The anti-CD20 antibody rituximab augments the immunospecific therapeutic effectiveness of an anti-CD19 immunotoxin directed against human B-cell lymphoma


  • David J. Flavell,

    1. The Simon Flavell Leukaemia Research Unit and Cancer Research UK, Division of Cancer Sciences, Department of Medical Oncology, University of Southampton Medical School, Southampton General Hospital, Southampton, Hampshire, UK
    Search for more papers by this author
  • Sarah L. Warnes,

    1. The Simon Flavell Leukaemia Research Unit and Cancer Research UK, Division of Cancer Sciences, Department of Medical Oncology, University of Southampton Medical School, Southampton General Hospital, Southampton, Hampshire, UK
    Search for more papers by this author
  • Christine J. Bryson,

    1. The Simon Flavell Leukaemia Research Unit and Cancer Research UK, Division of Cancer Sciences, Department of Medical Oncology, University of Southampton Medical School, Southampton General Hospital, Southampton, Hampshire, UK
    Search for more papers by this author
  • Sarah A. Field,

    1. The Simon Flavell Leukaemia Research Unit and Cancer Research UK, Division of Cancer Sciences, Department of Medical Oncology, University of Southampton Medical School, Southampton General Hospital, Southampton, Hampshire, UK
    Search for more papers by this author
  • Armorel L. Noss,

    1. The Simon Flavell Leukaemia Research Unit and Cancer Research UK, Division of Cancer Sciences, Department of Medical Oncology, University of Southampton Medical School, Southampton General Hospital, Southampton, Hampshire, UK
    Search for more papers by this author
  • Graham Packham,

    1. The Simon Flavell Leukaemia Research Unit and Cancer Research UK, Division of Cancer Sciences, Department of Medical Oncology, University of Southampton Medical School, Southampton General Hospital, Southampton, Hampshire, UK
    Search for more papers by this author
  • Sopsamorn U. Flavell

    1. The Simon Flavell Leukaemia Research Unit and Cancer Research UK, Division of Cancer Sciences, Department of Medical Oncology, University of Southampton Medical School, Southampton General Hospital, Southampton, Hampshire, UK
    Search for more papers by this author

Dr David J Flavell, The Simon Flavell Leukaemia Research Unit, Southampton General Hospital, Southampton, Hampshire SO16 6YD, UK. E-mail:


The chimaeric anti-CD20 antibody rituximab (Rituxan) sensitises lymphoma cells to small molecule cytotoxic drugs and to protein toxins. We have explored the augmentive effect of rituximab on the anti-CD19 immunotoxin BU12-SAPORIN in a model of human lymphoma. Intact rituximab and its F(ab)2 derivative both augmented the immunospecific protein synthesis inhibitory effects of BU12-SAPORIN in a complement-independent manner. A combination of rituximab + BU12-SAPORIN completely abolished the proliferation of Ramos cells in vitro and also induced a significantly greater degree of apoptosis in these cells. Treatment with rituximab, BU12-SAPORIN or a combination of both induced poly(ADPribose) polymerase and caspase 3 cleavage, although this was always consistently greater in combination-treated cells. zVAD almost completely inhibited apoptosis in rituximab- or BU12-SAPORIN-treated cells but only partially in combination-treated cells. In severe combined immunodeficient (SCID)-Ramos mice the combination of rituximab + BU12-SAPORIN was significantly better therapeutically than either single agent. The immunological fidelity of the therapeutic effect because of combination treatment was demonstrated through the failure of rituximab to augment an irrelevant anti-CD7 immunotoxin. The therapeutic efficacy of rituximab and combination treatment was reduced in SCID-Ramos mice depleted of serum complement while natural killer cell depletion failed to show any convincing role for antibody-dependent cellular cytotoxicity. This study shows a clear therapeutic advantage from using rituximab to immunospecifically augment immunotoxin cytotoxicity warranting further investigation.

Recent advances have given a new impetus to both passive and active immunotherapeutic approaches for the treatment of human haematological malignancies. The great attraction in harnessing immunological mechanisms to eliminate cancer cells lies in the promise of the potentially high degree of selective tumour cell kill that might be achievable via mechanisms that do not overlap with small molecule cytotoxic drugs, thus potentially overcoming drug resistance problems. The introduction of the chimaeric anti-CD20 antibody rituximab (Rituxan) into mainstream clinical practice as an alternative or complementary treatment for follicular lymphoma (FL) (McLaughlin et al, 1998) and other B-cell malignancies (Byrd et al, 2001) represented the first major success for an immunotherapeutic in cancer. Rituximab induces complete remissions in a significant proportion of FL patients (Foran et al, 2000) but the treatment is non-curative with the majority eventually having a recurrence of their disease (Davis et al, 2000).

