INTRODUCTION

Natural killer (NK) cells are large granular lymphocytes of the innate immune system that are especially abundant in the placenta and are also known as “uterine granular lymphocytes”1. According to the “missing self” hypo-thesis 2, NK cells will recognize and eliminate those cells who fail to express self major histocompatibility complex (MHC) class I molecules. But why the maternal NK cells can tolerate semiallogenenic fetal tissues on which do not express maternal MHC class I molecules? Human leukocyte antigen-G (HLA-G), a nonclassical MHC I molecule, which is predominantly expressed by fetal extravillous trophoblasts, may be the answer3. Unlike classical MHC I antigens, HLA-G exhibits a tissue-restricted distribution pattern and its mRNA is alternatively spliced to encode seven different isoforms, namely the four membrane-bound proteins, HLA-G1, -G2, -G3, -G4 and three soluble isoforms, HLA-G5 (also termed sHLA-G1), -G6, -G74, 5. HLA-G plays an important role not only during pregnancy, but also during allograft and xenograft6. In addition, it provides new strategies to cure tumors and inflammation because HLA-G is also expressed in some tumors and inflamed cells7, 8.

Soluble HLA-G1α chain (sHLA-G1α chain), encoded by a mRNA retaining intron 4, is a protein that lacks the transmembrane and cytoplasmic domains9. In physiological condition, a large quantity of sHLA-G1 is secreted in the amniotic fluid and released into the maternal peripheral blood where they may exhibit systemic immunoinhibitory functions10. It was shown that the soluble HLA-G could trigger CD95/CD95 ligand-mediated apoptosis in activated CD8+ cells11 and suppress the allo-proliferative response12. These might be the important mechanisms implicated in the immunotolerance of the fetal allograft.

To date, three NK inhibitory receptors that directly interact with HLA-G have been identified on NK cells: (1) KIR2DL4 (p49)13, 14, which is specific for HLA-G and is composed of two extracellular Ig-like domains and a single ITIM in its cytoplasmic tail; (2) ILT-2 (or LIR-1)15, which belongs to a new family of Ig-SF-receptors and has four extracellular Ig domains and four ITIMs in its intracytoplasmic tail; (3) ILT-416, which is selectively expressed in monocytes, macrophages and dendritic cells (DCs), can bind to both classical class I molecules and nonclassical class I molecules HLA-G. Little is known about how soluble HLA-G interacts with its receptors on NK cells, and how the balance between activating and inhibitory signals in NK cells is kept. To answer these questions, the sHLA-G1α chain was expressed as a fusion protein with glutathione S-transferase (GST) and the mechanism of inhibitory effect of this protein was studied.

MATERIALS AND METHODS

Cell culture

Human erythroleukemia HLA-A, -B, -C, and -G-negative K562 cells were maintained in RPMI 1640 medium with 10% fetal calf serum (Hyclone). NK cell line NK92 cells (provided by Immune Medicine Inc, Canada) were cultured in α -MEM supplemented with 12.5% fetal calf serum, 12.5% equine serum, 0.2 mM inositol, 0.1 mM β-mercaptoethanol, 0.02 mM folic acid and 100-200 U/ml recombinant IL-2 (BD Pharmingen) at 37°C in a 5% CO2 humidified incubator.

The expression and purification of GST-sHLA-G1α chain

The extracellular domain of HLA-G1 was ligated into a pGEX-4T2 vector (Pharmacia Biotech). The new construct was sequenced to confirm the correct open-reading frame. E. coli BL21 cells were then transformed with the construct pGEX-4T2-sHLA-G1α chain and the control vector pGEX-4T2 respectively. Isopropyl-beta D-thiogalactopyranoside (IPTG, GIBCO) was used to induce high production of control protein GST and target protein GST-sHLA-G1α chain. GST and GST-sHLA-G1α chain products were recovered by gradient dialysis and purified with glutathione-sepharose affinity column according to manual. The target protein was confirmed by immunoblotting. The purity of target protein was determined by SDS-PAGE. Endotoxin content in testing system was determined by using Tachypleus Amebocyte Lysate Kit (Zhanjiang A&C Biological Limited Company). Endotoxin standards were obtained from National Institute for the Control of Pharmaceutical and Biological Products in China.

