The Genotype of the NK Cell Receptor, KIR2DL4, Influences Infg Secretion

Molecular Human Reproduction, Vol.15, No.8 pp. 489–497, 2009 Advanced Access publication on June 9, 2009 doi:10.1093/molehr/gap039

ORIGINAL RESEARCH The genotype of the NK cell receptor, KIR2DL4, inﬂuences INFg secretion by decidual natural killer cells

J.P. Goodridge1,†, L.J. Lathbury2,†, E. John3, A.K. Charles2,4, Downloaded from https://academic.oup.com/molehr/article/15/8/489/1252362 by guest on 25 September 2021 F.T. Christiansen2,3, and C.S. Witt3,5 1Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue North Seattle, Seattle, WA 98109, Australia 2School of Pathology and Laboratory Medicine, University of Western Australia, Nedlands, WA 6009, Australia 3Department of Clinical Immunology and Immunogenetics, PathWest, Royal Perth Hospital, Wellington St, Perth, WA 6847, Australia 4Department of Pathology, PathWest, King Edward Memorial/Princess Margaret Hospitals, Subiaco, WA 6008, Australia

5Correspondence address. Tel: þ61-8-9224-2899; Fax: þ61-8-9224-2920; E-mail: [email protected]

abstract: Natural killer (NK) cells are the predominant leukocyte in first trimester decidua and play a role in vascular remodelling through interferon gamma (IFNg) secretion. Membrane expression of the killer immunoglobulin-like receptor (KIR) KIR2DL4 on peripheral blood NK (pNK) cells is controlled by the 9A/10A transmembrane genetic polymorphism. On peripheral NK cells (pNK), KIR2DL4 can only be detected on the membrane of cells from individuals with at least one copy of the 10A allele and ligation of KIR2DL4 results in IFNg secretion. In this study, we assessed KIR2DL4 expression and IFNg secretion as a result of KIR2DL4 ligation, by decidual NK (dNK) cells. The 9A/10A transmembrane polymorphism was shown to control KIR2DL4 expression by dNK, as previously shown for pNK cells. Freshly isolated dNK cells from subjects with at least one 10A allele expressed KIR2DL4 whereas those from 9A homozygous subjects did not. Although freshly isolated dNK did not secrete IFNg in response to KIR2DL4 ligation regardless of KIR2DL4 genotype, activation by in vitro culture with IL-2 enabled dNK cells from individuals with at least one 10A allele, but not those without a 10A allele, to secrete IFNg in response to KIR2DL4 ligation. This study confirms that expression of KIR2DL4 by dNK is dependent on the 9A/10A polymorphism and that this polymorphism influences IFNg secretion by dNK cells.

Key words: CD158d / cell surface molecules / cytokines / KIR2DL4 / natural killer cells

Introduction et al., 2002) although there is as yet no evidence of a role for IFNg in humans. As ligation of KIR2DL4 on peripheral blood NK (pNK) NK cells are the predominant lymphoid cell population at the site of cells by anti-KIR2DL4 monoclonal antibody (mAb) results in IFNg implantation, and constitute 70% of the infiltrating CD45þ leukocytes secretion (Rajagopalan et al., 2001; Goodridge et al., 2007), in first trimester decidua (Moffett-King, 2002). Receptors that enable KIR2DL4 expression on dNK cells may play a role in placentation. decidual NK (dNK) cells to recognize HLA class I molecules include Little is known about the expression of KIR2DL4 on dNK cells, the lectin-like family (CD94/NKG2), leukocyte Ig-like receptor-1, although it has been detected on such cells using a polyclonal antibody and the immunoglobulin-like receptor (KIR) family (Bashirova et al., in a single report (Ponte et al., 1999). In contrast, much is now known 2006). It has been postulated that activation of dNK cells via these about the expression of KIR2DL4 on pNK cells. KIR2DL4 is only receptors may be an important event in human placentation. expressed on the surface of freshly isolated CD56bright pNK cells, Indeed, particular combinations of maternal KIR and fetal HLA-C although it can be induced on CD56dim pNK cells by culture (Good- alleles have been shown to predispose to pre-eclampsia and recurrent ridge et al., 2003; Kikuchi-Maki et al., 2003). Based on the presence of miscarriage (Hiby et al., 2004, 2008), syndromes characterized by a sequence of 9 or 10 adenine residues at the end of exon 6, two shallow placentation. The immunological events leading to pathological common allele groups, 9A and 10A, each having a gene frequency syndromes of pregnancy such as pre-eclampsia are not well-defined. In of approximately 0.5, have been described (Witt et al. 2000, 2002). the mouse, it has been demonstrated that interferon gamma (IFNg) The deletion of one adenine in 9A alleles results in a frame-shift and secretion by dNK cells is an important event in placentation (Croy the production of either a protein with a truncated cytoplasmic tail

† These authors contributed equally to the work described in this manuscript.

