Maternal KIR repertoire is not associated with recurrent spontaneous abortion

C.S. Witt1,4,6, J. Goodridge2, M.G. Gerbase-DeLima3, S. Daher5 and F.T. Christiansen1,2,4

1 Department of Clinical Immunology, Royal Perth Hospital, Perth, West Australia, 6000, 2 School of Surgery and Pathology, University of West Australia, Perth, West Australia, 6912, 3 Immunogenetics Division, Pediatrics Department, Universidade Federal de São Paulo, 4 West Australian Institute for Medical Research, Royal Perth Hospital, Perth West Australia, 6000 and 5 Obstetrics Department, Universidade Federal de Sao Paulo, Brazil

6 To whom correspondence should be addressed. Email: campbell.witt{at}health.wa.gov.au


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
BACKGROUND: In view of evidence suggesting an immunological cause of recurrent spontaneous abortions (RSA) and the large number of maternal natural killer (NK) cells present in the pregnant uterus, we investigated the genetic polymorphism of the killer cell immunoglobulin-like receptors (KIR) in women with RSA. METHODS: KIR gene repertoire and KIR2DL4 (a receptor for HLA-G) genotyping were determined by SSP and SSCP respectively, in women experiencing RSA and controls. RESULTS: The KIR repertoire did not differ between RSA patients and controls in terms of: (i) the number of inhibitory receptors; (ii) the number of activating receptors; (iii) the ratio of inhibitory to activating receptors. KIR2DL4, a receptor for HLA-G, has different transmembrane alleles, which produce functionally different phenotypes. The frequency of KIR2DL4 transmembrane genotypes differed significantly between RSA patients and controls (P=0.03). However, although homozygosity for a membrane-bound receptor was more frequent in patients (25%) than controls (10%), other genotypes that would produce the same phenotype were not more frequent in patients than controls. CONCLUSIONS: The data provide little evidence that KIR polymorphism plays a role in predisposition to RSA.

Key words: killer cell immunoglobulin-like receptors/NK cell/recurrent spontaneous abortion


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
There is evidence that recurrent spontaneous abortion (RSA) is immunologically mediated (reviewed in Piccinni et al., 2001Go; Daher et al., 2003Go). The most prevalent lymphocyte in the decidua during the first and second trimesters is the natural killer (NK) cell (Starkey et al., 1988Go; Trundley and Moffett, 2004Go). The precise role of uterine NK (uNK) cells is not clear but there is evidence that their secretion of interferon-{gamma} (IFN-{gamma}) may be involved in maternal blood vessel remodelling (Guimond et al., 1997Go; Ashkar et al., 2000Go; Li et al., 2001Go; Croy et al., 2003Go) and lower numbers of decidual NK cells have been reported in RSA (Yamamoto et al., 1999Go). The fetal trophoblast does not express the polymorphic HLA-A and HLA-B antigens and this may protect it from attack by maternal T cells. However, the lack of class I HLA antigens might be expected to make it susceptible to attack by NK cells. The possibility that uterine NK cells interact with the fetal trophoblast is suggested by the observations that uNK cells have a series of receptors including the KIR family and CD94/NKG2 family (Hiby et al., 1997Go; Verma et al., 1997Go), and the ligands for these receptors, HLA-C, -E and -G, are expressed on trophoblast (Hutter et al., 1996Go; Proll et al., 1999Go; King et al., 2000aGo,bGo). Indeed the frequency of certain null alleles of HLA-G has been reported to be increased in couples with RSA (Aldrich et al., 2001Go). However, it is not clear whether the purpose of these class I ligands is to interact with the activating receptors on NK cells, thereby stimulating them to secrete blood vessel-modifying cytokines, or to interact with inhibitory receptors, thereby protecting the trophoblast from NK-mediated lysis. In either case, it could be hypothesized that as there is considerable genetically determined variation in the repertoire of KIR receptors between different individuals, a particular KIR repertoire might predispose to RSA. Indeed, Varla-Leftherioti et al. (2003)Go reported that RSA women had fewer inhibitory KIR in a study of 26 patients.

