A Functional Role of HLA-G Expression in Human Gliomas: An Alternative Strategy of Immune Escape

Abs and reagents

The following Abs were used: 87G (IgG2b, anti-HLA-G; kindly provided by D. Geraghty, Fred Hutchinson Cancer Research Center, Seattle, WA), 4H84 (mIgG1, anti-HLA-G; kindly provided by M. McMaster, University of California, San Francisco, CA); MEM-G/9 (mIgG1, anti-HLA-G; Exbio, Prague, Czech Republic); BFL1.1.2 (mIgG2b, anti-HLA-G; Immunotech, Marseille, France); W6/32 (mIgG2a, anti-HLA-A, -B, -C, -G, -E and β₂-microglobulin (β₂m); Biozol, Eching, Germany); L243 (mIgG2a, anti-HLA-DR), BB7.2 (mIgG2b, anti-HLA-A2; American Type Culture Collection, Manassas, VA); RPA-T4 (mIgG1κ, FITC- and PE-labeled anti-CD4); HIT8A (mIgG1κ, PE-labeled anti-CD8), Annexin V-FITC (BD PharMingen, Heidelberg, Germany); B9.11 (mIgG1, FITC-labeled anti-CD8), N901 (NKH-1; mIgG1, PE-labeled anti-CD56) (Immunotech); goat anti-mouse F(ab′)₂ (Dianova, Hamburg, Germany); and peroxidase-labeled rabbit anti-human β₂m (DAKO, Hamburg, Germany). The Abs were titrated for flow cytometry and used at concentrations indicated (usually 20 μg/ml) in the functional assays. Soluble CD95L was obtained from CD95L-transfected murine N2A neuroblastoma cells (20). Propidium iodide (PI) was purchased from Sigma-Aldrich (St. Louis, MO).

Cell culture

The human malignant glioma cell lines LN-18, U138MG, U87MG, LN-428, D247MG, T98G, LN-319, LN-229, A172, U251MG, U373MG, and LN-308 (kindly provided by Dr. N. de Tribolet, Neurosurgical Service, Centre Hospitalier Universitaire, Vaudois, Lausanne, Switzerland) and SV40 fetal human astrocytes non-neoplastic astrocytes (kindly provided by A. Muruganandam, Institute for Biological Sciences, National Research Council of Canada, Ottawa, Canada) were cultured in 75-cm³ Falcon plastic flasks (BD Biosciences, Heidelberg, Germany) using DMEM supplemented with 1% glutamine (Life Technologies, Paisley, U.K.), 10% FCS (Biochrom, Berlin, Germany), and penicillin (100 IU/ml)/streptomycin (100 μg/ml) (Life Technologies). The JEG-3 human choriocarcinoma cell line and the K-562 human erythroleukemia cell line (American Type Culture Collection) were maintained in RPMI 1640 supplemented with 1 mM sodium pyruvate, penicillin (100 IU/ml)/streptomycin (100 μg/ml), and 10% FCS. Cells were routinely tested for contamination with mycoplasma. Where indicated, the cells were irradiated using a Gammacell 1000 Elite (Nordion, Ottawa, Ontario, Canada).

Immunohistochemistry

Tumor biopsy specimens were surgically removed from four patients with glioblastoma (grade IV, World Health Organization; male, age range 32–64 years old, mean 46 years old) and one patient with anaplastic oligo-astrocytoma (grade III, World Health Organization; female, 42 years old). Frozen sections 10 μm thick were cut, fixed in acetone, and immunostained with anti-HLA-G Ab (87G, BFL1.1.2). The reaction product was visualized with the streptavidin-biotin method (reagents from DAKO) using diaminobenzidine (Serva, Heidelberg, Germany) as an electron donor. As control tissue, placenta was used.

