Mol. Hum. Reprod. Advance Access originally published online on December 5, 2005
Molecular Human Reproduction 2005 11(10):699-710; doi:10.1093/molehr/gah185
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The soluble pool of HLA-G produced by human trophoblasts does not include detectable levels of the intron 4-containing HLA-G5 and HLA-G6 isoforms
1Institute of Cell Biology, Histology and Embryology/Center of Molecular Medicine, Medical University of Graz, Harrachgasse 21, A-8010 Graz, Austria, 2Institut für Immungenetik, Charité-Universitätsmedizin Berlin, Spandauer Damm 130, 14050 Berlin, Germany and 3Clinic of Obstetrics and Gynecology, Medical University of Graz, Auenbruggerplatz 14, A-8036 Graz, Austria
4 To whom correspondence should be addressed at: Institute of Cell Biology, Histology and Embryology/Center of Molecular Medicine, Medical University of Graz, Harrachgasse 21/7, A-8010 Graz, Austria. E-mail: gottfried.dohr{at}meduni-graz.at
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
In the context of implantation and pregnancy, several immunomodulating functions have been attributed to the different HLA-G isoforms. Increasing attention is now being addressed to the actively secreted soluble forms, because they might have a systemic function or could be useful as diagnostic tools. However, the cellular source of secretion, even during pregnancy, where HLA-G expression level is known to be highest, is still under debate. To elucidate the conflicting results, we investigated the isoform distribution in human first trimester and term placentas in situ and in vitro. Results obtained by applying immunohistochemistry, western blot, enzyme-linked immunosorbent assay (ELISA) and RT–PCR show that (1) all of the
1 domain-containing HLA-G isoforms are restrictedly expressed in the extravillous cytotrophoblasts (EVCTs) and very few first-trimester syncytiotrophoblasts, which directly cover cell columns, whereas mesenchymal cells of the villous chorion do not express HLA-G; (2) as demonstrated in western blots, trophoblasts express only the HLA-G1 isoform; (3) HLA-G5 and -G6 transcripts could be detected in human term placenta and isolated first-trimester trophoblasts but levels are extremely low; and (4) conditioned media of primary first-trimester trophoblasts, and the chorion laeve-derived trophoblastic cell line AC1-M59 do contain HLA-G1 fragments shed from the cell surface. Our data provide substantial evidence that none of the intron 4-containing isoforms, the so-called actively secreted, soluble HLA-G5 or -G6, are produced by human trophoblasts in situ or in vitro. Key words: HLA-G/human placenta/reproductive immunology/MHC class I/trophoblast
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Since 1987, when HLA-G was first detected by Geraghty et al. (1987)
, lots of effort has been undertaken to investigate expression and function of the nonclassical class Ib major histocompatibility complex (MHC) gene. Initially it was thought that the HLA-G protein was exclusively expressed during pregnancy at the feto–maternal interface (Ellis et al., 1986; Kovats et al., 1990; McMaster et al., 1995), and on a subset of thymic epithelial cells (Crisa et al., 1997; Mallet et al., 1999), involved in maintaining the tolerance of the maternal immune system towards the semi-allogeneic foetus. It has been shown that HLA-G1–expressing extravillous cytotrophoblasts (EVCTs) interact with the local maternal immune system by inhibiting both decidual natural killer (NK) and T cell-mediated cell lysis either directly or indirectly by providing a peptide resulting from their leader sequence for HLA-E, which, thus stabilized, reaches the cell surface and interacts with the CD94/NKG2 receptor (Llano et al., 1998; Ponte et al., 1999). Although basically similar, HLA-G differs from the classical HLA class Ia molecules by its low polymorphism, a shortened cytoplasmic tail, which prevents endocytosis and prolonged cell surface expression (Davis et al., 1997; Park et al., 2001), the ability to load only a restricted set of peptides (Lee et al., 1995) and its tendency to alternative splicing (Ishitani and Geraghty, 1992; Kirszenbaum et al., 1994), yielding four membrane-bound proteins HLA-G1, -G2, -G3 and -G4 and three secreted products HLA-G5 (sHLA-G1), -G6 (sHLA-G2) and HLA-G7 (Le Bouteiller and Blaschitz, 1999; Paul et al., 2000). The so-called ‘actively secreted’ HLA-G5, -G6 and -G7 isoforms are highly unusual, as they result from intron inherent premature termination codons left behind by the incomplete splicing process. Recently, an additional soluble but not actively secreted form, cleaved from the membrane by metalloproteinases, was identified as the functionally active shed HLA-G1 (Park et al., 2004). Since then further studies have exposed an increasing diversity of HLA-G’s functional aspects, and growing interest is focused especially on the soluble forms, because potential systemic effects have been suggested for these, and utilization as a diagnostic parameter in IVF, pre-eclampsia, inflammatory and neoplastic disease seems to be reasonable (Le Bouteiller et al., 2003; LeMaoult et al., 2003; Singer et al., 2003; Steinborn et al., 2003; Noci et al., 2005). HLA-G has been demonstrated to induce apoptosis of activated CD8+ T cells (Fournel et al., 2000a) to suppress the allo proliferation of CD4+ T cells in mixed lymphocyte reaction (Bainbridge et al., 2000b; LeMaoult et al., 2004), to induce a Th2 cytokine profile (Rieger et al., 2002) and to suppress or augment the allo-CTL responses in a dose-dependent manner (Kapasi et al., 2000). Furthermore, cells with HLA-G surface expression were shown to be less susceptible to human cytomegalovirus (HCMV) infection, and the modulation of endothelial cell activity was demonstrated in vitro by soluble HLA-G (reviewed by Le Bouteiller et al., 2003). However, most of the experiments were designed using HLA-G-transfected cell lines, or their conditioned media and some of the results could not be reproduced when trophoblast-derived HLA-G was employed (Kapasi et al., 2000).
The analysis of the HLA-G tissue distribution under physiological and pathological conditions revealed even more diversity (Le Discorde et al., 2003). Rebmann et al. (2003a) detected peripheral blood monocytes as pregnancy independent cellular source of HLA-G secretion. In patients suffering from cancer, HLA-G expression was attributed either to tumor cells or to antigen-presenting cells (APCs), which were thought to play an inhibiting role in the tolerogenic cascade of events (Seliger et al., 2003; Le Friec et al., 2004). HLA-G was also found in well-accepted transplanted organs and during the course of inflammatory diseases (Lila et al., 2002) or HIV infections (LeMaoult et al., 2003). However, some of the studies published so far present controversial results (Polakova and Russ, 2000; Wagner et al., 2000; Bainbridge et al., 2001; Davies et al., 2001; Hurks et al., 2001; Seliger et al., 2003; Ibrahim el et al., 2004). Fuzzi et al. (2002) reported that embryos growing in vitro express HLA-G1 and -G5, whereas Van Lierop et al. (2002) demonstrated no HLA-G secretion by preimplantation embryos.