KIR2DL4 is an NK cell receptor for HLA-G (Ponte et al., 1999Go; Rajagopalan and Long, 1999Go). KIR2DL4 has been variously reported to have activating activity for IFN-{gamma} secretion (Rajagopalan et al., 2001Go; Kikuchi-Maki et al., 2003Go), inhibitory activity for cytotoxicity (Ponte et al., 1999Go), or activating activity for cytotoxicity (Rajagopalan et al., 2001Go; Goodridge et al., 2003Go; Kikuchi-Maki et al., 2003Go). Part of the explanation for this apparent diversity of functions may lie in the fact that there are two functionally different alleles. One allele codes for a type I membrane receptor whereas the other codes for receptors that are poorly expressed and may be secreted (Rajagopalan et al., 2001Go; Witt et al., 2002Go). This polymorphism might influence the outcome of NK cell interaction with HLA-G on the trophoblast.

In this study, we determined the KIR repertoire in 52 Brazilian RSA patients. In addition we examined the frequency of the alternate KIR2DL4 genotypes in these patients.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Subjects
The patient group comprised 51 Brazilian Caucasian women (aged 18–45 years) with unexplained RSA consecutively referred. All of them had a history of at least three spontaneous miscarriages (mean 4, range 3–7) and no previous successful pregnancy. They all attended the Outpatient Clinic of Spontaneous Abortion of the Obstetrics Department of UNIFESP-EPM, São Paulo, Brazil. The clinic is a referral centre that receives only women with a history of habitual abortion. Of 110 patients referred, 59 were excluded because the cause of the miscarriage was identified. Thus, the studied patients represent 46% of the whole RSA population. The following tests were performed to exclude known causes of abortion: hysteroscopy, hysterosalpingography, serial ultrasound, parental karyotypes, investigation of luteal phase insufficiency (repeated serum progesterone concentration measurements and endometrial biopsy), prolactin dosage, glycaemic curve, thyroid hormone levels, investigation of toxoplasmosis, cytomegalovirus, rubella, HIV, group B streptococci, Chlamydia trachomatis, hepatitis B and C and bacterial vaginosis, investigation of antiphospholipid antibodies (anticardiolipin antibody and lupus anticoagulant). The control group consisted of women (aged 18–45 years) who have had previously had at least two successful pregnancies with no history of miscarriage, pre-eclampsia, ectopic pregnancy or preterm delivery.

A group of 51 patients with RSA and 55 controls was used for KIR repertoire. For KIR2DL4 genotyping, 52 patients with RSA and 72 controls were used. Different cohorts were evaluated because the sample amount in some cases was not enough to perform both tests. There were 19 patients and 43 controls in common between the groups. The protocol of this investigation was approved by the Medical Ethics Committee of our Institution and informed consent to participate in the study was obtained from all individuals.

KIR repertoire
The presence or absence of genes for each of the KIR receptors was determined by PCR–SSP using the primers shown in Table I. Two separate reactions were included in the detection of KIR2DL1 as outlined in Table I. The ‘2DL1’ primers detect all alleles of KIR2DL1 except those found on the ‘B’ KIR haplotype. The ‘2DL1v’ primers detect all alleles of KIR2DL1 except for a rare variant found on some ‘A’ haplotypes. Samples giving positive reactions with either 2DL1 or 2DL1v primers were considered to have the KIR2DL1 gene. As the majority of KIR2DS4 genes contain a 22 bp deletion and are not translated, primers that amplify only the expressed alleles of KIR2DS4 were used. Each PCR reaction contained 12.5 µl patient DNA at 20 ng/µl, 2.5 µl 10 x PCR buffer, 0.2 IU Taq polymerase (Amplitaq; Applied Biosystems, USA), 1 µl 40 mmol/l dNTP, 10 pmol of each primer and water to a final volume of 25 µl.


