A Role for Tissue Transglutaminase in Stabilization of ... - BioOne

4 downloads 0 Views 1MB Size Report
Jul 11, 2007 - Nicola J. Robinson, Philip N. Baker, Carolyn J.P. Jones, and John D. Aplin1. Maternal and Fetal Health Research Centre, Division of Human ...
BIOLOGY OF REPRODUCTION 77, 648–657 (2007) Published online before print 11 July 2007. DOI 10.1095/biolreprod.107.061747

A Role for Tissue Transglutaminase in Stabilization of Membrane-Cytoskeletal Particles Shed from the Human Placenta Nicola J. Robinson, Philip N. Baker, Carolyn J.P. Jones, and John D. Aplin1 Maternal and Fetal Health Research Centre, Division of Human Development, University of Manchester, St. Mary’s Hospital, Manchester M13 0JH, United Kingdom apoptotic cells by formation of intracellular cross-linked protein polymers, preventing leakage of intracellular components into the surrounding tissues. This protein scaffold may stabilize dying cells, ensuring their clearance by phagocytosis and thereby preventing inflammatory reactions [6]. TGM2 has been identified as the major autoantigen in celiac disease [7], which several authors suggest is associated with adverse pregnancy outcome, including recurrent miscarriage, low birth weight babies, stillbirths, and intrauterine growth restriction in untreated women [8–12]. The reasons for the poor outcomes are unknown. TGM2 is widely expressed in human placenta, with strong activity at the syncytial microvillous membrane [13, 14], which is directly exposed to maternal blood and autoantibodies. During pregnancy, particulate material continuously sheds from this surface into maternal circulation [15, 16], where it is thought to interact with, and skew, responses in maternal immune cells. Identification of the principal target proteins of TGM2 at this interface may help elucidate its normal function as well as the pathogenic mechanism of altered pregnancy outcome in celiac disease. We have used a proteomic strategy to identify TGM2 target proteins in the placental microvillous membrane (MVM). To test the hypothesis that material shed from terminally differentiated syncytiotrophoblasts may be cross-linked by TGM2, shed microparticulate material was isolated from primary cytotrophoblast cells in the presence and absence of TGM2 inhibitors. Trophoblast cell fusion and differentiation was also assessed following TGM2 inhibition. The results suggest a role for TGM2 in the organization and turnover of the MVM and its associated cytoskeleton.

ABSTRACT Tissue transglutaminase (TGM2; also known as TG2 or tTG) localizes to the syncytial microvillous membrane (MVM) of the human placenta, the primary interface between maternal and fetal tissue. To identify TGM2 substrates in the MVM, membrane vesicles were prepared and labeled with biotinylated acyl donor or acceptor probes. Biotinylated species were selected on an avidin affinity matrix and identified by mass spectrometry of tryptic peptides. The most abundant were cytoskeletal (actin, tubulin, and cytokeratin) and membrane-associated (annexins, integrins, and placental alkaline phosphatase) proteins. During pregnancy, apoptotic particulate material, the end product of the trophoblast life cycle, is shed from the MVM into maternal circulation. Shed material was isolated from primary trophoblast cultures in which syncytial-like masses develop by fusion. A substantial fraction of actin in the particles was in the form of covalent polymeric aggregates, in contrast to cellular actin, which dissociated completely into monomer in SDS-PAGE. When cells were cultured in the presence of transglutaminase inhibitors, actin in the shed particles remained exclusively in monomeric form, and a reduction in trophoblast intercellular fusion and differentiation was observed. These findings suggest that transglutaminase-mediated cross-linking stabilizes the particulate material shed from the placenta. actin, cytoskeleton, immunology, placenta, plasma membrane, p r e g n a n c y, s h e d d i n g , s y n c y t i o t r o p h o b l a s t , t i s s u e transglutaminase, trophoblast

INTRODUCTION Tissue transglutaminase (TGM2; also referred to as transglutaminase type 2, TG2, or tTG) belongs to a family of cross-linking enzymes responsible for catalyzing Ca2 þ-dependent acyl-transfer reactions, resulting in the formation of an isopeptide bond [1]. In addition to its transamidating capacity, TGM2 possesses GTPase activity, with the ability to hydrolyze both GTP and ATP, and intrinsic kinase activity [2]. TGM2 is a cytosolic protein that has also been observed in the nucleus and can be externalized to the cell surface or extracellular matrix [3, 4]. Despite ubiquitous expression, its functions remain poorly understood. The observation that TGM2 activity correlates with cellular regression in rat livers after induction of hyperplasia first suggested its involvement in apoptosis [5]. TGM2mediated cross-linking is thought to be important in stabilizing

MATERIALS AND METHODS Tissue Local Ethical Committee Approval, with patient consent, enabled the collection of placental tissue. Term placenta was obtained at cesarean section or vaginal delivery following uncomplicated pregnancy.

MVM Preparation The preparation of MVMs was based on an established method [17] and used either method 1 or method 3. Greater purification was obtained with vesicles made from method 1, as demonstrated by the higher enrichment of alkaline phosphatase activity. The protein recovery (milligram per gram of placenta) was 5-fold higher by method 3 (stir-prep). The final vesicle preparation was spun down from 150 mM NaCl, 4 mM KCl, 2 mM MgCl2, and 2 mM Tris-HCl, pH 7.4, and stored at 808C.

1

Correspondence: John Aplin, Division of Human Development, University of Manchester, St. Mary’s Hospital, Research Floor Room 74, Hathersage Rd., Manchester M13 0JH, U.K. FAX: 44 0 161 276 6134; e-mail: [email protected]

Membrane-Associated TGM2 Activity Assay with AcylAcceptor/Donor Probes

Received: 30 March 2007. First decision: 30 April 2007. Accepted: 1 July 2007. Ó 2007 by the Society for the Study of Reproduction, Inc. ISSN: 0006-3363. http://www.biolreprod.org

Two biotinylated probes were used for labeling TGM2 substrate proteins in MVM vesicles: biotin-cadaverine (BTC), which represents the acyl-acceptor probe, and a biotinylated glutamine-containing hexapeptide (biotinylTVQQEL) [18], which represents the acyl-donor probe in the cross-linking reaction catalyzed by TGM2. Aliquots of frozen MVM isolated from term

648

TGM2 AND SHED PLACENTAL MICROPARTICLES human placental syncytiotrophoblasts were spun down in buffer at 23 000 3 g, 15 min, 48C. One hundred micrograms of MVM protein was resuspended in TGM2 activity buffer containing 5 mM CaCl2, 100 mM Tris-HCl (pH 8.3), 0.15 M NaCl, 5 mM acyl-donor or -acceptor probe, and 10 mM dithiothreitol (DTT) (added just before the incubation) in a total volume of 0.2 ml. This was incubated at 378C for 18 h, and then SDS-PAGE/Western blotting was carried out. N,N 0 -dimethylcasein was labeled with guinea pig TGM2 as a positive control for the assay.

