Tissue Inhibitors of Metalloproteinases in Endometrium of ...

8 downloads 0 Views 1MB Size Report
Expression of mRNA for TIMP-1 and TIMP-2 was examined by Northern analysis of endometrial RNA derived from steroid-treated ovariectomized ewes and from ...
BIOLOGY OF REPRODUCTION 53, 302-311 (1995)

Tissue Inhibitors of Metalloproteinases in Endometrium of Ovariectomized Steroid-Treated Ewes and during the Estrous Cycle and Early Pregnancy' 2 Anne L. Hampton, Anna R. Butt, Simon C. Riley, and Lois A. Salamonsen

PrinceHenry'sInstitute of Medical Research, Clayton, Victoria 3168, Australia ABSTRACT Tissue inhibitors of metalloproteinases (TIMPs) have an important role inremodeling of tissues and are likelyto be implicated inuterine function, including embryo implantation and placentation. Expression of mRNA for TIMP-1 and TIMP-2 was examined by Northern analysis of endometrial RNA derived from steroid-treated ovariectomized ewes and from intact ewes during the estrous cycle and early pregnancy. Expression of mRNA for TIMP-1 (transcript size 0.9 kb), high inovariectomized ewes, was substantially reduced by estrogen and to a lesser extent by progesterone. Incyclic and pregnant animals, abundance remained low until Day 10 and then increased, with high abundance continuing to Day 20 inthe pregnant animals. Two transcripts for TIMP-2 were detected in ovine tissues-the 3.5-kb transcript and, in greater abundance, the 1.0-kb transcript. Inovariectomized ewes, endometrial abundance of both transcripts was low, and it decreased following estrogen treatment but was stimulated by progesterone alone or progesterone inthe presence of estrogen. Abundance of TIMP2 mRNA increased from Day 4to Day 14 of the cycle. During early pregnancy, expression of the 1.0-kb transcript increased from Day 4to Days 12-14 and was maintained at a high level to Day 20, whereas the 3.5-kb transcript decreased after Day 14 to very low levels by Day 20. Incontrast with this pattern of regulated expression of TIMP, mRNA for proMMP-1 and for proMMP-3 was not detectable inany of the same tissues by Northern analysis. TIMP-1 protein was immunolocalized to both epithelium and stroma of intact endometrium, and the intensity of immunostaining was correlated with mRNA levels. TIMP-1 was secreted by both epithelial and stromal cells inprimary culture, and its identity was confirmed by Western analysis, while reverse zymography demonstrated TIMP-1 and TIMP-2 along with a putative ovine TIMP-3 inthe culture medium from both cell types. The precise role of TIMP inthe endometrium remains to be established. INTRODUCTION Tissue remodeling in the endometrium occurs during the normal ovarian cycle in association with angiogenesis, implantation, and placentation [11. The matrix metalloproteinase (MMP) family of enzymes is critical for the degradation of extracellular matrix associated with tissue remodeling [2, 3], and these enzymes are specifically inhibited by tissue inhibitors of metalloproteinases (TIMPs). TIMP type 1 (TIMP-1) is a 29-kDa glycoprotein, whereas TIMP-2 of 21 kDa is not glycosylated. The two proteins share 38% identity of nucleotide and 66% similarity of amino acid sequence [4]. A third, more recently described member of the family, TIMP-3, is a 24-kDa protein that binds strongly to extracellular matrix [5, 6]. TIMP-1 binds specifically with 1:1 stoichiometry and inactivates the connective tissue MMPs (MMP-1, collagenase; MMP-3, stromelysin-1; and MMP-2 and -9, gelatinases A and B, respectively). TIMP-1 also binds to the latent form of MMP-9. TIMP-2 binds active forms of the same enzymes but also binds to the latent form of MMP2 [7-9]1. Both TIMP-1 and TIMP-2 are mitogenic for a number of cell types [10, 11]. Synthesis and secretion of these inhibitors are regulated largely by the modification of mRNA transcription rates. Factors regulating TIMP-1 include estrogen (E) and progesterone (P) [12, 13], transforming growth factor [14], interleukin-1 1151, interleukin-6 [16] and tumor

necrosis factor [17, 181. However, it is clearly evident that different mechanisms exist for regulation of TIMP-1, TIMP2, and TIMP-3 expression in a variety of tissues and in normal and transformed cells [6, 19-21]. Little is known of the expression of TIMP in the uterus. Cultured ovine endometrial stromal, but not epithelial, cells release proMMP-1, proMMP-2, and proMMP-3 [22, 23], and MMPs were detected in ovine uterine flushings in early pregnancy [1], but production of TIMP was not examined. The present study was undertaken to establish 1) whether mRNA for TIMP-1, TIMP-2, proMMP-1, and proMMP-3 are detectable by Northern analysis in ovine endometrium, 2) whether their expression is regulated by the ovarian steroids, E and P, and 3) whether there are any temporal changes during the estrous cycle or early pregnancy. Secretion of TIMPs by mixed and individual endometrial cell types was demonstrated by reverse zymography and Western analysis, and the cellular localization of TIMP-1 in ovine endometrium was established by immunohistochemistry. MATERIALS AND METHODS Tissues All animal experimentation was approved by the Animal Ethics Committees at the Monash Medical Centre and at the Victorian Institute for Animal Science. Steroid-treated, ovariectomized ewes. Parous Corriedale ewes were ovariectomized (OVX) at least 1 mo before the start of the experiment. Animals were randomly allocated to four groups (n = 4 per group) of similar live

Accepted April 4, 1995. Received December 21, 1993. 'Supported by the NH&MRC of Australia. 2 Correspondence: Dr. Lois Salamonsen, Prince Henry's Institute of Medical Research, PO Box 5152, Clayton, Victoria 3168, Australia. FAX: + 61 3 550 6125.