HLA-G transfectants

HLA-G transfectants were obtained as described elsewhere (18). In brief, HLA-G1 (encoding the full-length HLA-G transcript) and HLA-G5 (coding for the secreted HLA-G H chain) plasmids were generated by cloning HLA-G1 and HLA-G5 cDNAs into a GFP vector (pIRES2-pEGFP; Clontech Laboratories, Palo Alto, CA). GFP vector transfectants (pIRES2-pEGFP) were used as controls. Transfection was done using Effectene transfection reagent (Qiagen, Hilden, Germany). Cells were selected in medium containing 0.5 mg/ml neomycin (G418; Calbiochem, Bad Soden, Germany). The following transfectants were used in the study: K-562-HLA-G1, U87MG-pIRES2-EGFP (referred to as U87MG-pEGFP), U87 MG-HLA-G1, and U87MG-HLA-G5.

PBMCs and purified lymphocyte populations

PBMCs were isolated from the peripheral blood of normal healthy volunteers by density gradient centrifugation using Biocoll Separating Solution (Biochrom). Cells were analyzed for their HLA-A2 isotype by flow cytometry. CD4 and CD8 T cells were purified by positive selection with CD4 or CD8 Microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany). NK cells were isolated by an NK cell-negative isolation kit (Dynal Biotech, Hamburg, Germany). The purity of the isolated subpopulations used in the experiments was >95% as determined by flow cytometry.

Flow cytometry

Adherent cell lines were collected after treatment with cell dissociation buffer (Life Technologies) at 37°C for 5 min. JEG-3, K-562-HLA-G1, or K-562 cells were used as positive and negative control cell lines for HLA-G or MHC expression when indicated. Cells were washed with PBS containing 0.1% BSA and 0.1% sodium azide and blocked with human Igs (Alphaglobin; Grifols, Langen, Germany) for 10 min at 4°C. After one washing step, the unlabeled first Ab was added at the final concentration. Isotype control mAbs were used at the same concentration as the primary Ab. Incubation was done on ice for 30 min, followed by two washes. Goat anti-mouse IgG (F(ab′)₂-PE (5 μg/ml; Sigma-Aldrich) or IgG (F(ab′)₂-di-chlorotriazinyl-fluoresceine (10 μg/ml; Dianova) were used as secondary Abs. After washing, PI was added to the cells at a final concentration of 0.02 μM. PI-positive (nonviable) cells were excluded from analysis. Specific fluorescence indexes (SFIs) were calculated by dividing mean fluorescence obtained with specific Ab by mean fluorescence obtained with isotype control Ab. Analyses of PBMCs and lymphocyte subsets were performed similarly. Cells were blocked with human Igs and stained either with directly labeled Abs or with the respective combinations of an unlabeled first Ab and a fluorescent secondary Ab. Flow cytometry was performed using a FACSCalibur (BD Biosciences, Heidelberg, Germany).

Detection of apoptosis by Annexin V binding

PBMCs or purified lymphocyte subsets undergoing apoptotic cell death were analyzed by staining with FITC-labeled Annexin V (BD PharMingen). After various incubation times with the glioma cells, lymphocytes were collected, washed with PBS, and resuspended in a buffer containing 10 mM HEPES/NaOH (pH 7.4), 140 mM NaCl, and 2.5 mM CaCl₂. Then Annexin V-FITC and PI were added. After incubation for 30 min, the cells were analyzed by flow cytometry. When analysis of lymphocyte subsets was desired, PBMCs were first incubated with respective PE-labeled cell surface Abs (CD4-PE and CD8-PE from BD PharMingen; CD56-PE from Immunotech) before the Annexin V staining was performed.

Primary alloreactive proliferation

To suppress the proliferation of stimulator cells, glioma cells were incubated in serum-free medium for 24 h, irradiated at 50–200 Gy, and maintained in serum-free medium for another 24 h. Cells were detached using cell dissociation buffer (Life Technologies), counted, and seeded in RPMI 1640 medium. HLA-A2-mismatched responder PBMCs (10⁵) were coincubated with 10⁴ irradiated glioma cells in a total volume of 100 μl. The assay was performed in 96-well plates in triplicate. PBMCs without glioma cells, glioma cells without PBMCs, and PHA-stimulated PBMCs (10 μg/ml) were used as controls. After 4 days, the cells were pulsed for 24 h with 0.5 μCi of [methyl-³H]thymidine (Amersham Pharmacia Biotech, Freiburg, Germany). Cells were harvested with a cell harvester (Inotech, Dottikon, Switzerland), and the incorporated radioactivity was bound to a glass fiber filtermat (Wallac, Turku, Finland). The filtermat was wetted with Ultima Gold Scintillation Cocktail (Packard, Dreieich, Germany), and radioactivity was determined in a 1450 Microbeta Plus Liquid Scintillation Counter (Wallac).

