CELLULAR IMMUNOLOGY 129,435-448 (1990)

Human Trophoblast and JEG Choriocarcinoma Cells Are Sensitive to Lysis by IL-2-Stimulated Decidual NK Cells

ASHLEY KING AND Y. W. LOKE

Cambridge University Department ofPathology, Tennis Court Road, Cambridge2 CB2 I QPV United Kingdom

Received March 9,199O; accepted March 27, 1990

Freshly isolated decidual large granular lymphocytes (LGL) show natural killer (NK) activity against K562 cells but not against normal human trophoblast. We now show that these decidual LGL proliferate in vitro in response to recombinant interleukin-2 (rIL-2) and that these rIL-2- stimulated cells acquire a broad cytolytic potential that is characteristic of lymphokine-activated killer (LAK) cells. Both fetal fibroblasts and JEG3 choriocarcinoma cells are resistant to Iysis by freshly isolated decidual effecters but are readily kitled by IL2-stimulated decidual LGL. The ability to kill these target cells is acquired after only 18 hr exposure to rIL-2. rIL-2-activated decidual LGL also kill cultured normal trophoblast cells but much lower levels of cytolysis were seen even after the effecters had been stimulated with rIL-2 for 4-6 days. The preferential killing of malignant over normal human trophoblast cells raises questions about the potential role of IL-Zactivated de&dual LGL in the control of unduly invasive or malignant trophoblast popula- tiOIIS in ViVO. 0 1990 Academic Press. Inc.

INTRODUCTION

It is now generally accepted that the leucocytes present in both nonpregnant endo- metrium and decidua are predominantly large granular lymphocytes (LGL) with the unusual phenotype CD3-, CD1 6-, and CD56bmt+ ( l-4). The functions of these uter- ine LGL have not been established, although their preponderance in endometrium at the time of implantation and in early decidua suggests a possible role in controlling trophoblast development. Uterine LGL are most prevalent during the first trimester of pregnancy in humans, coinciding with the phase of extensive trophoblast migra- tion into maternal decidua and spiral arteries. It is essential that this migration is closely controlled because, although transformation of uterine arteries by trophoblast is crucial for the establishment of an adequate blood supply to the developing feto- placental unit, the trophoblast cells must not be allowed to penetrate the uterine wall (5).

We have recently demonstrated that freshly isolated human decidual LGL cannot kill first-trimester trophoblast cells, despite being able to lyse the NK-sensitive cell line K562 (6). We now report that in vitro stimulation of decidual LGL with recombinant interleukin-2 (rIL-2) generates a population of lymphokine-activated killer (LAK) cells which do possess the capacity to lyse trophoblast cells. In addition, these decidual LAK cells also acquire the ability to kill JEG choriocarcinoma cells and fetal fibro-

435

0008-8749/90$3.00 Copyright 0 1990 by Academic Press, Inc. All rights of reproduction in any form rserved.

436 KING AND LORE

TABLE 1

Monoclonal Antibodies Used to Character& Trophoblast

Mab Source Specificities

PKKl Labsystems HMFG-2 Unipath

W6/32 Sera-Lab 71.1 Dr. D. Antczak

Cytokeratin Human milk fat globulin-stains uterine glandular

epithelial cells. Monomorphic determinant of Class 1 HLA-A, B, C. Generated against horse endometrial cup cells and

found to cross-react with human villous

CTl0.18.4

18B/AS

Prof. P. Johnson

Dr. Y. W. Loke

cytotrophoblast but not extravillous trophoblast. Generated against cytotrophoblast cultured in our

laboratory. Reacts against all cytotrophoblast populations but

not villous syncytiotrophoblast

blasts, and do so to a much greater extent than normal trophoblast. Since IL2 has been shown to be present in villous syncytiotrophoblast (7,8), and cytolysis of tropho- blast does not normally occur during pregnancy, it would appear that the IL-2 activa- tion of decidual LGL must be finely regulated in vivo.

MATERIALS AND METHODS

Tissues

Placental tissue, fetal skin, and decidua were obtained from elective terminations of pregnancy at gestations between 6 and 12 weeks. The fragments were washed in RPM1 1640 (Flow) and separated macroscopically.

