No evidence for apoptosis of decidual leucocytes in normal and molar pregnancy: implications for immune privilege

Abstract

Complete hydatidiform moles are totally paternally derived and represent complete allografts that might be expected to provoke maternal immune rejection. Our previous and other studies have shown expression of Fas by increased numbers of activated decidual CD4⁺ T cells in both complete and partial molar pregnancy as well as increased FasL⁺ expression by molar trophoblasts compared with trophoblasts in normal pregnancies. As the Fas/FasL system represents a major apoptotic pathway that can play a role in immune privilege, the aim of this study was to investigate whether apoptosis of decidual immune cells, particularly T cells, could be responsible for maternal immune tolerance in molar pregnancy. Using terminal deoxynucleotidyl transferase (TdT)-mediated nick end-labelling (TUNEL), a significant increase in TUNEL⁺ cells was demonstrated in decidua associated with partial (P = 0·0052) and complete (P = 0·0096) hydatidiform mole compared with normal early pregnancy. Co-labelling immunoperoxidase studies showed that the TUNEL⁺ cells in both normal and molar pregnancies were not activated CD45RO⁺ immune cells, CD3⁺ T cells, CD56⁺ uterine natural killer (NK) cells or CD14⁺ CD68⁺ macrophages. Double immunohistochemical labelling with antiactive caspase-3 and leucocyte markers confirmed the lack of leucocyte apoptosis. Double immunostaining with anticytokeratin to detect trophoblast and M30 CytoDeath, which detects a neoepitope of cytokeratin 18 revealed after caspase-mediated cleavage, revealed apoptotic extravillous trophoblast cells within decidual tissue. We conclude that there is no evidence that apoptosis of decidual leucocytes plays a role in maintaining maternal tolerance in either normal or molar pregnancy.

INTRODUCTION

Hydatidiform mole is a benign gestational trophoblastic disease in which placental trophoblast exhibits excessive proliferation. One in 10 molar pregnancies will progress to some form of persistent gestational trophoblastic disease, including invasive mole and choriocarcinoma [1–3]. Complete hydatidiform moles are usually diploid with a 46, XX karyotype and are of androgenetic origin [4–6]. Partial hydatidiform moles, in contrast, are generally triploid with an excess set of paternal chromosomes resulting from the fertilization of a normal egg by two sperms [7,8]. Because complete hydatidiform moles are totally paternally derived, they represent a complete intrauterine allograft within the mother and thus may be expected to stimulate maternal immune responses leading to fetal rejection. Partial moles, with their excess set of paternal genes, may also stimulate an altered immune response compared to normal pregnancy. We have reported previously an increase in CD45RO⁺ activated/memory CD4⁺ T cells in maternal decidua in partial and complete hydatidiform mole compared with normal early pregnancy, suggesting an altered maternal immune response to molar trophoblast [9] but their precise specificity is unknown. Recently other workers have detected increased numbers of decidual CD3⁺, CD8⁺ T cells in molar pregnancy [10]. However, molar trophoblast survives without provoking adverse maternal immunity and the mechanisms underlying this immune privilege remain elusive.

Apoptosis, or programmed cell death, is recognized as a key mechanism in development and homeostasis and apoptosis of leucocytes has been proposed as a mechanism for maintaining immune privilege [11]. Fas ligand (CD95 ligand/FasL), a Type II transmembrane trimeric protein belonging to the tumour necrosis factor (TNF) family [12], induces apoptosis of activated immune cells that express its receptor Fas [11]. Fas-mediated apoptosis may play an integral role in the survival of semi-allogeneic fetal trophoblast [13–21]. In complete hydatidiform mole, trophoblast cells up-regulate FasL expression [22] and CD45RO⁺ decidual activated/memory CD4⁺ T cells show increased Fas expression compared with normal pregnancy [23]. Apoptosis of activated Fas⁺ CD4₊ T cells in decidua could be mediated by FasL⁺ molar trophoblast, thereby facilitating survival of the molar pregnancy. Others have reported apoptosis of decidual CD45⁺ leucocytes in normal early pregnancy [24], possibly Fas⁺ CD3⁺ T cells [16]. The aim of this study was to compare apoptosis in decidua from complete and partial hydatidiform moles with that from normal early pregnancy.

MATERIALS AND METHODS

Tissues

Formalin-fixed paraffin-embedded decidual tissues from 12 gestationally matched normal first-trimester pregnancy terminations (8–12 weeks from last menstrual period), 10 complete and 18 partial hydatidiform moles were retrieved from archive files of the Department of Pathology, Royal Victoria Infirmary (RVI), University of Newcastle upon Tyne, UK and the Anatomical Pathology Department, Buddachinaraj Hospital, Phitsanulok, Thailand.

