NK resistance of tumor cells from multiple myeloma and chronic ...

Leukemia (2008) 22, 998–1006 & 2008 Nature Publishing Group All rights reserved 0887-6924/08 $30.00 www.nature.com/leu

ORIGINAL ARTICLE NK resistance of tumor cells from multiple myeloma and chronic lymphocytic leukemia patients: implication of HLA-G G Maki1,2, GM Hayes1, A Naji2, T Tyler1, ED Carosella2, N Rouas-Freiss2 and SA Gregory1 1 Section of Hematology and Bone Marrow Transplantation, Division of Hematology/Oncology, Rush University Medical Center, Chicago, IL, USA and 2Service de Recherches en He´matologie-Immunologie, Commissariat a l’Energie Atomique, Hopital Saint-Louis, Paris, France

Exploiting the antitumor effect of natural killer (NK) cells has regained interest in light of data from preclinical and clinical work on the potential of alloreactive NK cells. Multiple myeloma (MM) and chronic lymphocytic leukemia (CLL) represent the two most prevalent adult hematological malignancies in the western hemisphere. To evaluate the role of NK cells in the immune surveillance and their therapeutic potential for CLL and MM, tumor cell susceptibility to NK-mediated killing was investigated. Results show relative resistance of tumor cells from CLL as well as MM (73 and 70% of the patients, respectively) to NK-mediated killing. To gain insight into molecular mechanisms of this resistance, the expression of the tolerogenic HLA-G molecule in CLL and MM and its relevance to susceptibility to NK-mediated killing were investigated. HLA-G transcript was found in tumor cells from 89% (n ¼ 19) of CLL and 100% (n ¼ 9) of MM patients examined. HLAG1 surface expression was observed in CLL and was very low or undetectable in MM. Notably, blocking of HLA-G1 with specific antibody on CLL samples increased their susceptibility to NK-mediated killing, demonstrating that HLA-G participates in protecting CLL cells from NK-mediated killing and may thus contribute to their immune escape in vivo. Leukemia (2008) 22, 998–1006; doi:10.1038/leu.2008.15; published online 21 February 2008 Keywords: NK resistance; CLL; multiple myeloma; HLA-G

Introduction Multiple myeloma (MM) and chronic lymphocytic leukemia (CLL), two clonal B-cell malignancies, represent the two most prevalent adult hematological malignancies in the western hemisphere. MM is characterized by the outgrowth of malignant plasma cells and their accumulation within the bone marrow, which provides the appropriate support for myeloma cell growth and survival.1 CLL is characterized by an accumulation of clonal B cells in the bone marrow, blood and lymphoid organs.2 Despite initial response to conventional treatments and promising results with newer agents, such as immunomodulatory drugs (Thalidomide, Revlimid), proteasome inhibitors (Velcade) and antibody-based immunotherapy, success has been hindered by the emergence of resistant clones, or no response, and MM and CLL remain incurable.3,4 Cellular immunotherapy exploiting the antitumor effect of immune effector cells provides an attractive treatment modality, potenCorrespondence: Dr G Maki, Service de Recherches en He´matologieImmunologie, Commissariat a` l’Energie Atomique, Hopital SaintLouis, 1 Avenue Claude Vellefaux, Paris Cedex 10 75475, France. E-mail:[email protected] Received 8 August 2007; revised 14 January 2008; accepted 15 January 2008; published online 21 February 2008

tially complementing these drugs, and particularly aiming at the eradication of minimal residual disease. Currently, hematopoietic stem cell transplantation, in particular allogeneic stem cell transplantation, that exploits the antitumor effect of immune effector cells can be considered the only potentially curative strategy.5,6 Natural killer (NK) cells are important components of this antitumor activity and NK-based immunotherapy represents a promising approach, which is particularly interesting in the setting of stem cell transplantation where tumor burden is low. Moreover, combining different therapeutic modalities that target the tumor cell through different mechanisms is expected to be more effective in eliminating tumor cells and to provide means for the development of curative strategies for CLL and MM. Natural killer cells are important effectors of innate immunity and regulators of adaptive immunity via production of various cytokines.7,8 NK cells, particularly upon activation with different cytokines, are able to kill tumor cells without prior sensitization and, therefore, play an important role as a first-line defense in the control of malignancies. However, some tumors escape this immune surveillance. The occurrence of malignancies resulting via tumor escape from the immune system may be due to either a defective immune function, in particular of NK cells, or to inherent features of tumor cells causing lack of recognition by NK cells and their subsequent activation to fulfill their effector function. Thus, elucidating those aspects regulating NK interaction with tumor cells is important to understand the role of NK cells in the control of malignancies as well as their potential as a means of immunotherapy for these diseases. NK cytotoxic activity is regulated by a fine balance between signals through activating and inhibitory receptors expressed by NK cells. Inhibitory receptors regulating NK effector functions bind to MHC class I molecules or their homologs on target or opposing cells.9,10 According to the ‘missing self’ hypothesis, NK cells recognize and kill tumor cells mainly because of their lack of or altered expression of MHC-class I molecules.11 Because of the lack of inhibitory receptors corresponding to MHC class I antigens present on target cell, NK cells from allogeneic donors are believed to be more effective in killing tumor cells. This has encouraged a body of preclinical and clinical work on the basis of using allogeneic NK cells to eliminate tumor cells and generated a renewal of interest in the development of NK-based immunotherapy for the treatment of malignancies and in particular hematological malignancies.12–14 Most NK inhibitory receptors are clonally distributed and present on only a small proportion of circulating NK cells, however, receptors for the nonclassical HLA class I molecule, HLA-G, namely ILT2 /LIR1/CD85j15,16 and KIR2DL4,17 are not clonally distributed. KIR2DL4 is expressed on all NK cells of an individual and ILT2 on a subset of NK cells varying in size among different donors. HLA-G expression by tumor cells may

