The chimeric anti-CD20 antibody rituximab induces apoptosis in B-cell chronic lymphocytic leukemia cells through a p38 mitogen activated protein–kinase–dependent mechanism

B-cell chronic lymphocytic leukemia (B-CLL) is the most common form of leukemia in the Western world, with an annual incidence exceeding 5 new cases per 100 000 individuals, and the disease must still be regarded as incurable.1,2 The pathogenesis of B-CLL is poorly understood, but deregulated control of apoptosis is believed to be critical for the characteristic accumulation of B lymphocytes in this disease.1,3 B-CLL cells are long-lived, small, mature CD5⁺CD19⁺CD23⁺ B cells that appear to be arrested early in the G₁-phase of the cell cycle.1-5 

CD20 is a 33- to 37-kd membrane protein of unknown function, expressed on mature B cells and their B-lineage bone marrow precursors.6-8 Because CD20 neither sheds from the cell surface nor internalizes upon antibody binding, sustained binding of anti-CD20 antibodies can activate effector mechanisms, such as complement-mediated lysis and antibody-dependent cellular cytotoxicity (ADCC), and immunotherapy with anti-CD20 antibodies has demonstrated significant efficacy in the treatment of B-cell neoplasms.9-14 However, in B-cell lines, cultured under conditions that preclude activation of these effector mechanisms, anti-CD20 antibodies also induce apoptosis.15-17CD20-induced apoptosis in these cell lines has been shown to be partially dependent on protein tyrosine kinase activity, Ca⁺⁺ mobilization, and effector caspase activation.16 

In this study, we have used freshly isolated primary leukemia cells from patients with B-CLL to analyze the functional consequences of binding of a mouse/human chimeric anti-CD20 antibody (rituximab) to CD20 expressed in the membrane of the malignant cells. We demonstrate that rituximab activates a CD20-mediated signaling pathway that results in apoptosis and is dependent on p38 mitogen activated protein (MAP)–kinase activation.

Fresh blood samples were obtained from patients previously diagnosed with CD19⁺CD5⁺CD23⁺B-CLL, who had been off all treatment, including corticosteroids, for at least 30 days. The mean percentage of CD5⁺CD20⁺ cells in the mononuclear fraction was 96.1% (range, 90%-99%). All patients signed informed consent. The study was approved by the regional committee on scientific ethics in Copenhagen.

Mononuclear cells were isolated from fresh blood by Lymphoprep separation (Nycomed Pharma, Oslo, Norway), washed, and stained with fluorochrome-conjugated antibodies against CD20 and CD5 (Becton Dickinson, Erembodegem, Belgium). Buffy coats from 16 healthy donors and tumor samples from 11 patients with low-grade non-Hodgkin lymphoma (NHLs) were used as controls. For analysis of CD20^dim cells in normal buffy coats, directly conjugated antibodies against CD3, CD4, CD8, CD13, CD14, CD19, CD22, CD23, CD33, CD34, CD38, CD56, CD95, kappa or lambda immunoglobulin light chains (all from Becton Dickinson) or FMC7, immunoglobulin (Ig)–M or IgG (DAKO, Glostrup, Denmark) were applied. We acquired 5000 to 30 000 ungated events and analyzed them on a FACsort flow cytometer (Becton Dickinson) with the use of the CellQuest software.

Mononuclear cells were cultured at 2 × 10⁶/mL in RPMI 1640 containing 10% heat inactivated fetal calf serum (Harlan Sera-Lab, Sussex, England) and antibiotics, in a humidified atmosphere containing 5% CO₂. Cell suspensions were supplemented with 4 μg/mL, unless otherwise indicated, of chimeric anti-CD20 antibody rituximab (kindly provided by IDEC Pharmaceuticals, San Diego, CA) or a control human IgG-preparation (SSI-IVIG, a kind gift from Dr Flemming Balstrup, Statens Seruminstitut, Copenhagen, Denmark), containing greater than 95% IgG with 61% IgG₁. For cross-linking, the cultures were further supplemented with anti–human IgG₁F(ab)₂ fragment (AO407, DAKO) at concentrations 5 times that of the primary antibody. For inhibition of p38 MAP-kinase activity, cells were preincubated for 60 minutes in the presence of 2.5 to 10 μM SB203580, an inhibitor of the p38 pathway, or 10 to 25 μM UO126, an inhibitor of the extracellular signal–regulated kinase (ERK) pathway (Alexis Biochemicals, San Diego, CA) and subsequently supplemented with rituximab or the control preparation.