Secondary alloreactive lysis (⁵¹Cr release)

To test for lytic activity of primed cytotoxic T cells after coculture with HLA-A2-mismatched glioma cells (U87MG-pEGFP, U87MG-HLA-G1, U87MG-HLA-G5) for 5 days, a ⁵¹Cr release assay was performed. Growth-suppressed glioma cells were seeded into 25-cm² flasks to a density of 10⁶ cells. HLA-A2-mismatched PBMCs (1.5 × 10⁷) were added in 5 ml of RPMI 1640 containing 10% FCS. Cells were cocultured for 5 days. Glioma cells (10⁶) were labeled by addition of 100 μCi ⁵¹Cr (catalog no. NEZ147; NEN, Boston, MA) in 1 ml, and 10⁴ labeled glioma cells were seeded as target cells in a U-shaped 96-well plate. Primed alloreactive cytotoxic lymphocytes obtained after 5 days of coculturing were removed and added to the labeled target cells at different E:T ratios (ranging from 10 to 80) in a volume of 100 μl. After coincubation for 4 h, 50 μl of the supernatant was transferred to a Luma-Plate 96 (Packard), dried overnight, and measured. To correct for spontaneous release of ⁵¹Cr, a control of labeled target cells in medium only was included (0% specific lysis). The maximum ⁵¹Cr release possible was determined by addition of Nonidet P-40 (100% lysis). Lysis was calculated as follows: [cpm (effector cells) − cpm (spontaneous)]/[cpm (Nonidet P-40) − cpm (spontaneous)] × 100%.

Primary alloreactive lysis (⁵¹Cr release)

Primary alloreactive lytic damage of glioma target cells by HLA-A2-mismatched lymphocytes was measured in a ⁵¹Cr release assay. Target glioma cells were radioactively labeled and seeded into 96-well U-bottom wells as described above. For blocking experiments, glioma cells or HLA-G transfectants were preincubated with specific Abs or the corresponding isotype control Abs. After incubation for 1 h at 37°C, cross-linking by a secondary Ab (goat anti-mouse F(ab′)₂, 10 μg/ml; Dianova) was performed to immobilize the Ab on the glioma cells. Effector cells (PBMCs, CD4, CD8, NK cells) were added at E:T ratios ranging from 1 to 100 in a total volume of 100 μl. After coincubation for 8 h, 50 μl of the supernatant were transferred to a Luma-Plate 96 (Packard), dried overnight, and measured. In experiments using purified NK cells, incubation was done for 4 h. To correct for spontaneous ⁵¹Cr release, a control of labeled target cells in medium only was included (0% specific lysis). The maximum ⁵¹Cr release possible was determined by addition of Nonidet P-40 (100% lysis).

sHLA-G ELISA

Secreted HLA-G molecules were measured as total soluble HLA class I molecules in the culture supernatants. MHC class I Ab (W6/32, concentration 5 μg/ml, volume 100 μl) was coated onto a 96-well microtiter plate (Nunc, Roskilde, Denmark) and incubated overnight at 4°C. After three washes with PBS containing 0.05% Tween 20 (PBS-Tween), the plate was saturated with 250 μl of PBS containing 2% BSA. After incubation for 2 h at 37°C, the plate was washed three times with PBS-Tween. Cell culture supernatants (100 μl) were added to each well at various dilutions and incubated for 2 h at 37°C. Plates were washed another three times with PBS-Tween, and peroxidase-labeled rabbit anti-mouse β₂m Ab (dilution, 1/500; volume, 100 μl) was added. After incubation for 1 h at 37°C and washing three times with PBS-Tween, 100 μl of ortho-phenyl-diamine peroxidase substrate was added (Abbott Diagnostics, Wiesbaden-Delkenheim, Germany) for 30 min at room temperature avoiding exposure to light. The reaction was stopped by addition of 50 μl of 2 M H₂SO₄. The concentration was estimated from the absorption measured at 490 nm (triplicate wells) on an ELISA microplate reader (Wallac).

