2.3 µM) of each enzyme were utilized. The reaction mixtures were stopped with SDS sample buffer containing dithiothreitol, boiled, and then subjected to polyacrylamide gel electrophoresis. 10% chondroitan sulfate and type IV collagen, 8% fibronectin and entactin, and 6% laminin slab gels were stained with 1% Coomassie Brilliant Blue.
To further estimate the concentration of rHME or matrilysin required to produce 50% substrate cleavage, varying concentrations of both metalloproteinases were incubated with substrates as otherwise described above. For entactin, duration of incubation was reduced to 15 min and reaction temperature was reduced to 25 °C. Samples were resolved by SDS-polyacrylamide gel electrophoresis and stained with 1% Coomassie Brilliant Blue. Subsequently, bands representing substrate alone as well as residual non-cleaved substrate were scanned with an Apple Color One ScannerTM to permit quantitative comparison. Degradation was interpolated linearly to estimate the concentration of metalloproteinase required to produce 50% substrate cleavage.
Assessment of Elastolytic ActivityElastolytic activity was quantified by measuring solubilization of insoluble 3H-elastin. Bovine ligament elastin (Elastin Products, Owensville, MO) was radiolabeled with [3H]sodium borohydride (DuPont NEN), as described previously (5, 9). Elastin degradation was quantified by measuring solubilization of insoluble 3H-elastin at 37 °C and pH 7.5. Data are expressed as micrograms of elastin degraded from at least three separate experiments each performed in duplicate. These calculations are based on measurements of 3H-elastin counts/min, corrected for buffer blanks. The radiolabeled elastin used for these studies had a specific activity of 1900 cpm/µg.
Degradation of1-AT (10 µg) was incubated at 37 °C for
18 h with rHME or interstitial collagenase in a final reaction
mixture volume of 60 µl with enzyme concentrations of 10 and 100 nM. The reactions were quenched by addition of SDS sample
buffer containing dithiothreitol and the mixtures were boiled and
subjected to polyacrylamide gel electrophoresis. 8% Slab gels were
stained with 1% Coomassie Brilliant Blue. To quantify the degradation
of
1-AT, gels were scanned using a Gilford
spectrophotometer set at 560 nm equipped with a linear gel scanning
device (12).
The effect of the cleavage of 1-AT, by rHME, on its
ability to inhibit the elastase activity of HLE was examined.
Specifically, 3H-labeled insoluble elastin was incubated
for 24 h at 37 °C with 0.4 µg of HLE alone, 0.4 µg of HLE
preincubated with molar excess (40 µg) of intact
1-AT
or
1-AT previously incubated with rHME for 18 h at
37 °C.
Amino acid sequence analysis was performed on
1-AT degraded by purified rHME. 5 µg of
1-AT was incubated with 44 ng of rHME for 10 min at
37 °C and subsequently resolved by 12% SDS-polyacrylamide gel
electrophoresis. To resolve lower molecular weight degradation products, incubation time was extended to 30 min and a 4-20% gradient gel (Bio-Rad) was utilized. Proteins were transferred to a Problott membrane (Applied Biosystems) and visualized with 0.1% Coomassie Blue,
excised, and sequenced by Edman degradation using an Applied Biosystems
473 sequenator.
All peptides were synthesized as described in Ref. 13. The
initial rates of hydrolysis (vi) of the peptides
were measured fluorometrically at 37 °C using excitation at 278 nm and emission at 358 nm (14) in a Perkin-Elmer LS-50B Luminescence Spectrometer. The assays were conducted in 50 mM Tris-Cl,
pH 7.5, 200 mM NaCl, 10 mM CaCl2,
2.5% Me2SO, and 0.005% Brij-35. Measurements of
vi were made at 50, 100, 150, 200, and 250 µM substrate. The data were then fit to the equation:
vi = Vmax[So]/(Km + [So]). For some substrates Km values could not be obtained due to limited substrate solubility. In these cases,
only kcat/Km was determined
by measuring vi at % Km. Verification of the proper cleavage products was
made by reversed-phase high performance liquid chromatography.
rHME was
expressed in Escherichia coli and, after cell lysis,
solubilized in urea, dialyzed, and sequentially subjected to SP-Sepharose ion exchange chromatography, Sephacryl S-200 column gel
filtration, and heparin-agarose affinity/ion exchange chromatography (Fig. 1). After the first chromatographic step, all of
the rHME migrated with the apparent molecular mass of 22 kDa
corresponding to the mature processed form. At each purification step,
equal amounts of protein were incubated with insoluble
3H-elastin. The specific activity increased greater than
250-fold after the final purification step (Table I).