For morphology studies, cytospin preparations were stained with May-Grünwald-Giemsa. To quantitate the extent of apoptosis, cells were harvested, washed in phosphate-buffered saline (PBS), and gently resuspended in 0.4 mL hypotonic fluorochrome solution (0.05 mg/mL propidium iodide (PI) in 0.1% sodium citrate, 0.37% NP40, and 0.02 mg/mL RNase A), as previously described.18-20There were 5000 events, at a rate of 100 to 200 events per second acquired on a FACSort flow cytometer, with the use of a threshold of 20 in FL-2 linear mode. For phosphatidylserine exposure and membrane integrity analysis, cells were double-stained with fluorescein isothiocyanate (FITC)–labeled annexin-V and PI with the Apoptosis Detection Kit I (6693KK; PharMingen, San Diego, CA) according to the manufacturer's instructions.21 There were 30 000 ungated events acquired on a FACsort flow cytometer; these were analyzed by means of the CellQuest software.

Cells were washed once in ice-cold PBS and lysed in ice-cold lysis buffer (10 mM Hepes [pH 7.3], 0.1% Triton X-100, 400 mM KCl, 2 mM EDTA, 1 mM ethyleneglycotetraacetic acid, 1 mM dithiothreitol [DTT], 1 mM Na₃VO₄, 1 mM phenylmethyl sulfonyl fluoride, 5 μg/mL leupeptin, 5 μg/mL aprotinin, 5 μg/mL pepstatin). Sodium dodecyl sulfate–polyacrylamide gel electrophoresis was performed with the use of equal amounts of whole cell lysate, followed by transfer to nitrocellulose filters. After blocking with 5% milk and overnight incubation with the primary antibodies (c-Jun NH2-terminal protein kinase [JNK] [sc-571], ERK [sc-94-G], p38 [sc-535] (Santa Cruz Biotechnology, Santa Cruz, CA) and pY-JNK (9255S), pY-ERK (9106S), and pY-p38 (9211S) (New England BioLabs, Beverly, MA), the nitrocellulose sheet was further processed by means of horseradish peroxidase-conjugated secondary antibody (PO260, PO448, or PO160) (DAKO) and developed by means of an enhanced chemiluminescence system (Amersham Pharmacia Biotech, Little Chalfont, England).

First, 1 × 10⁸ cells were lysed for 30 minutes on ice in 1 mL lysis buffer (50 mM Tris-base, 0.5% NP-40, 150 mM NaCl, 10 mM NaF, 10 mM sodium pyrophosphate, 1 mM EDTA, 10 μg/mL leupeptin, 10 μg/mL aprotinin). Cellular debris was removed by centrifugation at 12 000g for 10 minutes at 4°C. Total MAPKAP K2 was immunoprecipitated by addition of 3 μg primary antibody (06-534) (Upstate Biotechnology, Lake Placid, NY) overnight, followed by 10 μL protein G sepharose for an additional hour. The immunoprecipitates were washed 3 times in wash buffer (50 mM Tris base, 0.1% NP-40, 150 mM NaCl, 10 mM NaF, 10 mM sodium pyrophosphate, 1 mM EDTA, 10 μg/mL leupeptin, 10 μg/mL aprotinin) and once in kinase assay buffer (25 mM Hepes [pH 7.4], 25 mM β-glycerophosphate, 25 mM MgCl₂, 0.1 mM Na₃VO₄, 2 mM DTT). Then, 50 μL beads in assay buffer were left as a 1:1 suspension, and 15 μL MAPKAP K2 substrate (12-240) (Upstate Biotechnology) was added. In vitro kinase reactions were initiated by the addition of 5 μCi (1.85 × 10⁵ Bq) [γ³²-P]adenosine triphosphate and incubated at 30°C for 20 minutes. Reactions were spotted on P81 Whatman paper. The samples were washed 3 times for 5 minutes in 75 mM phosphoric acid and once in acetone and then air dried; their radioactivity was determined in a β-counter.

