CD56dim NK cells from CLL patients - Nature

Leukemia (2011) 25, 101–109 & 2011 Macmillan Publishers Limited All rights reserved 0887-6924/11 www.nature.com/leu

ORIGINAL ARTICLE Analysis of CD16 þ CD56dim NK cells from CLL patients: evidence supporting a therapeutic strategy with optimized anti-CD20 monoclonal antibodies M Le Garff-Tavernier1,2,3, J Decocq1,3,4, C de Romeuf4, C Parizot5, CA Dutertre4,6,7,8, E Chapiro2, F Davi1,2,3, P Debre´1,3,5, JF Prost4, JL Teillaud6,7,8, H Merle-Beral2,6,7,9 and V Vieillard1,3,9 1

INSERM, UMR-S 945, Paris, France; 2AP-HP, Hoˆpital Pitie´-Salpeˆtrie`re, Service d’He´matologie Biologique, Paris, France; UPMC Universite´ Paris 06, UMR-S 945, Paris, France; 4Laboratoire Franc¸ais de Fractionnement et des Biotechnologies (LFB), Les Ulis, France; 5Laboratoire d’Immunologie Cellulaire et Tissulaire, AP-HP, Hoˆpital Pitie´-Salpeˆtrie`re, Paris, France; 6 INSERM, UMR-S 872, Paris, France; 7Centre de Recherche des Cordeliers, UPMC Universite´ Paris 06, UMR-S 872, Paris, France and 8Universite´ Paris Descartes, UMR-S 872, Paris, France 3

Although anti-CD20 monoclonal antibodies (mAbs) show promise for the treatment of chronic lymphocytic leukemia (CLL), the success of the anti-CD20 mAb rituximab in CLL treatment has been limited. Novel anti-CD20 mAbs with more potent cytotoxic activity have recently been engineered, but so far most have only been tested in vitro with natural killer (NK) cells from healthy donors. Because it is still unclear whether these optimized cytotoxic mAbs will improve NK-cell killing of tumor cells in CLL patients, we characterized the relevant phenotypic and functional features of NK cells from CLL patients in detail. Expression of inhibitory and activating NK-cell receptors and of Fc gamma receptor IIIA (FccRIIIA) is well preserved in CD16 þ CD56dim cytotoxic NK cells from these patients, independently of disease progression. These cells are fully functional following cytokine stimulation. In addition, the FccRIIIA-optimized LFB-R603 anti-CD20 mAb mediates 100 times greater antibody-dependent cell-mediated cytotoxicity by NK cells from CLL patients and healthy donors than rituximab. Enhanced degranulation against autologous B-CLL cells is observed at lower concentrations of LFB-R603 than rituximab, regardless of CLL prognostic factors. These findings strongly justify further clinical development of anti-CD20 mAbs optimized for FccR engagement in CLL patients. Leukemia (2011) 25, 101–109; doi:10.1038/leu.2010.240; published online 26 October 2010 Keywords: chronic lymphocytic leukemia (CLL); natural killer (NK) cells; anti-CD20; monoclonal antibodies

Introduction Chronic lymphocytic leukemia (CLL) is an accumulation of CD5 þ /CD19 þ malignant monoclonal B lymphocytes with a highly variable clinical course. The introduction of therapeutic monoclonal antibodies (mAbs) has brought a remarkable change in treatment. The anti-CD52 alemtuzumab and anti-CD20 rituximab mAbs have been widely used in progressive disease. In contrast to alemtuzumab, rituximab is generally well tolerated with infrequent serious adverse effects that are generally well controlled. Although the combination of fludarabine and cyclophosphamide with rituximab is currently considered the Correspondence: Dr M Le Garff-Tavernier, INSERM UMR-S 945, Baˆtiment CERVI, GH Pitie´-Salpeˆtrie`re, 83 bd de l’Hoˆpital, 75651 Paris Cedex 13, France. E-mail: [email protected] 9 These authors are senior co-authors. Received 19 February 2010; revised 12 August 2010; accepted 1 September 2010; published online 26 October 2010

‘gold standard’ for the treatment of CLL,1,2 it does not allow full-disease eradication and the disease can become resistant. The activity of rituximab is thought to be mediated through several mechanisms, including complement-dependent cytotoxicity, apoptosis, antibody-dependent cellular cytotoxicity (ADCC) and/or phagocytosis.3 Strategies pursued to improve the efficacy of mAb therapy have included modifying the antibody structure to increase its affinity to the Fc gamma receptor IIIA (FcgRIIIA, CD16), an intermediate-affinity activating FcgR strongly expressed on natural killer (NK) cells.4 Optimized mAbs, including AME-133, GA-101, veltuzumab (hA20) and ofatumumab (HuMax-CD20)5–7 are at different stages of development. We previously described a novel chimeric anti-CD20 mAb, EMAB-6, characterized by low fucose content in its Fc region. EMAB-6 shows improved FcgRIIIA binding, resulting in enhanced ADCC, mediated particularly by macrophages8 and NK cells.9 The NK cells constitute a unique component of the innate immune system, able to recognize various targets without specific sensitization. The NK cells are heterogeneous and differ in their proliferative potential, homing characteristics, functional capabilities and in responses to a wide range of cytokines. Around 10% of NK cells in the peripheral blood are CD16CD56bright. This immunoregulatory subset of NK cells produces a wide range of cytokines and chemokines, but its ability to kill target cells spontaneously is poor. In contrast, the rest (90%), corresponding to the CD16 þ CD56dim cytotoxic NK-cell subset, produces relatively lower cytokine levels, but possesses an abundance of cytolytic granules and can spontaneously lyse susceptible target cells.10 The complexity of the NK-cell compartment is attributable to the vast network of inhibitory or activating receptors that allow these cells to recognize target cells. The functional state of NK cells is, therefore, dictated by the balance between opposing signals.11 Moreover, in the presence of mAbs, strong expression of FcgRIIIA allows CD16 þ CD56dim NK cells to kill targets through ADCC, regardless of the presence of any inhibitory receptors. ADCC has been identified as a critical mechanism underlying the clinical efficacy of therapeutic antibodies, including rituximab.12,13 We focused our attention on NK cells in patients with CLL, because the size of this cell compartment is an independent factor predictive of the rate of disease progression in patients with newly diagnosed CLL.14 Several studies have reported functional alterations in NK cells in these patients, including defective cytotoxic activity, which are particularly pronounced in advanced disease and that could be because of the accumulation of leukemic cells rather than to any intrinsic