N-terminal sequence analysis was performed on the final material to
confirm its purity and identify the N terminus of the active form of
recombinant HME. A single Phe-Arg-Glu sequence was identified that
corresponds to cleavage at the His99-Phe100
bond, just C-terminal to the conserved cysteine switch motif. It should
be noted, however, that the N terminus of mature native HME has not
been identified to date. Finally, this expression and purification
procedure gave ~500 µg of purified rHME from 1 liter of E. coli.
|
The
basement membrane proteins fibronectin, entactin, laminin, chondroitan
sulfate, and heparan sulfate were incubated with rHME or matrilysin at
37 °C as described under "Materials and Methods." Native and
pepsinized type IV collagen was also incubated with rHME or 92-kDa
gelatinase under similar conditions at 25 °C. A lower incubation
temperature was selected to avoid conditions that would favor
denaturation of the collagen molecule in solution. Cleaved products
were resolved by SDS-polyacrylamide gel electrophoresis and stained
with 1% Coomassie Brilliant Blue (Fig. 2). The amounts of rHME or matrilysin required to produce 50% cleavage of fibronectin, entactin, laminin, and chondroitan sulfate are summarized in Table II. rHME and matrilysin both effectively cleave
entactin. rHME, however, is more potent than matrilysin in cleaving
fibronectin, laminin, and chondroitan sulfate. rHME and matrilysin also
efficiently degrade the proteoglycan heparan sulfate (data not shown).
Finally, rHME is comparable to 92-kDa gelatinase in its ability to
degrade pepsinized type IV collagen. Neither enzyme, however,
effectively cleaved native type IV collagen (data not shown).
|
The elastolytic activity of rHME
was measured by quantifying solubilization of 3H-elastin as
measured by the release of 3H into the supernatant.
Activity is expressed as micrograms of elastin degraded. Because
elastin is an insoluble substrate, classic Michaelis-Menten kinetics
could not be applied in comparing rHME and HLE. Under conditions of
substrate excess with small amounts of rHME or HLE, the amount of
elastin degraded was linearly related to enzyme concentration (Fig.
3). The specific activity of HME as calculated from the
first three data points is 33 µg of elastin degraded/mg of
enzyme/min. The specific activity of rHME against elastin was one-third
that of HLE (Fig. 4). Incubation of rHME with insoluble
elastin was extended for up to 96 h and catalysis continued to
proceed in a linear manner (data not shown), demonstrating that the 22 kDa species is stable under the conditions tested.
Cleavage of
Human
1-AT was incubated with rHME or interstitial collagenase
at concentrations of 10 and 100 nM for 18 h at
37 °C. The reaction products were resolved by SDS-polyacrylamide gel
electrophoresis. As shown in Fig. 5, rHME completely
converted 54-kDa
1-AT into a stable 50-kDa species at
both concentrations used, whereas interstitial collagenase demonstrated
incomplete conversion even at 100 nM. Incubation of 10 µg
of
1-AT with varying concentrations of rHME at 37 °C
for 30 min yielded 50% degradation of substrate at a concentration of
2 nM. Under similar conditions, interstitial collagenase
and matrilysin yielded 50% degradation of substrate at concentrations
of 75 and 47 nM, respectively. Preincubation of
1-AT with rHME abolished its ability to inhibit the
HLE-mediated degradation of elastin (data not shown). Thus, the stable
50-kDa form of
1-AT is functionally inactive.
N-terminal Amino Acid Sequence Analysis of
Previous studies of the hydrolysis of
1-AT by MMPs have shown cleavage sites within the
reactive loop (12, 15, 16). To determine initial HME cleavage sites in
1-AT, the reaction was carried out for 10 min at
37 °C and the products analyzed by SDS-PAGE (Fig. 6).
Cleavage products with molecular masses of 50, 29, 25, and 4 kDa were
identified, corresponding to two scissions of intact
1-AT to yield 50 and 4 kDa or 29 and 25 kDa products,
respectively. N-terminal amino acid sequence analysis identified the
two major cleavage sites: Phe352-Leu353,
generating the 50- and 4-kDa degradation products and
Glu199-Val200, generating the 29- and 25-kDa
fragments. As noted above, however, upon prolonged incubation the only
stable degradation product was the 50-kDa species, corresponding to
cleavage at Phe352-Leu353 within the reactive
loop.
Peptide Substrate Specificity of Cleavage by HME
The most
important subsite in determining MMP substrate specificity is
P1. Thus, the relative rates of hydrolysis of nine octapeptides differing only in the residues in subsite P
1
have been quantified for HME (Table III). The preference
follows the order: Leu
Ala > Lys > Phe > Tyr > Trp > Arg > Ser > Glu. In general, HME
tolerates a variety of large and small residues at the P
1
position. This is consistent with the expectation that HME should have
a deep S
1 pocket based on its predicted amino acid
sequence (see "Discussion").
|
Elastin is a highly cross-linked and hydrophobic insoluble extracellular matrix protein that imparts elastic recoil to a variety of tissues including musculoskeletal ligaments, arterial vessels, and lung parenchyma (18). These properties contribute to its extreme stability and resistance to proteolysis by all but a limited number of proteinases. When compared in parallel assays, rHME has approximately one-third the elastolytic capacity of HLE. A similar difference in elastolytic capability between recombinant 92-kDa gelatinase and HLE has been observed (9). It is probable, however, that under most circumstances HLE remains within the neutrophil, degrading internalized foreign material, and that only inadvertent release of HLE from the cell, during "frustrated phagocytosis" or death of the short-lived neutrophil, may cause tissue destruction. On the other hand, in response to mediators of inflammation, MMPs, including HME, are characteristically secreted into the extracellular space where they modulate matrix remodeling. With sustained accumulation of macrophages and neutrophils, e.g. in chronic inflammation of the lung induced by cigarette smoking, HME and HLE expression may be poorly regulated with consequent elastin destruction, ultimately producing the distinctive changes of pulmonary emphysema.
