Abstract
Invasion by extravillous trophoblast into uterine decidua and myometrium with remodeling of spiral arteries is essential for normal human pregnancy and is tightly regulated. Uterine natural killer (uNK) cells appear to be a major maternal regulator of placentation through the secretion of growth factors, cytokines and proteinases.
Matrix metalloproteinase (MMP)-2 and MMP-9 activity in placental bed biopsies was studied using in situ gelatin zymography. Expression by uNK cells of MMP-2, MMP-9 and their tissue inhibitors, TIMP-1, TIMP-2 and TIMP-3, was localized in the placental bed by immunohistochemistry. Levels of MMP-2, MMP-9, TIMP-1, TIMP-2 and TIMP-3 secreted into 24 h cell culture supernatants of isolated uNK and unseparated (total) decidual cells (8–10 and 12–14 weeks gestation, n = 10 each group) were determined by gelatin gel zymography or western blot as appropriate. RESULTS: Gelatinase activity in situ appeared greater in decidua than myometrium. uNK cells showed strong immunostaining for MMP-2 and TIMP-2. MMP-9 activity was lower in uNK cells than total decidual supernatants (8–10 weeks: P = 0.0003; 12–14 weeks: P = 0.0012). In contrast, there was no difference in MMP-2 secreted by either uNK cell or total decidual cultures.
uNK cells from early human pregnancy decidua possess innate protease activity, especially MMP-2, providing further evidence for a role for these cells in regulation of trophoblast invasion and spiral artery remodeling in early placentation.
Introduction
Trophoblast invasion and spiral artery remodeling during early human pregnancy are essential for a successful uncomplicated pregnancy. Extravillous trophoblast cells (EVT) invade the decidualized uterine endometrium and reach the inner third of the myometrium as early as 8 weeks of gestation (interstitial EVT) (Pijnenborg et al., 1981; Pijnenborg et al., 1983). EVT also migrate within the lumen of uterine spiral arteries in a retrograde direction, reaching myometrial portions of the vessels from 10 weeks gestation (endovascular EVT). During early pregnancy uterine spiral arteries are remodeled; their musculoelastic wall is lost and replaced by amorphous fibrinoid material in which intramural EVT are embedded (Pijnenborg et al., 2006). Trophoblast invasion and spiral artery remodeling are regulated by a large number of autocrine and paracrine factors, including cytokines, hormones and oxygen (Bischof et al., 2000; Lala and Chakraborty, 2003; Bischof and Irminger-Finger, 2005). Many of these have been shown to regulate protease activity, especially matrix metalloproteinases (MMPs) and the urokinase plasminogen activator (uPA) system, which play pivotal roles in degradation of basement membranes and extracellular matrix (ECM) (Bischof et al., 2000; Lala and Chakraborty, 2003; Bischof and Irminger-Finger, 2005).
MMPs are proteolytic zinc-requiring enzymes which include the collagenases (MMP-1, -4, -8), the stromelysins (MMP-3, -10, -11) and the gelatinases (MMP-2 and -9) (Nagase and Woessner, 1999). MMP-2 and MMP-9 (also known as gelatinase A and B) are regarded as key enzymes in degradation of the basement membrane, which consists mainly of type IV collagen (Nagase and Woessner, 1999). Their activities also rely on activity of other proteases, which transform pro-MMPs into their active form (Lijnen, 2001). Three tissue inhibitors for MMP (TIMP-1, TIMP-2 and TIMP-3) regulate protease activity. Each TIMP inhibits various MMPs, but TIMP-1 forms complexes specifically with MMP-9 (Goldberg et al., 1992), and TIMP-2 is involved in regulation of MMP-2 activity (Itoh et al., 1998; Morrison et al., 2001; Zhao et al., 2004). Interestingly, TIMP-3 supports activation of MMP-2 via membrane-type MMP, as well as inhibition (Zhao et al., 2004), and thus TIMP-3 is regarded as a major regulator of MMPs in vivo.
