- Email: [email protected]
Purification and properties of rat uterine procollagenase
ARCHIVES OF BIOCHEMISTRY AND BIOPHYSICS Vol. 225, No. 1, August, pp. 285-295, 1983
Purification WILLIAM Divisitm
and Properties
of Rat Uterine Procollagenase’
T. ROSWIT, J. HALME,
AND
JOHN J. JEFFREY’
of Dermatobgy, Department of Medicine, and the Department of Biological Chemistry, Washington University School of Medicine, St. Louis, Missouri 63110 Received February
28, 1983, and in revised form March 31, 1983
A procollagenase from monolayer cultures of postpartum rat uterine cells has been purified. The crucial step in the purification is the binding of the procollagenase from crude, fetal bovine serum-containing culture medium to heparin-Sepharose, followed by elution with extremely low concentrations (5-10 IIM) of dextran sulfate. Resultant eluates contain 8-10s procollagenase. Purification is completed by ion-exchange chromatography on DEAE-Sepharose, gel filtration on AcA-44, and chromatography on blue-Sepharose. Rat uterine procollagenase appears as a protein doublet of M, - 58,000, as indicated by two polyacrylamide gel electrophoresis systems, by AcA-44 chromatography, and by equilibrium sedimentation ultracentrifugal analysis. The proenzyme forms are converted by trypsin to an active enzyme doublet of M, - 48,000. Small amounts of active enzyme, which are often generated during the purification, are electrophoretically indistinguishable from trypsin-activated collagenase. Active collagenase can be separated from the zymogen forms by DEAE-Sepharose chromatography. The two forms of the proenzyme doublet can be partially separated by gel filtration on AcA-44 and preliminary analysis indicates each has equal collagenolytic activity. The amino acid analysis of rat uterine collagenase reveals it to be markedly different from two other vertebrate collagenases whose composition is known. The uterine proenzyme is unusually rich in glycine and in the hydroxy amino acids and is considerably more acidic than the human skin fibroblast collagenase, consistent with the different ionexchange behavior of the two molecules. The specific activity of rat uterine collagenase at 37°C is approximately 3000 pg collagen/min/mg, using native reconstituted guinea pig skin type I collagen fibrils as substrate. The enzyme cleaves denatured collagen, but fails to attack a variety of noncollagen proteins.
Carefully controlled collagen degradation is a crucially important process which accompanies the postpartum involution of the mammalian uterus (1, 2). During this involutionary period, the collagen accumulated during pregnancy is removed in a very short time (2-4). Numerous studies have emphasized the importance of the enzyme collagenase as the initiator of the process of collagenolysis in a wide variety of tissues (5), including the uterus (6). In 1 Supported by USPHS Grant HD 05291, AM 12129, and AM 16184. ’ To whom correspondence should be addressed. 285
this latter tissue, collagenase activity appears shortly after delivery, is maintained at high levels throughout uterine remodeling, and then declines to low or undetectable levels when resorption is complete (7). A collagenase produced by the rat uterus has been isolated and partially purified from the medium of explant (6) and cell (8) cultures derived from this tissue. Studies on the hormonal regulation of this enzyme have suggested an important role for progesterone in preventing the appearance of active collagenase in the rat uterus (g-11), raising the possibility that a major physiological function of the re0003-9861/%3 $3.00 Copyright 0 1983by AcademicPress,Inc. All rights of reproductionin any form reserved.
286
ROSWIT,
HALME,
productive hormone is to regulate the rate and the geometry of collagen degradation in the uterus. Studies on the properties of collagenolytic enzymes have, in general, been hampered by the availability of only small quantities of protein, often in impure form. A number of these enzymes have been purified in amounts usually too small to allow extensive characterization (5, 12-14). The collagenase of tadpole skin, however, has been purified and studied with respect to amino acid composition (14) and certain kinetic properties (15,16). One collagenase, that from cultured human skin fibroblasts, has been purified as a zymogen in continuously available chemically significant quantities (17-18). The availability of milligram amounts of this proenzyme has permitted studies on substrate binding (19), fundamental kinetic parameters (20), collagen type specificity (20-22), and zymogen activation (23). Thus, to more precisely define not only the nature of the chemistry of the collagenolytic process itself in the uterus, but also the mechanisms of the physiologic regulation of the process, we have employed mass cultures of rat uterine cells, described previously, as a source of crude uterine collagenase. The present communication details the complete purification of rat uterine procollagenase from the medium of these cells. METHODS
AND
MATERIALS
CeU culture wudmok Primary cultures of postpartum rat uterine cells were grown as previously described (8), either in plastic trays (600 cm’, Nunc, Inc.) or plastic roller bottles (850 cmz, Falcon) in Dulbecco’s modified Eagle’s medium, high glucose, containing glutamine (KC Biologicals), 30 mM Hepes3 (pH 7.6), 200 units/ml penicillin, 200 pi/ml streptomycin, and 10% heat-inactivated fetal bovine serum (KC Biologicals). Harvested medium was brought to 50 mM Tris, 10 mM CaC1, and stored at -20°C until used. In a typical purification, 40-80 liters of harvested medium was employed as starting material. Assay procedures. Collagenase activity was mea-
’ Abbreviations used: Hepes, 4-(2-hydroxyethyl)-lpiperazineethanesulfonic acid; SDS, sodium dodecyl sulfate; CM, 0-(carboxymethyl).
AND
JEFFREY
sured at 3’7°C using native [‘4Clglycine-labeled reconstituted guinea pig skin collagen (24) as a substrate. Typically, 2550 pl of a 0.4% solution of collagen (sp act approx 50,000 cpm/mg) was allowed to aggregate at 3’7°C for at least 18 h before use in the assay. After appropriate incubation with enzyme solutions, usually in a volume of 100 ~1, the assays were terminated by the addition of 200 ~1 of 50 mM Tris, 10 mM CaCIz, pH 7.6, followed by centrifugation at 12,OOOgfor 15 min to sediment unreacted substrate. The entire supernatant solution was then removed and its radioactivity determined by liquid scintillation spectrometry. Protein concentration was determined by the spectrophotometric method of Groves et al (25) using bovine serum albumin as a standard. Chrmmdographic procedures. All chromatography was performed at 4°C. Flow rates were maintained with a Gilson peristaltic pump. Absorbance of eluates was monitored using a Gibson double-beam spectrophotometer. Heparin-Sepharose chromatography. Heparin-Sepharose (Pharmacia) was packed in a 5 X lo-cm column and equilibrated with 50 mM Tris, 10 mM CaClz, 200 mM NaCl, pH 7.5. Crude culture medium (lo-20 liters) was clarified by passage through a 3-pm filter (Gelman) and solid sodium chloride was added until the conductivity was equal to that of the column buffer. The medium was then pumped through the column at 550 ml/h. The column was washed with 50 mM Tris-HCl, 10 mM CaClz and then eluted with a linear gradient of dextran sulfate of 0 to 50 pg/ml in a total volume of 2 liters at a flow rate of 350 ml/ h. Fractions were assayed for collagenase as described above, active fractions were pooled, and a sufficient volume of a 30% solution of the nonionic detergent Brij was added to a final concentration of 0.05%. DEAE-Sephurose chromatography. The pooled, collagenase-containing fractions eluted from heparinSepharose were applied to a 2.6 X g-cm column of DEAE-Sepharose equilibrated with 50 mM Tris-HCl, 10 mM CaClz, 0.05% Brij, pH 7.5. The sample was applied at 175 ml/h, and the column was washed with the same buffer. Enzyme elution was accomplished with a 700-ml gradient of 0 to 0.3 M NaCl in column buffer; the flow rate was 175 ml/h. Collagenase assays were performed as described, enzyme-containing fractions were pooled, and the NaCl concentration of the pool was lowered by a brief (2-h) dialysis vs 50 mM Tris, 10 mM CaClz, 0.05% Brij. Gel jltratiox The pooled, dialyzed DEAE eluate was applied to a 1.6 X 1.5-cm column of heparinSepharose equilibrated in the dialysis buffer. The enzyme protein was eluted directly onto a 1.6 X 200-cm column of Ultrogel AcA-44 (LKB) with 50 mM Tris, 10 mM CaCI,, 0.05% Brij, containing 0.8 M NaCl. The AcA-44 column was equilibrated with 50 mM Tris, 10 mM CaClz, 0.3 M NaCl, 0.05% Brij, and the column
PURIFICATION
OF RAT
UTERINE
was run at a flow rate of 15 ml/h. Collagenase was assayed as described, and enzyme-containing fractions were pooled and dialyzed for 4 h vs 50 mM Tris, 10 mM CaClr, 0.05% Brij. Blue-Sepharose chromatography. The pooled collagenase-containing fractions from AcA-44 were applied to a 1.6 x 3-cm column of blue-Sepharose (Pierce) equilibrated with 50 mM Tris, 10 mhf CaCl,, 0.05% Brij at a flow rate of 50 ml/h. The column was eluted with a 170-ml linear gradient of O-O.9 M NaCl in the same buffer. Collagenase was assayed in the column fractions and active fractions were pooled and dialyzed vs 50 mM Tris, 10 mM CaCl,, 0.05% Brij. Polyacrylamidegel electrophwesis. As needed, samples from the above column chromatographic procedures were prepared for electrophoresis by dialysis against 10,000 vol of 0.1 N HAc followed by lyophylization. The dried samples were dissolved directly in the appropriate sample buffer for the electrophoretic method selected. Protein standards routinely included on all gels were obtained from Pharmacia and included phosphorylase b (94,000), serum albumin (67,000), ovalbumin (43,000), carbonic anhydrase (30,000). soybean trypsin inhibitor (20,100), and a-lactalbumin (14,000). Two electrophoretic systems were employed: the discontinuous gel system of King and Laemmli (26) and the continuous gel system of Fairbanks et a2. (27). Following electrophoresis the protein bands were stained with 1% Coomassie brilliant blue. UltracentrZfugal analysis. Sedimentation equilibrium analysis was performed in a Spinco Model E analytical ultracentrifuge, equipped with an ultraviolet scanning system at 280 mm, at 30,000 rpm and 4°C by the method of Yphantis (28). Protein samples were dissolved in 50 mM Tris, 10 mM CaCl,, 0.05% Brij. After equilibrium was attained, the data were plotted as the log of ODza, vs the square of the distance (x2) from the center of the rotor. Amino acid analysis. Amino acid analyses were performed by the Washington University Protein Sequencing Facility, using a Waters high-performance liquid chromatography amino acid analyzer equipped with a fluorescence detector.