NK resistance of CLL and MM: role of HLA-G G Maki et al

999 therefore represent a more universal mechanism of tumor escape from NK rejection. HLA-G is distinct from the other HLA class I molecules as it exhibits limited polymorphism, restricted tissue distribution and a number of protein variants. Alternative splicing of the primary transcript can generate four membrane-bound (G1–G4) and three soluble (G5–G7) isoforms.18 HLA-G1 may also be found as a soluble protein upon cleavage from cell surface. HLA-G is expressed physiologically on cytotrophoblasts at the feto– maternal interface, where it plays an important role in maternal tolerance of the semiallogenic fetus.19,20 Further, HLA-G expression has been reported in a number of malignancies21 where it may function as a means of tumor escape from immune surveillance.21–23 Membrane-bound and soluble HLA-G exerts immunosuppressive effects via interaction with inhibitory receptors ILT2, ILT4 and KIR2DL4, differentially expressed by NK, T cells and antigen-presenting cells.18,21 Alternatively, HLA-G may serve as a leader sequence allowing for the expression of HLA-E, a cell surface inhibitory protein for which most NK cells express the receptor (CD94/NKG2A/B).24 Recently, a new pathway of immune suppression by HLA-G has been described, which is mediated by immunosuppressive T and NK cells generated by acquisition of HLA-G through a process of membrane transfer called ‘trogocytosis’.25,26 The impact of this immune suppression can, therefore, be beneficial or detrimental depending on its scope, since by its immunoregulatory role, it may facilitate the acceptance of the fetus, allografts and tumors. The aim of this study was to investigate the expression of HLA-G in CLL and MM and determine its relevance in tumorcell susceptibility to NK-mediated killing. Here, we report the expression of multiple HLA-G mRNA isoforms by tumor cells from CLL and MM patients, while being absent from normal counterparts. Furthermore, we found that these cells were mostly resistant to NK-mediated killing and that blocking of membrane-bound HLA-G on CLL samples increased their susceptibility to NK-mediated killing. We propose that HLA-G participates in protecting CLL cells from NK-mediated attack and contributes to their immune escape. These data justify further investigation into the role of HLA-G expression on tumor cells of B origin in the pathophysiology of the disease.

Materials and methods

Patient samples and cell lines Peripheral blood from 19 patients with the diagnosis of CLL, bone marrow aspirates from 13 myeloma patients, peripheral blood and bone marrow aspirates from healthy volunteers as control were obtained according to institutional review board approval. Peripheral blood and bone marrow mononuclear cells (PBMC) were isolated by density centrifugation on FicollHypaque (Pharmacia, Amersham Pharmacia, Uppsala, Sweden). B lymphocytes were then isolated by negative selection magnetic bead using a B-cell isolation kit (Miltenyi Biotech, Auburn, CA, USA), and myeloma cells by CD138 magnetic bead selection. NK cells were isolated by negative selection using an NK cell isolation kit (Miltenyi Biotech) from healthy donors’ peripheral blood mononuclear cells (PBMCs) and kept briefly on low-dose IL-2. NK-sensitive erythroleukemia cell line K562, myeloma cell lines U266, RPMI 8226, IM9, the M8 melanoma cell line and its stable transfectant M8-HLA-G1 were used as target for NK-mediated cytotoxicity assay and maintained in culture in RPMI 1640 supplemented with 10% fetal

bovine serum (FBS). NK-92, a highly cytotoxic human NK cell line, with characteristics of activated NK cells27,28 was used as an effector cell population and maintained in X-VIVO 10 supplemented with 5% human AB serum and 100 U ml–1 interleukin-2 (Chiron, Emervyille, CA, USA). All cell lines were obtained from ATCC (Manassa, VA, USA).

Reagents, antibodies and flow cytometry B-cell isolation kit, CD138 micro beads, NK-cell isolation kit and MACS columns were obtained from Miltenyi Biotech. The following conjugated monoclonal antibodies (mAbs) were used in this study: CD5-FITC, CD19-PE, CD20-FITC and -PE, CD45-FITC, CD56-FITC and -PE, CD16-FITC (all from BD Bioscience/BD Pharmingen, San Diego, CA, USA), CD138-PE, CD107a-FITC clone H4A3 (IgG1) from Diaclone (Besanc¸on, France). The HLA-G-specific antibodies recognizing membranebound HLA-G1 isoform were used: clone MEM-G/9 (mouse IgG1) was obtained from Serotec (Raleigh, NC, USA), and clone 87G (IgG2a) from Exbio (Prague, Czech Republic). Appropriate isotype-matched antibodies were used for controls. Briefly, 2 105 cells were washed in PBS with 2% FBS and 0.2% sodium azide, incubated with labeled antibodies or isotype control for 30 min at 4 1C in the dark and analyzed following three washes. Live cells, gated on forward and side scatter, were analyzed for antigen expression using a Becton Dickinson FACScan or FACSort flow cytometer. A minimum of 5000 events were collected and analyzed for each sample using Hewlett Packard Lysis 2 or Macintosh Cellquest software.

Cytotoxicity assay Chromium release assay. Natural Killer cell cytotoxicity

was performed in standard 4-h 51Cr release assays as previously described.27 Briefly, target cells were labeled with 50 mCi Na2 51 CrO4 (New England Nuclear, Boston, MA, USA) and resuspended at 1 105 cells ml–1, 100 ml of target cells were mixed in triplicate with 100 ml of effector NK cells at various effector/target (E/T) cell ratios ranging from 20:1 to 1:1. For determination of spontaneous release, target cells were incubated with medium alone and for maximum release, with lysing solution (complete medium þ 10% Triton X-100). Following incubation for 4 h, 100 ml of supernatant was transferred to Ready Cap filter scintillants (Beckman Fullerton, CA, USA) and the amount of 51Cr released due to target cell lysis was measured using a Beckman LS6000 scintillation counter.

Calcein assay. Target cells were labeled with 12.5 mM

Calcein AM (Molecular Probes, Eugene, OR, USA)29 and washed twice with phosphate-buffered saline þ 5% FBS (PBS-FBS) and resuspended at 1.5 105 cells ml–1. A quantity of 100 ml of target cells was mixed in triplicate with 100 ml of effector NK cells and was added to each well at various E/T ratios ranging from 10:1 to 1:1. For spontaneous release, target cells were incubated with medium alone (PBS-FBS) and for maximum release with lysing solution (50 mM sodium tetraborate þ 0.1% Triton X-100, pH 9.0). After 3 h incubation at 37 1C, 150 ml of supernatant was removed from each well and transferred to a 96-well flat bottom microtiter plate and fluorescence measured on a Cytofluor II Fluorescence Plate Reader. The specific cytotoxicity in percent was determined as the ratio of experimental release to maximum release (both corrected for spontaneous release). Leukemia


1000

HLA-G-blocking cytotoxicity assay. For HLA-G blocking experiments, target cells were incubated with azide-free HLA-G mAb 87G at a concentration of 10 mg ml–1 for 30 min prior to coincubation with NK cells and then cytotoxicity assay (CD107 mobilization or CytoxiLux) was performed as described. CytoxiLux PLUS assay. The CytoxiLux PLUS assay (OncoImmunin, Inc., Gaithisburg, MD, USA) is a fluorogenic cytotixicity assay measuring the activation of the caspase pathway, which results in the development of a green fluorescence due to the cleavage of a cell-permeable fluorogenic caspase substrate.30 Briefly, target cells were fluorescently labeled (FL2 channel, red fluorescence) and then incubated with effector cells for 90 min at 37 1C in 5% CO2. Cells were then washed and further incubated for 30–45 min in the presence of a fluorogenic caspase substrate (FL1 channel, green fluorescence). Following incubation and washing, samples were analyzed by flow cytometry. Cleavage of the caspase substrate results in increased green fluorescence in dying cells, which will be visualized as double positive events.