The most conflicting results concerning the secreted HLA-G products (HLA-G5, -G6 and -G7) derive from the placenta itself. Several reports in the past have shown that among the different trophoblast populations, only the EVCT expresses HLA-G, whereas the villous trophoblast expresses neither HLA class Ia nor HLA class Ib molecules (Hutter et al., 1996; Loke and King, 2000). Surprisingly, HLA-G5 secretion has recently been attributed to the syncytiotrophoblast (Solier et al., 2002; Morales et al., 2003), which constitutes an extensive surface responsible for the metabolic exchange between mother and foetus. Such HLA-G secretion into the maternal circulation should consequently result in increased serum levels during gestation, but HLA-G5 serum levels measured in pregnant women were almost equal to those of healthy nonpregnant females and even lower than those of males (Rebmann et al., 1999). Hunt et al. (2000a) have proposed that mainly HLA-G6 circulates in maternal blood during pregnancy, and Morales et al. (2003) immunolocalized soluble HLA-G6 and/or HLA-G2 molecules in EVCTs exclusively. Although several groups have been successful in expressing and detecting the shorter HLA-G isoforms (Menier et al., 2000; Riteau et al., 2001; Morales et al., 2003), Ulbrecht et al. (2004) and others have demonstrated that the truncated isoforms are retained in the endoplasmic reticulum and do not reach the endpoint of their secretory pathway (Bainbridge et al., 2000a; Mallet et al., 2000). Finally, Menier et al. (2004) showed with mAb 5A6G7 that besides extravillous trophoblasts, foetal erythroid progenitor cells might secrete HLA-G5 and/or -G6 throughout prenatal life, but several immunolocalization reports, investigating histological sections of human first trimester or term placentas, failed to detect HLA-G within foetal blood cells, which are frequently seen in such sections. Additionally, soluble HLA-G was not detectable in umbilical cord blood samples (Fournel et al., 1999, 2000b; Hunt et al., 2000b). Most of these conflicting data might reflect differences in methodologies used for HLA-G detection, and lack of specificity has been demonstrated for some anti-HLA-G antibodies (Sedlmayr et al., 2002; Seliger et al., 2003; Polakova et al., 2004).
In the light of the conflicting data and the poorly defined source of HLA-G secretion in human placenta, we designed this study, which compares for the first time results of two 1 domain specific pan HLA-G markers (4H84, MEM-G/1) with two intron 4-binding HLA-G5 and -G6 antibodies (16G1, 5A6G7) under strictly controlled conditions. We investigated serial sections of first trimester and term placenta samples immunohistochemically. Additionally, a very sensitive western blot technique and an isoform specific RT–PCR with even higher sensitivity were applied to analyse the HLA-G expression in term placenta samples taken from three different topographic regions as well as from primary culture trophoblasts generated from first-trimester placenta. Finally, we addressed the question as to whether media conditioned by freshly isolated first-trimester trophoblasts and supernatants from the term chorion laeve-derived, immortalized trophoblastic cell line AC1-M59 contained actively secreted HLA-G5 and/or cleaved soluble HLA-G1 products.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Tissue collection and isolation of first-trimester trophoblasts
Placentas were provided by the Clinic of Obstetrics and Gynecology according to a protocol approved by the ethics committee of the Medical University of Graz. Term placenta samples freshly obtained from uncomplicated normal deliveries (n = 5) were cut from three different topographic regions: the villous chorion, the basal plate and the chorion laeve. Foetal membranes were separated by peeling off the amnion epithelium from the chorion laeve; both parts were morphologically controlled by instantaneous cryosections. All samples were measured in weight and immediately frozen in liquid nitrogen for immunohistochemical investigation, RT–PCR and western blotting. A part of each sample was fixed in 4% phosphate-buffered paraformaldehyde and embedded in paraffin. First-trimester placenta samples were collected from pregnancy terminations for psychosocial reasons at week 6–9 (n = 5), immediately cooled on ice and processed for immunohistochemistry and/or trophoblast isolation. Trophoblasts including the villous and extravillous populations were isolated, as previously described (Blaschitz et al., 2000b
). In brief, first-trimester villous chorion was digested using Dispase/DNase and Trypsin, and certrifuged on a Percoll gradient. Enriched trophoblasts were further purified by immunodepletion using anti-CD45RB (leucocyte common antigen, Dako, Glostrup, Denmark) and anti-fibroblast–specific antigen (FSA, Dianova, Hamburg, Germany) in combination with immunomagnetic beads (Dynabeads, Dynal, Oslo, Norway). Purified, cytokeratin 7 positive trophoblasts (98%) were analysed with mAb MEM-G/1 immunocytochemically and identified as 
60% HLA-G positive.