MMP-1, MMP-2, MMP-3, MMP-9, TIMP-1 and TIMP-2 have been localized in the placental bed using immunohistochemistry (Huppertz et al., 1998; Huisman et al., 2004; Seval et al., 2004) or in situ hybridization (Bjorn et al., 1997). Using immunohistochemistry, MMP-1 and MMP-3 were localized to EVT (Huppertz et al., 1998). Immunoreactivity for MMP-2 was detected in both decidual cells and EVT, but MMP-9 staining was only observed in areas with abundant EVT (Huppertz et al., 1998; Huisman et al., 2004). TIMP-1 and TIMP-2 expression was also observed on EVT (Huppertz et al., 1998; Seval et al., 2004). However, previous studies have only localized immunostaining of EVT in the placental bed. In addition, there have been no reports of in situ gelatinase activity in human placental bed or decidua to complement immunohistochemical studies, although MMP-2 and MMP-9 activity has been reported in trophoblast cell culture (Graham and McCrae, 1996; Xu et al., 2001; Bauer et al., 2004; Lash et al., 2005) and uterine leukocyte subset (Shi et al., 1995) supernatants by gelatin gel zymography. In situ zymography has also been used to demonstrate gelatinase activity in EVT in placental samples (Isakada et al., 2003).
Uterine natural killer (uNK) cells, characterized by a CD16dimCD56bright phenotype (Saito et al., 1993) account for only 1% of peripheral NK cells but make up ∼70–80% of leukocytes in early pregnancy decidua (Bulmer et al., 1991). uNK cells secrete various cytokines, including tumor necrosis factor (TNF)-α, interferon (IFN)-γ, transforming growth factor (TGF)-β1 and interleukin (IL)-1β (Saito et al., 1993; Jokhi et al., 1997). uNK cells may therefore be a major maternal coordinator of immunotolerance to the semiallogeneic fetus (Bulmer and Lash, 2005), as well as playing a role in regulation of trophoblast invasion (Bulmer and Lash, 2005) and spiral artery remodeling (Lash et al., 2006). The protease activity of uNK cells has not been well characterized, although Shi et al. (1995) demonstrated gelatinase activity in uterine CD56+ cell supernatants.
We hypothesized that uNK cells are a major source of proteases in uterine decidua. The aim of the current study was to examine both the expression and the proteolytic activity of MMP-2 and MMP-9 and their inhibitors in the placental bed of normal early pregnancy and to compare activity in uNK cells and unseparated decidual cells at different gestational ages.
Materials and Methods
Samples
Placental bed biopsies were obtained from healthy women with apparently normal pregnancies undergoing elective termination of pregnancy at the Royal Victoria Infirmary, Newcastle upon Tyne at 8–10 (n = 6), 12–14 (n = 6) and 16–20 weeks (n = 6) gestation. Placental bed biopsies were taken as previously described using biopsy forceps (Richard Wolf Endoscopes, Wimbledon, UK) by one operator (S.C.R.) (Robson et al., 2002). Biopsies were either snap-frozen in liquid nitrogen-cooled isopentane and stored at −80°C for cryostat sections, or fixed in 10% buffered formalin for 24 h and routinely embedded in paraffin wax. Only biopsies containing both decidua and myometrium, interstitial EVT and at least one cross-section of a spiral artery were included in the study. Decidua parietalis was obtained from a separate group of women undergoing termination of apparently normal pregnancies at 8–10 (n = 15) and 12–14 weeks (n = 15) of gestation as previously described (Vassiliadou and Bulmer, 1998; Lash et al., 2006). All women gave informed written consent before termination of pregnancy. The study was approved by the Joint Ethics Committee of Newcastle upon Tyne Health Authority and the University of Newcastle.
Gelatin in situ zymography
Gelatinase activity was localized by gelatin in situ zymography using un-quenched gelatin-Oregon green (Molecular Probes, Leiden, The Netherlands) (Lindsey et al., 2001; Frederiks and Mook, 2004). Freshly cut cryosections of placental bed (10 µm) were air dried at room temperature, washed in phosphate-buffered saline (PBS) and incubated at 37°C for 1 h with PBS including 5 µM MMP-2 inhibitor (EMD Biosciences, Darmstadt, Germany), PBS including 20 nM MMP-9 inhibitor (EMD Biosciences), PBS alone or 50 nM EDTA (to inhibit all zinc-requiring enzymes). Gelatin-Oregon green was diluted to 0.1 mg/ml in developing buffer [50 mM Tris base, 40 mM HCl, 200 mM NaCl, 5 mM CaCl2-2H2O, 50 mM phenylmethane sulfonyl fluoride and 0.2% (w/v) Brij-35]. Each section was overlaid with 150 µl buffer and incubated at 37°C for 3 h in a humid container. After washing to remove unbound gelatin, sections were counterstained with 100 ng/ml 4′,6-diamidino-2-phenylindole for 30 min to identify nuclei. Fluorescence was detected using a fluorescent microscope (Leica Microsystems, Wetzler, Germany) with excitation at 460–500 nm for 1/16 s and emission at 512–542 nm. Gelatinase activity was shown as a loss of fluorescence (Frederiks and Mook, 2004). In situ zymography was performed on four different placental bed biopsies (9–19 weeks gestation).