CD107mobilization assay. A recently developed flow cytometry-based CD107 mobilization assay31,32 was used to evaluate the effect of blocking HLA-G on CLL cells and their sensitivity to NK-mediated killing. CD107a is expressed in the membrane of cytotoxic granules of cytotoxic lymphocytes and become exposed on the cell surface upon degranulation. This is a measurement of effector activation to kill the target cells rather than the lysis of the target cells and thus provides an accurate assessment of the impact of inhibitory signals. Effector NK-92 and target CLL cells were washed twice and suspended in complete medium (RPMI1640 supplemented with 5% human AB serum) at cell density of 1 106 ml–1 (NK cell) and 5 106 ml–1 (target cells). A quantity of 100 ml of each effector and target cells were mixed and coincubated for 4 h at 37 1C in 5% CO2. Thereafter cells were stained with anti-CD56PE (NK marker) and anti-CD107a-FITC. Cells were then washed

Table 1a PIN CLL1d CLL2d CLL3d CLL4 CLL5 CLL6 CLL7 CLL8 CLL9 CLL10 CLL11 CLL12d CLL13 CLL14 CLL15 CLL16 CLL17 CLL18 CLL19

twice and suspended in PBS-FBS (2%), and analyzed by flow cytometry.

RT-PCR analysis Isolated CLL and MM pellets were lysed in Trizol (Invitrogen, Carlsbad, CA, USA) at a concentration of 1 106 cells ml–1 Trizol. RNA was purified from 100 ml of each Trizol suspension using Qiagen’s (Valencia, CA, USA) RNEasy columns as per manufacturer’s protocol and RNA integrity was confirmed by gel electrophoresis and concentration determined via SmartSpec analysis (Bio-Rad). Purified RNA was treated with amplification grade DNase I (Invitrogen) for 15 min to remove any contaminating genomic DNA before DNase activity was subsequently heat inactivated. Each RNA in the quantity of 0.5 mg was converted to cDNA utilizing Promega’s (Madison, WI, USA) First Strand Synthesis kit. RT-PCR reactions were run (36 cycles at 60 1C annealing, elongation for 2 min at 72 1C) using 25 ng cDNA, Platinum Taq Polymerase (Invitrogen) and 70 nM of both HLA-G forward and reverse primers (primer sequences: 50 ggttctaaagtcctcgctcacc-30 and 50 -gcagcctgagagtagctcc-30 ). Fifteen-microliter aliquots of each reaction were resolved on 1% agarose and representative bands were excised, purified and sequenced to confirm their identities.

Results

Tumor cells from CLL and myeloma patients are resistant to NK-mediated killing Nineteen CLL and ten myeloma patients followed at Rush University Medical Center were included in this study. The median age was 59 years (range, 38–80years) for CLL patients and 57 years (range, 44–70 years) for myeloma patients. The percentage of CLL cells in the blood at the time of the study ranged from 60 to 97% (Table 1a). Four of the CLL patients had been treated and all others had not required treatment. The

Chronic lymphocytic leukemia (CLL) patients’ characteristics Sex/age (years)

Stagea

M/65 F/36 M/53 F/55 M/62 M/79 F/53 M/66

IV 0 0 I I II 0 I

M/75 F/38 M/80 F/64 M/59 M/48 M/68

0 0 IV 0 IV IV 0

CLL cells in sampleb (%)

HLA-G1 by FACSc

60 81 96 90 60 93 70 90 95 64 85 81 61 58 95 77 65 97 80

ND 4.8 1.7 13.7 ND Neg 14.6 1.52 ND 5.6 Neg ND 12.8 6 9.8 2.4 5.2 4.2 17

HLA-G isoforms by PCR Neg G1//G5, G2/G4, G3 G1/G5, G3 G1/G5, G2/G4, G3 G1/G5, G3 Neg G1/ G5, G3 G1 G3 G1/G5, G3 G3 G2/G4 G1/G5, G3 G1/G5, G3 G1/G5 G1/G5, G3 G1/G5, G2/G4, G3 G1/G5, G2/G4, G3 G1/G5, G2/G4, G3

Abbreviations: F, female; M, male; ND, not determined; PIN, patient’s identification number. a Stage defined according to RAI classification. b Percentage of CLL cells was determined by the percentage of PBMCs coexpressing CD5 and CD19 surface markers. c HLA-G surface expression was examined using MEM-G/9 mAb. The percentage of CLL cells positive for HLA-G1 expression is indicated. d Clinical parameters not available. Leukemia