Reverse transcription and amplification

First-strand DNA synthesis of 1 μg of total RNA primed with oligo(dT)_12–18 (Amersham Pharmacia Biotech) and Expand Reverse Transcriptase (Boehringer Mannheim, Mannheim, Germany) was done according to the manufacturer’s recommendations (22). One microliter of cDNA was amplified using the following primer pairs, which coamplify the HLA-G isoforms HLA-G1, -G2/4, and -G3: Ex1 (5′-CAAGGATGGTGGTCATGGCG-3′) and Ex6/3′-untranslated (UT) (5′-CAGCTGTTTCACATTGCAGCCTG-3′), 60°C annealing temperature, 40 cycles. PCR products were analyzed on 1.5% agarose gels containing ethidium bromide or on 8% native polyacrylamide gels followed by silver staining (23).

DNA probes and Northern blot analysis

RNA was isolated with the TriPure Isolation reagent (Boehringer Mannheim). A total of 20 μg of total RNA was separated on 1.2% formaldehyde agarose gels, transferred to Hybond-N⁺ membrane (Amersham Pharmacia Biotech) in 20× SSC overnight and cross-linked by baking at 80°C for 2 h. The filters were hybridized with 10 ng/ml specific digoxigenin-labeled probe and were detected as described (21). Hybridization probes specific for HLA-G transcripts were derived from the 3′-UT regions and were generated by PCR and incorporation of digoxigenin-labeled dideoxyuridine triphosphate (Boehringer Mannheim). The 486-bp HLA-G probe was labeled from a subcloned 287-bp amplificate with the primers M13 (-20) and M13 reverse (25). The 440-bp HLA-B probe was generated from the subcloned 3′-UT PvuII/PstI fragment using SP6 and T7 primers. To detect HLA-G transcripts in HLA-G transfectants, a 787-bp HLA-G4 probe was generated by PCR using the primer pair Ex1 and Ex6/3′-UT.

Immunoblot analysis

Cells were lysed for 30 min at 4°C in lysis buffer (20 mM Tris-HCl (pH 7.5), 1% Triton X-100, 2% trasylol; Bayer, Leverkusen, Germany) and 1 mM PMSF (Sigma-Aldrich). After removal of cellular debris by centrifugation, protein content of the lysates was quantified using a MicroBradford assay (24). Lysates equivalent to 100 μg of protein were separated by 10% SDS-PAGE and transferred onto nitrocellulose membranes (Schleicher and Schüell, Dassel, Germany) using standard blotting techniques. HLA-G H chains were detected with the mAb 4H84 and the ECL Western blotting analysis system (Amersham Pharmacia Biotech).

Statistical analysis

All data are representative of experiments performed at least three times with similar results. Significance was assessed by two-sided t test (∗, p < 0.05; ∗∗, p < 0.01).

