Journal of Reproductive Immunology 82 (2009) 24–31

Decidual leucocyte populations in early to late gestationnormal human pregnancy

P.J. Williams, R.F. Searle, S.C. Robson, B.A. Innes, J.N. Bulmer ∗Institute of Cellular Medicine, Newcastle University, Newcastle-upon-Tyne, Tyne and Wear NE2 4HH, United Kingdom

Received 5 May 2009; received in revised form 7 August 2009; accepted 14 August 2009

Abstract

Most research on human decidual leucocytes to date has focused on the predominant CD56+ uterine natural killer (uNK) cellpopulation in early pregnancy. Few reports have documented decidual leucocyte populations after 13 weeks gestation and in latepregnancy. Placental bed (decidua basalis) and non-placental bed (decidua parietalis) biopsies from normal pregnancies were takenfrom women undergoing termination of pregnancy in the 1st and 2nd trimesters and following Caesarean section in the 3rd trimester.Immunohistochemistry was used to quantify the numbers of decidual cells expressing CD56, CD3, CD8, CD94, NKG2A and CD14and double labelled CD161+CD3+ NKT-like cells. Although a significant reduction in CD56+ uNK cells was found in 3rd trimestersamples compared with 1st and 2nd trimester decidua, a substantial residual CD56+ leucocyte population was identified in 3rdtrimester decidua. Expression of the KIR CD94/NKG2A mirrored that of CD56 at all gestational ages, providing an explanationfor the absence of cytotoxic responses at the fetal–maternal interface. There was no difference in leucocyte populations betweendecidua basalis and decidua parietalis. Double immunohistochemical labelling revealed small numbers of decidual CD3+CD56+and CD8+CD56+ cells, which decreased in number at term, and CD161+CD3+ cells, which increased in number at term. No

differences in leucocyte populations were detected between decidua parietalis and decidua basalis. In contrast to previous reports,a substantial residual CD56+ cell population was demonstrated in 3rd trimester decidua. Decidual cytotoxic T-lymphocytes didnot alter in number during gestation, while in contrast CD14+ macrophages decreased at term, representing the smallest decidualpopulation assessed.© 2009 Published by Elsevier Ireland Ltd.

Keywords: Decidua; Leucocytes; Pregnancy

1. Introduction

Immunohistochemical studies have demonstratedthat approximately 30–40% of stromal cells in humandecidua in early pregnancy are leucocytes (Bulmer et al.,

1991). CD56bright CD16− uterine natural killer (uNK)cells account for up to 70% (Bulmer et al., 1991; Moffett-King, 2002), CD14+ macrophages up to 30% (Bulmer

∗ Corresponding author at: Institute of Cellular Medicine, 3rd Floor,William Leech Building, Newcastle University, Newcastle-upon-TyneNE2 4HH, United Kingdom.

E-mail address: j.n.bulmer@ncl.ac.uk (J.N. Bulmer).

0165-0378/$ – see front matter © 2009 Published by Elsevier Ireland Ltd.doi:10.1016/j.jri.2009.08.001

et al., 1991) and CD3+ T-lymphocytes, the majority ofwhich are CD8+ cytotoxic T-lymphocytes, fewer than20% of the total decidual leucocyte population (Bulmeret al., 1991).

Little is known about decidual leucocyte populationsafter the 1st trimester but it is generally accepted thatuNK cells decrease in number (Haller et al., 1993;Bulmer and Lash, 2005). The close association of uNKcells and extravillous trophoblast in the placental bed

(Loke and King, 1995) has led to the suggestion thatthese cells have a crucial role in regulating trophoblastinvasion and decidual artery remodelling (Craven etal., 1998; Hanna et al., 2006) and this is supported

product

bpwieh

ctfawe

aNestNucTiaa

pateoMCrf2

eempMic(aod

cs

P.J. Williams et al. / Journal of Re

y HLA-receptor interaction between extravillous tro-hoblast and uNK cells (Moffett-King, 2002). However,hile animal models demonstrate a role for uNK cells

n vascular remodelling (Greenwood et al., 2000), thextent to which these observations can be extrapolated touman pregnancy is not clear (Pijnenborg et al., 2006).

Regulation of cytolytic activity by NK cells and someytotoxic T-lymphocytes is via Killer Inhibitory Recep-ors (KIR), expression of which may account for theirailure to generate cytotoxic responses against the semi-llogeneic fetus as these receptors are able to interactith all three trophoblast HLA-class I molecules (Apps

t al., 2007; Sharkey et al., 2008).The cell surface antigens CD8 and CD56, however,

re not specific markers for cytotoxic T-lymphocytes andK cells, respectively. CD56 is a marker of cytolytic

ffector function of circulating T-lymphocytes followingtimulation via cytokines or CD3 (Pittet et al., 2000). Fur-hermore CD8 may be expressed by both NK cells andKT cells (Tsuda et al., 2001) and similarly, some pop-lations of CD8+ T-lymphocytes may co-express NKell markers (CD56, CD161) (Takayama et al., 2003).hus complete characterisation of cell surface antigens

s important since CD8+CD56+ T cells are function-lly distinct from CD8+CD56− T cells (Takayama etl., 2003).