View this table:
[in this window]
[in a new window]
 
Table I. Primer sequences and thermocycling profiles used for amplification of KIR genes

 
Thermocycling program 44 consisted of denaturation at 95°C for 5 min, five cycles of denaturation at 97°C for 20 s, annealing at 64°C for 45 s, extension at 72°C for 90 s, followed by 25 cycles of denaturation at 97°C for 20 s, annealing at 60°C for 45 s, extension at 72°C for 90 s, followed by a further extension at 72°C for 10 min. Program 39 was the same as program 44 except that the annealing temperatures were 66°C for the first five cycles and 62°C for 25 cycles. In program ‘2DL5’, the annealing temperatures were 67 and 63°C. In program ‘2DL4’, the annealing temperatures were 62 and 52°C. The MgCl2 concentration was 25 mmol/l for most genes and 20 mmol/l for KIR2DL1, KIR2DL1v, KIR2DL2, KIR3DL1, KIR2DL5 and KIR2DS4. Every PCR reaction included primers for human growth hormone (hGH) as a positive control for that reaction. Results could not be obtained for 2DL1, 2DL2, 2DL3, 2DL5, 3DL1, 2DS1, 2DS2, 2DS3, 2DS4, 2DS5, 3DS1 in 1, 3, 5, 0, 1, 0, 0, 4, 3, 0, 0 controls and 1, 0, 2, 1, 1, 0, 3, 3, 3, 5, 1 patients respectively, as these samples failed to amplify adequately as judged by the absence of a PCR product for the hGH primers. Only one patient sample failed to amplify with KIR2DL4 primers.

KIR2DL4 genotyping
KIR2DL4 genotyping was performed as previously described (Witt et al., 2000Go). Briefly, a stretch of DNA including exon 6, intron 6 and exon 7 was amplified by PCR and the genotype of the PCR products was determined by capillary electrophoresis.

Statistical analysis
Differences between patients and controls in the frequency of each KIR gene were tested by the conventional {chi}2-test. Differences in the total number of activating, inhibitory KIR, the ratio of activating to inhibitory KIR, and the distribution of KIR2DL4 genotypes was assessed by {chi}2-test using the CLUMP program (Sham and Curtis, 1995Go).


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
KIR repertoire
There was no significant difference in the frequency of any KIR gene between controls and patients (Table II). There was no significant difference in the total number of inhibitory KIR, nor in the total number of activating KIR between patients and controls (Figure 1a, b). Thirty-five per cent of controls and 50% of patients had all three HLA-C inhibitory KIR (P=0.22) (KIR2DL1, KIR2DL2, KIR2DL3). There was no significant difference in the proportion of patients and controls having the ‘A’, ‘B’ or ‘AB’ genotypes, where A is defined by the presence of KIR2DL3 and absence of KIR2DL2, B is defined as the absence of KIR2DL3 and presence of KIR2DL2, and AB is defined by the presence of both KIR2DL3 and KIR2DL2. Figure 2 shows that there was no difference in the ratio of activating to inhibitory KIR between patients and controls.


View this table:
[in this window]
[in a new window]
 
Table II. Frequency of each KIR gene in recurrent spontaneous abortion patients and controls

 


View larger version (22K):
[in this window]
[in a new window]
 
Figure 1. (a) Number of inhibitory KIR per individual in patients and controls. The number of inhibitory KIR per individual does not differ significantly between patients and controls (P>0.05). (b) Number of activating KIR per individual in patients and controls. The number of activating KIR per individual does not differ significantly between patients and controls (P>0.05).

 


View larger version (17K):
[in this window]
[in a new window]
 
Figure 2. Number of activating and inhibitory KIR in patients (PATS) and controls (CTLS). The number of activating (Act) KIR is indicated on the top line of the horizontal axis label and the number of inhibitory (Inhib) KIR is shown on the bottom line of the horizontal axis label. The distribution of activating and inhibitory KIR does not differ significantly between patients and controls (P>0.05).

 
KIR2DL4 genotype
Table III shows the frequency of KIR2DL4 transmembrane genotypes in patients and controls.