NK cells and optimized anti-CD20 in CLL M Le Garff-Tavernier et al

102

defect in the former.15,16 This hypothesis was suggested by the restoration of their functional activity after interferon (IFN)-a and/or interleukin (IL)-2 treatment.17,18 We conducted the first extensive characterization of the phenotypic and cytotoxic qualities of NK cells from CLL patients and compared them with those of healthy subjects. Our data reveal that NK cells from these patients have the major phenotypic features of competent NK cells and are able to function effectively. We also explored the preclinical activity of LFB-R603, a novel optimized anti-CD20 mAb, and show that at low doses this antibody promotes a significantly higher percentage of NK cell-mediated degranulation against autologous CLL cells than rituximab.

Patients and methods

CD335/NKp46 (BAB281), CD85j/ILT2 (HP-F1), 2B4 (C1.7) and CD69 (TP1.55.3) (Beckman-Coulter), CD16 (3G8) and LAIR1 (DX26) (BD Biosciences), NKG2C (134591) and NKp80/KLFR1 (239127) (R&D Systems, Lille, France) and CD337/NKp30 (AF29-4D12) (Miltenyi Biotec, Paris, France). Erythrocytes were lysed with a FACS lysing solution kit (BD Biosciences). After extensive washing, at least 5 103 CD3CD56dim NK cells were analyzed on a FACSCanto I with FACSDiva 5.03 software (BD Biosciences). The level of CD16 expression on CD3CD56dim NK cells was quantified using QuantiBRITE standard beads (BD Biosciences), as recommended by the manufacturer. For intracellular IFN-g staining, peripheral blood mononuclear cells (PBMC) were incubated overnight in the presence of IL-12 (10 ng/ml) and IL-18 (100 ng/ml) (R&D Systems). Cells were fixed and permeabilized with a cytofix/ cytoperm kit (BD Biosciences), and stained with IFN-g mAb (B27; BD Biosciences), as described.21

Patients and control donors Fresh blood samples were collected from 57 CLL patients followed at Pitie´-Salpeˆtrie`re Hospital (Paris, France). All patients met the morphological and phenotypic criteria for typical CLL (Matutes score X4).19 They had not been treated previously or had not received any treatment during the 6 months before the study. None of these patients had received rituximab or alemtuzumab. Binet stage A patients were considered to have an unfavorable prognosis if they met any two of the following four criteria: (1) Presence of unmutated Immunoglobulin heavy variable chain sequences with a VH gene sequence p2% of sequence alterations compared with the published germline sequence; (2) High expression of surface CD38 (420%); (3) High expression of cytoplasmic ZAP-70 (420%); and (4) Clonal cytogenetic abnormalities, including deletion of chromosomes 17p or 11q or trisomy of chromosome 12, revealed by fluorescent-labeled DNA probes used in interphase fluorescence in situ hybridization (probes 17p13.1, ATM 11q23 and CEP12, respectively).20 Patients with del (13q) (probe 13q14), have an excellent prognosis if they do not have additional defects. Patients with 17p were considered to have an unfavorable prognosis even if all their other criteria were favorable. All other Binet stage A patients were considered to have a favorable prognosis. The control group comprised sex- and age-matched outpatient volunteers from the rheumatology (n ¼ 29), who had consulted for noninflammatory arthritic pathology such as osteoarthritis, or gerontology (n ¼ 38) departments at Pitie´Salpeˆtrie`re or Charles-Foix Hospital, respectively. Our study applied restrictive health selection criteria, including no infectious or malignant diseases 6 months before the study and without acute illness at the time of sampling. None of the control patients had an autoimmune disease. All case and control patients provided a written, informed consent before participation. The local ethics committee approved the study.

Flow cytometry and absolute B cell, T cell, and NK cell counts Freshly collected blood cells underwent cell-surface staining for four-color cytometric analysis. Isotype-matched immunoglobulin served as the negative control. NK cells were analyzed after staining with an appropriate antibody cocktail: CD3-FITC (UCHT1; Beckman-Coulter, Marseille, France), CD19-PerCP Cy5.5 (SJ25C1; BD Biosciences, Le Pont de Claix, France), CD56-APC (N901; Beckman-Coulter), and the following PEconjugated antibodies: KIR2DL1/KIR2DS1 (EB6B), KIR2DL2/ KIR2DL3/KIR2DS2 (GL183), KIR3DL1/KIR3DS1 (Z27), CD159a/ NKG2A (Z199), NKG2D (ON72), CD336/NKp44 (Z231), Leukemia

Direct cytolytic activity of NK cells NK cell-enriched PBMC (8 CLL patients and 5 controls) were tested for direct cytotoxicity. NK cell-enriched PBMC fractions were obtained after negative depletion (anti-CD19-coated beads) followed by positive selection (anti-CD56-coated beads) (Miltenyi Biotec, Bergish Gladbach, Germany). Similar proportions of NK cells were detected in NK cell-enriched PBMC fractions from CLL patients (38±11%) and controls (42±10%). The B cells accounted for 30±20% of the residual cells for CLL patients and 1±2% for controls, whereas the proportions for T cells were 31±17% and 55±11%, respectively. Direct cytotoxicity was assessed after IL-2 stimulation (proleukin-2, 1000 IU/ml) for 72 h, with a standard 4-h 51 Cr-release assay against the human leukocyte antigen (HLA) class I-deficient human K562 cell line.21

Anti-CD20 LFB-R603 antibody LFB-R603 is a chimeric mAb directed against CD20 with a glycosylation profile that allows increased binding to CD16. This mAb has been granted an orphan-drug status in Europe and in the USA for CLL treatment (http://www.lfb.fr/en/home.html). A previous generation of this antibody, named EMAB-6, sharing the same specificity and a similar glycosylation pattern, was described in de Romeuf et al.9

Purification of NK cells and ADCC assay

Purified NK cells (498% CD3CD56 þ cells) from 7 CLL patients and 6 controls were obtained from peripheral blood after centrifugation on a Ficoll/Hypaque gradient, followed by negative selection with anti-CD19-coated beads (Miltenyi Biotec) and cytofluorometric selection of CD3CD56 þ NK cells on a FACSVantage (BD Biosciences). ADCC assays were performed with purified NK cells and 51Cr-labeled Raji cells at an effector-target (E:T) ratio of 10:1. Experiments were performed in the presence of LFB-R603 (LFB) or rituximab (Roche Ltd, Basel, Switzerland) at concentrations ranging from 0.1 to 10 000 ng/ml.

