Altered phenotype of HLA-G expressing trophoblast and decidual natural killer cells in pathological pregnancies

Peter M. Emmer1, Eric A.P. Steegers2, Harold M.J. Kerstens3, Johan Bulten3, Willianne L.D.M. Nelen2, Kees Boer4 and Irma Joosten1,5

1 Department of Blood Transfusion and Transplantation Immunology, 2 Department of Obstetrics and Gynaecology and 3 Department of Pathology, University Medical Centre Nijmegen, 6500 HB, Nijmegen and 4 Department of Obstetrics and Gynaecology, Academic Medical Centre Amsterdam, 1105 AZ Amsterdam, The Netherlands


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
BACKGROUND: The interaction between decidual natural killer (NK) cells and alloantigens expressed on fetal trophoblast cells are thought to be essential for successful implantation and placentation. Consequently, a disturbed interaction during the first trimester of pregnancy might well lead to a subsequent pregnancy failure. METHODS: We investigated the expression of HLA-G and NK cell markers in tissue sections from recurrent miscarriage (n = 9) and ectopic tubal pregnancies (n = 5), and two hysterectomy specimens of healthy pregnancy as well as decidual biopsies (n = 9) were used as controls. RESULTS: We show in normal pregnancy not only a decrease, but also a morphological change in CD56+ NK cells upon interaction with HLA-G-expressing trophoblasts. The cells appear to be transitioning from a blast-like (activation) state into a state of apoptosis. The number of CD16+ NK cells was low. In contrast, in recurrent miscarriage tissue a sustained NK cell marker expression of both CD56 and CD16 was paralleled by a decreased expression of HLA-G. No morphological changes from the blast-like stage were apparent. Finally, in ectopic pregnancies HLA-G expression in the absence of decidual NK cells was associated with a disturbed trophoblast differentiation. CONCLUSIONS: In pathological pregnancies we show an in-situ altered phenotype of trophoblast and NK cells.

Key words: ectopic pregnancy/HLA-G/human/recurrent miscarriage/trophoblast


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Evidence is accumulating that some idiopathic pregnancy disorders might have an immunological background (Regan, 1992Go). The presence of fetal cells in the maternal circulation (Van Wijk et al., 2001Go) clearly shows that the placenta is not an immunological barrier. Furthermore, expression of paternally MHC derived HLA molecules at the fetal–maternal interface (Houlihan et al., 1995Go; Hutter et al., 1996Go; King et al., 1996Go) can be expected to lead to recognition of the foreign antigens by the maternal immune system. This stresses the need for modulation of the maternal immune response. This modulation of the local immune system might take place in the form of a redirection or suppression of uterine natural killer (NK), and to a lesser extent T cell, responses upon recognition of the paternal antigens.

Of particular interest is the MHC-derived (non-classical) class Ib molecule HLA-G. HLA-G is expressed at the placental interface in particular by the trophoblast cells and it is the major HLA molecule with prolonged and significant levels of protein expression. HLA-G is most likely co-dominantly expressed (Hiby et al., 1999Go) and differs from the classical class I molecules such that it shows low polymorphism at the protein level, a short cytoplasmatic tail, extensive post-transcriptional alternative splicing and protein expression which is highly restricted.

Early studies using HLA-G-specific mRNA probes revealed the presence of HLA-G message in both first trimester villous and extravillous cytotrophoblasts, whereas villous syncitiotrophoblasts were negative (Yelavarthi et al., 1991Go). Staining with antibodies specific for class I molecules confirmed the presence of class I proteins in all extravillous, and absence in villous, trophoblast populations (Shorter et al., 1993Go). Antibodies were generated against various HLA-G epitopes resulting in recognition of overlapping trophoblast populations (McMaster et al., 1995Go; Yang et al., 1996Go; Loke et al., 1997Go; McMaster et al., 1998Go). Reactivity against villous core components appears to be the major difference of the putative HLA-G-specific antibodies. With regard to the restricted expression, it has recently been reported that HLA-G protein might be expressed in malignant tissue (Cabestre et al., 1999Go) and thymus (Mallet et al., 1999Go) as well.

In-vitro studies have shown that the cytotoxic reactivity of uterine NK and T cells against HLA-G expressing target cells of various origin was inhibited (Rouas Freiss et al., 1997Go; Le Gal et al., 1999Go; Rajagopalan and Long, 1999Go). This suggests that the interaction between HLA-G and immunocompetent cells at the placental interface could be critical in determining the outcome of pregnancy. In this regard, it may be helpful to look for deviations in these interactions by studying early pregnancy disorders of unknown aetiology.

It has been estimated that 1.0% of reproducing couples will suffer from recurrent unexplained terminations of early pregnancies (Regan, 1992Go). In case of ectopic pregnancy most implantation occurs in the tubal duct (Solima and Luciano, 1997Go). Both disorders occur during implantation/placentation when trophoblast proliferation and differentiation is most crucial. It is therefore quite conceivable that an altered expression of HLA-G might be associated with such pregnancy failure.