Positive selection and culture of CD56+ cells
Total decidual and uNK cell cultures were prepared as previously described (Vassiliadou and Bulmer, 1998; Lash et al., 2006). The purity of CD56+ cells selected by this method has been previously shown to be >95% (Jones et al., 1997). At the end of the culture period, viability of CD56+ enriched cell cultures was assessed by trypan blue exclusion and was consistently >95%. CD56+ uNK cells account for ∼30% of the total decidual cell fraction (data not shown) (Ritson and Bulmer, 1987). In addition, since decidua parietalis was used there were no EVT cells in the total decidual fraction (data not shown). Briefly, decidua was finely minced and digested for 2 × 40 min at room temperature using 0.1% collagenase type II and 0.01% DNase type IV (Sigma Chemical Co., Poole, UK) in incomplete medium (RPMI-1640; Sigma) supplemented with 2 mM glutamine, 50 U/ml penicillin and 50 µg/ml streptomycin (Gibco, Paisley, UK). The supernatant was sieved through a 40 µm cell strainer to remove undigested tissue, resuspended in 500 µl complete medium [RPMI1640 with 10% fetal calf serum (Sigma)] and incubated overnight in 5% CO2 at 37°C to remove adherent endometrial stromal and epithelial cells. After resuspension in incomplete medium, decidual cells were incubated with anti-CD56 antibody (NKH-1-RD1, Coulter Immunology, Luton, Bedfordshire, UK) for 30 min at 4°C. The cells were then washed twice in cold PBS, pH 7.3, resuspended in ice-cold separation buffer (PBS with 5 mM EDTA and 0.5% BSA, pH 7.4) and 2 µl Macs indirect IgG microbeads (Miltenyi Biotec Ltd, Camberley, UK) per 1 × 106 total cells and incubated at 4°C for 15min. The cells were then adjusted to 1 ml total volume with separation buffer and positively selected using a Midi Macs Separation Column (Miltenyi Biotec Ltd). Total decidual cells and the CD56+ fraction (uNK cells) were adjusted to 1 × 106/ml in complete medium, cultured for 24 h in a standard 37°C 5% CO2 in air incubator and supernatants collected and stored at −80°C until required for analysis. In addition, freshly isolated (t = 0) CD56+ cells were collected and stored at −80°C until required for RT–PCR analysis.
Gelatin gel zymography
To determine the activity of secreted MMP-2 and MMP-9, zymographic analysis was performed as previously described (Graham and McCrae, 1996; Lash et al., 2005). Preliminary experiments indicated that 10 µg total protein (Bio-Rad Protein Assay, Bio-Rad Laboratories, Hercules, CA, USA) did not give saturation of gelatin cleavage and therefore 10 µg total protein was mixed with zymogram sample buffer (Bio-Rad) and resolved in a 12% SDS–PAGE containing 2 mg/ml gelatin. After electrophoresis, the gels were washed in 2.5% Triton X-100 to remove SDS, rinsed with water and incubated overnight at 37°C in a solution of 50 mM Tris and 5 mM CaCl2 (pH 7.0) to allow substrate digestion. After incubation, gels were stained with 0.4% Coomassie Brilliant Blue R250 in 45% methanol/10% acetic acid, destained with 30% methanol/10% acetic acid, preserved and dried. The dried gels were scanned, digitized and densitometry was performed (Unscan-It, Silk Scientific Co., Orem, UT, USA). Comparison of total decidual and uNK cells was performed with Wilcoxon t-test (Statview 5.0, Abacus Concepts Inc., Berkley, CA, USA). Differences were considered statistically significant at P < 0.05.