302

TIMP EXPRESSION IN OVINE ENDOMETRIUM

weight: 1) not treated (OVX); 2) E alone, Days 1-12; 3) P alone, Days 3-12, and 4) E+P, Days 1-2 and 3-12, respectively. E was administered as estradiol-17P from silastic implants, and P was administered intravaginally [24]. The ewes were hysterectomized at the end of the treatment period. Endometrium (combined caruncular and intercaruncular regions) was dissected from myometrium and frozen in liquid nitrogen. Cyclic andpregnant ewes. Corriedale ewes (n = 47) were run with either vasectomized or intact rams fitted with marking harnesses for detection of ewes in estrus (Day 0). On Days 4, 10, 12, 14, or 16 after estrus (n = 3 or 4 per group) or on Days 4, 7, 8, 9, 10, 12, 14, 15, 16, 17, and 20 of pregnancy (n = 2-4 per group), the uterus of each ewe was flushed with saline, pregnancy was confirmed by the presence of one or more embryos, and the ewes were hysterectomized. On Days 17 and 20 of pregnancy, flushing was not performed and pregnancy was confirmed by the presence of attached trophoblast tissue, which was removed manually from the endometrium after hysterectomy and before the endometrium was dissected from the myometrium (Day 17 = first day of definitive attachment). Endometrial tissue was collected as above. In addition, wedges of endometrium plus myometrium were fixed for 6 h in Carnoy's fixative, washed with 90% ethanol, and embedded in wax. Range of tissues. Samples of adrenal gland, aorta, atrium, cerebellum, CL, frontal cortex, kidney cortex, liver, whole ovary (without CL), pituitary, placentome, skeletal muscle, spleen, and submandibular gland were collected from an entire ewe at slaughter and snap-frozen in liquid nitrogen. Slices of corpora lutea were fixed in Camoy's solution. Ovine endometrial cells. Mixed epithelial and stromal cells or highly purified preparations of each of these cell types were prepared by digestion of endometrium from E + P-treated ewes with bacterial collagenase (type III; Worthington, Freehold, NJ; [20, 21]) and plated into flasks at a density of 1 x 106 cells/mL in medium 199 (M199) with 10% fetal calf serum and antibiotics. Where separate cell types were required, they were prepared from the mixed cells by differential trypsinization [23] and replated into fresh flasks. After 48 h, the cells were washed free of serum and incubated in serum-free M199 either without or with 40 nmol/L phorbol myristate acetate (PMA) for 48 h. Then the medium was collected, centrifuged to remove cellular debris, and stored at - 20C. The cells were washed, harvested with trypsin/versene, and taken for immediate extraction of RNA. NorthernAnalysis Complementary DNA probes against ovine TIMP-1 (900 bp) and TIMP-2 (438 bp) were labeled with a3 2P-dCTP (Bre-

303

satec, Adelaide, Australia) to specific activities of 1-3 X 109 cpm/pg. cDNA probes against human proMMP-1 (1.6 kb) and human proMMP-3 (1.4 kb) were similarly labeled. In previous studies [1, 23] these probes have been shown to hybridize to proMMP-1 and proMMP-3 mRNA from ovine endometrial cells treated in vitro with PMA. A rat glyceraldehyde 3-phosphate dehydrogenase (GAPDH) cRNA probe, complementary to nucleotides 96 to 660 of the rat GAPDH cDNA clone, was labeled with a32P-UTP to a specific activity of 108 cpm/gLg, and an oligonucleotide probe (30mer) for rat 18S ribosomal RNA was end-labeled with y- 32 P-ATP to a specific activity of 108-109 cpm/ipg. The latter two probes have previously been shown to cross-hybridize with ovine RNA [23]. Total RNA was isolated from tissue samples (500 mg) using guanidinium isothiocyanate lysis and centrifugation through cesium chloride [25] and from cultured cells by a single step extraction with guanidinium isothiocyanate, phenol, and chloroform [25]. For Northern blotting, 20 jig of total RNA was denatured in 1 mol/L glyoxal with 50% dimethyl sulfoxide, subjected to electrophoresis in a 1.2% agarose gel, and transferred to Hybond nylon membranes (Amersham International, Sydney, Australia) by capillary blotting. An RNA ladder (0.24-9.5 kb; Bethesda Research Laboratories, Bethesda, MD) was used for size identification of bands. Each nylon membrane was baked at 80°C for 2 h, UV cross-linked for 10 min, and prehybridized for 4 h at 42°C in hybridization buffer (50% formamide, singlestrength Denhardt's solution [50 strength is 1% each of Ficoll, polyvinylpyrrolidone, and BSA], 5-strength SSPE [0.75 mol/L NaCI, 50 mmol/L NaH 2PO 4.2H 20, 5 mmol/L EDTA, pH 7.4], and 150 mg/mL preboiled herring sperm DNA). Blots were hybridized with the appropriate 32P-labeled TIMP probe (2 X 106 cpm/mL) for 16-18 h at 42°C. The blots were washed at 42°C in double-strength SSC (singlestrength SSC is 0.15 mol/L sodium chloride, 0.015 mol/L sodium citrate, pH 7.4) in 0.1% SDS. There was an additional wash step in 0.1-strength SSC in 0.1% SDS for the TIMP-1 probe and in 0.5-strength SSC in 0.1% SDS for the TIMP-2 probe. The blots containing endometrial tissues were also probed with the 3 2P-labeled proMMP-1 and proMMP-3 probes for 17 h at 42°C and washed in double-strength SSC in 0.1% SDS at 42°C (proMMP-1) or in single-strength SSC in 0.1% SDS at 42°C (proMMP-3). For the GAPDH probe (2 X 10 6 cpm/ml), prehybridization was for 4 h at 42°C in a solution containing 50% formamide, 5-strength SSPE, 0.15 mol/L Tris-HCl (pH 8) in 1% SDS, and 125 IU/ml heparin. Hybridization was for 17 h in the same solution at 60°C, and the blot was washed with 0.2-strength SSC in 0.1% SDS at 70°C. For the 18S probe, hybridization and washing conditions were as described previously [24]. Autoradiography was performed with Fuji RX film (Fuji Photo Film Co. Ltd., Tokyo, Japan). Prior to rehybridization with a different probe, the blot was washed in 0.1% SDS at 950C, allowed

304

HAMPTON ET AL.