To assess the validity of this hypothesis, we investigated HLA-G protein expression in combination with the distribution of maternal NK cells in tissue obtained from recurrent spontaneous miscarriage and ectopic pregnancy.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Study groups
Tissue samples were obtained from two hysterectomy specimens of otherwise healthy pregnancies at 12 and 13 weeks of menstrual age respectively, diagnosed for cervical carcinoma. These specimens gave additional insight into the HLA-G expression and NK cell distribution in the less superficial layers of decidua. Additionally, decidual biopsies of nine control women were obtained inadvertently as part of a routine procedure of chorionic villous biopsy by means of a flexible bended forceps (2 mm). Initially the biopsies were kept in Hanks' buffered saline solution (Merck, Hohenbrunn, Germany). After light microscopic establishment of the nature of the biopsy, they were rejected for karyotyping and put into formaldehyde.

A group of nine women with a history of recurrent spontaneous miscarriages was studied; a population in part previously described, but without overlap in material studied (Nelen et al., 1997Go, 1998Go; Emmer et al., 2000Go). A miscarriage was defined as a spontaneous pregnancy loss before 16 weeks menstrual age. Only women with a history of at least two consecutive spontaneous idiopathic miscarriages before 16 weeks menstrual age with the same partner were included. Women were excluded in cases of chromosomal rearrangement in either partner, Lupus coagulant, anti-cardiolipin antibodies, or medication used to treat disorders other than vitamin B6 and B12 deficiencies, diabetes mellitus, polycystic ovarian syndrome or thyroid dysfunction. Women received folic acid supplementation (0.5 mg daily) over a period of at least 2 months preceding conception. In these nine women, trophoblastic tissue samples were collected after spontaneous miscarriage or curettage of non-vital pregnancies diagnosed by ultrasonography. These nine women had experienced on average five miscarriages (range three to nine) with the index pregnancy being the most recent pregnancy. The mean menstrual age of this miscarriage tissue was 8.9 weeks (range 4.3–14.6).

We finally obtained tissue from five ectopic tubal pregnancies by means of tubal extirpation at a mean menstrual age of 6.9 weeks (range 5.6–8.3).

The Institutional Review Board of the University Medical Centre Nijmegen approved the study and women gave written informed consent, before participation.

Immunohistochemisty
To reveal HLA-G expression and NK cell distribution in the tissues, we employed an immunohistochemical method using formaldehyde-fixed tissue slides and HLA-G and NK cell-specific monoclonal antibodies. Tissues were dissected and snap-frozen in liquid nitrogen or fixed for 2–4 h in 4% neutral buffered formaldehyde and embedded in paraffin using an automated tissue processor (Tissue-Tek® VIP150; Sakura) under standard conditions for surgical biopsies. Sections of 4 µm were cut, mounted on SuperFrost/Plus glass slides (Menzel-Gläser, Germany) and dried overnight at 37°C. HLA-G expression and NK cell antigens were demonstrated by standard immunohistochemical procedures using antibodies described in Table IGo.


View this table:
[in this window]
[in a new window]
 
Table I. Antibodies used to characterize HLA-G expression and natural killer (NK) cell antigens
 
Paraffin sections were dewaxed in xylol and methanol, and endogenous peroxidase was blocked for 15 min in 1% H2O2 in methanol. This was, depending on the primary antibody used, followed by unmasking of antigens by enzyme digestion with pronase (0.5%) or microwave treatment (buffered 0.01 mol/l citrate solution, pH 6.0). Antibodies were diluted in phosphate-buffered saline containing 0.05% Tween-20 (PBST; Merck, Hohenbrunn, Germany) and 1.0% BSA (PBSTB; Sigma, Steinheim, Germany). All incubations were at 37°C and washings between the incubations were performed in PBST. Preincubating the slides for 10 min in PBSTB reduced background staining. Excess buffer was wiped off and slides were incubated with the primary antibodies for 1 h, followed by incubation with biotinylated horse anti-mouse (1/200; Vector Laboratories, Burlingame, USA) and avidin–biotin–horse-radish peroxidase complex (ABC; Vector Laboratories) for 30 min. Finally, peroxidase was visualized with 0.05% diaminobenzidine (DAB; Sigma)/0.15% H2O2 (Merck) in PBS/0.65% Imidazol (Merck; pH 7.7) for 5 min at room temperature. Specimens were counterstained with haematein and mounted in Permount (Fisher Scientific, Fair Lawn, NJ, USA). To determine the specificity of staining, slides were incubated with isotype-matched negative control antibodies directed against Aspergillus niger glucose oxidase (Dako, Glostrup, Denmark).