The mechanism(s) by which rituximab induces apoptotic cell death in lymphoma cells is not precisely understood and evidence exists for the involvement of complement-mediated lysis (Golay et al, 2000; Harjunpaa et al, 2000; Cragg et al, 2003; Manches et al, 2003), antibody-dependent cellular cytotoxicity (ADCC) (Stockmeyer et al, 2000; Voso et al, 2002) and the direct triggering of apoptotic cell signalling cascades involving Src-family kinases (Hofmeister et al, 2000), mitogen-activated protein kinases (Pedersen et al, 2002), cellular calcium influx (Shan et al, 2000) through caspase-dependent (Byrd et al, 2002) and -independent (van der Kolk et al, 2002) mechanisms.

Rituximab also sensitises lymphoma cells to the action of a range of small molecule cytotoxic drugs (Demidem et al, 1997; Emmanouilides et al, 2002; Rose et al, 2002). The precise mechanism of sensitisation is unknown but there is experimental evidence to suggest that in some instances this may be via inactivation of signal transducers and activators of transcription 3(STAT3) and subsequent inhibition of the interleukin 10 autocrine/paracrine loop with a resultant downregulation of the mitochondrial anti-apoptotic protein Bcl-2 (Alas & Bonavida, 2001; Alas et al, 2001). There are likely to be other, as yet undescribed, mechanisms operative that may be exploitable in the future. Such laboratory-based observations have led to early stage clinical trials in low grade lymphoma (Czuczman et al, 1999; Czuczman, 1999) to evaluate the therapeutic benefits of a rituximab given together with cytotoxic chemotherapy. A recent study indicates that rituximab + CHOP (cyclophosphamide/doxorubicin/vincristine/prednisone) therapy in elderly patients with low grade follicular and diffuse large cell lymphoma has positive benefits with greater response rates and significant increases in event-free and overall survival (Coiffier et al, 2002). Rituximab also increases the sensitivity of lymphoma cells to protein toxins, such as ricin A chain (Ghetie et al, 2001), saporin (Flavell et al, 2000a) and B-cell-specific immunotoxins, based on these two molecules (Flavell et al, 2000b; Ghetie et al, 2001). The present study describes the immunospecific augmentive effect that rituximab exerts on the therapeutic activity of a saporin-based immunotoxin directed against the CD19 molecule on the surface of a human lymphoma cell line and delineates some of the molecular events that define this.

Materials and methods

Ramos and NALM-6 cell lines

The human Burkitt lymphoma cell line Ramos (Klein et al, 1975) and paediatric B-lineage acute lymphoblastic leukaemia (ALL) cell line NALM-6 (Hurwitz et al, 1979) were obtained from the European Collection of Cell Cultures and grown in RPMI 1640 medium containing 10% fetal calf serum (FCS) (R10).

Protein synthesis inhibition assay

Inhibition of protein synthesis in target Ramos or NALM-6 cells was determined using a 3[H]-leucine incorporation assay as described previously (Flavell et al, 1991).

Antibodies, toxins, caspase inhibitors and immunotoxins

Rituximab (MabThera; Roche, Welwyn Garden City, UK) was obtained from the hospital pharmacy. Anti-poly (ADPribose) polymerase (PARP) murine monoclonal antibody was obtained from R&D Systems (Oxford, UK). Murine monoclonal caspase 3 antibody was obtained from New England Biolabs (Hitchin, UK). Antibodies recognising bovine and murine C1q and C9 complement components and CD55-R phyocerythrin (RPE)- and CD59-fluorescein isothiocyanate (FITC)-conjugated antibodies were obtained from Serotec Ltd (Kidlington, UK). A rabbit polyclonal anti-asialo GM1 antibody was obtained from Wako Chemicals (Richmond, VA, USA). Cobra venom factor (CVF) was obtained from Venom Supplies pty Ltd (Tanunda, Australia). The broad spectrum caspase inhibitor z-VAD-fmk was obtained from Calbiochem (Nottingham, UK). The ribosome inactivating protein Saporin was extracted from the seeds of Saponaria officinalis supplied by Chiltern Seeds, Ulverston, Cumbria, UK (Stirpe et al, 1983) and used to construct immunotoxins as described previously (Morland et al, 1994; Flavell et al, 1995).

Flow cytometry

A Coulter Epics XL flow cytometer [Beckman Coulter (UK) Ltd, High Wycombe, UK] equipped for three-colour analysis and running Expo 32 software was used for all the flow cytometric analyses described. For single and dual colour analysis of CD19 and CD20 expression levels by Ramos and NALM-6 cells, cells were incubated with a saturating concentration of BU12 or rituximab antibody in phosphate-buffered saline (PBS) containing 0·05% sodium azide at 4°C for 30 min. Cells were washed by centrifugation in cold PBS also containing azide and the pellets resuspended and incubated for a further 30 min in cold PBS containing FITC-labelled goat anti-murine whole immunoglobin IgG antisera (10 μg/ml) (Sigma Chemical Company, Poole, UK) or FITC-labelled sheep anti-human whole IgG (10 μg/ml) (Dako Ltd, High Wycombe, UK) respectively. For two-colour analysis of CD55 and CD59 expression levels on Ramos cells phycoerythrin and FITC directly conjugated antibodies were utilised.