to cool for 30 min while shaking, and exposed to Kodak XAR film (Eastman Kodak, Rochester, NY) overnight to check completeness of probe removal. Densitometric analysis was performed using an Olympus Cue 2 (Olympus Corporation, New York, NY) image analyzer. Relative abundance was calculated following correction for loading according to the relative signals for GAPDH or 18S mRNA, and comparisons were made only within a single transcript and within an individual blot. Every blot contained a sample of RNA from mixed endometrial cells as a positive control to confirm appropriate probe labeling and hybridization conditions. Immunohistochemistry The Carnoy's-fixed tissue was embedded in wax, and six micron sections on poly-L-lysine-coated slides were subjected to immunohistochemistry. The primary antiserum was raised in rabbits against purified ovine TIMP-1. Sections were overlaid sequentially with the following reagents at in methanol for 30 min, Trisroom temperature: 3% HO 2 buffered saline (TBS) for 10 min; 0.05% Tween 20 in TBS for 1 h, TBS rinse for 10 min, primary antiserum or normal rabbit serum (NRS) (each 1:3000) diluted in TBS with 0.1% normal horse serum for 3 days at 4°C, TBS wash for 1 h, biotinylated horse anti-rabbit IgG (Vector Laboratories, Burlington, CA) in TBS for 1 h, TBS wash for 10 min, Vectastain ABC reagent (Vector Laboratories) for 45 min, and peroxidase substrate (diaminobenzidine tetrahydrochloride) 1 mg/ml with 3.3 tl 6% H2 O2 for up to 10 min. Parallel sections were treated with primary antiserum and with NRS as negative control. Ovine CL was used as a positive control, and a section was included within each batch of sections to assess day to day variability. All sections were counterstained with 1:10 Harris' hematoxylin (Fisher Scientific, Orangeburg, NY), and the stained sections were dehydrated, mounted, and examined under an Olympus BH2 microscope by two independent observers. Western Blot Analysis Samples of conditioned medium from mixed endometrial cells and from purified cultures of stromal fibroblasts or epithelial endometrial cells (treated or untreated with PMA) were concentrated 20-fold on Centricon filters (molecular weight cut-off, 10 000: Amicon) and subjected to SDS gel electrophoresis under reducing conditions on 10% polyacrylamide gels. The proteins were transferred electrophoretically to Hybond membrane (Amersham). Nonspecific binding of IgG was blocked with 5% BSA (Sigma Chemical Co., St. Louis, MO) in TBS, and the filters were reacted with anti-ovine TIMP-1 at 1:5000 dilution in 5% BSA in TBS at room temperature for 18 h. After extensive washing of the filters with TBS, anti-TIMP antibody bound to the antigen was complexed with biotinylated anti-rabbit IgG

(Vector Laboratories, 1:5000) for 1 h at room temperature and washed in TBS. Protein bands were visualized with the ECL detection system (Amersham) and exposure to X-Omat film (Kodak, Melbourne, Australia) for 15 sec. Reverse Zymography Culture media from mixed endometrial cells were concentrated 20-fold as above and analyzed for the presence of TIMP by zymography of 5 1 aliquots on SDS polyacrylamide gel electrophoresis (12% gel containing 1% gelatin and an MMP preparation) under nonreducing conditions, with use of a kit purchased from University Technologies International Inc. (Calgary, Canada). After thorough washing in a solution of 2.5% Triton X100, 50 mmol/L Tris buffer (pH 7.5), and 5 mmol/L CaCl2, the gel was incubated for 23 h in the same buffer without Triton X100. The gel was stained with Coomassie brilliant blue R-250 (Bio-Rad Laboratories, Richmond, CA), and the presence of TIMP was visualized by the presence of dark blue bands on a cleared background. Controls were a sample of culture medium from ovine CL slices containing both TIMP-1 and TIMP-2 (a gift from Dr. M. Smith, University of Missouri, Columbia, MO), and standards containing mouse TIMP-1 + TIMP-2 and TIMP-3, which were provided with the kit. StatisticalAnalysis Data from densitometric estimations of the relative abundance of each mRNA transcript for each treatment or day within each study group were analyzed for variance using the Kruskal-Wallis test (following correction for loading), and individual comparisons between groups were assessed with the Mann-Whitney U test. Differences were considered significant at the 0.05 level. RESULTS Northern Analysis The ratios of the abundance of mRNA for 18S to that for GAPDH were compared for all samples. There was no consistent variation with any of the treatments in this study, and therefore both 18S and GAPDH were considered to be appropriate for assessment of RNA loading. Range of tissues. Northern blot analyses of total RNA from a range of ovine tissues and from endometrial cells (Fig. 1) showed substantial differences in hybridization of the two TIMP probes to the various tissues. Corrections for loading of total RNA in the samples were made by comparing the hybridization of the test probe (cDNA for TIMP1 or TIMP-2) and the hybridization of the oligonucleotide probe for 18S ribosomal RNA to the RNA in each sample. Both TIMP-1 and TIMP-2 probes hybridized to total RNA derived from ovine endometrial cells that had been subjected to PMA for 48 h (Fig. 1). A single transcript of 0.9 kb

TIMP EXPRESSION IN OVINE ENDOMETRIUM

305

FIG. 1. Northern blot analyses of total RNA (20 pg/lane) from a range of ovine tissues as detailed and from ovine endometrial 32 cells treated in vitro with phorbol myristate acetate (10 Ag total RNA/lane). Hybridization was to P-labeled probes for TIMP-1, TIMP-2, and 18S. Autoradiograms were generated after sequential hybridizations and were for 14 days, 4 days, and 3 days, respectively.

was identified for TIMP-1. Two transcripts-1-.0 kb and 3.5 kb-were identified for TIMP-2, with the 1.0-kb transcript dominant (ratio 1.8:1). By far the greatest abundance of mRNA encoding TIMP1 was in the CL, although there was also substantial hybridization to mRNA from the placentome and the adrenal gland (relative abundance 100:8.5:1.6, respectively). Hybridization was at a lower level, in decreasing order, in ovary, aorta, atrium, submandibular gland, pituitary, and spleen and was below the level of detection for other tissues tested at the exposure times used. In the adrenal gland, a second transcript of 3.0 kb (abundance 0.8 times that of the 0.9-kb transcript) was present. Messenger RNA for TIMP-2 was present in more of the tissues tested than was mRNA for TIMP-1. The relative abundance of mRNA encoding the 1.0-kb transcript of TIMP-2, expressed relative to CL (100) was as follows: placentome (17), aorta (4.5), ovary (4.4), adrenal (3.6), cerebellum (2.4), atrium (2.3), kidney cortex (2.3), frontal cortex (2.2), pituitary (1.1), spleen (0.7). Only very low levels were present in the submandibular gland, liver, and skeletal muscle. The 3.5-kb transcript of TIMP-2 was also detected in all tissues except the CL and the placentome. The relative abundance