RESULTS
The binding of rat uterine collagenase from fetal bovine serum-containing culture medium to heparin-Sepharose and the elution of the enzyme from this matrix with dextran sulfate is illustrated in Fig. 1. The vast bulk of the protein of the medium fails to bind to this matrix. In addition, the collagenase is eluted with dextran sulfate (M, - 500,000) at a concentration of approximately 10 nM (5 pg/ml),
PROCOLLAGENASE
287
resulting in a typical purification of some 7000-fold (Table I). Generally this single purification step yields solutions in which collagenase constitutes 8-10% of the total protein. Beyond this point in the purification of the enzyme, 0.05% Brij was routinely included in all buffers as first suggested by Cawston et al. (13) for porcine synovial collagenase in order to minimize loss of activity, particularly severe when solutions of collagenase were frozen and thawed. The eluates from heparin-Sepharose are subjected to chromatography on DEAESepharose (Fig. 2), and the collagenase is eluted with a gradient of sodium chloride. This step both removes the dextran sulfate from the solution and separates the enzyme from the bulk of the major contaminant at this point in the purification, bovine serum albumin. The resultant collagenase-containing fractions are concentrated by binding to a small (1.6 X 1.5-cm) column of heparinSepharose and eluted with 0.8 M NaCI directly onto a 1.6 X 200-cm column of AcA44. The enzyme emerges from the column as a nearly symmetrical peak with an apparent M, of 57,000 (Fig. 3). At this stage of the purification, the collagenase exists predominantly in a trypsin-activatable form. A small percentage, however, which appears to vary with batches of starting material, is present as active enzyme. Final purification of rat uterine collagenase is achieved by absorption of the enzyme to Cibachron blue-Sepharose followed by elution with a sodium chloride gradient (Fig. 4). The collagenase is retained on the column until the salt concentration is quite high (0.4-0.5 M). The purification of the rat uterine cell collagenase is detailed in Table I. The overall fold-purification from crude serum-containing medium is approximately 75,000, with an overall yield of -30%. Some 25-50 pg of pure collagenase is obtained per liter of culture medium. Final specific activity of the uterine enzyme is approximately 3000 mg collagen/min/mg protein at 37°C (Table I), with native, reconstituted guinea pig skin collagen fibrils as substrate.
288
ROSWIT,
HALME,
FRACTION
AND
JEFFREY
NUMBER
FIG. 1. Heparin-Sepharose chromatography. Twelve liters of culture medium was prepared as under Materials and Methods of heparin-Sepharose at a flow rate of 550 ml/h. Elution was gradient of dextran sulfate, O-50 pg/ml, in 50 mM Tris-HCl, pH fractions were collected. Fractions 44-68 were pooled, brought
Electrophoresis of the uterine collagenase in the SDS-containing polyacrylamide gel electrophoresis system of King and Laemmli (26) reveals a proenzyme protein doublet with apparent molecular weights of approx 64,000 and 60,000. These forms are shown in the composite polyacrylamide gel in Fig. 5. Invariably, there is very much less of the upper component of the doublet than of the lower. Trypsin activation converts the proenzyme doublet to a corresponding active enzyme doublet, with molecular weights of 52,000 and 48,000, respectively. The active collagenase, which is often generated during the purification, migrates in the same position as do the trypsin-activated enzyme forms. Again, these species are shown in Fig. 5. These values are the same whether the sample has been reduced or not. In addition, a pro-
TABLE
10% fetal bovine serum-containing and applied to a 5 X lo-cm column accomplished with a 2-liter linear 7.5, 10 mM CaClr. Sixteen-milliliter to 0.05% Brij, and frozen.
tein band, roughly corresponding in molecular weight to approximately that of a dimer of the zymogen (-120,000), often appears on the gel. The corresponding dimer of the active enzyme has not been observed. When the continuous SDSpolyacrylamide electrophoresis system of Fairbanks et al. (27) is employed, the apparent molecular weight of the zymogen forms is approximately 58,000, and of the active forms, 48,000(not shown). Of further interest is the fact that the apparent dimer of proenzyme, which is visible when the system of King and Laemmli is used for electrophoresis, is invariably indetectable using the system of Fairbanks et cd, in which no stacking gel is employed and in which the SDS concentration is lo-fold higher. The apparent molecular weight of rat
I
PURIFICATION OF RAT UTERUS COLLAGENASE
Stage Crude medium Heparin-Sepharose DEAE-Sepharose AcA-44 Blue-Sepharose
Protein hg) 251,000 27.8 5.1 1.6 1.3
Specific activity (pg collagen/min/mg) .05 360 966 2786 3280
Purification (fold) 1 7,200 19,320 55,720 65,600
Total units (pg collagen/min) 13,350 10,008 4,888 4,502 4,087
Yield (%) 100 75 37 34 31
PURIFICATION
OF RAT
UTERINE
FRACTION
PROCOLLAGENASE
289
NUMBER
FIG. 2. DEAE-Sepharose chromatography. Four pooled dextran sulfate elutions from heparinSepharose columns as shown in Fig. 1 (total volume = 950 ml) were applied to a 2.6 X S-cm column of DEAE-Sepharose as described under Materials and Methods. Elution was accomplished with a linear gradient of NaCl, O-O.3 M in 700 ml of 50 mM Tris-HCl, pH 7.5, 10 mM Ca&, 0.05% Brij. Fraction size was 11 ml. Fractions 33-43 were pooled.
uterine collagenase as assessed by gel filtration varies with the matrix employed. Using Sephadex G-150, the molecular weight of the active enzyme appears to be approximately 67,000 (not shown), whereas when AcA-44 is used as the column material an apparent M, of -57,000 is obtained for the zymogen form and 47,000 for the active species. In order to resolve these disparate values, both the zymogen and the active form of the enzyme were subjected to equilibrium ultracentrifugation analysis. Plots of log OD vs 2’ were linear, and molecular weights were esti-
FRACTION FIG. 3. Gel filtration
mated from the slopes of these plots. The values obtained (data not shown) were 58,000 and 47,000, respectively, in excellent agreement with that obtained by electrophoresis in the systems of King and Laemmli, of Fairbanks et al, and by gel filtration on AcA-44. In addition, there was no evidence for higher molecular weight species, such as dimeric forms of either zymogen or active collagenase. Figure 6 is a summary of the molecular weight data using calibration curves from (A) the electrophoresis system of King and Laemmli, (B) the electrophoresis system of Fair-
NUMBER
on AcA-44. The pooled collagenase-containing fractions from DEAE-Sepharose chromatography (total volume = 120 ml) were prepared for application to a 1.6 X 200-cm column of AcA-44 as described under Materials and Methods. Five-milliliter fractions were collected. Fractions 39-46 (40 ml) were pooled.