Cell culture of primary trophoblasts, cell lines, transfectants and collection of conditioned media
All nutrient media were obtained from Invitrogen (DH Breda, The Netherlands) and supplemented with d-glucose (4.5 mg/ml), l-glutamine (0.584 mg/ml), penicillin (100 U/ml), streptomycin (0.1 mg/ml) and 1 mM sodium pyruvate (all PAA, Pasching, Austria). Cell lines were cultivated with or without the addition of 10% fetal calf serum (FCS), because western blot analysis benefits from a reduced or omitted FCS proportion in the sample by increasing the relative amount of cell-specific proteins loaded on each lane. Additionally, mAb 16G1 in particular required serum-free conditions due to a strong cross-reaction to FCS, which was evaluated in control experiments analysing neat media containing 10% FCS. None or only minor FCS supplements to the culture media allowed us to trace HLA-G5 with mAb 16G1. About 2 x 106 freshly isolated primary first-trimester trophoblasts were cultured in Dulbecco’s modified Eagle’s medium (DMEM); the AC1-M59 cell line (Funayama et al., 1997; Gaus et al., 1997), a naturally HLA-G-expressing cell line originating from term extravillous chorion laeve trophoblast and immortalized by fusion with the choriocarcinoma cell line Jeg3, was maintained in nutrient mixture F12 HAM. Jeg3 cells, also naturally expressing HLA-G, and the HLA class I negative choriocarcinoma cell line Jar American Type Culture Collection (ATCC) were cultivated in minimum essential medium (MEM). In our experiments, AC1-M59 cells turned out to be a better model for term placenta trophoblast than Jeg3, because immunocytochemical evaluation with mAb MEM-G/1 revealed a stronger HLA-G signal expressed by a higher number of cells. The lymphoblastoid cell line 721.221 (.221) and its transfectants .221-G1 and .221-G5 (Fujii et al., 1994) were maintained in Roswell Park Memorial Institute 1640 (medium) (RPMI 1640) and selected in media containing 1 mg/ml of geneticin (Gibco Life Technologies, Lofer, Austria). The myeloid K562 cell line was transfected by electroporation with artificial HLA-G2 cDNA constructs that were produced by PCRs using HLA-G2 isoform-specific primers and a previously used HLA-G cDNA clone as template DNA. In brief, PCR products coding for HLA-G2 were cloned into PGEM-T plasmid (Promega, Mannheim, Germany) and sequenced (MWG-Biotech, Ebersberg, Germany). After digestion by XbaI/KpnI, the fragments were subcloned into the pcDNA3 or pcDNA3.1/hygro expression vector (Invitrogen, DH Breda, The Netherlands). Cells were maintained in RPMI medium supplemented with 1 µg/ml of gentamicin. Conditioned media of all cell lines were harvested after 24 h, aliquoted and stored frozen. Additionally, media from primary trophoblasts, AC1-M59, Jeg3 and Jar cells were concentrated 5- to 10-fold using a 10 kDa Sartorius filter membrane (Vivaspin 6TM Concentrator, VS0601, Hanover, Germany).
Generation of recombinant HLA-G ectodomain heavy chains
Recombinant HLA-G1 ectodomain heavy chains were generated in Escherichia coli BL21 (DE3) pLysS (Novagen/VWR, Vienna, Austria) as inclusion bodies using an HLA-G1 ectodomain-containing pGMT7 expression vector, kindly provided by J. Boyson (Boyson et al., 2002); rHLA-A2 ectodomain and rß2m were provided by A. Ziegler (Garboczi et al., 1992). The samples were solubilized in 8 M of urea and diluted in phosphate-buffered saline (PBS).
Reverse transcription–preamplification PCR
Frozen basal plate, chorion leave and villous chorion samples were immersed for 1 week in RNAlater®ICE (Ambion, Huntingdon, UK) at –20°C to avoid RNA degradation during thawing. After thawing, fragments were homogenized in Trifast (Peqlab, Erlangen, Germany) with a tissue homogenizer (Potter-Elvehjem, Novodirekt, Kehl, Germany). Isolated first-trimester trophoblasts were directly lysed with Trifast in the culture dish. RNA was isolated according to the manufacturer’s instructions. After DNase I digestion (Ambion), RNA was quantified photometrically. Approximately 2.5 µg of total RNA and 10 pmol of RT primer (5'-CTCTCAAGGATCTTACCGCTTTTTT- TTTTTTTTTTTTTVN) were used for a 20 µl reverse transcription (RT) reaction using the Improm 2 system (Promega). About 0.5 µl of first-strand cDNA was used for second-strand synthesis in a 20 µl PCR [reaction solution: 1x Gentherm PCR buffer, 2 mM of Mg2Cl, 100 nM of dNTPs, 1.4% acetamide, 500 fMol of HLA-G2ndS-Primer (5'-CGACTGACTCTATCTAATGCTCCAGAGGAGACACGGAACAC), 500 fMol of BACT2ndS-Primer (5'-CTGAC- TCTATCTAATGCTCCACGAAACTACCTTCAACTC), 0.025 U/µl of Gentherm Taqpolymerase (Rapidozym, Berlin, Germany), 0.025 U/µl of DeepVent exo polymerase (New England Biolabs, Frankfurt/Main, Germany)] with the following cycling conditions: three cycles at 95.5°C for 15 s, 50°C for 30 s and 72°C for 3 min. Reactions were kept at 72°C after the last step, and another 20 µl of reaction solution [same as above but with the following primers: preamplification forward (5'-CGACTGACTCTATCTAATGCTCC) and preamplification reverse (5'-CTCTCAAGGATCTTACCGCTTT)] primers were added to each sample. After 23 cycles (at 95.5°C for 15 s, 54°C for 20 s, 72°C for 3 min), the reaction was chilled to 4°C and 1 µl was used for gene/isotype-specific PCR. The HLA-G2ndS primer was designed in a way that allows differentiation between HLA-G and all other HLA-transcripts by at least two bases. To evaluate HLA-G specificity of preamplification, the HLA-C transcript content of the preamplification reaction was also quantified, because HLA-C is known to be expressed in trophoblasts. This preamplification method was derived from the nongene-specific Three Prime End Amplification (TPEA) (Dixon et al., 1998).
Gene-/isotype-specific PCR
Quantification of HLA-G isotypes and ß-actin was done by real-time PCR using a Rotorgene cycler (Corbett research, Mortlake, Australia). The reaction solution (see Reverse transcription-preamplification PRC) was supplemented with 2 mM of tetra-methyl-ammonium oxalate and 1/100 000 v/v CyberGreen (Invitrogen, Karlsruhe, Germany). The following primers were used: for HLA-G1 (G_2–3For 5'-CCAGAGCGAGGCCAGTTCTC/G_4–5Rev 5'-GGGCAGaGAAGACTGCTTCCA), the 7th base indicated by a small letter does not reflect the HLA-G sequence and was introduced to avoid an internal primer loop with a very high melting temperature; for HLA-G5 (G_2–3For/G_4-i4Rev 5'-GCCTCCATCTCCCTCCTTA- CTCC); for HLA-G6 (G_2–4For 5'-ACCAGAGCGAGGCCAACC/G_4-i4Rev); for HLA-C (C-For 5'-GAGACCAGGCCAGCAGGA/C-Rev 5'-TGGACGCA- GCCTGAGAGC); and for ß-actin (BACT-For 6'-CGTGGACATCCGCAAAGACC/BACT-Rev 5'-ACATCTGCTGGAAGGTGGAC).
For amplification, 35 cycles (at 96°C for 20 s, 54°C for 15 s, 72°C for 90 s) were conducted. Additionally to melting point analysis, all products were checked by acrylamide gel electrophoresis. To eliminate different PCR efficiencies, we used calibration curves for the standardization of the quantifications. Calibration curves for each analysed HLA-G isotype were generated by conducting PCR with dilution series made from DNA of the respective isotype. The ß-actin values were used to check whether comparable amounts of RNA were used from different tissue preparations.