Degranulation of NK cells The degranulation marker CD107a was detected on freshly isolated PBMC from both CLL patients and controls by co-staining for NK markers (CD3CD56 þ NK cells). Cells were incubated with anti-CD107a mAb (H4A3; BD Biosciences), with 10 or 1000 ng/ml of either anti-CD20 rituximab (Roche) or LFB-R603 (LFB, Les Ulis, France), as described.21


Statistical analysis

Phenotypic analysis of NK receptors in CLL patients

Statistical analyses were performed with Prism 4 software (GraphPad Software, San Diego, CA, USA). Intergroup comparisons were assessed with the Mann–Whitney U-test or Wilcoxon rank-sum test. P-values less than 0.05 were considered statistically significant; *Po0.05, **Po0.01, ***Po0.001.

Results

CLL patients Table 1 and Supplementary Table 1 show the main clinical and biological prognostic characteristics of CLL patients included in this study. Because the potential benefit of early treatment for unfavorable prognosis Binet stage A patients is a key therapeutic question, patients were divided into two groups according to their biological prognostic factors. Binet stage A CLL patients with an unfavorable prognosis were classified with Binet stage B/C patients (group 2, n ¼ 23) and compared with stage A patients with a favorable prognosis (group 1, n ¼ 34). There was no significant difference in the sex ratio or median age between groups 1 and 2.

Distribution of lymphocyte subsets in CLL patients

Flow cytometry was used to measure CD19 þ B, CD3 þ T and CD3CD56 þ NK subsets in CLL patients and controls. The B-cell counts were significantly higher in CLL patients (group 1: median 10 592/mm3 (range: 317–103 402); group 2: 28 084/ mm3 (2918–198 407)) than in controls (median 111/mm3 (22–943)) (groups 1 and 2 vs controls: Po0.0001). The distribution of lymphocyte subsets showed that CLL patients had significantly lower T-cell and NK-cell frequencies than controls (groups 1 and 2 vs controls: Po0.0001 for T cells and Po0.0001 for NK cells) (Figures 1a and c, respectively). Interestingly, CLL patients had significantly higher absolute T-cell and NK-cell counts than controls (groups 1 and 2 vs controls: Po0.0001 for T cells, group 1 vs controls: P ¼ 0.0066 and group 2 vs controls: P ¼ 0.0003 for NK cells) (Figures 1b and d, respectively). However, despite higher lymphocyte counts in group 2 patients, T-cell and NK-cell counts did not differ significantly between the two groups of CLL patients. Similarly, the distribution of the CD56dim NK-cell subset did not differ between the CLL groups and controls (Figure 1e). Finally, there was no difference in the frequency of CD16 þ /CD56dim NK cells (Figure 1f) or surface density of CD16, as detected by the Quantibrite assay (data not shown), between these groups. Similar results were obtained using the classical stage A vs stages B/C of the Binet classification (data not shown). Table 1

103

The phenotypic properties of the CD3CD56dim cytotoxic NKcell subset were investigated further by gating out CD3CD56bright NK cells. CD56dim NK cells from CLL patients, irrespective of their prognostic status, were indistinguishable from those of controls in terms of cell-surface expression of the major NK receptors studied, including c-lectin receptors (NKG2A, NKG2C and NKG2D), natural cytotoxicity receptors (NKp30, NKp44 and NKp46) and other NK-cell markers such as NKp80, ILT-2, LAIR-1 and 2B4 (Figure 2a, upper panels; data not shown). Similar phenotypic results were observed according to the Binet stages (A vs B/C) (data not shown). This phenotype was then studied longitudinally, over a 1-year period, comparing the initial samples with samples taken 2, 4, 6 and/or 12 months later. A remarkable phenotypic stability was observed for each NK receptor studied. Figure 2a (lower panels) shows representative CLL patients from each of the groups (patients 1 and 2, group 1; patient 3, group 2; patient 4, group 1 (at first evaluation) and group 2 (at 12 months)). These results were confirmed with three other patients (data not shown). In addition, patient 4’s NK receptor repertoire remained stable during disease progression between stages A (group 1) and C (group 2) after 1 year. The only difference observed in the KIR was a weak, but significant decrease in expression of KIR2DL1/DS1, which recognizes group-2 HLA-Cw alleles, in CLL patients with an unfavorable prognosis compared with controls (P ¼ 0.0370). Binet B/C patients presented a nonsignificant decrease compared with controls (data not shown). No differences between groups were observed for the other KIRs tested, including KIR2DL2-3/DS2, which binds group-1 HLA Cw ligands, and KIR3DL1/DS1, which recognizes HLA-Bw4 (Figure 2b). The expression pattern of the early activation marker CD69 on CD3CD56 þ NK cells was also investigated. As shown in Figure 2c, CD69 expression in CD3CD56dim NK cells was slightly, but significantly lower in CLL patients than in controls (group 1 vs controls: P ¼ 0.0228, and group 2 vs controls: P ¼ 0.0044), even if a certain degree of overlap of CD69 expression in CD3CD56dim NK cells was observed between both groups of CLL patients and controls. Similarly, CD69 expression decreased significantly in CD3CD56bright immunoregulatory NK cells from group 1 patients compared with controls (P ¼ 0.0037). Furthermore, HLA-DR expression, a late cell-activation marker, remained close to background levels in all control and CLL subjects, in all NK-cell subsets (data not shown). Similar results were obtained according to Binet stages (data not shown).