290
ROSWIT,
HALME,
FRACTION
AND
JEFFREY
NUMBER
FIG. 4. Blue-Sepharose chromatography. The pooled collagenase-containing fractions from gel filtration were prepared for application to a 1.6 X 3-cm column of Cibachron blue-Sepharose as described under Materials and Methods. Elution was accomplished with a linear gradient of NaCl, O-O.9 M in 170 ml. Fraction size was 5 ml. Fractions 11-17 (35 ml) were pooled and dialyzed against 4 liters of 50 mM Tris, pH 7.5, 10 mM CaClr, 0.05% Brij.
banks et al, and (C) gel filtration on Ultrogel AcA 44. The position and molecular weight of the standards are noted below the lines; the corresponding positions of the various forms of uterine collagenase
and their apparent molecular weights are noted above the lines. It should be noted that in contrast to the human skin fibroblast procollagenase, the corresponding components of the rat
DIMER
PRO ACTIVE
I FIG. 5. Polyacrylamide
gel electrophoresis of rat uterine collagenase species I. Appropriate samples of purified collagenase were prepared for electrophoresis in the system of King and Laemmli as described under Materials and Methods. Lane I: 20 fig of purified collagenase; Lane II: 40 pg of purified collagenase; Lane III: 25 gg of purified collagenase; Lane IV: 25 pg of purified collagenase activated with 2 pg of trypsin; 10 pg of soybean trypsin inhibitor was then added to inhibit further trypsin activity. These proteins appear at the bottom of Lane IV. The position of the molecular weight markers used as standards are indicated between lanes II and III.
OF RAT
PURIFICATION
UTERINE
200 120,OCi- Proenzyme Domer 100
80 60
40
Phorphorylase
b
20
Carbanx
10’
Anhydrose J
I
h
0 0
B
Try,,sm Inh,b,,or
‘Oo.2 L
b
1
1
1
1
0.3
04
0.5
06 h
0.7
200
0.8 1
-J
091
C
k 57,000
Chymotrypr,nagen
'"l.O I
J
1.0
12 1
14I
1.6 I
Proenzyme
A
1.8 I
2.0 I
2I 2
2.4 I 2.6 I
"e'",
FIG. 6. Molecular weight estimation. Calibration curves were constructed from the mobilities of standard proteins on the polyacrylamide gel electrophoresis systems of King and Laemmli (A) and Fairbanks et al (B) and the elution position of standard proteins on a 1.6 X 266-cm column of AcA-44 (C). The position and the apparent molecular weight of each collagenase species is indicated in each panel.
uterine proenzyme have not been fully separated. Gel filtration on AcA-44, however, has allowed a partial separation of the zymogen doublet bands (Fig. 7). Analysis of these partially separated species indicates that each has essentially equal collagenolytic activity. Although the final product of the purification scheme is predominantly in trypsin-activatable form, some active collagenase is often generated during the purification process. The amount of this active collagenase varies from batch to batch, but
PROCOLLAGENASE
291
is normally no more than 20% of the total. Effective resolution of active- from trypsinactivable enzyme is obtained by a final chromatographic separation on DEAESepharose, with elution accomplished by a sodium chloride gradient. Active collagenase, when present, elutes from the column before the inactive species (Fig. 8). In the experiment depicted in Fig. 8, the protein comprising pool I displayed full, trypsin-independent activity whereas that in pool III was completely inactive until trypsin activation was accomplished. The protein in pool II, as expected, displayed approximately 60% of full activity in the absence of trypsin activation. The amino acid composition of rat uterine procollagenase is presented in Table II, together with the composition of human skin fibroblast procollagenase and of the tadpole collagenase as reported by Nagai (14) for comparison. Rat uterine procollagenase is unusually rich in glycine (17’7 residues0000 residues) and the hydroxy amino acid (serine + threonine = 211 residues/1000 residues). It differs markedly in composition from the human skin fibroblast and the tadpole skin enzymes. Rat uterine procollagenase is considerably more acidic than is the human skin fibroblast zymogen. The uterine proenzyme contains no detectable hydroxyproline, hydroxylysine, or amino sugars. Preliminary studies on the action of rat uterine collagenase indicate that the active enzyme is capable of degrading gelatin but is unable to cleave a variety of noncollagen proteins, including cytochrome c, chymotrypsinogen A, and ovalbumin. DISCUSSION
Rat uterine collagenase represents a second mammalian collagenolytic enzyme continuously available in sufficient amount to allow the study of its chemical, physical, and enzymatic properties. Human skin fibroblast collagenase was the first such enzyme to be purified in such quantities, and subsequent studies of this enzyme have yielded considerable knowledge on the enzymology of collagenolysis. In the case of the rat uterine collagenase, 30-40 liters of
292
ROSWIT,
HALME,
AND
JEFFREY
0.3 s a
COLLAGENASE ACTIVITY “C - GLYCINE CPM
0.2
3-5
45
40
FRACTION
50
NUMBER
FIG. 7. Partial separation of zymogen doublet forms on AeA-44. Fractions were pooled as indicated from a typical gel-filtration run on AcA-44 as described under Materials and Methods. Equal amounts of protein were subjected to electrophoresis in the system of King and Laemmli, and the resultant stained bands are displayed in the inset. Similar aliquots of each pool were activated with trypsin and assayed for collagenase activity. The results of the assay are displayed beneath each lane of the polyacrylamide gel.
medium typically yield l-2 mg of pure collagenase. This quantity of medium is now routinely available approximately once a month, with an easily maintained cell culture system consisting of 12-16 600~cm* Nunc culture trays. Thus, ample quantities of the enzyme are available for kinetic analysis, for production of antibodies, and for physical and comparative biochemical studies. The purification of the rat uterine collagenase proved to be considerably more difficult than that of the human fibroblast enzyme, in large part because of a lower initial concentration in crude medium, but also because of important differences between the two proteins. The uterine procollagenase is much less stable than is the human, and requires the continuous presence of a nonionic detergent for the preservation of its activity. In this respect, it is similar to the porcine synovial collagenase described by Cawston et aL (13). Second, the uterine collagenase exhibits very different ion-exchange behavior from
that of the human fibroblast enzyme. For human fibroblast collagenase, chromatography on CM-cellulose at pH 7.5 represents the crucial step in the purification (1’7);this step is not as useful for the rat enzyme. Although uterine collagenase does bind to CM-Sepharose at neutral pH, significant losses of activity occur together with activation of the remaining proenzyme. The crucial feature of the purification of rat uterine collagenase is the avid binding of the zymogen to heparin conjugated to Sepharose. Purifications of at least ?OOOto 7500-fold are routinely achieved. Under optimized conditions, using smaller volumes, fold-purifications as high as 15,000 have been observed. In our large preparative runs the usual 7500-fold purification results in solutions containing 8-10% collagenase. It is tempting to suggest that the high affinity displayed by rat uterine collagenase for acidic glycosaminoglycans has physiologic relevance in addition to the usefulness of these molecules as tools in
PURIFICATION
OF RAT
UTERINE
0
5
FRACTION
293
PROCOLLAGENASE
10
15
20
NUMBER
FIG. 8. Separation of zymogen and active species of rat uterine collagenase on DEAE-Sepharose. Purified collagenase (1 mg) was applied to a 1.6 X 2.0-cm DEAE-sepharose column equilibrated with 50 mM Tris-HCl, 10 mM CaCla, 0.05% Brij. Elution was accomplished with a 200-ml gradient, O-O.3 M NaCl. Fractions were pooled as indicated by the roman numerals. Aliquots of each pool containing approximately equal amounts of protein were subjected to electrophoresis in the system of King and Laemmli and the results displayed in the inset.