of the two transcripts for TIMP-2 varied between tissues from a ratio of 15:1 in the pituitary to 1.4:1 in the adrenal gland. Ovariectomized and steroid-treated animals. The relative hybridization of probes for TIMP-1, TIMP-2, and GAPDH to representative samples of total RNA from the four treatment groups is shown in Figure 2, and the mean relative abundance of each TIMP transcript following densitometry and correction for loading is presented in Table 1. Expression of mRNA for both TIMP-1 and TIMP-2 was regulated by E and P. For TIMP-1 mRNA, the highest level of expression was in OVX ewes; this was significantly reduced by E (p < 0.025) and by P (p < 0.05). The effect of E was significantly greater than that of P (p < 0.05). Expression in the presence of E + P was significantly different from the other groups (p < 0.05) and was intermediate between that seen for each steroid separately. Both 1.0-kb and 3.5-kb mRNA transcripts for TIMP-2 were present in the endometrium of OVX animals and in those treated with E, and there was a significant increase (p < 0.025) in the 1.0-kb transcript following P treatment that was maintained in the presence of E. P also increased the abundance of the 3.5-kb transcript (p = 0.056), but this

306

HAMPTON ET AL. TABLE 2. Mean relative abundance* of mRNA for TIMP-1 and TIMP-2 in endometrium from cycling ewes.

Day of cycle

n

4 10

4 4

12

4

14 16

4 3

TIMP-1 0.9 kb 1.1 0 .9

TIMP-2 3.5 kb

0.3 0.1 bc

1.7 + 0.2b aC

2.4 3.0

0.4 1.5

1.0 kb

a

0.1 ab a 0.9 ± 0.1

0.1 5 1.3

1.1 + 0

1

1b

1.8 +±0.4c 1.0 + 0.9

0.4abc 0. def 1

.8 + 0.1 ad

g

3.1 + 0.2beg 2.5 0.7cf

*Each value represents the mean densitometric unit SEM following correction for loading by comparison with GAPDH mRNA for each sample. 8 a Within each transcript, values with the same superscripts differ significantly from each other (p < 0.05).

FIG. 2. Representative Northern blot analyses showing hybridization of 32P-labeled probes for TIMP-1, TIMP-2, and GAPDH to total RNA (20 pg) from two animals in each of the four in vivo steroid-treatment groups. OVX: ovariectomized, E:estrogen, P: progesterone. Autoradiograms were generated after sequential hybridizations and were for 7 days, 4 days and 4 h, respectively.

increase was significant only when compared with the Etreated animals (p = 0.021). In general, the effects of P on the abundance of the 1.0-kb transcript were greater than those on the 3.5-kb transcript, resulting in a change in the ratio of the two transcripts (3.5 kb:1.0 kb) from 1:1 to 1:2 when P was present. In contrast, no hybridization was seen of any samples in the OVX and steroid-treated groups to the 3 2P-cDNA probes for human proMMP-1 and proMMP-3. Cyclic animals. The relative abundance of mRNA for TIMP-1 increased from Day 4 to Day 16 of the estrous cycle (Table 2) with significant (p < 0.05) differences between its relative abundance on Day 4 and Day 14 and between Day 10 and Days 12 and 14. Although the mean was high for Day 16, there was considerable variability between animals. A similar increase in abundance of the 1.0-kb transcript of TIMP-2 was seen as the cycle progressed, with sigTABLE 1. Mean relative abundance* of mRNA for TIMP-1 and TIMP-2 in endometrium from ovariectomized steroid-treated ewes.

Treatment**

n

OVX

4

OVX-E OVX-P OVX-E + P

4 4 4

TIMP-1 0.9 kb 3.4

0.3

nificant (p < 0.05) increases above both Day 4 and Day 10 levels on Days 12, 14, and 16, and from Day 12 to Day 14. Changes in the 3.5-kb transcript paralleled those of the 1.0kb transcript, increasing significantly (p < 0.05) from Day 4 to Days 10, 12, and 14 but falling again on Day 16. No hybridization was seen to the probes for proMMP-1 and proMMP-3. Early pregnantanimals. Data from groups of animals that were not statistically different from one another and were in adjacent groups were pooled for final analyses. During early pregnancy, the relative abundance of mRNA for TIMP-1 increased significantly (p < 0.05) from Days 910 to Days 12-14 and then remained high to Day 20 (Table 3). The relative abundance of the 1.0-kb transcript for TIMP2 mRNA also showed significant increases (p < 0.05) at Days 12-14, relative to Days 4-8 and 9-10, and remained high to Day 20. The rise on Days 12-14 was paralleled by an overall increase in the 3.5-kb transcript, but the abundance of this transcript decreased again from Days 15-16 and was very low on Days 17-20. There was no hybridization to the probes for proMMP-1 and proMMP-3. Immunohistochemistry Positive staining for TIMP-1 was always seen in cells of the CL (data not shown), and little or no staining was seen in the control sections of endometrium in which normal rabbit serum was substituted for primary antiserum (Fig.

TABLE 3. Mean relative abundance* of mRNA for TIMP-1 and TIMP-2 in endometrium from pregnant ewes.

TIMP-2 3.5 kb a

0.7+ 0.18a 1.9 + 0.4a 1.1 + 0.1a

0.5

0.2

0.2.3 ±+ 0.8 + 0.2a 0.6 0.2

Day of pregnancy

n

TIMP-1 0.9 kb

4 +7 +8 9 + 10 12 + 14 15 + 16 17 + 20

8 5 6 5 4

0.9 + 0.1 bc + 0.9 0.2de 3.6 + 1.1ad 2.9 + 0 .3be cf 5.8 ± 2.2

1.0 kb 0.5

0 .2ab

Cd 0.1 1.5 + 0.1ac 1.5 0.4 bd

*Each value represents the mean relative densitometric unit + SEM following correction for loading by comparison with GAPDH mRNA for each sample. **VX: ovariectomized, OVX-E: with estrogen treatment, OVX-P: with progesterone treatment, OVX-E + P: with estrogen plus progesterone. a'dWithin each transcript, values with the same superscripts differ significantly from each other (p < 0.05).