Antibodies
Three monoclonal anti-HLA-G antibodies (pan HLA-G mAbs) capable of binding an epitope present on the 1 domain, which is part of all possible isoforms, were used in this study: MEM-G/1 (IgG1, Exbio, Prague, Czech Republic) (Menier et al., 2003), 4H84 (IgG1) (McMaster et al., 1998) and HCA2 (IgG1) (Seitz et al., 1998). HCA2 was shown to bind all HLA-G isoforms but also HLA-A, and a few -B and -E molecules. Monoclonal antibody MEM-G/9 (IgG1) (Menier et al., 2003), shown to recognize at least HLA-G1 and -G5 but not -G2, was only used for enzyme-linked immunosorbent assay (ELISA) and immunostaining of cryosections and cytospins, but not on paraffin sections and in western blots because both techniques destroyed the MEM-G/9 antigenic-binding site irretrievably. Experiments searching for soluble HLA-G5 and -G6 employed mAb 16G1 (IgG1) (Lee et al., 1995) and/or mAb 5A6G7 (IgG1 Exbio, Prague, Czech Republic) (LeMaoult et al., 2003; Le Rond et al., 2004), which were both produced against slightly different peptides deriving from the intron 4-encoded part of HLA-G5 and -G6. For some ELISA experiments, we used biotinylated W6/32 and rabbit anti-ß2m (Dako Cytomation). Isotype-matched negative control antibodies (Dako Cytomation or Sigma Aldrich, Vienna, Austria) were included in all of the experiments.
Immunohistochemistry and immunocytochemistry
Serial sections of frozen tissues and cytospun cells were labelled using a biotin streptavidine horse-radish peroxidase (HRP) system and 3-amino-9-ethylcarbazole (AEC) (Labvision, Fremont, CA, USA) following a previously described protocol (Blaschitz et al., 2000a). Sections cut from paraffin blocks were processed by heat-mediated antigen retrieval in citrate buffer at pH 6.9 under steam pressure (120°C) and blocking of endogenous peroxidase with 3% H2O2, and visualized by the EnVision-HRP-DAB detection system (Dako Cytomation) according to the manufacturer’s protocol. Primary antibodies HCA2 [culture supernatant (CS), 1:10], 4H84 (CS, 1:1000), MEM-G/1 and MEM-G/9 (2 µg/ml), 16G1 (5 µg/ml), 5A6G7 (10 µg/ml) and mouse IgG1 isotype control antibody (10 µg/ml) were diluted in protein-protecting diluent buffer (Labvision) and incubated for 30 min at room temperature. Between incubation steps, the slides were washed in Tris-buffered saline (TBS), finally counterstained with Mayer’s hemalum (VWR) and mounted either aqueously for AEC or permanently for 3,3'-diaminobenzidine-labelled (DAB-labelled) tissues.
ELISA
The 96-well plates (Nunc maxi sorbTM, VWR) were coated with capture antibodies or isotype-matched negative control antibodies (10 µg/ml in carbonate/bicarbonate buffer, pH 9.6) and incubated at 4°C overnight. After a blocking step with post-coating buffer [2% bovine serum albumin (BSA) in PBS, pH 7.4] for 1 h at 37°C, the samples were incubated at 4°C overnight. Biotinylated detection antibody (1 µg/ml in post-coating buffer) was added and incubated for 1 h at room temperature. AMDEXTM streptavidine poly-HRP (Amersham, Vienna, Austria) diluted 1:4000 was added and incubated for 30 min at room temperature. HRP was developed with freshly prepared ABTS (2,2’-azinobis[3-ethylbenzo-thiazoline-6-sulfonic acid]-diamonium salt) solution (Roche, Vienna, Austria), and the optical density (OD) was analysed at 405 nm. All steps were separated by washing procedures (three times in PBS, pH 7.4, containing 0.05% Tween-20, 300 µl per well). All samples, neat or concentrated, were tested in duplicate. Detection limits of the ELISA assays were determined by analysing the .221-G5 CS with quantitative western blot, using a dilution series of rHLA-G1 ectodomain heavy chains and employing the AlphaDigiDoc 1000TM gel documentation and image analysis system (Alpha Innotech, San Leandro CA, USA).
Western blot analysis
Cell monolayers were scraped from the culture dish, and cells growing in suspension were collected by centrifugation. After two ice cold PBS wash steps, the pellets were lysed in hot lysis buffer made up of 0.01 M of Tris, pH 7.4, 1% sodium dodecyl sulphate (SDS), 1 mM of sodium orthovanadate (Sigma Aldrich) and a protease inhibitor cocktail (Sigma Aldrich), containing aprotinin, leupeptin, E-64, bestatin EDTA and AEBSF (4-(2-aminoethy1)benzenesulfonylfluoride). Frozen placental samples were thawed. Every 200 mg of tissue were supplemented with 600 µl of hot lysis buffer, homogenized carefully with a metal-blade homogenizer and chilled on ice. All samples were cleared by centrifugation at 1000 x g at 4°C for 15 min. Protein concentrations of the respective supernatants were estimated by the protocol of Lowry at OD750, and aliquots were frozen for later use. Equal protein amounts from each sample (10 µg per lane) were analysed in a 10% SDS polyacrylamide gel electrophoresis (PAGE; NuPAGE, Invitrogen, Carlsbad, CA, USA) under reducing conditions and transferred onto nitro-cellulose membranes. Only supernatants of transfectants were used at considerably lower protein concentration (about 0.05 µg per lane) and cell-conditioned media were analysed, neat or concentrated as indicated. Western blots were processed with primary antibodies MEM-G/1 (0.5 µg/ml), 4H84 (1:10,000), 16G1 (1 µg/ml), 5A6G7 (1 µg/ml), HCA2 (1:50) and subsequently with the highly sensitive alkaline phosphatase-labelled Western breeze chemiluminescence kit, according to the manufacturer’s protocol (Invitrogen, Carlsbad, CA, USA). Results were visualized by exposure to hyperfilm (Amersham). All reactions were checked for background signals using irrelevant isotype-matched negative control antibodies (IgG1 1 µg/ml, Dako Cytomation). The enzymatic deglycosylation kit, including PNGase F, NANase II and O-glycosidase DS (Bio-Rad, Munich, Germany) cleaving all N- and most O-linked oligosaccharides from glycoproteins was employed for some experiments to clarify the glycosylation status of trophoblast-derived HLA-G and for molecular weight comparison with deglycosylated HLA-G molecules from transfected cell lines.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Immunolocalization with pan-HLA-G antibodies and the identification of intron 4-retaining soluble HLA-G5 and -G6 isoforms
Control cell lines
For immunolocalization, we first controlled our set of antibodies using wild-type and transfected cell lines .221, .221-G5, .221-G1, K562, K562-G1 and K562-G2. Especially for the detection of the soluble isoforms, it was necessary to analyse whether the various histological techniques affect the results, e.g. by washing away or masking the respective antigens. We therefore compared the immunocytochemical results of acetone-fixed cytocentrifuged cells with those obtained with cryo- and paraffin-sections of cell pellets. Antibodies MEM-G/1, 4H84 and HCA2 have been demonstrated to bind an epitope located on the
1 domain of HLA-G, which is part of all seven HLA-G isoforms, including cleaved or shed products. They are therefore termed pan-HLA-G antibodies (Figure 1a). We found that all three pan-HLA-G antibodies recognized HLA-G1, -G2 and -G5 in cell lines transfected with the respective constructs and in all histological preparations used (data not shown). Untransfected wild types were left unstained. The antibody MEM-G/9, specific for native ß2m–HLA-G1 and -G5 complexes, did not recognize HLA-G2 and could not be used after the paraffin-embedding procedure. Both intron 4-specific antibodies 16G1 and 5A6G7 specifically labelled the HLA-G5–transfected cells regardless of the preparation technique used. Staining was visible even when prepared as acetone-fixed cryosections, indicating that the soluble products were not lost during the washing and staining procedure (Figure 2b and d inset).