Detection of apoptotic cells

Single TUNEL labelling

The terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick end-labelling (TUNEL) assay (Apoptosis Detection System, fluorescein: Promega, Southampton, UK) was carried out following the manufacturer's protocol for formalin-fixed paraffin-embedded tissues. The sections were mounted using Vectashield mounting medium with propidium iodide (Vector Laboratories, Peterborough, UK) to lightly counterstain the nuclei and covered with glass coverslips. The sections were stored at 4°C in the dark and were analysed and photographed within 24 h.

Two positive controls were included: a section of poorly differentiated endometrial carcinoma showing histological evidence of extensive apoptosis and a section of normal early pregnancy decidua treated with deoxyribonuclease (DNase) Type I to produce fragmentation of chromosomal DNA. The DNase positive control was washed separately to prevent any residual DNase activity affecting other sections in the assay. Negative controls were performed for each test section by replacing 1 µl of TdT enzyme in the TdT incubation solution with 1 µl distilled water in order to assess both non-specific binding and the extent of autofluorescence within the tissue section.

Characterization of apoptotic cells

Single immunohistochemical labelling

Single immunohistochemical labelling using monoclonal antibodies (MoAb) against cytokeratin to detect trophoblast, leucocyte common antigen (CD45), which reacts with all leucocyte populations, and M30 CytoDeath, which detects a neoepitope of cytokeratin 18 revealed after caspase-mediated cleavage [25] was performed using an avidin–biotin peroxidase complex (ABC) method (Vectastain Elite kit; Vector Laboratories, Peterborough, UK) as described previously [9]. Briefly, after deparaffinization, rehydration, appropriate pretreatment (Table 1) and quenching of endogenous peroxidase activity, non-specific staining was blocked by incubation with the supplied horse IgG for 10 min. The sections were then incubated sequentially with primary MoAb for 60 min (Table 1), biotinylated horse antimouse immunoglobulins for 30 min and avidin–biotin peroxidase complex for 30 min. The reaction was developed with DAB (3,3′-diaminobenzidine substrate: Sigma Chemical Co., Poole UK) containing 0·1% hydrogen peroxide to give a brown reaction product. Positive tissue controls for each MoAb were included in each staining run and negative controls in which the primary antibody was replaced with non-immune mouse IgG were performed for each sample.

Table 1.

Primary antibodies

Antibody1	Source	Specificity	Dilution	Pretreatment
Active caspase-32	R&D Systems3	p17 subunit of caspase-3	1/400	None
M30 CytoDeath (clone M30)	Roche4	Caspase cleaved epitope of the human cytokeratin 18 cytoskeleton protein	1/200	Trypsin 10 min
Cytokeratin (cytokeratin 8/18; NCL-5D3 and cytokeratin 5/6/18; NCL-LP34)	Novocastra5	Simple and glandularepithelium	[50NCL-5D3 + 12·5NCL-LP 34]/1000	Trypsin 10 min
LCA (NCL-LCA-RP)	Novocastra	Leucocytes	1/100	Trypsin 5 min
CD3 (NCL-CD3-PS1)	Novocastra	T cells	1/50	Pressure cooker, citrate buffer pH 6·0
CD14 (NCL-CD14)	Novocastra	Monocytes, macrophages	1/20	Pressure cooker, citrate buffer pH 6·0
CD45RO (NCL-UCHL1)	Novocastra	Primed/memory T cells, monocytes, macrophages	1/400	Pressure cooker, citrate buffer pH 6·0
CD56 (NCL-CD56–1B6)	Novocastra	NK cells, a subset of activated T cells	1/500	Pressure cooker, citrate buffer pH 6·0
CD68 (KP1)	Dako6	Macrophages	1/50	Trypsin 10 min

¹All antibodies are mouse MoAb unless otherwise stated.

²Rabbit polyclonal antibody.

³R&D Systems Inc., MN, USA.

⁴Roche Applied Science, Mannheim, Germany.

⁵Novocastra Laboratories, Newcastle upon Tyne, UK.

⁶Dako, Ely, UK.