TIMP-2 3.5 kb a

0.

1.0 2.3 1.1 0.05

1ab + 0.2cd a ce ± 0.4 ± 0.4 0.05bde

1.0 kb ab

3.9 + 0.6 de 3.7 0.3 l ad 8.5 2.7 6.9 0 .8be c 9.1 + 1.4

*Each value represents the mean densitometric unit ± SEM following correction for loading by comparison with GAPDH mRNA for each sample. a-fWithin each transcript, values with the same superscripts differ significantly from each other (p < 0.05).

TIMP EXPRESSION IN OVINE ENDOMETRIUM

307

FIG. 3. Immunohistochemistry for TIMP-1 inendometrium. a) Control in which normal rabbit serum issubstituted for primary antiserum, b) intercaruncular tissue on Day 4 of pregnancy, c) intercaruncular tissue on Day 20 of pregnancy, d) blood vessels at the endometrial-myometrial interface on Day 12. Scale bars = 50 gM.

3a). There was considerable variation in the intensity of immunostaining in the endometrial tissues from individual animals within the same treatment groups. Positive staining was seen both in the epithelium (luminal and glandular) and in the stroma and was generally of greater intensity in inter-caruncular than in caruncular areas within the same tissue, and in glands than in luminal epithelium. Staining in tissue from Day 4 of the cycle or of pregnancy (Fig. 3b) was considerably less than that in tissue taken later in the cycle, and maximal immunostaining was seen in intercaruncular tissue from Day 20 of pregnancy (Fig. 3c). In the caruncular stroma, intense staining could be seen around blood vessels (Fig. 3d), particularly around Day 12. Western Blot Analysis The data from Northern analysis and immunohistochemistry are supported by Western analysis of medium from

culture of endometrial cells. Figure 4 shows a representative Western blot for TIMP-1 containing a positive control sample of culture medium from CL and samples from culture of mixed endometrial cells and from two separate cultures each of purified stromal and epithelial cells. A band of approximately 30 kDa was seen in all samples, confirming the release of TIMP-1 into culture medium by the endometrial cells. No consistent change in TIMP-1 was detected when PMA was included in the cultures (data not shown). Reverse Zymography TIMP-1 and -2, and an additional band with relative mobility corresponding to that of the TIMP-3 standard, were demonstrated in culture medium from ovine endometrial cells analyzed by reverse zymography on a gel that included standards for all three TIMPs and culture medium from ovine luteal cells (Fig. 5). The dominant TIMPs secreted by

308

HAMPTON ET AL.

FIG. 4. Western blot analysis of medium from culture of: lane 1, ovine CL; lane 2, mixed ovine endometrial cells; lanes 3-4, endometrial stromal cells from 2 ewes; lanes 5-6, endometrial epithelial cells from 2 ewes. Molecular weights are derived from markers run on the same gel (x 10-3).

ovine endometrial cells were TIMP-1 and the putative TIMP3. Both of these appeared as doublets. A band of TIMP-2 was also present but much fainter than those for TIMP-1 and the putative TIMP-3. By contrast, medium from culture of luteal cells predominantly contained TIMP-1 and TIMP2. There was no substantial difference between TIMPs in the medium derived from mixed endometrial cells or from highly purified cultures of stromal or epithelial cells without or with PMA stimulation. DISCUSSION Messenger RNA for both TIMP-1 and TIMP-2 but not for proMMP-1 or proMMP-3 was detected by Northern analysis in the ovine endometrium, although mRNA for all of these were detected in cultured ovine endometrial cells treated with PMA (see also [1, 23]). The predominance of one transcript at 0.9 kb for TIMP-1 is in accord with studies in other species [19, 20], although the reason for the presence of two transcripts for TIMP-1 in the adrenal gland alone is not clear. RNA transcripts of two sizes for TIMP-2 were found in the ovine endometrium, and similar transcripts have been demonstrated in both normal and tumor cells of human [19] and murine [20] origin. The origin of these two transcripts re-

FIG. 5. Reverse zymography. Lane 1, mouse TIMP-1 + TIMP-2; lane 2, mouse TIMP-3; lane 3, culture medium from ovine CL; lane 4, culture medium from mixed endometrial cells; lanes 5-6, culture medium from endometrial stromal cells without or with stimulation by phorbol myristate acetate (PMA), respectively; lanes 78, culture medium from endometrial epithelial cells without or with PMA stimulation, respectively. Molecular weight/standards x 1 0- 3.

mains to be determined, although it has been suggested that alternative splicing of 5' untranslated regions may occur [19]. In the previous studies in human and mouse cells, the 3.5-kb species was dominant, compared with the consistent dominance of the 1.0-kb transcript in ovine endometrium in the present study. This appears to be a species rather than a tissue difference, as it is also seen when comparing RNA derived from PMA-treated endometrial cells from human and sheep (Salamonsen and Hampton, unpublished). The levels of expression of mRNA for endometrial TIMP1 and TIMP-2 were lower than those in the CL or the placentome (TIMP-1 and TIMP-2) or the adrenal gland (TIMP2), but they were substantially greater than those in the other sheep tissues tested. In the mouse [27, 28], the highest levels of TIMP-1 mRNA in adult tissues were found in ovary, adrenal gland, and bone, while TIMP-2 was expressed in only the ovary, lung, and thymus of the tissues tested. Further, only low levels of expression of mRNA for TIMP-1 and TIMP-2 were seen until after implantation in these studies, possibly reflecting the use of entire uteri by these authors compared with the use of endometrium in our study. It was not possible to examine the expression of TIMP-3 in ovine tissues. In the human and the mouse, TIMP-3 is expressed in many tissues including the uterus and the placenta [5, 6]. TIMP-1 and TIMP-2 mRNA are regulated by ovarian steroids in ovine endometrium. Both E and P independently down-regulated the expression of TIMP-1 mRNA compared with its high level of expression in the OVX animals, however, the effect of E was much greater than that of P. By contrast, TIMP-2 mRNA was increased by P, whereas E had a small inhibitory effect particularly on the 3.5-kb transcript. In general, the two transcripts for TIMP-2 were similarly regulated, although the relative increase in the 1.0-kb transcript with P was greater than that for the 3.5-kb transcript. Independent regulation of TIMP-1 and TIMP-2 has previously been demonstrated in cell culture [19, 20] and in vivo [21]. E and P have been implicated in regulation of TIMP-1 in the ovary, with a positive correlation between the levels of TIMP-1 mRNA and the ratio of follicular E:P concentration [291, and P stimulated TIMP-1 production from cultured rabbit uterine cervical fibroblasts [12, 30]. Furthermore TIMP-1 production by human endometrial explants was increased by P, suggesting a role for this hormone in the cyclical tissue remodeling that is so pronounced in the human [13]. TIMP2 mRNA levels and protein expression were down-regulated by the combination of estradiol plus synthetic progestin, but not by estradiol alone in human breast cancer cells [31]. Thus the present studies are in general agreement with the concept of independent regulation of TIMP-1 and TIMP2 by E and P, but there appear to be differences between tissues or cell types. The endometrium contains a range of cell types [32], and Northern analysis of whole tissue indicates only the abundance of mRNA throughout the tissue but not local changes in different cell types. Furthermore,