View this table:
[in this window]
[in a new window]
 
Table III. Frequency of KIR2DL4 transmembrane genotypes in patients and controls

 
The ‘B’ allele has the wild-type sequence in exons 6 and 7 and differs from the ‘A’ allele by two SNP in intron 6, but does not produce a membrane-bound receptor (C.S.Witt, J.Goodridge, F.T.Christiansen; unpublished dataGo). The overall distribution of KIR2DL4 genotypes was significantly different between patients and controls (P=0.03). The frequency of the AA genotype, which corresponds to homozygosity for a membrane-bound receptor phenotype, was the most divergent between patients (25%) and controls (10%). However, the difference in frequency for this genotype compared to all others was not significantly different (P=0.14). If a membrane-bound receptor phenotype predisposed to RSA, then it might be expected that other genotypes encoding a membrane-bound receptor would also be more frequent among the RSA patients. Although heterozygous, the AB and AC genotypes encode a membrane-bound receptor, but these genotypes were less frequent in the patients than controls. Thus, the increased frequency of the AA genotype among RSA patients, suggesting the importance of a membrane-bound receptor, is not supported by the frequencies of the AB and AC genotypes. In addition, the BC genotype, which is more frequent among patients than controls, does not encode a membrane-bound receptor. Thus there does not appear to be any consistent phenotype associated with the RSA.


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
No association between KIR repertoire or KIR2DL4 genotype could be demonstrated in this study. Varla-Leftherioti et al. (2003)Go reported that RSA patients have fewer HLA-C inhibitory KIR. Specifically, only 31% of RSA patients but 69% of controls had all three HLA-C inhibitory KIR. Furthermore, 19% of patients had only a single HLA-C inhibitory KIR whereas this was true for only 4% of controls. This contrasts with the present study in which a greater number of patients than controls possessed all three HLA-C inhibitory KIR and none of the patients or controls were found to have only a single HLA-C inhibitory KIR. Although the primers used for the amplification of KIR2DL1 by Varla-Leftherioti et al. (2003)Go were not specified, the relatively high frequency of KIR2DL1-negative samples suggests that the primers described by Uhrberg et al. (1997)Go were used. The Uhrberg primers only amplify the alleles of KIR2DL1 found on the A haplotype and do not amplify the KIR2DL1*004 allele which is commonly found on the KIR B haplotype (Valiante et al., 1998Go) in association with KIR2DL2. In the present study we used both the Uhrberg primers and the primers described by Gomez-Lozano and Vilches (2002)Go which, between them, detect all known KIR2DL1 variants. However, the difference in results between this and the Varla-Leftherioti study cannot be attributed to the difference in KIR2DL1 primers, as our values changed minimally when we only counted Uhrberg primer-positive samples as being positive for KIR2DL1. [Only subjects who are homozygous for the B KIR haplotype (KIR2DL2-containing haplotype) would be expected to be discrepant for the Uhrberg and Gomez-Lozano primers, and such individuals are rare.]

The AA KIR2DL4 transmembrane genotype, which represents homozygosity for a membrane-bound receptor for HLA-G, was marginally increased in frequency in RSA patients. However, the other genotypes that would be expected to code for a membrane-bound receptor were not increased in RSA patients, suggesting that the increase in the AA genotype is most likely a type I error. The absence of any association of KIR2DL4 genotype with RSA or with pre-eclampsia (Witt et al., 2002Go) suggests that if KIR2DL4 has a critical role in pregnancy, then it must be only in rare circumstances. This is consistent with the finding of a woman who was homozygous null for KIR2DL4 but who had had multiple successful pregnancies (Gomez-Lozano et al., 2003Go).

Although no association between KIR genotype and RSA was observed, and no disturbance of HLA-C allele frequencies was observed in a previous study (Christiansen et al., 1997Go), it may be premature to conclude that KIR receptors are not relevant to pregnancy or RSA. As the KIR receptors would be expected to interact with HLA-C on the trophoblast, it will be necessary to study the KIR repertoire in RSA cases in which the HLA-C genotype of the fetus can be established in order to determine whether the presence or absence of certain KIR receptor–HLA ligand interactions are necessary. HLA-B alleles are believed not to be expressed on the trophoblast, but a lack of HLA-B alleles with the Bw4 epitope has been observerd in RSA couples (Christiansen et al., 1997Go). It will therefore be important that future studies do not neglect any possible interaction between Bw4 on the fetus and the maternal NK receptors for Bw4, KIR3DL1 and KIR3DS1.