|
|
Placenta
A brief introduction about the structural features of human placenta is given in Figure 1b.
Regardless of the histological preparation technique used, three pan-HLA-G markers indicated the restricted expression of HLA-G within the EVCT population of first-trimester cell islands (Figure 2a), term chorion laeve and basal plate (Figure 2e) and as migrating interstitial cells within the decidua (Figure 2g and i). In first-trimester placenta, at few special sites, where the syncytiotrophoblast directly covers the proliferating EVCT at the basis of an anchoring villus, we also detected HLA-G in syncytiotrophoblast (data not shown). However, the remaining large syncytiotrophoblastic surface of all villous trees and their underlying villous cytotrophoblasts never stained with pan-HLA-G markers (Figure 2a and e). As expected, the EVCT labelling with mAb MEM-G/9 (Figure 2c), which does not include the HLA-G2 isoform, was similar to that obtained with pan-HLA-G markers. However, proved by transfected cell lines and schematized in Figure 1a, HLA-G5- and -G6-specific antibodies should be able to reveal results unaffected by other HLA isotypes and stain at least part of pan-HLA-G. However, the intron 4 labelling resulted in a tissue distribution very different from the pan-HLA-G–staining pattern. In cryosections, none of the placenta cells were stained with 16G1 and 5A6G7 (Figure 2b, d and f), but using paraffin serial sections the syncytiotrophoblast, decidua cells (DC) and the uterine gland (UG) epithelium were labelled with 16G1 (Figure 2h). Investigations of decidua basalis with mAb 5A6G7 resulted in a predominant labelling of UG epithelium and no staining of migrating trophoblasts (Figure 2k), which is in complete contrast to the pattern given by MEM-G/1 in a corresponding serial section (Figure 2i).
Western blot analysis of recombinant HLA-G1 ectodomain heavy chains and HLA-G1, -G5 and -G2 transfectants
Because immunohistochemistry raised the problem of different HLA-G5 and -G6 results depending on the histological technique used, we employed the western blot technique for further investigations. To establish a reliable detection system, we analysed rHLA-G1 ectodo-mains and lysates of transfected cell lines with pan-HLA-G mAbs. Examples of these western blot experiments with mAbs 4H84, MEM-G/1 and HCA2 are shown in Figure 3a. As expected, rHLA-A2 was only recognized by mAb HCA2. Neither of the antibodies cross-reacted with rß2m, which served as negative control. To estimate the detection limit of the Western breeze system, which is said to show a greater sensitivity than the HRP-based enhanced chemiluminescence (ECL) systems, we titrated rHLA-G1 and found the corresponding signal down to about 40 pg of total protein per lane (Figure 3a). To demonstrate that the isoforms are distinguishable from each other by their different molecular weights, we used .221 and K562 cell lines transfected with different HLA-G isoforms and found signals at the expected sizes. No reaction was found for the wild-type cell lines (Figure 3b). HLA-G5, which is slightly smaller than HLA-G1, appeared not only in the cells lysates, but also in the CS as secreted product, with identical size. HLA-G1 transfectants were found to release HLA-G fragments smaller than the full-length protein into the medium. This indicates the production of a soluble molecule by shedding or cleavage. K562-G2 transfectants showed a strong band close to 30 kDa, but the respective conditioned media only revealed a signal when applied in a 10-fold concentration (data not shown). Jeg3 lysates contained mainly HLA-G1.
|
Intron 4-containing HLA-G isoforms are not detectable in term villous trophoblasts or in villous mesenchymal cells
Because it has been proposed that the villous cyto- and syncytiotrophoblasts might be a cellular source of the HLA-G5 secretion, which could not be confirmed by our comparative immunolocalization results, we analysed the pan-HLA-G expression pattern of term villous and extravillous trophoblast populations in comparison with the distribution of the intron 4 motif, using highly sensitive western blots. We found pan-HLA-G signals with mAbs MEM-G/1 and 4H84 in the basal plate and chorion laeve of all five placentas; examples are given in (Figure 4a). The pan-HLA-G signals were distributed over a broad array of molecular weights, and therefore, the presence of HLA-G5 could not initially be ruled out by its difference in molecular size. However, deglycosylation of basal plate lysates and HLA-G1 transfectants showed clearly that only unusually glycosylated HLA-G1 was hidden in the broad range of signals (Figure 4a). Pan-HLA-G markers showed, similarly to our negative control cell line Jar, that none of the possible isoforms could be detected in the villous part of any of the five placentas. The respective analysis using mAb 16G1 revealed a specific binding to the intron 4-containing HLA-G5 in transfectants (Figure 4a) and several strongly labelled bands corresponding to sizes present even in lysates of the HLA-G negative control cell line Jar. An 43 kDa band close to HLA-G5 was detectable in all investigated samples, including JAR and .221 wild-type cells (Figure 5b). We did not identify a signal corresponding to the HLA-G5- or -G6-specific molecular weights in any of the five villous samples, and even chorion laeve and basal plate were shown to lack these isoforms.
