Cytokine regulation of HLA-G expression in human trophoblast cell lines

E L s EV I E R

Journal of Reproductive Immunology 29 (1995) 179- 195

Cytokine regulation of HLA-G expression in human trophoblast cell lines Yaping Yanga, Daniel E. Geraghtyb, Joan S. Hunt*a’c ‘Department of Pathology and Laboratory Medicine, University of Kansas Medical Center, 3901 Rainbow Boulevard, Kansas City, KS 66160-7400, USA ‘Human Immunogenetics Program, Fred Hutchinson Cancer Research Center, Seattle, WA 98104, USA ‘Department of Anatomy and Cell Biology, University of Kansas Medical Center, 3901 Rainbow Boulevard, Kansas City, KS 66160-7400. USA

Received 20 June 1995; revision received 14 August 1995; accepted 16 August 1995

Abstract

HLA class I genes are differentially expressed among subpopulations of cells in first trimester human placentas. In this study, HLA class I protein was detected in extravillous cytotrophoblast cells by immunohistochemistry using the monoclonal antibody W6/32. In the same trophoblast subpopulation, class Ib proteins were identified with two monoclonal antibodies, 87G (anti-HLA-G) and 131 (anti-HLA-A/G) and class Ia protein was detected with the monoclonal antibody, 4E (anti-HLA-B/C). All of the antibodies also identified antigens on the human trophoblastderived choriocarcinoma cell line, JEG-3. Therefore, the JEG-3 cells were used as a model system to study cytokine regulation of HLA-G in trophoblast ceils. Northern blot hybridization studies showed that interferons (IFN-cq IFN-8, IFN-7) and tumor necrosis factor-o (TNF-(r) modestly enhanced steady state levels of HLA-G mRNA. Yet analysis of HLA-G protein by immunocytochemistry and flow cytometry failed to identify any changes in intracellular or membrane expression of HLA-G protein following cytokine treatment. Resistance to upregulation of HLA class I antigens was not a general feature of JEG-3 cells; IFNs enhanced expression of HLA-B/C as well as HLA class 1 light chain, fl2microglobulin. HLA null Jar choriocarcinoma cells did not contain HLA-G mRNA or antigen and exposure to cytokines had no effect on HLA-G. The results of this study are consistent with the postulate that trophoblast cell expression of HLA-G is stringently regulated and is controlled in part by post-transcriptional mechanisms. l Corresponding author, Tel.: (+l-913) 588 7270; Fax: (+l-913) 588-2710; e-mail:jhunt@ kumc.wpo.ukans.edu.

0165-0378/95M9.50 0 1995 Elsevier Science Ireland Ltd. All rights reserved SSDI 0165-0378(95)00942-E

180

Y. Yang et al. /Journal of Reproductive Immunology 29 (1995) 179-195

Cytokines; HLA-C; HLA-G; Human; lmmunohistochemistry; Messenger ribonucleic acid; P@enta; Trophoblast; Tumor necrosis factor

Keywork

Interferons;

1. Introduction

Expression of the major histocompatibility (HLA) antigens is strictly controlled in trophoblast cells that comprise the fetal component of the maternal-fetal interface (reviewed by Hunt and Orr, 1992; Schmidt and Orr, 1993; Geraghty, 1993). Some subpopulations of trophoblast cells are entirely devoid of HLA class I antigens. Others, notably extravillous cytotrophoblast cells found in maternal blood Iacunae and decidua during the first trimester of pregnancy, express the HLA class Ib gene, HLA-G (Ellis et al., 1990; Kovats et al., 1990; Yelavarthi et al., 1991; Chumbley et al., 1993). HLA-G has several novel features that may be useful in pregnancy, including comparatively low polymorphism (Morales et al., 1993; van der Ven and Ober, 1994) and alternative splice sites that yield both soluble and cell bound forms of the antigen (Ishitani and Geraghty, 1992; Kirszenbaum et al., 1994; Fujii et al., 1994). This extravillous cytotrophoblast cell subpopulation has also been reported to express an HLA class Ia gene, HLA-C (Ellis et al., 1989; Shorter et al., 1993). Because of the risk of graft rejection due to maternal immune cell recognition of paternally-derived HLA, it is important to learn how trophoblast cells respond to HLA class I-inducing cytokines. Human placental cytokines include interferons, IFN-CY, IFN-P and IFN-7 as well as tumor necrosis factor-o (TNF-(r) (Bulmer et al., 1990; Chen et al., 1991; Haynes et al., 1993; Paulesu et al., 1994). Of these, only IFNy effects have been extensively assessed. Several studies have shown that subpopulations of trophoblast cells respond differently to this potent modulator; IFN-7 enhances HLA class I antigens in cultures of extravillous cytotrophoblast-like cells isolated from first trimester placentas (Grabowska et al., 1990) but has no effect on antigen expression by trophoblast cells contained within the villi (Hunt et al., 1987). The specific class I loci affected by IFN-y were not identified in these studies and may or may not have included HLA-G. The HLA-G gene contains a 16bp deletion in its interferon consensus sequence (ICS) and its enhancer A region has three base changes from the palindromic sequence present in other class I genes that prevent binding of the transcription factors, KBFl and NFKB (Geraghty et al., 1987; Mendiola et al., 1994). These alterations might confer a measure of stability, preventing interferons from enhancing the rate of transcription of the HLA-G gene. In this study we investigated the potential of uteroplacental cytokines to regulate HLA class I antigen expression in trophoblast cells. Display of HLA

Y. Yang et al. /Awnat

of Reproductive Immunology 29 (1995) 179-195

181

class Ia and Ib antigens by first trimester trophoblast cells and two trophoblastic tumor cell lines, JEG-3 and Jar cells, was first mapped by immunohistology and flow cytometry using a panel of monoclonal antibodies. Subsequently, the effects of cytokines on HLA-G mRNA and HLA class I proteins in JEG-3 and Jar cells were evaluated by northern blot hybridization, immunocytochemistry and flow cytometry. 2. Materials and methods 2.1. Tissues and cell lines