Identification of Biotinylated Peptides The biotinylated TGM2 target polypeptides in MVM were separated from the unbiotinylated peptides by avidin affinity chromatography. MVM protein was biotinylated as described and spun down at 13 000 3 g for 5 min to recover the membrane fraction; the pellet was solubilized in 1.0% SDS/Tris-HCl (100 mM, pH 8.3), with approximately a 1:5 ratio of MVM:SDS. This was boiled for 3 min and diluted 10-fold to 0.1% SDS. Dialysis against 0.1% SDS was carried out to remove free biotin, and the solution was mixed overnight with prewashed avidin beads (Softlink soft release avidin resin; Promega) that had been prepared by 33 washes in Tris-HCl (pH 8.3), with a final wash and a 30-min incubation at room temperature in 0.1% Tris-HCl. Following the overnight incubation, the beads were washed in Tris-HCl (pH 8.3), and the supernatant containing the unbiotinylated peptides was retained. The biotinylated peptides were released from the beads by boiling in 50 ll of SDS sample buffer. The bound and unbound fractions were loaded onto SDS-PAGE gels and visualized by silver staining.

SDS-PAGE/Silver Staining/Protein Blotting Protein concentration was determined with the BCA Protein Assay Kit (Pierce) according to the manufacturer’s instructions. Aliquots of the MVM vesicles obtained were run on 7.5% SDS-PAGE, and filter transfers were analyzed with avidin peroxidase to detect biotinylated proteins. Silver staining of SDS-PAGE gels was performed by briefly washing gels in dH2O and then fixing them in 50% (v/v) methanol and 5% (v/v) acetic acid for 20 min; this was followed by fixation in 50% (v/v) methanol for 10 min. Gels were sensitized in 0.02% (w/v) sodium thiosulfate for 1 min and then washed twice for 1 min each in dH20. Gels were then submerged in 0.1% (w/v) silver nitrate for 20 min at 48C. After two 1-min washes in dH2O, gel staining was developed in 0.04% (v/v) formalin and 2% (w/v) sodium carbonate until bands were visible (5–10 min). The development was stopped with 5% acetic acid, and the gels could be stored in 1% acetic acid at 48C. Western blotting, antibody probing, and enhanced chemiluminescence detections were carried out according to the procedure previously described [14]. Monoclonal mouse anti-b-actin antibody was used at a final concentration of 0.5 lg/ml, and polyclonal goat anti-mouse-HRP (DAKO) was used at 1 lg/ml.

Mass Spectrometry This was carried out in the Biomolecular Analysis Core Facility in the Faculty of Life Sciences at the University of Manchester. Gel bands from the fraction bound to the avidin beads were excised with a disposable glass pipette, reduced with 10 mM DTT in 25 mM NH4HCO3 at 568C for 1 h, and then alkylated with 55 mM iodoacetamide in 25 mM NH4HCO3 for 45 min at room temperature in the dark. After washing in acetonitrile, proteins were digested overnight with ;5 ll of trypsin at 12.5 ng/ll (sequencing grade; Promega) in ;50 ll of 25 mM NH4HCO3. The resultant peptides were extracted from the gel with 25 mM NH4HCO3, with further extraction with 5% formic acid in 50% acetonitrile. Samples were dried to approximately 20 ll, of which 6 ll was analyzed by liquid chromatography-tandem mass spectrometry (LC-MSMS) with a Q-TOF Micromass spectrometer. Data acquired were searched against both SWISSPROT and TrEMBL by ProteinLynxGlobalServer software. Positive identification was defined as three or more distinct peptides from the same polypeptide sequence by tandem MS. Two or one peptide was classed as a tentative identification.

Immunoprecipitation Biotinylated MVM protein was spun down to recover the membrane pellet and solubilized in 1% (v/v) Brij35 (Surfact-Amps 35; Pierce) in Tris-buffered saline containing 1% (v/v) commercial protease inhibitors (containing pepstatin A, bestatin, E-64, leupepin, and aprotinin) and mixed at 48C for 4 h. The extract was centrifuged at 13 000 3 g at 48C for 15 min to separate Brij35soluble and -insoluble proteins. The pellet was taken up into 40 ll of SDS sample buffer and stored at 208C. The supernatant, containing detergentsoluble proteins, was retained at this stage for use in SDS-PAGE/Western

649

blotting. For immunoprecipitation experiments, rabbit polyclonal anti-human annexin A5 antibody (final concentration, 4 lg/ml; Abcam) was added to the supernatant and mixed for 1 h at room temperature. Protein A beads were swollen in buffer A (0.02 M NaH2PO4 and 0.15 M NaCl, pH 8.0) for 30 min and then washed in buffer A (23), with a final wash in Brij35/buffer A. The MVM plus antibody solution was then added to the Protein A beads and mixed overnight at room temperature. The beads were spun down, and the first supernatant or ‘‘flow-through’’ was retained. The beads were washed in Brij35/ buffer A (33) with a final wash in buffer A without Brij35. The antibody plus antigen complex was released from the beads by boiling in SDS sample buffer. Mouse monoclonal anti-human annexin A5 antibody (final concentration, 0.4 lg/ml; Abcam) was used for Western blotting.

Isolation of Primary Trophoblast Cells from Term Placenta The method of isolation employed was based on methods used by Kliman et al. [19] with later modifications [20]. Mononucleate cytotrophoblast cells were cultured in Dulbecco modified Eagle medium/F12/10% fetal calf serum aggregate and fused for 2–4 days in culture [21]. The culture medium contained 1.05 mM calcium. Cytotrophoblasts from term placenta were plated onto glass coverslips precoated with 10% (v/v) collagen/0.1 M acetic acid at a density of 1 3 106 per coverslip. Cells were fixed with ice-cold methanol at 66 h. Anticytokeratin 7 antibody was used as a positive control for trophoblast cells, and anti-vimentin antibody was used to detect fibroblast or macrophage contamination.

Immunocytochemistry Antibody staining and imaging were carried out as previously described [14]. Trophoblast cells were incubated with primary antibodies for 1 h at room temperature. Mouse monoclonal anti-TGM2 antibody (CUB7402; Neomarkers) was used at a final concentration of 1 lg/ml, mouse monoclonal anti-annexin A5 (Abcam) was used at a final concentration of 4 lg/ml, and goat polyclonal anti-annexin A2 (Abcam) was used at a final concentration of 5 lg/ml. Cells were incubated with fluorescein-conjugated secondary antibodies for 1 h at room temperature, either polyclonal goat anti-mouse fluorescein isothiocyanate (FITC; DAKO) or polyclonal rabbit anti-goat FITC, both at a final concentration of 40 lg/ml. Nuclei were counterstained with propidium iodide (PI).