|
|
Only marginal amounts of HLA-G5 and -G6 transcripts can be detected in term placenta biopsies
To obtain a final answer about HLA-G5 and -G6 expression, we verified our immunological results by RT–PCR experiments. First trials to quantify HLA-G transcripts with nonisotype-specific primers revealed that a high melting point of most of the respective products complicated amplification. Because primer design possibilities are very limited if the different isotypes have to be differentiated, isotype-specific quantification by PCR could not be achieved by direct RT–PCR, although many different conditions and additives were tested. We decided therefore to preamplify HLA-G together with ß-actin transcripts using a modified version of three primed end amplification (Dixon et al., 1998). This method is, after adaptor addition, able to amplify all HLA-G isotype- and ß-actin-transcripts with the same primer set and is therefore very reliable. In comparison with a previously described preamplification protocol (Morales et al., 2003), the low specificity of our protocol, which is due to a nonspecific 3' adaptor primer, has the advantage that so far undescribed 3' splice variants will not be lost and ß-actin may be coamplified. The preamplified material was then used for the quantification of the HLA-G1, -G5 and -G6 isotypes and ß-actin by real-time PCR, whereas ß-actin results were used to check for similar amounts of initial RNA input. Dilution series of DNA from the respective isotypes were amplified in parallel, to determine the isotype abundance independently of the PCR efficiency. The results clearly show that HLA-G1 expression is several-fold higher in first-trimester trophoblasts if compared with all other samples and isotypes. The soluble forms revealed only low abundance of 1.5% and 0.2% for HLA-G5 and -G6, respectively. No HLA-G6 mRNA could be found in basal plate and in villous chorion. In villous chorion, the level of HLA-G5 mRNA (preamplification, 23 PCR cycles) was even below the HLA-C mRNA level (no preamplification). All results of transcript quantification including HLA-C as a control to demonstrate specificity of the preamplification step are shown in Figure 4b. Gel analysis of the PCR products after quantification showed bands at the expected sizes (Figure 4b). Identity of the respective bands were randomly tested by restriction endonuclease digest or sequencing (data not shown).
The soluble pool of HLA-G in vitro
Soluble HLA-G fragments are released from membrane-bound HLA-G1, but no intron 4-containing isoforms are detectable. In situ investigations of term placenta tissues failed to detect any HLA-G5. To resolve conflicting data and to exclude antigen loss by rapid secretion or washing effects during sample preparation, we investigated the composition of the soluble HLA-G fraction in vitro using term and first-trimester trophoblasts. We employed the AC1-M59 cell line as a model for term chorion leave trophoblast. AC1-M59 cells were tested immunocytochemically for their pan-HLA-G expression and were found to be strongly positive, apparently more positive than Jeg3 cells. Compared with term, first-trimester trophoblasts have a much higher viability in culture and were used as primary trophoblast mixture containing all of the trophoblast subpopulations. However, by pan-HLA-G staining, we identified about 60% of the purified trophoblast cells as HLA-G positive EVCT. ELISA analysis of CSs (Figure 5a) of these cells were compared with transfectant-conditioned media and revealed that the 4H84–ELISA has the highest sensitivity for the set of soluble HLA-G produced by first-trimester trophoblasts and AC1-M59 cells, whereas the MEM-G/9-based ELISA was the better tool for testing transfectant CSs. ELISA data revealed considerable levels of soluble HLA-G in trophoblast-conditioned media, especially with the 4H84–MEM-G/1biotin format, but surprisingly this soluble pool could not be attributed mainly to HLA-G5 or -G6 secretion, because only very weak signals, slightly above isotype control, were found in seven-fold–concentrated trophoblast-conditioned media, using 16G1-capture formats. These low signals indicated, if at all, only a marginal secretion of HLA-G5 or -G6 by first-trimester trophoblasts. However, questions on the reliability of the 16G1-based ELISA formats were raised, because we also found signals with media conditioned by the HLA-G negative cell line Jar. The detection limit of our 16G1 ELISAs was approximately 5 ng/ml HLA-G5 (data not shown), whereas we detected rHLA-G in western blots even at concentrations of about 2 ng/ml (Figure 3a). We employed the western blot technique for further analysis of the soluble HLA-G pool generated by first-trimester trophoblasts and the trophoblast-derived AC1-M59 cell line. Analysing lysed cells and their conditioned media with pan HLA-G mAbs 4H84 and MEM-G/1 (Figure 5b, upper panel), we found that first trimester and term trophoblast-derived cells mainly express HLA-G1, which was released into the medium as fragments (HLA-G1shed) smaller than HLA-G5. These fragments were detectable only in concentrated supernatants and led to a more prominent band in media conditioned by AC1-M59 cells. We also found a full-length HLA-G1 band in both of the trophoblast-conditioned media, indicating the presence of membrane vesicles or some cell debris fragments in the samples. Only AC1-M59 but not first-trimester trophoblasts contained an HLA-G2–sized band and again, similarly to term placenta basal plate, only HLA-G1 was identified in lysates of primary mixed trophoblast culture after deglycosylation. The same set of samples labelled with the two intron 4 antibodies 16G1 and 5A6G7 show that neither first-trimester trophoblasts nor the AC1-M59 cell line secreted intron 4-containing HLA-G in vitro (Figure 5b). Even in five-fold–concentrated trophoblast supernatants, HLA-G5- or -G6-related signals were not detectable, whereas the HLA-G5 control resulted in a strong signal. Almost no background was caused by mAbs MEM-G/1 and 5A6G7; only one high molecular weight band due to FCS supplementation was visible in all of our blots. Other than pan HLA-G or 5A6G7 antibodies, 16G1 caused various cross-reacting bands at molecular sizes different from HLA-G5 or -G6, e.g. a 43 kDa band also present in .221 wild-type cells (Figure 5b).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Antibodies
In this study, three HLA-G
1 domain-specific monoclonal antibodies have been used. All of the three pan-HLA-G antibodies, MEM-G/1, 4H84 and HCA2 were tested on equally prepared transfected cell lines for their ability to recognize HLA-G1, -G2 and -G5 under the different methodological settings used in this study. Immunohistochemical staining of placenta serial sections has confirmed again the previously described expression of HLA-G being restricted to the EVCT population throughout placental development (Shorter et al., 1993; McMaster et al., 1995; Blaschitz et al., 2001), whereas no expression could be demonstrated for the villous trophoblasts. At early stages of gestation, the syncytiotrophoblast which directly covers EVCTs was found to be HLA-G positive. This special staining might not surprise, because syncytial growth is dependent on the supply of the underlying cells. Usually these supplying cells resemble the HLA-G negative villous cytotrophoblasts, but at special sites, the so-called cell islands and cell columns, the underlying extravillous HLA-G positive trophoblasts might also support this syncytial growth. Together with previous data, including the well-known antibody W6/32, our immunohistochemical results provide substantial evidence that none of the possible seven isoforms is expressed in the villous trophoblast population, neither cyto- nor syncytiotrophoblasts. This is in contrast to previous findings of Chu et al. (1998), Solier et al. (2002) and Ishitani et al. (2003), who found the villous cyto- and/or syncytiotrophoblast as a cellular source of HLA-G5 secretion. However, these authors used either mAb 16G1 on paraformaldehyde-fixed cryosections (Ishitani et al., 2003) or another intron 4-specific antibody 16D3 and paraffin-embedded tissue sections (Chu et al., 1998). Our data, based on antibody 16G1 obtained with paraffin sections, initially also seemed to support this possibility, but internal tissue controls and results obtained from cryosections revealed fundamental discrepancies. Although HLA-G5-transfected cell lines showed a strong reaction with 16G1 or 5A6G7 even in cryosections, no signals were detectable in cryosections of placenta. These clear-cut results rule out the proposed possibility that soluble antigens might be undetectable in cryosections due to rapid secretion or antigen loss by washing and staining procedures. However, immunohistochemical staining of paraformaldehyde fixed, paraffin embedded tissue sections sometimes results in unspecific staining patterns, which reflect artefacts provoked by the formation of neo epitopes through the fixation, embedding or antigen-retrieval procedures (Boenisch, 2003).