TIMP EXPRESSION IN OVINE ENDOMETRIUM

comparison between in vivo and in vitro studies is particularly difficult in terms of steroid regulation, given the intercellular events stimulated by steroid hormones (for review see [33]). As antiserum specific for ovine TIMP-2 is not currently available, we were unable to explore whether these differences between TIMP-1 and TIMP-2 in steadystate levels of mRNA were reflected in protein levels or distribution. In the sheep estrous cycle, P levels rise by Day 4, reaching a maximum at about Day 8, and start to fall after Day 13. By Day 16, P levels are low, and E is rising rapidly [34]. During the estrous cycle both TIMP-1 and TIMP-2 mRNA levels increased after Day 10 when P was maximal, and TIMP-2 but not TIMP-1 mRNA levels decreased again on Day 16. The greater variability between abundance of the mRNA in different animals on Day 16 may result from very different E levels in individual animals in vivo at a time when E is rising very rapidly. Similar changes for both TIMP1 and TIMP-2 were somewhat surprising because their responsiveness to ovarian steroids differed and because levels of TIMP-1 mRNA remained high while P levels fell. During early pregnancy, TIMP expression remained high, as would be expected in the presence of continuing high levels of P after Day 12. The very low abundance of the 3.5-kb transcript for TIMP-2 on Days 17 and 20 of pregnancy and the marked change in the ratio of the two transcripts at this time is of particular note and may suggest an effect of the conceptus on the expression of the 3.5-kb transcript. The significance of these relative changes in the two TIMP-2 mRNA transcripts during this period requires further investigation. TIMP-1 was immunolocalized in both stromal and epithelial cells of the endometrium. The marked change in the intensity of staining in both these cell types with time of the estrous cycle and of early pregnancy reflects the changes in the mRNA for TIMP-1, suggesting that alterations in mRNA are reflected by changes in the translation product and that both cell types respond to regulatory influences. Western analysis demonstrated that both epithelial and stromal cells in culture released the protein, providing further evidence that both are sites of synthesis in vivo. Both TIMP-1 and TIMP-2 can be synthesized by macrophages [21], and in these cells as in others, are readily subject to control by a variety of regulatory signals. In the mouse there is considerable infiltration of the endometrium by macrophages at the time of implantation [35]. It is unknown whether a similar influx occurs in the sheep uterus or whether TIMPs of macrophage origin contribute to the content of TIMP in the endometrium. However, the immunohistochemical data presented here did not suggest the presence of TIMP-1 in such cells. Reverse zymography of culture medium from endometrial cells in culture further confirmed the production of TIMPs and showed that TIMP-1, TIMP-2, and a putative ovine TIMP-3 are produced by both stromal and epithelial

309

cell types in vitro. Interestingly, there is substantially less TIMP-2 than TIMP-1 or the putative TIMP-3 in these media, and there is no evidence that PMA markedly alters the production of any of these moieties from either cell type under the culture conditions used. Although TIMP-3 is generally found associated with the extracellular matrix, it has previously been found in conditioned medium from cultured chick embryo fibroblasts [36]. In the present study we did not examine the extracellular matrix underlying the cultured cells for the presence of TIMP-3. The nature of this putative ovine TIMP-3, its location in vivo, and whether it is regulated independently and differently from TIMP-1 and TIMP2 in the endometrium remain to be established. In mouse C3H 10T1/2 fibroblasts it is highly inducible but with faster on/off transcription kinetics than for TIMP-1 [6]. The lack of detectable mRNA for proMMP-1 and proMMP-3 in the ovine endometrium, particularly at the time of implantation and placentation, was somewhat surprising since there is substantial remodeling of the endometrial tissue following implantation in association with angiogenesis. However, as MMPs are generally produced only at focal points in a tissue undergoing remodeling, it is likely that more sensitive techniques may be needed to demonstrate the presence of these enzymes or their mRNA. In human endometrium, although there is considerable remodeling of endometrium throughout the menstrual cycle, mRNA for proMMP-1 and proMMP-3 was detected by Northem analysis only in some menstrual and perimenstrual tissue when the degradation of the extracellular matrix was maximal [37]. By contrast, mRNA for TIMP-1 and TIMP-2 was detectable in all human endometrial tissues [371 as in the present study. The roles of TIMP-1, TIMP-2, and TIMP-3 in the normal physiological functions of the endometrium are not yet known. TIMPs have been implicated in regulation of angiogenesis [38], cell growth [10, 39] and differentiation [401, and embryonic development [411, all of which are part of normal uterine physiology. The mitogenic actions of TIMPs [10, 11] may be important in the phenomenal preimplantation growth of ruminant trophoblast between Day 12 and Day 16 [42]. It is important to establish whether these mitogens are present in the uterine lumen and whether specific receptors are present on trophoblast. TIMPs are also likely to have a role in regulating the trophoblast invasion at implantation. In the mouse, addition of TIMP-1 inhibited matrix degradation by trophoblast outgrowths in cultured blastocysts [43] and reduced trophoblast invasion in vitro [44]. In the sheep and other ruminants, invasion is limited to breaching of the subepithelial basal lamina and minor incursions of trophoblast into the underlying stroma [45, 46], but there is extensive and rapid tissue remodeling during placentome formation. MMPs are produced by endometrial cells and by sheep trophoblast, and are present in the uterine lumen at the time of implantation [1]. Fur-