We set out to compare the staining pattern of the CAM5.2 antibody, a pan-trophoblast marker to the staining pattern of the HLA-G-specific monoclonal antibodies 4H84, 87G, 16G1 and BFL.1. Because of the extremely small size of the (miscarriage) tissue, we had no choice but to use formaldehyde-fixed preparations. It thus appeared that only the 4H84 antibody gave significant staining of trophoblast cells with acceptable levels of background staining. The 4H84 antibody was induced against a peptide derived from the {alpha}-1 domain and should therefore recognize all HLA-G isoforms (McMaster et al., 1998Go). Staining intensity of the different antibodies was scored independently by two pathologists. The following subjective index was used: –, no staining; +, some cells positive some negative; ++, strongly positive; +++, very strongly positive; absence of cell populations was scored as not present. Digital images were recorded using a Leica Dialux transmission microscope (Leica, Wetzlar, Germany) equipped for digital image capture with a 3CCD colour video camera (DXC-950P; Sony, Tokyo, Japan) and a computer-interfaced frame grabber (Intellicam; Matrox Electronic Systems, Dorval, QC, Canada). To quantify differences in HLA-G expression, the luminosity of the digital images was analysed by means of Adobe® PhotoShop® 5.5 software. In images of serial sections stained with 4H84 and CAM5.2, rectangular selections were superimposed, inverted to negative and the average luminosity determined. Luminosity ratios of 4H84/CAM5.2 were calculated to exclude effects of background staining.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
HLA-G and NK cell expression in normal pregnancy
During early pregnancy, tree-like structures called villi evolve from the embryonic sac and attach to uterine endometrium. Upon contact, the endometrium and stroma are decidualized, the maternal spiral arteries are relayed and fill the inter villous space with blood, thereby forming the haemochorial placenta. At places where the fetal villi anchor to the decidua, a cell column of proliferating cytotrophoblast is formed (Figure 1AGo: frames A and B, area marked `av'). Table IIGo shows the expression of HLA-G as stained by the 4H84 antibody in trophoblast tissue from two hysterectomy preparations, and nine decidual biopsies. In the hysterectomy preparations we observed that these cell columns display a gradient of increasing staining intensity for 4H84 towards the decidua (Figure 1AGo: frames A and B area marked `cc'). Upon contact with the decidua, the cell column cells start differentiating and migrate into the decidua, thereby forming a population of intermediate extravillous trophoblast cells whilst retaining a 4H84 staining intensity similar to that of the most proximal cell column cells. These intermediate trophoblast cells further migrate towards the myometrium and differentiate, thereby forming intramural and endovascular populations. Finally at the junction with the myometrium, differentiation/migration is halted, resulting in the multinucleate placental bed giant cells. All populations of extravillous trophoblast were stained by 4H84 with an intensity similar to the strongest staining cells of the cell column. In contrast to the extravillous trophoblast populations, we did not observe 4H84-specific staining of both villous syncytio- and cytotrophoblast cell populations or any villous core components (Table IIGo). In the nine decidual biopsies, we observed expression of HLA-G and cytokeratin similar to that found in the hysterectomy preparations. We concluded that in the hysterectomy preparations and decidual biopsies the HLA-G expression pattern and levels corresponded with those observed in previous studies on material collected from elective terminations of pregnancy (Loke et al., 1997Go; McMaster et al., 1998Go).



View larger version (140K):
[in this window]
[in a new window]
 
Figure 1. (A) Representative examples of staining for 4H84, CAM5.2, DJ130C and 1B6 in a hysterectomy preparation of an otherwise healthy pregnancy. Frames A, B and C show serial sections of an anchoring villus stained for 4H84 (A and B) and CAM5.2 (C) at x40 (A) and x100 (B, C) magnification. A and B show very strong staining for 4H84 on extravillous trophoblast cells, thereby exceeding staining of CAM5.2 (C). (B) Frames D-L show serial sections stained for 4H84 (D, G, J), DJ130C (E, H, K) and 1B6 (F, I, L) at x250 magnification of respectively trophoblast-free decidua (D, E and F), decidua at the marginal zone of the placental bed (G, H and I), and decidua at the centre of the placental bed (J, K and L). The phenotype of 1B6+ NK cells changes upon contact with 4H84+ extravillous trophoblast (compare frames F and I). 1B6+ NK cells become blast-like (frame I), whereas prolonged contact leads to loss of 1B6 expression while the NK cells reduce in size and with condensed nuclei (frame L). av = anchoring villus; cc = cell column; d = decidua.