Trophoblast

Trophoblast was cultured as previously described (9) with certain modifications. Briefly, chorionic villi were minced into small pieces and incubated for 10 min at 37°C in 0.25% trypsin 1:250 (Difco), 0.02% EDTA (Sigma). The resultant cell suspen- sion was filtered through muslin, and then centrifuged at 400g for 5 min. The pellet was resuspended in RPM1 1640 and layered onto Lymphoprep (Flow), then centri- fuged at 800g for 15 min to pellet red blood cells. The band of cells at the interface was resuspended in Hams F12 culture medium (Flow), containing 20% fetal calf se- rum (FCS) (GIBCO), and 2 mML-glutamine and antibiotics. The cells were adjusted to a concentration of lo6 cells/ml and plated onto 35-mm culture dishes which had been precoated with 20 pg/ml laminin (Sigma) for 45 min at room temperature. The dishes were incubated in a continuous flow of 95% air, 5% CO2 at 37°C. These cul- tures routinely contain 80-90% trophoblast cells with characteristics of extravillous cytotrophoblast populations as defined by a panel of monoclonal antibodies (Mabs) (9). The Mabs we now use are shown in Table 1 (10). Trophoblast cells are PKKl+, HMFG-2-, W6/32+, 18B/A5+, 7 1. 1-, CT. 10.18.4+. The trophoblast cultures were used on Day 3-4 for cytotoxicity assays.

DECIDUAL LAK CELLS KILL HUMAN TROPHOBLAST 437

K.562

K562 cells (Flow) were maintained in suspension by twice weekly subculture in RPM1 1640 and 10% FCS with antibiotics and 2 mM L-glutamine.

JEG-3 Choriocarcinoma Cells

JEG-3 choriocarcinoma cells (a gift from Dr. R. Sutcliffe, Glasgow) were main- tained in RPM1 (Dutch modification; Flow) and 10% FCS with antibiotics and 2 mM L-glutamine.

Fetal Fibroblasts

Fetal skin fragments were disaggregated in a way similar to that of trophoblast, but were subsequently suspended in RPM1 1640 and 20% FCS with antibiotics and 2 mA4 L-glutamine and seeded on to untreated plastic flasks. They were used on Day 3-4 for the cytotoxicity assays.

Decidual Eflectors

Decidual leucocytes were extracted as before (6). Briefly, the decidual fragments were minced and then pushed through a 53-pm stainless steel sieve (Gallenkamp). The resulting cell suspension was washed and layered onto Lymphoprep and spun at 400g for 20 minutes. The cells at the interface were removed, washed, and resus- pended in RPM1 1640 and 20% FCS at 2 X 106/ml. They were cultured at 37°C 5% CO;! with rIL-2 (Sigma). The optimum dose of rIL-2 to ensure maximum prolifera- tion was determined initially. The cells were cultured in 24-well plates with varying doses of rIL-2 in a total volume of 2 ml. After overnight incubation the nonadherent cells were transferred to 96-well plates in a volume of 150 ~1, leaving behind adherent macrophages and stromal cells. On Day 3, 1 &/well of [3H]thymidine was added. Eighteen hours later the cells were harvested onto filter papers using a Dynatech Au- tomash 2000. Filters were placed in scintillation fluid (Optiphase Hisafe II) and counted in an LKB 12 16 Rackbeta scintillation counter. Following these experi- ments, nonadherent decidual lymphocytes were routinely grown in suspension with 100 U/ml rIL-2 in 25 cm2 flasks for l-8 days. After 3-4 days the cultures were supple- mented with fresh medium containing 100 U/ml rIL-2.

Chromium Release Assays

These were performed as previously described (6), with some modifications. Tro- phoblast cells were removed from tissue culture dishes on Day 3-4 with 0.25% Tryp- sin/0.02% EDTA and transferred to flat-bottomed 96-well plates (Falcon) (1 X lo4 cells/well) that had been precoated with laminin. After incubating overnight at 37°C 5% COz, the cells were labelled with 3 &i sodium “chromate (51Cr; Amersham) per well for 90 min in not more than 40 ~1 of RPM1 1640. In some experiments, the cells were labelled with “Cr overnight in 50 ~1 of Hams F12 and 20% FCS in a moist box. This resulted in better labelling of cells and lower levels of spontaneous release. JEG- 3 choriocarcinoma cells and fetal fibroblasts were also labelled overnight with 3 &i