HLA-G expression in human glioma cell lines

Twelve human glioma cell lines were tested for HLA-G mRNA expression in the absence or presence of IFN-γ by Northern blotting using a probe specific for all HLA-G isoforms (Fig. 1⇓). Six glioma cell lines transcribed HLA-G constitutively (U251MG, U373MG, D247MG, T98G, LN-428, and A172). The presence of different alternative splicing patterns were confirmed by RT-PCR specific for transmembraneous and soluble isoforms of HLA-G (data not shown). After stimulation with IFN-γ, HLA-G transcripts became detectable in 10 of the 12 glioma lines (Fig. 1⇓). Only U87MG and LN-308 cells were refractory to IFN-γ-mediated induction of HLA-G expression. PCR analysis paralleled these observations, confirming strong up-regulation of different HLA-G isoforms after IFN-γ induction (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Human glioma cells express HLA-G mRNA. HLA-G mRNA levels were assessed by Northern blot analysis. A total of 20 μg of total RNA from untreated (−) or IFN-γ-treated (+) cells (500 U/ml; 48 h) (12 different glioma cell lines) was separated on a formaldehyde agarose gel and hybridized with HLA-G specific probes. The amount of RNA loaded is visualized by ethidium bromide staining of the 28S rRNA (lower panel). The human choriocarcinoma line JEG-3 was used as a positive control.

To test whether the transcripts detected were translated into protein, flow cytometry was performed using the anti-HLA-G mAbs 87G and MEM-G/9. A weak membrane expression of HLA-G was observed in four of 12 untreated glioma cell lines (U373MG, LN-319, T98G, and A172) (Fig. 2⇓). After stimulation with IFN-γ, HLA-G membrane expression increased and was detectable in eight of the 12 tested glioma cell lines, using a cutoff SFI value of 1.3 (Fig. 2⇓). The expression of class I (HLA-A, -B, -C mAb W6/32; HLA-A2 mAb BB7.2) and class II (HLA-DR; mAb L243) molecules was assessed in parallel, showing that all glioma cells expressed HLA class I molecules constitutively. IFN-γ enhanced the surface expression of MHC class I in all cell lines except A172. The MHC class II DR-molecule was expressed constitutively in four of 12 glioma cell lines. IFN-γ induced HLA-DR in all except two cell lines (LN-308 and U138MG). The human SV40-transformed astrocytic cell line SV40-fetal human astrocytes did not show HLA-G RNA or protein expression in the absence or presence of IFN-γ (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Human glioma cells express HLA-G protein. MHC class I, class II, and HLA-G expression by human glioma cell lines in the absence or presence of IFN-γ was determined by flow cytometry. The cell lines were incubated for 48 h with or without IFN-γ (500 U/ml), stained with the MHC class I-reactive mAb W6/32 (HLA-A, -B, -C, -G, -E plus β₂m), the MHC class II-reactive mAb L243 (HLA-DR), or the HLA-G-specific mAb 87G (HLA-G). JEG-3 choriocarcinoma were used as positive control and K-562 was used as negative control (data not shown). The numbers in the upper right corner of each box indicate SFI values. An SFI value of 1.3 was considered positive. Data are representative of four independent experiments with similar results.

HLA-G expression in human brain tissue

Immunohistochemistry detected HLA-G expression in four of five brain tumors. Up to 50% of tumor cells expressed HLA-G in three of the four tested glioblastomas and in one anaplastic oligo-astrocytoma (Fig. 3⇓; staining shown here with mAb BFL1.1.2 and mAb 87G). Positive cells were evenly scattered throughout the specimens. One glioblastoma was negative for HLA-G with all tested Abs.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Analysis of HLA-G expression in vivo. Frozen tissue sections of a glioblastoma (A, C, and D) and term placenta as control (B) were immunostained with the HLA-G-specific mAb BFL1.1.2. (A and B), mAb 87G (C), or IgG-isotype control (D). About 50% of tumor cells expressed HLA-G (glioblastoma, 32 year old male; A, magnification ×185, C, magnification ×420). Positive cells were evenly scattered throughout the specimen. Isotype control (D) of glioblastoma is shown at magnification ×420. Endothelial cells of three blood vessels (asterisks) in placenta control tissue are HLA-G-positive (B, magnification ×395).