310

HAMPTON ET AL.

thermore, pig trophoblast, which like that of the sheep is noninvasive in vivo, has invasive capacity, shown by its ability to penetrate epithelium and develop in ectopic sites [47]. The expression of TIMPs within the endometrium may therefore be of great importance in limiting trophoblast invasion. We postulate that the regulation of individual TIMPs is coordinated so that they are expressed at appropriate sites and times to regulate the MMPs for which they are most specific. Sequential attainment of the correct balance between individual TIMPs and MMPs would allow both the remodeling of the uterus during the estrous cycle and implantation of the trophoblast at specific caruncular sites. ACKNOWLEDGMENTS We are grateful to Dr. Michael Smith (University of Missouri, Columbia, MO) for providing the ovine TIMP cDNA probes, the anti-ovine TIMP-1, and the culture medium from corpus luteum; to Dr. Hideaki Nagase (University of Kansas Medical Center, Kansas City, KS) for the cDNA probes to human proMMP-1 and proMMP-3; and to Dr. Peter Fuller and Dr. Judith Clements (Prince Henry's Institute of Medical Research, Melbourne, Australia) for the probes for GAPDH and 18S.We thank Bruce Doughton, Richard Fry, and Daphne Vogiagis for assistance with animal handling and surgery, Richard Young for assistance with cell cultures, and Jock Findlay for critically reading the manuscript.

REFERENCES 1. Salamonsen LA, Nagase H, Woolley DE. Matrix metalloproteinases, a role in implantation? In: Glasser SR, Mulholland J, Psychoyos A (eds.), Endocrinology of Embryo-Endometrial Interactions. New York: Plenum Press; 1994: 365-378. 2. Woessner JF Jr. Matrix metalloproteinases and their inhibitors in connective tissue remodeling. FASEB J 1991; 5:2145-2154. 3. Birkedal-Hansen H, Moore WGI, Bodden MK, Windsor LJ, Birkedal-Hansen B, De Carlo A, EnglerJA. Matrix metalloproteinases: a review. Crit Rev Oral Biol Med 1993; 4:199250. 4. Boone TC, Johnson MJ, De Clerck YA, Langley KE, cDNA cloning and expression of a metalloproteinase inhibitor related to tissue inhibitor of metalloproteinases. Proc Natl Acad Sci USA 1990; 87:2800-2804. 5. Apte SS, Mattel MG, Olsen BR. Cloning of the cDNA encoding human tissue inhibitor of metalloproteinases-3 (TIMP-3) and mapping of the TIMP3 gene to chromosome 22. Genomics 1994; 19:86-90. 6. LecoKJ, Khokha R, Pavloff N, Hawkes SP, Edwards DR. Tissue inhibitor of metalloproteinases-3 (TIMP-3) is an extracellular matrix-associated protein with a distinctive pattern of expression in mouse cells and tissues. J Biol Chem 1994; 269:9352-9360. 7. Docherty AJP, Murphy G. The tissue metalloproteinases family and the inhibitor TIMP: a study using cDNAs and recombinant proteins. Ann Rheum Dis 1990; 49:469-479. 8. Stetler-Stevenson WG, Krutzsch HC, Liotta LA. Tissue inhibitor of metalloproteinase (TIMP-2): a new member of the metalloproteinase inhibitor family. J Biol Chem 1989; 264:17374-17378. 9. Goldberg GI, Manner BL, Grant GA, Eisen AZ, Wilhelm S, He C. Human 72k type IV collagenase forms a complex with a tissue inhibitor of metalloproteinases designated TIMP-2. Proc Natl Acad Sci USA 1989; 86:8207-8211. 10. Hayakawa T, Yamashita K, Tanzawa K, Uchijima E, Iwata K. Tissue inhibitor of metalloproteinases from human bone marrow stromal cell line KM 102 has erythroid-potentiating activity, suggesting its possibly bifunctional role in the hematopoietic environment. FEBS Lett 1992; 298:29-32. 11. Nemeth JA, Goolsby CL. TIMP-2, a growth-stimulatory protein from SV40-transformed human fibroblasts. Exp Cell Res 1993; 207:376-382. 12. Sato T, Ito A, Mori Y, Yamashita K, Hayakawa T, Nagase H. Hormonal regulation of collagenolysis in uterine cervical fibroblasts. Modulation of synthesis of procollagenase, prostromelysin and tissue inhibitor of metalloproteinases (TIMP) by progesterone and estradiol-17-beta. BiochemJ 1991; 275:645-650. 13. Marbaix E, Donnez J, Courtoy PJ, Eeckhout Y. Progesterone regulates the activity of collagenase and related gelatinases A and B in human endometrial explants. Proc Natl Acad Sci USA 1992; 89:11789-11793. 14. Edwards DR, Murphy G, Reynolds JJ, Whitham SE, Docherty AJP, Angel P, Heath JK.

