A predominance of T helper (Th)2-type cytokines and a weakening of Th1 responses seem to be critical for the maintenance of a successful gestation. Among Th2-type cytokines, interleukin (IL)-10 is produced by human cytotrophoblasts and defects in this production result in specific pathological conditions of pregnancy. The current opinion is that IL-10 serves to protect the fetus from a harmful maternal immune response. However, production of the cytokine and its direct effect on uterine natural killer (uNK) cells, which represent the predominant lymphocyte population infiltrating the pregnant endometrium, are largely unknown. Thus, to shed light on the cytokine network at the maternal–fetal barrier during early pregnancy, we investigated the IL-10 system in uNK cells. We showed that uNK cells express the mRNA transcripts for IL-10 and IL-10 receptor. Production of IL-10 by the uNK cells was enhanced by both IL-2 and IL-12. Treatment with IL-10 alone enhanced uNK cell cytotoxic activity. In contrast, the cytokine did not modify the basal or stimulated production of interferon (IFN)-γ by uNK. Thus, IL-10 does not act as a direct antagonist of uNK cell function and activation. However, IL-10 produced by uNK cells in response to IL-12 and IL-2 may still have a feedback inhibitory effect on the production of deleterious cytokines within the uterine microenvironment.
The predominant uterine lymphocytes in human early gestation are CD56bright granulated natural killer (NK) cells that do not express CD16 or membrane CD3 and these account for ~70% of bone marrow-derived cells (Loke et al., 1995; King et al., 1998, 1999). The morphological and phenotypic properties of decidual CD56+ cells suggest that they are specific for the uterus since only a very small proportion (<2%) of peripheral blood lymphocytes (PBL) appears to have similar characteristics (Kammerer et al., 1998). Defining the functions of uterine natural killer (uNK) cells has been a protracted endeavour (King, 2000). In mice deficient for uNK cells, the decidua lacks normal cell density while decidual arterioles fail to undergo modifications in their smooth-muscle coats and display endothelial cell damage (Croy et al., 2000). An increase in perinatal death and a reduced weight at birth and during adulthood has been observed in the offspring from gestation of these strains of mice. On the other hand, activation of NK cells in pregnant mice, following injection with double-strand RNA poly-IC, induces spontaneous fetal resorption (Raghupathy, 1997). This effect seems to be associated with the NK cell-enhanced production of interferon (IFN)-γ and tumour necrosis factor (TNF)-α which are required to trigger abortion by vascular injury (Clark et al., 1998).
In humans, uNK cells which are in direct contact with fetal cytotrophoblasts exhibit moderate cytotoxicity toward major histocompatibility (MHC) class I-negative cells but do not appear to be constitutively able to lyse fetal cytotrophoblasts in vitro. Indeed, the expression of human leukocyte antigen (HLA)-G and HLA-E has been demonstrated to protect cytotrophoblast cells from the lysis mediated by uNK cells (Somigliana et al., 1999; King et al., 2000). However, uNK cells may cause fetal loss by developing, in the presence of TNF and interleukin (IL)-2, into lymphocyte activated killer (LAK) cells which have been shown to be capable of killing trophoblast cells (Raghupathy, 1997); and, consistent with a role for uNK cells in spontaneous abortions, several investigators have documented a higher risk of miscarriage in women with an increased infiltration and activation of endometrial NK cells (Kodama et al., 1998; Clifford et al., 1999; Quenby et al., 1999).
Whether or not activated uNK cells are involved in the killing of the implanted embryo by direct lysis or by cytokine-triggered thrombotic processes, the production of specific cytokines and their effect on uNK cells are indeed of critical relevance (Raghupathy, 1997; Somigliana et al., 1999). In this context, it is important to note that a successful pregnancy is thought to be accompanied by a bias away from T helper (Th)1-type cytokines and toward Th2-type cytokines (Hill et al., 1995; Marzi et al., 1996; Raghupathy, 1997). While IFN-γ, TNF-α and IL-2 are known to be potentially deleterious for the conceptus, IL-4, IL-6 and IL-10, together with transforming growth factor (TGF)-β, are supposed to favour implantation and to control harmful maternal responses (Krishnan et al., 1996). Specifically, it has been claimed that IL-10 is involved in the maintenance of human pregnancy through a dichotomous effect on HLA expression, inducing HLA-G expression, while down-regulating classical class I and class II antigens (Moreau et al., 1999). Moreover, IL-10 has been postulated to be involved in the direct inhibition of Th1-type cytokine synthesis and in the suppression of a significant proportion of NK-like cells and other inflammatory cells at the uteroplacental interface (Lin et al., 1993; Cadet et al., 1995; Chaouat et al., 1996; Fried et al., 1998; Hashii et al., 1998; Rivera et al., 1998; Hennessy et al., 1999).