Functional analysis of NK cells in CLL patients The functional capacities of the NK-cell compartment of CLL patients were investigated further. As shown in Figure 3a, intracellular IFN-g in CD3CD56dim NK cells from CLL patients

Main clinical and biological characteristics of CLL patients

Patients (n)

Age median (range) Sex

F

Binet stage

M

Group 1 (34) 11 23 Group 2 (23) 8 15

Prognostic factors

Lymphocytosis (109/l) median (range)

A 62 (34–83) 66 (40–79)

14.9 (5.1–106.6) 34 33.8 (7.3–212.2) 9

B/C 0 14

IgHV mutational status Ma

UMa

CD38

ZAP-70

+

+

Cytogenetic abnormalities 17p 11q 13q 12 +

31/31 0/31 27/34 7/34 27/31 4/31 7/21 14/21 15/23 8/23 11/22 11/22

0/33 4/23

0/33 6/23

18/33 12/23

1/33 2/23

Abbreviations: CLL, chronic lymphocytic leukemia; F, female; IgHV, immunoglobulin variable heavy chain; M, male. a M, mutated; UM, unmutated. Leukemia


104 ***

***

***

100

10 CD3+/mm3

75 % CD3+

***

4

50 25

103

102 1

0 Ctl

group 1

group 1

Ctl

group 2

*** 10

*

30 20 10 0 Ctl

group 1

103 102 101 1 Ctl

group 2

group 1

group 2

100

% CD16+/CD3-CD56dim

100

% CD56dim/CD3-CD56+

**

4

CD3-CD56+/mm3

% CD3-CD56+

***

***

40

group 2

75 50 25 0 Ctl

group 1

group 2

75 50 25 0 Ctl

group 1

group 2

Figure 1 Distribution of lymphocyte subsets in CLL patients. FACS analysis of percentages and absolute cell counts of: (a and b) CD3 þ T cells; and (c and d) CD3CD56 þ NK cells, gated on lymphocytes from peripheral blood. (e) Expression of CD56dim gated on CD3CD56 þ NK cells. (f) Expression of CD16 þ gated on CD3CD56dim NK cells. Cells were collected from controls (Ctl, J) and CLL patients in group 1 (}) and group 2 (&). Horizontal bars represent the median values. P-values refer to the comparison between controls and CLL patients, assessed with the Mann–Whitney U-test. *Po0.05, **Po0.01, ***Po0.001.

(groups 1 and 2) treated with IL-12 and IL-18 was expressed at a similar level to that in NK cells from controls. Similar results were observed in the CD3CD56bright NK-cell subset even if IFN-g expression increased slightly in CLL patients (group 1 and group 2) compared with controls (P ¼ 0.0359 and P ¼ 0.0176, respectively) (Figure 3b). The direct cytotoxicity of NK cells was then evaluated after IL-2 activation by their capacity to lyse HLA class-I negative K562 sensitive-target cells and Figure 3c shows that the cytotoxicity of NK cell-enriched PBMC from CLL patients was similar to that in controls.

ADCC potency of NK cells from CLL patients is related to the concentration of rituximab and LFB-R603 NK-cell cytolysis can also be induced indirectly through Fc receptors, by antibodies mediating ADCC. ADCC assays with purified NK cells against Raji target cells, in the presence of serial concentrations of rituximab or LFB-R603, produced similar ADCC curves in controls and CLL patients with both anti-CD20 mAbs, including an exponential phase followed by a plateau (Figure 4a). The effect in CLL patients was similar irrespective of their prognostic status. However, the potency was dose-dependent for both rituximab and LFB-R603, and the Leukemia

maximum ADCC was detected at 1 ng/ml of LFB-R603, but at 100 ng/ml of rituximab. This indicates that the potency of LFB-R603 is superior to that of rituximab (Figure 4a). At 1 ng/ml, the median ADCC value with LFB-R603 was 62.5% for controls and 65.1% for CLL patients, whereas at a similar dose of rituximab, the ADCC was significantly lower (P ¼ 0.03) at 7.3 and 8.5%, respectively (Figure 4b, upper panel). In contrast, the ADCC with both anti-CD20 mAbs was similar at 100 ng/ml in CLL patients and controls (Figure 4b, lower panel). As shown in Figure 4c, in CLL patients, EC50 values were 10.5 ng/ml for rituximab and 0.13 ng/ml for LFB-R603, which indicates that the ADCC activity of the latter was 81 times higher than that of rituximab. In the control subjects, EC50 values were 16.2 ng/ml for rituximab and 0.16 ng/ml for LFB-R603, with LFB-R603 ADCC activity 101 times higher than that of rituximab. Taken together, these results confirm the strong ADCC capacity of NK cells from CLL patients and suggest that, at least ex vivo, LFB-R603 is more potent than rituximab at low concentrations. The degranulation capacities of NK cells from CLL patients in the presence of their autologous B-CLL cells provide additional support for these data because ADCC is influenced not only by binding of NK cells to their target but also and mainly by their ability to degranulate. This ex vivo model evaluated the


105

50 25 Ctl

100

75

% ILT-2

% NKp30

75

0

50 25 0

group 1 group 2

Ctl

50 25 0

100

100

75

75

75

50 25 #2

#3

50 25 0

#4

#1

CLL patients

% KIR2DL2-3/DS2

60 40 20 0

0

#4

#1

40 20 0 group1 group 2

**

50 25 Ctl

group1 group 2

#2

#3

#4

CLL patients

60

Ctl

*

75

0

25

80

group1 group 2

% CD69+/CD3-CD56dim

Ctl

#3

group1 group 2 0 2 4 6 12

CLL patients

*

80

#2

Ctl

50

% KIR3DL1/DS1

#1

% ILT-2

100

0

% KIR2DL1/DS1

75

group 1 group 2

% CD69+/CD3-CD56bright

% NKG2A

100

% NKp30

% NKG2A

100

80 60 40 20 0 Ctl

group1 group 2

**

50

25

0

Ctl

group1 group 2

Figure 2 Pattern of receptor expression on NK cells from CLL patients. (a) Expression of NKG2A, NKp30 and ILT-2. Upper panels show phenotypic expression of each marker gated on CD3CD56dim NK cells. Lower panels show a longitudinal phenotypic study of four representative CLL patients (patients 1, 2, and 4 in group 1; patient 3 in group 2), every 2 months during a 12-month period after the first phenotypic evaluation. (b) Expression of killer cell Ig-like receptors, including KIR2DL1/DS1, KIR2DL2-3/DS2, and KIR3DL1/DS1. (c) Expression of the early cell-activation marker CD69 on CD3CD56dim NK cells and CD3CD56bright NK cells. Cells were collected from controls (Ctl, J) and CLL patients in group 1 (}) and group 2 (&). Horizontal bars represent the median value. P-values refer to the comparison between controls and CLL patients, assessed with the Mann-Whitney U-test. *Po0.05, **Po0.01.