Reverse transcription polymerase chain reaction
RNA was extracted from t = 0 CD56+ cells (8–10 and 12–14 weeks gestation, n = 5 each group) using the RNaqueous RNA extraction kit (Ambion, Austin, TX, USA) according to the manufacturer’s instructions. 500 ng total RNA was reverse transcribed using Superscript III according to the manufacturers instruction’s (Invitrogen, Paisley, UK). RT–PCR was performed for MMP-2, MMP-9 and GAPDH using either commercially available primer pairs (MMP-2 and MMP-9, R&D Systems, Abingdon, UK) or GAPDH F 5′-CGACCACTTTGTCAAGCTCA and GAPDH R 5′-AGGGGTCTACATGGCAACT. Samples were amplified for 40 cycles, with each cycle consisting of 45 s at 95°C, 45 s at 55°C (MMP-2 and MMP-9) or 45 s at 62°C (GAPDH) and 45 s at 72°C. The positive controls supplied with the primer pairs are artificial plasmids, which include several primer sequences, and therefore give slightly shorter RT–PCR product lengths than expected for endogenous MMP-2 or MMP-9. RT–PCR products were separated on a 1% agarose gel and stained with ethidium bromide. Data were not quantified.
Western blot analysis for TIMPs
Thirty micrograms of total protein (in cell supernatants) were mixed with reducing dye, denatured and resolved in a 15% SDS–PAGE. Proteins were transblotted to Hybond-P membrane (Amersham Biosciences, Buckinghamshire, UK) and blocked overnight in PBS containing 0.05% Tween 20 (PBS-T) and 5% skimmed milk at 4°C. Membranes were probed with mouse anti-human TIMP-1 (Chemicon International Inc., Temecula, CA, USA; 1:200), mouse anti-human TIMP-2 antibody (Chemicon; 1:100) or mouse anti-human TIMP-3 (Chemicon; 1:200) in PBS-T containing 3% skimmed milk for 2 h at room temperature. Membranes were incubated with goat anti-mouse horseradish peroxidase-conjugated secondary antibody (Bio-rad; 1:1000) and developed using an ECL-plus kit (Amersham Biosciences) and X-OMAT film (Eastman Kodak Co., Rochester, NY, USA). Membranes were then washed thoroughly in PBS-T and re-probed with mouse anti-β-actin (Sigma; 1:1000) as an internal control. Films were scanned, digitized and densitometry was performed (Unscan-It, Silk Scientific Co.). All data are shown as density ratio per respective β-actin density. Statistical significances were determined with Mann–Whitney U-test, and differences were considered statistically significant at P < 0.05.
Immunohistochemistry
Immunohistochemistry was performed on formalin fixed placental bed biopsies in three gestational age groups (8–10 weeks, n = 5; 12–14 weeks, n = 4; 16–20 weeks, n = 5) using an avidin–biotin peroxidase method (Vectastatin Elite, Vector Laboratories, Peterborough, UK) as previously described (Lash et al., 2006). Primary antibody specificity, dilutions, pretreatments and incubation conditions are shown in Table I. Some tissue sections were double labeled with MMP-9, TIMP-1 or TIMP-3 and anti-CD56 allowing co-localization of uNK cells. If double labeling was not possible because of incompatibility of the required pretreatments, a 3 µm serial section was immunostained with anti-CD56. The reaction was developed with diaminobenzidine (Sigma) for single labeling or the first step of double labeling, and with VIP (Vector VIP kit, Vector Laboratories) for the second step of double labeling. Washes between each step were performed in 0.15 M Tris buffered 0.05 M saline buffer (pH 7.6). Single-labeled sections were lightly counterstained in Mayer’s hematoxylin (BDH, Poole, UK) and mounted with DPX synthetic resin (Raymond Lamb, London, UK). Double-labeled sections were not counterstained.