First trimester placentas (8-12 weeks of gestation; n = 3) were obtained from elective pregnancy terminations in cooperation with the Department of Obstetrics and Gynecology. The protocol was approved by the Human Subjects Committee of this institution. Placentas were separated by manual dissection from other components, were dissected into 0.5-1.0 cm3 portions, and were flash frozen in liquid N2 for later sectioning by cryostat. Mouse fibroblasts that had been transfected or not by electroporation with the 6.0-kb Hind111 fragment of genomic DNA containing the HLA-G gene were gifts from B. Koller, University of North Carolina, and H.T. Orr, University of Minnesota. HLA-G transfected Sl4/8 cells and their untransfected counterparts, termed LM cells, were maintained in D-MEM medium containing glutamine, antibiotics, and 10% FCS. JEG-3 human choriocarcinoma cells (HTB-36) were purchased from the American Type Culture Collection (ATCC, Rockville, MD). The Jar human choriocarcinoma cell line was a gift from R.A. Patillo, Medical College of Wisconsin, Madison, WI. JEG-3 and Jar cells were maintained as previously described in RPM1 1640 containing antibiotics, glutamine and 10% FBS (Yang et al., 1993). 2.2. Cytokines and induction experiments

Human recombinant IFN-cr/2B (5.7 x lo5 U/mg) and human Iibroblast IFN-P (1.4 x lo6 U/mg) were purchased from Lee Biomolecular Research Inc., San Diego, CA. Human recombinant IFN-r (2.5 x lo7 Ufmg) and recombinant TNF-CY (1.9 x 10” U/mg) were purchased from Endogen, Boston, MT, and recombinant mouse IFN-7 (1 x 10’ U/mg) was purchased from Genzyme Corporation, Cambridge, MA. For Northern analysis and flow cytometry experiments, subconfluent cell layers in 75 cm2 tissue culture flasks were treated (or not, controls) with media containing species-specific cytokines (IFN-CYand IFN-8, 1000 U/ml; IFN-y, 100 U/ml; TNF-CY,1 U/ml) for 48 h (Hunt et al., 1990). The conditions of dosage and time course experiments done on JEG-3 cells are described in Section 3, Results. The cell layers were washed extensively and harvested by brief trypsinization and were either pelleted and stored at -80°C or were tested for antigen expression by flow cytometry as described below.

182

Y. Yang et al. I Journal of Reproductive Immunology 29 (1995) 179-195

2.3. Mitochondrial enzyme activity assay (MTT assay)

The effects of cytokines on cell lines were assessed as described (Hunt et al., 1990) using a modified MTT assay, where mitochondrial enzyme activity shows a linear relationship with numbers of viable cells (Mossmann, 1983). Cells were seeded into flat-bottom 96-well microplates (S14/8 and LM cells, 4 x lo3 cells/well; JEG-3 cells, 3 x lo3 cells/well; Jar cells, 2.5 x lo3 cells/well) and were treated with cytokines as described above. Medium was removed from the wells and 100 ~1 of 1% MTT in phenol red-free and serumfree RPM1 1640 medium was added to each well. After incubation for 4 h, the MTT solution was carefully removed from the wells. Anhydrous isopropanol(lO0 ~1, Sigma Chemical Co. St. Louis, MO) was added and the plates were shaken vigorously on an orbital shaker. The optical density was read at 570 nm using a Titer-Tek Multiskan microplate reader (Flow Laboratories, Inc., McLean, VA) after blanking on isopropanol. 2.4. Northern blot hybridizations A 450-bp PvuII/PvuII fragment from the 3 ‘-untranslated region of HLAG (HLA-6p.1, a gift from H.T. Orr) was subcloned into the Hind111 site of plasmid pGEM1 (Promega, Madison, MI). Antisense HLA-G cRNA was transcribed from linearized plasmid using T7 RNA polymerase as described dehydrogenase (Yelavarthi et al., 1991). Glyceraldehyde-3-phosphate (G3PDH) mRNA was identified by using a cDNA probe synthesized from a 1.2-kb EcoRI fragment of G3PDH cDNA (R.W. Allen, American Red Cross Blood Services, St. Louis, MO) cloned into pGEM3z plasmid (Promega). Hybridizations with 32P-labeled probes were performed as previously described (Yang et al., 1993; Chen et al., 1994). In brief, total RNA was isolated from samples by extraction with guanidine isothiocyanate (Chomczynski and Sacchi, 1987) and poly(A)+ RNA was prepared using a Mini RiboSepTM Ultra Kit from Becton Dickinson Labware, Bedford, MA. The RNA preparations were subjected to electrophoresis on 1% agarose gels, then were transferred to nitrocellulose filters (Schleicher and Schuell, Keene, NH). The membranes were hybridized with 32P-labeled antisense HLA-G cRNA probe (2 x lo6 counts/min/ml), and were exposed to Kodak XAR film with intensifying screens for one to three days. Subsequently, the membranes were stripped by boiling for 15 min, were rehybridized with the radiolabeled G3PDH cDNA (2 x lo6 counts/mm/ml) and were exposed to X-ray film for 15-30 min. 2.5. Antibodies The mouse monoclonal antibody, W6l32, which identities a combinatorial determinant on HLA class I heavy + light chains (Parham et al., 1979), was generated from hybridoma cells (ATCC, HB95). Antibody was partially