The diversity in HLA-G immunolocalization results published so far might also reflect the uncertain specificity of HLA-G antibodies used, e.g. problems caused by the high affinity of mAb 87G to FcR (Sedlmayr et al., 2002) due to its IgG2a isotype, or mAb BFL.1 which is now unavailable because the specific clone has been lost (Seliger et al., 2003). Recently, also mAb 4H84 was suggested to cross-react with certain classical HLA class I molecules (Polakova et al., 2004) especially after exposing cells to a mild acidic environment (Polakova et al., 2003). However, we could rule out a cross-reaction of mAb 4H84 at least for rHLA-A2, which has the highest homology to HLA-G among the class I HLA molecules.
Using western blot analysis, we were able to demonstrate that besides the specific HLA-G5 protein, various other bands are labelled by intron 4-specific mAb 16G1, even in HLA-G negative control cell lines. Recently Le Friec et al. (2004) proposed an 43 kDa band obtained with 16G1 as new HLA-G5. However, we identified a band of similar size as unspecific labelling, because we found the same signal in .221 wild-type cells and also in Jar cells. In addition, the 16G1–W6/32-based ELISA system showed an HLA-G5 signal in Jar-conditioned media. A weak signal seen with the 16G1–MEM-G1 biotin ELISA in first-trimester trophoblast-conditioned media could not be confirmed by analysing the same samples with the more sensitive western blot. Similarly, paradox 16G1 ELISA results were shown previously by Solier et al. (2002), where the 16G1 ELISA signal was higher than the MEM-G/9 signal when analysing supernatant of trophoblast cultures, although the MEM-G/9-based ELISA format was demonstrated to be definitely the more sensitive approach for the detection of HLA-G5. The specificity of mAb 16G1 is further challenged by data obtained from some of our ELISA experiments using term placenta extracts and a 16G1–TP25.99 biotin assay (data not shown). 16G1 was clearly able to capture proteins, detectable by mAb TP25.99, which binds to the 3 domain of HLA-class I molecules (Desai et al., 2000), but other than W6/32, does not include HLA-G in its class I-binding spectrum. This might suggest an additional 16G1-mediated capture of some classical HLA class I molecules. In this context, the 16G1-based affinity purification of placental HLA-G5 used by Ishitani et al. (2003) must be considered as arguable.
Similar to 16G1, mAb 5A6G7 did not label invasive trophoblasts but bound strongly to UG epithelium. Whether this might really be a hint for the consideration that UGs secrete HLA-G remains to be determined. Facing these problems of antibody specificity, we included antibody MEM-G/1 in this study, an 1 domain-specific pan-HLA-G antibody, which has never been shown to produce any cross-reactions.
Cell and tissue analysis
Using a highly sensitive western blot technique, we demonstrated that none of the villous samples of term placenta contain measurable amounts of HLA-G. This result implies that the villous cyto- and syncytiotrophoblast and the villous mesenchymal core, composed of various foetally derived cells (e.g. fibroblasts, Hofbauer cells, endothelial cells, blood cells and all components of the maternal intervillous space blood) do not express HLA-G. On the protein level, we found HLA-G in term basal plate, chorion laeve and first-trimester mixed trophoblasts, but we did not find HLA-G5 or -G6 in either of the placenta preparations. Initially, lysates of term basal plate, chorion laeve and first-trimester trophoblasts appeared to contain several HLA-G isoforms hidden in a broad band ranging over several molecular weights, but the deglycosylation of the proteins resulted in only one single band corresponding to the HLA-G1 size. We hereby confirmed with antibody MEM-G1 the previous 4H84-based observations by McMaster et al. (1998) that trophoblast HLA-G isoforms are due to unusual glycosylation rather than translation of different transcripts.
Our protein data are directly supported by the quantitative RT–PCR results: marginal amounts of HLA-G1 transcripts were found in term villous chorion samples. It seems more likely that they result from the few remaining cell island EVTCs, which are commonly found there, than from abundant Hofbauer cells or even endothelial cells, as suggested previously (Blaschitz et al., 1997; Chu et al., 1998; Rebmann et al., 1999). The intron 4-containing mRNAs were hardly detectable at all in our placenta and trophoblast samples. Hiby et al. (1999) also observed low HLA-G5- and -G6-transcript levels, but concerning mRNA for membrane-bound molecules, our results differ. We found relatively higher amounts of HLA-G5 mRNA in first-trimester trophoblast samples, maybe because we analysed viable purified trophoblasts, whereas Hiby et al. (1999) used whole first-trimester placenta tissue.