Classical NKT cells have also been detected in earlyregnancy decidua (Tsuda et al., 2001), and their lig-nd CD1d is expressed on both villous and extravillousrophoblast (Boyson et al., 2002). Classical NKT cellsxpress a semi-invariant TCR (V�24J�18) that rec-gnizes glycolipids (alpha-GalCer) presented by theHC-like molecule CD1d and NKT cells also expressD161 (Peralbo et al., 2007). NKT cells have been

eported to be present in decidua at approximately ten-old their frequency in peripheral blood (Boyson et al.,002).

Differences in leucocyte distribution and cytokinexpression between decidua basalis and decidua pari-talis have led to the suggestion that distinct cytokineicroenvironments are important for regulation of tro-

hoblast invasion (Von Rango et al., 2003; Trundley andoffett, 2004). In agreement with this are reports of

ncreased numbers of cytotoxic T-lymphocytes and uNKells in decidua parietalis compared to decidua basalisSindram-Trujillo et al., 2003a,b). In contrast, Haller etl. (1995) failed to show any difference in the frequencyf immunocompetent cells between decidua basalis and

ecidua parietalis.

The objectives of the present study were to preciselyharacterise and quantify cytolytic decidual leucocyteubpopulations including NK, NKT, CD8+ T cells and

ive Immunology 82 (2009) 24–31 25

macrophages throughout normal pregnancy, to comparethe presence of these leucocytes in decidua basalis anddecidua parietalis and to assess whether the expressionof KIR CD94/NKG2A changed during pregnancy.

2. Materials and methods

After approval from Newcastle and North Tyne-side Joint Ethics Committee and appropriate informedconsent, placental bed and non-placental bed biopsieswere obtained from women undergoing elective preg-nancy termination during the 1st trimester and early2nd trimester and from women undergoing elective Cae-sarean section, prior to the onset of labour, in the 3rdtrimester as previously described (Robson et al., 2002).Ultrasound scan was performed following the inductionof anesthesia with the patient in the lithotomy position.This scan determined viability, gestational age (basedon fetal crown rump length or biparietal diameter), andthe position of the placenta. All women who underwentpregnancy termination at <12 weeks gestation, weregiven misoprostol (400 �g per vaginum) 2–3 h beforethe procedure. Women undergoing pregnancy termina-tion at >14 weeks of gestation underwent dilatation andevacuation following administration of gemeprost (1 �gper vaginum) at 3 and 6 h before the procedure. Noneof the subjects had reported vaginal bleeding prior totermination of pregnancy. Sample numbers and gesta-tional ages are given in Table 1. Placental bed biopsieswere defined as those containing extravillous trophoblast(identified by cytokeratin immunostaining), whereasnon-placental bed biopsies did not contain trophoblast;these correspond to decidua basalis and decidua pari-etalis, respectively. Biopsies were snap-frozen and storedat −70 ◦C.

2.1. Monoclonal antibodies (mAbs)

Eight mAbs were employed for immunostaining offrozen sections (Table 2). The optimal dilution for eachmAb was determined in positive control tissue (frozensections of either tonsil or decidua), selected on thebasis of maximal specific reactivity and minimal back-ground staining. Incubation with primary antibody wasfor 30 min at room temperature except for TCR V�24(60 min) and CD161 (overnight at 4 ◦C).

2.2. Single immunohistochemical labelling

Serial cryostat sections (7 �m) were immunostainedusing an avidin–biotin complex (ABC) peroxidasemethod (Vectastain Elite, Vector Laboratories, Peterbor-

26 P.J. Williams et al. / Journal of Reproductive Immunology 82 (2009) 24–31

Table 1Details of subject groups.

Subject group Mean (SEM) gestational age (weeks) Gestational age range (weeks) Number of subjects

1st trimester decidua basalis 9.92 (0.46) 7–12 131st trimester decidua parietalis 9.83 (0.54) 8–12 62nd trimester decidua basalis 14.67 (0.50) 13–19 122nd trimester decidua parietalis 14.7 (0.42) 13–16 7

3rd trimester decidua basalis 37.0 (1.0)3rd trimester decidua parietalis 37.58 (0.43)

ough, UK) as previously described (Robson et al., 2002).The reaction was developed using 3,3′ diaminobenzi-dine (DAB) (Sigma Chemical Co., Poole, UK) to givea brown reaction product. Positive (frozen sections oftonsil or decidua) and negative (test sections in whichprimary mAb was replaced by normal serum) controlswere included in each staining run and for each mAb.