Y, Yang et al. /Journal of Reproductive Immunology 29 (1995) 179-195

183

purified from ascites by ammonium sulfate precipitation. The mouse monoclonal antibody, 87G, which identifies both membrane-bound and soluble forms of HLA-G, was generated and partially purified by D.E. Geraghty (Odum et al., 1992; Manuscript in preparation). Mouse monoclonal antibodies that identify HLA-A (13 1, IgG fraction; Spear et al., 1985) and HLAB/C (4E, hybridoma culture supernatant; Yang et al,, 1984) were generous gifts from J. Kornbluth, University of Arkansas, and S. Yang, Memorial Sloan-Kettering Cancer Institute, respectively. B 1G6, a mouse monoclonal antibody to HLA class I light chain (@2-microglobulin, P2m) was purchased from AMAC, Inc. (Westbard, ME). 2.6. Immunohistochemistry Immunohistochemical staining procedures have been described (Yang et al., 1993). Five-micron thick sections of human first trimester placentas were. sectioned onto glass slides and fixed for 10 min in cold acetone. JEG-3, Jar, S14/8 and LM cells were grown to confluency on Lab-Tek Tissue Culture Chamber slides (Nunc, Inc., Naperville, IL). The chamber assemblies were removed and the cells were fixed for 10 min in 4% paraformaldehyde. The primary antibodies were used at the following concentrations: W6/32, 1:100; 87G, 10 &ml; anti-/32m, 10 &ml; 131, 1:100; 4E, 1:10. In each experiment, an equivalent concentration of normal mouse IgG (Sigma) was substituted for the specific antibody on duplicate tissue sections or cell layers in order to identify any nonspecific reactivity in the visualization system. Binding was detected using an anti-mouse IgG avidin-biotin immunoperoxidase staining kit from Zymed, South San Francisco, CA, which yields a red reaction product at the site of antibody binding. The cell layers were counterstained with Gill’s hematoxylin, overlaid with Crystal Mount (Biomeda, Foster City, CA) to prevent loss of color, dried and coverslipped with Permount (Fisher Scientific, Pittsburgh, PA) for analysis by light microscopy. Intensities of the immunostains on the majority of the cells in each trophoblast subpopulation in tissue sections (Table 1) were estimated by two independent readers as 0, negative; +/O to ++++, weak to strongly positive and the results were averaged. The same system was used to assess the staining intensities of JEG-3 and Jar cells (Table 2). 2.7. Flow cytometry Flow cytometry was performed as described (Hunt and Soares, 1988). Aliquots of 1 x lo6 cells were washed three times in ice cold PBS (pH 7.25) containing 0.02% NaNs, pelleted and incubated at 4°C with 100 ~1 of primary monoclonal antibody or normal mouse IgG for 1 h. W6/32 was used at a 1600 dilution and other monoclonal antibodies were used at 5 pg/ml. After three washes in ice cold PBS, the cells were stained with FITC-

184

Y. Yang et al. /Journal of Reproductive Immunology 29 (1995) I79- I95

Table 1 Summary of immunohistochemical trimester placentas”

Syncytiotrophoblast Villous cytotrophoblast Cytotrophoblast column Proximal to villus Distal to villus

stains for HLA class I antigens on frozen sections of first

W&32

87G

131

4E

mIgGb

0 0

0 0

0 0

0 0

0 0

0 ++++

0 ++++

0 ++++

0 +

0 0

‘Intensities of the innnunostains on the majority of the cells in each subpopulation were estimated as 0, negative; +/O to ++++, weak to strongly positive staining. bW6/32 binds to all HLA class I heavy chains that are associated with light chains (P2m); 87G identifies HLA-G; 131 binds to both HLA-A and HLA-G; 4E recognizes HLA-B and HLA-C. mIgG, normal mouse IgG.

conjugated goat anti-mouse IgG from Vector Laboratories, Burlingame, CA. The cells were washed, pelleted and fixed in 500 ~1 of 1% paraformaldehydePBS for analysis by forward light scatter at 488 nm (EPICS 752, Coulter, Hialeah, FL). Five thousand or more cells were counted in each experiment for each antibody.

Table 2 Summary of immunocytochemical Cells

JEG-3

Jar

Treatment

None +IFN-a +IFN-/I +IFN-y +TNF-(Y None +IFN-(w +IFN-/3 +IFN-y +TNF-u

stains on JEG-3 and Jar human choriocarcinoma

cells”

Antibodyb W6l32

87G

82m

+ ++ +++ +++ +

++ ++ ++ ++ ++

0

0

0

0

0

0

0

0

0

0

++ +++ +++ +++ ++ + + + + +

mIgG 0

0 0 0 0 0 0 0 0 0

“Intensities of the immtmostains were estimated by two independent readers and averaged: 0, negative; + to +++, weakly to strongly positive. bw6/32, anti-HLA class I H + L chain combination; 87G, anti-HLA-G; /32m, anti-fl2rn L chain; mIgG, mouse IgG.

Y. Yang et al. /Journal

of Reproductive Immunology 29 (1995)