Expression of HLA-G in glioma cells protects from alloreactive lytic killing

To elucidate the functional relevance of constitutive and inducible HLA-G expression by glioma cells, different alloreactivity assays were performed. Freshly isolated PBMCs were used as effector cells and incubated with HLA-A2-mismatched glioma targets that had previously been cultured in the absence or presence of IFN-γ. The relevance of HLA-G molecules for the protection of glioma cells from alloreactive lysis was shown by preincubating glioma cells with HLA-G-specific Ab 87G (Fig. 4⇓). HLA-G expressed on U373MG and T98G participated in the protection of glioma cells against direct cytolytic activity of PBMCs as recorded in an 8-h ⁵¹Cr release assay. The MHC class I-specific Ab W6/32 was used as a control. Although W6/32 blocks alloreactive lysis by MHC class I-restricted CD8 T cells, the observed net effect on PBMC lysis in the presence of W6/32 was an increase in specific lysis. MHC class I expression regulates NK cell lysis via interaction with killer inhibitory receptors, therefore blocking MHC class I might have facilitated lysis by NK cells present within the PBMC population. Because W6/32 cross-reacts with HLA-G and the negative immune regulatory molecule HLA-E, part of the observed effects of this Ab on lysis may also have been mediated by an interaction with HLA-E. In U87MG, a glioma cell line lacking HLA-G expression, neutralization with 87G had no significant influence on lysis by alloreactive PBMCs (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

HLA-G expressed on human glioma cells inhibits lysis by alloreactive PBMCs. T98G or U373MG glioma cells were cultured in the absence or presence of IFN-γ (500 U/ml) for 48 h and used as targets. Target cells were preincubated with 87G, W6/32, or isotype control Ab for 1 h. Freshly purified HLA-A2-mismatched PBMCs were added as effector cells at different E:T ratios (shown, 50:1). Data are expressed as percentages of specific lysis recorded in an 8-h ⁵¹Cr release assay (∗, p < 0.05; ∗∗, p < 0.01 compared with isotype control Ab). Data are representative of three independent experiments with similar results.

Gene transfer of HLA-G1 or -G5 into HLA-G-negative glioma cells

To elucidate the effects of the two major isoforms, HLA-G1 and HLA-G5, on antitumoral immune responses, HLA-G transfectants of an HLA-G-negative glioma line (U87MG) were generated. Transfectants showed high levels of HLA-G expression as documented by immunoblot, flow cytometry, and ELISA for soluble HLA-G (sHLA-G) (Fig. 5⇓). The forced expression of HLA-G isoforms did not alter the expression levels or inducibility by IFN-γ of classical MHC class I or II molecules as determined by flow cytometry (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Expression of HLA-G1 and HLA-G5 after gene transfer into U87 MG glioma cells. A, The transfectants were examined by immunoblot analysis using the 4H84 mAb specific for the denatured HLA-G H chain. JEG-3 choriocarcinoma cells and K-562-HLA-G1 were used as positive controls. The HLA-G protein migrates at 39 kDa. B, Cell surface expression of HLA-G1 protein in the HLA-G transfectants was assessed by flow cytometry using the HLA-G-specific Abs 87G (left), MEM-G/9 (right), and BFL1.1.2 (data not shown). Filled histograms indicate specific fluorescence; open histograms correspond to the isotype control (IgG2b) for HLA-G. C, Levels of sHLA-G released into the supernatant by U87MG, U87MG-HLA-G5, and U87MG-HLA-G1 transfectants were determined by ELISA. Data are expressed as arbitrary units corresponding to the OD measured at 490 nm. Data are representative of at least three independent experiments with similar results.

Gene transfer of HLA-G1 or -G5 into U87MG glioma renders glioma cells resistant to primary alloreactive lysis