As the TUNEL reaction is unable to discriminate apoptotic from necrotic cells, active caspase-3, a marker of early apoptosis, was localized using a rabbit polyclonal antibody (pAb; Table 1) and a streptavidin–biotin complex method. After deparaffinization, rehydration and quenching of endogenous peroxidase activity, non-specific labelling was blocked by incubating sections overnight at 4°C with 10% normal swine serum (NSS) diluted in TBS. The sections were then incubated sequentially with anti-active caspase-3 for 45 min, biotinylated swine antirabbit immunoglobulins (Dako; diluted 1/500 in NSS) for 30 min followed by streptavidin–biotin peroxidase complex (Dako; 1/100 in NSS) for 30 min. The reaction was developed with DAB to give a brown reaction product. A negative control was performed for each test section by replacing the primary antibody with non-immune rabbit IgG. Tonsil was used as the positive control.

Combined immunoperoxidase and TUNEL

In order to investigate the identity of the TUNEL⁺ cells in the decidua, a combined labelling technique with an ABC method to stain leucocytes followed by TUNEL staining was designed. Antibodies against leucocyte antigens (CD3, CD14, CD56, CD68) and the activation molecule CD45RO were used to demonstrate decidual leucocyte subpopulations (Table 1).

First, decidual leucocytes were labelled by the ABC method and the reaction was developed with DAB. Positive and negative controls were included as described above. After washing in phosphate buffered saline (PBS) the TUNEL assay was then performed as described above.

Double immunohistochemical labelling

In order to characterize further the apoptotic cells, anti-active caspase-3 was used in combination with anti-CD3, anti-CD14 and anti-CD56 antibodies to detect T cells, macrophages and natural killer (NK) cells, respectively. Double-labelling was performed using a streptavidin–biotin peroxidase complex method for active caspase-3 staining. After the reaction was developed with DAB, sections were double-labelled for CD3, CD14 or CD56 using an ABC method as described above except that the reaction was developed with Vector VIP peroxidase substrate (Vector Laboratories) to give a purple reaction product.

To resolve the identity of the apoptotic cells within decidua double-labelling was performed using anticytokeratin and M30 CytoDeath. Initially sections were labelled with anticytokeratin using the ABC method and developed with DAB. M30 was detected using an alkaline phosphatase kit (Vector Laboratories) and developed with Vector Blue (Vector Laboratories) to give a blue reaction product. Endogenous alkaline phosphatase was inhibited using levamisole solution (Vector Laboratories) according to the manufacturer's recommendations.

Quantification and statistical analysis

TUNEL single-labelling

For assessment of apoptosis, slides were analysed under ultraviolet (UV) light with appropriate wavelength filters to visualize green (fragmented DNA) and red (propidium iodide nuclear counterstain) fluorescence. The number of TUNEL⁺ cells in the decidual stroma was counted in 15 random high-power (×400) fields. Statistical analysis was performed using the non-parametric Mann–Whitney test with P < 0·05 as the limit of significant differences between two groups.

Combined ABC/TUNEL labelling

Positive cells were assessed in the whole tissue section. Leucocytes (CD45) and CD45RO positive cells were visualized using light microscopy and photographs were taken using an Olympus EX60 video image grabber. Without changing the microscopic field, TUNEL⁺ cells were then visualized with UV light and photographs of the same field were taken. The two image files of the same microscope field were processed using Adobe Photoshop to allow simultaneous visualization of DAB- and TUNEL⁺ cells. Double-positive cells were identified as bright red (DAB⁺) and bright green (TUNEL⁺) staining on the same cells. Single-positive cells were either bright red or bright green.

Double immunohistochemical labelling

Double-positive cells were identified as brown (DAB; active caspase-3 or cytokeratin) and either purple (Vector VIP; CD3, CD14 or CD56) or blue (Vector blue; M30) staining on the same cells. Single-positive cells were brown (active caspase-3, cytokeratin), purple (CD3, CD14, CD56) or blue (M30) cells.

RESULTS

Single TUNEL staining

Both positive controls, DNAse-treated decidua and poorly differentiated endometrial carcinoma, showed numerous TUNEL⁺ cells (Fig. 1a,b). Negative controls showed no labelling (Fig. 1c). A number of TUNEL⁺ cells were detected in decidual stroma in normal early pregnancy (Fig. 1d), partial (Fig. 1e) and complete hydatidiform (Fig. 1f) mole.

Fig. 1.