In Situ TGM2 Activity Assay An assay was developed for measuring in situ tTG cross-linking activity in the trophoblast cell cultures with a biotinylated probe. By this method, a biotinylated primary amine such as BTC acts as the acyl-acceptor in the transamidating reaction catalyzed by tTG and becomes incorporated into endogenous protein substrates of tTG [22]. Cells were preincubated with 1 mM BTC for 1 h at 378C prior to fixation, and fluorescently labeled streptavidin (streptavidin-FITC) was used to detect the biotinylated product.

Exposure of Trophoblast Cultures to TGM2 Inhibitors The competitive and irreversible TGM2 inhibitor cystamine [23, 24], which is thought to undergo disulfide exchange with the active-site cysteinyl residue, and the primary amine competitor monodansylcadaverine (MDC) [25], which has been widely used to inhibit TGM2 activity in vivo, were introduced into the cytotrophoblast culture system. Following the determination of the optimum nontoxic inhibitor doses, cystamine concentrations of 1.5 mM and MDC of 150 lM were used in the cultures. As TGM2 is known to be important for cellmatrix adhesion [26, 27], the inhibitors were added to the cultures after 18 h to allow sufficient time for stable cell attachment and the formation of multicellular aggregates.

Assessment of Trophoblast Differentiation Trophoblast differentiation was assessed by measuring the extent of multinucleated trophoblast formation at the end of the culture period [28]. Cells were fixed and permeabilized with methanol as described. Cell-cell borders and nuclei were detected by staining cultures simultaneously with antibody to Ecadherin, which detects intertrophoblastic contact surfaces, and with PI. Total numbers of nuclei were counted from three random fields per culture flask. At least 50 nuclei were counted per field. Nuclei present in a single cell (cells were identified by E-cadherin staining at cell-cell borders) were also counted. Nuclei present as two or more in a single cell were expressed as a percentage of the total nuclei and were defined as multinucleated.

650

ROBINSON ET AL.

FIG. 1. Identification and fractionation of acyl-donor and acceptor TGM2 protein substrates in the MVM. TGM2-mediated labeling of MVM vesicles with (A) the acylacceptor probe BTC or (B) the acyl-donor probe biotinyl-TVQQEL. MVM protein (M) from term placenta was reacted with probe in the presence of Ca2þ. Aliquots were fractionated on avidin-conjugated beads. S shows nonretained material, while E1 and E2 are proteins eluted from the beads. Replicate gels were stained with silver (top), or blots were prepared and probed with avidin peroxidase to detect TGM2 substrate proteins (bottom). Bands in a wide range of Mr values are present in the unfractionated MVM and in eluates from the beads, but no bands are present in the supernatant fractions, indicating efficient selection of biotinylated proteins. Note that the two probes yield different protein profiles. *, þ, # ¼ bands in common from eluted fraction between silver-stained gel and immunoblot. Positions of molecular weight standards (kiloDaltons) are indicated. C) TGM2 protein localization in 66-h trophoblast cultures. Primary term trophoblast cells were fixed in methanol after 66 h of culture and (left) stained with mouse monoclonal antiTGM2 antibody and an FITC-conjugated secondary antibody (green). TGM2 is observed at lateral cell boundaries, especially in residual mononucleate cells, in association with stress fibers, and more weakly in punctate deposits. D) TGM2 transamidating activity in 66-h trophoblast cultures. TGM2 cross-linking activity was visualized with a BTC incorporation assay with streptavidinFITC to detect biotinylated products (right, green). Labeling is partially seen in punctate deposits, with activity not obviously associated with lateral cell boundaries or major cytoskeletal elements. Nuclei were counterstained with propidium iodide (PI) (red). Bar ¼ 20 lm.

hCG Assay

Statistical Analysis

Levels of immunoreactive hCG in trophoblast culture media were determined with a solid-phase enzyme-linked sandwich immunosorbent assay (DRG Diagnostics). Medium was collected from 18–66 h of culture. The concentration of p-nitrophenol product was assessed by measuring the absorbance at 405 nm. All measurements were carried out in duplicate.

All experiments were performed three times, and each experiment used cells obtained from different placentas. Cells were not pooled. Within an experiment, means were obtained from duplicate determinations. Data were compared by the nonparametric Kruskal-Wallis test for repeated-measures analysis of variance with the Dunn post hoc test by GraphPad Prism software, version 4 (GraphPad Software). Significance was taken as P , 0.05.

Isolation of Syncytiotrophoblast Microparticles from Trophoblast Culture To isolate shed syncytiotrophoblast microparticles, supernatants from 18 to 66 h in culture from three or more 25-cm3 flasks of trophoblast cells were collected and subjected to a three-step centrifugation procedure at 48C. The supernatant was spun at 1000 3 g for 10 min, at 10 000 3 g for 10 min to remove cellular debris, and finally at 70 000 3 g for 90 min to pellet the microparticulate material [16]. The final pellet was collected and resuspended into 20 ll of SDS sample buffer and stored at 208C until use.

Electron Microscopy Sample Preparation For electron microscopy preparation, specimens were resuspended in human serum and embedded as previously described [29].

RESULTS Proteomic Analysis of Target Proteins for TGM2-Mediated Cross-Linking in the Placental MVM TGM2 transamidating activity was detected in isolated MVM vesicles with two different biotinylated affinity probes in the presence of 5 mM calcium. The biotinylated primary amine BTC acted as an acyl-acceptor (Lys-donor) [30], and the hexapeptide biotinyl-TVQQEL [18] acted as an acyl-donor (Glu-donor) probe. Both were incorporated into numerous endogenous protein substrates of TGM2 in the MVM, as shown by separating MVM reaction mixtures on SDS-PAGE

651

TGM2 AND SHED PLACENTAL MICROPARTICLES TABLE 1. Functional classification of identified TGM2 substrates. Functional group Cytoskeletal network

Membrane-associated

Ca2þ and phospholipid-bindingproteins

Transport processes

Protein folding Coagulation Miscellaneous

Proteins Actin; beta, gamma 1 and alpha 2 Myosin heavy chain 9 Myosin light chain 6 Cofilin 1 Tubulin Heat shock protein 27-kDa protein 1 Transferrin Receptor Transferrin Alkaline phosphatase, placental Chloride intracellular channel protein 4 Integrin alpha-V Integrin alpha-IIb Annexin A1 Annexin A2 Annexin A4 Annexin A5 Annexin A6 Clathrin heavy chain Solute carrier family 2 (facilitated glucose transporter) ATPase, Naþ/Kþ transporting, alpha 1 polypeptide Protein disulphide isomerase A3 Heat shock protein HSP90-alpha Heat shock protein HSP90-beta Fibrinogen alpha chain Fibrinogen beta chain Fibrinogen gamma chain Tissue transglutaminase Calpain 6 S100 calcium-binding protein P S100 calcium-binding protein A8 S100 calcium-binding protein A11 Guanine nucleotide-binding protein (G protein) beta polypeptide 1 Chorionic somatomammotropin hormone Ras-related protein Rab-1A Pyruvate kinase, isoforms M1/M2 Monoamine oxidase type A