179-195

185

3. Results

3.1. Expression of HLA class Ia and class Ib in first trimester trophoblast cells and trophoblast cell lines Immunohistology was used to establish the binding patterns of the monoclonal antibodies used in this study to subpopulations of trophoblast cells in acetone-fixed frozen sections of lirst trimester placentas. The results of the immunohistochemical stains are summarized in Table 1. To a greater or lesser degree, all four monoclonal antibodies bound to cytotrophoblast cells in columns, with the intensity of the staining signals increasing with distance from the villi. Immunostaining patterns obtained by using W6/32 (antiheavy + light chain), 87G (anti-HLA-G) and 131, which is reportedly specific for HLA-A (Spear et al., 1985), were very similar. Binding of 4E (anti-HLAB,-C) to extravillous cytotrophoblastic cells was invariably weaker than binding of other anti-HLA class I reagents. These findings were consistent for the three placentas tested in this study. There was essentially no interexperiment variation when immunohistochemical stains were performed on the same tissues a second time. In order to confirm detection of HLA-G by W6/32 and 87G and to investigate the possibility that 131 or 4E recognized HLA-G, binding of the four monoclonal antibodies to HLA-G-transfected S14/8 cells and untransfected LM cells were tested by flow cytometry. S14/8 cells gave the following results: W6132, peak channel of fluorescence, 145, 99% positive cells; 87G, peak channel, 14599% positive cells; 131, peak channel, 150,99% positive cells; 4E, no binding; normal mouse IgG, no binding. None of the monoclonal antibodies bound to untransfected LM cells. These experiments verified recognition of HLA-G antigens by W6/32 and 87G and lack of recognition by 4E. However, the monoclonal antibody, 131, which is believed to be specific for HLA-A, detected HLA-G on S14/8 mouse libroblasts; 131 is therefore referred to in the following paragraphs as anti-HLA-A/G. The trophoblastic cell lines, JEG-3 and Jar, were tested with the same panel of monoclonal antibodies. As described in detail below, JEG-3 cells bound W6/32, 87G, 131 and 4E as well as anti-/32m whereas Jar cells were positive only with anti-P2m. These lines were then used to analyze regulation of HLA class I in the trophoblast lineage. 3.2. Cytokine effects on steady state levels of HLA-G mRNA in trophoblast cell lines Experiments were done to determine whether or not cytokines influenced steady state levels of HLA-G mRNA in JEG-3 or Jar cells. Fig. 1 shows that the HLA-G antisense cRNA probe identified a doublet of transcripts migrating to approximately 1.9 and 1.8 kb in poly(A)+ RNA obtained from

Y. Yang et al. /Journal

186

JEG-3 z.’ 2

HLAG 1.9 Kb

zg

I 4r

of‘ Reproductive

Immunology

29 (1995)

/ 79- 195

Jar

PlBr

JEG-3

t 4 B EZZZ 00 LL - ‘L l& ?

=

1.8 Kb

G3PDH

control

IFN-a

IFN- B

IFN-7

TNF

Fig. 1. Northern blot hybridization analysis of HLA-G and G3PDH mRNA in JEG-3 and Jar cells treated with 1000 U/ml of IFN-cr, 1000 U/ml of IFN-P, 100 U/ml of IFN--( or 1 U/ml of rTNF-cr for 48 h. Details of the experiments are given in Section 2, Materials and Methods. Panel on the right shows ratios of HLA-G:G3PDH mRNA signals in JEG-3 cells obtained by using a scanning densitometer.

JEG-3 choriocarcinoma cells. Steady state levels of HLA-G mRNA were elevated by treating the JEG-3 cells for 48 h with IFN-a, IFN-6, IFN--y and TNF-a. IFN-/3 was the most effective inducer (84% increase over control). Jar cells did not contain HLA-G mRNA and none was detected following incubation with cytokines (Fig. 1). HLA-G-transfected S14/8 cells and their untransfected counterparts, LM cells, were used to control for specificity of the northern analyses. The HLAG antisense cRNA probe identified a single size of transcript, - 1.9 kb, in samples collected from S14/8 cells whereas LM cells did not contain any RNA sequences that cross-hybridized with the HLA-G cRNA probe (Fig. 2).

Fig. 2. Detection of HLA-G mRNA in HLA-G-transfected S14/8 mouse tibroblasts and untransfected LM cells by northern blot hybridization. Five microgrames of poly(A)+ RNA were loaded into each lane.

Y. Yang et al. /Journal of Reproductive Immunology 29 (1995) 179-195

187

3.3. Cytokine effects on HLA-G expression: analysis by immunocytochemistry Having established that the HLA-G gene is transcribed in JEG-3 cells and that steady state levels of specific message are increased by exposing the cells to cytokines, immunocytochemistry was done to determine whether or not HLA-G protein was produced and whether or not levels differed in cytokinetreated and untreated cells. Table 2 gives a summary of the JEG-3 and Jar immunocytochemical staining results with a panel of monoclonal antibodies. 87G readily identified constitutive HLA-G in both untreated and cytokine-treated JEG-3 cells but the intensity of staining was not altered by treatment with cytokines. W6/32 detected assembled HLA class I antigens in untreated JEG-3 cells, and the intensities of the W6/32 immunostains were elevated by treatment with IFNCX,-0 and -7. This pattern was duplicated when anti-fl2m was used. TNF-ar failed to alter expression of any of the class I antigens. Jar cells contained no detectable HLA class I heavy chain antigen but, as anticipated (Trowsdale et al., 1980), exhibited low levels of /32m. Light chain expression was not elevated by exposure to cytokines, an unexpected observation that was not pursued further. No binding to either JEG-3 or Jar cells was detected when normal mouse IgG was substituted for primary antibody. Trypan blue exclusion studies, visual examination of cell layers and MTT assays performed after exposing the cells for 48 h to 1000 U/ml of IFN-CY, 1000 U/ml of IFN-/3, 100 U/ml of IFN-y or 1 U/ml of TNF-ar, the conditions used in these experiments, indicated that essentially all of the cells were viable (data not shown). Fig. 3 illustrates immunostaining and localization of the antigen detected by 87G in JEG-3 and S14/8 cell layers that had been incubated (or not) with IFN-7. HLA-G antigen expression by JEG-3 cells was not uniform; approximately 40% of the cells were weakly to strongly positive with 87G and 60% appeared negative (Fig. 2~). No differences in the intensity of HLA-G signal were observed after treatment with IFN-y (Fig. 2g). Nor was localization altered; membrane staining was observed in both untreated and IFN-y-treated cells and was most prominent on small, vacuolated cells with condensed nuclei. HLA-G-transfected S14/8 cells used as controls in these experiments exhibited intense staining with 87G whether or not the cells had been exposed to IFN-y (Fig. 2a,e). While it was not possible to detect a change in the intensity of the immunostain in cytokine-treated S14/8 cells, membrane staining appeared to be more prominent. Neither LM (Fig. 2b,f) nor Jar (Fig. 2d,h) cells bound 87G and no binding was detected to any of the cell layers when normal mouse IgG was substituted for 87G. 3.4. Cytokine effects on HLA-G expression: analysis by flow cytometry The results of the experiments reported above showed clearly that JEG-3