Cytotoxicity assay

The 4 h 51Cr-release assay was performed to determine the cytolytic activities of NK92 cells. K562 cells were used as target cells (T). The fusion protein GST-HLA-Glα chain or the GST control was preincubated for 30 min at 4°C with NK92 cells before adding 5×103 51Cr-labeled K562 cells (100 μCi of 51Cr sodium chromate/106 cells, Amersham, UK) at different effector/target (E/T) ratios of 5:1 and 2.5:1. After 4 h incubation at 37°C, 50 μl of the supernatant was collected for liquid scintillation counting (Wallac 1410, Pharmacia). Spontaneous release was measured in target cells incubated with media alone and maximal release was measured by treatment of target cells with 1% Triton X-100. Percent specific lysis was determined by the following formula: %specific lysis = (experimental release - spontaneous release)/( maximum release - spontaneous release)×100%. Results were presented as means ± SD of triplicate samples.

Detection of receptor ILT-2 on NK92 cells

NK92 cells were washed with PBS and incubated with the monoclonal antibody against ILT-2 (anti-CD85, BD Pharmingen) at 4°C for 2 h. After thoroughly washing, they were incubated with FITC-goat-anti-mouse IgG (Southern Biotechnology Associates) at 4°C for 2 h and washed again, and the expression of ILT-2 was detected by flow cytometric analysis (FACSCalibur, BD, USA). Data was analyzed with Cell Quest 3.0 from BD Corporation.

Fixation of target cells17

K562 cells were washed with PBS and incubated with 1% paraformaldehyde on ice for 30 min. Cells were then washed again to remove paraformaldehyde.

Immunoprecipitation and Western blotting

The NK92 cells were incubated with fixed K562 cells with or without the presence of GST-sHLA-G1α chain respectively for 5,15, 30, 60 min respectively. The cells were lysed and the debris was cleared. The lysate was then incubated with anti-ILT-2 antibody overnight and subsequently with protein A-agarose for 4 h at 4°C. After washing, the beads were boiled in SDS buffer and the supernant was collected. All samples were separated by SDS-PAGE followed by Western blotting. Antibodies used were: 87G, (a anti-sHLA-G1 mAb, gifted by Dr. D. Geraghty, Fred Hutchinson Cancer Research Center, USA), PY20, a specific antibody against the phosphorylated tyrosine (Phosphotyrosine Ab-1, Neomarkers, USA), anti-phospho-MEK (Cell Signaling Technology, Inc), anti-MEK (Cell Signaling Technology, Inc), anti-phospho-ERK (Cell Signaling Technology, Inc), anti-ERK (sc-94, Santa Cruz Biotechnology, Inc), CTD110.6 (anti-O-GlcNAc, Covance, USA) and PTP1C/SHP1 (p17320, BD Transduction Laboratories). Proteins were detected by NBT and BCIP response.

RESULTS

Identification of the GST-sHLA-G1α chain fusion protein

The sHLA-G1α chain gene sequence was identical to the extracellular fragment of HLA-G1α chain whose gene bank number is AF226990. BL21 cells transformed with the pGEX-4T2-sHLA-G1α chain vector or the pGEX-4T2 control vector were induced with IPTG, and the bacterial protein products were examined. Coomassie blue-stained sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) of induced or uninduced cell lysates showed that 59 kD fusion protein was overproduced in the pGEX-4T2-sHLA-G1α chain transformed BL21 cells (Fig 1A), while the pGEX-4T2 transformants produced only a 29 kD GST protein. Western blot profiles contained a band corresponding to the 59 kD protein, which was specifically recognized by 87G mAb12, a specific antibody against sHLA-G1 (Fig 1B), confirming that the fusion protein was derived from the HLA-G1α chain coding sequence. Soluble target protein was purified by gluta-thione-sepharose affinity column to 98% purity determined by SDS-PAGE. Immunoblots showed that the purified protein was GST-sHLA-G1α chain. The final endotoxin content was 2.2 EU/mg of protein.

Figure 1
figure 1

Detection of GST-sHLA-G1α chain by SDS-PAGE and Western blotting. (A) Detection of GST-sHLA-G1α chain by 12% SDS-PAGE. (B) Detection of GST-sHLA-G1α chain by Western blotting with antibody 87G. Lane 1, total cell proteins before induction. Lane 2, total cell proteins after induction with IPTG. Lane 3, GST-sHLA-G1α chain eluted from gluthathione-sepharose column. M, standard protein.