CD20 expression was analyzed with the unpaired nonparametrical Mann-Whitney test, and statistical analysis was carried out by means of GraphPad Prism software (San Diego, CA).

CD5⁺ B-CLL cells have been reported to express CD20 with weaker fluorescence intensity in their plasma membrane than B cells of CD5⁻ B-cell malignancies.20 We examined the expression of CD20 on B-CLL cells and compared this expression with normal CD5⁺ and CD5⁻ B cells and CD5⁻ low-grade NHL cells (Figure1). Using dual-color staining and flow cytometry, we were able to define 3 separate CD20⁺populations in healthy donors: (1) CD5^brightCD20^dim, (2) CD5^dimCD20^bright, and (3) CD5⁻CD20^bright (Figure 1, upper panel, right top). All B-CLL samples (n = 44) expressed CD20 but in intensities varying from CD20^dim to CD20^bright (Figure 1, left, bottom). CD20 expression on B-CLL cells was significantly weaker than on the CD20^brightCD5^dim or negative healthy populations (P < .001, Mann-Whitney), and the CD20^bright NHL cells (P < .004, Mann-Whitney). However, the intensity of CD20 staining on B-CLL cells was not significantly different from that of the healthy CD5^brightCD20^dim population (population 1 in Figure 1, top). Surprisingly, further analysis of this healthy CLL-like population demonstrated that these cells did not express B-cell markers such as CD19, CD22, CD23, or surface membrane immunoglobulin; they did not express myeloid markers such as CD13, CD14, or CD33; and they did not express the NK-cell marker CD56 (data not shown) (n = 6). However, the CD5^brightCD20^dim cells stained positive for the pan–T-cell marker CD3; they had an equal distribution between CD4⁺ and CD8⁺ cells; and they expressed predominantly the αβ-chain of the T-cell receptor although γδ–positive cells were also identified in this population. Thus, although normal CD5^brightCD20^dim cells express CD5 and CD20 with intensities very comparable to those of B-CLL cells, the healthy population appears to be of T lineage, not of B lineage (Figure1).

We studied the functional effect of rituximab in primary leukemia cells from patients with B-CLL, using 3 corroborative apoptosis assays: morphology analysis, PI staining of cell DNA, and detection of phospholipid asymmetry and plasma membrane integrity, by double staining with annexin V–FITC and PI. We cultured freshly isolated B-CLL cells in the presence or absence of rituximab. When cross-linked, rituximab induced morphological changes typical of apoptosis and significantly increased the fraction of hypodiploid nuclei (observed in 25 independent experiments) (Figure2A). This induction of apoptosis was dose and time dependent, with the maximal response at a dose of 4 μg rituximab per milliliter medium (Figure 2B-C).

Annexin V binds the membrane phospholipid phosphatidylserine, which is externalized from the inner to the outer leaflet of the plasma membrane in the early stage of apoptosis. When membrane integrity is lost, as seen in the later stage of cell death resulting from either the apoptotic or the necrotic processes, PI staining becomes positive. According to the results of our time-course study, annexin-positive apoptotic cells, after 24 hours of culture, were almost equally distributed between PI⁻ early apoptotic (mean, 43%; range, 26%-71%; n = 7) and PI⁺ late apoptotic (mean, 35%; range, 22%-59%; n = 7) cells (Figure 2C). In some, but not all cases, we did observe some induction of apoptosis in the presence of the secondary antibody alone, or in the presence of the cross-linked IgG1 control, but cross-linked rituximab was always more potent in induction of apoptosis (Figure 2B). To address the role of the secondary antibody itself, we performed experiments in which the culture plates were coated with immobilized IgG1 or rituximab, and even under these conditions rituximab induced apoptosis, whereas IgG1 did not (data not shown). Furthermore, antibodies against CD19, alone or cross-linked, failed to induce apoptosis in B-CLL cells, and the rituximab preparation did not induce apoptosis in primary human T lymphocytes cultured under identical conditions (data not shown). These results demonstrate that cross-linking of rituximab bound to CD20 on the surface of freshly isolated B-CLL cells induces apoptosis.