(a–c) TUNEL⁺ controls. (a, DNase-treated decidua; b, endometrial carcinoma) show numerous green TUNEL⁺ cells. Negative control normal decidua shows no fluorescence (c). (d–e) TUNEL labelling of decidua from normal early pregnancy (d), partial (e) and complete hydatidiform mole (f) shows increased TUNEL⁺ cells (arrows) in molar compared with normal pregnancy. (g–j) Combined immunoperoxidase/TUNEL labelling of decidua from complete hydatidiform mole using light (g, i) and fluorescence (h, j) microscopy. Thin arrows indicate single-labelled CD45RO⁺ (g, h) or CD3⁺ (i, j) cells and thick arrows indicate TUNEL⁺ cells. Note the lack of double-positive cells. (k, l, n, o) Consecutive sections of normal early pregnancy decidua labelled for active caspase-3 (k), LCA (l), M30 (n) and cytokeratin (o). The distribution of caspase-3⁺ cells differs from LCA⁺ cells but the distribution of M30⁺ cells and cytokeratin⁺ cells is similar to that of caspase 3⁺ cells. (m, p) Double-labelling of formalin-fixed paraffin-embedded sections of complete hydatidiform mole for active caspase-3 and CD3 (m) and cytokeratin and M30 Cytodeath (p) reveals single-labelled caspase-3⁺ (brown) and CD3⁺ (purple) cells (arrows) but no double-positive cells. In contrast, double-labelled cytokeratin⁺/M30⁺ trophoblast cells (dark blue–brown) are seen easily (thick arrows). Thin arrows indicate single-labelled cytokeratin⁺ extravillous trophoblast cells (brown). Original magnification: ×200 (1 a−1f, 1k, 1 l, 1n, 1o); ×400 (1g−1j, 1 m, 1p).

The quantitative analysis of TUNEL⁺ cells in decidua associated with normal early and molar pregnancies is shown in Fig. 2. In normal decidua there were scanty individual TUNEL⁺ cells, sometimes localized adjacent to endometrial glands. In contrast there were significantly increased numbers of TUNEL⁺ cells per field in partial (median, range: 2·93, 0·73–10·00; P = 0·0052) and complete mole (3·80, 0·73–7·46; P = 0·0096) compared with normal pregnancy (0·80, 0·13–4·00).

Fig. 2.

TUNEL⁺ cells in decidua from normal early pregnancy (NEP, n = 12), partial (PHM, n = 18) and complete hydatidiform mole (CHM, n = 10). Each point represents an individual sample and the horizontal bars are median values.

Combined ABC and TUNEL staining

Negative controls for both avidin–biotin peroxidase and TUNEL methods showed no non-specific staining. In both normal and molar pregnancy decidua, combined labelling demonstrated clearly that none of the CD3⁺, CD14⁺, CD56⁺, CD68⁺ or CD45RO⁺ cells were also TUNEL⁺ (Fig. 1g–j). As there were no double-positive cells, quantification was not performed.

Immunostaining for active caspase-3 and leucocyte populations

The findings were similar for normal and molar pregnancy. Single-labelling for active caspase-3 showed positive cells within decidual stroma but LCA⁺ cells on consecutive sections showed different localization (Fig. 1k,l). Double-labelling for active caspase-3 with CD3, CD14 or CD56 revealed no double-labelling (Fig. 1m). It was not possible to double-label for LCA and active caspase-3 due to high levels of non-specific background staining.

Immunostaining for cytokeratin and M30 cytoDeath

Single-labelling for cytokeratin demonstrated extravillous trophoblast cells dispersed within decidua. M30 immunostaining on consecutive sections localized M30⁺ cells in the same areas as cytokeratin⁺ cells (Fig. 1n,o). Double-labelling for cytokeratin and M30 demonstrated substantial numbers of M30⁺ cytokeratin⁺ extravillous trophoblast cells within decidua from both normal pregnancy and hydatidiform mole (Fig. 1p).

DISCUSSION

A substantial number of leucocytes are present in endometrium throughout the menstrual cycle and in early pregnancy; their numbers increase dramatically in the late secretory phase and in the first trimester of pregnancy, accounting for at least 20% and 30% of stromal cells, respectively [26]. There are three major leucocyte populations in early pregnancy decidua: CD3⁺ T lymphocytes, CD56⁺ bright CD16^- uterine NK cells (endometrial granulated lymphocytes) and CD68⁺ CD14⁺ macrophages [27]. The present study assessed whether apoptosis of decidual leucocytes is likely to play a role in survival of molar trophoblast. There was a significant increase in the number of TUNEL⁺ cells in decidua in partial and complete hydatidiform mole compared with normal early pregnancy, suggesting an altered maternal response to the molar pregnancy. Co-labelling, however, demonstrated unequivocally that none of the TUNEL⁺ cells from either normal or molar pregnancy were T cells (CD3⁺), macrophages (CD14⁺, CD68⁺), uterine NK cells (CD56⁺) or activated leucocytes (CD45RO⁺). This lack of apoptosis was confirmed by double-labelling of various leucocyte subsets for active caspase-3.