Primary antibodies used for immunohistochemistry.
| Antibody specificity | Clone | Pretreatment | Dilution | Incubation time |
|---|---|---|---|---|
| MMP-2a | CA-4001 | none | 1/100 | 60 min |
| MMP-9a | GE-213 | Trypsin, 10 min, 37°C | 1/50 | Overnight, 4°C |
| TIMP-1a | 102D1 | none | 1/20 | Overnight, 4°C |
| TIMP-2a | 2TMP05 | none | 1/200 | 60 min |
| TIMP-3a | 136-13H4 | Citrate buffer, pH 6.0; pressure cook 1 min | 1/100 | 30 min |
| CD56b | CD564 | Citrate buffer, pH 6.0; pressure cook 1 min | 1/50 | 30 min |
| Pan-Cytokeratinb | LP34+5D3 | Trypsin, 10 min, 37°C | LP34 1/80+5D3 1/20 | 30 min |
| Antibody specificity | Clone | Pretreatment | Dilution | Incubation time |
|---|---|---|---|---|
| MMP-2a | CA-4001 | none | 1/100 | 60 min |
| MMP-9a | GE-213 | Trypsin, 10 min, 37°C | 1/50 | Overnight, 4°C |
| TIMP-1a | 102D1 | none | 1/20 | Overnight, 4°C |
| TIMP-2a | 2TMP05 | none | 1/200 | 60 min |
| TIMP-3a | 136-13H4 | Citrate buffer, pH 6.0; pressure cook 1 min | 1/100 | 30 min |
| CD56b | CD564 | Citrate buffer, pH 6.0; pressure cook 1 min | 1/50 | 30 min |
| Pan-Cytokeratinb | LP34+5D3 | Trypsin, 10 min, 37°C | LP34 1/80+5D3 1/20 | 30 min |
Distributor: aChemicon International Inc. bNovocastra Laboratories, Newcastle upon Tyne, UK. All incubations were at room temperature unless otherwise stated.
Primary antibodies used for immunohistochemistry.
| Antibody specificity | Clone | Pretreatment | Dilution | Incubation time |
|---|---|---|---|---|
| MMP-2a | CA-4001 | none | 1/100 | 60 min |
| MMP-9a | GE-213 | Trypsin, 10 min, 37°C | 1/50 | Overnight, 4°C |
| TIMP-1a | 102D1 | none | 1/20 | Overnight, 4°C |
| TIMP-2a | 2TMP05 | none | 1/200 | 60 min |
| TIMP-3a | 136-13H4 | Citrate buffer, pH 6.0; pressure cook 1 min | 1/100 | 30 min |
| CD56b | CD564 | Citrate buffer, pH 6.0; pressure cook 1 min | 1/50 | 30 min |
| Pan-Cytokeratinb | LP34+5D3 | Trypsin, 10 min, 37°C | LP34 1/80+5D3 1/20 | 30 min |
| Antibody specificity | Clone | Pretreatment | Dilution | Incubation time |
|---|---|---|---|---|
| MMP-2a | CA-4001 | none | 1/100 | 60 min |
| MMP-9a | GE-213 | Trypsin, 10 min, 37°C | 1/50 | Overnight, 4°C |
| TIMP-1a | 102D1 | none | 1/20 | Overnight, 4°C |
| TIMP-2a | 2TMP05 | none | 1/200 | 60 min |
| TIMP-3a | 136-13H4 | Citrate buffer, pH 6.0; pressure cook 1 min | 1/100 | 30 min |
| CD56b | CD564 | Citrate buffer, pH 6.0; pressure cook 1 min | 1/50 | 30 min |
| Pan-Cytokeratinb | LP34+5D3 | Trypsin, 10 min, 37°C | LP34 1/80+5D3 1/20 | 30 min |
Distributor: aChemicon International Inc. bNovocastra Laboratories, Newcastle upon Tyne, UK. All incubations were at room temperature unless otherwise stated.
Results
Gelatinase activity in the placental bed
Gelatinase activity was observed in all areas of the placental bed, as determined by loss of fluorescence, due to Oregon-green gelatin digestion, compared with the negative control (Fig. 1A and B). Gelatinase activity was stronger in decidua than myometrium (Fig. 1B– D). Pre-incubation of sections with either MMP-2 or MMP-9 inhibitor decreased gelatinase activity (Fig. 1C and D). Inhibition was stronger in the presence of MMP-9 inhibitor (Fig. 1D) compared with MMP-2 inhibitor (Fig. 1C) in both decidua and myometrium, suggesting that MMP-9 is the predominant gelatinase in the placental bed.
Gelatinase activity in 19 weeks gestation placental bed biopsy detected by gelatin in situ zymography.
(A) Negative control: all zinc-requiring enzyme activity is inhibited. (B) Incubated with zinc-containing buffer: gelatinase activity is represented by loss of fluorescence. (C) Incubated with zinc-containing buffer after pretreatment with MMP-2 inhibitor. (D) Incubated with zinc-containing buffer after pretreatment with MMP-9 inhibitor. (E) DAPI stain showing cell nuclei. (F) Serial section stained with Hematoxylin and Eosin. Magnification ×100.