2.3. Double immunohistochemical labelling

In order to further quantify expression of CD8 andCD56 by NK, T and NKT cells, double immuno-histochemical labelling was performed to identifyCD3+CD56+ and CD8+CD56+ activated T or NKT cellsand CD161+CD3+ NK T-like double positive cells. Sec-tions were first labelled with CD3, CD8 or CD161 usingthe ABC-peroxidase method described above which wasdeveloped using NovaRed (Vector Laboratories) to givea red reaction product. The slides were then incubatedfor 30 min with the appropriate second primary antibody

(CD56 or CD3). Sections were then sequentially incu-bated with rabbit-anti-mouse immunoglobulins (Dako;30 min) and alkaline phosphatase anti-alkaline phos-phatase (APAAP) (Dako; 30 min). The reaction was

Table 2Primary antibodies used. Negative controls were performed by replacing the

Antibody Supplier Specificity

CD3 Novocastraa T cell associateCD8 DAKOb Suppressor/cytCD14 Serotecc Macrophages aCD56 Novocastra N-CAM expre

cytotoxic T celCD94 Serotec CD94 moleculNKG2A Professors Alessandro and Lorenzo

Moretta, Univerita di GenovaNKG2A molec

CD161 Serotec C-type lectin eTCR V�24 Serotec V�24 variant o

a Novocastra Laboratories Ltd., Newcastle-upon-Tyne, UK.b DAKO Ltd., Cambridgeshire, UK.c Serotec Ltd., Oxford, UK.

26–38 1233–39 12

developed using the alkaline phosphatase substrate IIIkit (Vecta Blue) (Vector Laboratories).

2.4. Quantification and analysis of results

Positively labelled cells were counted in five ×250magnification fields using a 10 mm × 10 mm graticule.Counting was performed in equivalent fields in serialsections of each case for each mAb. The results wereanalysed by calculating the mean number of positivelabelled cells ± standard error of mean (SEM) per ×250medium power field for each positive cell population.When counting double labelled slides both single pos-itive (red for first antibody, blue for second antibody)and double positive cells (labelled both red and blue)were counted in order to gain a full picture of leucocytepopulations present.

2.5. Statistical analysis

Descriptive statistical analysis and assessment of nor-mality was performed using Prism (GraphPad SoftwareInc.). Data shown were normally distributed and the stu-dent’s t-test was used to compare decidual leucocyte

primary antibody with non-immune mouse serum.

Clone Dilution

d CD3 antigen UCHT1 1:200otoxic T cells and subset of NK cells DK25 1:100nd monocytes UCHM1 1:20

ssed by uNK cells and subset ofls

ERIC1 1:100

e of CD94/NKG2 receptors HP-3B1 1:200ule of CD94/NKG2A receptor Z�99�2B 1:10

xpressed by NK, NKT and T cells B199.2 1:100f TCR C15 1:100

P.J. Williams et al. / Journal of Reproduct

Fig. 1. Graphs illustrating leucocyte populations in 1st trimester, 2ndtrimester and 3rd trimester (a) decidua basalis and (b) decidua pari-etalis. Numbers of CD14 and CD56 positive cells decreased betweenthe 1st and 2nd trimester and the 3rd trimester of pregnancy. Dataac*

pangw

3

i(

TC

S

1

2

3

N

re expressed as mean and SEM of 12–13 cases per group. Statisti-al differences were determined using the student’s t-test. *P < 0.05,*P < 0.001, ***P < 0.0001.

opulations between 1st, 2nd and 3rd trimester decidua,s well as between placental bed (decidua basalis) andon-placental bed (decidua parietalis) decidua for eachestational age group. The conventional level of P < 0.05as taken as the limit of significance.

. Results

All leucocyte subpopulations assessed were presentn decidua from 1st, 2nd and 3rd trimester pregnancyFig. 1 and Table 3).

able 3omparison of decidual leucocyte populations between decidual basalis and d

ample CD3 CD8

st trimesterDecidua basalis 34.9 ± 3.7 31.7 ± 3.1Decidua parietalis 38.8 ± 4.6 28.7 ± 1.3

nd trimesterDecidua basalis 34.6 ± 4.9 27.0 ± 4.6Decidua parietalis 43.3 ± 3.0 35.3 ± 3.2

rd trimesterDecidua basalis 40.3 ± 6.8 34.8 ± 5.5Decidua parietalis 44.0 ± 6.5 36.7 ± 5.0

umbers represent mean number of positive cells per ×250 field ± SEM.

ive Immunology 82 (2009) 24–31 27

CD3+ and CD8+ cell numbers did not vary signifi-cantly between 1st, 2nd and 3rd trimester decidua basalis(Table 3, Figs. 1 and 2a). To confirm specificity of CD3labelling TCR�� (Serotec, UK) staining was also per-formed (data not shown). Similar staining patterns wereobserved for CD3 and TCR��·

No significant difference in CD14+ cell numbers wasobserved between 1st and 2nd trimester decidua basalis(Table 3, Fig. 1). Numbers of CD14+ cells were signifi-cantly reduced in 3rd trimester decidua compared to both1st (P < 0.0001) and 2nd trimesters (P < 0.0001).