purification. Collagenase exists and acts in the extracellular space, which contains a wide variety of such species, as well as acidic glycoproteins. Interactions of collagenase in viva with these macromolecules could play a major role in the availability or the geographic distribution of those enzymes. The rat uterine enzyme displays certain similarities to human skin fibroblast collagenase, as well as some marked differences. In common with the human enzyme, the uterine collagenase appears to be secreted as a pair of zymogen molecules of very similar molecular weights. Invariably there is much more of the lighter component of the zymogen than of the heavier. Both components are converted by trypsin to a pair of active enzyme molecules, with an apparent loss of approx 10,000 D. Interestingly, recent studies of the collagenase produced by cultured rabbit synovial cells indicate that the enzyme is also secreted from the cells as a doublet, one of which appears to be glycosylated, the other not (28). On the other hand, human skin
fibroblast procollagenase contains no detectable carbohydrate (18); hence its doublet nature cannot be explained by selective glycosylation. Whether differential glycosylation is responsible for the doublet comprising rat uterine collagenase is not known at this time. The two components of the doublet have been partially separated by gel filtration on AcA-44, and each appears to have approximately equal activity against native guinea pig Type I collagen fibrils (Fig. 8). Because the upper component is invariably present in smaller amounts than the lower species, however, subtle differences in enzymatic behavior may exist and must await the availability of large amounts of upper component for rigorous study. Another similarity between the rat uterine and the human skin collagenase is the existence of puzzling differences in the apparent molecular weights displayed by these molecules, depending upon the gelfiltration matrix used to assess this parameter. Nevertheless, values for the molecular weight of uterine procollagenase as
ROSWIT,
HALME,
AND
TABLE
JEFFREY
II
AMINO ACID COMPOSITION OF RAT UTERINE PROCOLLAGENASE: COMPARISON WITH HUMAN SKIN AND TADPOLE ENZYMES
Amino acid Asp Glu Thr Ser Pro GUY Ala Val Leu Ileu CYS Met 5r Phe His LYS Aw
Uterus proenzyme (res/molecule) 55 67 27 96 26 102 42 17 29 11 11 6 21 18 30 22 16
Uterus proenzyme (res/lOOO res)
Human skin proenyzme (res/lOOO res)
Tadpole collagenase (res/lOOO res)
95 115 46 165 45 177 72 29 51 18 19 10 37 31 52 38 28
117 55 55 55 61 78 66 56 61 39 13 16 39 71 36 64 54
139 98 57 57 65 83 67 54 78 46 5 10 44 70 26 52 34
well as that of the active form, as determined by two electrophoresis systems and equilibrium sedimentation ultracentrifugation, were in good agreement. Hence, pending further physicochemical studies, we are assigning a value of 58,000 D to rat uterine procollagenase and of 48,000 D to the active species. The amino acid composition of the rat uterine enzyme is clearly different fromthat of either human skin fibroblast collagenase or the tadpole collagenase reported by Nagai (14). In fact, the composition of the human skin fibroblast enzyme more closely resembles that of the tadpole than of the uterine collagenase. The human skin and tadpole enzymes display startling similarities in their compositional content of hydroxy, amino, and nonpolar amino acids. In common with these enzymes, however, uterine procollagenase contains no hydroxyproline or hydroxylysine and no amino sugars. The relative predominance of acidic amino acids in the uterine zymogen, when compared to human skin fibroblast proenzyme, is consistent with the anion-exchange chromatographic properties
of the rat collagenase when compared with the human, but fails to explain the affinity of the enzyme for acidic mucopolysaccharides or its ability to bind to CM-cellulose at neutral pH. It may well be that some structural feature of the molecule results in a spatial separation of clusters of oppositely charged amino acids, allowing for interaction with either positively or negatively charged matrices under identical conditions. Interestingly, the content of glycine (177 residues/1000 residues) in rat uterine collagenase is quite high. If a large portion of these glycine residues was concentrated in one area of the collagenase molecule a triple-helical, collagen-like structure could exist, as it does in the Ci, components of the complement system (29). Such a rigid rod-like structure might be useful for maintaining two domains of the enzyme separated in space. Uterine collagenase cleaves native collagen in solution into the classical TC* and TCBfragments. As noted in an earlier study (8), the further cleavages catalyzed in collagen by crude preparations of the enzyme are not observed with the pure molecule.
PURIFICATION
OF RAT
UTERINE
Uterine collagenase, however, does cleave gelatin effectively, as do the human skin fibroblast (17,23) and tadpole collagenases (14). The uterine enzyme, however, is completely devoid of activity against denatured and cytochrome c, chymotrypsinogen, ovalbumin, all of which contain glycylleucyl or glycyl-isoleucyl bonds potentially susceptible to attack by collagenase. It thus appears to be quite specific for collagenous substrates. As a collagenase, the specific activity of the rat uterine enzyme is approximately 3000 pg collagen/min/mg protein at 37”C, using native, reconstituted guinea pig skin collagen fibrils as substrate. This value is approximately threefold higher than the value determined for human skin fibroblast collagenase using the identical collagen as substrate. Other reported values for the specific activities of pure collagenases have ranged from 2800 (mouse bone collagenase) to nearly 54,000 (porcine synovial collagenase). It should be emphasized that no two values have been determined under identical conditions, and this wide range of values presently in the literature must be viewed in this light. Nevertheless, it is clear from the present investigation that major differences can exist. Studies on the fundamental kinetic parameters of rat uterine collagenase should contribute to our understanding of the nature of these differences. ACKNOWLEDGMENTS The authors are indebted to Dr. Gregory Grant and Mr. James Sacchettini for performing the amino acid analyses, and to Mr. Walter Nulty for the ultracentrifugal analyses. REFERENCES 1. HARKNESS, M. L. R., AND HARKNESS, R. D. (1956) J. Phgsiol 132, 492-501. 2. HARKNESS, R.D., AND MORALEE, B. E. (1956) J. Physiol 132, 502-508. 3. MORRIONE, T. G., AND SEIFTER, S. (1962) J. Ex~. Med 115, 357-362. 4. WOESSNER, J. F., AND BREWER, T. H. (1963) Biochem J. 89, 75-81.
PROCOLLAGENASE
295
5. GROSS, J. (1982) in Cell Biology of Extracellular Matrix (Hay, E. D., ed.), pp. 217-258, Plenum, New York. 6. JEFFREY, J. J., AND GROSS,J. (19’70) Biochelnistry 9,268-273. 7. JEFFREY, J. J., COFFEY, R. J., AND EISEN, A. Z. (1971) Biochim Biophys. Ada 252,136-142. 8. HALME, J., TYREE, B. T., AND JEFFREY, J. J. (1980) Arch, Biochem. Biophys. 199, 51-60. 9. JEFFREY, J. J., COFFEY, R. J., AND EISEN, A. Z. (1971) Biochim. Biophys. A& 252, 143-150. 10. KOOB, T. J., AND JEFFREY, J. J. (1974) B&him Biophys. Acta 354, 61-70. 11. TYREE, B. T., HALME, J., AND JEFFREY, J. J. (1980) Arch Biochem Biophys. 202, 314-317. 12. SAKAMOTO, S., SAKAMOTO, M., GOLLIHABER, P., AND GLIMCHER, M. (1978) Arch Biochem. Biophys. 188, 438-449. 13. CAWSTON, T. E., AND TYLER, J. E. (1979) Biodwm. J 183, 647-686. 14. HORI, H., AND NAGAI, Y. (1979) Biochim. Biophgs. 566, 211-221. 15. HAYASHI, T., NAKAMURA, T., HORI, H., AND NAGAI, Y. (1980) J. Biockem. 87, 993-995. 16. SUNABA, H., HAYASHI, T., HORI, H., AND NAGAI, Y. (1981) Co11 Res. 1, 177-185. 17. STRICKLIN, G. P., BAUER, E. A., JEFFREY, J. J., AND EISEN, A. Z. (1977) Biochemistry 16, X07-1614. 18. STRICKLIN, G. P., EISEN, A. Z., BAUER, E. A., AND JEFFREY, J. J. (1978) Bioch.emistry 17, 23312337. 19. WELGUS, H. G., JEFFREY, J. J., STRICKLIN, G. P., ROSWIT, W. T., AND EISEN, A. 2. (1980) J. Bid Chem. 255, 6806-6813. 20. WELGUS, H. G., JEFFREY, J. J., AND EISEN, A. Z. (1981) J. Biol CAewz. 256, 9511-9515. 21. WELGUS, H. G., JEFFREY, J. J., AND EISEN, A. Z. (1981) J. Bid Chem. 256, 9516-9520. 22. STRICKLIN, G. P., JEFFREY, J. J., ROSWIT, W. T., AND EISEN, A. Z. (1983) Biochemistry 22, 6168. 23. WELGUS, H. G., JEFFREY, J. J., STRICKLIN, G. P., AND EISEN, A. Z. (1982) J. Biol. Chem. 257, 11534-11539. 24. NAGAI, Y., LAPIERE, C. M., AND GROSS, J. (1966) Biochemistry 5, 3123-3130. 25. GROVES, W. E., DAVIS, F. C., AND SELLS, B. (1968) AnaL Biochem 22, 195-210. 26. KING, J., AND LAEMMLI, V. K. (1971) J. Mol. Bid 62, 465-477. 27. FAIRBANKS, G., STECH, T. L., AND WALLACH, D. F. H. (1971) Biochemistry 10, 2606-2616. 28. YPHANTIS, D. A. (1964) Biochemistry 3,297-317. 29. NAGASE, H., JACKSON, R. C., BRINCKERHOFF, C. E., VATER, C. A., AND HARRIS, E. D. (1981) J. Bid Chem 256, 11951-11954.