Full size image

Inhibition of NK92 Cytotoxicity by GST-sHLA-G1α chain

The 4 h 51Cr-release assay showed that 10 μg/ml GST-sHLA-G1α chain protein was able to inhibit NK92 cytolytic activity at the E/T ratio of 2.5:1, while the control GST protein could not (Fig 2).

Figure 2
figure 2

Effect of GST-sHLA-G1α chain on the cytotoxic activity of NK92 cells. Purified GST-sHLA-G1α chain and GST were used at indicated concentrations. Results were presented as the percentage lysis in a 4 h 51Cr-release assay. At the E/T ratio of 5:1, 10 μg/ml GST-sHLA-G1α chain could decrease NK92 cytotoxic activity compared to medium and GST (control) groups. At the (E/T) ratio of 2.5:1, 10 μg/ml GST-sHLA-G1α chain protein was able to inhibit NK92 cytotoxic activity.

Full size image

Detection of the ILT-2 receptor by flow cytometric analysis

Three HLA-G inhibitory receptors have been identified on NK cells: KIR2DL4, ILT-2, ILT-4. ILT-2 can be identi-fied by anti-CD85 antibody. The flow cytometric analysis result showed that ILT-2 was expressed on NK92 cell line (Fig 3).

Figure 3
figure 3

Flow cytometry analysis of ILT-2 receptor on NK92 cell surface. ILT-2 receptor expression was detected by flow cytometry analysis using anti-CD85 antibody and FITC-goat-anti-mouse IgG. Left: control, which was NK92 cells without incubation with antibody, M1: the expression ratio of ILT-2 on NK92 cells, which was 93.85%.

Full size image

Effect of GST-sHLA-G1α chain on the MEK-ERK pathway in NK92 cells

The protein-tyrosine kinases (PTKs) have been proved to provide early and requisite signals for cytotoxic function18, 19. Since the phosphorylation of tyrosine can represent PTK activity, we next examined the effect of GST-sHLA-G1α chain on the tyrosine phosphorylation in NK92 cells induced by susceptible targets. Western blot results showed that tyrosine phosphorylation increased rapidly and reached its peak at 15 min after stimulation with K562 cells (Fig 4A).

Figure 4
figure 4

Inhibition of the activating signals in NK92 cells by GST-sHLA-G1α chain. NK92 cells interacted with K562 for 0-60 min. +, the presence of GST-sHLA-G1α chain in the incubation system. (A) Detection of tyrosine phosphorylation by Western blotting with PY20. Arrowheads represent some proteins whose phosphorylation was completely inhibited by GST-sHLA-G1α chain. M, the protein markers. (B) Detection of phosphorylated MEK1/2 by Western blotting with anti-p-MEK (top). The blots were probed with anti-MEK1/2 to show equal loading of all lanes (bottom). (C) Detection of phosphorylated ERK1/2.

Full size image

Mostly the phosphorylation of tyrosine kinases triggers downstream events via Raf-MEK-ERK pathways. Downstream molecules, MEK and ERK were essential to regulate perforin and granzyme B mobilization and redistribution towards the contact zone between NK cells and target cells during NK cell cytotoxicity20. The MEK-ERK pathway was rapidly activated upon target engagement in human NK92 cells and played an important role in cytotoxicity of NK92 cells. We found that GST-sHLA-G1α chain inhibited some protein tyrosine phosphorylation (Fig 4A, bands indicated by arrowheads) and subsequently blocked the activation of MEK-ERK (Fig 4B, 4C).

Recruitment of SHP-1 by ILT-2 receptor

ILT-2 has four immunoreceptor tyrosine-based inhibition motifs (ITIMs)21, with consensus sequence (I/V)xYxx(L/V) contributing to the recruitment of Src homology 2 domain-containing tyrosine phosphatases (SHP). GST-sHLA-G1α chain could bind to the receptor ILT-2 and then ILT-2 recruited SH2 domain-containing protein tyrosine phosphatase 1 (SHP-1) (Fig 5). In contrast, SHP-1 recruitment was not observed in NK92 cells alone or NK92 cells which were incubated with K562 cells. The results showed that GST-sHLA-G1α chain could bind to the receptor ILT-2 and the latter recruited SHP-1 which dephosphorylated some important PTKs and blocked the activation of downstream molecules such as MEK and ERK to inhibit the cytotoxicity of NK92 cells.