1001 percentage of myeloma cells in the marrow ranged from 5 to 60%. Because of the lack of inhibitory receptors corresponding to MHC class I antigens present on target cells, NK cells from allogeneic donors are believed to be more effective in killing tumor cells. Moreover, NK activity is often altered in these patients.33–35 We tested the sensitivity of patient-derived tumor cells to NK cells from unrelated healthy donors as well as NK-92, a highly cytotoxic human NK cell line with characteristics of activated NK cells27,28 and which represents a model system to investigate susceptibility to NK-mediated killing and mechanisms of tumor escape. Susceptibility of CLL cells to NK-mediated killing was determined on B cells isolated from patient PBMCs, using a B-cell isolation kit, or on unselected PBMCs when the percentage of CLL cells was determined as X90% via coexpression of CD19 and CD5. Calcein assay and flow cytometry-based cytotoxicity assays were performed as described for 19 CLL patients (Figure 1) and even at the highest effector to target ratio of 20:1 at the best 25% of CLL cells were lysed. Only five CLL samples were relatively sensitive to killing by NK-92 cells (X15% killing at an E/T ratio of 20:1) (Figure 1a), while all others were mostly resistant ( no killing or less than 10% killing at an E/T ratio of 20:1) (Figure 1b). These results demonstrate CLL tumor cell resistance to NK-mediated killing. Parallel experiments investigated the sensitivity of MM cells to NK-mediated killing. CD138 is a marker for plasma cells and the pattern of expression of CD138 and CD45 allows discrimination between normal and malignant plasma cell (myeloma cells).36 Myeloma cells were isolated by CD138 selection of bone marrow aspirates from 10 myeloma patients and myeloma-cell susceptibility to NK-mediated killing was evaluated using both these patient-derived myeloma cells and myeloma cell lines (Figure 1). Myeloma cell lines were sensitive to NK-mediated killing (Figure 1c) while patient-derived primary myeloma cells were mostly resistant in these assays demonstrated by only 15% cytotoxicity at the greatest effector/target ratio of 10:1 (Figure 1d). These data show general resistance of primary CLL and myeloma cells to NK-mediated killing. CLL tumor cells were resistant to NK-92-mediated killing, with cytotoxic values of 7.9% ±1.8, 5.4% ±1.4 and 3±1 at E/T ratio of 20:1, 10:1 and 5:1, respectively (n ¼ 19), while MM cells were killed at 6.9% ±2, 1.9% ±2.2 and 0.5% ±0.2 at 10:1, 5:1 and 1:1, respectively (n ¼ 10). No killing of primary bone marrow CD138 þ plasma and B cells from healthy donors was observed (data not shown).

HLA-G is expressed on CLL and myeloma cells To gain insight into the mechanism responsible for this NK resistance, expression of HLA-G and its isoforms on CLL cells was investigated as the expression of HLA-G has been demonstrated in other malignancies to impede NK-mediated killing.22,23 B cells isolated from PBMCs of CLL patients were analyzed by flow cytometry and RT-PCR. Surface expression of HLA-G was determined by staining with mAb MEM-G/9, which detects the full-length isoform HLA-G1. Since current commercially available antibodies allow only the detection of HLA-G1/ G5 isoforms, PCR analysis was performed in parallel to more fully characterize HLA-G expression. Representative PCR bands were purified and sequenced to confirm the HLA-G variants expressed by each sample, and the results are listed in Table 1a. Surface expression of HLA-G on CLL cells varied between 1.5 to 17% of cells (Figure 2). With the exception of 2 (CLL1 and CLL6)

Figure 1 Killing of chronic lymphocytic leukemia (CLL) and myeloma cells by natural killer (NK) cells. CLL cell susceptibility to NK-92 was determined in a calcein assay on B cells isolated from patients PBMCs. Only five CLL samples were relatively sensitive to killing by NK-92 cells (X15% killing at an E/T ratio of 20:1) (a) while all others were mostly resistant (no killing or less than 10% killing at an E/T ratio of 20:1) (b). Myeloma cell lines killing by NK-92 cells was measured in a 4 h 51Cr release assay at various E/T ratios. Results are expressed as mean values of three independent experiments±s.d. (c). Primary myeloma cells isolated from the bone marrow of myeloma patients were tested in a calcein assay. Results show the killing of myeloma cells from nine different patients by NK-92 and for patients MM1, MM2 and MM8 by primary NK cells from three healthy donors (d).

of 19 CLL patients examined, HLA-G transcripts were detected in CLL samples for at least one HLA-G isoform by PCR (Table 1a). All samples exhibiting cell surface HLA-G had concordant expression of the HLA-G1/5 transcript by molecular analysis. B cells isolated from PBMCs of three healthy donors Leukemia


1002 CLL2

CLL3

104

104

104

3

3

3

102

102

81

1

10

4.8

10 0

10

10 0

1

10

0

101 102 103 104

10

10 0

104

104

103

103

103

10

10

CD19

103

10

103

64

102

5.6

101

0

10

0

10 0 101 102 103 104

10

0

10

10

0

10

10 0 101 102 103 104

104

104

103

103

103

10

2.4

102

101

10

103

65

10

10 0 101 102 103 104

101

0

10

10 0 101 102 103 104

CLL18 104

104

103

103

103

0

10

10 0 101 102 103 104

0

10

10 0 101 102 103 104

HLA-G1

10 0 101 102 103 104

HLA-G1

MM13 CD138-

MM12 CD138+

Normal BM

MWM

e

0

CD5

MM12 CD138-

CD5

17

102 101

MM13 unsorted

10

10 0 101 102 103 104

103

80

101

MM11 CD138+

0

10 0 101 102 103 104 104

102

101

MM10 CD138+

10

4.2

102

MM9 CD138+

101

0

CLL19

104

97

5.2

102

101

0

10 0 101 102 103 104

102

10 0 101 102 103 104 104

102

101

0

0

CLL17

104

77

12.8

101

CLL16

102

102

101 10

4

103

61

102

10 0 101 102 103 104

0

10 0 101 102 103 104

CLL13

4

103

102

101 10

4

14.6

1

10 0 101 102 103 104

CD19

10

10

10 0 101 102 103 104

CLL10

4

102

10

0

10 0 101 102 103 104

103

70

1

10

0

104

102

1

10 10

13.7

102

1

10 0 101 102 103 104

CLL7

104

90

0

101 102 103 104

CLL4

102

1.7

102

10

0

101 102 103 104

103

96

1

10

0

104

102

1

10 10

10

MWM

10

MM13 CD138+

10

G1/G5 G2/G4 G3

Figure 2 HLA-G expression on chronic lymphocytic leukemia (CLL) and myeloma cells. CLL cells were analyzed for HLA-G cell surface expression by flow cytometry. Percentage of B-CLL cells was determined by the percentage of PBMCs coexpressing CD5 and CD19 surface markers (panel a and c), surface expression of HLA-G was determined using HLA-G-specific mAb, MEM/G9 (panel b and d). Myeloma cells were isolated from the bone marrow of five multiple myeloma (MM) patients (MM9–MM13) were examined for HLA-G transcripts via RT-PCR and 15 ml aliquots of each reaction were resolved on 1% Agarose. Lanes represent the CD138-selected fraction ( þ ) for patients MM9–MM13, along with the CD138-negative fraction () for MM12 and MM13. For MM13, unsorted bone marrow cells were also tested as were bone marrow cells derived from a healthy donor (normal bone marrow) to serve as control. (MWMFmolecular weight marker) (e).