    Acknowledgements
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
This work was supported by grants from the West Australian Institute for Medical Research and the National Health and Medical Research Council of Australia.


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Aldrich CL, Stephenson MD, Karrison T, Odem RR, Branch DW, Scott JR, Schreiber JR and Ober C (2001) HLA-G genotypes and pregnancy outcome in couples with unexplained recurrent miscarriage. Mol Hum Reprod 7, 1167–1172.[Abstract/Free Full Text]

Ashkar AA, Di Santo JP and Croy BA (2000) Interferon-{gamma} contributes to initiation of uterine vascular modification, decidual integrity and uterine NK cell maturation during normal murine pregnancy. J Exp Med 192, 259–270.[Abstract/Free Full Text]

Christiansen OB, Mohapeloa HP, Steffensen R and Jersild C (1997) HLA-C and -Bw typing of couples with unexplained recurrent miscarriages. J Reprod Immunol 37, 63–77.[CrossRef][ISI][Medline]

Croy BA, Esadeg S, Chantakru S, van den HM, Paffaro VA, He H, Black GP, Ashkar AA, Kiso Y and Zhang J (2003) Update on pathways regulating the activation of uterine natural killer cells, their interactions with decidual spiral arteries and homing of their precursors to the uterus. J Reprod Immunol 59, 175–191.[CrossRef][ISI][Medline]

Daher S, Shulzhenko N, Morgun A, Mattar R, Rampim GF, Camano L and DeLima MG (2003) Associations between cytokine gene polymorphisms and recurrent pregnancy loss. J Reprod Immunol 58, 69–77.[CrossRef][ISI][Medline]

Gomez-Lozano N and Vilches C (2002) Genotyping of human killer-cell immunoglobulin-like receptor genes by polymerase chain reaction with sequence-specific primers: an update. Tiss Antigens 59, 184–193.[CrossRef][ISI]

Gomez-Lozano N, De Pablo R, Puente S and Vilches C (2003) Recognition of HLA-G by the NK cell receptor KIR2DL4 is not essential for human reproduction. Eur J Immunol 33, 639–644.[CrossRef][ISI][Medline]

Goodridge JP, Witt CS, Christiansen FT and Warren HS (2003) KIR2DL4 (CD158d) genotype influences expression and function in NK cells. J Immunol 171, 1768–1774.[Abstract/Free Full Text]

Guimond M-J, Luross JA, Wang B, Terhost C, Danial S and Croy BA (1997) Absence of natural killer cells druing murine pregnancy is associated with reproductive compromise in TgE26 mice. Biol Reprod 56, 169–179.[Abstract]

Hiby SE, King A, Sharkey AM and Loke YW (1997) Human uterine NK cells have a similar repertoire of killer inhibitory and activatory receptors to those found in blood, as demonstrated by RT–PCR and sequencing. Mol Immunol 34, 419–430.[CrossRef][ISI][Medline]

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. [Published erratum appears in Cell Tiss Res (1997) Mar 287(3) 625.][CrossRef][ISI][Medline]

Kikuchi-Maki A, Yusa S, Catina TL and Campbell KS (2003) KIR2DL4 is an IL-2-regulated NK cell receptor that exhibits limited expression in humans but triggers strong IFN-gamma production. J Immunol 171, 3415–3425.[Abstract/Free Full Text]

King A, Allan DS, Bowen M, Powis SJ, Joseph S, Verma S, Hiby SE, McMichael AJ, Loke YW and Braud VM (2000a) HLA-E is expressed on trophoblast and interacts with CD94/NKG2 receptors on decidual NK cells. Eur J Immunol 30, 1623–1631.[CrossRef][ISI][Medline]

King A, Burrows TD, Hiby SE, Bowen JM, Joseph S, Verma S, Lim PB, Gardner L, Le Bouteiller P, Ziegler A et al. (2000b) Surface expression of HLA-C antigen by human extravillous trophoblast. Placenta 21, 376–387.[CrossRef][ISI][Medline]