188

Y. Yang et al. /Journal of Reproductive Immunology 29 (1995) 179-195

S14/8

LM

JEG-3

Jar

Fig. 3. Immunocytochemical staining of (a,e) Sl4/8 cells, (b,f) LM cells, (c,g) JEG-3 cells, and (d,h) Jar cells that had been incubated in medium (a-d) or in medium containing 100 U/ml of IFN-7 (e-h) for 48 h. The monoclonal antibody to HLA-G, 87G, identified immunoreactivity in Sl4/8 and JEG-3 cells but not in LM or Jar cells. Treatment with IFN-y appeared to enhance membrane staining in S14/8 cells (arrows). Membrane staining was common in both untreated and IFN-y-treated JEG-3 cells, and appeared strongest in vacuolated cells with condensed nuclei (arrows). Inserts show that normal mouse IgG used at the same concentration did not bind to cell monolayers.

cells contain HLA class I proteins and suggested that the class Ib protein, HLA-G, might be regulated differently from other class I antigens synthesized in JEG-3 cells. Flow cytometric analysis was then performed to establish the effects of cytokines on cell membrane expression of HLA-G. Fig. 4 shows that as predicted by the immunocytochemistry results, HLAG was entirely resistant to modulation by cytokines; none enhanced binding of 87G and the results obtained using anti-A/G (13 1) were identical. In dose response studies, levels of surface HLA-G detected by 87G remained unchanged even when the JEG-3 cells were treated with up to 1000 U/ml of IFN-y, 10 000 U/ml of IFN-cr or IFN-@ or 100 U/ml of TNF-o in dose response experiments (data not shown). Time course studies done after 24,48 and 72 h of stimulation with cytokines also failed to reveal any IFNmediated increase in HLA-G antigen expression by JEG-3 cells.

Y. Yang et al. /Journal of Reproductive Immunology 29 (1995)

W6132

67G

P2m

131

4E

189

179-19s

mlg G

Control IFN-a

IFN-P

IFN-Y

TNF- a

Fluorescence

intensity

Fig. 4. Flow cytometry profiles for JEG-3 cells incubated with medium alone (control) or medium containing IFN-o, IFN-8, IFN-7 or TNF-cx for 48 h as described in Section 2, Materials and Methods, then tested for binding of monoclonal antibodies to HLA heavy + light chains (W6/32), HLA-G heavy chains (87G), @2mlight chains, HLA-A/G heavy chains (13 I), HLAB/C heavy chains (4E) and normal mouse IgG (mIgG).

By contrast, all of the interferons enhanced binding of W6/32 and antiHLA-B/C (4E) to JEG-3 cell membranes. These experiments demonstrated that JEG-3 cells are not intrinsically refractory to cytokine induction of HLA class I antigens. TNF-ar had no effect on binding of W6132 or 4E although light chain expression @2m) was enhanced. Thus, the TNF-CYused in the experiments was effective. Normal mouse IgG did not bind to either untreated (Control) or cytokine-treated JEG-3 cells. Flow cytometry experiments were also done on untreated and cytokinetreated Jar choriocarcinoma cells. Anti-HLA class I heavy chain reagents did not bind to Jar cells whether or not the cells had been exposed to cytokines (data not shown). 4. Discussion Considerable new information on the expression of HLA class I antigens in trophoblast cells and the potential of endogenous placental cytokines to modulate expression emerged from the present experiments. Specific class I antigens were first identified in placental cells in situ. As predicted-from previous reports (Ellis et al., 1990; Kovats et al., 1990; Yelavarthi et al., 1991; Chumbley et al., 1994; McMaster et al., 1995), HLA-G was detectable in invasive cytotrophoblastic cells in frozen sections of early gestation placentas.

190

Y. Yang et al. /Journal of Reproductive immunology 29 (1995) 179-195

The results obtained with 87G were confirmed with the monoclonal antibody, 131, which heretofore has been considered specific for the class Ia antigen, HLA-A (Spear et al., 1985). Binding to Sl4/8 cells showed clearly that 131 also recognizes HLA-G, a structurally similar protein (Morales et al., 1993; Messer et al., 1992). The monoclonal antibody, 131, probably did not detect any HLA-A in extravillous trophoblast cells; Redman et al. (1984) reported that monoclonal antibodies to paternal HLA-A alleles do not bind to these cells. Consistent with observations made by using in situ hybridization to detect HLA-G mRNA (Hunt et al., 1991), staining of trophoblast cells in columns increased with distance from the villi, suggesting that HLAG becomes increasingly important as fetal trophoblast cells contact maternal decidual cells. An antibody to HLA-B/C, 4E, bound weakly to the same extravillous trophoblast cells that were HLA-G positive. Studies on Sl4/8 cells showed that 4E does not recognize HLA-G. Although no experiments were done to eliminate the possibility that 4E recognized HLA-B in trophoblast cells, it is more likely that the antibody recognized an HLA-C variant (Ellis et al., 1989; Shorter et al., 1993). HLA-C antigens have several distinguishing characteristics, including poor display on cell surfaces, generation of multiple proteins with different isoelectric points and reduced ability to stimulate immune responses in comparison with HLA-A,-B (Mizuno et al., 1989). Thus, HLA-C expression in trophoblast cells may not pose a risk to semiallogeneic pregnancy, but the antigens might have a unique role at the maternal-fetal interface. Because all of the monoclonal antibodies that recognized extravillous trophoblast cells in tissue sections also bound to JEG-3 choriocarcinoma cells, we considered the line appropriate for investigating cytokine regulation of HLA class I antigens in trophoblast cells. Our results were consistent with those shown by Rinke de Wit et al. (1990) where binding of both W6/32 and 4E was modestly enhanced on IFN-y-treated JEG-3 cells. Studying a cell line has the advantage of avoiding contamination by other types of cells but has the disadvantage of raising the question as to how closely regulation in the tumor cells resembles regulation in normal cells. To substantiate the results reported here further experiments should be done on purified, untransformed trophoblast cells. Major observations made in this study relative to cytokine regulation of trophoblast cell HLA class I gene expression were as follows: (1) cytokines enhance steady state levels of HLA-G mRNA in JEG-3 cells but do not induce specific message in Jar cells, (2) cytokines do not enhance intracellular or cell membrane expression of HLA-G in either JEG-3 or Jar cells, and (3) expression of HLA-G and class Ia antigens are differently regulated in trophoblast cells. Thus, trophoblast cells express both a cytokine-resistant HLA