were similarly analyzed but no HLA-G transcripts were detected (data not shown). Similar HLA-G expression analysis was performed on myeloma cells isolated by CD138 selection of bone marrow aspirates from eight myeloma patients. Multiple HLA-G transcripts were detected by RT-PCR in CD138-positive cell fractions (Table 1b) but not in either the CD138-negative cell fractions nor in bone marrow cells derived from a healthy donor Leukemia

and Figure 2e shows data from five representative MM patients (MM9-MM13). In contrast to the concordance between RT-PCR and flow cytometry analysis for the CLL samples, HLA-G cellsurface expression was low or undetectable on primary myeloma cells and cell lines via flow cytometry (data not shown), suggesting expression of other isoforms including soluble isoforms and/or post-transcriptional regulatory mechanisms may be responsible for the absence of HLA-G protein on


1003 Table 1b

Myeloma patients’ HLA-G PCR

PIN

HLA-G isoforms by PCR

MM1 MM2 MM3 MM4 MM5 MM6 MM7 MM8 MM9 MM10 MM11 MM12 MM13

G1/G5 ND ND ND G1/G5,G3 ND ND G1/G5 G1/G5, G2/G4, G1/G5, G2/G4, G3 G1/G5, G2/G4 ±G2/G4 G1/G5, ±G2/G4, G3

Abbreviations: MM, Multiple myeloma; ND, not determined; PIN, patient’s identification number.

myeloma cells’ surface. Since significant cell-surface HLA-G expression could not be detected in MM cells of any patient, the role of HLA-G in NK susceptibility was not further investigated in myeloma cells but only in CLL cells.

NK-92 cells express the HLA-G receptor ILT2, which can inhibit their cytotoxic activity against HLA-Gexpressing target cells (M8-HLA-G1) The expression of HLA-G inhibitory receptors in NK-92 was analyzed. In contrast to KIR2DL4 surface expression, which is detectable but very low on NK-92,17,28 these cells express high levels of ILT2. To determine the importance of the HLA-G/ILT2 interaction in delivering inhibitory signals during NK-92-mediated killing, we examined the sensitivity of melanoma cell line M8 and its stable transfectant M8-HLA-G1 cells to NK92 cytotoxicity using a flow cytometry-based cytotoxicity assay. Parental M8 cells are HLA-Ipositive but do not express HLA-G protein, while a stable transfectant of this cell line, M8-HLA-G1, constructed and utilized as described in detail previously37,38 expresses the cell surface HLA-G1 isoform. CD107 mobilization is a measurement of effector activation to kill the target rather than the lysis of the target cells themselves and, therefore, provides a more accurate assessment of the impact of inhibitory signals. Results demonstrate that killing of M8-HLA-G1 by NK-92 was significantly lower than that of parental M8 cells (Supplementary Figure 1). Sensitivity to NK92-mediated killing could be restored to M8HLA-G1 cells by blocking HLA-G1 with the specific mAb 87G (Supplementary Figure 1) suggesting that the ILT2/HLA-G pathway is indeed important in controlling NK-92-mediated cytotoxicity toward HLA-G1-expressing targets.

Blocking of HLA-G increases the sensitivity of CLL cells to NK-mediated killing To assess the contribution of HLA-G expression on CLL cells to their susceptibility to NK-mediated killing, cytotoxicity assays were performed in the absence and presence of a blocking antibody against HLA-G. Target cells were incubated with azide-free HLA-G mAb 87G at a concentration of 10 mg ml–1 for 30 min prior to coincubation with NK cells, and then cytotoxicity assay was performed as described. Tumor-cell killing by NK-92 cells was significantly increased when HLA-G was blocked on CLL tumor cells as assessed in a cytotoxicity assay measuring caspase activation in target cell

(Figure 3a). Coordinately, in CD107 mobilization assays measuring NK cell activation to kill the target, blocking of HLA-G significantly increased NK cell degranulation in all samples tested (Figure 3b). These results provide strong support for a contribution of HLA-G in protecting CLL cells from NKmediated attack.

Discussion The nonclassical HLA class-I molecule, HLA-G, was first identified on cytotrophoblast, where it was shown to protect the semi-allogeneic fetus from maternal NK cells while its immunoregulatory action has been shown to contribute to graft acceptance and tumor escape. To date, the expression of HLA-G in hematological malignancies has been investigated in a number of studies39–43 with conflicting findings, presumably due to the heterogeneity of the studied population or technical factors. Amiot et al.39 investigating HLA-G expression in normal and malignant hematopoiesis found HLA-G transcripts only in T and B lymphocytes and a variable proportion of their malignant counterparts (6/15 ALL, 100% of B-NHLs, 20/30 CLL), no protein antigens were detected at the cell surface or in the cytosol of any of these cells. These authors also found increased plasma level of sHLA-G in 70% of CLL 54% of B-NHL and 45% of T-NHL and more recently in acute myeloid leukemia (AML).40 A retrospective study investigating 51 leukemia samples, including 8 CLL did not detect HLA-G surface expression nor transcripts in CLL.41 Notably, HLA-G expression has recently been reported in BCLL and associated with an unfavorable outcome.42 Flow cytometry analysis of CLL samples (n ¼ 47) showed surface expression of HLA-G in a variable proportion of tumor cells in all CLL samples examined that correlated with survival. The authors found that patients with o23% HLA-G-positive leukemic cells (n ¼ 37) had a longer progression-free survival than those with 423% HLA-G (n ¼ 10). Moreover, the level of HLA-G expression correlated with degree of immunosuppression as assessed by levels of IgG, total T cell numbers and CD4 T cells. Based upon these results, HLA-G was hypothesized to be a better prognostic factor than Zap-70 or CD38 status. Leleu et al.43 evaluated both total soluble HLA-class I and sHLA-G in 103 patients with newly diagnosed MM, 30 monoclonal gammopathy of undetermined significance (MGUS) and 30 healthy individuals to determine its prognostic value in relation to b2microglobulin (b2m), a common prognostic factor for MM. These parameters were found to be significantly lower in healthy individuals than in MM or MGUS. While sHLA-I varied across stages of MM, sHLA-G was higher in MM and MGUS regardless of the stage of the disease, concluding that sHLA-I but not sHLA-G could be used as an additional prognostic factor along with b2m. The immunoregulatory role of HLA-G is now well established, including its inhibition of NK cytotoxicity against melanoma22 and glioma,23 however, no data exist demonstrating its potential contribution in protecting tumor cells from NK-mediated killing in hematological malignancies. The aim of this study was to evaluate the role of NK cells in MM and CLL, the two most prevalent adult hematological malignancies in the western hemisphere, first in the occurrence of the disease as well as their potential as a means of immunotherapy in these diseases. In this regard, we sought to assess the expression of HLA-G and its isoforms in CLL and MM and to determine its relevance to susceptibility to NK-mediated killing. Leukemia