438 KING AND LOKE

“Cr/well. After labelling, the cells were washed three times and 100 gl of complete medium was added. K562 cells were labelled with 100 &i of “Cr in not more than 200 ~1 RPM1 1640 for 90 min at 37°C. They were washed three times and resus- pended in complete medium for 1 hr at 37°C. After washing a further time the cells were resuspended at 1 X 105/ml and 100 &well was added to round-bottomed 96- well plates (Coming). The decidual lymphocytes, cultured in rIG2 for variable num- bers of days, were counted on the day of the assay and resuspended in the minimum volume required for the assay. One hundred microliters of effecters were added to each well to achieve a total volume of 200 ~1. The effector:target ratios are, therefore, different for each experiment. The plates were incubated for 5 hr at 37°C 5% CO;!, after which time 100 ~1 of supematant was removed from each well and counted in a gamma counter. Spontaneous “Cr-release was measured in wells containing target cells alone. This was < 10% for K562 and < 15% for trophoblast cells, fetal fibroblasts and JEG-3-choriocarcinoma cells. Maximum “Cr release was determined by the ad- dition of 100 ~1 of 2.5% Triton X-100 (BDH Chemicals Ltd) to the wells containing labelled target cells. Each assay was set up in triplicate and the mean cytotoxicity was calculated from:

(experimental 5 ‘Cr-release) - (spontaneous 5’Cr-release) x 1 o. (maximal “Cr-release) - (spontaneous “0-release) ’

Cytolysis of Fetal Fibroblasts and Trophoblast on Glass Slides

Lysis was also visualised directly by plating trophoblast cells and fetal fibroblasts on to glass Lab-tek tissue culture chamber/slides (Nunc), (those receiving trophoblast were precoated with laminin), and adding decidual effecters as before. Control cham- bers received no effector cells. After 5 hr at 37°C 5% CO* the slides were washed thoroughly to remove effector cells and viewed under phase contrast. Trophoblast cells were subsequently identified after acetone fixation (4°C 5 min) using PKKl (Labsystems), diluted l/200, a monoclonal antibody to cytokeratin. After incubation with the primary antibody for 45 min, the slides were covered with biotinylated rabbit anti-mouse immunoglobulins, 1 / 100 (Dakopatts) for 30 min. Peroxidase-conjugated avidin, l/400 (Dakopatts) was then applied to the slides for 30 min and the reaction was developed with 0.05% 3,3-diaminobenzidine (DAB; Sigma) with 0.03% hydrogen peroxide. All incubations were carried out at room temperature and the slides were washed twice with phosphate-buffered saline between each stage. The slides were counterstained with haematoxylin, dehydrated, and mounted. The slides plated with fetal fibroblasts were fixed in methanol and stained with eosin and methylene blue (Hema-Gurr, BDH Ltd).

Cytospin Preparations

Decidual LGL grown in suspension with rIG2 were resuspended at 2 X 106/ml in RPM1 1640 and 10% FCS and cytospin smears made with a Shandon cytocentrifuge. These slides were air-dried and fixed in acetone at 4°C for 5 min before storing at -20°C. A few slides were fixed in methanol for 10 min and then stained with Giemsa. Immunostaining of the cytospin slides was performed using an avidin-biotin method

DECIDUAL LAK CELLS KILL HUMAN TROPHOBLAST 439

rlL-2 (U/ml)

FIG. 1. The proliferative response of decidual LGL to rIL-2. Nonadherent lymphocytes were cultured with varying doses of rIL-2 for 4 days. Proliferation was assessed by [3H]thymidine incorporation.

as described above using CD45 (LCA, Dakopatts) and CD56 (Leu 19, Becton-Dick- inson). Both Mabs were diluted l/50.

RESULTS

CD56 + Cells Proliferate in Response to IL-2

The nonadherent decidual lymphocytes proliferate in response to rIL-2 in a dose- dependent fashion (Fig. 1). There was minimal proliferation of lymphocytes grown in the absence of IL-2 and, on Giemsa-stained smears, these cells did not appear to be very healthy with numerous dead cells seen. Cytospin smears of r&2-stimulated decidual lymphocytes revealed that the cells were present in clumps and the vast majority of the cells in each clump were CD45+, CD56+ cells. Mitotic figures were also present among the groups of CD56+ cells (Fig. 2) confirming active proliferation. Giemsa staining demonstrated that many of the cells contained the cytoplasmic gran- ules characteristic of LGL (Fig. 3). In addition, preliminary data obtained using flow cytometry of decidual cell preparations grown in rIL-2 indicates that 70% of the cells are CD56+ after extraction, with the percentage of CD56+ cells rising to 85 and 90% after exposure to rIL-2 for 1 and 5 days, respectively (unpublished). These findings indicate that it is the decidual LGL which preferentially proliferate in response to rIL-2 in vitro.