Binding of anti-CD20 antibodies to CD20 induces phosphorylation of protein tyrosine kinases and calcium mobilization in B-cell lines.15-17 These proximal membrane signals could theoretically lead to activation of MAP kinases in a CD20-mediated signaling pathway.22 We analyzed whether this was the case in the CD20-mediated apoptosis signaling pathway. MAP-kinase activation was assessed by Western blot analysis on whole cell lysates with anti–phospho kinase (MAPK) antibodies, which specifically recognize an activated form of MAPK (ie, MAPK phosphorylated at tyrosine and serine residues). We cultured freshly isolated B-CLL cells in the presence or absence of rituximab and the cross-linking F(ab)₂ fragment. Rituximab, when added alone, induced weak phosphorylation of JNK and ERK and had little or no impact on the phosphorylation of p38 (Figure 3A, top). However, cross-linking of rituximab resulted in strong and sustained phosphorylation of all 3 MAP kinases (observed in 3 out of 3 independent experiments) (Figure 3A, bottom). To determine if this rituximab-induced MAP-kinase activation was related to the induction of apoptosis, we next analyzed the effect of inhibitors of the MAP-kinase pathways. The MEK-inhibitor UO126 had no inhibitory effect on CD20-mediated apoptosis, even though it completely inhibited rituximab-induced ERK activation (data not shown). In contrast, the p38 inhibitor SB203580 completely abrogated rituximab-induced MAPKAP K2 activity downstream of p38 (observed in 3 out of 3 independent experiments) (Figure 3B, bottom), and decreased rituximab-induced apoptosis by a mean of 41% (range, 16%-67%) (Figure 3B, top). In additional dose-response experiments, SB203580 significantly inhibited CD20-mediated apoptosis, even at doses from 2.5 μM (data not shown).

Anti-CD20–based immunotherapy of B-cell malignancies has been developed in an attempt to activate immune-effector mechanisms, such as ADCC and complement-mediated lysis, and thereby induce cell death in the malignant clone.10 Activation of both effector mechanisms have been demonstrated in vitro and in vivo, but these observations do not preclude the possibility that the clinical efficacy of rituximab may involve additional effector mechanisms.13,14 23 

Because CD5⁺ B-CLL cells have been reported to express less CD20 than other B cells,24 we first examined the phenotype of B-CLL cells in terms of CD5 and CD20 coexpression. We confirmed that CD20 expression is weaker on B-CLL cells compared with normal CD5⁺ and CD5⁻ B lymphocytes and malignant CD5⁻ B cells from samples of low-grade NHLs. In healthy donors, we identified a small, but very consistent, population of cells that have a CD5^brightCD20^dim phenotype, indistinguishable from that of B-CLL cells. Interestingly, as previously reported by others,25 this population is in fact a small population of CD20^dim T lymphocytes. Studies are underway to characterize this population further. Given that the dim expression of CD20 has been noted on some T-cell malignancies,26 including both leukemias and lymphomas, it is important to assess the exact frequency of CD20 expression on T-cell malignancies and, in particular, to investigate the functional consequences of binding of rituximab to such cells.