Gene transfer of HLA-G1 and HLA-G5 rendered U87MG cells less susceptible to direct alloreactive lytic killing by HLA-A2-mismatched PBMCs as assessed by ⁵¹Cr release of the target population (Fig. 6⇓A). Glioma cell death was also monitored by fluorescence microscopy. Modulation of glioma cell lysis after incubation of tumor targets with PBMCs is demonstrated by the significantly higher survival rates of the HLA-G transfectants compared with the vector control transfectants at 48 h (Fig. 6⇓B). To further characterize the effector populations inhibited by HLA-G expressed on U87MG cells, we performed primary alloreactivity assays with purified lymphocyte subpopulations (CD56, CD4, CD8). U87MG parental cells showed resistance to NK cell lysis (Fig. 6⇓C). Therefore, effects of ectopic HLA-G expression could not be attributed to NK cell inhibition. In contrast, MHC class I-negative K-562 cells were clearly protected from NK lysis by the expression of HLA-G1. Unexpectedly, lytic assays with CD8 and CD4 effector cell subsets demonstrated that HLA-G1 directly inhibited lysis mediated by alloreactive CD4 and CD8 effector cell populations (Fig. 6⇓D).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Inhibitory effects of HLA-G in U87MG glioma cell transfectants on different subsets of immune effector cells under primary alloreactive coculture conditions. A, U87MG-pEGFP, U87MG-HLA-G1, or U87MG-HLA-G5 cells were used as targets in cocultures with freshly isolated HLA-A2-mismatched PBMCs at different E:T ratios (shown 100:1). Data are expressed as percentages of specific lysis recorded in an 8-h ⁵¹Cr release assay (∗, p < 0.05; ∗∗, p < 0.01 compared with U87MG-pEGFP) B, U87MG-pEGFP or U87MG-HLA-G1 were cocultured in six-well culture plates with HLA-A2-mismatched PBMCs for 48 h. Visualization over time was performed by fluorescence microscopy. C, CD56-positive NK effector cells were added to U87MG-pEGFP, U87MG-HLA-G1, or U87MG-HLA-G5 target cells at the indicated E:T ratios. As a positive control, the MHC class I-negative K-562 and K-562-HLA-G1-transfectants were used. D, Freshly isolated, HLA-A2-mismatched PBMCs or purified CD4, CD8, or CD56 subsets were incubated with U87MG-pEGFP or U87MG-HLA-G1 target cells at different E:T ratios (shown 80:1) for 8 h, and specific lysis of the target glioma cells was quantified by release of ⁵¹Cr (∗∗, p < 0.01 compared with U87MG-pEGFP vector control cells). Data are representative of at least three independent experiments with similar results.

HLA-G gene transfer inhibits alloproliferation and prevents effective priming of Ag-specific cells

To study the effects of HLA-G on the generation of primary and secondary immune responses, we first quantified the primary proliferative response of HLA-A2-mismatched responder PBMCs against growth-arrested glioma cells (U87MG-pEGFP, U87MG-HLA-G1, or U87MG-HLA-G5) by measuring [methyl-³H]thymidine incorporation. The proliferative response toward glioma cells was significantly suppressed by U87MG-HLA-G1 and U87MG-HLA-G5 cells (Fig. 7⇓A). To assess whether induction of apoptosis in the effector cells (19) contributes to these inhibitory effects, the cellular viability within the effector cell population was assessed during coculture of freshly isolated lymphocyte subpopulations (CD8, CD4) with U87MG-pEGFP, U87MG-HLA-G1, or U87MG-HLA-G5 transfectants. HLA-G1 did not induce apoptosis in the effector cell populations as detected by Annexin V staining up to 72 h after coincubation of freshly isolated and not previously stimulated PBMCs or lymphocyte subsets with target cells (Fig. 7⇓B). As a positive control for the induction of apoptosis, PHA-stimulated PBMCs were treated with CD95L.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