Human IL-10 is an 18 kDa polypeptide that lacks detectable glycosylation and is expressed as a non-covalent homodimer (Mosmann, 1994). IL-10 was originally identified as a secretory product of Th2 cells that inhibits the production of IL-2 and IFN-γ by Th1 cells via its ability to impede many aspects of macrophage activation. While production and role of IL-10 in human peripheral blood NK cells have been extensively investigated (Carson et al., 1995; Mehrotra et al., 1998; Cai et al., 1999), the IL-10 system in NK cells resident in the uterus has yet to be fully characterized. Therefore, in this study we analysed mRNA expression and production of IL-10 by purified populations of uNK cells both in the basal condition and following IL-2 and IL-12 stimulation. Moreover, we evaluated the expression of IL-10 receptor (IL-10R) transcripts in uNK cells and investigated specific functional consequences of the direct binding of IL-10 to these cells, namely cytotoxic activity and production of the Th1-type cytokine, IFN-γ.
Materials and methods
Cytokines used in this study were purified recombinant proteins of human origin and reconstituted in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with 0.1% human albumin. IL-10 was obtained from R&D Systems, Inc. (Minneapolis, MN, USA), IL-12 was from Roche, Milano Ricerche (Milan, Italy), and IL-2 was from Amersham International (Amersham, Bucks, UK). Anti-CD45–fluorescein isothiocyanate (FITC)/CD14–phycoerythrin (PE) (2D1, MØP9 clone, mouse IgG1/IgG2b) was from Becton Dickinson (Mountain View, CA, USA). Anti-CD3–FITC/CD56–PE (UCHT1, B159 clone, mouse IgG1/IgG1) and unconjugated anti-CD3 mAb (X35 clone, mouse IgG2a) were from Immunotech (Marseille, France). Culture medium consisted of RPMI 1640 medium (Euroclone, UK) supplemented with 2 mmol/l l-glutamine (Euroclone), antibiotics (Euroclone), 2.5 μg/ml fungizone (Euroclone) and 10% heat-inactivated fetal calf serum (FCS) (Euroclone). Collagenase A was purchased from Roche (Mi, Italy) and hyaluronidase was from Sigma. Oligonucleotide primers for IL-10, IL-10R and hypoxanthine phosphoribosyltransferase 1 (HPRT) were from Amersham–Pharmacia (Milan, Italy).
Sample collection and preparation of uterine NK cells
Decidual tissue was obtained from healthy women undergoing elective abortion of a normal pregnancy at 6–12 weeks gestation. Patients were informed that tissue samples would be used for research purposes and gave written consent. Approval for this study was granted by the local Human Institutional Investigation Committee. Specimens were extensively washed with phosphate-buffered saline, minced thoroughly between two scalpels into fragments of ~1 mm3 and digested for 1 h at 37°C with gentle agitation in Ham's F-10 with 0.1% collagenase and 0.2% hyaluronidase. Cell suspensions were plated and after an overnight incubation at 37°C, non-adherent lymphocytes were harvested from the culture flask, leaving behind adherent stromal cells and macrophages. The cell suspension obtained was then separated from dead cells and red cells by Ficoll-Hypaque gradient 1077. CD3+T cells were depleted from lymphocyte preparations by negative selection using an anti-CD3 mAb. Cells were then analysed in a flow cytometer (FACstar Plus; Becton Dickinson). Flow cytometric analysis of uterine lymphocytes prepared in this way revealed that > 98% were CD45+ of which >99% were CD14– and >97% were CD56+ (Figure 1A and B).