& The Author 2009. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved. For Permissions, please email: [email protected] 490 Goodridge et al. or one lacking the transmembrane region, both of which are not IgM (Beckman Coulter, Australia) and IgG2a (R&D Systems). Leaf-purified expressed at the cell surface. In contrast, the 10A alleles, which can anti-human CD16 (IgG1, clone 3G8, Biolegend, USA) and an equivalent be further divided into the 10A-A and 10A-B subgroups, encode IgG1 isotype control (clone MG1-45, Biolegend) were used for the receptors that may be expressed at the cell surface (Goodridge in vitro stimulation of NK cells. The secondary antibody used with mAb et al., 2003). Monoclonal anti-KIR2DL4 antibodies detect KIR2DL4 #33 to detect KIR2DL4 was FITC-conjugated Affinipure F(Ab’)2 fragment goat anti-mouse IgG (Fcg specific, Beckman Coulter, Australia). The sec- expression on the CD56bright population of freshly isolated pNK ondary antibody used with mAb #64 was FITC-conjugated goat anti- cells from individuals who have at least one copy of the 10A-A mouse IgM (m-chain specific, Beckman Coulter) and when mAb #2238 allele. The receptor produced by the 10A-B allele is expressed on was used to detect KIR2DL4, biotinylated goat anti-mouse IgG2a these cells but cannot be detected because the available mAbs (Southern Biotechnology Associates, USA), followed by streptavidin- react with the D0-domain of KIR2DL4 and the mRNA produced by DTAF (Beckman Coulter) was used. Expression of CD9, a dNK cell the 10A-B allele in freshly isolated CD56bright NK omits the marker (Koopman et al., 2003), was identified using FITC-conjugated anti- D0-domain (Goodridge et al., 2003). However, after in vitro culture body (clone P1/33/2, Dako Cytomation, Australia). Downloaded from https://academic.oup.com/molehr/article/15/8/489/1252362 by guest on 25 September 2021 of pNK cells, the full-length 10A-B receptor, including the D0 domain, is expressed and can be detected. Flow cytometry Although controversial (Allan et al., 1999; Boyson et al., 2002), When unconjugated primary antibodies were used, incubations with the there is evidence that HLA-G is the ligand for KIR2DL4 (Ponte primary and secondary antibody were completed prior to staining with et al., 1999; Rajagopalan and Long, 1999; Rajagopalan et al., 2005). directly conjugated antibodies. Each tube of cells was stained for three Since IFNg is an important mediator of vascular changes associated markers, including directly conjugated anti-CD56 and anti-CD3 antibodies with placentation, interaction between KIR2DL4 on dNK cells and (Beckman Coulter) for NK cell identification. The third antibody to detect HLA-G on trophoblast cells may be important for IFNg secretion a third marker was either a directly conjugated antibody or an unconjugated anti-KIR2DL4 antibody that was detected as described above. All and placental development. If KIR2DL4 is not expressed on the mem- antibodies were incubated for 30 min on ice, followed by two washes. brane of dNK cells from individuals who are homozygous for the 9A NK cells were identified as CD56þ, CD32 lymphocytes and CD56bright allele, then these cells may secrete less IFNg, thereby affecting placen- and CD56dim subsets were clearly distinguished. CD158a (clone tal development. The aims of this study were to determine whether HP-3E4), CD158b (clone CH-L) and CD158e (clone DX9) were detected KIR2DL4 is expressed on dNK cells using anti-KIR2DL4 mAbs and using PE-labelled antibodies (BD Biosciences, Franklin Lakes, USA). Flow to determine whether the 9A/10A polymorphism influences mem- cytometry was performed on a CoulterR EPICS XL-MCLTM flow cyt- brane expression and IFNg secretion by dNK cells. ometer (Beckman Coulter).

NK cell culture Materials and Methods 3 RosetteSep-purified, freshly isolated dNK or pNK cells (5 10 cells per well) were cultured for 8 days with g-irradiated Daudi cells (3 104 cells Cells per well, 60 Gy) in the presence of 200 U/ml recombinant human IL-2 Peripheral blood and placental tissue samples were obtained with consent (Chiron, USA). These proliferating, primary cultures of NK cells were from 56 donors undergoing elective first trimester termination of normal maintained by subculture every second day with fresh medium. Secondary pregnancy at the Marie Stopes International Clinic (Perth, Australia). NK cell cultures were established by restimulating primary cultured NK Ethics approval was obtained from the Human Ethics Committee, Univer- cells (3 103 cells per well) with 3 104 g-irradiated Daudi cells per sity of Western Australia. pNK cells were isolated from blood samples by well in the presence of 200 U/ml rIL-2 for 12 days. Prior to any functional Ficoll density gradient centrifugation with RosetteSep human NK cell assay, flow cytometric analysis of both primary and secondary cultured NK enrichment cocktail (Stem Cell Technologies, Vancouver, Canada). Decid- cells for CD56, CD3 and KIR2DL4 expression was performed. ual tissue was macroscopically dissected from non-decidual tissue, and mononuclear cells were isolated by mechanical disruption, followed by fil- SSCP for KIR2DL4 9A/10A transmembrane tration through a 70 mm sieve and centrifugation over Ficoll. dNK cells were then obtained by incubating the decidual mononuclear cells with genotype the RosetteSep human NK cell enrichment antibody cocktail and 200 ml DNA was extracted from 200 ml of buffy coat cells using a QIAamp DNA peripheral blood (g-irradiated to 30 Gy to prevent proliferation of con- Blood Mini Kit (Qiagen, Australia). Single stranded conformational poly- taminating peripheral NK cells) from the tissue donor as a source of red morphism (SSCP) was used to distinguish the three different transmem- blood cells (a specific ratio between mononuclear cells and red blood brane genotypes of exons 6–7 as described previously (Witt et al., 2000). cells is required for the purification of NK cells via the RosetteSep pro- cedure), followed by a second centrifugation over Ficoll. Due to limited Stimulation of cells for IFNg production cell numbers, not all of the assays described below were performed on and IFNg ELISA all samples. Non-tissue culture-treated, flat-bottomed, 96 well plates (BD Biosciences) were coated overnight with antibody or isotype control diluted in bicar- Antibodies bonate buffer (pH 9.6) and then washed twice with tissue culture Expression of KIR2DL4 on the cell surface and stimulation of NK cells medium. Freshly isolated, purified NK cells were added directly to the through KIR2DL4 was examined using several different anti-KIR2DL4 stimulation plates. Secondary cultured NK cells were harvested from mAbs. MAb #33 (IgG1) and #64 (IgM), were both kindly provided by their original culture wells, washed and then added to the stimulation Dr E. Long and S. Rajagopalan (Laboratory of Immunogenetics, NIH, plates. The plates were then incubated for 20 h at 378C. In some exper- MD, USA). Mab mAb #2238 was purchased from R&D Systems (IgG2a, iments, soluble mAb #33 or IgG1 isotype control antibodies were clone 181703, R&D Systems, USA). Isotype controls included IgG1 and included in the tissue culture medium at a concentration of 1 mg/ml, KIR2DL4 genotype influences INFg secretion 491 previously shown to be optimal for IFNg stimulation. Cell viability was not altered as a result of the stimulation. Supernatants were harvested and IFNg was quantified using the OPTeia Human IFNg ELISA (BD Pharmin- gen, USA).