The lack of apoptotic leucocytes in normal and molar pregnancy concurs partially with another study that failed to detect apoptosis of isolated decidual T cells from normal pregnancy using gel electrophoresis to detect DNA fragmentation [16]. Our findings, however, conflict with a report that apoptotic cells in decidua are CD45⁺ leucocytes [24]. The latter authors noted the tendency of apoptotic leucocytes to down-regulate expression of CD antigens and this would be the simplest explanation for our failure to detect apoptotic leucocytes. However, the combined immunoperoxidase/TUNEL assay used in the present study has the advantage of heat pretreatment which increases the sensitivity of the TUNEL assay with no loss of specificity [28,29]. Moreover, the failure to detect apoptotic leucocytes by TUNEL reactivity was strengthened by the lack of double-labelling for caspase-3 activity, a marker of early rather than late apoptosis.

It has been proposed that clonal deletion of antipaternal alloantigenic-specific maternal immune cells establishes immune privilege during pregnancy [30] and this pregnancy state restricted tolerance may be mediated by the Fas/FasL-mediated apoptotic pathway operating within the placenta [13,17]. However, pancreatic islets studies have shown that membrane-associated FasL expression in vivo promotes allograft rejection rather than immune privilege [31]. Recently Abrahams et al. [32] reported that isolated first-trimester trophoblast cells lack membranal FasL but rather express a cytoplasmic form, which is associated with a specialized lysosomal pathway and is secreted constitutively. After the disruption of these microvesicles, secreted FasL can induce T cell apoptosis in vitro through activation of the Fas pathway. The present study indicates that Fas-mediated apoptosis of maternal leucocytes plays no role in the survival of molar trophoblast in vivo and therefore immune privilege is not likely to be critical for survival of molar trophoblast in vivo. Others have also questioned the contribution of the Fas/FasL system towards fetal survival [33].

A striking finding in this study is the demonstration, using double-labelling for M30 CytoDeath and cytokeratin, that apoptotic cells in decidua from both normal and molar pregnancy are extravillous trophoblast cells. This finding is in accord with other reports of apoptosis of extravillous trophoblast [34–36]. It is likely that apoptosis of extravillous trophoblast is part of a controlled mechanism for trophoblast proliferation and invasion during normal pregnancy which is perturbed in molar pregnancy. In addition, based on morphological criteria, at least some of the apoptotic cells were very likely to be typical decidualized stromal cells, although the lack of a specific marker for decidualized stromal cells limited our ability to demonstrate this unequivocally. Although they express CD10 in vitro[36,37], expression of CD10 by decidual stromal cells in situ is much less consistent, with many cells being negative or focally positive, unlike non-decidualized endometrial stromal cells which are consistently CD10⁺[38,39]. Future studies using insulin-like growth factor binding protein (IGFBP-1), another marker of decidualization [40], may be possible but it should be noted that the IGFBP-1 status of decidua in molar pregnancy is unknown.

The mechanisms that protect the increased activated decidual T cells in partial and complete hydatidiform mole [9] from apoptosis mediated by FasL⁺ molar trophoblast remain unknown. Inhibition of Fas-mediated apoptosis can occur through the anti-apoptotic protein, FLIP, which inhibits the action of a protease enzyme caspase 8 or FLICE [41–44], but to date no data are available regarding their expression in human pregnancy. The function of the activated decidual T cells in hydatidiform mole is not known. Decidual T cells can produce cytokines including interleukin (IL)-3, IL-4, transforming growth factor (TGF)-β₁, TNF-α and interferon (IFN)-γ[42–45] and it has been suggested that deficient production by decidual T cells of cytokines, such as IL-4 and LIF, may lead to pregnancy failure [47]. Recently, the non-classical class I HLA-G and HLA-E molecules, both of which are expressed by fetal trophoblast cells, have been reported to suppress production by isolated first-trimester uterine NK cells of cytokines including IL-10, IL-13, TNF-α, granulocyte macrophage colony-stimulating factor (GM-CSF) and IFN-γ[49]. Whether HLA-G and HLA-E regulate the production of cytokines by decidual T cells and macrophages is unknown. We suggest that the increased population of activated decidual T cells in molar pregnancy fail to undergo apoptosis but may secrete cytokines that influence proliferation and invasion of the molar trophoblast.