Immunofluoresecence microscopy

Cytospin preparations of Ramos cells previously incubated with rituximab at 37°C in the presence of heat-inactivated (HI) FCS or HI FCS supplemented with severe combined immunodeficient (SCID) mouse serum were fixed in 4% paraformaldehyde and then stained with anti-C1q or C9 antibody. Bound antibody was visualised with a FITC-labelled affinity purified F(ab)2 fraction of sheep anti-mouse immunoglobulin (Sigma Chemical Co.). Slides were coverslipped in mountant containing anti-fade 1,4-diazabicyclo[2·2·2]octane and inspected on a Leica TCS 4D confocal microscope [Leica Microsystems (UK) Ltd, Milton Keynes, UK].

Radioiodinated antibody binding

Rituximab and BU12 antibodies were radiolabelled with 125I using IodoBeads (Pierce & Warriner, Warrington, UK) according to the manufacturer's instructions. The binding characteristics of radiolabelled antibody to Ramos or NALM-6 cells was determined on cell pellets by gamma-counting in the presence of cold antibody. The number of CD19 and CD20 molecules expressed on the cell surface was subsequently determined by Scatchard analysis as described previously (Scatchard, 1949)

Annexin V apoptosis assay

A flow cytometric technique employing an annexin V/propidium iodide (PI) kit supplied by Immunotech/Beckman Coulter (High Wycombe, UK) was used according to the manufacturer's instructions to quantify apoptosis in Ramos and NALM-6 cells.

Western blotting

Western blotting was conducted as described (Bryson, 2004) so as to quantify the extent of PARP and caspase 3 cleavage in Ramos cells following the various treatments. Cell pellets for analysis were extracted with a deoxycholate/nonidet P40-based extraction buffer (radioimmunoprecipitation assay buffer) for PARP analysis or a 3,3-cholamidopropyl-dimethylammonio-1-propanesulphonate (CHAPS) extraction buffer for caspase 3 analysis.

ADCC and complement-dependent cytotoxicity (CDC) chromium release assay

A chromium release assay that we previously described (Flavell et al, 1998) was used to quantify ADCC utilising SCID mouse splenocytes as effector cells or for complement-dependent cytotoxicity (CDC). FCS or SCID mouse serum was used as a source of complement. For ADCC assays splenocytes were harvested from SCID mice 24 h after injection of 100 μg of Poly inosinic:cytidylic acid (Poly I:C) (Sigma Chemical Co.) and used at a target:effector cell ratio of 1:50. All ADCC assays were conducted in HI FCS.

SCID mice

Pathogen-free CB.17 scid/scid (SCID) mice (Bosma et al, 1983) of both sexes 6–10 weeks of age were produced from our own breeding colony and used in all the experimental work described here. Animals for experimental use were transferred to autoclaved filter top micro-isolator cages and housed on sterile bedding as five single sex animals per cage and provided with sterile water and food ad libitum.

SCID mouse experiments

Decomplementation with CVF

SCID-Ramos mice were injected intravenously (i.v.) with 1 μg of CVF 24 h prior to therapy. Decomplemented serum was also obtained from SCID mice similarly injected from which blood was collected 24 h later. We have previously confirmed that this single treatment is sufficient to completely abolish the ability of serum to initiate CDC for at least 72 h (D.J. Flavell & S.L. Warnes, unpublished observations).

In vivo elimination of natural killer cells from SCID mice with Asialo GM1 antibody

SCID mice were injected i.v. with 10 μg of a rabbit anti-asialo GM1 antibody (Wako Chemicals, Richmond, VA, USA) and used in therapy studies 24 h later. We have previously confirmed that this single treatment is sufficient to effectively eliminate the functionally lytic splenocyte population for up to 10 d after treatment (Flavell et al, 1998).

SCID mouse therapy studies

Groups of 10 SCID mice were injected via a tail vein with 2 × 106 viable Ramos (SCID-Ramos mice) or NALM-6 (SCID-NALM-6 mice) cells in a 100 μl volume of R10 medium. Both the SCID-Ramos (Flavell et al, 1997) and SCID-NALM-6 (Flavell et al, 1995) models have been described in detail previously. Seven days later, animals in each group were treated with individual drugs (10 μg rituximab, 10 μg BU12-SAPORIN or a molar equivalent 11·67 μg F(ab)2 rituximab) or combinations of these (10 + 10 μg or 10 + 11·67 μg in the case of F(ab)2 rituximab) by injection into the tail vein in a 100 μl volume of vehicle PBS (pH 7·2). Control animals were sham treated with the same volume of PBS only. The experiment end point was the appearance of hind leg paralysis or a deteriorating condition whereupon animals were killed and autopsies conducted to confirm the presence of tumour.