1004 CLL19 +NK-92

/ 0

Anti HLA-G Ab 87G

+

Red Fluorescence (CLL cells)

10

-

CLL18 +NK-92 Gated on CLL

+NK-92

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

1%

10

10

10

10

10

10

10

10

10

10

10

10

10

17%

10 10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

1%

10

10

10

10

10

10

10

10

10

10

10

17%

10

10

10

10

10

10

30%

22%

10

10 10

10

10

10

10

10

10

10

10

10

Green Fluorescence (Caspase Substrate) Anti HLA-G Ab 87G +

60 40

49 51

45 55

CD56

65 35

CLL1 8

% CD56+CD107a+cells

CLL1 9 100 P = 0.006

75 50

∆=1.56

25 0 B-CLL

67 33

B-CLL+87G

38 62 CLL7

CD107 a (LAMP-1 )

Figure 3 Effect of blocking HLA-G1 on the sensitivity of chronic lymphocytic leukemia (CLL) to natural killer (NK) cells. Cytotoxicity assay was performed as described in the absence or presence of HLA-G-blocking Ab 87G using (a) CytoxiLux assay measuring lysis of target cell via increase in green fluorescence due to caspase activation in target cells (previously fluorescently labeled in red) and (b) CD107 mobilization assay where the lytic activity of NK cells was measured according to cell surface CD107a molecules expression on NK cells (CD56 positive). Results from CD107 mobilization assay are presented for three independent experiments and as mean values±s.d. of the three independent experiments (c). Numbers in the upper right quadrant represent the percentage of apoptotic target cells in A and the percentage of activated NK cells in B.

Cytotoxic activity of NK cells is regulated by a delicate balance between signals through activating and inhibitory receptors expressed by NK cells. While most NK inhibitory receptors are clonally distributed and, therefore, present on only a small proportion of circulating NK cells, LIR1/ILT2/CD85j and KIR2DL4, the two receptors for HLA-G are not clonally distributed. KIR2DL4 is expressed on all NK cells of an individual, and ILT2 on a subset of NK cells variable in size among different donors and also on most myelomonocytic cells, B cells, dendritic cells and subset of T cells. HLA-G expression may therefore represent a more universal mechanism of tumor escape. Because of the lack of inhibitory receptors corresponding to MHC class-I antigens present on target cell, and the absence of inhibitory signal, NK cells from Leukemia

allogeneic donors are thought to be more effective in killing tumor cells. This has encouraged a body of preclinical and clinical work on the basis of using allogeneic NK cells and initiated great interest and effort in exploiting NK-cell alloreactivity in bone marrow/stem cell transplantation to eliminate tumor cells in particular in hematological malignancies. However, tumor-cell susceptibility to NK cells can vary with the type of disease, as there can be heterogeneity within a specific malignancy. Chronic myelogenous leukemia (CML) cells are known to be very sensitive to NK cells and the most encouraging data from alloreactive NK trials were obtained with tumor cells of myeloid origin (AML). This knowledge is thus important to predict both the applicability and efficacy of any such therapy.


1005 Alternatively, NK activity itself may be altered in CML, AML and myelodysplastic syndromes (MDS) patients33–35 as well as CLL and MM (G Maki, unpublished data). CLL and MM tumorcell susceptibility to NK-mediated killing was, therefore, investigated using both NK cells from unrelated healthy donors, as well as the highly cytotoxic NK cell line, NK-92. Results presented here utilize NK-92 as it has a very high cytotoxic activity and represents an excellent model system to assess NK susceptibility and possible means of tumor escape. We found that while myeloma cell lines are sensitive to NK-mediated killing, primary tumor cells from MM as well as CLL are overwhelmingly resistant to NK-mediated killing. To gain insight into molecular mechanisms of this resistance, we sought to assess the expression of HLA-G and its isoforms in CLL and MM and to determine its relevance to susceptibility to NK-mediated killing. We examined both HLA-G transcript and cell-surface protein expression to elucidate its role in protecting tumor cells from NK cells. In contrast to KIR2DL4 surface expression, which is almost undetectable on NK-92,17,28 these cells express high levels of ILT2. For NK-92 interaction with HLA-G, ILT2 is the relevant receptor. Using the melanoma cell line M8 and its HLA-G1expressing transfectant M8-HLA-G1, the effectiveness of ILT2/ HLA-G pathway in protecting target cells from NK-92-mediated killing was confirmed. The importance of ILT2 as an inhibitory receptor for HLA-G has also been demonstrated via another NK cell line, NKL, despite high levels of surface expression of both ILT2 and KIR2DL4.38 Herein we report the relative resistance of primary tumor cells from CLL and myeloma patients to NK-mediated killing and the presence of HLA-G-encoding mRNAs in these cells. With the exception of 2 out of 19 CLL, HLA-G transcripts were found in tumor cells from both CLL and MM patients, while being absent from their normal counterparts. Surface expression of HLA-G1 could be detected on variable proportion of CLL cells and was low or undetectable on primary myeloma cells and myeloma cell lines. In contrast to myeloma cell lines that were very sensitive to NK-mediated killing, patient-derived myeloma cells were mostly resistant to NK cells. Of the 19 CLL samples examined, only 5 were moderately sensitive (X15% kill at 20:1) while 14 were resistant (killing less than 10% at 20:1). Notably, blocking of HLA-G1 on CLL cells increased their susceptibility to NK-mediated killing regardless of their initial level of susceptibility to NK-mediated killing. With regard to CLL and myeloma samples with low or no surface expression of HLA-G1, we cannot exclude the contribution of other isoforms of HLA-G, specifically sHLA-G in vivo. Indeed both membrane-bound (mHLA-G) and soluble HLA-G (sHLA-G, either shed from the membrane or produced by soluble isoform-encoding mRNAs) have been shown to have immunosuppressive effect by either triggering of inhibitory signals on NK and T cells, or by induction of apoptosis on these cells.18,21 This immunosuppressive role can also be mediated through a recently defined process of ‘acquired’ HLA-G through a process of membrane transfer called ‘trogocytosis’. The expression of HLA-G (mHLAG or sHLA-G) on CLL or myeloma cells, bone marrow or blood can, therefore, be relevant to tumor escape not only by directly inhibiting NK cytotoxic function and conferring resistance to NK-mediated killing but also as a more general immunosuppressive factor affecting other immune effector cells such as T lymphocytes and dendritic cells. Moreover, HLA-G1 and –G5 molecules can exist as either b2m-free or-associated chain as monomers, disulfide-linked homodimers or oligomers44 with differences in their binding affinity to receptors. Recent data support preferential binding of ILT2 for HLA-G dimers.45