 

View this table:
[in this window]
[in a new window]
 
Table II. Expression of trophoblast and natural killer (NK) cell markers in tissue collected from two hysterectomy specimens (H) of early pregnancies and nine decidual biopsies (DB) of pregnancies of normal neonatal outcome
 
Upon decidualization of the endometrium, the extravillous trophoblast cells come in close contact with the maternal NK cells and large mesenchymal cells. To study the expression of NK cell markers and to be able to distinguish peripheral and uterine NK cells, we used antibodies directed against either the CD16 or CD56 antigen (Figure 1BGo). The DJ130c antibody directed against the CD16 antigen strongly stained peripheral NK cells, whereas the 1B6 antibody directed against CD56 weakly stained peripheral NK cells, showing strong staining of decidual NK cells. DJ130c staining was restricted to NK cells found in the proximity of the small blood vessels with a strong staining intensity and sites of focal decidual necrosis. By comparison, DJ130c staining intensity of decidual NK cells was hardly detectable. In the hysterectomy preparations, strong dense staining for 1B6 was mainly found in decidua not yet invaded by trophoblast cells (Figure 1BGo frames D, E, F). Upon contact with invading trophoblast at the marginal zone of the placental bed, NK cells enlarged and 1B6 expression, although more diffuse, seemed to increase (Figure 1BGo frames G, H, I). Finally, 1B6 staining of placental bed NK cells, in prolonged contact with trophoblast, was almost completely absent whereas NK cells reduced in size and the nuclei condensed (Figure 1BGo frames J, K, L). This shift in staining intensity supports the interaction between trophoblast invasion and NK cell distribution and suggests a role for HLA-G. Overall the decidual biopsies showed expression of CD16 and CD56 similar to the hysterectomy preparations (Table IIGo).

HLA-G and NK cell expression in recurrent miscarriage tissue
In all preparations where extravillous trophoblast cells were present, HLA-G was expressed in tissue obtained from recurrent miscarriage (Figure 2Go, Table IIIGo). At places where the fetal villi anchored to the decidua, a cell column of proliferating cytotrophoblast can be seen in which we observed a gradient of increasing staining intensity for 4H84 towards the decidua. Upon contact of the cell column with the decidua, cells migrate into the decidua and form a population of intermediate extravillous trophoblast that differentiate into intramural and endovascular trophoblast cells, all staining for 4H84. Staining of the placental bed giant cells could not be elucidated due to their absence caused by the nature of the miscarriage material. Thus the 4H84 staining pattern of the observed extravillous trophoblast populations in miscarriage tissue resembled that of the hysterectomy preparation. However, when looking at the staining intensity we observed a marked difference. For this purpose we compared in serial sections the ratio of staining intensities of 4H84 to CAM5.2 between both miscarriage and hysterectomy tissue. While the CAM5.2 staining intensity was similar between the preparations (compare Figure 1AGo frame C with Figure 2Go frame C), remarkably, in material obtained after miscarriage, the ratio between 4H84 and CAM5.2 staining at the anchoring site of the villi appeared decreased as compared with the hysterectomy preparation (compare Figure 1AGo staining ratio of frames B and C to Figure 2Go staining ratio frames B and C). We observed this phenomenon in all preparations where anchoring villi were present. In an effort to quantify this decrease, we analysed the luminosity of the (inverted) digital images by means of Adobe PhotoShop 5.5 software. This resulted in a mean luminosity ratio of 4H84/CAM5.2 staining of 0.91 in recurrent miscarriage (n = 6) and 1.11 (n = 11) in hysterectomy preparations and decidual biopsies resulting in a significant difference in the ratio of 0.20 (T-test P < 0.018). This clearly suggests a decreased expression of HLA-G in tissue obtained after recurrent miscarriage.



View larger version (134K):
[in this window]
[in a new window]
 
Figure 2. Representative examples of staining for 4H84, CAM5.2, DJ130C and 1B6 in tissue from recurrent miscarriage. Frames A, B and C show serial sections of an anchoring villus stained for 4H84 (A and B) and CAM5.2 (C) at x40 (A) and x100 (B, C) magnification. Although extravillous trophoblast stains for 4H84, the ratio 4H84/CAM5.2 is decreased as compared with normal pregnancy (compare staining intensity ratio of frames B and C and compare to Figure 1AGo, frame B versus C). Frames D, E and F show serial sections of decidua stained for 4H84 (D), DJ130C (E) and 1B6 (F) at x250 magnification. Both DJ130C and 1B6 show strong staining of NK cells. av = anchoring villus; cc = cell column; d = decidua.

 

View this table:
[in this window]
[in a new window]
 
Table III. Expression of trophoblast and natural killer (NK) cell markers in tissue collected from nine women after recurrent miscarriage at 8.9 weeks (range 4.3-14.6) of gestation
 
With the observation of a decreased expression of HLA-G we were interested whether this could affect the distribution of the maternal NK cells in the decidua (Figure 2Go, frames D, E, F). In the hysterectomy material, DJ130c staining was only observed in secretory endometrium and some isolated areas of focal superficial decidual necrosis. Staining with a granulocyte-specific marker revealed that these areas were infiltrated with granulocytes known for their expression of CD16 (data not shown).

Whilst DJ130c staining of lymphocytes in hysterectomy material was restricted, in miscarriage tissue lymphocytes localized in decidua basalis were abundantly stained (Figure 2Go, frame E). Most surprisingly, in three out of nine tissue samples we also found CD16+ (CD56–) cells that had infiltrated the chorionic villi. Whereas in the hysterectomy preparation 1B6 staining of decidual lymphocytes was low, in contrast, in miscarriage tissue both 1B6 and DJ130c strongly stained the decidual NK cells. This altered NK cell phenotype could point to a disturbed interaction between trophoblast and NK cells and might be caused by the decreased expression of HLA-G.