MMP-9 and MMP-2 activity in uNK cell and total decidual cell culture supernatants
MMP-2 and MMP-9 were observed in all cell culture supernatants from total decidual cell and uNK cell isolates, as determined by gelatin gel zymography. The predominant form of MMP-9 observed was the latent form and this was assessed by densitometry. However, the predominant MMP-2 form which was assessed by densitometry was the active form. Levels of MMP-9 were significantly lower in uNK cell supernatants than total decidual cell isolate supernatants, at both 8–10 (n = 10, P = 0.0003) and 12–14 weeks (n = 10, P = 0.0012) gestation (Fig. 2A and B). In contrast, there was no difference in MMP-2 levels between the uNK cell and total decidual cell isolate supernatants from either gestational age (Fig. 2A and B), suggesting that uNK cells are a major source of decidual MMP-2. There was no difference in levels of secreted MMP-2 or MMP-9 by either cell isolate at the different gestational ages. However, only active MMP-2 (62 kDa band) was seen in samples from 12–14 weeks gestational age.
Gelatin gel zymography for MMP-2 and MMP-9.
(A) Representative gelatin gel zymogram of total and uNK cell supernatants from both 8–10 and 12–14 weeks gestation. The latent form of MMP-9 is 92 kDa, while its active form is 82 kDa; similarly the latent form of MMP-2 is 72 kDa and its active form is 62 kDa. (B) Graphical representation of data, presented as mean average pixel ± SEM. n = 10 for each group. (C) Representative RT–PCR for MMP-2, MMP-9 and GAPDH of freshly isolated uNK cells. The positive controls for both MMP-2 and MMP-9 provided with the primer pairs give a PCR product slightly smaller than for cDNA (MMP-2, cDNA—449 bp, positive control—380 bp; MMP-9, cDNA—564 bp, positive control—380 bp). The positive controls supplied with the primer pairs are artificial plasmids, which include several primer sequences.
RT–PCR of freshly isolated CD56+ uNK cells confirmed expression of both MMP-2 and MMP-9 by uNK cells isolated from decidua in both gestational age groups studied (8–10 and 12–14 weeks gestation, Fig. 2C).
TIMP secretion by uNK cell and total decidual cell culture supernatants
TIMP-1, -2 and -3 protein secretion was observed in all samples as determined by western blot analysis (n = 10; 8–10 and 12–14 weeks gestation). There was no difference in the level of TIMP-1 or TIMP-3 protein secreted by uNK cells compared with total decidual cell isolates from both 8–10 and 12–14 weeks of gestation (Fig. 3). However, uNK cells from 8–10 weeks gestation secreted higher levels of TIMP-2 protein compared with total decidual isolates (P = 0.041, Fig. 3). There was no difference in secretion of TIMP-2 protein between the two cell isolates from 12–14 weeks gestation.
TIMP-1, TIMP-2, TIMP-3 expression by western blotting.
(Top) Representative western blots. (Bottom) Graphical representation of data, presented as mean density ratio per β-actin density ± SEM. n = 10 for each group.
Immunohistochemistry
Representative photomicrographs of immunolocalization of MMP-2, MMP-9, TIMP-1, TIMP-2, TIMP-3 and CD56 are shown in Fig. 4. MMP-2 was immunolocalized in EVT and uNK cells. MMP-9 was also detected in EVT, but only weakly in uNK cells. TIMP-1 was localized to interstitial EVT. In the later gestational age samples, trophoblast giant cells were also immunostained, but were weaker than mononuclear interstitial EVT. TIMP-2 was also widely immunolocalized throughout placental bed, with strong localization to uNK cells. TIMP-3 immunostaining was weak on interstitial and endovascular EVT and uNK cells. There was no specific staining of uterine myometrium for any of the MMPs and TIMPs studied (data not shown).
MMP and TIMP immunostaining in placental bed.