Similarly, CD56+ cell numbers in decidua basalis(Fig. 2b) did not differ between the 1st and 2nd trimesters(Table 3), but there was a significant reduction inCD56+ cells in the 3rd trimester compared with both1st trimester (P = 0.001) and 2nd trimester (P = 0.001)decidua (Fig. 1). Anti-CD94 labelling mirrored CD56labelling with a significant decrease in 3rd trimesterdecidua (P = 0.03 compared to 1st trimester; P = 0.007compared to 2nd trimester) (Fig. 1). NKG2A labellingwas also reduced in 3rd trimester decidua basalis com-pared to 1st trimester decidua (P = 0.04) (Fig. 2c) and2nd trimester decidua (P = 0.01). In both 1st trimesterand 3rd trimester decidua CD16 labelling was sparse(1st trimester: 7.42 ± 2.0; 3rd trimester: 10.9 ± 2.8); thislow level of CD16 labelling confirmed the NK pop-ulation to be of the CD16−CD56+ uNK type ratherthan representing the CD16+ peripheral blood NKphenotype.

Although a low number (0.4 ± 0.5%) of decid-ual NKT cells have been reported to express TCRV�24 (Tsuda et al., 2001), single immunohistochemicallabelling of both 1st and 3rd trimester decidual samplesfailed to show any expression of TCR V�24 by decid-ual lymphocytes. Anti-TCR V�24 antibody activity wasdemonstrated by specific staining of tonsil tissue.

The gestational age differences described for deciduabasalis were also detected in decidua parietalis (Fig. 1).Gestational age matched placental bed (decidua basalis)and non-placental bed (decidua parietalis) biopsies in

ecidua parietalis in the 1st, 2nd and 3rd trimester of pregnancy.

CD14 CD56 CD94 NKG2A

34.4 ± 2.8 45.2 ± 2.8 44.1 ± 3.4 40.6 ± 3.726.3 ± 1.5 39.7 ± 4.5 37.5 ± 3.0 39.3 ± 2.6

33.3 ± 1.9 48.7 ± 4.0 46.0 ± 3.2 43.4 ± 3.633.3 ± 2.8 49.1 ± 3.6 43.6 ± 2.6 45.7 ± 3.6

17.3 ± 1.2 29.0 ± 3.3 33.1 ± 2.8 30.1 ± 3.019.8 ± 1.9 31.8 ± 1.7 32.1 ± 1.9 28.9 ± 2.0

28 P.J. Williams et al. / Journal of Reproductive Immunology 82 (2009) 24–31

basalisD8/CD

cells apof the

are CD8+ cells (Bulmer et al., 1991). The similar stain-ing pattern between CD3 and TCR�� expression inthe present study (data not shown) confirms previousreports that the �� heterodimer is the preferential form

Fig. 2. Photomicrograph (×100) of 1st trimester leucocytes in decidua(c) NKG2A single positive cells (positive cells appear brown) and (d) CCD56 single positive cells appear blue, double labelled CD8+CD56+to colour in this figure legend, the reader is referred to the web version

each of the gestational age groups showed no significantdifference in expression of CD3, CD8, CD14, CD56,CD94 or NKG2A in 1st, 2nd and 3rd trimester betweendecidua basalis and decidua parietalis (Table 3).

3.1. Double immunohistochemcial labelling

A significant decrease in CD3+CD56+ cells wasobserved in decidua basalis (Fig. 3) from 1st trimesterto 3rd trimester (P = 0.0044). A similar staining pat-tern was also seen for CD8+CD56+ double labelling(P = 0.0191). In contrast there was a significant increasein CD161+CD3+ cells in 3rd trimester compared to earlypregnancy decidua basalis (P = 0.0452).

4. Discussion

In the present study all the decidual leucocyte popula-tions demonstrated in the 1st trimester of pregnancy werealso present in both the 2nd and 3rd trimester. CD14+

macrophages were present in lower numbers in termdecidua compared with CD3+ T-lymphocytes. Further-more, the numbers of decidual CD56+ cells did not alterbetween 1st and 2nd trimester pregnancy samples, and

showing (a) CD8 single positive cells, (b) CD56 single positive cells,56 double positive cells (×200) (CD8 single positive cells appear red,

pear purple [indicated by arrow]). (For interpretation of the referencesarticle.)

there was a substantial residual CD56+ cell populationin late third trimester samples.