HLA-G and NK cell expression in ectopic pregnancies
We observed that HLA-G was expressed on all extravillous trophoblast cell populations in ectopic pregnancy (Table IVGo). Concerning expression of HLA-G in ectopic trophoblast, due to the absence of a decidual layer it was difficult to compare the morphology of the `tubal placental bed' with that of an uterine pregnancy (compare Figure 3Go, frame A with Figure 1AGo, frame A). Although we could recognize structures similar to the anchoring villi, we also observed marked differences. Most striking was the observation that pre- and end-stages of trophoblast differentiation seem to be present in close proximity to each other. Multinucleate giant cells, positive for 4H84 (Figure 3Go, frames A and B see arrowheads), resembling the placental bed giant cells in the hysterectomy preparations were found intermingled with the highly proliferative cytotrophoblast of the cell column (Figure 3Go, frames A and B area marked `cc'). In analogy with the placental bed giant cells, these multinucleate giants cells could represent an end stage in trophoblast differentiation. To find these pre- and end-stage trophoblast cells at the same place suggests a deranged regulation of trophoblast differentiation in ectopic pregnancy. Furthermore there were 4H84+ cells that could be identified as endovascular trophoblast cells present in lacunae that resembled blood vessels. This could indicate that although trophoblast differentiation was changed, these cells are in part still able to effect their putative function, possibly through the expression of HLA-G.


View this table:
[in this window]
[in a new window]
 
Table IV. Expression of trophoblast and natural killer (NK) cell markers in tissue collected from five ectopic pregnancies at 8 weeks (range 5.6-8.3) of gestation
 


View larger version (134K):
[in this window]
[in a new window]
 
Figure 3. Representative examples of staining for 4H84, CAM5.2, DJ130C and 1B6 in ectopic tubal pregnancies. Frames A, B and C show serial sections of an anchoring villus stained for 4H84 (A and B) and CAM5.2 (C) at x40 (A) and x100 (B, C) magnification. Clearly present are the giant multinucleate trophoblast cells with varying levels of 4H84 staining (A and B arrowhead). Frames D, E and F show serial sections of the tubal implantation site stained for 4H84 (D), DJ130C (E) and 1B6 (F) at x250 magnification. There is no significant staining of 1B6; DJ130C predominantly stains polymorphonuclear cells. av = anchoring villus; cc = cell column.

 
With the observation that HLA-G was expressed at the interface of trophoblast cells and tubal serosa or mucosa in ectopic pregnancy, we still wondered whether this expression of HLA-G could affect NK cells. Due to the absence of a `real decidua' we expected and indeed found no significant staining for 1B6 and or DJ130c at the transition where trophoblast cells invaded the tubal serosa (Figure 3Go, frames D, E and F). The only expression was found on peripheral NK cells in blood vessels lining the tuba and in the large blood-filled lacunae surrounding the villi. This could indicate that although HLA-G expression is switched on, the presence of a decidual layer is crucial for trophoblast differentiation.


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
In this study we have conclusively shown that HLA-G protein is expressed at the cell surface of extravillous trophoblast cells in tissue from both recurrent miscarriage and ectopic pregnancies. With the observation of HLA-G protein expression, we ruled out the possibility that such abnormal pregnancy outcome could be explained solely by the absence of surface HLA-G protein. In control pregnancies the HLA-G expression pattern and levels corresponded with those found in previous studies on tissue from first trimester elective termination of pregnancies (Loke et al., 1997Go; McMaster et al., 1998Go). Staining of the hysterectomy preparations for the uterine NK cell marker, CD56, showed that expression intensity at the centre of the placental bed in areas with extensive trophoblast invasion was diminished as compared with the marginal zone of the placental bed. Furthermore this decrease was paralleled by a morphological change in which the CD56+ NK cells seem to be transitioning from a blast-like (activation) state into a state of apoptosis upon interaction with HLA-G-expressing trophoblasts. The presence of focal superficial decidual necrosis in the hysterectomy material has been described (Pijnenborg et al., 1980Go) and might be caused by the remodelling of the spiral arteries associated with the redirection of the maternal blood flow. The expression of HLA-G in these areas further illustrates that HLA-G not only seems to play a role in trophoblast invasion, but also might activate maternal immuno-competent cells without compromising pregnancy. This confirms the interaction between NK cells and trophoblast cells that takes place during placentation and suggests a role for HLA-G in this interaction.