Statistical analysis

Log-rank analysis (Peto's method) was undertaken using the SOLO Statistics Software application (BMDP, Los Angeles, CA, USA). P-values of 0·05 or less were considered as statistically significant. For statistical analysis of apoptosis data, two-sample Student's t-tests were performed. Levene's test was used to accept (P > 0·05) or reject (P < 0·05) the assumption of equal variances of both ranges of data.


Cell surface expression levels of CD19, CD20, CD55 and CD59 on Ramos and NALM-6 cells

We first determined, by flow cytometry and radio-iodinated antibody binding, the levels of CD19 and CD20 expression on the surface of Ramos (B-NHL) and NALM-6 (B-ALL) cells. Both the flow cytometric (Fig 1A and B) and Scatchard analysis of radio-iodinated antibody binding (Fig 1C) were in general agreement and both methods showed that Ramos cells expressed approximately 10 times more CD20 (mean 330 000 molecules/cell) than CD19 (mean 32 000 molecules/cell) molecules on their surface (Fig 1C). NALM-6 cells expressed CD19 at an approximately equivalent density (mean 36 000 molecules/cell) but CD20 was expressed only at very low levels (mean 653 molecules/cell) (Fig 1C). Two-colour flow cytometry demonstrated that the majority of Ramos cells (80·58%) were negative for both CD55 (DAF) and CD59 (protectin) with the same 15·5% subpopulation expressing both (Fig 1D).

Figure 1.

 Antigen expression levels on Ramos and NALM-6 cells stained with saturating concentrations of antibodies. Ramos (A) and NALM-6 (B) cells were stained with the anti-CD19 antibody BU-12 (solid histogram), the anti-CD20 antibody rituximab (—) or an irrelevant isotype matched control antibody (- - -) and analysed by flow cytometry. (C) The average numbers of CD19 and CD20 molecules expressed on the surface of Ramos and NALM-6 cells determined by Scatchard analysis. (D) Expression of CD55 and CD59 by Ramos cells determined by two-colour flow cytometric analysis.

Rituximab antibody is cytolytic for Ramos cells by CDC and ADCC

Rituximab antibody but not its F(ab)2 fragment produced increasing dose-dependent CDC lysis of target Ramos cells with 10% standard FCS as complement source (Fig 2A). When the FCS was HI, rituximab did not cause target cell lysis. BU12 antibody (anti-CD19) was not lytic to Ramos cells over the same concentration range. The lytic activity of Rituximab could be partially restored to HI FCS by the addition of 10% normal SCID mouse serum but not with CVF de-complemented SCID mouse serum (Fig 2B). Neither rituximab nor BU12 antibody were lytic by CDC for NALM-6 cells (data not shown).

Figure 2.

 Lysis of Ramos cells as measured by chromium release following exposure to increasing molar concentrations of antibodies. (A) Concentration-dependent lysis of Ramos cells exposed to whole rituximab antibody (bsl00001), F(ab)2 rituximab (•) or BU12 antibody (bsl00000) in standard fetal calf serum (FCS) or to whole rituximab in heat inactivated (HI) FCS (bsl00066). (B) Restoration of lytic activity of rituximab for Ramos cells in HI FCS by severe combined immunodeficient (SCID) mouse serum. Ramos cells were exposed to increasing molar concentrations of whole rituximab antibody in untreated (bsl00001) and HI (bsl00000) FCS and in HI FCS supplemented with 10% SCID mouse serum (bsl00043) or to BU12 antibody (bsl00084) in HI FCS supplemented with 10% SCID mouse serum. (C) Antibody-dependent cellular cytotoxicity (ADCC) elicited against Ramos target cells by intact rituximab (•), a F(ab)2 derivative of rituximab (bsl00066) or BU12 antibody (bsl00001). (D–M) Ramos cells stained for C1q (D–H) or C9 (I–M) following incubation with rituximab + HI FCS (D and I), rituximab + FCS (E) rituximab + SCID mouse serum (J), rituximab + HI FCS + CVF treated SCID mouse serum (K), no primary antibody + HI FCS + SCID mouse serum (G and L) and BU12 antibody + HI FCS + SCID mouse serum (H and M). (Scale bar = 20 μm).

We next conducted an experiment to determine if rituximab was capable of eliciting ADCC against Ramos cells with SCID mouse splenocytes as cytotoxic effector cells. Intact rituximab but not the F(ab)2 rituximab derivative was capable of eliciting ADCC in a dose-dependent manner (Fig 2C). BU12 antibody showed a small degree of lytic activity in ADCC but this was relatively minor and was demonstrable only at the highest antibody concentration.