A similar staining pattern was seen for CD3 and CD8,confirming that the majority of decidual T-lymphocytes

Fig. 3. Graph illustrating double labelled leucocyte populations in1st trimester and 3rd trimester decidua basalis. Data are expressedas mean and SEM of 12–13 cases per group. Statistical differenceswere determined using the student’s t-test. *P < 0.01, **P < 0.05.

product

oeTed3iwbtwpP

tcumcptenaaraspa(

dwlCldbstlciT3tddmms

P.J. Williams et al. / Journal of Re

f TCR expressed by decidual T-lymphocytes (Diverst al., 1994; Morii et al., 1993), although expression ofCR �� by a minor decidual T cell population cannot bexcluded. No significant difference in CD3+ and CD8+ecidual leucocytes was observed in either 1st, 2nd orrd trimester, indicating that these cells remain constantn number throughout gestation. This result contrastsith that of Haller et al. (1993) who reported a dou-ling of CD3+ cell numbers in 3rd trimester comparedo 1st trimester decidua. In the present study CD3+ cellsere identified using the UCHT1 clone with an ABCeroxidase method, whereas Haller used OKT3 and theAP method.

In the 3rd trimester CD3+ T-lymphocytes representhe largest immune cell population due to the con-omitant decrease in CD56+ uNK cells. The finding ofnaltered T lymphocyte numbers throughout gestationay reflect an immunomodulatory function for these

ells in decidua. Proposed functions for decidual T lym-hocytes have focused on the Th1/Th2 balance thoughto be associated with successful pregnancy (Wegmannt al., 1993). However, the Th1/Th2 paradigm appearsow to be an oversimplification and Th1 cytokines suchs interferon gamma (IFN�) and tumor necrosis factorlpha are thought to be important for uterine vascularemodelling and implantation, respectively (Chaouat etl., 2007). Furthermore, in an earlier study we havehown that CD8+ T lymphocytes from normal earlyregnancy decidua have cytolytic function and producerange of cytokines, including IFN� and interleukin-8

Scaife et al., 2006).In both placental bed and non-placental bed

ecidua from all gestational ages CD14+ macrophagesere detected in lower numbers than CD3+ T-

ymphocytes, conflicting with previous reports thatD14+ macrophages comprise up to 30% of decidual

eucocytes in early pregnancy (Bulmer et al., 1991). Theifference in macrophage numbers may be explainedy use of different CD14 antibody clones in othertudies, the present investigation using UCHM1 ratherhan Leu-M3 (Bulmer et al., 1991). The finding of aarger CD3+ T-lymphocyte than CD14+ macrophageell population in 3rd trimester decidua is, however,n agreement with others (Abadia-Molina et al., 1996).he significant reduction in CD14+ macrophages inrd trimester decidua compared with both 1st and 2ndrimester decidua may be explained by the suggested roleecidual macrophages play in regulation of apoptosis

uring the implantation process and placental develop-ent (Abrahams et al., 2004). Furthermore, decidualacrophage function may be modulated by their expres-

ion of leucocyte immunoglobulin-like receptors which

ive Immunology 82 (2009) 24–31 29

can interact with dimeric HLA-G expressed by extravil-lous trophoblast, providing a potential mechanism fora local immunomodulatory role within decidua viacytokine secretion (Apps et al., 2007).

Although the level of CD56 expression in 1sttrimester decidua concurs with levels of expressionreported in previous studies (Bulmer et al., 1991), anunexpected finding was unaltered CD56 expressionin early 2nd trimester samples, with high numbers ofCD56+ cells also detected in 3rd trimester decidua.Previous reports had suggested that CD56+ cell num-bers decline following the 1st trimester of pregnancy,being virtually absent from term decidua (Haller et al.,1993; Saito, 2000). The present report of a substantialCD56+ cell population in late pregnancy decidua is inagreement with Vargas et al. (1993) who reported thepresence of a CD56+ cell population in 3rd trimesterdecidua. Sparse expression of CD16 indicates that theCD56+ cells represent the uNK cell population, ratherthan infiltrating CD16+ NK cells from the peripheralcirculation. A possible explanation for the identificationof a residual CD56+ cell population in term placentalbed decidua which contrasts with earlier reports, may bethat the early studies identified these cells by their phlox-inophilic cytoplasmic granules (Hamperl and Hellweg,1958). Term decidual uNK cells may have been over-looked: we have recently shown a lower proportion ofdecidual CD56+ cells with phloxinophilic cytoplasmicgranules or cytoplasmic immunostaining for perforinand granzyme at 16–20 weeks gestation compared with8–10 and 12–14 weeks gestation, suggesting that aproportion of CD56+ uNK cells at later gestational agesare agranular (Bulmer et al., in press). The loss of lyticgranules by uNK cells in term pregnancy may reflectdifferent roles of these cells during gestation. The pres-ence of uNK cell lytic granules in 1st and 2nd trimesterpregnancy, which is associated with cytolytic activity(Jones et al., 1997), may indicate a role in regulatingEVT invasion and immune surveillance. As pregnancyprogresses the loss of lytic granules could indicate a shiftin function to cytokine production, which may be neededfor labour and parturition. Further investigation of uNKcell function in term decidua would be worthwhile.