Although we observed no differences in the distribution of HLA-G over the different populations of trophoblast cells, we here present data suggesting that HLA-G expression intensity was decreased in trophoblast tissue from first trimester recurrent miscarriage. This decrease in HLA-G protein was paralleled by the unusual presence of CD16+ lymphocytes. Expression of the CD16 marker is limited in healthy late secretory endometrium and during normal early pregnancy (Geiselhart et al., 1995Go; Quenby et al., 1999Go) and therefore the presence of CD16 in miscarriage tissue might indicate a disturbed interaction between NK cells and the semi-allogeneic trophoblast cells. The limited variation of the HLA-G antigen binding site formed by the {alpha}-1 and {alpha}-2 domains suggests that the HLA-G molecule should always be recognized by the putative receptor on maternal effector cells such as NK cells (Hiby et al., 1999Go). Inadequate signalling is most likely due to the absence or expression of sub-threshold levels of HLA-G or its putative receptor. This is in part confirmed by a recent study on the in-vitro interaction between HLA molecules expressed by tumour cells and NK cells, in which tumour cells lacking appropriate HLA class I expression induce infiltration, cytotoxic activation and transcription of IFN-{gamma} in NK cells (Glas et al., 2000Go). Obviously, HLA-G is not the sole ligand for uterine NK cells. HLA-E is another class I molecule expressed on placental tissues, that interacts with these cells (King et al., 2000Go). We have no data on its expression level in the tissues studied but it is clear that its expression is influenced by HLA-G expression levels. The HLA-G leader peptide is considered a very strong binder to HLA-E and stabilizes the molecular complex (Lee et al., 1998Go). Thus HLA-G also indirectly interacts with uterine NK cells via HLA-E. The increase in expression of the CD16 antigen in miscarriage tissue is in concert with a flow cytometric analysis of endometrial biopsies (Lachapelle et al., 1996Go) which found an increase in the CD16+ uterine NK cells in late secretory endometrium of women with a history of recurrent miscarriage. Another immunohistochemical study showed that (non-pregnant) endometrial biopsies from women who subsequently miscarried had elevated numbers of CD16+ cells (Quenby et al., 1999Go). Regarding the CD16 expression in some chorionic villi in miscarriage tissue, it could be argued that this reflects non-specific binding to the Fc{gamma} receptor-bearing Hofbauer cells of fetal origin. We never saw any non-specific staining of the chorionic villi with our (IgG1) control antibodies. This supports our view that these cells are of maternal origin and have infiltrated the fetal villi. An important issue to be resolved in future studies is whether changed HLA-G expression and NK cell phenotype are related to inadequate trophoblast invasion or the result of embryonic/fetal death. However, this will be difficult, as trophoblast function and embryonic/fetal viability are intimately linked (Cross, 2001Go).

The expression of HLA-G in ectopic trophoblastic tissue as described in this study is in tune with previous reports (Loke et al., 1997Go; Goldman-Wohl et al., 2000Go; Rabreau et al., 2000Go) and could indicate that the expression of HLA-G is regulated autonomically. The lack of NK cells at the tubal implantation site as demonstrated by this study and by Proll et al. suggests that recruitment of NK cells is not affected by HLA-G (Proll et al., 2000Go). However, the reported deranged differentiation of trophoblast in ectopic pregnancy is in concert with the unlimited spontaneous trophoblast growth, resulting in tubal rupture. The observation that blood-filled lacunae and endovascular trophoblast are present in tubal pregnancies might indicate (functional) remodelling of tubal blood vessels by trophoblast cells. In this regard it is interesting that, in a recent study, s.c. injection of HLA-G-expressing chorion carcinoma cells into nude mice led to the formation of similar blood-filled lacunae instead of neo-vascularization found after injection of non-trophoblastic tumour cells (Grummer et al., 1999Go). This supports the current view that HLA-G expression on trophoblast cells might play a role in the remodelling of spiral arteries and the establishment of a haemochorial placenta (Colbern et al., 1994Go; Hara et al., 1996Go). Finally, it stresses the need for strict regulation of trophoblast differentiation, presumably imposed by decidual NK cells or stromal cells.

In the present study, we have shown that in normal pregnancy the expression of the common NK cell antigen CD56 was decreased upon interaction with HLA-G-expressing trophoblast cells. In contrast, we have found that a decreased HLA-G expression in trophoblastic tissue obtained from women with a history of recurrent miscarriages was paralleled by an increased expression of both peripheral and uterine NK cell antigens. Finally, we have shown in ectopic pregnancies that expression of HLA-G was switched on in the absence of a decidual layer, associated with disturbed trophoblast differentiation. These findings are in concert with the hypothesis that bi-directional regulation of infiltrating fetal cells and maternal immune response is needed for successful pregnancy. The full extent of this relationship should be further elucidated.


    Acknowledgements
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
We thank Dr A.McMaster and Dr S.Fisher for the kind gift of the 4H84 antibody, Dr D.Geraghty for the kind gift of the 87G and 16G1 antibody, N.J.Hamel-van Bruggen for helping us to organize the intake of recurrent miscarriage patients and G.H.Schuring-Blom for storage of the decidual biopsies. We are grateful to A.Blaschitz for helpful discussion. This work was supported by a grant from the `Zorg Onderzoek Nederland' Foundation (grant no. 2810063).