Y. Yang et al. /Journal of Reproductive Immunology 29 (1995) 179-195

191

class Ib antigen, HLA-G, and an interferon-inducible class Ia antigen that is likely to be HLA-C. Despite the lack of classical Enhancer A and ICS elements in the HLA-G gene, steady state levels of specific message were increased following treatment with cytokines. Rinke de Wit et al. (1990) did not identify this enhancement. However, the conditions used for modulation (2000-5000 U/ml of recombinant IFN-y, 24 h) were different from the conditions in the present study and scanning densitometry was not done to identify subtle alterations. A doublet of HLA-G mRNA was present in JEG-3 cells but not Sl4/8 cells. Splice variants in JEG-3 cells were predicted by the results of Fujii et al. (1994), who observed an additional band in JEG-3 but not in 0.221-G (a HLA-G-transfected lymphoblastoid cell line) cDNA when cr3 and 3’ untranslated primers were used. It was proposed that the second band may have been due to a ll-base insertion in the 3 ’ untranslated region of the JEG-3 sequence that is not found in the 0.221-G sequence. The insertion is 36 bases downstream from the normal splice junction, and could generate a transcript of approximately the size observed in these experiments. Risk and Johnson (1990) reported a single 1.95kb mRNA in BeWo cells but overexposure of the blot may have prevented identification of two closely migrating transcripts. The second band in the JEG-3 cells is probably not an HLA-C message cross-hybridizing with the HLA-G cRNA because the HLA-C mRNA found in BeWo cells is considerably smaller, - 1.5 kb (Ellis et al., 1989). Although interferons had similar if not identical effects on HLA class I in trophoblast cells, TNF-a! was entirely different. Despite reports in the literature of upregulation of both mRNA and class Ia antigens in endothelial cells and fibroblasts (Collins et al., 1986), TNF-ar did not influence either class Ia or Ib heavy chain antigen in trophoblast cell lines. This is consistent with the results of previous studies on rat trophoblast cells, which failed to identify TNF-o-mediated upregulation of class I RTI antigens (Hunt et al., 1990; Roby et al., 1994). A comparatively low dose of TNF-(r (1 U/ml) was used, but dose response experiments failed to show any class I enhancement when the cells were treated with higher concentrations. Light chain was clearly increased by TNF-a, suggesting a heavy chain-specific resistance mechanism. Johnson and Pober (1994) have recently reported that a 40-bp region containing a kappa B-like element is necessary for the response to TNF in transfected HeLa cells and that the variant kappa B found in HLA-C does not respond to TNF. Further investigation is required to uncover the mechanism(s) that permits cytokines to elevate HLA-G mRNA in trophoblast cells while failing to increase intracellular and membrane HLA-G protein. Perhaps, higher levels of soluble HLA-G, which would not be detected by the methods used here,

192

Y. Yang et al. /Journal of Reproductive Immunology 29 (1995) 179-195

was produced and exported. Alternatively, cytokines might affect neither the transcription nor translation of HLA-G in trophoblast cells. The cytokines might diminish message destabilizing elements or enhance stabilizing elements, thus extending the half-life of the message. In summary, our data predict that trophoblast cell HLA-G will remain stable regardless of encounter with HLA class I-enhancing cytokines. This interpretation is supported by the observation that HLA-G expression by invasive cytotrophoblast cells is unchanged in pathological conditions of pregnancy such as preeclampsia, gestation hypertension and intrauterine growth retardation (Colbern et al., 1994). However, confirmation for our prediction requires that levels of cytokines be matched with levels of HLA-G antigen, and this has not been done in situations of pregnancy failure. Stable expression is likely to be of the utmost importance; HLA-G antigen protects HLA null lymphoblastoid cells against lysis by cloned human decidual natural killer cells (Deniz et al., 1994) and the antigen may have other important, nonimmunological functions (reviewed by Hunt and Orr, 1992; Schmidt and Orr, 1993). Our findings also predict that trophoblast cell HLA-C will be.increased by interferons. Such elevation may not disturb maternal-fetal immunological relationships; Heybome et al. (1994) have recently reported that recognition of trophoblast cells by y6 T cells, the major population of TcR+ cells in pregnant uteri, is not MHC-restricted. Acknowledgements This work was supported by grants from the National Institutes of Health to J.S.H. (HD26429) and D.E.G. (AI31874), and by core facilites of the Kansas Mental Retardation Research Center (HDO2528). Y. Yang is the recipient of a predoctoral fellowship from the Kansas Health Foundation, Wichita, KS. The authors thank H.T. Orr for sharing the HLA-G cDNA, B. Koller and H.T. Orr for giving us the HLA-G transfected mouse fibroblast cell lines, J. Kornbluth for the monoclonal antibody 131, and S. Yang for the monoclonal antibody 4E. We appreciate the excellent technical assistance provided by D. Vassmer. References Bulmer, J.N., Morrison, L., Johnson, P.M. and Meyer, A. (1990) lmmunohistochemical localization of interferons in human placental tissues in normal, ectopic and molar pregnancy. Am. J. Reprod. Immunol. 22, 109-116. Chen, H.-L., Kamath, R., Pace, J.L., Russell, SW. and Hunt, J.S. (1994) Expression of the interferon-c receptor gene in mouse placentas is related to stage of gestation and is restricted to specific subpopulations of trophoblast cells. Placenta 15, 109- 12I.