The lack of detection of HLA-G on the surface of myeloma cells that we tested, may therefore, also be due to these conformational features of HLA-G molecules which may not be recognized by our antibodies (MEM-G/9 and 87G). This work investigated only tumor cell interaction with NK cells, and sHLA-G in the blood or BM was not investigated. Our results from HLA-G-blocking experiments demonstrate the relevance of HLA-G in the tumor escape from NK cells in these malignancies and imply that other mechanisms of NK resistance may also be in place. Although not directly investigated for MM in this study because of very low or undetectable surface expression, the relative resistance to NK-mediated killing and the presence of HLA-G mRNA in all myeloma samples examined strongly suggest a role for HLA-G in this resistance. These data suggest that although it may not on its own to determine resistance to NK cells, HLA-G may contribute to protecting CLL and potentially MM cells from NK-mediated killing. Our findings on the expression of HLA-G in CLL cells confirm previous study by Nu¨ckel et al. reporting surface expression of HLA-G in a variable proportion of the cells on all CLL samples examined. While this study established a correlation between HLA-G expression level and the degree of immunosuppression and survival with a cutoff level of 23% HLA-G positivity, our data demonstrate for the first time the role of HLA-G in tumor NK resistance in hematological malignancies. This is the first report implicating HLA-G in protecting CLL and potentially MM cells from NK cells in primary patient derived tumor cells rather than in cell lines. Data with cell lines may not translate accurately to primary tumor cells particularly for diseases such as CLL and MM where the in vivo microenvironment plays important role in the pathophysilogy of the disease. These data justify further investigation into the role of HLA-G expression on these tumor cells of B origin in the pathophysiology of the disease and potential benefit from modulating HLA-G expression in improving the therapeutic outcome in the future. This knowledge could help predict efficacy and response to immunotherapy and allow designing better NK-based immunotherapy that could broaden the therapeutic spectrum in patients with B-CLL and multiple myeloma.

Acknowledgements This work was supported in part by grants from ‘The Leukemia & Lymphoma Society’ and ‘Grant CLL-63119, Section of Hematology, Rush University Medical Center’.

References 1 Hideshima T, Bergsagel PL, Kuehl WM, Anderson KC. Advances in biology of multiple myeloma: clinical applications. Blood 2004; 104: 607–618. 2 Hamblin TJ. Chronic lymphocytic leukaemia: clinical translations of biological features. Curr Top Microbiol Immunol 2005; 294: 165–185. 3 Hideshima T, Richardson P, Anderson KC. Novel therapeutic approaches for multiple myeloma. Immunol Rev 2003; 194: 164–176. 4 Keating MJ, Chiorazzi N, Messmer B, Damle RN, Allen SL, Rai KR et al. Biology and treatment of chronic lymphocytic leukemia. Hematology Am Soc Hematol Educ Program 2003, 153–175. 5 Pant S, Copelan EA. Hematopoietic stem cell transplantation in multiple myeloma. Biol Blood Marrow Transplant 2007; 13: 877–885. Leukemia


1006 6 Kharfan-Dabaja MA, Anasetti C, Santos ES. Hematopoietic cell transplantation for chronic lymphocytic leukemia: an evolving concept. Biol Blood Marrow Transplant 2007; 13: 373–385. 7 Trinchieri G. Biology of natural killer cells. Adv Immunol 1989; 47: 187–376. 8 Lanier LL. NK cell recognition. Annu Rev Immunol 2005; 23: 225–274. 9 Lopez-Botet M, Llano M, Navarro F, Bellon T. NK cell recognition of non-classical HLA class I molecules. Semin Immunol 2000; 12: 109–119. 10 Colonna M. Specificity and function of immunoglobulin superfamily NK cell inhibitory and stimulatory receptors. Immunol Rev 1997; 155: 127–133. 11 Karre K. NK cells, MHC class I molecules and the missing self. Scand J Immunol 2002; 55: 221–228. 12 Ruggeri L, Capanni M, Casucci M, Volpi I, Tosti A, Perruccio K et al. Role of natural killer cell alloreactivity in HLA-mismatched hematopoietic stem cell transplantation. Blood 1999; 94: 333–339. 13 Ruggeri L, Capanni M, Urbani E, Perruccio K, Shlomchik WD, Tosti A et al. Effectiveness of donor natural killer cell alloreactivity in mismatched hematopoietic transplants. Science 2002; 295: 2097–2100. 14 Parham P, McQueen KL. Alloreactive killer cells: hindrance and help for haematopoietic transplants. Nat Rev Immunol 2003; 3: 108–122. 15 Cosman D, Fanger N, Borges L, Kubin M, Chin W, Peterson L et al. A novel immunoglobulin superfamily receptor for cellular and viral MHC class I molecules. Immunity 1997; 7: 273–282. 16 Saverino D, Fabbi M, Ghiotto F, Merlo A, Bruno S, Zarcone D et al. The CD85/LIR-1/ILT-2 inhibitory receptor is expressed by all human T lymphocytes and down regulates their functions. J Immunol 2000; 165: 3742–3755. 17 Rajagopalan S, Long EO. A human histocompatibility leukocyte antigen (HLA)-G-specific receptor expressed on all natural killer cells. J Exp Med 1999; 189: 1093–1100. 18 Carosella ED, Moreau P, Le Maoult J, Le Discorde M, Dausset J, Rouas-Freiss N. HLA-G molecules: from maternal–fetal tolerance to tissue acceptance. Adv Immunol 2003; 81: 199–252. 19 Kovats S, Main EK, Librach C, Stubblebine M, Fisher SJ, DeMars R. A class I antigen, HLA-G, expressed in human trophoblasts. Science 1990; 248: 220–223. 20 King A, Hiby SE, Verma S, Burrows T, Gardner L, Loke YW. Uterine NK cells and trophoblast HLA class I molecules. Am J Reprod Immunol 1997; 37: 459–462. 21 Rouas-Freiss N, Moreau P, Ferrone S, Carosella ED. HLA-G proteins in cancer: do they provide tumor cells with an escape mechanism? Cancer Res 2005; 65: 10139–10144. 22 Paul P, Rouas-Freiss N, Khalil-Daher I, Moreau P, Riteau B, Le Gal FA et al. HLA-G expression in melanoma: a way for tumor cells to escape from immunosurveillance. Proc Natl Acad Sci USA 1998; 95: 4510–4515. 23 Wiendl H, Mitsdoerffer M, Hofmeister V, Wischhusen J, Bornemann A, Meyermann R et al. A functional role of HLA-G expression in human gliomas: an alternative strategy of immune escape. J Immunol 2002; 168: 4772–4780. 24 Braud VM, Allan DS, O’Callaghan CA, Soderstrom K, D’Andrea A, Ogg GS et al. HLA-E binds to natural killer cell receptors CD94/ NKG2A, B and C. Nature 1998; 391: 795–799. 25 LeMaoult J, Caumartin J, Daouya M, Favier B, Le Rond S, Gonzalez A et al. Immune regulation by pretenders: cell-to-cell transfers of HLA-G make effector T cells act as regulatory cells. Blood 2007; 109: 2040–2048. 26 Caumartin J, Favier B, Daouya M, Guillard C, Moreau P, Carosella ED et al. Trogocytosis-based generation of suppressive NK cells. EMBO J 2007; 26: 1423–1433. 27 Gong JH, Maki G, Klingemann HG. Characterization of a human cell line (NK-92) with phenotypical and functional characteristics of activated natural killer cells. Leukemia 1994; 8: 652–658.