Li XF, Charnock-Jones DS, Zhang E, Hiby S, Malik S, Day K, Licence D, Bowen JM, Gardner L, King A et al. (2001) Angiogenic growth factor messenger ribonucleic acids in uterine natural killer cells. J Clin Endocrinol Metab 86, 1823–1834.[Abstract/Free Full Text]

Piccinni MP, Scaletti C, Vultaggio A, Maggi E and Romagnani S (2001) Defective production of LIF, M-CSF and Th2-type cytokines by T cells at fetomaternal interface is associated with pregnancy loss. J Reprod Immunol 52, 35–43.[CrossRef][ISI][Medline]

Ponte M, Cantoni C, Biassoni R, Tradori-Cappai A, Bentivoglio G, Vitale C, Bertone S, Moretta A, Moretta L and Mingari MC (1999) Inhibitory receptors sensing HLA-G1 molecules in pregnancy: decidua-associated natural killer cells express LIR-1 and CD94/NKG2A and acquire p49, an HLA-G1-specific receptor [see comments]. Proc Natl Acad Sci USA 96, 5674–5679.[Abstract/Free Full Text]

Proll J, Blaschitz A, Hutter H and Dohr G (1999) First trimester human endovascular trophoblast cells express both HLA-C and HLA-G. Am J Reprod Immunol 42, 30–36.[ISI][Medline]

Rajagopalan S and Long EO (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]

Rajagopalan S, Fu J and Long EO (2001) Cutting edge: induction of IFN-gamma production but not cytotoxicity by the killer cell Ig-like receptor KIR2DL4 (CD158d) in resting NK cells. J Immunol 167, 1877–1881.[Abstract/Free Full Text]

Sham PC and Curtis D (1995) Monte Carlo tests for associations between disease and alleles at highly polymorphic loci. Ann Hum Genet 59, 97–105.[ISI][Medline]

Starkey PM, Sargent IL and Redman CWG (1988) Cell populations in human early pregnancy decidua: characterization and isolation of large granular lymphocytes by flow cytometry. Immunology 65, 129–134.[ISI][Medline]

Trundley A and Moffett A (2004) Human uterine leukocytes and pregnancy. Tiss Antigens 63, 1–12.[CrossRef][ISI]

Uhrberg M, Valiante NM, Shum BP, Shilling HG, Lienert-Weidenbach K, Corliss B, Tyan D, Lanier LL and Parham P (1997) Human diversity in killer cell inhibitory receptor genes. Immunity 7, 753–763.[ISI][Medline]

Valiante NM, Uhrberg M, Shilling HG, Lienert-Weidenbach K, Arnett KL, D'Andrea A, Phillips JH, Lanier LL and Parham P (1998) Functionally and structurally distinct NK cell receptor repertoires in the peripheral blood of two human donors. Immunity 7, 739–751.[CrossRef][ISI]

Varla-Leftherioti M, Spyropoulou-Vlachou M, Niokou D, Keramitsoglou T, Darlamitsou A, Tsekoura C, Papadimitropoulos M, LePage V, Balafoutas C and Stavropoulos-Giokas C (2003) Natural killer (NK) cell receptors' repertoire in couples with recurrent spontaneous abortions. Am J Reprod Immunol 49, 183–191.[ISI][Medline]

Verma S, King A and Loke YW (1997) Expression of killer cell inhibitory receptors on human uterine natural killer cells. Eur J Immunol 27, 979–983.[ISI][Medline]

Witt CS, Martin A and Christiansen FT (2000) Detection of KIR2DL4 alleles by sequencing and SSCP reveals a common allele with a shortened cytoplasmic tail. Tiss Antigens 56, 248–257.[CrossRef][ISI]

Witt CS, Whiteway JM, Warren HS, Barden A, Rogers M, Martin A, Beilin L and Christiansen FT (2002) Alleles of the KIR2DL4 receptor and their lack of association with pre-eclampsia. Eur J Immunol 32, 18–29.[CrossRef][ISI][Medline]

Yamamoto T, Takahashi Y, Kase N and Mori H (1999) Decidual natural killer cells in recurrent spontaneous abortion with normal chromosomal content. Am J Reprod Immunol 41, 337–342.[ISI][Medline]

Submitted on May 25, 2004; accepted on July 26, 2004.