    Notes
 
5 To whom correspondence should be addressed at: Department of Blood Transfusion and Transplantation Immunology, Geert Grooteplein 10, PO Box 9101, 6500 HB, Nijmegen 603/ABTI, The Netherlands. E-mail: i.joosten{at}utdts.azn.nl Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Cabestre, F.A., Lefebvre, S., Moreau, P., Rouas-Friess, N., Dausset, J., Carosella, E.D. and Paul, P. (1999) HLA-G expression: immune privilege for tumour cells? Semin. Cancer. Biol., 9, 27–36.[ISI][Medline]

Colbern, 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.[ISI][Medline]

Cross, J.C. (2001) Genes regulating embryonic and fetal survival. Theriogenology, 55, 193–207.[ISI][Medline]

Emmer, P.M., Nelen, W.L., Steegers, E.A., Hendriks, J.C., Veerhoek, M. and Joosten, I. (2000) Peripheral natural killer cytotoxicity and CD56+CD16+ cells increase during early pregnancy in women with a history of recurrent spontaneous abortion. Hum. Reprod., 15, 1163–1169.[Abstract/Free Full Text]

Geiselhart, A., Dietl, J., Marzusch, K., Ruck, P., Ruck, M., Horny, H.P., Kaiserling, E. and Handgretinger, R. (1995) Comparative analysis of the immunophenotypes of decidual and peripheral blood large granular lymphocytes and T cells during early human pregnancy. Am. J. Reprod. Immunol., 33, 315–322.[ISI][Medline]

Glas, R., Franksson, L., Une, C., Eloranta, M.L., Ohlen, C., Orn, A. and Karre, K. (2000) Recruitment and activation of natural killer (NK) cells in vivo determined by the target cell phenotype. An adaptive component of NK cell-mediated responses. J. Exp. Med., 191, 129–138.[Abstract/Free Full Text]

Goldman-Wohl, D.S., Ariel, I., Greenfield, C., Hanoch, J. and Yagel, S. (2000) HLA-G expression in extravillous trophoblasts is an intrinsic property of cell differentiation: a lesson learned from ectopic pregnancies. Mol. Hum. Reprod., 6, 535–540.[Abstract/Free Full Text]

Grummer, R., Donner, A. and Winterhager, E. (1999) Characteristic growth of human choriocarcinoma xenografts in nude mice. Placenta, 20, 547–553.[ISI][Medline]

Hara, N., Fujii, T., Yamashita, T., Kozuma, S., Okai, T. and Taketani, Y. (1996) Altered expression of human leukocyte antigen G (HLA-G) on extravillous trophoblasts in preeclampsia: immunohistological demonstration with anti-HLA-G specific antibody `87G' and anti-cytokeratin antibody `CAM5. 2'. Am. J. Reprod. Immunol., 36, 349–358.[ISI][Medline]

Hiby, S.E., King, A., Sharkey, A. and Loke, Y.W. (1999) Molecular studies of trophoblast HLA-G: polymorphism, isoforms, imprinting and expression in preimplantation embryo. Tiss. Antigens, 53, 1–13.[ISI]

Houlihan, J.M., Biro, P.A., Harper, H.M., Jenkinson, H.J. and Holmes, C.H. (1995) The human amnion is a site of MHC class Ib expression: evidence for the expression of HLA-E and HLA-G. J. Immunol., 154, 5665–5674.[Abstract/Free Full Text]

Hutter, H., Hammer, A., Blaschitz, A., Hartmann, M., Ebbesen, P., Dohr, G., Ziegler, A. and Uchanska Ziegler, B. (1996) Expression of HLA class I molecules in human first trimester and term placenta trophoblast. Cell Tiss. Res., 286, 439–447.[ISI][Medline]

King, A., Boocock, C., Sharkey, A.M., Gardner, L., Beretta, A., Siccardi, A.G. and Loke, Y.W. (1996) Evidence for the expression of HLAA-C class I mRNA and protein by human first trimester trophoblast. J. Immunol., 156, 2068–2076.[Abstract]

King, A., Allan, D.S., Bowen, M., Powis, S.J., Joseph, S., Verma, S., Hiby, S.E., McMichael, A.J., Loke, Y.W. and Braud, V.M. (2000) HLA-E is expressed on trophoblast and interacts with CD94/NKG2 receptors on decidual NK cells. Eur. J. Immunol., 30, 1623–1631.[ISI][Medline]

Lachapelle, M.H., Miron, P., Hemmings, R. and Roy, D.C. (1996) Endometrial T, B, and NK cells in patients with recurrent spontaneous abortion. Altered profile and pregnancy outcome. J. Immunol., 156, 4027–4034.[Abstract]

Le Gal, F.A., Riteau, B., Sedlik, C., Khalil-Daher, I., Menier, C., Dausset, J., Guillet, J.G., Carosella, E.D. and Rouas-Freiss, N. (1999) HLA-G-mediated inhibition of antigen-specific cytotoxic T lymphocytes. Int. Immunol., 11, 1351–1356.[Abstract/Free Full Text]