Figure 5
figure 5

Recruition of SHP-1 by ILT-2 receptor. The cell lysates were immunoprecipitated with anti-human ILT-2 Ab (anti-CD85). Western blots were probed with anti-SHP-1 Ab. 0 represents sample from NK92 cells without stimulation. Numbers, the time after stimulation by K562 cells. +, NK92 cells incubated with GST-sHLA-G1α chain.

Full size image

Effect of GST-sHLA-G1α chain on the O-GlcNAc modification in NK92 cells

The dynamic glycosylation of serine or threonine residues on nuclear and cytosolic proteins by O-linked β-N-acetylglucosamine (O-GlcNAc) is adundant in all multicellular eukaryotes and O-GlcNAc modification has been reported to be a new regulatory modification important to signal transduction cascades22. To the best of our knowledge, our results showed that O-GlcNAc modification decreased during NK cell cytotoxicity and that GST-sHLA-G1α chain could inhibit this decrease (Fig 6), indicating that O-GlcNAc might get involved in the signal transduction of NK cells.

Figure 6
figure 6

Detection of the O-GlcNAc modification in NK92 cells. Whole cell lysates were analyzed by Western blotting with CTD110.6. + represents the system with the presence of GST-sHLA-G1α chain. Numbers represent the time after stimulation by K562 cells.

Full size image

DISCUSSION

GST-sHLA-G1α chain was produced as a fusion protein in E. coli and recognized by the anti-sHLA-G1 antibody, 87G mAb. According to the cytotoxic assays, this soluble GST-sHLA-G1α chain protein was able to inhibit NK92 cytotoxic activity at the effector / target (NK92/K562) ratio of 2.5:1, while the control GST protein could not, which indicates that the inhibition of NK92 cytotoxic activity was exerted by GST-sHLA-G1α chain, but not by GST fragment. The current study afforded a recombinant protein with biologic activity and lower endotoxin content. Furthermore, this GST-sHLA-G1α chain fusion protein was not glycosylated or conjugated with α2 microglobulin. The results showed that in addition to membrane-bound HLA-G1 molecules, the soluble GST-sHLA-G1α chain could also induce immunotolerance by inhibiting NK cell cytotoxicity.

In this study, ILT-2, a HLA-G receptor, was found to be expressed on NK92 cells by flow cytometry. In addition, NK92 cells also expressed an endogenous KIR2DL4 at low level14. This GST-sHLA-G1α chain may also be useful for identifying new HLA-G receptors on NK cells. MHC-recognizing receptors have been reported to recruit SHP-1 or SHP-2 to dephosphorylate23. Our results showed that GST-sHLA-G1α chain bound to ILT-2 and then the latter recruited SHP-1 to dephosphorylate some important and early signal molecules such as PTKs so that the downstream MEK-ERK pathway was consequently blocked.

The balance of activating and inhibitory signal transduction of NK cells is a complicated multifactorial process. Recently O-GlcNAc modification has been reported to be involved in regulating signal transduction in eukaryotes24. Regulation of O-GlcNAc levels by O-GlcNAc transferase and GlcNAcase may be analogous to the regulation of phosphorylation and dephosphorylation by kinases and phosphatases. O-GlcNAc modification of certain proteins is known to fluctuate during T cell activation, insulin signaling, glucose metabolism, and cell cycle progression24. For examples, the elevation of O-GlcNAc levels was associated with insulin resistance in adipocytes25 and the O-GlcNAc modification of tau, AP3 and so on was perturbed in Alzheimers disease patients26. In the present study, we reported that O-GlcNAc modification was involved in cytotoxic signal transduction in NK cells and that GST-sHLA-G1α chain could inhibit the alteration of O-GlcNAc level in NK cells. Several groups have documented reciprocity between O-GlcNAc and O-phophorylation24. A given serine/threonine may exist in three states: glycosylated, phosphorylated, or unmodified. O-GlcNAc could have an effect by blocking phosphorylation and/or by an independent mechanism, such as the modulation of protein-protein interactions. Which is the real one in NK cells during cytotoxic activity is under further study.

In conclusion, the GST-sHLA-G1α chain was produced and its function was studied. It could inhibit NK cell cytotoxicity by interacting with receptor ILT-2 and consequently recruiting SHP-1, blocking some protein tyrosine phosphorylation, inhibiting the activation of MEK and ERK signaling pathway. The results also showed that the modification of O-GlcNAc was involved in NK activating and inhibitory signals. Therefore, further studies are needed not only on the signaling pathway initiated by receptors but also how the multiple signals are integrated for regulation of NK cell responses.