Purification WILLIAM Divisitm
and Properties
of Rat Uterine Procollagenase’
T. ROSWIT, J. HALME,
AND
JOHN J. JEFFREY’
of Dermatobgy, Department of Medicine, and the Department of Biological Chemistry, Washington University School of Medicine, St. Louis, Missouri 63110 Received February
28, 1983, and in revised form March 31, 1983
A procollagenase from monolayer cultures of postpartum rat uterine cells has been purified. The crucial step in the purification is the binding of the procollagenase from crude, fetal bovine serum-containing culture medium to heparin-Sepharose, followed by elution with extremely low concentrations (5-10 IIM) of dextran sulfate. Resultant eluates contain 8-10s procollagenase. Purification is completed by ion-exchange chromatography on DEAE-Sepharose, gel filtration on AcA-44, and chromatography on blue-Sepharose. Rat uterine procollagenase appears as a protein doublet of M, - 58,000, as indicated by two polyacrylamide gel electrophoresis systems, by AcA-44 chromatography, and by equilibrium sedimentation ultracentrifugal analysis. The proenzyme forms are converted by trypsin to an active enzyme doublet of M, - 48,000. Small amounts of active enzyme, which are often generated during the purification, are electrophoretically indistinguishable from trypsin-activated collagenase. Active collagenase can be separated from the zymogen forms by DEAE-Sepharose chromatography. The two forms of the proenzyme doublet can be partially separated by gel filtration on AcA-44 and preliminary analysis indicates each has equal collagenolytic activity. The amino acid analysis of rat uterine collagenase reveals it to be markedly different from two other vertebrate collagenases whose composition is known. The uterine proenzyme is unusually rich in glycine and in the hydroxy amino acids and is considerably more acidic than the human skin fibroblast collagenase, consistent with the different ionexchange behavior of the two molecules. The specific activity of rat uterine collagenase at 37°C is approximately 3000 pg collagen/min/mg, using native reconstituted guinea pig skin type I collagen fibrils as substrate. The enzyme cleaves denatured collagen, but fails to attack a variety of noncollagen proteins.
Carefully controlled collagen degradation is a crucially important process which accompanies the postpartum involution of the mammalian uterus (1, 2). During this involutionary period, the collagen accumulated during pregnancy is removed in a very short time (2-4). Numerous studies have emphasized the importance of the enzyme collagenase as the initiator of the process of collagenolysis in a wide variety of tissues (5), including the uterus (6). In 1 Supported by USPHS Grant HD 05291, AM 12129, and AM 16184. ’ To whom correspondence should be addressed. 285
this latter tissue, collagenase activity appears shortly after delivery, is maintained at high levels throughout uterine remodeling, and then declines to low or undetectable levels when resorption is complete (7). A collagenase produced by the rat uterus has been isolated and partially purified from the medium of explant (6) and cell (8) cultures derived from this tissue. Studies on the hormonal regulation of this enzyme have suggested an important role for progesterone in preventing the appearance of active collagenase in the rat uterus (g-11), raising the possibility that a major physiological function of the re0003-9861/%3 $3.00 Copyright 0 1983by AcademicPress,Inc. All rights of reproductionin any form reserved.
286
ROSWIT,
HALME,
productive hormone is to regulate the rate and the geometry of collagen degradation in the uterus. Studies on the properties of collagenolytic enzymes have, in general, been hampered by the availability of only small quantities of protein, often in impure form. A number of these enzymes have been purified in amounts usually too small to allow extensive characterization (5, 12-14). The collagenase of tadpole skin, however, has been purified and studied with respect to amino acid composition (14) and certain kinetic properties (15,16). One collagenase, that from cultured human skin fibroblasts, has been purified as a zymogen in continuously available chemically significant quantities (17-18). The availability of milligram amounts of this proenzyme has permitted studies on substrate binding (19), fundamental kinetic parameters (20), collagen type specificity (20-22), and zymogen activation (23). Thus, to more precisely define not only the nature of the chemistry of the collagenolytic process itself in the uterus, but also the mechanisms of the physiologic regulation of the process, we have employed mass cultures of rat uterine cells, described previously, as a source of crude uterine collagenase. The present communication details the complete purification of rat uterine procollagenase from the medium of these cells. METHODS
AND
MATERIALS
CeU culture wudmok Primary cultures of postpartum rat uterine cells were grown as previously described (8), either in plastic trays (600 cm’, Nunc, Inc.) or plastic roller bottles (850 cmz, Falcon) in Dulbecco’s modified Eagle’s medium, high glucose, containing glutamine (KC Biologicals), 30 mM Hepes3 (pH 7.6), 200 units/ml penicillin, 200 pi/ml streptomycin, and 10% heat-inactivated fetal bovine serum (KC Biologicals). Harvested medium was brought to 50 mM Tris, 10 mM CaC1, and stored at -20°C until used. In a typical purification, 40-80 liters of harvested medium was employed as starting material. Assay procedures. Collagenase activity was mea-
’ Abbreviations used: Hepes, 4-(2-hydroxyethyl)-lpiperazineethanesulfonic acid; SDS, sodium dodecyl sulfate; CM, 0-(carboxymethyl).
AND
JEFFREY
sured at 3’7°C using native [‘4Clglycine-labeled reconstituted guinea pig skin collagen (24) as a substrate. Typically, 2550 pl of a 0.4% solution of collagen (sp act approx 50,000 cpm/mg) was allowed to aggregate at 3’7°C for at least 18 h before use in the assay. After appropriate incubation with enzyme solutions, usually in a volume of 100 ~1, the assays were terminated by the addition of 200 ~1 of 50 mM Tris, 10 mM CaCIz, pH 7.6, followed by centrifugation at 12,OOOgfor 15 min to sediment unreacted substrate. The entire supernatant solution was then removed and its radioactivity determined by liquid scintillation spectrometry. Protein concentration was determined by the spectrophotometric method of Groves et al (25) using bovine serum albumin as a standard. Chrmmdographic procedures. All chromatography was performed at 4°C. Flow rates were maintained with a Gilson peristaltic pump. Absorbance of eluates was monitored using a Gibson double-beam spectrophotometer. Heparin-Sepharose chromatography. Heparin-Sepharose (Pharmacia) was packed in a 5 X lo-cm column and equilibrated with 50 mM Tris, 10 mM CaClz, 200 mM NaCl, pH 7.5. Crude culture medium (lo-20 liters) was clarified by passage through a 3-pm filter (Gelman) and solid sodium chloride was added until the conductivity was equal to that of the column buffer. The medium was then pumped through the column at 550 ml/h. The column was washed with 50 mM Tris-HCl, 10 mM CaClz and then eluted with a linear gradient of dextran sulfate of 0 to 50 pg/ml in a total volume of 2 liters at a flow rate of 350 ml/ h. Fractions were assayed for collagenase as described above, active fractions were pooled, and a sufficient volume of a 30% solution of the nonionic detergent Brij was added to a final concentration of 0.05%. DEAE-Sephurose chromatography. The pooled, collagenase-containing fractions eluted from heparinSepharose were applied to a 2.6 X g-cm column of DEAE-Sepharose equilibrated with 50 mM Tris-HCl, 10 mM CaClz, 0.05% Brij, pH 7.5. The sample was applied at 175 ml/h, and the column was washed with the same buffer. Enzyme elution was accomplished with a 700-ml gradient of 0 to 0.3 M NaCl in column buffer; the flow rate was 175 ml/h. Collagenase assays were performed as described, enzyme-containing fractions were pooled, and the NaCl concentration of the pool was lowered by a brief (2-h) dialysis vs 50 mM Tris, 10 mM CaClz, 0.05% Brij. Gel jltratiox The pooled, dialyzed DEAE eluate was applied to a 1.6 X 1.5-cm column of heparinSepharose equilibrated in the dialysis buffer. The enzyme protein was eluted directly onto a 1.6 X 200-cm column of Ultrogel AcA-44 (LKB) with 50 mM Tris, 10 mM CaCI,, 0.05% Brij, containing 0.8 M NaCl. The AcA-44 column was equilibrated with 50 mM Tris, 10 mM CaClz, 0.3 M NaCl, 0.05% Brij, and the column
PURIFICATION
OF RAT
UTERINE
was run at a flow rate of 15 ml/h. Collagenase was assayed as described, and enzyme-containing fractions were pooled and dialyzed for 4 h vs 50 mM Tris, 10 mM CaClr, 0.05% Brij. Blue-Sepharose chromatography. The pooled collagenase-containing fractions from AcA-44 were applied to a 1.6 x 3-cm column of blue-Sepharose (Pierce) equilibrated with 50 mM Tris, 10 mhf CaCl,, 0.05% Brij at a flow rate of 50 ml/h. The column was eluted with a 170-ml linear gradient of O-O.9 M NaCl in the same buffer. Collagenase was assayed in the column fractions and active fractions were pooled and dialyzed vs 50 mM Tris, 10 mM CaCl,, 0.05% Brij. Polyacrylamidegel electrophwesis. As needed, samples from the above column chromatographic procedures were prepared for electrophoresis by dialysis against 10,000 vol of 0.1 N HAc followed by lyophylization. The dried samples were dissolved directly in the appropriate sample buffer for the electrophoretic method selected. Protein standards routinely included on all gels were obtained from Pharmacia and included phosphorylase b (94,000), serum albumin (67,000), ovalbumin (43,000), carbonic anhydrase (30,000). soybean trypsin inhibitor (20,100), and a-lactalbumin (14,000). Two electrophoretic systems were employed: the discontinuous gel system of King and Laemmli (26) and the continuous gel system of Fairbanks et a2. (27). Following electrophoresis the protein bands were stained with 1% Coomassie brilliant blue. UltracentrZfugal analysis. Sedimentation equilibrium analysis was performed in a Spinco Model E analytical ultracentrifuge, equipped with an ultraviolet scanning system at 280 mm, at 30,000 rpm and 4°C by the method of Yphantis (28). Protein samples were dissolved in 50 mM Tris, 10 mM CaCl,, 0.05% Brij. After equilibrium was attained, the data were plotted as the log of ODza, vs the square of the distance (x2) from the center of the rotor. Amino acid analysis. Amino acid analyses were performed by the Washington University Protein Sequencing Facility, using a Waters high-performance liquid chromatography amino acid analyzer equipped with a fluorescence detector.