and probing blots with avidin peroxidase (Fig. 1, A and B). Substrates ranged from ;15 to 230 kDa, with high-molecularmass cross-linked products sometimes observed. No bands were present when nonbiotinylated MVM was run on SDSPAGE and probed with avidin peroxidase (data not shown). Biotinylated target proteins in the MVM were isolated with an avidin-affinity column. The presence of 0.1% SDS in the washing buffer ensured that only directly biotinylated targets would be recovered. Multiple proteins were present in fractions eluted from the beads. No biotinylated products remained in the supernatants when either the acyl-acceptor or acyl-donor probe was used, indicating that efficient separation had occurred in both cases. Fractions eluted from beads were run on SDS-PAGE. Several regions of each gel were excised, alkylated, and digested with trypsin. Samples were then directly analyzed by LC-MSMS. Supplemental Tables 1 and 2 (available online at www.biolreprod.org) show the range of acyl-donor and acyl-acceptor TGM2 substrate proteins identified, respectively, in descending order of apparent abundance. This proteomic strategy led to the identification of .30 TGM2 protein substrates in the MVM. Table 1 is a functional classification of the TGM2 protein substrates identified. Some target proteins were identified with both probes, indicating that they are able to function as both acyl-acceptor and -donor in the transamidation reaction and therefore form homo-oligomers. The two main target types are cytoskeletal and membrane-associated proteins, suggesting a role for TGM2

Acyl donor/acceptor Donor and Donor Donor Donor Acceptor Acceptor Donor and Donor and Donor and Donor Acceptor Acceptor Donor and Donor and Donor and Donor Donor and Donor Donor and

acceptor

acceptor acceptor acceptor

acceptor acceptor acceptor acceptor acceptor

Acceptor Donor and acceptor Acceptor Acceptor Acceptor Acceptor Acceptor Acceptor Acceptor Acceptor Acceptor Acceptor Acceptor Acceptor Acceptor Acceptor Acceptor

in the organization and turnover of trophoblast plasma membranes and associated cytoskeleton. TGM2 Distribution and Activity in Cultured Cytotrophoblasts Having established the presence of TGM2 [14] and identified target proteins in an in vitro assay with MVM isolated from placental tissue, it was important both to characterize the enzyme activity and confirm selected target proteins in living cells. Primary cytotrophoblasts isolated from term placenta were used for studying in vitro TGM2 expression and transamidation activity. Mononucleate cells aggregate and fuse over 2–4 days in culture [21]. TGM2 localization in the 66-h cultures was widespread and strong. It was prominent at the lateral borders of the cells, both at cell contacts and at noncontact surfaces, in the cytoplasm and in association with stress fibers and peripheral actin bundles (Fig. 1C). To detect acyl-donor (Gln-donor) TGM2 substrates, the acyl-acceptor probe BTC was added to the cultures and visualized with FITC-labeled streptavidin. TGM2 0 s transamidating activity appeared much more variable between different cells, perhaps consistent with differentiation dependence. It was most prominently seen in a punctate cytoplasmic compartment (Fig. 1D). Generally, the activity of TGM2 was far more restricted than the enzyme’s distribution. Biotinylation

652 FIG. 2. Annexins are TGM2 substrates in MVM and cultured trophoblast. A) MVM vesicles were reacted with BTC, detergent solubilized, and fractionated on avidin beads. The top panel shows annexin A5 immunoreactivity (33 kDa) detected with a mouse monoclonal anti-human annexin A5 antibody in the unfractionated MVM (M), the pellet (P) and the detergent solubilizate (S), the flow-through (FT) after solubilization of the MVM vesicles with nonionic detergent, and the immunoprecipitate (IP) with a rabbit polyclonal anti-human annexin A5 antibody. The lower panel shows an immunoblot of the MVM (M), flow-through (FT), and immunoprecipitate (IP) probed with avidin peroxidase, indicating that immunoprecipitated annexin A5 is biotinylated and also that a significant fraction of biotinylated A5 is not retained on the beads. B) Immunolocalization of TGM2 substrates annexin A2 and A5 (green) in term trophoblast at 66 h in culture. Punctate annexin A2 staining is observed (arrowheads) mostly around the edges of the syncytium (arrows). Annexin A5 appears mostly associated with mononucleate unfused cells (arrows), with less staining observed in the large syncytia (arrowheads). C) Serial Z-sections of annexin A2 localization within multinucleated 66-h trophoblast cells, from the basal (top left) to the apical (bottom right) surface. Focal annexin A2 immunoreactivity is detected at the apical surface (arrows), above the plane of the nuclei. Nuclei were counterstained with PI (red). Bar ¼ 25 lm (B) and 20 lm (C).

ROBINSON ET AL.

TGM2 AND SHED PLACENTAL MICROPARTICLES

653

FIG. 3. The effect of TGM2 on trophoblast turnover. A) TGM2 cross-linking activity in primary trophoblast cells fixed at 66 h, measured with the biotincadaverine incorporation assay with streptavidin-FITC to detect biotinylated product (green). TGM2 cross-linking activity apparent in the control culture (upper left) is lost from the cultures treated with either TGM2 inhibitor (upper middle and right). Minimal cross-linking activity is observed in the cystamine (125 mM) treated culture (arrows), with no evidence of TGM2 cross-linking in the MDC (150 lM) treated culture. Bar ¼ 25 lm. B) Cytotrophoblast cells were cultured in the presence or absence of TGM2 inhibitors, and cultures were stained with E-cadherin protein to detect sites of cell-cell contract (green). In untreated trophoblast cells at 66 h, little E-cadherin staining is apparent (lower left), suggesting fusion to form multinucleated syncytia has occurred. More E-cadherin staining is observed between trophoblast cells cultured with either cystamine or MDC (lower middle and right), indicating less fusion with a higher proportion of mononucleate cells remaining. Nuclei were counterstained with propidium iodide (PI) (red) in each case. Bar ¼ 20 lm. C) By E-cadherin immunoperoxidase staining, nuclei were counted, and those present as single mononucleate cells with a complete Ecadherin cell border staining were scored in the presence or absence of either TGM2 inhibitor. The extent of multinucleated cell (i.e., syncytiotrophoblast) formation was calculated as described in Materials and Methods. Use of either inhibitor resulted in a significant reduction in syncytium formation. Asterisks indicate values significantly different from untreated controls (*P , 0.05, ***P , 0.001, Kruskal-Wallis ANOVA followed by the Dunn post hoc test). D) Culture media were collected from trophoblast cells cultured in the presence or absence of TGM2 inhibitors, and the secretion of bioactive hCG was analyzed as described in Materials and Methods. A significant reduction (***P , 0.001, the Dunn post hoc test) in hCG secretion was seen in cultures treated with either cystamine or MDC. After removal of culture supernatants, the protein content of the adherent cells was measured (E). No significant differences in the protein content were determined after treatment with either TGM2 inhibitor.