Magnification ×400. (A–D) Fourteen-week placental bed biopsy. (A) Single immunostaining for MMP-2: arrows (T) show positive EVT. (B) Double immunostaining for MMP-9 (brown) and CD56 (purple): sharp arrows show MMP-9 positive uNK cells and round arrows show MMP-9 negative uNK cell cells. The asterix shows MMP-9 positive endovascular extravillous trophoblast cells. (C and D) Serial sections single labeled for MMP-2 (C) and CD56 (D): arrows indicate examples of MMP-2 positive uNK cells. (E and F) Serial sections of 8-week placental bed biopsy single labeled for TIMP-2 (E) and CD56 (F): arrows indicate TIMP-2 positive uNK cells; most uNK cells are positive at this gestational age. (G and H) Serial sections of 13-week placental bed biopsy single labeled for TIMP-2 (G) and CD56 (H): closed arrows indicate TIMP-2 positive uNK cells and open arrows show TIMP-2 negative uNK cells; a lower proportion of uNK cells are positive at this gestational age. (I) Double immunolabeling of 10-week placental bed biopsy for TIMP-1 (brown) and CD56 (purple): arrows show TIMP-1 positive uNK cells. (J) Double immunolabeling of 18-week placental bed biopsy for TIMP-3 (brown) and CD56 (purple): arrow shows a TIMP-3 positive uNK cell.
Discussion
In the current study, we have demonstrated in situ activity and localization of the MMPs, MMP-2 and MMP-9, and their inhibitors, TIMP-1, TIMP-2 and TIMP-3, in the placental bed during early pregnancy. This is the first comprehensive characterization of the gelatinases and their inhibitors in the placental bed during early pregnancy. MMP-2 and TIMP-2 were localized strongly to uNK cells in the placental bed throughout early pregnancy. Furthermore, uNK cells appeared to be a major in vitro decidual source of MMP-2.
MMP-2 and MMP-9 activity in trophoblast cells and term placenta has been well described (Bischof et al., 2000), and MMP-2 and MMP-9 have also been localized in human placental bed biopsies using immunohistochemistry (Huppertz et al., 1998; Huisman et al., 2004; Seval et al., 2004) and in situ hybridization (Bjorn et al., 1997). In general, MMP-2 and MMP-9 reactivity of interstitial EVT has been reported, but localization to decidual cells has not been well described. Bjorn et al. (1997) reported MMP-2 and MMP-9 activity in homogenates of early pregnancy decidua using gel zymography and also demonstrated co-expression of MMP-2 and membrane type-1 (MT1) MMP by EVT using in situ hybridization (Bjorn et al., 1997). The current study supports the finding of Shi et al. (1995) who demonstrated gelatinase activity in supernatants of decidual cell fractions positively selected by immunomagnetic separation. In concordance with the results of the present study, CD56+ cells were the predominant source of MMP-2 in early pregnancy decidua. Shi et al. (1995) also reported that type IV gelatinolytic activity was greatest in CD56+ cell supernatants compared with other endometrial and decidual leucocyte populations and increased across the menstrual cycle and into early pregnancy. In the current study, we demonstrated MMP-2 and MMP-9 activity in cell culture supernatants of both total decidual and uNK cell isolates and uNK cells appeared to be a major source of active MMP-2 in these cultures. In addition, we demonstrated mRNA for both MMP-2 and MMP-9 in uNK cells isolated from decidual tissue at both 8–10 and 12–14 weeks gestation. Seval et al. (2004) have proposed that MMP-2 is the key regulator of trophoblast invasion in very early pregnancy, before 8 weeks gestation.
Expression of TIMP-1 and TIMP-2 has been reported in EVT in early pregnancy decidua (Huppertz et al., 1998; Seval et al., 2004) but other sources of these proteins within decidua/placental bed were not investigated. mRNA for TIMP-1, TIMP-2 and TIMP-3 has also been demonstrated in cultured cytotrophoblast cells by northern blot analysis (Xu et al., 2001). However, there are no reports demonstrating TIMP-3 immunolocalization in placental bed or decidua in humans. In the current study, we demonstrated weak secretion of TIMP-1 and TIMP-3 by total decidual and uNK cells by western blot analysis, as well as weak immunostaining of placental bed biopsies for these proteins. TIMP-2 was the predominant TIMP expressed in the placental bed as determined by immunohistochemistry and western blot analysis. TIMP-2 not only inhibits MMP-2 activity, but can also activate pro-MMP-2 (Imai et al., 1996) through binding to MT1-MMP (Strongin et al., 1995). However, TIMP-2 is regarded as a better inhibitor of MT1-MMP than other MMP’s (Zhao et al., 2004). More evidence is required to verify that MMP-2 activity in early pregnant decidua is regulated by TIMP-2 expression by uNK cells.