CD94 and NKG2A labelling mirrored that for CD56and may reflect a role for this receptor in the immunolog-ical maintenance of pregnancy. For each subject groupthere was no significant difference in CD94 and NKG2Aexpression, suggesting that CD94/NKG2A is the pre-

dominant type II membrane glycoprotein KIR receptorexpressed within the decidua. The similar levels ofCD56 and CD94 expression by decidual leucocytes inall gestational age groups is in agreement with Verma et

product

30 P.J. Williams et al. / Journal of Re

al. (1997) who reported that all CD56bright cells werealso CD94bright. CD94 expression by CD56+ decid-ual leucocytes may explain why cytotoxic responsesare not generated during successful pregnancy, sinceCD94/NKG2A receptor binding to non-classical classI MHC antigen HLA-E expressed by extravillous tro-phoblast cells leads to inhibition of cytolytic activity(Braud et al., 1998; Lee et al., 1998).

The failure to detect a significant difference in leuco-cyte numbers between placental bed and non-placentalbed decidua in normal pregnancy is in agreement withprevious reports from both Khong (1987), who reportedno difference in leucocyte populations between placentalbed and non-placental bed biopsies in multigravid termpregnancies, and Haller et al. (1995), who reported nodifference in T-lymphocyte, B-lymphocyte, macrophageand natural killer cell numbers between decidua basalisand decidua parietalis in early pregnancy. In contrast,Sindram-Trujillo et al. (2003a,b) reported increased per-centages of CD16+ cells and decreased CD56+ uNKcells in decidua basalis compared to decidua parietalisat term. Differences in the findings between the twogroups are likely to be attributable to the different tech-niques used to examine decidual leucocyte populations.The present study used immunohistochemical methodswhich allowed definite identification of decidua pari-etalis and decidua basalis, whereas Sindram-Trujilloet al. (2003a,b) assessed decidual leucocytes by flowcytometry. Furthermore, the biopsy technique used inthe present study to obtain samples of placental bedand non-placental bed decidua allows access to a deeperlayer of the maternal–fetal interface than may be avail-able from the macroscopically dissected material usedby Sindram-Trujillo et al. (2003a,b).

While the present study assessed the major cytotoxicdecidual leucocyte populations, it is important to notethat there are also populations of T regulatory cells(CD4+CD25+FoxP3+) which have been demonstratedthroughout pregnancy (Heikkinen et al., 2004; Tilburgset al., 2008), small numbers of B lymphocytes whichdo not vary during the menstrual cycle or pregnancyas well as small numbers of decidual dendritic cells, ofthe immature phenotype, whose function is as yet notestablished (Gardner and Moffett, 2003).

APAAP double labelling of CD3+CD56+ andCD8+CD56+ leucocytes revealed a small populationof these double-positive cells in both early and latepregnancy decidua, with a significant reduction in both

populations in late pregnancy. The demonstration ofthese double positive cell populations has implicationsfor those reports which have used CD56 positive selec-tion for in vitro studies as these CD56 positive cells

ive Immunology 82 (2009) 24–31

may not represent a pure NK cell population and arelikely to include a population of activated T cells.Double immunohistochemical labelling also identifiedCD161+CD3+ leucocytes in early pregnancy deciduawith a significant increase in late pregnancy decidua.In contrast to other studies which used flow cytome-try (Tsuda et al., 2001), no TCR V�24 positive NKTcells were detected. The CD161+CD3+ population arethought to represent an NKT-like cell, a subset of con-ventional alpha/beta T lymphocytes that express killerreceptors (Godfrey et al., 2004).

Further studies to evaluate the functional role ofthese decidual leucocytes is required, particularly forlater stages of pregnancy which have been largely over-looked in functional studies to date, thereby allowingfurther elucidation of the role of the leucocyte popula-tions in decidua in the immunological maintenance ofpregnancy.

Acknowledgements

The authors thank Professors Alessandro andLorenzo Moretta, Universita di Genova for kindly pro-viding the antibody for NKG2A, all the patients whoconsented to give placental bed biopsies, and also all theclinical staff at the Royal Victoria Infirmary, Newcastle-upon-Tyne for their help in obtaining samples.

Funding: PW was supported by the Dr. WilliamEdmund Harker Foundation, RVI, Newcastle-upon-Tyne.

References

Abadia-Molina, A.C., Ruiz, C., Montes, M.J., King, A., Loke, Y.W.,Olivares, E.G., 1996. Immune phenotype and cytotoxic activityof lymphocytes from human term decidua against trophoblast. J.Reprod. Immunol. 31, 109–123.

Abrahams, V.M., Kim, Y.M., Straszewski, S.L., Romero, R., Mor, G.,2004. Macrophages and apoptotic cell clearance during pregnancy.Am. J. Reprod. Immunol. 51, 275–282.

Apps, R., Gardner, L., Sharkey, A.M., Holmes, N., Moffett, A., 2007.A homodimeric complex of HLA-G on normal trophoblast cellsmodulates antigen-presenting cells via LILRB1. Eur. J. Immunol.37, 1924–1937.