proportion of CD107a þ NK cells in whole PBMC from CLL patients and controls. As shown in Figure 5a, analysis of degranulation in the presence of autologous target cells confirmed the greater potency of LFB-R603 compared with rituximab: although NK-cell degranulation was not observed in samples from any of the 21 CLL patients at low concentrations of rituximab (10 ng/ml), the median values of degranulation with the same concentration of LFB-R603 were 16.3% in group 1 and 15.2% in group 2 (Figure 5a). Similar results were obtained using the Binet stages (data not shown). At a higher concentration (1000 ng/ml), NK-cell degranulation remained significantly higher with LFB-R603 than with rituximab (group 1, P ¼ 0.0005 and group 2, P ¼ 0.0078) (Figure 5b). According to Binet stages, this difference remained significant in the A patient group but not in the B/C patient group, probably because of the low number of patients in this latter group (data not shown). Finally, we could not exclude that CLL cells inhibited NK-cell function. To test this hypothesis, we added Raji cells into the CD107 degranulation assay performed with PBMC from CLL patients or healthy controls. As shown in the Supplementary Figure 1, in the presence of Raji cells, the CD107 level in NK cells was similar in CLL patients and healthy controls, suggesting an absence of effect of CLL cells on NK degranulation. Together, these data demonstrate that in the presence of autologous B-CLL target

cells, the proportion of degranulating NK cells was considerably higher with LFB-R603 than with rituximab.

Discussion This report describes the first extensive phenotypic and functional study of CD16 þ CD56dim NK cells in CLL patients and age-matched controls. CLL patients were found to have a significantly higher number of NK cells, as previously described,22 but expression of both inhibitory and activating NK-cell receptors was well preserved. In addition, this NK-cell repertoire remained remarkably stable despite disease progression. Intriguingly, only a slight decrease in KIR2DL1/DS1 expression was noted in patients with an unfavorable prognosis, whereas the frequency of HLA-Cw*06, a KIR2DL1 ligand, was significantly higher in CLL cohorts than in controls.23,24 The phenotypic features of these CD16 þ CD56dim NK cells cannot explain the lack of antitumor activity or disease control. This may suggest a failure to express some NK ligands on B-CLL cells, such as MICA and ULBPs, as previously reported.25 In contrast, the major phenotypic alterations in NK cells observed in acute leukemia, including the downregulation of natural cytotoxicity receptors,26,27 suggest that chronic and acute leukemia use Leukemia


106 different strategies to escape from NK-cell immunity. Poor expression of CD69 was also observed on CD16 þ CD56dim NK cells from CLL patients; this deficit could explain the * % IFNγ+ cells/ CD3-CD56bright

50 25 Ctl

group1 group 2

75 50 25 0

Ctl

group1 group 2 10 ng/ml

75

***

75

1000 ng/ml **

25 /1

5/ 1 2. 5/ 1 1. 25 /1

10

20

/1

0

**

+

50 25

15.2

16.3 0.0

0

1.4

group 1

50

42.6 35.2

25 12.4

9.5

0

RTX LFB-R603 RTX LFB-R603

E/T ratio

Figure 3 IFN-g expression and direct cytotoxic activity of NK cells from CLL patients. Intracellular expression of IFN-g in: (a) CD3CD56dim and (b) CD3CD56bright NK cells from controls (J) and CLL patients in group 1 (}) and group 2 (&), after overnight incubation with IL-12 and IL-18. Horizontal bars represent the median values. (c) Cytotoxicity of enriched NK cells from controls (J) and CLL patients (~) measured by 51Cr release-assays against K562 target cells. Mean percentage values±s.d. (vertical bars) are shown. P-values refer to the comparison between controls and CLL patients, assessed with the Mann–Whitney U-test. *Po0.05.

***

75

+

50

% CD107a

% NK lysis

100

% CD107a

% IFN+ cells/ CD3-CD56dim

75

0

*

100

100

previously reported reduction in NK cytotoxic activity in the absence of exogenous stimulation.15,28 This relative NK-cell anergy could be due to the ability of CLL cells and their microenvironment to produce cytokines, such as IL-10, which suppress the proliferation of antigen-specific Th1 cells and downregulate co-stimulatory molecules on antigen-presenting cells.29–31 Importantly, NK-cell function was perfectly recovered following activation. This included the ability to produce large amounts of IFN-g after IL-12 and IL-18 activation, and effective cytolysis against major histocompatibility complex

RTX LFB-R603 RTX LFB-R603

group 2

group 1

group 2

Figure 5 Degranulation response determined by CD107a expression on CD3CD56dim NK cells from CLL patients. PBMC from CLL patients in group 1 (n ¼ 12) and group 2 (n ¼ 9) were incubated with: (a) 10, and (b) 1000 ng/ml of rituximab (RTX) or LFB-R603. The median value of the % of CD107a þ NK cells without mAb was 3.5 (range: 0.8–15.0). Horizontal black bars represent median values. P-values were calculated with the Wilcoxon rank-sum test and refer to the comparison between rituximab and LFB-R603. **Po0.01, ***Po0.001.