Y. Yang et al. /Journal of Reproductive Immunology 29 (1995) 179-195

193

Chen, H.-L., Yang, Y., Hu, X.-L., Yelavarthi, K.K., Fishback, J.L. and Hunt, J.S. (1991) Tumor necrosis factor-alpha mRNA and protein are present in human placental and uterine cells at early and late stages of gestation. Am. J. Pathol. 139, 327-335. Chomczynski, P. and Sacchi, N. (1987) Single step method of RNA isolation by acid guanidine thiocynate phenol chloroform extraction. Anal. Biochem. 162, 156- 159. Chumbley, G., King, A., Gardner, L., Howlett, S., Holmes, N. and Loke, Y.W. (1994) Generation of an antibody to HLA-G in transgenic mice and demonstration of the tissue reactivity of this antibody. J. Reprod. Immunol. 27, 173-186. Chumbley, G., King, A., Holmes, N. and Loke, Y.W. (1993) In situ hybridization and northem blot demonstration of HLA-G mRNA in human trophoblast populations by locusspecific oligonucleotide. Hum. Immunol. 37, 17-22. Colbem, G.T., Chiang, M.H. and Main, E.K. (1994) Expression of the nonclassic histocompatibility antigen HLA-G by preeclamptic placenta. Am. J. Obstet. Gynecol. 170, 1244-1250. Collins, T., LaPierre, L.A., Fiers, W., Strominger, J.L. and Pober, J.S. (1986) Recombinant human tumor necrosis factor increases mRNA levels and surface expression of HLA-A,B antigens in vascular endothelial cells and dermal fibroblasts in vitro. Proc. Natl. Acad. Sci. USA 83, 446-450. Deniz, G., Christmas, S.E., Brew, R. and Johnson, P.M. (1994) Phenotypic and functional cellular differences between human CD3- decidual and peripheral blood leukocytes. J. Immunol. 152, 4255-4261. Ellis, S.A., Palmer, MS. and McMichael, A.J. (1990) Human trophoblast and the choriocarcinema cell line BeWo express a truncated HLA class I molecule. J. Immunol. 144, 731-735. Ellis, S.A., Strachan, T., Palmer, M.S. and McMichael, A.J. (1989) Complete nucleotide sequence of a unique HLA class 1 C locus product expressed on the human choriocarcinoma cell line BeWo. J. Immunol. 142, 3281-3285. Fujii, T., ishitani, A. and Geraghty, D.E. (1994) A soluble form of the HLA-G antigen is encoded by an mRNA containing intron 4. J. Immunol. 153, 5516-5524. Geraghty, D.E. (1993) Structure of the HLA class I region and expression of its resident genes. Curr. Opin. Immunol. 5, 3-7. Geraghty, D.E., Koller, B.H. and Orr, H.T. (1987) A human major histocompatibility complex class I gene that encodes a protein with a shortened cytoplasmic segment. Proc. Natl. Acad. Sci. USA. 84, 9145-9149. Grabowska, A., Chumbley, G., Carter, N. and Loke, Y.W. (1990) Interferon-c enhances mRNA and surface expression of class I antigen on human extravillous trophoblast. Placenta 11, 301-308. Haynes, M.K., Jackson, L.G., Tuan, R.S., Shepley, K.J. and Smith, J.B. (1993) Cytokine production in first trimester chorionic villi: detection of mRNAs and protein products in situ. Cell. Immunol. 151, 300-308. Heybome, K., Fu, Y.-X., Nelson, A., Farr, A., O’Brien, R. and Born, W. (1994) Recognition of trophoblasts by cd T cells. J. Immunol. 153, 2918-2926. Hunt, J.S. and Orr, H.T. (1992) HLA and maternal-fetal recognition. FASEB J. 6.2344-2348. Hunt, J.S. and Soares, M.J. (1988) Expression of histocompatibility antigens, transferrin receptors, intermediate filaments, and alkaline phosphatase by in vitro cultured rat placental cells and rat placental cells in situ. Placenta 9, 159- 171. Hunt, J-S., Andrews, G.K. and Wood, G.W. (1987) Normal trophoblasts resist induction of class I HLA. J. Immunol. 138, 2481-2487. Hunt, J.S., Hsi, B.-L., King, C.R. and Fishback, J.L. (1991) Detection of class 1 MHC mRNA

194

Y. Yang et al. /Journal

in subpopulations

of Reproductive

Immunology

29 (1995)