Statistics Comparisons of the percentage of KIR2DL4 positive cells between individuals with different KIR2DL4 genotypes were performed by 2-tailed t-tests. Comparisons of fold-increase in IFNg secretion were between cell types, between KIR2DL4 genotypes and between different stimulating antibodies were performed by 2-tailed t-tests. Downloaded from https://academic.oup.com/molehr/article/15/8/489/1252362 by guest on 25 September 2021 Results Figure 1 dNK cells are CD56 bright, CD9 positive. NK cells were isolated from decidua and the CD9 phenotype of each population was determined. Cells with a lymphocyte profile on the FSC, SSC plot Credentials of anti-KIR2DL4 mAbs were selected for analysis. Dot plots of decidual cell populations from two The reactivity of the anti-KIR2DL4 mAbs has been previously published donors (38 and 20, in A and B, respectively) are shown. A relatively pure bright þ and is summarized here. According to the manufacturer, mAb #2238 CD56 , CD9 population was obtained for sample 38 (A), whereas the preparation for donor 20 contained relatively fewer CD56bright, CD9þ cells detects KIR2DL4 on transfected BaF/3 cells and does not recognize (B). The numbers represent the percentage of gated lymphocytes in each transfectants expressing KIR2DL1, 2DL2, 2DL3, 2DL5, 2DS1, 2DS2, quadrant of the dot plots. 2DS4, 3DL1, 3DL2 or 3DS1. MAbs #33 and #64 have also been shown to be specific for KIR2DL4 in similar transfection experiments (Rajagopalan et al., 2001). We have compared the reactivity of the levels of CD9. However, a few dNK preparations (Fig. 1B), had a sig- monoclonal anti-KIR2DL4 antibodies, mAb #33, #64 and #2238 by nificant number of cells that were not CD56bright and which were testing them on freshly isolated pNK cells from subjects representing CD92. dNK cell preparations with ,50% CD56bright cells were con- all KIR2DL4 genotypes (homozygous 10A-A, 10A-B, 9A and all hetero- sidered to be possibly contaminated with peripheral blood cells and zygous combinations) (Goodridge et al., 2007 and unpublished obser- were excluded from functional studies. vations). All three antibodies gave equivalent results staining only the To further verify the identity of the cells in the RosetteSep purified CD56bright pNK cells of individuals with at least one 10A-A allele. populations, we examined the expression of three KIR that are known None of the mAb stained freshly isolated CD56dim pNK cells. All to be expressed on dNK cells; namely CD158a (KIR2DL1/2DS1), three antibodies also gave equivalent staining of in vitro activated pNK CD158b (KIR2DL2/3/2DS2) and CD158e (KIR3DL1). Expression cells from donors representing all genotypes (Goodridge et al., 2007 on freshly isolated CD56bright dNK was compared with that of and unpublished observations). That is, the majority of in vitro activated CD56dim pNK cells from the same donors. As shown in Fig. 2, the NK cells became KIR2DL4 positive but this only occurred for pNK cells proportion of cells expressing each of these receptors, particularly from donors with at least one 10A allele. In all cases staining was of only CD158a and CD158b, on freshly isolated CD56bright dNK cells was moderate intensity with no clear bimodal staining of distinct positive and generally higher than that observed for their matched CD56dim pNK negative cell populations. The only difference between the mAbs was cells. These results are in agreement with previous reports (Hiby that mAb #33 and mAb #2238 differed in their ability to induce IFNg et al., 1997; Verma et al., 1997) and are consistent with the majority secretion from pNK cells (unpublished observations). Although both of cells in these preparations being dNK cells (Verma et al., 1997). antibodies induced IFNg secretion from NK-92 and cultured pNK cells when solid-phase bound (SPB), only mAb #33 induced IFNg from freshly isolated peripheral NK cells when used as a soluble Freshly isolated dNK cells express KIR2DL4 ligand. MAb #64 has also been reported to stimulate IFNg when in accordance with their transmembrane used as a soluble ligand (Rajagopalan et al., 2001) but has not been genotype tested in this laboratory. As none of the available mAb are immunopre- We have previously shown that membrane expression of KIR2DL4 is cipitating, it is not possible to formally exclude cross reactivity with relatively weak compared with other NK cell markers and is restricted other surface antigens on primary NK cells. to the CD56bright subpopulation of freshly isolated pNK cells from subjects with at least one 10A-A allele of the transmembrane domain Identification of dNK cells (Goodridge et al., 2003). We therefore investigated whether this poly- To confirm the identity of the cells isolated from the decidual tissue morphism also influences KIR2DL4 membrane expression on dNK following the RosetteSep NK cell separation, samples were stained cells. The level of KIR2DL4 expression by CD56bright dNK cells iso- for CD56 and CD9, a marker that has recently been found on lated from 35 decidual tissue samples were determined and related freshly isolated endometrial NK cells (Eriksson et al., 2004) and to transmembrane genotype. dNK cells (Koopman et al., 2003). Representative examples are pre- As we have shown for CD56bright pNK cells (Goodridge et al., sented in Fig. 1. Most dNK cell preparations (Fig. 1A), in agreement 2003), dNK cells also exhibited low level expression of KIR2DL4 with previously published data (Trundley and Moffett, 2004), con- and the level of expression varied between individuals. In Fig. 3 dot tained a majority of cells that were CD56bright and expressed high plots of representative examples showing low (upper panels) and 492 Goodridge et al. Downloaded from https://academic.oup.com/molehr/article/15/8/489/1252362 by guest on 25 September 2021