Transforming growth factor beta modulates the expression of collagenase and metalloproteinase inhibitor. EMBO J 1987; 6:1899-1904. 15. Ito A,Sato T, Okima Y,Chen LC, Nagase H, Mori Y.Calmodulin differentially modulates the interleukin-1 induced biosynthesis of tissue inhibitor of metalloproteinases and matrix metalloproteinases in human uterine cervical fibroblasts. J Biol Chem 1991; 266: 13598-13601. 16. Sato T, Ito A, Mori Y. Interleukin-6 enhances the production of tissue inhibitor of metalloproteinases (TIMP) but not that of matrix metalloproteinases by human fibroblasts. Biochem Biophys Res Commun 1990; 170:824-829. 17. Chua CC, Chua BHL. Tumor necrosis factor induces mRNA for collagenase and TIMP in human skin fibroblasts. Conn Tissue Res 1990; 25:161-170. 18. Ito A, Sato T, Iga T, Mori Y. Tumor necrosis factor bifunctionally regulates matrix metalloproteinases and tissue inhibitor of metalloproteinases (TIMP) production by human fibroblasts. FEBS Lett 1990; 269:93-95. 19. Stetler-Stevenson WG, Brown PD, Onisto M, Levy AT, Liotta LA. Tissue inhibitor of metalloproteinases-2 (TIMP-2) mRNA expression in tumor cell lines and human tumor tissues. J Biol Chem 1990; 265:13933-13938. 20. Leco KJ, Hayden LJ, Sharma RR, Rocheleau H, Greenberg AH, Edwards DR. Differential regulation of TIMP-1 and TIMP-2 mRNA expression in normal and Ha-ras-transformed murine fibroblasts. Gene 1992; 117:209-217. 21. Shapiro SD, Kobayashi DK, Welgus HG. Identification of TIMP-2 in human alveolar macrophages. Regulation of biosynthesis is opposite to that of metalloproteinases and TliMP-. J Biol Chem 1992 267:13890-13894. 22. Salamonsen LA, Nagase H, Woolley DE. Production of matrix metalloproteinase 3 (stromelysin) by cultured ovine endometrial cells. J Cell Sci 1991; 100:381-385. 23. Salamonsen LA, Nagase H, Suzuki R, Woolley DE. Production of matrix metalloproteinase 1 (interstitial collagenase), and matrix metalloproteinase 2 (gelatinase A: 72 kDa gelatinase) by ovine endometrial cells in vitro: different regulation and preferential expression by stromal fibroblasts. J Reprod Fertil 1993; 93:583-589. 24. Salamonsen LA, Hampton AL, Clements JA, Findlay JK. Regulation of gene expression and cellular localization of prostaglandin synthase by oestrogen and progesterone in the ovine uterus. J Reprod Fertil 1991; 92:393-406. 25. Chirgwin JM, Przybyla AE, MacDonald RJ, Rutter WJ. Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease. Biochemistry 1979; 18:52945299. 26. Chomczynski P, Sacchi N. Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 1987: 162:156-159. 27. Nomura S, Hogan BLM,Willis AJ, HeathJK, Edwards DR. Developmental expression of tissue inhibitor of metalloproteinase (TIMP) RNA. Development 1989; 105:575-584. 28. Waterhouse P, Denhardt DT, Khokha R. Temporal expression of tissue inhibitors of metalloproteinases in mouse reproductive tissues during gestation. Mol Reprod Dev 1993; 35:219226. 29. Curry TE, Mann JS Estes RS, Jones PB. Alpha 2-macroglobulin and tissue inhibitor of metalloproteinases: collagenase inhibitors in human preovulatory ovaries. Endocrinology 1990; 127:63-68. 30. Sakyo K, Ito A,Ogawa C, Mori Y. Hormonal control of collagenase inhibitor production in rabbit uterine cervical fibroblast-like cells. Biochem Biophys Acta 1986; 883:517-522. 31. Van den Brule FA, EngelJ, Stetler-StevensonWG, Li u FT, Sobel ME, Castronovo V. Genes involved tumor in invasion and metastasis are differentially modulated by estradiol and progestin in human breast-cancer cells. IntJ Cancer 1992; 52:653-657. 32. Findlay JK, Salamonsen LA. Paracrine regulation of implantation and uterine function. In: Seppala M (ed.), Bailliere's Clinical Obstetrics and Gynaecology. Factors of ImportanceImplantation for 1990; 5:117-131. 33. Salamonsen LA.Local regulators and the establishment of pregnancy. Reprod Fertil Dev 1992; 4:125-134. 34. Baird DT, McNeilly AS. Gonadotrophic control of follicular development and function during the oestrous cycle of the ewe. J Reprod Fertil Suppl 1981; 30:119-133. 35. De M, Choudhuri R, Wood GW. Determination of the number and distribution of macrophages, lymphocytes and granulocytes in the mouse uterus from mating through implantation. J Leukocyte Biol 1991; 50:252-262. 36. BlenisJ, Hawkes SP. Characterization of a transformation-sensitive protein in the extracellular matrix of chicken embryo fibroblasts. J Biol Chem 1984; 259:11563-11570. 37. Hampton AL, Salamonsen LA. Expression of messenger ribonucleic acid encoding matrix

metalloproteinases and their tissue inhibitors is related to menstruation. J Endocrinol

1994; 141:R1-R3. 38. Moses MA, SudhalterJ, Langer R. Identificationof an inhibitor of neovascularization from cartilage. Science 1990; 248:1408-1410. 39. Westbrook CA, Gasson JC, Gerber SE, Selsted ME, Golde DW. Purification and characterization of human T-lymphocyte-derived erythroid-potentiating activity. J Biol Chem 1984; 259:9992-9996. 40. Docherty AJP, Lyons A,Smith BJ, Wright EM, Stephens PE, Harris TJR, Murphy G, Reyn-

TIMP EXPRESSION IN OVINE ENDOMETRIUM olds JJ. Sequence of human tissues inhibitor of metalloproteinases and its identity to erythroid-potentiating activity. Nature 1985; 318:66-69. 41. Nakanishi Y, Sugiura F, Kishi J. Collagenase inhibitor stimulates cleft formation during early morphogenesis of mouse salivary gland. Dev Biol 1986; 113:201-206. 42. Bindon BM. Systematic study of preimplantation stages of pregnancy in the sheep. Aust J Biol Sci 1971; 24:131-147. 43. Behrendtsen O, Alexander CM, Werb Z. Metalloproteinases mediate extracellular matrix degradation by cells from mouse blastocyst outgrowths. Development 1992; 114:447-456.

311

44. Lala PK, Graham CH. Mechanisms of trophoblast invasiveness and their control: the role of proteases and protease inhibitors. Cancer Metastasis Rev 1990; 369-379. 45. Boshier DP. A histological and histochemical examination of implantation and early placentome formation in sheep. J Reprod Fertil 1969; 19:51-61. 46. Wooding FB, Morgan G. Fetomatemal cell fusion at ruminant implantation. In: Yoshinaga K (ed.), Blastocyst Implantation. Boston: Adams Publishing Group; 1989:117-123. 47. Samuel CA. The development of pig trophoblast in ectopic sites. J Reprod Fertil 1971; 27:494-495.