1 ng/ml

Ctl (n=3)

75 50 25 0 0

0.1

* n=6

100 % NK ADCC

% NK ADCC

100

1

10

75

65.1

62.5

50 25 8.5

7.3

0

100 1000 10000

* n=7

100 ng/ml

% NK ADCC

% NK ADCC

100 100

CLL group 1 (n=4)

75 50 25 0

0

0.1

1

10

n=5

75

71.6 55.6

66.9

60.4

50 25 0

100 1000 10000

n=5

RTX

LFB-R603

RTX LFB-R603

Ctl 100

CLL group 2 (n=1) % NK ADCC

% NK ADCC

100 75 50 25 0

0

0.1

1

10

CLL

100 1000 10000

[mAb] ng/mL

75 50

RTX CLL LFB-R603 CLL RTX Ctl LFB-R603 Ctl

25 0

-2 -1

0

1

2

3

4

5

[mAb] ng/mL (LOG)

Figure 4 NK cell ADCC in the presence of rituximab or LFB-R603. (a) A dose-response study of ADCC against Raji cells was performed with or without anti-CD20 rituximab (n) or LFB-R603 (~), at six concentrations ranging from 0.1 to 10 000 ng/ml. Assays were performed with purified NK cells from controls and from CLL patients in group 1 and group 2. Mean percentage values±s.d. (vertical bars) are shown. (b) Comparison of ADCC activity against Raji cells, after treatment with 1 ng/ml or 100 ng/ml of rituximab (RTX) (n) and LFB-R603 (R603) (~). Horizontal black bars represent median values. (c) Percentage lysis of Raji cells induced by rituximab or LFB-R603 in the presence of NK cells from control individuals (n ¼ 3) or CLL patients (n ¼ 5). P-values were calculated with the Wilcoxon rank-sum test and refer to the comparison between rituximab and LFB-R603, *Po0.05. Leukemia


107 class-I-defective K562 target cells after IL-2 activation. Similar capacities were observed in all patients, regardless of their prognostic factors in agreement with previous reports.17,18 Altogether, these results contribute to our knowledge on NK-cell nature and function in CLL. Our results demonstrate the efficacy of NK cells from CLL patients, including patients with an unfavorable prognosis, at killing tumor Raji cells and degranulating against autologous B-CLL cells. Consequently, these data strongly support the use of anti-CD20 mAbs optimized for FcgR engagement to induce efficient ADCC. Surprisingly, very few studies have evaluated anti-CD20 mAb-mediated ADCC in CLL,9,32 with most anti-CD20 ADCC studies using non-Hodgkin’s lymphoma cell lines. Only two studies have investigated the ability of NK cells to mediate ADCC against primary CLL cells in the presence of anti-CD20 mAbs.9,32 Weitzman et al.32 showed that rituximab and veltuzumab, another anti-CD20 mAb optimized for enhanced complement-dependent cytotoxicity, resulted in similar levels of ADCC in the presence of primary CLL cells and NK cell lines. Recently, we have shown that EMAB-6, an anti-CD20 optimized for FcgRIIIA engagement, induced greater ADCC against CLL cells than rituximab, in the presence of NK cells from healthy donors.9 Of note, we focused our attention on NK cell-mediated ADCC. However, we cannot exclude a synergistic effect of macrophages and neutrophils with NK cells in this mechanism.8,33 Other studies have focused on the cytotoxic activity of anti-CD20 mAbs against non-Hodgkin’s lymphoma cells. With optimized GA-101 anti-CD20 mAb, the potency of ADCC against CD20-expressing non-Hodgkin’s lymphoma cells ranges from 5 to 100 times greater than with rituximab.34 Similarly, AME-D, another optimized anti-CD20 mAb, produces greater ADCC at lower mAb concentrations than rituximab against an non-Hodgkin’s lymphoma cell line.35 In the presence of NK cells from CLL patients, similar dose-response relationships were found for the induction of ADCC by the optimized anti-CD20 LFB-R603 mAb and rituximab, except that LFB-R603 could kill the same number of Raji cells at 1/100 the concentration of rituximab. ADCC did not increase with concentrations higher than 100 ng/ml for rituximab or 1 ng/ml for LFB-R603. Of note, LFB-R603 induces strong ADCC at a low concentration (1 ng/ml), regardless of the prognostic factors. Importantly, a robust increase in NK-cell degranulation against autologous CLL cells was observed in the presence of LFB-R603, in contrast to the results with rituximab. Together, these data indicate that the potency of ADCC with these two anti-CD20 mAbs differs considerably and that the anti-CD20 LFB-R603 mAb triggers better effector functions to clear CLL cells, which may be therapeutically beneficial. Intriguingly, considerable interindividual variation in degranulation efficacy was also observed, regardless of the patients’ prognosis. Previous studies have shown that the FcgRIIIA polymorphism is associated with differential affinity and binding to human IgG1 mAbs. However, in CLL, the FcgR polymorphism did not predict responses to low-fucosylated mAbs,32,36,37 suggesting that this polymorphism does not explain the variability in degranulation responses. Instead, NK-cell degranulation might be affected by the effector/target ratio, in agreement with Fischer et al.38 Finally, the individual variability might be also attributed to intrinsic resistance of CLL cells to degranulation as observed in 1/21 patients. Moreover, we noted a defect in NK-cell degranulation in two patients who had received rituximab before sampling (data not shown). Accordingly, functional characterization of NK cells against autologous B-CLL targets from each patient might be useful for predicting the quality of

the response to anti-CD20 treatment in relationship to clinical treatment history. In CLL, the introduction of anti-CD20 mAbs has brought about a remarkable change in treatment strategy, but limitations in their therapeutic use have also been noted. Rituximab is not widely used alone in CLL because of its low-response rate and the short-response duration. Furthermore, increased doses of rituximab, from 500 to 2250 mg/m2, are associated with significant toxicity without any enhancement of clinical benefits.39 On the basis of these success and limitations, different types of second-generation mAbs have been engineered. Ofatumumab (HuMax) and veltuzumab are designed to improve complement-dependent cytotoxicity, whereas GA-101 and LFB-R603, for example, are intended to increase the affinity of the mAb Fc region for FcgRIIIA, by amino-acid substitutions or modification of the carbohydrate structure of the mAb, or both. Lower doses of these antibodies must be sufficient to produce ADCC levels similar to or higher than those achieved with rituximab, as demonstrated in this study. Clinical trials with these antibodies may reveal that they have enhanced activity compared with rituximab, but without its side effects. Thus, an interim analysis in an international phase I trial of ofatumumab, used as single agent at a dose of 2000 mg in refractory CLL patients, reported a good safety profile and a high overall response rate in heavily pretreated patients.40 A multicenter phase I study of GA-101, used as single agent at escalating doses of 400–2000 mg, showed good responses with two cases of complete remission at intermediate doses (800/1200 mg), a tumor-burden decrease and good tolerance in 11/13 evaluable CLL patients.41 Finally, an ongoing phase I study is testing LFB-R603 at doses escalating from 75 to 1650 mg/m2. In summary, our findings clearly demonstrate that peripheral NK cells from CLL patients are functional in terms of degranulation and ADCC, although a certain degree of variability is observed. They support the idea that a functional characterization of NK cells from each CLL patient could be useful in predicting the outcome of clinical treatment with mAbs. Our data also show that LFB-R603 activity is more promising than that of rituximab. The potency of LFB-R603 at low doses may have an important impact in patients with severe disease or in those who are resistant to chemotherapy.