179-195

of first trimester cytotrophoblast cells by in situ hybridization. J. Reprod. Immunol. 19, 3 15-323. Hunt, J.S., Atherton, R.A. and Pace, J.L. (1990) Differential responses of rat trophoblast cells and embryonic tibroblasts to cytokines that regulate proliferation and class I MHC antigen expression. J. Immunol. 145, 184-189. Ishitani, A. and Geraghty, D.E. (1992) Alternative splicing of HLA-G transcripts yields proteins with primary structures resembling both class 1 and class II antigens. Proc. Natl. Acad. Sci. USA 89, 3947-3951. Johnson, D.R. and Pober, J.S. (1994) HLA class I heavy chain gene promoter elements mediating synergy between tumor necrosis factor and interferons. Mol. Cell. Biol. 14, 1322-1332. Kirszenbaum, M., Moreau, P., Gluckman, E., Dausset, J. and Carosella, E. (1994) An alternatively spliced form of HLA-G mRNA in human trophoblasts and evidence for the presence of HLA-G transcript in adult lymphocytes. Proc. Natl. Acad. Sci. USA 91, 4209-4213. Kovats, S., Main, E.K., Librach, C., Stubblebine, M., Fisher, S.J. and DeMars, R. (1990) A class I antigen, HLA-G, expressed in human trophoblasts. Science 248, 220-223. McMaster, M.T., Librach, C.L., Zhou, Y., Lim, K.-H., Janatpour, M.J., DeMars, R., Kovats, S., Damsky, C. and Fisher, S.J. (1995) Human placental HLA-G expression is restricted to differentiated cytotrophoblasts. J. Immunol. 154, 3441-3778. Mendiola, J., Zachow, K., Chiu, I. and Orr, H. T. (1994) Role of the enhancer A in the regulation of HLA class 1 gene expression in trophoblast cell lines. Transgenics 1, 147. Messer, G., Zemmour, J., Orr, H.T., Parham, P., Weiss, E.H. and Girdlestone, J. ( 1992) HLAJ, a second inactivated class I HLA gene related to HLA-G and HLA-A. Implications for the evolution of the HLA-A-related genes. J. Immunol. 148, 4043-4053. Mizuno, S., Kang, S.H., Lee, H.W., Trapani, J.A., DuPont, B., and Yang, S.Y. (1989) Isolation and expression of a cDNA clone encoding HLAXw6: unique characteristics of HLAC encoded gene products. Immunogenetics 29, 323-330. Morales, P., Corell, A., Martinez-Laso, J., Martin-Villa, M., Varela, P., Paz-Artal, E., Allende, L. and Arnaiz-Villena, A. (1993) Three new HLA-G alleles and their linkage disequilibria with HLA-A. Immunogenetics 38, 323-331. Mossmann, T. (1983) Rapid calorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays. J. Immunol. Methods 65, 55-63. Odum, N., Ledbetter, J.A., Martin, P., Geraghty, D., Tsu, T., Hansen, J.A. and Gladstone, P. (1992) Homotypic aggregation of human cell lines by HLA class II-, class Ia- and HLAG-specilic monoclonal antibodies. Eur. J. Immunol. 21, 2121-2131. Parham, P., Barnstable, C.J. and Bodmer, W.F. (1979) Use of a monoclonal antibody (W6/32) in structural studies of HLA-A,B,C antigens. J. Immunol. 123, 342-349. Paulesu, L., Romagnoli, R., Cintorino, M., Grazia Ricci, M. and Garotta, G. (1994) First trimester human trophoblast expresses both interferon-c and interferon-c-receptor. J. Reprod. Immunol. 27, 37-48. Redman, C.W., McMichael, A.J., Stirrat, G.M., Sunderland, CA. and Ting, A. (1984) Class 1 major histocompatibility complex antigens on human extra-villous trophoblast. Immunology 52, 457-468. Rinke de Wit, T.F., V&mans, S., Van den Elsen, P.J., Haworth, A. and Stern, P.L. (1990) Differential expression of the HLA class 1 multigene family by human embryonal carcinoma and choriocarcinoma cell lines. J. Immunol. 144, 1080-1087. Risk, J.M. and Johnson, P.M. (1990) Northern blot analysis of HLA-G expression by BeWo human choriocarcinoma cells. J. Reprod. Immunol. 18, 199-203.

Y. Yang et al. /Journal of Reproductive Immunology 29 (1995) 179-195

195

Roby, K.F., Hamlin, G.P., Soares, M.J. and Hunt, J.S. (1994) Responses of phenotypically distinct rat trophoblast cell lines to MHC class I-inducing cytokines. Placenta 15,577-590. Schmidt, C.M. and Orr, H.T. (1993) Maternal/fetal interactions: the role of the MHC class I molecule HLA-G. Crit. Rev. Immunol. 13, 207-224. Shorter, S.C., Starkey, P.M., Ferry, B.L., Clover, L.M., Sargent, IL. and Redman, C.W. (1993) Antigenic heterogeneity of human cytotrophoblast and evidence for the transient expression of MHC class I antigens distinct from HLA-G. Placenta 14, 571-582. Spear, B.T., Kombluth, J., Strominger, J.L. and Wilson, D.B. (1985) Evidence for a shared HLA-A intralocus determinant defined by monoclonal antibody 131. J. Exp. Med. 162, 1802-1810. Trowsdale, J., Travers, P., Bodmer, W.F. and Patillo, R.A. (1980) Expression of HLA-A, -B and -C and b2-microglobulin antigens in human choriocarcinoma cell lines. J. Exp. Med. 152, lls-17s. van der Ven, K. and Ober, C. (1994) HLA-G polymorphisms in African Americans. J. Immunol. 153, 5628-5633. Yang, S.Y ., Morishima, Y., Collins, N.H., Alton, T., Pollack, M.S., Yunis, E.J. and DuPont, B. (1984) Comparison of one-dimensional IEF patterns for serologically detectable HLAA and -B allotypes. Immunogenetics 19, 217-231. Yang, Y., Yelavarthi, K.K., Chen, H.-L., Pace, J.L., Terranova, P.F. and Hunt, J.S. (1993) Molecular, biochemical and functional characteristics of tumor necrosis factor-a produced by human placental cytotrophoblastic cells. J. Immunol. 150, 5614-5624. Yelavarthi, K.K., Fishback, J.L. and Hunt, J.S. (1991) Analysis of HLA-G mRNA in human placental and extraplacental membrane cells by in situ hybridization. J. Immunol. 146, 2847-2854.