Extraction of mRNA, RT–PCR and sequence analysis of human IL-10 and IL-10R
Reverse transcription–polymerase chain reaction (RT–PCR) specific for IL-10 and IL-10R transcripts was performed as previously described (Chen et al., 1994; Michel et al., 1997). Total RNA was isolated from the uNK cells by acid guanidnium isothiocyanate/ phenol chloroform extraction as previously described (Viganò et al., 1997). RNA (1 μg) was transcribed into cDNA using 1 mmol/l each dNTP, 1 unit RNasin, 100 pmol of random hexamer primers, 200 units of reverse transcriptase (all from Perkin–Elmer, Milano, Italy) in a total volume of 20 μl. The reaction mixture was run at 42°C for 1 h followed by a 5 min incubation at 95°C and then quick-chilled on ice. Oligonucleotide primers used for PCR reactions had the following sequences: human IL-10, forward 5′-CTGTGAAAACAAGAGCAAGGC-3′, reverse 5′-GAAGCTTCTGTTGGCTCCC-3′; human IL-10R, forward 5′-CCATCTTGCTGACAACTTCC-3′, reverse 5′-GTGTCTGATACTGTCTTGGC-3′. IL-10 primers amplified a 500 bp fragment; IL-10R primers were designed to amplify a fragment of 440 bp. The PCR reaction was performed on the entire cDNA product according to the instructions provided with the GeneAmp Amplification Reagent Kit (Perkin–Elmer). For IL-10 gene amplification, samples were initially denatured at 94°C for 3 min, then heated at 94°C (20 s), cooled at 61°C (20 s) and heated at 72°C (20 s) for 35 cycles with a final extension at 72°C for 5 min. For IL-10R gene amplification, samples were initially denatured at 94°C for 3 min, then heated at 94°C (1 min), cooled at 60°C (1 min) and heated at 72°C (1 min) for 30 cycles with a final extension at 72°C for 5 min. In each experiment, the negative control was prepared using all reagents and substituting 1 μl of water for the RNA. Integrity of the RNA and absence of genomic contamination was assessed by amplification of the HPRT gene with intron-spanning primers (upstream: nucleotides 570–589; downstream: nucleotides 647–667; Genbank Accession # NM000194) according to the following PCR protocol: 94°C (20 s), 55°C (20 s), 72°C (20 s) for 34 cycles. The small size of the intron (170 bp) ensures that both cDNA and genomic DNA are readily amplified, giving rise to 97 and 267 bp fragments respectively. A further control was obtained omitting the reverse transcriptase to detect the presence of any contaminating genomic DNA. The positive control consisted of amplification of IL-10 and IL-10R transcripts in phytohemaglutinin (PHA)-blasts. PCR products were visualized on a 4% agarose gel stained with ethidium bromide.
The identity of the PCR products with the primer-defined sequences was confirmed by sequence analysis. Sequencing reactions were prepared by means of ABI Big dye terminator chemistry (PE Applied Biosystems, Foster City, CA, USA) and analysed by using the ABI 310 Genetic Analyzer.
51Chromium release cytotoxicity assay
Uterine NK cell-enriched preparations were plated in 96-well V-bottom plates in 100 μl of RPMI 1640 media supplemented with 10% FCS with or without the indicated combination of cytokines. After 18 h of incubation at 37°C, plates were washed and 1×10451Cr-labelled K562 target cells were added to each well. Plates were centrifuged for 3 min at 200 g and incubated for an additional 4 h at 37°C. Plates were then centrifuged, the supernatant from each well was harvested and counted in a gamma counter to determine the isotope release. Minimum and maximum 51Cr release were determined in 10% FCS and 1% NP-40 detergent respectively. Specific lysis was determined as previously described (Mazzeo et al., 1998).
Human uNK cells cultured at 106 cells/ml were stimulated with cytokines as indicated in the figure legends. The concentrations of IL-10 protein in media conditioned by uNK cell cultures were measured using an enzyme-linked immunosorbent assay (ELISA) kit obtained from R&D Systems. A human INF-γ ELISA kit was purchased from Bender MedSystems (Vien, Austria). Assays were performed according to the manufacturer's instructions; the lower limits of IL-10 and of IFN-γ detection were 3.9 and 1.5 pg/ml respectively. All samples were assayed in duplicate. Results are expressed in pg/ml.