Cultured Human Trophoblast Cells Are Susceptible to Lysis by rIL-2-Stimulated Decidual Lymphocytes

The anti-trophoblast cytolytic activity of decidual lymphocytes cultured in rIL-2 for l-6 days was examined using a chromium release assay. Four experiments were

440 KING AND LOKE

DECIDUAL LAK CELLS KILL HUMAN TROPHOBLAST 441

performed and the three results are shown in Fig. 4 using effecters grown for 18 hr (a), 4 days (b), and 6 days (c). These demonstrate that trophoblast cells can be killed by IL-Zactivated decidual LGL although the percentage of lysis was always less than that seen with K562. Decidual lymphocytes that had been exposed to rIL-2 for 4-6 days were more lytic than those that had been exposed to rIL-2 overnight. The results of all experiments are summarised in Fig. 5.

There was some variation in the percentage of lysis seen in the trophoblast cultures exposed to LGL cultured for 3-6 days in rIL-2 which may be due to the different number of nontrophoblast-contaminating cells present in cultures prepared on different days. Therefore, to confirm that trophoblast cells themselves are lysed by rIL-Zstimulated decidual LGL, the cytolytic assay was performed on glass chamber/ slides. Decidual effecters cultured for 4 days in rIL-2 were incubated for 5 hr with trophoblast cells plated onto chamber/slides at an effector:target ratio of 1OO:l. At the end of the incubation period, the effecters were washed off and the slides viewed under phase contrast. There was obvious loss of trophoblast in chamber/slides incu- bated with decidual LAK cells (Figs. 6a, 6b), which was confirmed by PICK1 staining (data not shown).

Fetal Fibroblasts Are Susceptible to Lysis by IL-d-Stimulated Decidual Lymphocytes

Fetal fibroblasts were used as control targets as they are embryonic cells of an equiv- alent gestational age. Three experiments were performed and the results of one repre- sentative experiment are shown (Fig. 7). There was only minimal cytolysis of fetal fibroblasts by both PBL and freshly isolated decidual effecters. However, after only 18 hr rIL-2 stimulation, these effecters were capable of much higher levels of cytolysis against fetal fibroblasts, and this was always higher than that seen with trophoblast targets. These fetal fibroblast targets are, therefore, performing in these assays as LAK targets (11). In the chamber/slide assay virtually all fetal fibroblast cells were lysed at effector:target ratios of 50: 1 and 100: 1 (Fig. 8). This illustrates the differences in sensitivity between this assay and the chromium release assay.

JEG-3 Choriocarcinoma Cells Are Killed by IL-2Stimulated Decidual LGLs but Not Freshly Isolated Efectors

No lysis of JEG-3 choriocarcinoma cells was seen with PBL or unstimulated decid- ual effecters in three experiments. However, as with fetal fibroblasts, high levels of cytolysis were seen even after overnight stimulation of decidual LGL with rIL-2. This is in contrast to the results found with normal trophoblast where at least 3-4 days of culture with rIL-2 were needed to demonstrate significant lysis. The results of two representative experiments are summarized in Fig. 9.

FIG. 2. Cytospin smear of decidual LGL grown for 4 days in rIL-2 and stained for CD56. One clump of CD56+ cells is shown with a mitosis clearly visible (A) (X540).

FIG. 3. Cytospin smear of decidual LGL grown for 4 days in rIL-2 and stained with Giemsa, demonstrat- ing the presence of numerous granulated ceils. (Giemsa X540).

442 KING AND LOKE

al 70

.---a--- 60 **. Kc562

*. 60 l -,-

7 **-

” . . . .

120:1 60:1 302 163 7.6:1 3.8:l I.%,

Effector:Tarpet Ratio

b) 60 -I

60

60

.- 6O:l 30:1 15:1 7.5:1 4, 2:1 13

ElleQor:Targel Ratio

70 -

60.

60 -

40-

30 -

20 -

10-l ,703 66:l 42.5:1 213 10.6:1 531

Elleclor : Target Ratio

FIG. 4. Three chromium release experiments in which decidual effecters grown for varying number of days in rIG2 were tested against cultured trophqblast cells. The effecters were cultured with rIL-2 for (a) 18 hr, (b) 4 days, (c) 6 days.