28 Maki G, Klingemann HG, Martinson JA, Tam YK. Factors regulating the cytotoxic activity of the human natural killer cell line, NK-92. J Hematother Stem Cell Res 2001; 10: 369–383. 29 Lichtenfels R, Biddison WE, Schulz H, Vogt AB, Martin R. CARELASS (calcein-release-assay), an improved fluorescence-based test system to measure cytotoxic T lymphocyte activity. J Immunol Methods 1994; 172: 227–239. 30 Liu L, Chahroudi A, Silvestri G, Wernett ME, Kaiser WJ, Safrit JT et al. Visualization and quantification of T cell-mediated cytotoxicity using cell-permeable fluorogenic caspase substrates. Nat Med 2002; 8: 185–189. 31 Betts MR, Brenchley JM, Price DA, De Rosa SC, Douek DC, Roederer M et al. Sensitive and viable identification of antigenspecific CD8+ T cells by a flow cytometric assay for degranulation. J Immunol Methods 2003; 281: 65–78. 32 Bryceson YT, March ME, Barber DF, Ljunggren HG, Long EO. Cytolytic granule polarization and degranulation controlled by different receptors in resting NK cells. J Exp Med 2005; 202: 1001–1012. 33 Pierson BA, Miller JS. CD56+bright and CD56+dim natural killer cells in patients with chronic myelogenous leukemia progressively decrease in number, respond less to stimuli that recruit clonogenic natural killer cells, and exhibit decreased proliferation on a per cell basis. Blood 1996; 88: 2279–2287. 34 Costello RT, Sivori S, Marcenaro E, Lafage-Pochitaloff M, Mozziconacci MJ, Reviron D et al. Defective expression and function of natural killer cell-triggering receptors in patients with acute myeloid leukemia. Blood 2002; 99: 3661–3667. 35 Kiladjian JJ, Bourgeois E, Lobe I, Braun T, Visentin G, Bourhis JH et al. Cytolytic function and survival of natural killer cells are severely altered in myelodysplastic syndromes. Leukemia 2006; 20: 463–470. 36 Almeida J, Orfao A, Ocqueteau M, Mateo G, Corral M, Caballero MD et al. High-sensitive immunophenotyping and DNA ploidy studies for the investigation of minimal residual disease in multiple myeloma. Br J Haematol 1999; 107: 121–131. 37 Le Gal FA, Riteau B, Sedlik C, Khalil-Daher I, Menier C, Dausset J et al. HLA-G-mediated inhibition of antigen-specific cytotoxic T lymphocytes. Int Immunol 1999; 11: 1351–1356. 38 Menier C, Riteau B, Carosella ED, Rouas-Freiss N. MICA triggering signal for NK cell tumor lysis is counteracted by HLA-G1-mediated inhibitory signal. Int J Cancer 2002; 100: 63–70. 39 Amiot L, Onno M, Drenou B, Monvoisin C, Fauchet R. HLA-G class I gene expression in normal and malignant hematopoietic cells. Hum Immunol 1998; 59: 524–528. 40 Sebti Y, Le Friec G, Pangault C, Gros F, Drenou B, Guilloux V et al. Soluble HLA-G molecules are increased in lymphoproliferative disorders. Hum Immunol 2003; 64: 1093–1101. 41 Polakova K, Krcova M, Kuba D, Russ G. Analysis of HLA-G expression in malignant hematopoetic cells from leukemia patients. Leuk Res 2003; 27: 643–648. 42 Nuckel H, Rebmann V, Durig J, Duhrsen U, Grosse-Wilde H. HLA-G expression is associated with an unfavorable outcome and immunodeficiency in reb. Blood 2005; 105: 1694–1698. 43 Leleu X, Le Friec G, Facon T, Amiot L, Fauchet R, Hennache B et al. Total soluble HLA class I and soluble HLA-G in multiple myeloma and monoclonal gammopathy of undetermined significance. Clin Cancer Res 2005; 11: 7297–7303. 44 Apps R, Gardner L, Sharkey AM, Holmes N, Moffett A. A homodimeric complex of HLA-G on normal trophoblast cells modulates antigen-presenting cells via LILRB1. Eur J Immunol 2007; 37: 1924–1937. 45 Gonen-Gross T, Achdout H, Arnon TI, Gazit R, Stern N, Horejsi V et al. The CD85J/Leukocyte inhibitory receptor-1 distinguishes between conformed and beta 2-microglobulin-free HLA-G molecules. J Immunol 2005; 175: 4866–4874.

Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

Leukemia