Boyson, J.E., Rybalov, B., Koopman, L.A., Exley, M., Balk, S.P.,Racke, F.K., Schatz, F., Masch, R., Wilson, S.B., Strominger, J.L.,2002. CD1d and invariant NKT cells at the human maternal–fetalinterface. Proc. Natl. Acad. Sci. USA. 99, 13741–13746.

Braud, V.M., Allan, D.S., O’callaghan, C.A., Soderstrom, K.,D’andrea, A., Ogg, G.S., Lazetic, S., Young, N.T., Bell, J.I.,

Phillips, J.H., Lanier, L.L., Mcmichael, A.J., 1998. HLA-E bindsto natural killer cell receptors CD94/NKG2A, B and C. Nature 391,795–799.

Bulmer, J.N., Lash, G.E., 2005. Human uterine natural killer cells: areappraisal. Mol. Immunol. 42, 511–521.

product

B

B

C

C

D

G

G

G

H

H

H

H

H

J

K

L

L

M

M

P

P.J. Williams et al. / Journal of Re

ulmer, J.N., Morrison, L., Longfellow, M., Ritson, A., Pace, D., 1991.Granulated lymphocytes in human endometrium: histochemicaland immunohistochemical studies. Hum. Reprod. 6, 791–798.

ulmer, J.N., Williams, P.J., Lash, G.E. Immune cells in the placentalbed. Int. J. Dev. Biol., in press.

haouat, G., Ledee-Bataille, N., Dubanchet, S., 2007. Immune cellsin uteroplacental tissues throughout pregnancy: a brief review.Reprod. Biomed. Online 14, 256–266.

raven, C.M., Morgan, T., Ward, K., 1998. Decidual spiral arteryremodelling begins before cellular interaction with cytotro-phoblasts. Placenta 19, 241–252.

ivers, M., Bulmer, J., Miller, D., Lilford, R., 1994. Gamma delta Tcells in human decidua. Am. J. Obstet. Gynecol. 171, 578–579.

ardner, L., Moffett, A., 2003. Dendritic cells in the human decidua.Biol. Reprod. 69, 1438–1446.

odfrey, D.I., MacDonald, H.R., Kronenberg, M., Smyth, M.J., Van,K.L., 2004. NKT cells: what’s in a name? Nat. Rev. Immunol. 4,231–237.

reenwood, J.D., Minhas, K., Di Santo, J.P., Makita, M., Kiso, Y.,Croy, B.A., 2000. Ultrastructural studies of implantation sites frommice deficient in uterine natural killer cells. Placenta 21, 693–702.

aller, H., Radillo, O., Rukavina, D., Tedesco, F., Candussi, G.,Petrovic, O., Randic, L., 1993. An immunohistochemical study ofleucocytes in human endometrium, first and third trimester basaldecidua. J. Reprod. Immunol. 23, 41–49.

aller, H., Tedesco, F., Rukavina, D., Radillo, O., Gudelj, L., Beer,A.E., 1995. Decidual–trophoblast interactions: decidual lymphoidcell populations in basal and parietal decidua. J. Reprod. Immunol.28, 165–171.

amperl, H., Hellweg, G., 1958. Granular endometrial stroma cells.Obstet. Gynecol. 11, 379–387.

anna, J., Goldman-Wohl, D., Hamani, Y., Avraham, I., Greenfield,C., Natanson-Yaron, S., Prus, D., Cohen-Daniel, L., Arnon, T.I.,Manaster, I., Gazit, R., Yutkin, V., Benharroch, D., Porgador, A.,Keshet, E., Yagel, S., Mandelboim, O., 2006. Decidual NK cellsregulate key developmental processes at the human fetal–maternalinterface. Nat. Med. 12, 1065–1074.

eikkinen, J., Mottonen, M., Alanen, A., Lassila, O., 2004. Phenotypiccharacterization of regulatory T cells in the human decidua. Clin.Exp. Immunol. 136, 373–378.

ones, R.K., Bulmer, J.N., Searle, R.F., 1997. Cytotoxic activity ofendometrial granulated lymphocytes during the menstrual cycle inhumans. Biol. Reprod. 57, 1217–1222.

hong, T.Y., 1987. Immunohistologic study of the leukocytic infiltratein maternal uterine tissues in normal and preeclamptic pregnanciesat term. Am. J. Reprod. Immunol. Microbiol. 15, 1–8.

ee, N., Llano, M., Carretero, M., Ishitani, A., Navarro, F., Lopez-Botet, M., Geraghty, D.E., 1998. HLA-E is a major ligand for thenatural killer inhibitory receptor CD94/NKG2A. Proc. Natl. Acad.Sci. U.S.A. 95, 5199–5204.

oke, Y.W., King, A., 1995. Human Implantation. Cambridge Univer-sity Press.