DISCUSSION

We have shown in this study that decidual LGL can be stimulated by rIL-2 in vitro to proliferate and to kill normal human first-trimester trophoblast. In contrast, freshly isolated decidual LGL do not lyse these target cells (6). The decidual LGL probably undergo transformation by IL-2 into lymphokine-activated killer (LAK) cells, a phe- nomenon which is well recognised in NK cells from peripheral blood (12, 13). Al- though some LAK activity has been attributed to T cells, most activity is found in

DECIDUAL LAK CELLS KILL HUMAN TROPHOBLAST 443

n Trophcblast

1 K562

FIG. 5. Cytolytic activity of decidual effector cells cultured in rIL-2 for 1-6 days. Each bar represents a single assay against both K562 and trophoblast at an effector:target ratio of 50: 1. Two assays were per- formed using effecters cultured overnight (Dee o/n) or for 6 days (Dee d6). One assay was performed using effecters cultured for 3 days (Dee d3), 4 days (Dee d4), and 5 days (Dee d5).

NK-lineage cells ( 14). LAK cells are capable of a wider range of cytolytic activity than unstimulated NK cells, and in particular are noted for their lysis of a number of NK- resistant targets, such as freshly isolated tumour cells and cell lines derived from solid tumours ( 12- 14). We have shown that embryonic fetal fibroblasts are very sensitive to lysis by decidual LAK cells, even those generated after only 18 hr rIL-2 stimulation. Cultured trophoblast cells, despite being embryonic cells of an equivalent gestational age, are not as sensitive to this effector population with lower levels of cytolysis being seen even after 6 days rIL-2 stimulation. Lysis of previously resistant targets by LAK cells may be due to IL-Zinduced expression of new surface receptors on the effector cells ( 14). In view of our recent observation that the resistance of trophoblast to lysis by freshly isolated decidual LGL may be due to lack of the appropriate NK target ( 15), it is possible that decidual LAK cells are expressing novel receptors which recog- nize target molecules different from those recognized by decidual NK cells. The de- cidual NK cells are characterized by an unusual surface phenotype, CD3-, CD 16-, CD56b”gh’+, which is present on only -2% of PBLs in normal individuals (16). Al- though unstimulated CD56b”ph’+ PBL cells are less cytotoxic than CD56di”+ NK cells, the most effective LAK effecters are conversely CD56b”gh’+ (17). Our findings with decidual LGL concur with this since unstimulated decidual cells are less cytotoxic than NK cells from peripheral blood (4, 6), but are easily, stimulated by rIL-2 to proliferate and become potent cytolytic effecters at low E:T ratios. The function of the CD56 molecule is unknown, but has recently been shown to be the embryonic form of NCAM ( 18, 19). This molecule may function as an adhesion molecule in NK-target cell interaction but only when the target cells also express NCAM (20).

Our results are in accord with the findings in mouse studies where NK activity has been demonstrated in decidua against NK-sensitive target cells, but not against

444 KING AND LOKE

Fl G. 6. Cytotoxicity assay using decidual effecters grown for 4 days with rIL-2 against trophoblast on a I Lab-tex chamber slide. The effector:target ratio was 100: 1. The cells are shown viewed under coni :rast at the end of the assay (a) with effecters and (b) no effectors.

plated

DECIDUAL LAK CELLS KILL HUMAN TROPHOBLAST 445

FIG. 7. Histogram showing the results ofa representative chromium release experiment using fetal fibro- blasts as targets. The effectors were PBL, freshly isolated decidual cells (Deck and decidual cells cultured in rIL-2 overnight (Dee o/n), for 4 days (Dee d4), and 8 days (Dee d8). Each bar represents pooled effecters collected on different days at an effector target ratio 50: 1.

blastocysts (2 1). Furthermore, splenic NK cells are not cytolytic towards purified mu- tine trophoblast populations unless first stimulated by IL-2 (22, 23). Of the three published human studies, one did demonstrate that decidual NK activity is increased after stimulation with IL-2, but these investigators have only tested their effecters against NK-sensitive K562 cells and not against trophoblast (4). The second study has demonstrated cytolysis of a normal human trophoblast cell line by IL-2-stimulated decidual cell preparations, but the unstimulated effecters appeared to have no NK activity at all, not even against NK-sensitive cells (24). A recent report showed no proliferative response at all by decidual CD56+ cells in response to IL-2 although the NK activity of unstimulated effecters was as high as that of PBLs when tested against K562 (25). The different extraction and purification procedures for decidual effecters in these studies may partly explain the different results.