Data are expressed as mean ± SEM. Differences between groups were compared by one-way analysis of variance (ANOVA). Fisher's least significant difference test was used as post-test to determine significant differences between groups. P < 0.05 was considered statistically significant.
IL-10 and IL-10R mRNA are expressed by uNK cells
It has been demonstrated that peripheral NK cells constitutively express the IL-10 and the IL-10R mRNA (Carson et al., 1995; Mehrotra et al., 1998). We evaluated uNK cells for the presence of IL-10 and IL-10R transcripts by RT–PCR and sequence analysis.
During cell isolation, a great effort was directed toward complete elimination of decidual cells, macrophages and T cells, as they are potential cytokine-producing contaminants. Phenotypic analysis was performed on each culture examined and only cell populations >99% free of CD14+ cells and with >97% CD56+ cells were considered (Figure 1A, B). Typically, the purified uNK cells comprised two distinct populations with respect to CD56 expression. Most were CD56bright and the remainder CD56dim (Figure 1B).
The purified populations of uNK cells were shown to express IL-10 and IL-10R transcripts as shown in Figures 2 and 3 respectively. The estimated and actual size of the PCR products were 500 and 440 bp for IL-10 and IL-10R respectively. The identity of the amplified products with the primer-defined human IL-10 and IL-10R DNA sequences was further confirmed by sequence analysis. DNA contamination of all RNA samples was controlled by RT–PCR of the constitutively expressed HPRT gene. All HPRT amplifications revealed the correct 97 bp DNA fragment but no 267 bp products that would have indicated genomic DNA contamination.
IL-2 and IL-12 increase IL-10 production by uNK cells
It has been demonstrated that peripheral blood NK cells do not produce measurable levels of IL-10 but can be primed for production of this cytokine (Peritt et al., 1996; Mehrotra et al., 1998; Peritt et al., 1998). Uterine NK cells are known to produce a number of cytokines, including colony stimulating factor (CSF)-1, TNF-α and IFN-γ but IL-10 production has not been extensively investigated (Loke et al., 1995; Deniz et al., 1996). We treated uNK cells with or without IL-2 and/or IL-12 for 18 h, then collected the supernatants and assayed for IL-10 by ELISA. Unstimulated uNK cells produced low concentrations of IL-10 that were significantly enhanced by IL-12 and even more by so IL-2. A slight synergistic effect on IL-10 production was observed when uNK cells were treated with IL-2 and IL-12 together (Figure 4).
IL-10 enhances uNK cell cytotoxicity
It has been shown that IL-10 is able to induce significant cytotoxic activity against tumour cell targets in both peripheral blood CD56bright and CD56dim NK cells in a dose-dependent fashion (Carson et al., 1995). Using highly purified populations of CD56+ uNK cells, we performed studies to determine the direct effect of human IL-10 on cytotoxic activity. Uterine NK cells showed a significant increase in cytotoxic activity against K562 targets when incubated overnight in medium containing 50 ng/ml of IL-10 compared with that observed when incubated in medium alone. In presence of 50 ng/ml of IL-10, uNK cell cytotoxicity against K562 cells was 25% compared to 12.5% observed in the absence of IL-10 (P < 0.05) (Figure 5A). A greater enhancement of uNK cell cytotoxicity could be achieved by the incubation of the uNK cells with either IL-2 or IL-12. IL-2 (10 ng/ml) or IL-12 (10 ng/ml) resulted in an ~1.5-fold increase in cytotoxicity against K562 cells. The combination of IL-12 and IL-10 had no additive effect on uNK cell cytotoxic activity, while IL-10 had a slight enhancement effect on IL-2-induced uNK cell cytotoxicity (Figure 5B).
IL-10 does not affect IFN-γ production by uNK cells
Peripheral blood NK cells have been shown to produce IFN-γ after stimulation with monocyte-derived factors such as IL-12 or TNF-α either alone or in combination with IL-2 (Carson et al., 1995). Moreover, while IL-10 alone does not have any effect on peripheral blood NK cell IFN-γ production (Carson et al., 1995; Cai et al., 1999), its action on the IL-2-mediated effect is still debated (Hsu et al., 1992; Carson et al., 1995).