HLA-G inhibits primary alloproliferative responses, priming of Ag-specific effector lymphocytes, and lytic activity of Ag-specific cytotoxic T-cells. A, PBMCs were cultured for 5 days in the absence or presence of growth-arrested stimulator cells (U87MG-pEGFP, U87MG-HLA-G1, U87MG-HLA-G5). Primary alloproliferation of the HLA-A2-mismatched responder population was determined by [³H]thymidine incorporation added for the last 24 h of coculture. Data are expressed as mean cpm ± SD of triplicate cultures (∗∗, p < 0.01 compared with U87MG control cells). B, U87MG-pEGFP or U87MG-HLA-G1 cells were cocultured with HLA-A2-mismatched PBMCs, CD4, or CD8 T cells, and the amount of apoptosis in the lymphocyte population was quantified by Annexin V-FITC at 48 h. As a positive control for apoptosis induction, PHA-stimulated PBMCs treated with CD95L are shown. C, Ag-specific cytotoxic T cells were generated by coculture of HLA-A2-mismatched PBMCs with U87MG-pEGFP or U87MG-HLA-G1 target cells for 5 days. Primed effector cells were removed and incubated with ⁵¹Cr-labeled U87MG-pEGFP target glioma cells at the indicated E:T ratios. Data are expressed as percentage of specific lysis recorded in an 8-h ⁵¹Cr release assay. D, Ag-specific cytotoxic T cells were generated by coculturing HLA-A2-mismatched PBMCs with U87MG-pEGFP (open bars) or U87MG-HLA-G1 (filled bars) as described in C. After removing primed lymphocytes from the cultures, effector cells were incubated with either U87MG-pEGFP or U87MG-HLA-G1 as targets. Data are expressed as percentage of specific lysis recorded in an 8-h ⁵¹Cr release assay. Data are representative of at least three independent experiments with similar results.

Next we investigated the effect of HLA-G1 on mounting secondary immune responses. Ag-specific cytotoxic lymphocytes were generated by coculturing U87MG-pEGFP or U87MG-HLA-G1 target cells with HLA-A2-mismatched PBMCs for 5 days. After transferring primed effector cells, specific killing of parental U87MG target cells was assessed by ⁵¹Cr release at different E:T ratios. The expression of HLA-G1 in the priming phase abrogated Ag-specific lysis by cytotoxic T cells (Fig. 7⇑C). The protective effect of HLA-G against specific killing by cytotoxic effector cells depended on the presence of HLA-G in the priming phase. When HLA-G was present in the target population after Ag-specific cells had been primed against U87MG-pEGFP control cells, subsequent lysis of U87MG-pEGFP cells was not significantly different from that of U87MG-HLA-G1 (Fig. 7⇑D). Similar results were obtained for U87MG transfectants expressing soluble HLA-G (U87MG-HLA-G5; data not shown).

Few HLA-G-positive cells are sufficient to inhibit alloreactive lysis of HLA-G-negative glioma cells

To assess the possible immune modulatory effects of a minor population of glioma cells expressing HLA-G within a given tumor cell population, ⁵¹Cr-labeled U87MG-HLA-G1 cells were mixed with ⁵¹Cr-labeled vector control U87MG-pEGFP cells, and lysis of targets was assessed after incubation with alloreactive, HLA-A2-mismatched PBMCs. Significant inhibition of lysis was observed when 10% or more of the target cells expressed HLA-G (Fig. 8⇓A). To delineate whether the release of soluble HLA-G as a result of shedding into the supernatant might account for this effect, the levels of soluble HLA molecules in the supernatants were quantified by ELISA. However, the levels of sHLA molecules (classical and nonclassical MHC molecules absorbed by W6/32 mAb) in the supernatant of HLA-G1 cells were unaltered upon coincubation of PBMCs with U87MG-HLA-G1 (Fig. 8⇓B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 8.

Few HLA-G-expressing tumor cells convey immune inhibitory effects. A, U87MG-HLA-G1 cells were mixed with U87MG-pEGFP cells and used as target cells. After coincubation with HLA-A2-mismatched PBMCs, the percentage of specific lysis was recorded in a 8-h ⁵¹Cr release assay. Significant protection was observed, when 10% or more percent of the cells expressed HLA-G (∗∗, p < 0.01; ∗, p < 0.05 compared with U87MG-pEGFP control cells). B, The levels of soluble HLA molecules were measured by ELISA in the supernatants of PBMCs alone, U87MG-pEGFP, U87MG-HLA-G1, U87MG-HLA-G5, or U87MG-HLA-G1 coincubated with PBMC for 48 h. Levels of sHLA molecules were unaltered during coculture of PBMCs and U87MG-HLA-G1 (∗∗, p < 0.01 compared with PBMCs). Data are representative of at least three independent experiments with similar results.