Our present study is the only one so far which uses human trophoblast cells with identifiable characteristics of the extravillous population. Since these are the cells which migrate into decidua in vivo, they are the most appropriate targets to use in assays against decidual LGL. Our results indicate that human decidual LGL clearly have the potential to kill trophoblast if they are appropriately stimulated. That this does not normally occur in vivo during successful pregnancy raises a number of ques- tions about possible regulatory mechanisms, since IL-2 has been identified in syncy- tiotrophoblast by both immunohistology and by in situ hybridisation (7, 8). Firstly, the levels of IL-2 required to generate decidual LAK cells may not be reached in vivo, since it can be shown experimentally that introduction of exogenous IL-2 to pregnant mice will induce abortion (26). Secondly, a complex suppressive network may be operative. Decidual and trophoblast-derived factors have been observed to inhibit IL- 2 stimulation of lymphocytes which may involve a variety of cytokines such as TGF- /3 (27), PGE-E2 (28), or OTP-1 in sheep (29). This latter observation is interesting as

446 KING AND LOKE

control

control

50:1

FIG. 8. Cytotoxicity assay using decidual effecters cultured for 4 days in rIL2 against fetal fibroblasts plated on a Lab-tek chamber/slide. After incubating for 5 hr the effecters were washed off and the slides stained with Hema-Gurr. No fetal fibroblasts remain at an E:T ratio of 100: 1 or 50: 1.

FIG. 9. Histogram of two chromium release experiments using JEG3 choriocarcinoma cells as targets. The effectorztarget ratio was 50: 1 and each bar represents pooled et%ctors collected on different days. The effector cells were PBL, freshly isolated decidual cells (Dee) and cells cultured in rIC2 overnigltt (Dee o/n), for 4 days (Dee d4), and 6 days (Dee d6).

DECIDUAL LAK CELLS KILL HUMAN TROPHOBLAST 447

IFN-(Y (which has extensive homology with OTP-1) may also inhibit the generation of LAK activity (30) and has been localized to placental trophoblast by immunohis- tology (3 1). Thirdly, trophoblast themselves may generate inhibitory signals against the LAK effecters, as it has been described that monolayers of target cells may inhibit IL-2 activated killer cells (32).

Such complex systems of control are presumably necessary to prevent IL-2-medi- ated transformation of decidual LGL into LAK cells in vivo with resultant cytolytic activity against normal trophoblast. However, it would be fair to suppose that the ability of decidual LGL to develop potent cytolytic activity must subserve some other important function. As we have observed in the present study, several days of culture with IL-2 are necessary before decidual LAK cells exhibit significant killing of tropho- blast and, even then, the extent of cytolysis is never as great as that against K562 or fetal fibroblasts. In contrast, decidual LGL incubated for only 18 hr with IL-2 are already highly cytolytic against JEG-3-choriocarcinoma cells, sometimes to a level exceeding that against K562. This is in accord with the report of Rosenberg et al. ( 1985) that LAK cells show selectivity toward malignant target cells (33). In addition, IL-Zactivated killer cells are more sensitive to the inhibitory signals of benign rather than malignant target cells (32) which may explain why normal trophoblast cells are killed at a lower level than JEG-3 choriocarcinoma cells. Our observation that malig- nant trophoblast cells are more susceptible to cytolysis by decidual LAK cells, suggest that a possible role for these decidual LAK cells in vivo may be to guard against the development of unduly invasive or malignant trophoblast populations during pregnancy.

ACKNOWLEDGMENTS

Supported by grants from The Wellcome Trust, Medical Research Council, East Anglian Regional Health Authority, World Health Organization, American Friends of Cambridge University, and Journal of Reproduction and Fertility. We are grateful to our obstetric colleagues at Addenbrooke’s Hospital, Cambridge for collecting the placental material.

REFERENCES

1. King, A., Wellings, V., Gardner, L., and Loke, Y. W., Human Immunol. 24,195, 1989. 2. Starkey, P. M., Sargent, I. L., and Redman, C. W. G., Zmmunology65, 129, 1988. 3. Bulmer, J. N., and R&son, A., The decidua in early pregnancy. In “Early Pregnancy Loss, Mechanisms

and Treatment” (R. W. Beard and F. Sharp, Eds.), p. 17 1. London Royal College of Obstetricians & Gynaecologists, London, 1988.