Conflict of interest J Decocq, C de Romeuf and JF Prost are employed by LFB, whose potential product was studied in the present report. CA Dutertre was supported by a CIFRE fellowship from the ANRT and LFB (#134/2004) between 2004 and 2007. J-L Teillaud was a LFB consultant until June 2007, which covers part of the study period. Three of the authors, C de Romeuf, JF Prost and JL Teillaud, are designated as inventors on patent application WO2006064121 owned by LFB Biotechnologies, and claim specific therapeutic use of the antibody studied in this report. The other authors declare no conflict of interest.

Acknowledgements We thank Z Azgui, K Maloum, F Nguyen-Khac, JL Binet, N Marchay, H M’Kada and S Nguyen-Quoc, from the Hematology Department at Pitie´-Salpeˆtrie`re Hospital (AP-HP, Paris, France) for the recruitment of CLL patients. We thank F Gandjbakhch, C Poulain and the personnel from the Department of Rheumatology at Pitie´-Salpeˆtrie`re Hospital (AP-HP, Paris, France) and Leukemia


108 V Siguret, E Pautas and the personnel from the Gerontology Department at Charles-Foix Hospital (AP-HP, Ivry, France) for the recruitment of the control group. P Bonnemye, M Brissard, S Gueguen, M Boudjoghra and A Grelier are acknowledged for their technical assistance. This study was supported in part by funds from the Laboratoire Franc¸ais de Fractionnement et des Biotechnologies (LFB, Les Ulis, France), the association la Ligue contre le cancer (RS08/75-4) and INSERM. The anti-CD20 LFB-R603 antibody was purchased by LFB.

Author contributions MLT, CAD, JLT, HMB and VV conceived and designed the study; MLT and JD performed the flow cytometric and functional analyses; CP performed the flow cytometric selection; EC provided cytogenetic analyses; FD provided IgHV sequences; HMB cared for the involved patients, provided clinical and diagnosis/prognosis cytometric data; MLT, JD, CdR, CAD, JLT, HMB and VV analyzed the flow cytometric and functional data; MLT, JLT, HMB and VV wrote the manuscript; JD, CdR, CP, CAD, EC, FD, PD and JFP critically read and approved the manuscript.

References 1 Keating MJ, O’Brien S, Albitar M, Lerner S, Plunkett W, Giles F et al. Early results of a chemoimmunotherapy regimen of fludarabine, cyclophosphamide, and rituximab as initial therapy for chronic lymphocytic leukemia. J Clin Oncol 2005; 23: 4079–4088. 2 Tam CS, O’Brien S, Wierda W, Kantarjian H, Wen S, Do KA et al. Long-term results of the fludarabine, cyclophosphamide, and rituximab regimen as initial therapy of chronic lymphocytic leukemia. Blood 2008; 112: 975–980. 3 Cartron G, Watier H, Golay J, Solal-Celigny P. From the bench to the bedside: ways to improve rituximab efficacy. Blood 2004; 104: 2635–2642. 4 Lazar GA, Dang W, Karki S, Vafa O, Peng JS, Hyun L et al. Engineered antibody Fc variants with enhanced effector function. Proc Natl Acad Sci USA 2006; 103: 4005–4010. 5 Delgado J, Briones J, Sierra J. Emerging therapies for patients with advanced chronic lymphocytic leukaemia. Blood Rev 2009; 23: 217–224. 6 Pleyer L, Egle A, Hartmann TN, Greil R. Molecular and cellular mechanisms of CLL: novel therapeutic approaches. Nat Rev Clin Oncol 2009; 6: 405–418. 7 Cheson BD. Ofatumumab, a novel anti-CD20 monoclonal antibody for the treatment of B-cell malignancies. J Clin Oncol 2010; 28: 3525–3530. 8 Leidi M, Gotti E, Bologna L, Miranda E, Rimoldi M, Sica A et al. M2 macrophages phagocytose rituximab-opsonized leukemic targets more efficiently than m1 cells in vitro. J Immunol 2009; 182: 4415–4422. 9 de Romeuf C, Dutertre CA, Le Garff-Tavernier M, Fournier N, Gaucher C, Glacet A et al. Chronic lymphocytic leukaemia cells are efficiently killed by an anti-CD20 monoclonal antibody selected for improved engagement of FcgammaRIIIA/CD16. Br J Haematol 2008; 140: 635–643. 10 Freud AG, Caligiuri MA. Human natural killer cell development. Immunol Rev 2006; 214: 56–72. 11 Vivier E, Tomasello E, Baratin M, Walzer T, Ugolini S. Functions of natural killer cells. Nat Immunol 2008; 9: 503–510. 12 Voso MT, Pantel G, Rutella S, Weis M, D’Alo F, Urbano R et al. Rituximab reduces the number of peripheral blood B-cells in vitro mainly by effector cell-mediated mechanisms. Haematologica 2002; 87: 918–925. 13 Clynes RA, Towers TL, Presta LG, Ravetch JV. Inhibitory Fc receptors modulate in vivo cytoxicity against tumor targets. Nat Med 2000; 6: 443–446. 14 Palmer S, Hanson CA, Zent CS, Porrata LF, Laplant B, Geyer SM et al. Prognostic importance of T and NK-cells in a consecutive series of newly diagnosed patients with chronic lymphocytic leukaemia. Br J Haematol 2008; 141: 607–614. Leukemia