in living cells could be inhibited by the function-blocking antiTGM2 monoclonal antibody CUB7402 (data not shown). Annexins are one of the most abundant target protein families observed in the MVM (see Supplemental Tables 1 and 2 available online at www.biolreprod.org). Evidence of direct annexin biotinylation in the MVM was obtained with annexin A5 as target. Immunoprecipitation of annexin A5 was carried out from biotinylated MVM, and biotinylation was verified by the binding of avidin to the annexin A5 immunoreactive band (Fig. 2A). A fraction of biotinylated annexin A5 was not retained by the avidin beads and appeared in the flow-through, suggesting that the binding capacity of the beads was limiting. Immunolocalization of selected annexins was carried out with cultured cytotrophoblasts from term placenta. Punctate annexin A2 staining was observed in trophoblast cultures fixed at 66 h

(Fig. 2B), with the prominent localization seen around the edges of the syncytium. Annexin A5 appeared mainly associated with mononucleate, unfused cells, with less seen in the syncytia (Fig. 2B). Serial Z-sections were taken, showing some focal expression of annexins at the apical surface above the plane of the nuclei, possibly associated with membranebound fractions breaking away from the multinucleated syncytium. Annexin A2 immunoreactivity was observed, associated with blebs at the apical surface (Fig. 2C). Effect of Inhibiting TGM2 0 s Transamidating Activity on Cytotrophoblast Differentiation The main functional classifications of TGM2 substrates in the placental MVM were cytoskeletal and membrane-associ-

654

ROBINSON ET AL.

FIG. 4. Microvilli with actin cytoskeleton in shed trophoblast material. TEMs of material shed by trophoblast cultures and isolated by high-speed centrifugation. A) Low-power image of sheets of mononuclear cyto- (C) and syncytial (S) trophoblasts showing numerous microvilli (arrowheads) on the apical surface as well as processes basally. B) At high power, microvilli can be seen to contain longitudinal actin filaments (arrows), while microtubules and cytokeratin filaments are also observed. C) Actin filaments are also apparent in transverse section (arrows) of microvilli and larger cytoplasmic particles. Bar ¼ 5 lm (A), 0.25 lm (B), and 0.2 lm (C). D) TGM2 cross-linking activity in trophoblast cell homogenates and syncytiotrophoblasts shed material in the presence of TGM2 inhibitors. Western blot analysis of the 66-h trophoblast cell homogenates (C) and material shed (SM) from trophoblast cultures, probed with (i) avidin peroxidase or (ii) anti-b-actin antibody and in the presence of TGM2 inhibitors cystamine (Cys) or monodansylcadaverine (MDC). Note that actin immunoreactivity is present in a monomer band and in a high Mr fraction, with loss of the high Mr band from the shed material treated with either TGM2 inhibitor. Molecular weight markers are indicated.

ated proteins, suggesting a role for TGM2 in the organization and turnover of trophoblast plasma membranes and associated cytoskeleton. To explore a functional role for TGM2 in the life cycle of trophoblasts, TGM2 inhibitors were introduced into primary cytotrophoblast monolayer cultures. Following the addition of either the TGM2 competitive substrate inhibitor MDC [25] or the TGM2 active site inhibitor cystamine [23, 24], almost complete inhibition of TGM2 cross-linking activity was observed with the BTC incorporation assay. There was no evidence for transamidating activity in the MDC-treated cultures, and minimal levels of TGM2 cross-linking activity were detected in the cultures containing cystamine (Fig. 3A).

The normal process of trophoblast cell turnover in the placental villus consists of the proliferation and differentiation of underlying mononuclear cytotrophoblast cells, followed by intercellular fusion of differentiated cytotrophoblasts to form the continuous multinucleate syncytiotrophoblast layer, which continues to differentiate; finally, there is an apoptotic release of old material into the maternal circulation [31]. Differentiation was assessed in the cultures by measuring the extent of multinucleated trophoblast at 66 h, using a nuclear stain together with anti-E-cadherin to reveal intercellular borders. During cytotrophoblast aggregation, E-cadherin assembles at cell-cell contact surfaces, and its expression is down-regulated and finally lost as the mononucleate cells undergo cellular differentiation and fusion [28]. Figure 3B shows the effect of either TGM2 inhibitor on the E-cadherinnuclear-staining pattern. When cells were fixed after 66 h with no inhibitor present, some E-cadherin staining was apparent, but many of the cells were multinucleated, with three or more nuclei in a single cell. In contrast, when trophoblasts were cultured in the presence of either cystamine or MDC, increased E-cadherin expression was observed, with a higher proportion of mononucleate cells. Mononucleate cells with continuous Ecadherin-positive cell borders were scored in the presence or absence of TGM2 inhibitors (Fig. 3C). Nuclei present as two or more in a single cell were expressed as a percentage of the total nuclei. In the control cultures, more than 85% of the cells were multinucleated, with a significant reduction when either cystamine (64.5%) or MDC (56.5%) was used (Kruskal-Wallis test, P , 0.0001, and Dunn post hoc test, P , 0.05 control versus cystamine, P , 0.001 control versus MDC). Syncytiotrophoblast are the major source of the glycoprotein hCG, a widely used marker of trophoblast differentiation. It is thought that hCGa is initially produced and is followed by the production of hCGb as trophoblast differentiation progresses. Therefore, hCGb is the limiting factor for the production of hCG and is measured in the hCG assay as an indicator of the presence of total hCG. Levels of hCG were measured in the medium of cells cultured with and without the addition of TGM2 inhibitors (the untreated and treated cells were passaged and maintained in culture at the same time). The inhibitors were added to the trophoblasts before differentiation had occurred, as described in Materials and Methods. Culture supernatants were removed after 66 h for the assay of immunoreactive hCG. There were significant reductions in hCG concentration between control cultures and those treated with either cystamine or MDC (Fig. 3D). The hCG concentration with cystamine (52.2 6 4.5 mIU/ml) was reduced by 60.3% (control hCG, 131.56 11.3 mIU/ml), and competitive substrate inhibitor MDC reduced hCG production by 61.8% (50.3 6 4.0 mIU/ml). Use of either inhibitor caused a significant difference compared with the control (KruskalWallis test, P , 0.0001, and Dunn post hoc test, P , 0.001, in both cases). The cellular protein content of the cultures was not altered by treatment with either inhibitor (Fig. 3E), suggesting the effects seen of TGM2 inhibition on trophoblast differentiation were not just due to decreased cell viability. TGM2-Mediated Cross-Linking in Shed Placental Microparticles TGM2 is thought to stabilize apoptotic cells by preventing the leakage of intracellular components by means of crosslinking. TGM2 cross-linking of proteins in the MVM could therefore occur before the shedding of apoptotic particulate or vesicular material into the maternal circulation. Material shed from primary trophoblast monolayer cultures between 18 and