Figure 2 KIR are expressed on a greater proportion of dNK cells than pNK cells. KIR (CD158a, CD158b, CD158e) expression was determined on NK cells purified from decidua (dNK) and matched peripheral blood samples (pNK) (samples 1–19). For dNK cells (open symbols), a gate was set on CD56brightCD32 cells and for pNK cells (closed symbols) a gate was set on CD56dimCD32 cells and the percentage of these cells that expressed CD158a, CD158b or CD158e is Figure 3 The level of KIR2DL4 expression on both dNK and pNK shown. A bar represents the mean of each group. cells varies between individuals. NK cells were isolated from decidua and peripheral blood samples and then stained for KIR2DL4 (mAb #2238), CD56 and CD3. A gate was initially set on CD32 cells. Representative dot plots from two pNK (donors X and Y) high (lower panels) KIR2DL4 expression by dNK and pNK cells are and two dNK (donors W and Z) samples showing high and low levels of presented. As shown, even in those individuals with high expression, KIR2DL4 expression are presented. The numbers represent the percentage there is only a modest shift in the mean fluorescent intensity with non- of cells in each quadrant of the dot plots. Note that in all cases, KIR2DL4 expression occurs only on the CD56bright subpopulation. bimodal distribution between positive and negative cells. The small population of CD56 negative cells apparently staining for KIR2DL4 is also present in the isotype control tube (data not shown) and there- 2007). This was also true for pNK in this study (P ¼ 0.03 for geno- fore represent non-specific staining. In Fig. 4, the percentage of types with a 10A-A allele versus 10A-B/9A), and this genotype was CD56bright dNK and matched CD56bright pNK cells from the same also generally associated with low surface expression by dNK cells donors (except for three 10A-B/9A dNK donors for whom (P ¼ 0.02 for genotypes with a 10A-A allele versus 10A-B/9A). matched pNK cells were not available) that was KIR2DL4 positive However, three individuals with this genotype formed a distinct (as shown by mAb #33) is plotted for each KIR2DL4 genotype. Posi- cluster in which 40% of their dNK cells expressed KIR2DL4 tivity was defined as having fluorescence intensity .99% of the same (Fig. 4), despite their pNK cells being negative when stained with cells stained with an isotype control antibody. The 9A/10A trans- the same antibody (mAb #64). Further investigation is required to membrane genotype influenced KIR2DL4 expression by dNK in the explain this heterogeneity. Thus, with the exception of three 9A/ same way, we have previously reported for pNK (Goodridge et al., 10A-B heterozygotes, KIR2DL4 membrane expression by dNK cells 2007). Individuals with at least one 10A-A allele had a significantly was determined by the 9A/10A transmembrane genotype in the higher percentage of KIR2DL4 positive dNK cells (ranging from 8 to same way as for pNK cells. 66%), than dNK cells derived from 9A homozygous individuals (ranging from 0.7 to 9.0%) (P , 0.00001). This was also true for Freshly isolated dNK and pNK cells secrete pNK cells (P ¼ 0.03). Homozygous 10A-A individuals had the highest percentage of KIR2DL4 positive dNK and pNK cells. The little IFNg in response to stimulation single example of a homozygous 10A-B individual had ,5% positive through KIR2DL4 dNK and pNK, consistent with the inability of anti-KIR2DL4 antibodies To test the ability of dNK cells to secrete IFNg when stimulated via to react with the protein lacking the D0-domain made by freshly iso- KIR2DL4, 8 pNK and 11 freshly isolated dNK cell samples were stimulated pNK from such donors. lated for 20 h with SPB anti-KIR2DL4 (mAb #33), anti-CD16 as a We have previously shown that pNK cells from 9A/10A-B hetero- positive control (for the pNK cells) or an isotype control and the zygous individuals have low KIR2DL4 expression (Goodridge et al., IFNg content of the supernatants was measured by ELISA. Mean KIR2DL4 genotype influences INFg secretion 493 Downloaded from https://academic.oup.com/molehr/article/15/8/489/1252362 by guest on 25 September 2021