A wide range of uNK cell functions have been demonstrated in vitro including immunoregulation, cytokine production (Saito et al., 1993; Bulmer and Lash, 2005) and NK-type cytotoxic activity (Jones et al., 1997). The in vivo function of uNK cells is unclear, although a role in regulation of EVT invasion has been proposed, via the secretion of several cytokines and growth factors which are known to regulate EVT invasion in vitro (Bulmer and Lash, 2005). Indeed, we and others have demonstrated that uNK cell culture supernatants stimulate trophoblast invasion in vitro (Hanna et al., 2006; Lash et al., 2007). In the current study, we demonstrated reduced gelatinase activity and immunolocalization of MMPs and TIMPs in myometrium compared with decidua. This difference in protease activity between the myometrium and decidua may reflect a mechanism by which EVT invasion is restricted to the decidua and superficial myometrium (Pijnenborg et al., 2006). The area of maximal gelatinase activity is consistent with localization of uNK cells to decidua parietalis and decidua basalis and not to myometrium (Bulmer and Lash, 2005). In the placental bed, uNK cells are associated with invading EVT and secretion of proteases by uNK cells may facilitate movement of EVT through the decidua.
Uterine NK cells have also been proposed to play a role in uterine spiral artery remodeling. Early structural changes in decidual spiral arteries occur at a time when uNK cells are present, before direct cellular interactions with trophoblast within spiral arteries (Bulmer et al., 1991). In addition, it has been reported that up to 80% of dilated decidual spiral arteries are not associated with EVT (Craven et al., 1998; Kam et al., 1999). It is possible that uNK cells play a pivotal role in spiral artery remodeling, possibly through the activation of digestion of ECM of spiral arteries via secretion of MMP-2 and MMP-9 (Moses, 1997). Using an in vitro chorionic plate artery model, we have demonstrated that uNK cell culture supernatants are able to disrupt the vascular smooth muscle layer, as compared with total decidual cell culture supernatants (Naruse et al., 2007). uNK-derived proteases may play a direct role in the process of disruption of the ECM in these vessel walls. It is worth noting that uNK cell numbers are reduced after 20 weeks of pregnancy when spiral artery changes are complete (Bulmer and Lash, 2005).
The secretion of humoral factors by uNK cells appears to be important for the establishment of a successful pregnancy. However, the triggers for uNK cell activation in early pregnancy are not clear. A recent study showed that different subtypes of killer-cell immunoglobulin-like receptors (KIRs) and leukocyte immunoglobulin-like receptor-1 (LIR-1) on the uNK cell surface led to a variation in secretion of cytokines and growth factors, suggesting that regulation of uNK is by HLA-Cw4, HLA-Cw6 and HLA-G (Hanna et al., 2006). This finding suggests an EVT regulation of uNK activity, but does not explain the secretion of cytokines, growth factors and MMPs by uNK cells in the absence of EVT cells. Hormonal regulation of uNK cell activity may exist; they express estrogen (beta), prolactin and glucocorticoid receptors, but not progesterone receptor (Bulmer and Lash, 2005). A direct role for estradiol on uNK cell activity remains controversial, but progesterone-mediated up-regulation of IL-15 production by decidual stromal cells may drive uNK differentiation indirectly (van den Heuvel et al., 2005). Dramatic hormonal changes with the onset of pregnancy and antigen expression from/to EVT may cooperate to regulate uNK cell function in early pregnancy. Comparison of non-pregnant, progesterone-treated endometrium and uterine decidua associated with ectopic tubal pregnancy would help to determine the effect of trophoblast (direct or indirect) versus the effect of hormones on uNK cell MMP expression.
In summary, we have demonstrated that uterine decidua possesses protease activity in situ and in vitro and that uNK cells are a major source of MMP-2 and TIMP-2. These data further support a role of uNK cells in regulating EVT invasion and spiral artery remodeling in establishment of successful pregnancy.
Funding
This project was supported by funding from BBSRC (S19967).
Acknowledgements
The authors are grateful to the staff at the Royal Victoria Infirmary, Newcastle upon Tyne, for their assistance with sample collection.