TGM2 AND SHED PLACENTAL MICROPARTICLES

66 h was isolated by differential centrifugation [16]. Transmission electron micrographs (TEMs) of the shed material demonstrated the presence of various different sizes and types of material ranging from large, apical, syncytial fragments with intact microvilli to microvesicles (Fig. 4, A–C). Actin filaments were seen running longitudinally in microvilli, and cytokeratin filament bundles and microtubules, the protein constituents of which had been identified at TGM2 substrates (Table 1), were also evident. TGM2 protein was readily detectable by Western blotting in the shed material (data not shown). To ascertain whether this fraction of the enzyme was active, the acyl-acceptor probe BTC was added to the cells in the final few hours of culture (from ;63 to 66 h), a time point reflecting the final stages of trophoblast differentiation in the cell cultures, when most nuclei have become incorporated into syncytium, from which shedding of apical material occurs. Shed material was isolated from the culture medium, and blots of the high-speed pellet were probed with avidin peroxidase (Fig. 4D). Numerous bands were present in both fractions, indicating that TGM2 cross-linking activity is present in shed material as well as in living cells. Both the cell homogenate and shed material were found to contain biotinylated b-actin, consistent with its identification as a TGM2 substrate in situ (Table 1). Strikingly, a high-molecular-mass immunoreactive band appeared in shed material but not in live cells and contained actin immunoreactivity, suggesting cross-linking into large multimeric aggregates. The actin monomer at ;42 kDa was of a lower intensity in the shed material than in the cell homogenate, suggesting much of the actin in the shed trophoblast microparticles was multimeric. The TGM2 inhibitors MDC and cystamine were added to the trophoblast cultures to examine the role of TGM2 in the shedding process. Syncytiotrophoblast microparticulate fragments were collected from the medium from 18 to 66 h in culture in the presence or absence of an inhibitor. The presence of tTG inhibitors did not affect the quantity of shed material isolated, with an average yield of 123 6 10.8 lg of protein per 30 3 106 cells without inhibitor, 119 6 12.3 lg of protein per 30 3 106 cells in the presence of cystamine, and 121 6 8.0 lg of protein per 30 3 106 cells in the presence of MDC (n ¼ 5 preparations). Western blots of the shed material from inhibitor-treated cells showed that b-actin was exclusively in the monomeric 42-kDa form (Fig. 4D), with disappearance of the high-molecular-mass band apparent in the untreated shed material. DISCUSSION The syncytiotrophoblast MVM acts as an interface between the placenta and maternal blood in the intervillous space and shows strong TGM2 expression and reactivity [14, 32]. We aimed to identify substrate proteins in the MVM in order to advance understanding of TGM2’s function at this key tissue interface. Assays were carried out in the presence of calcium ions, which are required for the transamidating activity of TGM2. Of .30 proteins cataloged, some were known TGM2 substrates, whereas others had not been previously identified. Several proteins acted both as acyl-donor and -acceptor substrates in the transamidating reaction, strongly suggesting these targets are cross-linked by TGM2 into homo-oligomers. Substrates fall into several functional groups. Cytoskeletal proteins such as actin, myosin heavy and light chains, tubulin, and intermediate filament polypeptides of the cytokeratin family are highly represented. TGM2-mediated polymerization of cytoskeletal proteins has implications for the enzyme in apoptosis. Changes to cytoskeletal organization have been

655

proposed to determine the morphologic events of programmed cell death [3]. TGM2 is often highly expressed during programmed cell death, and its induction has been associated with apoptosis in several cell systems [6]. Despite the appearance of TGM2 in various cells and tissues undergoing apoptosis, it is unclear whether the increase in TGM2 activity causes cell death in various pathological processes or whether the two occurrences are related noncausally. Recent reports have supported the idea that TGM2’s GTP-bound form plays a role in promoting cell survival, with the calcium-dependent transamidating activity being involved in apoptosis [6]. Modification of cytoskeletal proteins such as actin or myosin by TGM2 could lead to the formation of a highly cross-linked protein scaffold that stabilizes apoptotic cells, preventing the release of cellular components into surrounding tissues and thus circumventing the inflammatory response [33]. Various cytoskeletal proteins have previously been identified as TGM2 substrates; for example, actin was identified as a TGM2 substrate in human leukemia cells undergoing apoptosis [34], and myosin and spectrin were recently identified as TGM2 substrates in an intestinal epithelial cell line [35]. MVM proteins, including transferrin receptor, placental alkaline phosphatase (ALPP, also known as PLAP-1), and integrins alpha V and alpha IIb, were also identified as TGM2 substrates. ALPP is attached to the outer leaflet via an inositol link, and its presence in the substrate fraction suggests that TGM2 is active at the outer face of the membrane as well as the inner face, as evidenced by the presence in the target pool of cytoskeletal components and several members of the annexin family of Ca2 þ-dependent phospholipid-binding proteins. Annexins have various functions, including anticoagulation, cytoskeletal interactions, anti-inflammatory activity, and signal transduction. The anticoagulant properties of annexin A5, which binds anionic phospholipids and shields coagulant sites, have been proposed to be crucial for maintaining placental integrity [36]. Annexin A5 may also play a role in trophoblast differentiation, as anti-annexin A5 antibody was demonstrated to block trophoblast fusion in culture, and antisense oligonucleotides to annexin A5 reduced fusion by about 50% [37, 38]. There is evidence that annexin A5 may be present at both the outer and inner surface of the MVM [36, 39, 40]. Syncytiotrophoblast membrane microparticles isolated from normal placenta have previously been shown to inhibit endothelial cell proliferation in vitro [41]. Eight proteins were identified from purified syncytiotrophoblast MVM and found to assemble a self-aggregating complex, which may be responsible for the antiproliferative activity [42]. Seven of these eight proteins have been identified in the current study as TGM2 substrates in MVM (transferrin and its receptor, integrins aV and a5, ALPP, a-actinin, and monoamine oxidase type A), suggesting that TGM2 transamidation is involved in the formation of this aggregating cluster of proteins that becomes shed from the syncytial MVM. In vitro, TGM2 is seen both in syncytio- and cytotrophoblasts in association with cytoskeletal structures, including peripheral actin bundles and stress fibers. Colocalization of TGM2 with stress fibers has previously been demonstrated in human umbilical vein endothelial cells [43], where immunoprecipitation studies suggested that association with stress fibers was due to TGM2’s cross-linking of myosin. The use of a biotinylated substrate allowed transamidating activity to be localized and compared with the protein distribution in live cells. TGM2 localized predominantly to lateral cell borders, whereas cross-linking activity was restricted to a small number of cells and was not obviously associated with lateral cell

656

ROBINSON ET AL.