Utilising antisera against complement components C1q (Fig 2D–H) and C9 (Fig 2I–M) we used confocal microscopy to visualise the rituximab-mediated fixation of these two complement components on the Ramos cell surface. C1q could be detected on the cell surface when rituximab-coated Ramos cells were incubated in HI FCS as C1q is not destroyed by heat treatment at 56°C (Fig 2D). In contrast, visualisation of cell surface C9, (i.e. destroyed by heat treatment), was much reduced in HI FCS (Fig 2I). This was fully restorable when SCID mouse serum was added (Fig 2J) but only partially so when SCID mouse serum from CVF-treated animals was added (Fig 2K). Only autofluorescence for C1q staining was detected when rituximab was omitted (Fig 2G) and, for C9 staining, intensity was very much reduced when rituximab was omitted (Fig 2L). Cells coated with BU12 antibody in the presence of HI FCS and SCID mouse serum did not stain for C1q (Fig 2H) indicating that this antibody was incapable of fixing C1q. Unexpectedly BU12-coated Ramos cells stained for C9 at the cell surface in the presence of HI FCS and SCID mouse serum (Fig 2M).

Rituximab antibody augments the protein synthesis and growth-inhibitory effects of BU12-SAPORIN for Ramos cells in both complement-dependent and -independent manners

Intact and F(ab)2 rituximab both augmented the protein synthesis inhibition (PSI) capability of BU12-SAPORIN approximately 20- and 35-fold respectively, in standard FCS (Fig 3A) and HI FCS (data not shown). The PSI activity of unconjugated saporin was also augmented by intact or F(ab)2 rituximab by 21- and 15-fold respectively (Fig S1A). Intact rituximab or its F(ab)2 derivative used alone had a relatively small inhibitory effect on protein synthesis in Ramos cells cultured in standard FCS (Fig 3A) or HI FCS (data not shown). When the irrelevant anti-CD7 saporin immunotoxin HB2-SAPORIN was used there was no significant effect on protein synthesis in Ramos cells used either alone or in combination with rituximab antibody (Fig S1B).

Figure 3.

 Protein synthesis inhibition (A) and cell proliferation inhibition (B and C) in Ramos cells exposed to intact or a F(ab)2 fragment of rituximab, BU12-SAPORIN, BU12 antibody or saporin, as single agents or in combination. In (A) triplicate cultures of Ramos cells were exposed to increasing molar concentrations of BU12-SAPORIN (bsl00001), whole rituximab (○), F(ab)2 rituximab (bsl00084), a combination of increasing concentrations of BU12-SAPORIN + 10 μg/ml whole rituximab (•) or increasing concentrations of BU12-SAPORIN + 11·67 μg/ml F(ab)2 rituximab (bsl00066) in heat inactivated fetal calf serum. Bars represent standard deviations for triplicate cultures. In (B and C) Ramos cell growth was followed in flask cultures exposed continuously to (B) 10 μg/ml whole rituximab (○), 11·67 μg/ml F(ab)2 rituximab (bsl00084), 10 μg/ml BU12-SAPORIN (bsl00001),10 μg/ml BU12-SAPORIN + 10 μg/ml whole rituximab (•), 10 μg/ml BU12-SAPORIN + 11·67 μg/ml F(ab)2 rituximab (bsl00066) or medium control (bsl00079) (C) 10 μg/ml BU12 antibody (bsl00063), 10 μg/ml whole rituximab (bsl00067), 10 mg/ml BU12 antibody + 10 μg/ml whole rituximab (○) or medium control (bsl00079). Bars represent the standard deviations obtained for cell counts on two duplicate flasks.

Intact rituximab, but not the F(ab)2 derivative, delayed cell growth (Fig 3B) but the final cell density eventually achieved was similar to that obtained in the control culture. In contrast BU12-SAPORIN had a more pronounced effect, delaying outgrowth longer and additionally reducing the final cell density achieved. When BU12-SAPORIN was combined with intact rituximab, cell proliferation was completely abolished, whereas when combined with F(ab)2 rituximab only a growth delay effect was achieved albeit of a longer duration than for that obtained with BU12-SAPORIN alone (15 d vs. 22 d) (Fig 3B). In HI FCS the anti-proliferative effects of the BU12-SAPORIN/rituximab combination was similarly reduced to a level intermediate between BU12-SAPORIN alone and the combination in standard serum (see Fig S1C).

Combination rituximab antibody/BU12-SAPORIN treatment of Ramos cells induces significantly more apoptosis in Ramos cells than either drug used alone

Annexin V/PI staining was used to quantify apoptosis in cultures of treated Ramos cells. Experiments were performed in culture medium containing standard or HI FCS. Representative quadrant dot plots for the three treatment and control groups are shown in Fig 4A for cells treated in standard FCS (data for cells cultured in HI FCS is not shown). The data from six separate similar experiments was pooled, averaged and reduced to histogram format by plotting the percentages of viable (Fig 4B, quadrant F3) or both early + late apoptotic representing the total apoptotic population present (Fig 4E, quadrants F2 + F4) cells cultured in both standard and HI FCS.