Figure 4 Freshly isolated dNK (upper) and pNK cells (lower) from pregnant women express KIR2DL4 in accordance with their trans- g membrane genotype. Figure 5 Freshly isolated dNK cells produce relatively little IFN in response to stimulation through KIR2DL4. Freshly isolated NK cells were purified from decidua and, with the exception of three 9A/10A donors, also from matched peripheral blood samples. A gate pNK (samples 27, 30, 31, 33–36, 38) and dNK (samples 20–24, 29, 33, 35– was initially set on CD32 cells. The percentage of the CD56bright cells that 38) cells were isolated and cultured for 20 h in the presence of SPB mAb expressed KIR2DL4 (closed symbols: mAb #2238; open symbols: mAb specific for KIR2DL4 (mAb #33) or CD16 (except dNK samples 20–22), or #64) was determined for each sample, and results grouped according to isotype control. Supernatants were harvested and the IFNg content deter- KIR2DL4 transmembrane genotype. The alleles present in each group are indi- mined by ELISA. (A) Results shown as mean fold-increase when compared cated by a ‘þ’ on the horizontal axes. Individuals with at least one 10A-A allele with isotype control. The shaded bars show the response to anti-KIR2DL4. had a significantly higher percentage of KIR2DL4 positive cells than individuals The white bars show the response to anti-CD16. Error bars represent one who were homozygous for the 9A allele (P , 0.00001 and P ¼ 0.02 for dNK SEM. (B) Responses of individual samples to isotype control (squares) or anti- and pNK, respectively). Individuals with at least one 10A-A allele also had a KIR2DL4 or anti-CD16 (triangles). significantly higher percentage of KIR2DL4 positive cells than individuals with the 10A-B/9A genotype (P ¼ 0.02 and P ¼ 0.03 for dNK and pNK, respectively). even from individuals with high expressing KIR2DL4 genotypes, is perhaps not surprising. Given that only the CD56bright pNK express background levels of IFNg secretion in response to isotype control KIR2DL4 and these cells constitute only 5–10% of the pNK cell popu- antibody were similarly low for both populations (pNK: 1.35 pg/105 lation, the number of responding cells may have been too low to allow cells; dNK: 1.94 pg/105 cells). pNK cells stimulated via CD16 had a detection of IFNg secretion. mean of a 43-fold-increase in INFg secretion, with all eight samples As expected, freshly isolated dNK cells, which do not express producing at least a 5-fold increase over the isotype control (range: CD16, responded poorly to stimulation via CD16 with a mean fold 6.2–165-fold increase) (Fig. 5). Stimulation of pNK via KIR2DL4 pro- increase in IFNg secretion of 7. They also responded weakly to stimu- duced a relatively small response with a mean 9-fold increase, signifi- lation via KIR2DL4 with a mean fold-increase of 4 (Fig. 5). Thus, cantly lower than for CD16 (P , 0.05). This was true even for the although the majority of dNK cells is CD56bright and expresses freshly isolated pNK cells from individuals whose CD56bright NK KIR2DL4 in donors with the appropriate KIR2DL4 genotype, dNK cells expressed high levels of KIR2DL4. Although the one pNK cell failed to make a vigorous response to stimulation via KIR2DL4. dNK sample with a substantial IFNg responses (60-fold increase) had from only two donors made a greater than 5-fold increase in IFNg high KIR2DL4 expression, a further three individuals with high secretion. Although one of these had high KIR2DL4 expression, KIR2DL4 expression failed to make a greater than 5-fold increase in dNK cells from a further four donors with high levels of KIR2DL4 IFNg. The failure to detect large INFg responses from pNK cells, failed to show such an increase. 494 Goodridge et al.