Solier et al. (2002) have proposed that experimental limitations may prevent the detection of the HLA-G5 and -G6 isoforms on the protein level. We have shown here that even the mRNA of HLA-G5 or -G6 was almost absent in all of the analysed placenta and trophoblast samples. Based on this observation, a proposed rapid secretion of substantial amounts of HLA-G5 or -G6 protein by any placental cell is quite implausible.
Trophoblast-conditioned media
Interestingly, the 4H84–MEM-G/1biotin ELISA format detected a comparatively high amount of soluble HLA-G heavy chains in primary trophoblast and AC1-M59-conditioned media, suggesting a possible release of soluble HLA-G2, -G3 and -G4 derivatives as well as HLA-G6 and -G7 by these cells. However, only lysed Jeg3 and AC1-M59 cells, but not primary trophoblasts, revealed a very faint band of the same size as HLA-G2, whereas all of the trophoblastic cells showed the prominent HLA-G1 band. The corresponding conditioned media showed a HLA-G1 band and a band smaller than HLA-G5, which might indicate that our high 4H84–ELISA signals were rather due to an elevated release of soluble HLA-G1 free heavy chains mediated by enzymes such as metalloproteinases (Dong et al., 2003). The high metalloproteinase activity found in invasive trophoblast cells (Bischof et al., 1991, 1995) might explain the increased proteolytic molecule fragmentation (Park et al., 2004).
We provide evidence that the active secretion of intron 4-containing soluble HLA-G by first-trimester trophoblast in vitro is unlikely or below the detection limit of about 600 pg/ml. This confirmed that the lack of HLA-G5 and -G6 secretion by trophoblast-derived cells is also in line with previous observations (McMaster et al., 1998; Ulbrecht et al., 2004). However, if HLA-G5 and -G6 secretion occurred at all in naturally HLA-G–expressing cells, it would only contribute very little to the whole pool of soluble HLA-G, mainly comprising different kinds of HLA-G1 fragments, for which we show in vitro, according to the findings of Park et al. (2004), that they enter the surrounding environment. Without intron 2-specific antibodies, we have not been able to investigate the intron 2-containing HLA-G7 isoform in this study. Therefore we cannot exclude the possibility of HLA-G7 secretion by extravillous trophoblasts, but RT–PCR results (data not shown) make this highly improbable.
|
Conclusions |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Bringing together the results of the parallel approaches in this study, it can be stated that:
- The expression of 1 domain-containing HLA-G in human placenta, including all isoforms described, is definitely restricted to the EVCTs.
- Tissues containing trophoblasts only express the HLA-G1 protein and not the HLA-G5 or -G6 isoforms.
- Only extremely low levels of HLA-G5 and -G6 transcripts can be detected in term placenta and first-trimester trophoblast.
- Freshly isolated first-trimester trophoblasts and term chorion laeve, like AC1-M59 cells, release measurable amounts of shed or cleaved soluble HLA-G1 fragments into the culture media but no soluble forms containing the intron 4-derived peptide.
Implications
Evidence is provided that antibody 16G1 and maybe also other intron 4-binding mAbs might cause misleading results and should be employed only in combination with pan-HLA-G antibodies. In addition, they must be strictly controlled on similarly prepared positive and negative control samples. We show that pan-HLA-G mAb MEM-G/1 can serve as a reliable specific-control antibody, useful for different techniques. The utilization of ELISA systems focusing solely on intron 4-containing soluble HLA-G isoforms (Hunt et al., 2000a) for the diagnosis of pregnancy and placenta-related disorders seems highly questionable. ELISAs including HLA-G1 fragments could be recommended as a supplementing helpful diagnostic tool.
The findings of low HLA-G levels measured in plasma of pregnant women (Rebmann et al., 1999; Hunt et al., 2000a) as compared with nonpregnant female or even male also substantiate our result of villous syncytiotrophoblasts lacking HLA-G5 and -G6 secretion. Nevertheless, a recent study described soluble HLA-G levels up to 4 µg/ml and more in serum of pregnant women, using a novel ELISA approach (Yie et al., 2004). This study does not support these results; in fact it seems unlikely that the endovascular EVTCs in direct contact with maternal blood could shed or cleave so much HLA-G1 into the circulation, especially considering the fact that, in healthy individuals, ubiquitously expressed classical HLA-class I leads to serum levels of only 0.2–3 µg/ml (Grumet et al., 1994).
Whether immunomodulating effects reported for HLA-G5 are also functionally true for the cleaved or shed soluble HLA-G1 fragments, thus contributing to the locally established tolerance in pregnancy (Park et al., 2004), remains to be determined. However, it seems likely as long as these effects are not mediated alone by the intron 4-derived peptide itself. In the light of this study, considering the discussed antibody problems and the fact that mainly HLA-G1 contributes to the soluble pool of HLA-G, further investigations of pregnancy-related body fluids and IVF culture media seem to be necessary.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
We thank Mrs. N. Prutsch for the careful technical assistance in isolating first-trimester trophoblasts. The AC1-M59 cells were kindly provided by H.G. Frank, Aachen, Germany. We thank D. Geraghty, Seattle, WA, USA for providing us with the lymphoblastoid cell line 721.221 (.221) and the transfectants .221-G1 and .221-G5. The K562 cell line was a gift from E. Weiss, Munich, Germany. Recombinant HLA-A2 ectodomain and rß2m were provided by A. Ziegler, Berlin, Germany. This work was supported financially by the Franz Lanyar-Stiftung of the Medical University of Graz, Austria and was part of key research of ‘Reproduction and Pregnancy’ of the Medical University of Graz, Austria. This study was supported in part by the Network of Excellence ‘EMBIC’, sponsored under the 6th Framework Programme of the European Union.
|
Notes |
|---|
* The authors equally contributed to this work.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Bainbridge DR, Ellis SA and Sargent IL (2000a) The short forms of HLA-G are unlikely to play a role in pregnancy because they are not expressed at the cell surface. J Reprod Immunol 47,1–16.[CrossRef][ISI][Medline]
Bainbridge DR, Ellis SA and Sargent IL (2000b) HLA-G suppresses proliferation of CD4 (+) T-lymphocytes. J Reprod Immunol 48,17–26.[CrossRef][ISI][Medline]
Bainbridge D, Ellis S, Le Bouteiller P and Sargent I (2001) HLA-G remains a mystery. Trends Immunol 22,548–552.[CrossRef][ISI][Medline]
Bischof P, Friedli E, Martelli M and Campana A (1991) Expression of extrac