Figure 4.

 (A). Representative Annexin V/PI staining data obtained for Ramos cells treated with Rituximab, BU12-SAPORIN or a combination of both reagents for 24 h and 48 h. Viable cells are seen in quadrant F3, Early apoptotic cells in F4, secondary necrotic cells (late apoptotic) in F2 and primary necrotic in F1. (B to E) Reduced annexin V/PI staining data observed in Ramos cells. Histograms for each treatment group represent (B) percentage viable cells (quadrant F3), (C) percentage of secondary necrotic cells (quadrant F2), (D) percentage of early apoptotic cells (quadrant F4) and (E) the combined total percentages of early apoptotic (quadrant F4) + late apoptotic (quadrant F2) cells as a measure of the total apoptotic population present. Results in (B–E) represent mean values obtained for six separate replicate experiments and bars represent standard errors for pooled data.

Treatment of Ramos cells with the BU12-SAPORIN + rituximab combination consistently induced significantly more apoptosis when compared to treatment with BU12-SAPORIN or rituximab individually. By 48 h, over 60% of Ramos cells treated with the combination were apoptotic (early + late apoptotic; quadrants F2 + F4) compared with 37% treated with BU12-SAPORIN and 28% treated with rituximab alone. This was also paralleled by a more rapid and significantly greater loss of viable cells in combination treated Ramos cells compared with cells treated with each individual agent. The differences between the combination treated and individual reagent treated cells were highly significant (see Table I). Near identical results were obtained when the same experiment was conducted in HI serum indicating that complement was not involved in this process (data not shown).

Table I.   Results of analysis of statistical differences between rituximab and combination treated or BU12-SAPORIN and combination treated Ramos cells for apoptosis and viability.
QuadrantTime (h)P-value (independent t-test)
Rituximab versus combinationBU12-SAPORIN versus combination
  1. The quadrant refers to those shown in Fig 4A, where F3 represents viable cells, F2 represents secondary necrotic cells and F4 early apoptotic cells. Statistical analysis was carried out with data from six separate identical experiments as shown in Fig 4.P-values were obtained by the independent t-test.

  2. *Equal variances not assumed (Levene's test).

F3 viable24<0·001<0·001
F2 secondary necrotic240·0670·042
F4 early apoptotic240·0190·006
F2 + F4 secondary necrotic + early apoptotic240·004<0·001

We also studied two molecular determinants of apoptosis by analysing PARP cleavage and cleavage activation of caspase 3 in treated Ramos cells. PARP cleavage was readily detected in cells treated with BU12-SAPORIN and rituximab individually or in combination (Fig 5C). Consistent with the results obtained for annexin V staining, there was considerably more PARP cleavage detected in Ramos cells treated with the two drug combination compared with that seen in cells treated with each drug individually (Fig 5C). In the case of caspase 3 the difference between BU12-SAPORIN and combination-treated Ramos cells was minor, with only a slight increase in combination-treated cells (Fig 5E).

Figure 5.

 The effects of the caspase inhibitor z-VADfmk on apoptosis in Ramos cells treated with rituximab, BU12-SAPORIN or a combination of both drugs. Percentage of (A) early apoptotic cells and (B) viable cells observed following annexin V/PI staining 24 h after treatment of Ramos cells in the absence (□) and presence (bsl00001) of 50 μmol/l zVAD. Percentages represent the mean values obtained from three separate identical experiments and error bars standard errors. (C–F) Western blot analysis of anti-poly(ADPribose) polymerase and caspase 3 in Ramos cells treated with rituximab (lane 2), BU12-SAPORIN (lane 3) or a combination of both (lane 4) in the absence and presence of 50 μmol/l zVAD. Lane 1: untreated control cells.

The caspase inhibitor zVADfmk only partially inhibits apoptosis in drug combination treated Ramos cells

To determine the role of caspases in cell killing, we investigated the effects of the broad-spectrum caspase inhibitor zVADfmk on physical (annexin V/PI positivity) and molecular (PARP and caspase 3 cleavage) determinants of apoptosis in treated Ramos cells. Results of a representative experiment chosen from three independent identical experiments are shown in Fig 5A–F. zVAD almost completely blocked the appearance of early apoptotic cells due to BU12-SAPORIN alone (Fig 5A) though this result was not statistically significant when compared with non-zVAD treated control cells. zVAD had a smaller effect on rituximab-treated cells that was also non-significant. However, zVAD only partially blocked the appearance of early apoptotic cells in combination-treated cultures (Fig 5A). When cell viabilities were examined, zVAD gave a only a partial increase in the viability of Ramos cells treated with the two-drug combination but almost fully restored the viability of cells treated with BU12-SAPORIN alone (Fig 5B), although this latter finding was not statistically significant. In contrast, zVAD treatment of combination-treated Ramos cells increased viability of these cells significantly (P = 0·021) compared with combination-treated cells in the absence of zVAD.