To assess the direct effect of IL-10 on uNK cell IFN-γ production, uNK cells were cultured for 24 h with IL-2 and/or IL-12 either alone or in combination with IL-10 and uNK cell culture supernatants were assayed for the cytokine production. Uterine NK cells produced low levels of IFN-γ but they could be stimulated to produce the cytokine either by IL-2 or more so, by IL-12 (Figure 6). Moreover, as already reported (Marzusch et al., 1997), when uNK cells were activated with IL-2 plus IL-12, IFN-γ production was synergistically enhanced. IL-10 alone or in combination with IL-12 or IL-2 did not have any effect on uNK cell cytokine production (Figure 6).
The augmentation of IFN-γ production, when present, appeared to result from an increase in the endogenous production of the cytokine, because uNK cell numbers under the different conditions were similar.
Development of the feto-placental unit in human pregnancy requires maternal immune tolerance to the semi-allogeneic fetus (Raghupathy, 1997). It is now well established that local immunity is modified by the expression of uncommon histocompatibility antigens and the synthesis of specific cytokines at the maternal–fetal interface. Indeed, the surface levels of HLA-G/HLA-C molecules or other still incompletely defined proteins, as well as the activation state of local immune cells, could influence trophoblast sensitivity to lysis and thus affect the equilibrium maintained at the implantation site (Somigliana et al., 1999). Specifically, this site is characterized by the presence of a significant population of NK-like cells whose response may be altered by the cytokine environment and other immune modifiers. These cells have been demonstrated to be cytokine producers themselves (Loke et al., 1995). They express mRNA for CSF-1, TNF-α, TGF-β and IFN-γ. However, the expression profile seems to be different from that seen in peripheral blood NK cells, thus confirming that uNK cells are a specialized population of lymphoid cells with unique functional characteristics.
In this study, we investigated the IL-10 system in uNK cells since IL-10 is one of the most important cytokines in the success of pregnancy (Rivera et al., 1998). Indeed, a physiological pregnancy is thought to be a condition characterized by a shift of cytokine production from Th1-type to Th2-type cytokines with a local predominance of IL-10 and IL-4 over TNF-α and IL-2. In agreement with this concept, reduced IL-10 production and elevation of Th1-type cytokines have been reported in pathological pregnancies, distinguishing them from normal ones (Hara et al., 1995; Hill et al., 1995; Hennessy et al., 1999; Lim et al, 2000). Expression levels of IL-2, IL-12 and IFN-γ are higher in endometrial samples derived from women with recurrent miscarriages (Lim et al., 2000) and IL-2 is present in decidual cells of pre-eclamptic patients but not in those of normal pregnant subjects (Hara et al., 1995). Furthermore, peripheral blood mononucleocytes (PBMC) from women with unexplained recurrent spontaneous abortions only rarely respond to trophoblast antigens by producing IL-10, whereas the cytokine can be detected in every trophoblast-activated PBMC supernatant from reproductively normal women (Hill et al., 1995). Finally, Hennessy et al. have demonstrated a significant alteration in IL-10 staining of the trophoblast in pre-eclamptic pregnancies compared with normal pregnancy and, in a similar group of patients, a decrease in circulating IL-10 concentrations in maternal serum was shown (Hennessy et al., 1999).