Cultured dNK cells and pNK cells are able 23.0-fold-increase in IFNg compared with a mean 3.5-fold-increase secrete IFNg in response to SPB from pNK cells from individuals without a 10A-A allele (P , 0.03) anti-KIR2DL4 mAb (Fig. 7C), a result consistent with our previous report (Goodridge et al., 2007). Similarly, dNK from six donors with a membrane expres- As freshly isolated dNK at best made minimal IFNg in response to SPB sing KIR2DL4 genotype had a mean 11.4-fold-increase, compared anti-KIR2DL4, we determined whether their activation by culture with with an increase of only 2.3-fold for the four donors that lacked a interleukin (IL)-2 would enable them to respond to this stimulus. In 10A-A allele (P , 0.02) (Fig. 7B). Thus, for cultured pNK and dNK order to obtain sufficient cell numbers for these experiments it was cells, the 9A/10A transmembrane polymorphism affects IFNg pro- necessary to expand the dNK cells by primary and secondary duction in response to stimulation with SPB anti-KIR2DL4. culture as described in the Materials and Methods. The cells expanded by such culture were confirmed to be NK cells, i.e. they expressed Freshly isolated dNK cells do not respond CD56, and not CD3, on their surface. Both primary and secondary cultured pNK (data not shown) and dNK cells (Fig. 6) continued to to stimulation with soluble mAb #33 Downloaded from https://academic.oup.com/molehr/article/15/8/489/1252362 by guest on 25 September 2021 express KIR2DL4 in accordance with their transmembrane genotypes, As Rajagopalan et al. (2005) have shown that freshly isolated pNK cells a finding we have previously shown for primary cultures of pNK cells secrete IFNg in response to stimulation with soluble mAb #33 via an (Witt et al., 2002). KIR2DL4 expression was generally higher on cul- unusual mechanism not requiring membrane expression of KIR2DL4, tured dNK cells from individuals with at least one 10A-A allele, we investigated whether freshly isolated dNK cells can respond to although those from 9A homozygous and 9A/10A-B individuals soluble mAb #33. As shown in Fig. 8, the dNK cell preparations remained essentially KIR2DL4 negative. studied produced no IFNg in response to stimulation with soluble We next tested whether activation by in vitro culture would enable mAb #33 with none of the mAb #33 stimulated preparations produ- dNK cells stimulated with SPB anti-KIR2DL4 to produce IFNg. cing more INFg than the isotype control. In contrast, and in agreement Although the majority of dNK cells from all donors uniformly acquired with Rajagopalan et al. (2005), freshly isolated pNK cells were capable CD16 following culture, the cultured dNK cells retained other charac- of producing IFNg in response to stimulation with soluble mAb #33 teristics of their decidual phenotype remaining predominantly CD9 with a mean fold-increase of 44 (P , 0.01 for pNK versus dNK). positive (mean of 75%), whereas the cultured pNK cells remained pre- However, the pNK cells from only half of the donors responded dominantly CD9 negative (mean of 6%) (data not shown). As (Fig. 8B). The ability to respond did not appear to be related to the expected, cultured pNK cells responded strongly to stimulation via 9A/10A transmembrane genotype (data not shown). CD16 with a mean 46-fold-increase in IFNg secretion whereas cultured dNK also responded strongly with a mean 37-fold-increase (Fig. 7A). When stimulated via KIR2DL4, pNK had a mean Discussion 15-fold-increase and dNK had a mean 8-fold-increase (Fig. 7A). The In this study, using two different mAbs, we have shown that KIR2DL4 IFNg response to KIR2DL4 was influenced by KIR2DL4 genotype. is expressed on dNK cells, confirming the earlier report using a poly- Cultured pNK cells from individuals with a membrane expressing clonal antibody (Ponte et al., 1999). However, as for CD56bright pNK KIR2DL4 allele (10A-A or homozygous 10A-B) produced a mean cells, the intensity of KIR2DL4 expression on dNK cells was low, such that distinct KIR2DL4 positive and negative populations reported for the polyclonal antibody, were not seen. Rather, there appears to be low level expression on the majority of the dNK cells. The low level of surface expression may be related to the report that KIR2DL4 resides predominantly in intracytoplasmic endosomes that cycle between the cytoplasm and cell surface (Rajagopalan et al., 2005). KIR2DL4 expression by freshly isolated CD56bright dNK was controlled by the 9A/10A transmembrane polymorphism in the same way that we have previously reported for the CD56bright pNK cells (Goodridge et al., 2003). Thus, women who were homozygous for the 9A transmembrane allele or heterozygous for the 9A/10A-B alleles lacked detectable KIR2DL4 surface expression on their dNK cells and this was not changed by activation (culture with IL-2). In subjects with a 10A-A allele, the majority of freshly isolated dNK cells expressed KIR2DL4, and 10A-A homozygous individuals tended to Figure 6 KIR2DL4 expression by cultured dNK cells is determined have higher expression than heterozygous individuals. In the single by transmembrane genotype. 10A-B homozygote, detectable KIR2DL4 expression was induced by The percentage of KIR2DL4 positive cells in primary (circle symbol: samples 1, in vitro culture on both dNK cells and pNK cells, as previously 3–9, 11, 13, 14, 16–19) and secondary (square symbol: samples 1, 3, 5, 6, 9, reported for pNK (Goodridge et al., 2007). The fact that membrane 11, 14, 16, 18) dNK cell cultures is shown. For each population, a gate was set bright on CD56þ CD32 cells and the percentage of these cells expressing KIR2DL4 expression of KIR2DL4 is found on freshly isolated CD56 pNK (mAb #2238) determined. The results are grouped according to their cells or cultured CD56dim pNK cells (which become CD56bright after KIR2DL4 transmembrane genotype. culture), and on freshly isolated and cultured dNK cells, all of which have the CD56bright phenotype, suggests that membrane expression KIR2DL4 genotype influences INFg secretion 495 Downloaded from https://academic.oup.com/molehr/article/15/8/489/1252362 by guest on 25 September 2021

Figure 7 Cultured dNK and pNK cells can produce IFNg in response to SPB anti-KIR2DL4 and anti-CD16. Secondary cultures of pNK (samples 1, 3, 5, 6, 8, 9, 13, 14, 16, 17, 19) and dNK (samples 1, 3, 5, 6, 9, 11, 13, 14, 16, 18) cells were harvested, stimulated for 20 h by SPB mAb speciﬁc for KIR2DL4 (#2238), CD16, or isotype control and the IFNg content in the supernatants determined by ELISA. (A) The mean (þSEM) fold-increase in IFNg secretion when compared with isotype control is shown for pNK and dNK cells when stimulated through SPB anti-KIR2DL4 (shaded bars) or SPB anti-CD16 (white bars). (B) Responses of individual samples after stimulation with isotype control (squares) or anti-KIR2DL4 or anti-CD16 (triangles). (C) The mean (þSEM) fold-increase in IFNg secretion when stimulated with anti-KIR2DL4 for pNK and dNK cells from donors with expressing KIR2DL4 genotypes (shaded bars) or non-expressing genotypes (white bars). (D) INFg responses of individual samples to isotype control (squares) and anti-KIR2DL4 (triangles). INFg responses were greater for pNK cells and dNK cells from individuals with a membrane expressing KIR2DL4 genotype than a non-expressing genotype (P , 0.03, P , 0.02, for pNK and dNK, respectively).