In comparison with non-zVAD-treated cells (Fig 5C) PARP cleavage was completely inhibited by zVAD in rituximab-treated cells and almost completely in BU12-SAPORIN-treated cells (Fig 5D). However, PARP cleavage was only partially inhibited by zVAD in Ramos cells treated with the two-drug combination (Fig 5D). Similarly, zVAD abolished caspase 3 cleavage due to rituximab but only partially so for BU12-SAPORIN or for the combination of both drugs (Fig 5F).

The therapeutic activity of combination rituximab/BU12-SAPORIN in SCID-Ramos mice is superior to either individual drug used alone

Figure 6 shows representative survival curves from a total of three separately conducted similar experiments in SCID-Ramos mouse groups treated as indicated in the figure legend. Rituximab alone exerted a significant therapeutic effect compared with PBS sham-treated control animals (P = 0·0008) with 50% of animals in this group surviving disease-free for the 120-d duration of the study. In contrast, F(ab)2 rituximab had no significant therapeutic activity. BU12-SAPORIN used individually gave a significant prolongation in survival (P = 0·02) but all animals eventually succumbed with disseminated disease by day 75. When BU12-SAPORIN and rituximab were used to treat animals in combination, all of the animals survived the 120-d duration of the study disease free and the difference between the PBS sham-treated control group and this combination-treated group was highly significant (P < 0·00001) as was the difference between the BU12-SAPORIN and combination-treated groups (P < 0·001) and the rituximab versus combination-treated groups (P = 0·01). In contrast, when BU12-SAPORIN was used in combination with F(ab)2 rituximab the therapeutic effect was significantly reduced when compared with the group treated with a combination of intact rituximab + BU12-SAPORIN (P < 0·01) giving an intermediate therapeutic effect somewhere between that seen for treatment with intact rituximab + BU12-SAPORIN and treatment with BU12-SAPORIN alone (Fig 6). The therapeutic effect of a combination of BU12-SAPORIN + F(ab)2 rituximab was significantly less effective than the combination of BU12-SAPORIN + intact rituximab (P = 0·00002). Treatment with a combination of BU12-SAPORIN + F(ab)2 rituximab appeared to be more effective than treatment with BU12-SAPORIN alone, although the difference between these two therapy groups was not significant (P = 0·071).

Figure 6.

 Survival of severe combined immunodeficient mice that received various treatment regimes seven days after intravenous (i.v.) injection of 2 × 106 Ramos cells. Animals received treatment with a single i.v. injection of phosphate-buffered saline vehicle only (bsl00079) 10 μg BU12-Saporin (bsl00001), 10 μg intact rituximab (•), 11·67 μg of a F(ab)2 derivative of rituximab (○), a combination of 10 μg each rituximab + BU12-Saporin (bsl00066) or a combination of 10 μg BU12-Saporin + 11·67 μg F(ab)2 rituximab (bsl00084).

We next conducted an experiment to determine whether the simultaneous co-ligation of CD19 and CD20 by BU12 antibody and rituximab respectively, might be responsible for the therapeutic effect of the rituximab + BU12-SAPORIN combination treatment in the SCID-Ramos model. Representative results from two identical experiments are given as supplemental on line information (see Fig S2A). BU12 antibody had no significant therapeutic activity in SCID-Ramos mice, whereas the survival of rituximab-treated animals was significantly prolonged compared with PBS sham-treated control animals (P = 0·05). A combination of BU12 antibody + rituximab did not perform significantly better than treatment with rituximab alone.

In order to confirm the immunospecificity of the in vivo supra-additive therapeutic effect of combination treatment we conducted two separate studies. In the first study, we investigated an irrelevant anti-CD7 IT (HB2-SAPORIN) used alone or in combination with Rituximab in SCID-Ramos mice. HB2-SAPORIN had no therapeutic effect in SCID-Ramos mice despite being highly active against CD7+ human T-ALL cells in SCID mice (Morland et al, 1994). When HB2-SAPORIN was combined with rituximab the therapeutic outcome was no better than that for rituximab alone (see Fig S2B). In the second study, we examined the therapeutic effectiveness of combination treatment in SCID mice xenografted with the human B-lineage ALL cell line NALM-6 (SCID-NALM6 mice) that expresses CD20 at very low levels (see Fig 1B and C). In this study, rituximab on its own exerted only a very minor and non-significant therapeutic effect in SCID-NALM-6 mice. Combination treatment was no better than BU12-SAPORIN alone in SCID-NALM-6 mice (see Fig S2C).

The effects of complement or natural killer cell depletion of SCID-Ramos mice on the therapeutic effects of rituximab/BU12-SAPORIN combination treatment

We next conducted two separate experiments in an attempt to determine the in vivo contribution of ADCC and complement-mediated effects on the therapeutic activity of BU12-SAPORIN + Rituximab combination therapy.