4. Manaseki, S., and Searle, R. F., Cell Immunol. 121, 166, 1989. 5. Robertson, W. B., Pathology of the pregnant uterus. In “Haines and Taylor, Obstetrical and Gynaeco-

logical Pathology.” (H. Fox, Ed.), 3rd ed. pp. 1149-l 176. Churchill Livingstone, New York, 1987. 6. King, A., Birkby, C., and Loke, Y. W., Cell. Immunol. 118,337, 1989.

7. Boehm, K., Kelley, M. F., Ban, J., and Ilan, J., Proc. Natl. Acad. Sci. USA 86,656, 1989. 8. Soubiran, P., Zapitelli, J-P., and SchalTar, L., J. Reprod. Immunol. 12,225, 1987. 9. Loke, Y. W., and Burland, K., Placenta 9,173, 1988.

10. Loke, Y. W., Gardner, L., and Grabowska, A., Placenta 10,407, 1989. 11. Braun, N., Papadopoulos, T., and Muller-Hennelink, H. K., Virch. Arch. B Cell Pathol. 56,25, 1988. 12. Herberman, R. B. et al., Immunol. Today8, 178, 1987. 13. Michon, J. M., and Fidman, W. H., Ann. Inst. Pasteur Immunol. 139,433, 1988. 14. Ortaldo, J. R., and Longo, D. L., J. Natl. Cancer Inst. 80,999, 1988. 15. King, A., Kalra, P., Loke, Y. W., Cell. Immunol., 127,230, 1990.

448 KING AND LOKE

16. Ianier, L. L., Le, A. M., Civin, C. I., Loken, M. R., and Phillips, J. H., J. Zmmunol. X%,4480, 1986. 17. Ellis, T. M., and Fisher, R. I., J. Zmmunol. 142,2949, 1989. 18. Husmann, M., Pietsch, T., Fleischer, B., Weisgerber, C., and Bitter-Suermann, D., Eur. J. Zmmunol.

19,1761, 1989. 19. Lanier, L. L., Testi, R., Binal, J., and Phillips, J. H., J. Exp. Med. 169,2233, 1989. 20. Nitta, T., Yagita, H., Sato, K., and Okumara, K., J. Exp. Med. 170, 1757, 1989. 2 1. Croy, B. A., Gambel, P., Rossant, J., and Wegmann, T. G., Cell. Zmmunol. 93,3 15, 1985. 22. Zuckerman, F. A., and Head, J. R., J. Zmmunol. 116,274,1989. 23. Drake, B. L., and Head, J. R., J. Zmmunol. 143,9, 1989. 24. Parhar, R. S., Yagel, S., and Iala, P. K., Cell. Zmmunol. 77,263, 1989. 25. R&son, A., and Bulmer, J. N., Clin. Exp. Zmmunol. 77,263, 1989. 26. Tezabwala, B. V., Johnson, P. M., and Rees, R. C., Immunology 67, 115, 1989. 27. Clark, D. A., Falbo, M., Rowley, R. B., Banwatt, D., and Stedrouska-Clark, J., J. Zmmunol. 141,3833,

1988. 28. Tawfik, 0. W., Hunt, J. S., and Wood, G. W., Amer. J. Reprod. Zmmunol. Microbial. 12, 11, 1986. 29. Newton, G. R., Vallet, J. R., Hansen, P. J., and Bazer, F. W., Amer. J. Reprod. Zmmunol. Microbial.

19,99, 1989. 30. Tokuda, Y., Ebina, N., and Golub, S. H., Cancer Zmmunol. Zmmunother. 30,205, 1989. 31. Howatson, A. G., Farquharson, M., Meager, A., McNicol, A. M., Foulis, A. K., J. Endocrinol. 119,

531,1988. 32. Heiskala, M., and Timonen, T., Cell. Zmmunol. 110,209, 1987. 33. Rosenberg, S., Lotze, M. T., Muul, L. M., Leitman, S., Chang, A. E., Ettinghansen, S. E., Mtony,

Y. L., Skibber, J. M., Shilom, E., Vetto, J. T., Seipp, C. A., Simpson, C., and Reichert, M., N. Engl. J. Med. 313,1485, 1985.