of KIR2DL4 is associated with bright expression of CD56. Our current et al. (2005) recently demonstrated that in freshly isolated pNK cells study confirms the lack of detectable KIR2DL4 expression on both soluble HLA-G can be taken up by intracellular KIR2DL4 via an pNK and dNK cells from 9A/10A-B heterozygotes even after unusual mechanism, resulting in IFNg secretion. We were able to culture, a phenomenon that currently lacks explanation, given that confirm that soluble mAb #33 elicits IFNg production by freshly iso- there is up-regulation of the same mRNA species as is seen in lated pNK cells from some but not all individuals but responsiveness 10A-B homozygotes following culture (Goodridge et al., 2007). was not related to KIR2DL4 genotype. However, freshly isolated Despite membrane expression of KIR2DL4, stimulation of freshly dNK cells from the same women were in all cases completely unre- isolated dNK cells with SPB mAb #33 resulted in very little IFNg sponsive to this stimulus. The responsiveness of pNK to soluble secretion. However, responsiveness to KIR2DL4 stimulation was mAb #33 does not appear to be related to the 9A/10A genotype recovered by in vitro culture. Both cultured dNK cells and cultured and further experiments are in progress to clarify this point and to pNK cells secreted IFNg in response to SPB anti-KIR2DL4 providing determine the basis for the differences between pNK and dNK in at least one 10A-A allele was present. It is not clear whether SPB their response to soluble ligands. or soluble ligand is more relevant to dNK cell activation in vivo. Our data suggest that freshly isolated dNK cells, in contrast to Several studies have suggested that the non-classical class I molecule, freshly isolated pNK cells, cannot produce IFNg in response to stimu- HLA-G, is a ligand for KIR2DL4 (Ponte et al., 1999; Rajagopalan and lation through KIR2DL4, regardless of whether the stimulus is SPB or Long, 1999), although this has not been confirmed by all groups soluble in nature. Basal production of IFNg by freshly isolated dNK (Allan et al., 1999; Ponte et al., 1999; Boyson et al., 2002). Rajagopolan cells has previously been reported to be very low (Vigano et al., 496 Goodridge et al.

freshly isolated pNK cells have been found to secrete a number of factors, including tumour necrosing factor-alpha, macrophage inﬂam- matory protein-1-alpha, IL-1beta and IL-8 (Rajagopalan et al., 2005). Whether freshly isolated dNK cells can produce any of these factors as a result of stimulation through KIR2DL4, and the effect of the 9A/10A transmembrane polymorphism on this function, remain to be established. If HLA-G is indeed the ligand for KIR2DL4, our data suggest that the 9A/10A polymorphism would control the response of dNK cells to membrane-bound HLA-G (equivalent to our solid- phase stimulation), and have the potential to inﬂuence placentation.

Acknowledgements Downloaded from https://academic.oup.com/molehr/article/15/8/489/1252362 by guest on 25 September 2021 We would like to acknowledge the cooperation of the manager and staff at Marie Stopes International, Perth. We would also like to thank Drs E. Long and S. Rajagopalan for generously providing antibodies for our studies.

Funding This work was funded by the NHMRC (Grant Number 303236) and L. Lathbury was supported by a Child Health Research Foundation of WA Research Fellowship.

References Figure 8 Freshly isolated dNK cells do not secrete IFNg in response to stimulation with soluble mAb #33. Allan DSJ, Colonna M, Lanier LL, Churakova TD, Abrams JS, Ellis SA, McMichael AJ, Braud VM. Tetrameric complexes of human Freshly isolated dNK (samples 47–56) and matched pNK cells (samples 39– histocompatibility leukocyte antigen (HLA)-G bind to peripheral blood 42, 47–56) were cultured for 20 h in the presence of soluble mAb specific for KIR2DL4 (mAb #33) or isotype control. Supernatants were harvested myelomonocytic cells. J Exp Med 1999;189:1149–1155. and the IFNg concentration determined by ELISA. (A) Mean (þSEM) of the Bashirova AA, Martin MP, McVicar DW, Carrington M. The killer fold-increase in INFg production for pNK and dNK cells. (B) INFg response immunoglobulin-like receptor gene cluster: tuning the genome for to isotype control (squares) and mAb #33 (triangles) for individual samples. defense. Annu Rev Genomics Hum Genet 2006;7:277–299. Boyson JE, Erskine R, Whitman MC, Chiu M, Lau JM, Koopman LA, Valter MM, Angelisova P, Horejsi V, Strominger JL. Disulfide bond-mediated dimerization of HLA-G on the cell surface. Proc Natl 2001), although in that report, IFNg production by dNK cells was Acad Sci USA 2002;99:16180–16185. increased in the presence of IL-12, either alone or in combination Croy BA, Chantakru S, Esadeg S, Ashkar AA, Wei Q. Decidual natural killer cells: key regulators of placental development (a review). with either IL-2 or IL-10. We found that activation by in vitro culture J Reprod Immunol 2002;57:151–168. with IL-2 enabled dNK cells to secrete INFg in response to SPB anti- Engert S, Rieger L, Kapp M, Becker JC, Dietl J, Kammerer U. Profiling KIR2DL4 mAb, providing the donor had at least one 10A-A allele. chemokines, cytokines and growth factors in human early pregnancy Although there is little evidence of significant IL-2 production in the decidua by protein array. Am J Reprod Immunol 2007;58:129–137. decidua, if dNK cells were activated in vivo by physiological stimuli, Eriksson M, Meadows SK, Wira CR, Sentman CL. Unique phenotype of their KIR2DL4 genotype might then influence their secretion of IFNg. human uterine NK cells and their regulation by endogenous Chemokines, cytokines and growth factors other than IFNg are TGF-beta. J Leukoc Biol 2004;76:667–675. thought to play an important role the development of the placenta Goodridge JP, Witt CS, Christiansen FT, Warren HS. KIR2DL4 (CD158d) and the maintenance of a successful pregnancy (Parmentier et al., genotype influences expression and function in NK cells. J Immunol 2003; 1991). Indeed, various studies have shown that dNK cells can 171:1768–1774. produce, amongst others, the angiogenic factors vascular endothelial Goodridge JP, Lathbury LJ, Steiner NK, Shulse CN, Pullikotil P, Seidah NG, Hurley CK, Christiansen FT, Witt CS. Three common alleles of growth factor, angiopoietin 2, and angiogenin; the macrophage KIR2DL4 (CD158d) encode constitutively expressed, inducible and colony-stimulating factor; the chemokines IL-8 monocyte protein-1 secreted receptors in NK cells. Eur J Immunol 2007;37:199–211. and Regulated upon Activation, Normal T-cell Expressed and Hiby SE, King A, Sharkey AM, Loke YW. Human uterine NK cells have a Secreted; the regulatory factor, tumour growth factor-beta; and the similar repertoire of killer inhibitory and activatory receptors to those cytokines IL-4 and IL-1beta (Lash et al., 2006; Engert et al., 2007). found in blood, as demonstrated by RT–PCR and sequencing. Mol In response to stimulation with soluble mAb #33 or sHLA-G, Immunol 1997;34:419–430. KIR2DL4 genotype influences INFg secretion 497