RESULTS
The binding of rat uterine collagenase from fetal bovine serum-containing culture medium to heparin-Sepharose and the elution of the enzyme from this matrix with dextran sulfate is illustrated in Fig. 1. The vast bulk of the protein of the medium fails to bind to this matrix. In addition, the collagenase is eluted with dextran sulfate (M, - 500,000) at a concentration of approximately 10 nM (5 pg/ml),
PROCOLLAGENASE
287
resulting in a typical purification of some 7000-fold (Table I). Generally this single purification step yields solutions in which collagenase constitutes 8-10% of the total protein. Beyond this point in the purification of the enzyme, 0.05% Brij was routinely included in all buffers as first suggested by Cawston et al. (13) for porcine synovial collagenase in order to minimize loss of activity, particularly severe when solutions of collagenase were frozen and thawed. The eluates from heparin-Sepharose are subjected to chromatography on DEAESepharose (Fig. 2), and the collagenase is eluted with a gradient of sodium chloride. This step both removes the dextran sulfate from the solution and separates the enzyme from the bulk of the major contaminant at this point in the purification, bovine serum albumin. The resultant collagenase-containing fractions are concentrated by binding to a small (1.6 X 1.5-cm) column of heparinSepharose and eluted with 0.8 M NaCI directly onto a 1.6 X 200-cm column of AcA44. The enzyme emerges from the column as a nearly symmetrical peak with an apparent M, of 57,000 (Fig. 3). At this stage of the purification, the collagenase exists predominantly in a trypsin-activatable form. A small percentage, however, which appears to vary with batches of starting material, is present as active enzyme. Final purification of rat uterine collagenase is achieved by absorption of the enzyme to Cibachron blue-Sepharose followed by elution with a sodium chloride gradient (Fig. 4). The collagenase is retained on the column until the salt concentration is quite high (0.4-0.5 M). The purification of the rat uterine cell collagenase is detailed in Table I. The overall fold-purification from crude serum-containing medium is approximately 75,000, with an overall yield of -30%. Some 25-50 pg of pure collagenase is obtained per liter of culture medium. Final specific activity of the uterine enzyme is approximately 3000 mg collagen/min/mg protein at 37°C (Table I), with native, reconstituted guinea pig skin collagen fibrils as substrate.
288
ROSWIT,
HALME,
FRACTION
AND
JEFFREY
NUMBER
FIG. 1. Heparin-Sepharose chromatography. Twelve liters of culture medium was prepared as under Materials and Methods of heparin-Sepharose at a flow rate of 550 ml/h. Elution was gradient of dextran sulfate, O-50 pg/ml, in 50 mM Tris-HCl, pH fractions were collected. Fractions 44-68 were pooled, brought
Electrophoresis of the uterine collagenase in the SDS-containing polyacrylamide gel electrophoresis system of King and Laemmli (26) reveals a proenzyme protein doublet with apparent molecular weights of approx 64,000 and 60,000. These forms are shown in the composite polyacrylamide gel in Fig. 5. Invariably, there is very much less of the upper component of the doublet than of the lower. Trypsin activation converts the proenzyme doublet to a corresponding active enzyme doublet, with molecular weights of 52,000 and 48,000, respectively. The active collagenase, which is often generated during the purification, migrates in the same position as do the trypsin-activated enzyme forms. Again, these species are shown in Fig. 5. These values are the same whether the sample has been reduced or not. In addition, a pro-
TABLE
10% fetal bovine serum-containing and applied to a 5 X lo-cm column accomplished with a 2-liter linear 7.5, 10 mM CaClr. Sixteen-milliliter to 0.05% Brij, and frozen.
tein band, roughly corresponding in molecular weight to approximately that of a dimer of the zymogen (-120,000), often appears on the gel. The corresponding dimer of the active enzyme has not been observed. When the continuous SDSpolyacrylamide electrophoresis system of Fairbanks et al. (27) is employed, the apparent molecular weight of the zymogen forms is approximately 58,000, and of the active forms, 48,000(not shown). Of further interest is the fact that the apparent dimer of proenzyme, which is visible when the system of King and Laemmli is used for electrophoresis, is invariably indetectable using the system of Fairbanks et cd, in which no stacking gel is employed and in which the SDS concentration is lo-fold higher. The apparent molecular weight of rat
I
PURIFICATION OF RAT UTERUS COLLAGENASE
Stage Crude medium Heparin-Sepharose DEAE-Sepharose AcA-44 Blue-Sepharose
Protein hg) 251,000 27.8 5.1 1.6 1.3
Specific activity (pg collagen/min/mg) .05 360 966 2786 3280
Purification (fold) 1 7,200 19,320 55,720 65,600
Total units (pg collagen/min) 13,350 10,008 4,888 4,502 4,087
Yield (%) 100 75 37 34 31
PURIFICATION
OF RAT
UTERINE
FRACTION
PROCOLLAGENASE
289
NUMBER
FIG. 2. DEAE-Sepharose chromatography. Four pooled dextran sulfate elutions from heparinSepharose columns as shown in Fig. 1 (total volume = 950 ml) were applied to a 2.6 X S-cm column of DEAE-Sepharose as described under Materials and Methods. Elution was accomplished with a linear gradient of NaCl, O-O.3 M in 700 ml of 50 mM Tris-HCl, pH 7.5, 10 mM Ca&, 0.05% Brij. Fraction size was 11 ml. Fractions 33-43 were pooled.
uterine collagenase as assessed by gel filtration varies with the matrix employed. Using Sephadex G-150, the molecular weight of the active enzyme appears to be approximately 67,000 (not shown), whereas when AcA-44 is used as the column material an apparent M, of -57,000 is obtained for the zymogen form and 47,000 for the active species. In order to resolve these disparate values, both the zymogen and the active form of the enzyme were subjected to equilibrium ultracentrifugation analysis. Plots of log OD vs 2’ were linear, and molecular weights were esti-
FRACTION FIG. 3. Gel filtration
mated from the slopes of these plots. The values obtained (data not shown) were 58,000 and 47,000, respectively, in excellent agreement with that obtained by electrophoresis in the systems of King and Laemmli, of Fairbanks et al, and by gel filtration on AcA-44. In addition, there was no evidence for higher molecular weight species, such as dimeric forms of either zymogen or active collagenase. Figure 6 is a summary of the molecular weight data using calibration curves from (A) the electrophoresis system of King and Laemmli, (B) the electrophoresis system of Fair-
NUMBER
on AcA-44. The pooled collagenase-containing fractions from DEAE-Sepharose chromatography (total volume = 120 ml) were prepared for application to a 1.6 X 200-cm column of AcA-44 as described under Materials and Methods. Five-milliliter fractions were collected. Fractions 39-46 (40 ml) were pooled.