Several circulating and locally secreted inhibitors of HLE have been
identified. The primary physiological inhibitor is 1-AT which forms a stable complex with the enzyme (19). A variety of MMPs
are capable of degrading
1-AT (12, 15, 16) and abolishing its ability to inhibit HLE. MMPs, therefore, may
significantly modulate the physiological role of
1-AT.
We now show that HME, in particular, is an order of magnitude more
active than matrilysin, previously shown to be the most potent MMP
capable of degrading
1-AT (12). In addition, two initial
cleavages were identified, rather than the single degradation product
reported for other MMPs. The first is a cleavage at the
Glu199-Val200 bond. Stromelysin is the only MMP
which makes Glu-Val cleavages (2). The second site is within the active
site loop at the Phe352-Leu353 bond. This is
the same site at which interstitial collagenase and the 92-kDa
gelatinase cleave
1-AT (20, 21). Surprisingly, however,
MME, the murine orthologue to HME, cleaves at the
Pro357-Met358 bond instead. Of note, when
1-AT is inactivated by MME, a potent chemotactic factor
for neutrophils is generated (15). It is therefore possible that, in
addition to HME's inherent elastolytic capacity, elastin degradation
may be further augmented by inactivating the primary inhibitor of HLE
and by attracting additional neutrophils to sites of inflammation. HLE,
conversely, can cleave and inactivate TIMP (19), releasing HME from
inhibition.
HME is able to mediate degradation of several extracellular matrix components. HME was unable to cleave interstitial collagens and was minimally active against denatured collagen or gelatin (data not shown). It is, however, at least as active as any other MMP against the matrix proteins that comprise basement membranes and readily degrades fibronectin, laminin, entactin, chondroitan sulfate, and heparan sulfate. HME also degrades pepsinized type IV collagen as efficiently as the 92-kDa gelatinase (17). These cleavages are likely limited to the non-helical collagen domains. It is interesting to note that type IV collagen is relatively resistant to all inflammatory cell proteinases when compared with other basement membrane proteins. An attractive hypothesis would suggest that inflammatory cells release proteinases which dissolve the more susceptible basement membrane substrates leaving the type IV collagen backbone intact. Cells may then migrate through the disrupted basement membrane while the structural architecture is maintained by type IV collagen which would then provide a lattice for renewed biosynthesis of matrix components.
To further probe the substrate specificity of HME, its action on a
series of synthetic peptides with different P1 residues has been examined. These peptides were chosen based on knowledge gained
from structure-function studies performed on other MMPs. The structures
for the catalytic domains of several members of the MMP family have
recently become available including human fibroblast collagenase
(MMP-1) (22-24), human stromelysin-1 (MMP-3) (25, 26), human
matrilysin (MMP-7) (27), and human neutrophil collagenase (MMP-8)
(28-30). These structures revealed that the S
1 subsite is
the most well defined pocket in these MMPs and consists of a
hydrophobic pocket which varies greatly in its depth. Mutational
analyses of the S
1 pocket (29) revealed that residue 214 (numbering according to Browner (27)) which lies at the bottom of the
S
1 pocket is critical in determining its shape. For MMP-1
and MMP-7, which have an Arg214 and a Tyr214,
respectively, these residues point into the S
1 pocket,
thus forming a shallow pocket. MMP-3 and MMP-8 each have a
Leu214 residue which points away from the pocket, thus
allowing a deep S
1 pocket to exist. These S
1
pocket "types" have been corroborated by substrate specificity
studies using libraries of synthetic peptides (30-32). MMP-1 and
MMP-7, which have shallow S
1 pockets, prefer small
hydrophobic amino acids at the P
1 position. In contrast, MMP-3 and MMP-8, which have deep S
1 pockets, can
accommodate large and small P
1 amino acids with similar
efficiency.
Although the structure for macrophage elastase has not yet been
determined, the prediction based on the Leu214 which this
enzyme possesses suggests that there should be a deep S1
pocket which can accept large and small P
1 substrates. The data in Table III support this hypothesis. HME is able to tolerate large hydrophobic P
1 residues, such as Trp and Phe, much
better than MMP-1, which contains a shallow S
1 pocket.
Overall, HME had a preference for Leu at the P
1 position
and, in fact, rHME cleaved the reactive loop of
1-AT at
Phe352-Leu353. Of interest, HME is the only MMP
known that can accommodate Arg at P
1. This finding could
help in the design and synthesis of a selective HME inhibitor. This may
be useful in conditions with aberrant inflammatory responses which may
lead to macrophage-mediated tissue destruction.
We thank Thomas J. Broekelmann and Robert P. Mecham for invaluable assistance regarding N-terminal amino acid sequencing, and Anders Persson, Robert M. Senior, and Howard G. Welgus for helpful advice.