boundaries. This suggests that subcellular regulation of TGM2 transamidating activity occurs within the MVM, perhaps by means of intracellular Ca2þ concentration [1]. Because there are significant amounts of intracellular TGM2 in trophoblasts that are not associated with the MVM, TGM2 may carry out distinct functions in different cellular compartments in this system. The involvement of TGM2-mediated cross-linking in trophoblast differentiation was examined by the use of TGM2 inhibitors during differentiation in vitro and monitoring of the formation of multinucleated syncytiotrophoblasts (which occurs by fusion) and hCG secretion. Fusion was reduced by treatment with either TGM2 inhibitor and was accompanied by no loss in cell viability. Cystamine or MDC was added to the cytotrophoblast cultures prior to the onset of differentiation and significantly inhibited hCG secretion. Undifferentiated cytotrophoblast cultures produce low levels of hCG, and a reduction in secretion in the presence of inhibitor is consistent with the inhibition of syncytiotrophoblast formation. Although care must be applied in extrapolation of these in vitro data to the in vivo situation, these findings suggest that TGM2 enzymatic activity is important for normal trophoblast turnover in the placental villus. In vivo, after cytotrophoblast differentiation and fusion to form syncytiotrophoblast, terminal differentiation occurs within the syncytium. This eventually leads to the packaging of apoptotic material into syncytial knots and smaller vesicular structures, which are shed into maternal circulation [44]. TGM2 cross-linking of cytoskeletal proteins has implications in this normal process of cell turnover in the placental villus. TGM2 cross-linking of actin in the MVM could occur at the end of the trophoblast life cycle, before shedding. Previous analytic and functional studies have utilized either particles isolated from maternal venous blood [15] or microvillusderived membrane fractions derived by various methods (perfusion, explant, and mechanical dissection) [16, 41] from villous tissue. In the former case, the particles represent a fraction that has survived in circulation, while in the latter case, the preparation is from the intact cell membrane and not derived by a natural process of shedding. We have now established that microparticulate fragments are shed by primary trophoblasts in culture, suggesting the suitability of this model for such studies. These microparticles were found to contain abundant TGM2 protein but had less TGM2 cross-linking activity than equivalent loadings of cell protein (data not shown). The biotinylated acyl-probe was added to the trophoblast cells from 63 to 66 h in culture; thus, the presence of biotinylated bands in the shed particles shows TGM2 activity occurring in the final few hours of culture. This suggests that the predominant TGM2 transamidating activity occurs before material is shed from the syncytiotrophoblast. TGM2 inhibitors did not prevent shedding in vitro, as indicated by the presence of b-actin in approximately equivalent amounts in both the inhibitor-treated and control shed fragments. In the presence of either inhibitor, actin monomer was present in the shed material, in comparison to the nontreated shed material, where high-molecular-mass multimeric actin was present. This is consistent with less actin becoming cross-linked into larger complexes in the shed material from inhibitor-treated cultures. TGM2’s cross-linking activity may be important in the preparation of apically associated membrane material for shedding, suggesting that it has as a vital role to play in placental homeostasis and function. TGM2 cross-linking of material destined to be shed from the placenta may enhance phagocytosis by the maternal reticuloendothelial system and

contribute to reducing the maternal immune reaction to the high quantities of fetal antigen shed during pregnancy. To summarize, substrates for TGM2 have been identified at the primary maternal-fetal interface, the most abundant types including cytoskeletal and membrane-associated proteins. Syncytial microparticles shed from primary trophoblast cultures contain abundant TGM2 protein and cross-linking activity. A substantial proportion of actin in the shed particles is in the form of polymeric aggregates, in contrast to cellular actin, which dissociates completely into monomers in SDSPAGE. Primary trophoblast cells cultured in the presence of transglutaminase inhibitors showed decreased trophoblast differentiation, and actin in the shed particulate material remained exclusively in monomeric form. These results suggest a role for TGM2-mediated cross-linking in stabilizing particulate material shed from the placenta and, more generally, in the organization and turnover of trophoblast plasma membrane and associated cytoskeleton. REFERENCES 1. Griffin M, Casadio R, Bergamini CM. Transglutaminases: nature’s biological glues. Biochem J 2002; 368:377–396. 2. Mishra S, Murphy LJ. Tissue transglutaminase has intrinsic kinase activity: identification of transglutaminase 2 as an insulin-like growth factor-binding protein-3 kinase. J Biol Chem 2004; 279:23863–23868. 3. Aeschlimann D, Kaupp O, Paulsson M. Transglutaminase-catalyzed matrix cross-linking in differentiating cartilage: identification of osteonectin as a major glutaminyl substrate. J Cell Biol 1995; 129:881–892. 4. Akimov SS, Belkin AM. Cell surface tissue transglutaminase is involved in adhesion and migration of monocytic cells on fibronectin. Blood 2001; 98:1567–1576. 5. Fesus L, Thomazy V, Falus A. Induction and activation of tissue transglutaminase during programmed cell death. FEBS Lett 1987; 224: 104–108. 6. Fesus L, Szondy Z. Transglutaminase 2 in the balance of cell death and survival. FEBS Lett 2005; 579:3297–3302. 7. Caputo I, D’Amato A, Troncone R, Auricchio S, Esposito C. Transglutaminase 2 in celiac disease: minireview article. Amino Acids 2004; 26:381–386. 8. Ludvigsson JF, Montgomery SM, Ekbom A. Celiac disease and risk of adverse fetal outcome: a population-based cohort study. Gastroenterology 2005; 129:454–463. 9. Sheiner E, Peleg R, Levy A. Pregnancy outcome of patients with known celiac disease. Eur J Obstet Gynecol Reprod Biol 2006; 129:41–45. 10. Norgard B, Fonager K, Sorensen HT, Olsen J. Birth outcomes of women with celiac disease: a nationwide historical cohort study. Am J Gastroenterol 1999; 94:2435–2440. 11. Martinelli P, Troncone R, Paparo F, Torre P, Trapanese E, Fasano C, Lamberti A, Budillon G, Nardone G, Greco L. Coeliac disease and unfavourable outcome of pregnancy. Gut 2000; 46:332–335. 12. Ciacci C, Cirillo M, Auriemma G, Di Dato G, Sabbatini F, Mazzacca G. Celiac disease and pregnancy outcome. Am J Gastroenterol 1996; 91:718– 722. 13. Hager H, Gliemann J, Hamilton-Dutoit S, Ebbesen P, Koppelhus U, Jensen PH. Developmental regulation of tissue transglutaminase during human placentation and expression in neoplastic trophoblast. J Pathol 1997; 181:106–110. 14. Robinson NJ, Glazier JD, Greenwood SL, Baker PN, Aplin JD. Tissue transglutaminase expression and activity in placenta. Placenta 2006; 27: 148–157. 15. Knight M, Redman CW, Linton EA, Sargent IL. Shedding of syncytiotrophoblast microvilli into the maternal circulation in preeclamptic pregnancies. Br J Obstet Gynaecol 1998; 105:632–640. 16. Gupta AK, Rusterholz C, Huppertz B, Malek A, Schneider H, Holzgreve W, Hahn S. A comparative study of the effect of three different syncytiotrophoblast micro-particles preparations on endothelial cells. Placenta 2005; 26:59–66. 17. Glazier JD, Jones CJ, Sibley CP. Purification and Na þ uptake by human placental microvillus membrane vesicles prepared by three different methods. Biochim Biophys Acta 1988; 945:127–134. 18. Ruoppolo M, Orru S, D’Amato A, Francese S, Rovero P, Marino G, Esposito C. Analysis of transglutaminase protein substrates by functional proteomics. Protein Sci 2003; 12:1290–1297.