290
ROSWIT,
HALME,
FRACTION
AND
JEFFREY
NUMBER
FIG. 4. Blue-Sepharose chromatography. The pooled collagenase-containing fractions from gel filtration were prepared for application to a 1.6 X 3-cm column of Cibachron blue-Sepharose as described under Materials and Methods. Elution was accomplished with a linear gradient of NaCl, O-O.9 M in 170 ml. Fraction size was 5 ml. Fractions 11-17 (35 ml) were pooled and dialyzed against 4 liters of 50 mM Tris, pH 7.5, 10 mM CaClr, 0.05% Brij.
banks et al, and (C) gel filtration on Ultrogel AcA 44. The position and molecular weight of the standards are noted below the lines; the corresponding positions of the various forms of uterine collagenase
and their apparent molecular weights are noted above the lines. It should be noted that in contrast to the human skin fibroblast procollagenase, the corresponding components of the rat
DIMER
PRO ACTIVE
I FIG. 5. Polyacrylamide
gel electrophoresis of rat uterine collagenase species I. Appropriate samples of purified collagenase were prepared for electrophoresis in the system of King and Laemmli as described under Materials and Methods. Lane I: 20 fig of purified collagenase; Lane II: 40 pg of purified collagenase; Lane III: 25 gg of purified collagenase; Lane IV: 25 pg of purified collagenase activated with 2 pg of trypsin; 10 pg of soybean trypsin inhibitor was then added to inhibit further trypsin activity. These proteins appear at the bottom of Lane IV. The position of the molecular weight markers used as standards are indicated between lanes II and III.
OF RAT
PURIFICATION
UTERINE
200 120,OCi- Proenzyme Domer 100
80 60
40
Phorphorylase
b
20
Carbanx
10’
Anhydrose J
I
h
0 0
B
Try,,sm Inh,b,,or
‘Oo.2 L
b
1
1
1
1
0.3
04
0.5
06 h
0.7
200
0.8 1
-J
091
C
k 57,000
Chymotrypr,nagen
'"l.O I
J
1.0
12 1
14I
1.6 I
Proenzyme
A
1.8 I
2.0 I
2I 2
2.4 I 2.6 I
"e'",
FIG. 6. Molecular weight estimation. Calibration curves were constructed from the mobilities of standard proteins on the polyacrylamide gel electrophoresis systems of King and Laemmli (A) and Fairbanks et al (B) and the elution position of standard proteins on a 1.6 X 266-cm column of AcA-44 (C). The position and the apparent molecular weight of each collagenase species is indicated in each panel.
uterine proenzyme have not been fully separated. Gel filtration on AcA-44, however, has allowed a partial separation of the zymogen doublet bands (Fig. 7). Analysis of these partially separated species indicates that each has essentially equal collagenolytic activity. Although the final product of the purification scheme is predominantly in trypsin-activatable form, some active collagenase is often generated during the purification process. The amount of this active collagenase varies from batch to batch, but
PROCOLLAGENASE
291
is normally no more than 20% of the total. Effective resolution of active- from trypsinactivable enzyme is obtained by a final chromatographic separation on DEAESepharose, with elution accomplished by a sodium chloride gradient. Active collagenase, when present, elutes from the column before the inactive species (Fig. 8). In the experiment depicted in Fig. 8, the protein comprising pool I displayed full, trypsin-independent activity whereas that in pool III was completely inactive until trypsin activation was accomplished. The protein in pool II, as expected, displayed approximately 60% of full activity in the absence of trypsin activation. The amino acid composition of rat uterine procollagenase is presented in Table II, together with the composition of human skin fibroblast procollagenase and of the tadpole collagenase as reported by Nagai (14) for comparison. Rat uterine procollagenase is unusually rich in glycine (17’7 residues0000 residues) and the hydroxy amino acid (serine + threonine = 211 residues/1000 residues). It differs markedly in composition from the human skin fibroblast and the tadpole skin enzymes. Rat uterine procollagenase is considerably more acidic than is the human skin fibroblast zymogen. The uterine proenzyme contains no detectable hydroxyproline, hydroxylysine, or amino sugars. Preliminary studies on the action of rat uterine collagenase indicate that the active enzyme is capable of degrading gelatin but is unable to cleave a variety of noncollagen proteins, including cytochrome c, chymotrypsinogen A, and ovalbumin. DISCUSSION
Rat uterine collagenase represents a second mammalian collagenolytic enzyme continuously available in sufficient amount to allow the study of its chemical, physical, and enzymatic properties. Human skin fibroblast collagenase was the first such enzyme to be purified in such quantities, and subsequent studies of this enzyme have yielded considerable knowledge on the enzymology of collagenolysis. In the case of the rat uterine collagenase, 30-40 liters of
292
ROSWIT,
HALME,
AND
JEFFREY
0.3 s a
COLLAGENASE ACTIVITY “C - GLYCINE CPM
0.2
3-5
45
40
FRACTION
50
NUMBER
FIG. 7. Partial separation of zymogen doublet forms on AeA-44. Fractions were pooled as indicated from a typical gel-filtration run on AcA-44 as described under Materials and Methods. Equal amounts of protein were subjected to electrophoresis in the system of King and Laemmli, and the resultant stained bands are displayed in the inset. Similar aliquots of each pool were activated with trypsin and assayed for collagenase activity. The results of the assay are displayed beneath each lane of the polyacrylamide gel.
medium typically yield l-2 mg of pure collagenase. This quantity of medium is now routinely available approximately once a month, with an easily maintained cell culture system consisting of 12-16 600~cm* Nunc culture trays. Thus, ample quantities of the enzyme are available for kinetic analysis, for production of antibodies, and for physical and comparative biochemical studies. The purification of the rat uterine collagenase proved to be considerably more difficult than that of the human fibroblast enzyme, in large part because of a lower initial concentration in crude medium, but also because of important differences between the two proteins. The uterine procollagenase is much less stable than is the human, and requires the continuous presence of a nonionic detergent for the preservation of its activity. In this respect, it is similar to the porcine synovial collagenase described by Cawston et aL (13). Second, the uterine collagenase exhibits very different ion-exchange behavior from
that of the human fibroblast enzyme. For human fibroblast collagenase, chromatography on CM-cellulose at pH 7.5 represents the crucial step in the purification (1’7);this step is not as useful for the rat enzyme. Although uterine collagenase does bind to CM-Sepharose at neutral pH, significant losses of activity occur together with activation of the remaining proenzyme. The crucial feature of the purification of rat uterine collagenase is the avid binding of the zymogen to heparin conjugated to Sepharose. Purifications of at least ?OOOto 7500-fold are routinely achieved. Under optimized conditions, using smaller volumes, fold-purifications as high as 15,000 have been observed. In our large preparative runs the usual 7500-fold purification results in solutions containing 8-10% collagenase. It is tempting to suggest that the high affinity displayed by rat uterine collagenase for acidic glycosaminoglycans has physiologic relevance in addition to the usefulness of these molecules as tools in
PURIFICATION
OF RAT
UTERINE
0
5
FRACTION
293
PROCOLLAGENASE
10
15
20
NUMBER
FIG. 8. Separation of zymogen and active species of rat uterine collagenase on DEAE-Sepharose. Purified collagenase (1 mg) was applied to a 1.6 X 2.0-cm DEAE-sepharose column equilibrated with 50 mM Tris-HCl, 10 mM CaCla, 0.05% Brij. Elution was accomplished with a 200-ml gradient, O-O.3 M NaCl. Fractions were pooled as indicated by the roman numerals. Aliquots of each pool containing approximately equal amounts of protein were subjected to electrophoresis in the system of King and Laemmli and the results displayed in the inset.