offett-King, A., 2002. Natural killer cells and pregnancy. Nat. Rev.Immunol. 2, 656–663.

orii, T., Nishikawa, K., Saito, S., Enomoto, M., Ito, A., Kurai, N.,Shimoyama, T., Ichijo, M., Narita, N., 1993. T-cell receptors are

expressed but down-regulated on intradecidual T lymphocytes.Am. J. Reprod. Immunol. 29, 1–4.

eralbo, E., Alonso, C., Solana, R., 2007. Invariant NKT and NKT-like lymphocytes: two different T cell subsets that are differentiallyaffected by ageing. Exp. Gerontol. 42, 703–708.

ive Immunology 82 (2009) 24–31 31

Pijnenborg, R., Vercruysse, L., Hanssens, M., 2006. The uterine spiralarteries in human pregnancy: facts and controversies. Placenta 27,939–958.

Pittet, M.J., Speiser, D.E., Valmori, D., Cerottini, J.C., Romero, P.,2000. Cutting edge: cytolytic effector function in human circulatingCD8+ T cells closely correlates with CD56 surface expression. J.Immunol. 164, 1148–1152.

Robson, S.C., Simpson, H., Ball, E., Lyall, F., Bulmer, J.N., 2002.Punch biopsy of the human placental bed. Am. J. Obstet. Gynecol.187, 1349–1355.

Saito, S., 2000. Cytokine network at the feto-maternal interface. J.Reprod. Immunol. 47, 87–103.

Scaife, P.J., Bulmer, J.N., Robson, S.C., Innes, B.A., Searle, R.F., 2006.Effector activity of decidual CD8+ T lymphocytes in early humanpregnancy. Biol. Reprod. 75, 562–567.

Sharkey, A.M., Gardner, L., Hiby, S., Farrell, L., Apps, R., Masters,L., Goodridge, J., Lathbury, L., Stewart, C.A., Verma, S., Moffett,A., 2008. Killer Ig-like receptor expression in uterine NK cells isbiased toward recognition of HLA-C and alters with gestationalage. J. Immunol. 181, 39–46.

Sindram-Trujillo, A., Scherjon, S., Kanhai, H., Roelen, D., Claas, F.,2003a. Increased T-cell activation in decidua parietalis comparedto decidua basalis in uncomplicated human term pregnancy. Am.J. Reprod. Immunol. 49, 261–268.

Sindram-Trujillo, A.P., Scherjon, S.A., Van Hulst-Van Miert, P.P., VanSchip, J.J., Kanhai, H.H., Roelen, D.L., Claas, F.H., 2003b. Dif-ferential distribution of NK cells in decidua basalis comparedwith decidua parietalis after uncomplicated human term pregnancy.Hum. Immunol. 64, 921–929.

Takayama, E., Koike, Y., Ohkawa, T., Majima, T., Fukasawa,M., Shinomiya, N., Yamaguchi, T., Konishi, M., Hiraide, H.,Tadakuma, T., Seki, S., 2003. Functional and Vbeta reper-toire characterization of human CD8+ T-cell subsets withnatural killer cell markers, CD56+ CD57− T cells, CD56+CD57+ T cells and CD56− CD57+ T cells. Immunology 108,211–219.

Tilburgs, T., Roelen, D.L., Van Der Mast, B.J., De Groot-Swings,G.M., Kleijburg, C., Scherjon, S.A., Claas, F.H., 2008. Evidencefor a selective migration of fetus-specific CD4+CD25 bright reg-ulatory T cells from the peripheral blood to the decidua in humanpregnancy. J. Immunol. 180, 5737–5745.

Trundley, A., Moffett, A., 2004. Human uterine leukocytes and preg-nancy. Tissue Antigens 63, 1–12.

Tsuda, H., Sakai, M., Michimata, T., Tanebe, K., Hayakawa, S.,Saito, S., 2001. Characterization of NKT cells in human peripheralblood and decidual lymphocytes. Am. J. Reprod. Immunol. 45,295–302.

Vargas, M.L., Santos, J.L., Ruiz, C., Montes, M.J., Aleman, P.,Garcia-Tortosa, C., Garcia-Olivares, E., 1993. Comparison of theproportions of leukocytes in early and term human decidua. Am.J. Reprod. Immunol. 29, 135–140.

Verma, S., King, A., Loke, Y.W., 1997. Expression of killer cellinhibitory receptors on human uterine natural killer cells. Eur. J.Immunol. 27, 979–983.

Von Rango, U., Classen-Linke, I., Raven, G., Bocken, F., Beier,H.M., 2003. Cytokine microenvironments in human first trimesterdecidua are dependent on trophoblast cells. Fertil. Steril. 79,

1176–1186.

Wegmann, T.G., Lin, H., Guilbert, L., Mosmann, T.R., 1993. Bidirec-tional cytokine interactions in the maternal–fetal relationship: issuccessful pregnancy a TH2 phenomenon? Immunol. Today 14,353–356.