TGM2 AND SHED PLACENTAL MICROPARTICLES 19. Kliman HJ, Nestler JE, Sermasi J, Sanger M, Strauss JFI. Purification, characterisation, and in vitro differentiation of cytotrophoblasts from human term placentae. Endocrinology 1986; 118:1567–1582. 20. Greenwood SL, Boyd RD, Sibley CP. Transtrophoblast and microvillus membrane potential difference in mature intermediate human placental villi. Am J Physiol 1993; 265:C460–466. 21. Greenwood SL, Clarson LH, Sides MK, Sibley CP. Membrane potential difference and intracellular cation concentrations in human placental trophoblast cells in culture. J Physiol 1996; 492(pt 3):629–640. 22. Lee KN, Arnold SA, Birckbichler PJ, Patterson MK Jr, Fraij BM, Takeuchi Y, Carter HA. Site-directed mutagenesis of human tissue transglutaminase: Cys-277 is essential for transglutaminase activity but not for GTPase activity. Biochim Biophys Acta 1993; 1202:1–6. 23. Siefring GE Jr, Apostol AB, Velasco PT, Lorand L. Enzymatic basis for the Ca2 þ-induced cross-linking of membrane proteins in intact human erythrocytes. Biochemistry 1978; 17:2598–2604. 24. Cooper AJ, Jeitner TM, Gentile V, Blass JP. Cross linking of polyglutamine domains catalyzed by tissue transglutaminase is greatly favored with pathological-length repeats: does transglutaminase activity play a role in (CAG)(n)/Q(n)-expansion diseases? Neurochem Int 2002; 40:53–67. 25. Stenberg P, Curtis CG, Wing D, Tong YS, Credo RB, Gray A, Lorand L. Transamidase kinetics. Amide formation in the enzymatic reactions of thiol esters with amines. Biochem J 1975; 147:155–163. 26. Gentile V, Thomazy V, Piacentini M, Fesus L, Davies PJ. Expression of tissue transglutaminase in Balb-C 3T3 fibroblasts: effects on cellular morphology and adhesion. J Cell Biol 1992; 119:463–474. 27. Jones RA, Nicholas B, Mian S, Davies PJ, Griffin M. Reduced expression of tissue transglutaminase in a human endothelial cell line leads to changes in cell spreading, cell adhesion and reduced polymerisation of fibronectin. J Cell Sci 1997; 110(pt 19):2461–2472. 28. Coutifaris C, Kao LC, Sehdev HM, Chin U, Babalola GO, Blaschuk OW, Strauss JF III. E-cadherin expression during the differentiation of human trophoblasts. Development 1991; 113:767–777. 29. Jones CJ, Fox H. Syncytial knots and intervillous bridges in the human placenta: an ultrastructural study. J Anat 1977; 124:275–286. 30. Smethurst PA, Bungay PJ, Griffin M. The use of a biotin-labelled amine to identify endogenous substrates of transglutaminase in pancreatic islets. Biochem Soc Trans 1993; 21:424S. 31. Jones CJ, Fox H. Ultrastructure of the normal human placenta. Electron Microsc Rev 1991; 4:129–178. 32. Hager H, Jensen PH, Hamilton-Dutoit S, Neilsen MS, Birckbichler P, Gliemann J. Expression of tissue transglutaminase in human bladder carcinoma. J Pathol 1997; 183:398–403. 33. Piredda L, Amendola A, Colizzi V, Davies PJ, Farrace MG, Fraziano M, Gentile V, Uray I, Piacentini M, Fesus L. Lack of ‘tissue’ transglutaminase

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

657

protein cross-linking leads to leakage of macromolecules from dying cells: relationship to development of autoimmunity in MRLIpr/Ipr mice. Cell Death Differ 1997; 4:463–472. Nemes Z Jr, Adany R, Balazs M, Boross P, Fesus L. Identification of cytoplasmic actin as an abundant glutaminyl substrate for tissue transglutaminase in HL-60 and U937 cells undergoing apoptosis. J Biol Chem 1997; 272:20577–20583. Orru S, Caputo I, D’Amato A, Ruoppolo M, Esposito C. Proteomics identification of acyl-acceptor and acyl-donor substrates for transglutaminase in a human intestinal epithelial cell line. Implications for celiac disease. J Biol Chem 2003; 278:31766–31773. Krikun G, Lockwood CJ, Wu XX, Zhou XD, Guller S, Calandri C, Guha A, Nemerson Y, Rand JH. The expression of the placental anticoagulant protein, annexin V, by villous trophoblasts: immunolocalization and in vitro regulation. Placenta 1994; 15:601–612. Vogt E, Ng AK, Rote NS. Antiphosphatidylserine antibody removes annexin-V and facilitates the binding of prothrombin at the surface of a choriocarcinoma model of trophoblast differentiation. Am J Obstet Gynecol 1997; 177:964–972. Rote NS, Das M, Xu B, Lin L, Kumar N. Monoclonal antiphospholipid and anti-annexin a5 antibodies prevent intercellular fusion in a model of human villous cytotrophoblasts. Eur J Clin Invest 2003; 33:62. Rand JH, Wu XX, Guller S, Gil J, Guha A, Scher J, Lockwood CJ. Reduction of annexin-V (placental anticoagulant protein-I) on placental villi of women with antiphospholipid antibodies and recurrent spontaneous abortion. Am J Obstet Gynecol 1994; 171:1566–1572. Wang X, Campos B, Kaetzel MA, Dedman JR. Annexin V is critical in the maintenance of murine placental integrity. Am J Obstet Gynecol 1999; 180:1008–1016. Smarason AK, Sargent IL, Starkey PM, Redman CW. The effect of placental syncytiotrophoblast microvillous membranes from normal and pre-eclamptic women on the growth of endothelial cells in vitro. Br J Obstet Gynaecol 1993; 100:943–949. Kertesz Z, Hurst G, Ward M, Willis AC, Caro H, Linton EA, Sargent IL, Redman CW. Purification and characterization of a complex from placental syncytiotrophoblast microvillous membranes which inhibits the proliferation of human umbilical vein endothelial cells. Placenta 1999; 20: 71–79. Chowdhury ZA, Barsigian C, Chalupowicz GD, Bach TL, Garcia-Manero G, Martinez J. Colocalization of tissue transglutaminase and stress fibers in human vascular smooth muscle cells and human umbilical vein endothelial cells. Exp Cell Res 1997; 231:38–49. Huppertz B, Frank HG, Kingdom JC, Reister F, Kaufmann P. Villous cytotrophoblast regulation of the syncytial apoptotic cascade in the human placenta. Histochem Cell Biol 1998; 110:495–508.