We next demonstrated that, despite the dim CD20 expression, rituximab induced apoptosis in freshly isolated B-CLL cells under conditions in which the contributions from the above-mentioned effector mechanisms are negligible. First, complement is not active in heat-inactivated serum. Second, our samples contained a median of 96.1% B-CLL cells, and the vast majority of the residual cells are CD5⁺T-cells. Therefore, the ratio between B-CLL cells and cells capable of mediating the cytotoxic response, ie, NK cells and monocytes, is at best 1:100 or lower, making it unlikely that cytotoxicity has any significant impact on the apoptotic response. The induction of apoptosis requires cross-linking or immobilization of the antibody and appears to be specific for rituximab. First, although the cross-linking antibody by itself, or together with an IgG1 control preparation, can induce a small degree of apoptosis, this response is not observed in all cases in which rituximab induces apoptosis. The mechanism for this unspecific induction of apoptosis is not clear, but may be attributed to interaction through Fc receptors or immunoglobulin on the surface of the B-CLL cell. Second, antibodies against CD19, both alone and cross-linked, failed to induce apoptosis in B-CLL cells. Third, the rituximab preparation did not induce a measurable apoptotic response in normal T lymphocytes. Therefore, we conclude that the major mechanism by which rituximab works in our system is induction of apoptosis.

In cell lines, rituximab-induced apoptosis has been reported to involve calcium mobilization and protein tyrosine kinase phosphorylation.16 In many instances, such signals result in activation of serine/threonine MAP kinases. The dynamic balance of activation of the MAP-kinase pathways is thought to be important in determining whether a cell survives or undergoes apoptosis.27 An important finding in our study is the association between activation of MAP kinases and rituximab-induced apoptosis. Specifically, our results indicate that CD20-mediated apoptosis is dependent on the activity of the p38 MAP-kinase in B-CLL cells. The data presented in this study demonstrate that only cross-linking of rituximab induces a strong, sustained activation of p38 and a significant degree of apoptosis. When the activity of p38 was completely blocked, as evidenced by the absence of MAPKAP K2 activity in SB203580-treated cultures, the response to rituximab in terms of apoptosis induction was significantly decreased in 7 out of 7 experiments. The ERK inhibitor UO126 did not prevent rituximab-induced apoptosis. Although the unavailability of a specific JNK inhibitor precludes any conclusions about the role of JNK activation in rituximab-induced apoptosis, our results strongly indicate that rituximab-induced apoptosis is dependent on p38 activation in B-CLL cells. The specificity of the p38 inhibitor we applied has recently been questioned, but the data on this issue are not equivocal.28-30 Furthermore, we observed the inhibitory effect even at low doses at which the specificity of the inhibitor is not questioned (data not shown).

The p38 MAP kinase has been found to be important in many cellular apoptosis systems. A dependence of p38 MAPK activity in B-cell–mediated apoptosis has, for example, been demonstrated in B-cell receptor (BCR)–mediated apoptosis in the B104 B-cell line.31 Recently, both rituximab and anti-BCR antibodies were shown to induce apoptosis and lead to ERK activation in B-cell lines at various maturational stages.32 In contrast, analysis of Fas-mediated apoptosis in the BJAB B-cell line showed no dependence on p38 activity.31 These data suggests that the p38 MAPK pathway plays a complex role in different apoptosis pathways in B cells. Although our results demonstrate dependence on p38 MAP-kinase activitation for the induction of CD20-mediated apoptosis, complete inhibition of p38 activity resulted in only partial inhibition of apoptosis. This observation is consistent with previous reports, in which CD20-mediated apoptosis was shown to be partially dependent on calcium mobilization as well as on protein tyrosine kinase activation.16 These studies, along with the data presented here, suggest that the ability of rituximab to induce apoptosis involves a specific receptor-ligand–like interaction between the antibody and CD20. Specific components of the resulting CD20-mediated signaling cascade appear to be important for the ability of the antibody to induce apoptosis in B lymphocytes, and our results indicate that such molecules are to be defined downstream of p38. Identification of these molecules, and dissection of the pathway(s) that link CD20 to the apoptotic machinery, may provide new therapeutic targets in B-CLL and possibly other B-cell neoplasms.