Hiby SE, Walker JJ, O’shaughnessy KM, Redman CW, Carrington M, Rajagopalan S, Long EO. A human histocompatibility leukocyte antigen Trowsdale J, Moffett A. Combinations of maternal KIR and fetal (HLA)-G-specific receptor expressed on all natural killer cells. J Exp HLA-C genes influence the risk of preeclampsia and reproductive Med 1999;189:1093–1100. success. J Exp Med 2004;200:957–965. Rajagopalan S, Fu J, Long EO. Cutting edge: induction of IFN-gamma Hiby SE, Regan L, Lo W, Farrell L, Carrington M, Moffett A. Association of production but not cytotoxicity by the killer cell Ig-like receptor maternal killer-cell immunoglobulin-like receptors and parental HLA-C KIR2DL4 (CD158d) in resting NK cells. J Immunol 2001; genotypes with recurrent miscarriage. Hum Reprod 2008;23:972–976. 167:1877–1881. Kikuchi-Maki A, Yusa S, Catina TL, Campbell KS. KIR2DL4 is an IL-2-regulated Rajagopalan S, Bryceson YT, Kuppusamy SP, Geraghty DE, Meer AV, NK cell receptor that exhibits limited expression in humans but triggers Joosten I, Long EO. Activation of NK Cells by an Endocytosed strong IFN-gamma production. JImmunol2003;171:3415–3425. Receptor for Soluble HLA-G. PLoS Biol 2005;4:e9. Koopman LA, Kopcow HD, Rybalov B, Boyson JE, Orange JS, Schatz F, Trundley A, Moffett A. Human uterine leukocytes and pregnancy. Tissue Masch R, Lockwood CJ, Schachter AD, Park PJ et al. Human decidual Antigens 2004;63:1–12. natural killer cells are a unique NK cell subset with immunomodulatory Verma S, King A, Loke YW. Expression of killer cell inhibitory Downloaded from https://academic.oup.com/molehr/article/15/8/489/1252362 by guest on 25 September 2021 potential. J Exp Med 2003;198:1201–1212. receptors on human uterine natural killer cells. Eur J Immunol 1997; Lash GE, Schiessl B, Kirkley M, Innes BA, Cooper A, Searle RF, Robson SC, 27:979–983. Bulmer JN. Expression of angiogenic growth factors by uterine natural Vigano P, Gaffuri B, Somigliana E, Infantino M, Vignali M, Di Blasio AM. killer cells during early pregnancy. J Leukoc Biol 2006;80:572–580. Interleukin-10 is produced by human uterine natural killer cells but Moffett-King A. Natural killer cells and pregnancy. Nat Rev Immunol 2002; does not affect their production of interferon-gamma. Mol Hum 2:656–663. Reprod 2001;7:971–977. Parmentier HK, van Wichen D, Peters PJ, Tschopp J, De Weger RA, Witt CS, Martin A, Christiansen FT. Detection of KIR2DL4 alleles by Schuurman HJ. No histological evidence for cytotoxic T cells in sequencing and SSCP reveals a common allele with a shortened destruction of lymph-node follicle centres after HIV infection? AIDS cytoplasmic tail. Tissue Antigens 2000;56:248–257. 1991;5:778–780. Witt CS, Whiteway JM, Warren HS, Barden A, Rogers M, Martin A, Ponte M, Cantoni C, Biassoni R, Tradori-Cappai A, Bentivoglio G, Vitale C, Beilin L, Christiansen FT. Alleles of the KIR2DL4 receptor and Bertone S, Moretta A, Moretta L, Mingari MC. Inhibitory receptors their lack of association with pre-eclampsia. Eur J Immunol 2002; sensing HLA-G1 molecules in pregnancy: decidua-associated natural killer 32:18–29. cells express LIR-1 and CD94/NKG2A and acquire p49, an HLA-G1-specific receptor [see comments]. Proc Natl Acad Sci USA 1999; Submitted on January 14, 2009; resubmitted on April 22, 2009; accepted on 96:5674–5679. May 29, 2009