purification. Collagenase exists and acts in the extracellular space, which contains a wide variety of such species, as well as acidic glycoproteins. Interactions of collagenase in viva with these macromolecules could play a major role in the availability or the geographic distribution of those enzymes. The rat uterine enzyme displays certain similarities to human skin fibroblast collagenase, as well as some marked differences. In common with the human enzyme, the uterine collagenase appears to be secreted as a pair of zymogen molecules of very similar molecular weights. Invariably there is much more of the lighter component of the zymogen than of the heavier. Both components are converted by trypsin to a pair of active enzyme molecules, with an apparent loss of approx 10,000 D. Interestingly, recent studies of the collagenase produced by cultured rabbit synovial cells indicate that the enzyme is also secreted from the cells as a doublet, one of which appears to be glycosylated, the other not (28). On the other hand, human skin
fibroblast procollagenase contains no detectable carbohydrate (18); hence its doublet nature cannot be explained by selective glycosylation. Whether differential glycosylation is responsible for the doublet comprising rat uterine collagenase is not known at this time. The two components of the doublet have been partially separated by gel filtration on AcA-44, and each appears to have approximately equal activity against native guinea pig Type I collagen fibrils (Fig. 8). Because the upper component is invariably present in smaller amounts than the lower species, however, subtle differences in enzymatic behavior may exist and must await the availability of large amounts of upper component for rigorous study. Another similarity between the rat uterine and the human skin collagenase is the existence of puzzling differences in the apparent molecular weights displayed by these molecules, depending upon the gelfiltration matrix used to assess this parameter. Nevertheless, values for the molecular weight of uterine procollagenase as
ROSWIT,
HALME,
AND
TABLE
JEFFREY
II
AMINO ACID COMPOSITION OF RAT UTERINE PROCOLLAGENASE: COMPARISON WITH HUMAN SKIN AND TADPOLE ENZYMES
Amino acid Asp Glu Thr Ser Pro GUY Ala Val Leu Ileu CYS Met 5r Phe His LYS Aw
Uterus proenzyme (res/molecule) 55 67 27 96 26 102 42 17 29 11 11 6 21 18 30 22 16
Uterus proenzyme (res/lOOO res)
Human skin proenyzme (res/lOOO res)
Tadpole collagenase (res/lOOO res)
95 115 46 165 45 177 72 29 51 18 19 10 37 31 52 38 28
117 55 55 55 61 78 66 56 61 39 13 16 39 71 36 64 54
139 98 57 57 65 83 67 54 78 46 5 10 44 70 26 52 34
well as that of the active form, as determined by two electrophoresis systems and equilibrium sedimentation ultracentrifugation, were in good agreement. Hence, pending further physicochemical studies, we are assigning a value of 58,000 D to rat uterine procollagenase and of 48,000 D to the active species. The amino acid composition of the rat uterine enzyme is clearly different fromthat of either human skin fibroblast collagenase or the tadpole collagenase reported by Nagai (14). In fact, the composition of the human skin fibroblast enzyme more closely resembles that of the tadpole than of the uterine collagenase. The human skin and tadpole enzymes display startling similarities in their compositional content of hydroxy, amino, and nonpolar amino acids. In common with these enzymes, however, uterine procollagenase contains no hydroxyproline or hydroxylysine and no amino sugars. The relative predominance of acidic amino acids in the uterine zymogen, when compared to human skin fibroblast proenzyme, is consistent with the anion-exchange chromatographic properties
of the rat collagenase when compared with the human, but fails to explain the affinity of the enzyme for acidic mucopolysaccharides or its ability to bind to CM-cellulose at neutral pH. It may well be that some structural feature of the molecule results in a spatial separation of clusters of oppositely charged amino acids, allowing for interaction with either positively or negatively charged matrices under identical conditions. Interestingly, the content of glycine (177 residues/1000 residues) in rat uterine collagenase is quite high. If a large portion of these glycine residues was concentrated in one area of the collagenase molecule a triple-helical, collagen-like structure could exist, as it does in the Ci, components of the complement system (29). Such a rigid rod-like structure might be useful for maintaining two domains of the enzyme separated in space. Uterine collagenase cleaves native collagen in solution into the classical TC* and TCBfragments. As noted in an earlier study (8), the further cleavages catalyzed in collagen by crude preparations of the enzyme are not observed with the pure molecule.
PURIFICATION
OF RAT
UTERINE
Uterine collagenase, however, does cleave gelatin effectively, as do the human skin fibroblast (17,23) and tadpole collagenases (14). The uterine enzyme, however, is completely devoid of activity against denatured and cytochrome c, chymotrypsinogen, ovalbumin, all of which contain glycylleucyl or glycyl-isoleucyl bonds potentially susceptible to attack by collagenase. It thus appears to be quite specific for collagenous substrates. As a collagenase, the specific activity of the rat uterine enzyme is approximately 3000 pg collagen/min/mg protein at 37”C, using native, reconstituted guinea pig skin collagen fibrils as substrate. This value is approximately threefold higher than the value determined for human skin fibroblast collagenase using the identical collagen as substrate. Other reported values for the specific activities of pure collagenases have ranged from 2800 (mouse bone collagenase) to nearly 54,000 (porcine synovial collagenase). It should be emphasized that no two values have been determined under identical conditions, and this wide range of values presently in the literature must be viewed in this light. Nevertheless, it is clear from the present investigation that major differences can exist. Studies on the fundamental kinetic parameters of rat uterine collagenase should contribute to our understanding of the nature of these differences. ACKNOWLEDGMENTS The authors are indebted to Dr. Gregory Grant and Mr. James Sacchettini for performing the amino acid analyses, and to Mr. Walter Nulty for the ultracentrifugal analyses. REFERENCES 1. HARKNESS, M. L. R., AND HARKNESS, R. D. (1956) J. Phgsiol 132, 492-501. 2. HARKNESS, R.D., AND MORALEE, B. E. (1956) J. Physiol 132, 502-508. 3. MORRIONE, T. G., AND SEIFTER, S. (1962) J. Ex~. Med 115, 357-362. 4. WOESSNER, J. F., AND BREWER, T. H. (1963) Biochem J. 89, 75-81.
PROCOLLAGENASE
295
5. GROSS, J. (1982) in Cell Biology of Extracellular Matrix (Hay, E. D., ed.), pp. 217-258, Plenum, New York. 6. JEFFREY, J. J., AND GROSS,J. (19’70) Biochelnistry 9,268-273. 7. JEFFREY, J. J., COFFEY, R. J., AND EISEN, A. Z. (1971) Biochim Biophys. Ada 252,136-142. 8. HALME, J., TYREE, B. T., AND JEFFREY, J. J. (1980) Arch, Biochem. Biophys. 199, 51-60. 9. JEFFREY, J. J., COFFEY, R. J., AND EISEN, A. Z. (1971) Biochim. Biophys. A& 252, 143-150. 10. KOOB, T. J., AND JEFFREY, J. J. (1974) B&him Biophys. Acta 354, 61-70. 11. TYREE, B. T., HALME, J., AND JEFFREY, J. J. (1980) Arch Biochem Biophys. 202, 314-317. 12. SAKAMOTO, S., SAKAMOTO, M., GOLLIHABER, P., AND GLIMCHER, M. (1978) Arch Biochem. Biophys. 188, 438-449. 13. CAWSTON, T. E., AND TYLER, J. E. (1979) Biodwm. J 183, 647-686. 14. HORI, H., AND NAGAI, Y. (1979) Biochim. Biophgs. 566, 211-221. 15. HAYASHI, T., NAKAMURA, T., HORI, H., AND NAGAI, Y. (1980) J. Biockem. 87, 993-995. 16. SUNABA, H., HAYASHI, T., HORI, H., AND NAGAI, Y. (1981) Co11 Res. 1, 177-185. 17. STRICKLIN, G. P., BAUER, E. A., JEFFREY, J. J., AND EISEN, A. Z. (1977) Biochemistry 16, X07-1614. 18. STRICKLIN, G. P., EISEN, A. Z., BAUER, E. A., AND JEFFREY, J. J. (1978) Bioch.emistry 17, 23312337. 19. WELGUS, H. G., JEFFREY, J. J., STRICKLIN, G. P., ROSWIT, W. T., AND EISEN, A. 2. (1980) J. Bid Chem. 255, 6806-6813. 20. WELGUS, H. G., JEFFREY, J. J., AND EISEN, A. Z. (1981) J. Biol CAewz. 256, 9511-9515. 21. WELGUS, H. G., JEFFREY, J. J., AND EISEN, A. Z. (1981) J. Bid Chem. 256, 9516-9520. 22. STRICKLIN, G. P., JEFFREY, J. J., ROSWIT, W. T., AND EISEN, A. Z. (1983) Biochemistry 22, 6168. 23. WELGUS, H. G., JEFFREY, J. J., STRICKLIN, G. P., AND EISEN, A. Z. (1982) J. Biol. Chem. 257, 11534-11539. 24. NAGAI, Y., LAPIERE, C. M., AND GROSS, J. (1966) Biochemistry 5, 3123-3130. 25. GROVES, W. E., DAVIS, F. C., AND SELLS, B. (1968) AnaL Biochem 22, 195-210. 26. KING, J., AND LAEMMLI, V. K. (1971) J. Mol. Bid 62, 465-477. 27. FAIRBANKS, G., STECH, T. L., AND WALLACH, D. F. H. (1971) Biochemistry 10, 2606-2616. 28. YPHANTIS, D. A. (1964) Biochemistry 3,297-317. 29. NAGASE, H., JACKSON, R. C., BRINCKERHOFF, C. E., VATER, C. A., AND HARRIS, E. D. (1981) J. Bid Chem 256, 11951-11954.