15 Ziegler HW, Kay NE, Zarling JM. Deficiency of natural killer cell activity in patients with chronic lymphocytic leukemia. Int J Cancer 1981; 27: 321–327. 16 Foa R, Lauria F, Lusso P, Giubellino MC, Fierro MT, Ferrando ML et al. Discrepancy between phenotypic and functional features of natural killer T-lymphocytes in B-cell chronic lymphocytic leukaemia. Br J Haematol 1984; 58: 509–516. 17 Kay NE, Zarling J. Restoration of impaired natural killer cell activity of B-chronic lymphocytic leukemia patients by recombinant interleukin-2. Am J Hematol 1987; 24: 161–167. 18 Guven H, Gilljam M, Chambers BJ, Ljunggren HG, Christensson B, Kimby E et al. Expansion of natural killer (NK) and natural killer-like T (NKT)-cell populations derived from patients with B-chronic lymphocytic leukemia (B-CLL): a potential source for cellular immunotherapy. Leukemia 2003; 17: 1973–1980. 19 Matutes E, Owusu-Ankomah K, Morilla R, Garcia Marco J, Houlihan A, Que TH et al. The immunological profile of B-cell disorders and proposal of a scoring system for the diagnosis of CLL. Leukemia 1994; 8: 1640–1645. 20 Van Bockstaele F, Verhasselt B, Philippe J. Prognostic markers in chronic lymphocytic leukemia: a comprehensive review. Blood Rev 2009; 23: 25–47. 21 Beziat V, Nguyen S, Lapusan S, Hervier B, Dhedin N, Bories D et al. Fully functional NK cells after unrelated cord blood transplantation. Leukemia 2009; 23: 721–728. 22 Vuillier F, Tortevoye P, Binet JL, Dighiero G. CD4, CD8 and NK subsets in B-CLL. Nouv Rev Fr Hematol 1988; 30: 331–334. 23 Linet MS, Bias WB, Dorgan JF, McCaffrey LD, Humphrey RL. HLA antigens in chronic lymphocytic leukemia. Tissue Antigens 1988; 31: 71–78. 24 Mueller LP, Machulla HK. Increased frequency of homozygosity for HLA class II loci in female patients with chronic lymphocytic leukemia. Leuk Lymphoma 2002; 43: 1013–1019. 25 Poggi A, Venturino C, Catellani S, Clavio M, Miglino M, Gobbi M et al. Vdelta1 T lymphocytes from B-CLL patients recognize ULBP3 expressed on leukemic B cells and up-regulated by trans-retinoic acid. Cancer Res 2004; 64: 9172–9179. 26 Verheyden S, Demanet C. NK cell receptors and their ligands in leukemia. Leukemia 2008; 22: 249–257. 27 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. 28 Jewell AP, Worman CP, Giles FJ, Goldstone AH, Lydyard PM. Resistance of chronic lymphocytic leukaemia cells to interferon-alpha generated lymphokine activated killer cells. Leuk Lymphoma 1992; 7: 473–480. 29 Stevenson FK, Caligaris-Cappio F. Chronic lymphocytic leukemia: revelations from the B-cell receptor. Blood 2004; 103: 4389–4395. 30 Jak M, Mous R, Remmerswaal EB, Spijker R, Jaspers A, Yague A et al. Enhanced formation and survival of CD4+ CD25hi Foxp3+ T-cells in chronic lymphocytic leukemia. Leuk Lymphoma 2009; 50: 788–801. 31 Pallasch CP, Ulbrich S, Brinker R, Hallek M, Uger RA, Wendtner CM. Disruption of T cell suppression in chronic lymphocytic leukemia by CD200 blockade. Leuk Res 2009; 33: 460–464. 32 Weitzman J, Betancur M, Boissel L, Rabinowitz AP, Klein A, Klingemann H. Variable contribution of monoclonal antibodies to ADCC in patients with chronic lymphocytic leukemia. Leuk Lymphoma 2009; 50: 1361–1368. 33 Lefebvre ML, Krause SW, Salcedo M, Nardin A. Ex vivo-activated human macrophages kill chronic lymphocytic leukemia cells in the presence of rituximab: mechanism of antibody-dependent cellular cytotoxicity and impact of human serum. J Immunother 2006; 29: 388–397. 34 Umana P, Moessner E, Bruenker P, Unsin G, Puentener U, Suter T et al. Novel 3rd generation humanized type II CD20 antibody with glycoengineered Fc and modified elbow hinge for enhanced ADCC and superior apoptosis induction. Blood Rev 2006; 108: Abs 229. 35 Bowles JA, Wang SY, Link BK, Allan B, Beuerlein G, Campbell MA et al. Anti-CD20 monoclonal antibody with enhanced affinity for CD16 activates NK cells at lower concentrations and more effectively than rituximab. Blood 2006; 108: 2648–2654.


109 36 Farag SS, Flinn IW, Modali R, Lehman TA, Young D, Byrd JC. Fc gamma RIIIa and Fc gamma RIIa polymorphisms do not predict response to rituximab in B-cell chronic lymphocytic leukemia. Blood 2004; 103: 1472–1474. 37 Niwa R, Hatanaka S, Shoji-Hosaka E, Sakurada M, Kobayashi Y, Uehara A et al. Enhancement of the antibody-dependent cellular cytotoxicity of low-fucose IgG1 Is independent of FcgammaRIIIa functional polymorphism. Clin Cancer Res 2004; 10: 6248–6255. 38 Fischer L, Penack O, Gentilini C, Nogai A, Muessig A, Thiel E et al. The anti-lymphoma effect of antibody-mediated immunotherapy is based on an increased degranulation of peripheral

blood natural killer (NK) cells. Exp Hematol 2006; 34: 753–759. 39 O’Brien SM, Kantarjian H, Thomas DA, Giles FJ, Freireich EJ, Cortes J et al. Rituximab dose-escalation trial in chronic lymphocytic leukemia. J Clin Oncol 2001; 19: 2165–2170. 40 Wierda WG, Kipps TJ, Mayer J, Stilgenbauer S, Williams CD, Hellmann A et al. Ofatumumab as single-agent CD20 immunotherapy in fludarabine-refractory chronic lymphocytic leukemia. J Clin Oncol 2010; 28: 1749–1755. 41 Morschhauser F, Cartron G, Lamy T, Milpied NJ, Thieblemont C, Tilly H et al. Phase I study of RO5072759 (GA101) in relapsed/ refractory chronic lymphocytic leukemia. Blood 2009: Abs 884.

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

Leukemia