Lee, N., Goodlett, D.R., Ishitani, A., Marquardt, H. and Geraghty, D.E. (1998) HLA-E surface expression depends on binding of TAP-dependent peptides derived from certain HLA class I signal sequences. J. Immunol., 160, 4951–4960.[Abstract/Free Full Text]

Loke, Y.W., King, A., Burrows, T., Gardner, L., Bowen, M., Hiby, S., Howlett, S., Holmes, N. and Jacobs, D. (1997) Evaluation of trophoblast HLA-G antigen with a specific monoclonal antibody. Tiss. Antigens, 50, 135–146.[ISI]

Mallet, V., Fournel, S., Schmitt, C., Campan, A., Lenfant, F. and Le Bouteiller, P. (1999) Primary cultured human thymic epithelial cells express both membrane-bound and soluble HLA-G translated products. [In Process Citation.] J. Reprod. Immunol., 43, 225–234.[ISI][Medline]

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, 3771–3778.[Abstract/Free Full Text]

McMaster, M.T., Zhou, Y., Shorter, S., Kapasi, K., Geraghty, D., Lim, K.H. and Fisher, S. (1998) HLA-G isoforms produced by placental cytotrophoblasts and found in amniotic fluid are due to unusual glycosylation. J. Immunol., 160, 5922–5928.[Abstract/Free Full Text]

Nelen, W.L., Steegers, E.A., Eskes, T.K. and Blom, H.J. (1997) Genetic risk factor for unexplained recurrent early pregnancy loss. Lancet, 350, 861.[ISI][Medline]

Nelen, W.L., Blom, H.J., Thomas, C.M., Steegers, E.A., Boers, G.H. and Eskes, T.K. (1998) Methylenetetrahydrofolate reductase polymorphism affects the change in homocysteine and folate concentrations resulting from low dose folic acid supplementation in women with unexplained recurrent miscarriages. J. Nutr., 128, 1336–1341.[Abstract/Free Full Text]

Pijnenborg, R., Dixon, G., Robertson, W.B. and Brosens, I. (1980) Trophoblastic invasion of human decidua from 8 to 18 weeks of pregnancy. Placenta, 1, 3–19.[ISI][Medline]

Proll, J., Bensussan, A., Goffin, F., Foidart, J.M., Berrebi, A. and Le Bouteiller, P. (2000) Tubal versus uterine placentation: similar HLA-G expressing extravillous cytotrophoblast invasion but different maternal leukocyte recruitment. Tiss. Antigens, 56, 479–491.[ISI]

Quenby, S., Bates, M., Doig, T., Brewster, J., Lewis-Jones, D.I., Johnson, P.M. and Vince, G. (1999) Pre-implantation endometrial leukocytes in women with recurrent miscarriage. Hum. Reprod., 14, 2386–2391.[Abstract/Free Full Text]

Rabreau, M., Rouas-Freiss, N., Landi, M., Le Danff, C. and Carosella, E.D. (2000) HLA-G expression in trophoblast cells is independent of embryonic development. Hum. Immunol., 61, 1108–1112.[ISI][Medline]

Rajagopalan, S. and Long, E.O. (1999) A human histocompatibility leukocyte antigen (HLA)-G-specific receptor expressed on all natural killer cells. J. Exp. Med., 189, 1093–1100.[Abstract/Free Full Text]

Regan, L. (1992) Recurrent early pregnancy failure. Curr. Opin. Obstet. Gynecol., 4, 220–228.[ISI][Medline]

Rouas Freiss, N., Goncalves, R.M., Menier, C., Dausset, J. and Carosella, E.D. (1997) Direct evidence to support the role of HLA-G in protecting the fetus from maternal uterine natural killer cytolysis. Proc. Natl Acad. Sci. USA, 94, 11520–11525.[Abstract/Free Full Text]

Shorter, S.C., Starkey, P.M., Ferry, B.L., Clover, L.M., Sargent, I.L. 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.[ISI][Medline]

Solima, E. and Luciano, A.A. (1997) Ectopic pregnancy. Ann. NY Acad. Sci., 828, 300–315.[ISI][Medline]

Van Wijk, I.J., Griffioen, S., Tjoa, M.L., Mulders, M.A., Van Vugt, J.M., Loke, Y.W. and Oudejans, C.B. (2001) HLA-G expression in trophoblast cells circulating in maternal peripheral blood during early pregnancy. Am. J. Obstet. Gynecol., 184, 991–997.[ISI][Medline]

Yang, Y., Chu, W., Geraghty, D.E. and Hunt, J.S. (1996) Expression of HLA-G in human mononuclear phagocytes and selective induction by IFN-gamma. J. Immunol., 156, 4224–4231.[Abstract]

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.[Abstract/Free Full Text]

Submitted on February 5, 2001; resubmitted on September 17, 2001; accepted on November 29, 2001.