It has been suggested that, in human pregnancy, the main source of IL-10 in the uterus is represented by cytotrophoblast cells that produce consistent quantities of the cytokine both spontaneously and after activation by lipopolysaccharide (LPS), although a certain production has been also associated with the relatively scant decidual T cells (Roth et al., 1996; Piccinni et al., 1998). However, the precise identity of all cell types producing this cytokine at the maternal–fetal interface is still not clear. Most importantly, although the current opinion is that the localized production of IL-10 would serve to protect the fetus from a harmful maternal immune response (Chaouat et al., 1996), the direct effect of the cytokine on the predominant lymphocyte population infiltrating the pregnant endometrium is completely unknown. We herein confirm previous observations indicating that purified populations of human uNK cells can secrete low amounts of IL-10 (Deniz et al., 1996) and we extend these findings by demonstrating that these cells express the gene encoding IL-10 and that they may be stimulated to secrete IL-10 protein by both IL-2 and IL-12. In general, these observations are in line with those reported for peripheral NK cells that express IL-10 mRNA and can be primed for IL-10 production (Mehrotra et al., 1998). Specifically, Mehrotra et al. have recently demonstrated that IL-2 stimulates freshly purified human peripheral blood NK cells to produce low concentrations of IL-10 while they failed to show a similar effect for IL-12 (Mehrotra et al., 1998). However, in a study supporting the observation that NK cells in the presence of IL-12 or IL-4 differentiate into cell populations with distinct patterns of cytokine secretion similar to Th1 and Th2 cells (Peritt et al., 1998), it was shown that IL-12 is also able to enhance IL-10 production by NK cells. Therefore, it is becoming evident that IL-10 can be secreted from a variety of cell types including uNK cells, although it should be noted that the amount of IL-10 secreted by this lymphoid population represents a limited concentration if compared to that produced by trophoblasts (Roth et al., 1996). However, since uNK cells can be primed for the production of this cytokine, a role of these cells in contributing to the increase in IL-10 at the maternal–fetal interface cannot be discounted.
With regard to the specific functional action of IL-10, the cytokine plays a positive role in preventing spontaneous murine pregnancy failure. Supplemental IL-10 prevents fetal wastage in the abortion-prone strain CBA/J×DBA/2J (Chaouat et al., 1995). Moreover, in normal rats, IL-10 is able to reduce the LPS-induced fetal demise from 43 to 22% (Rivera et al., 1998). The mechanism through which IL-10 exerts such an effect under these circumstances is thought to be multifactorial, related to its growth-promoting action but mostly to its ability to counteract specific cytokines which are deleterious to the success of pregnancy. In this context, the results demonstrated that the functional consequences of IL-10 binding to uNK cells is not a direct inhibition of their specific functions. Indeed, we showed that human IL-10 alone induces a certain, although not potent, stimulation of uNK cell cytotoxic activity and that there is a small enhancing effect when IL-10 is used in combination with IL-2. It should be noted that, since IL-12, IL-2 and IL-10 utilize the same signalling pathway, the JAK/STAT pathway, it is unlikely that IL-10 would greatly increase the effects of IL-12 or IL-2 (Finbloom and Winestock, 1995). Moreover, we showed that IL-10 alone or in combination with IL-2 or IL-12 does not show any effect on uNK cell production of INF-γ which has been indicated as potentially deleterious for the conceptus (Fried et al., 1998). Therefore, any potential beneficial effect of the locally produced IL-10 in early pregnancy maintenance should not be viewed as a direct immunosuppression on uNK cells. However, these results do not rule out that an effect on type-1 cytokine predominance may occur indirectly via suppression of monocyte/macrophage cytokine production. Macrophages are able to produce large amounts of IL-12, which appears to be a potent inducer of uNK cells since it stimulates both production of the Th1-type cytokine IFN-γ and cytotoxic activity. Since IL-10 is a strong inhibitor of IL-12 secretion by monocytes/macrophages (D'Andrea et al., 1993), its production by uNK cells may represent a negative feedback loop for IL-12 overexpression. Thus, this feedback loop, which has already been proposed for peripheral blood NK cells (Mehrotra et al., 1998), may contribute to control excessive uNK cell activation and may have an inhibitory effect on abortifacient cytokines produced within the microenvironment of these cells. Studies are currently underway to verify this hypothesis. It should also be noted that recent evidence emphasizes the putative time-dependency of the functions of specific cytokines, such as IFN-γ in the decidual compartment (Ashkar et al., 2000); thus, the possibility of different actions of IL-10 in relation to early or later stages of embryo implantation and development should not be disregarded.
In conclusion, the results of this study support the following conclusions: (i) human uNK cells express the gene for IL-10 and produce IL-10 upon IL-2 and IL-12 stimulation; (i) human uNK cells also express the gene for IL-10 receptor; (iii) IL-10 does not have any constitutive or additive direct effect on uNK IFN-γ production and is able to augment both basal and IL-2-induced uNK cell cytotoxic activity. Thus, the potential use of supplemental IL-10 in specific complications of pregnancy warrants further investigation.