In the first experiment, SCID-Ramos mice were depleted of functional complement activity with CVF prior to treatment and compared the therapy outcomes with those obtained in complement intact SCID-Ramos mice receiving the same therapy. Representative results of two separate experiments are shown in Fig 7A. CVF treatment alone did not affect the survival of SCID-Ramos sham-treated control mice. There was no major difference between complement-intact and -depleted SCID-Ramos mice treated with BU12-SAPORIN alone. However, complement depletion of SCID-Ramos mice reduced the mean survival times and overall survival rate in comparison with complement intact SCID-Ramos mice following treatment with rituximab (54 d vs. 12 d and 45% vs. 10% respectively) or with combination therapy (84 d vs. 54 d and 70% vs. 50% respectively). Although there was an apparent difference between the survival characteristics of these two therapy groups, this was not significant when analysed by log-rank analysis.

Figure 7.

 Effects of (A) complement or (B) natural killer (NK) cell depletion on the survival of severe combined immunodeficient-Ramos mice that received various treatment modalities seven days after intravenous (i.v.) injection of 2 × 106 Ramos cells. In (A) half of the animals were treated i.v. with 1 μg cobra venom factor (CVF) to deplete complement activity 24 h before treatment (see Materials and methods); the remaining half were sham-treated at the same time with phosphate-buffered (PBS). Experimental treatment groups were comprised of PBS + PBS sham-treated controls (bsl00072), CVF + PBS sham-treated controls (bsl00083), PBS + 10 μg BU12-SAPORIN (bsl00001), CVF + 10 μg BU12-SAPORIN (bsl00000), PBS + 10 μg rituximab (•), CVF + 10 μg rituximab (○), PBS + combination (bsl00066), CVF + combination (bsl00084). In (B) half of the animals were treated i.v. with 10 μg rabbit anti-asialo GM1antibody (anti-GM1) to deplete them of NK cells 24 h prior to receiving treatment (see Materials and methods), the other half received sham PBS treatment at the same time and were effectively NK cell-intact. Experimental treatment groups were comprised PBS + PBS sham-treated controls (bsl00072), anti-GM1 + PBS sham treated controls (bsl00083), PBS + 10 μg rituximab (•), anti-GM1 + 10 μg rituximab (○), PBS + combination (bsl00066), anti-GM1 + combination (bsl00084).

To determine the role of natural killer (NK) cells we depleted SCID-Ramos mice of these cells with anti-asialo GM1 antibody and then treated these animals with rituximab or a combination of both BU12-SAPORIN + rituximab. The results obtained for one representative experiment of two separate studies conducted are shown in Fig 7B. Control SCID-Ramos mice treated with anti-asialo GM1 antibody alone (i.e. NK cell-depleted) and then sham-treated with PBS showed a small but significant prolongation in survival compared with PBS sham-treated controls (P < 0·05). We have observed this effect consistently on several different occasions and paradoxically seems to infer that SCID NK cells may actually promote lymphoma growth and thus makes the interpretation of these results more difficult. A similar effect was apparent in rituximab-treated SCID-Ramos mice where NK cell-depleted animals actually survived longer than NK cell intact animals (Fig 7B). However, the difference between these two groups was not significant (P = 0·314). There was only a relatively minor improvement in survival of NK cell intact combination-treated animals over similarly treated NK-depleted animals but this difference was not significant. We conclude from these results that NK cell-mediated ADCC or other direct NK cell effects do not play any significant role in the therapeutic activity of rituximab used alone or in combination with BU12-SAPORIN in the SCID-Ramos model used here.


This study clearly demonstrated that rituximab augments the immunospecific cytotoxic activity of an anti-CD19 immunotoxin for the human B-cell lymphoma cell line Ramos. Used in combination, these two immunotherapeutic agents were shown to induce a significantly greater degree of apoptosis and PSI in this CD19+ CD20+ cell line in a complement-independent manner. Interestingly the action of these two immunotherapeutics was additive with respect to apoptosis induction but supra-additive or synergistic with respect to PSI in target Ramos cells. Rituximab has previously been shown to sensitise CD20 expressing lymphoma cell lines to a variety of cytotoxic drugs (Demidem et al, 1997), corticosteroids (Rose et al, 2002) and to a ricin A chain-based anti-CD22 immunotoxin (Ghetie et al, 2001) through mechanisms that are not entirely clear.

By employing two immunotherapeutic reagents directed against different B-lineage molecules on the target cell surface we have increased both the potency and fidelity of the immunospecific attack. This was demonstrable through lack of augmentation when an irrelevant anti-CD7 immunotoxin was used together with rituximab and through studies with NALM-6, a cell line sensitive to BU12-SAPORIN (Flavell et al, 1995) but lacking expression of CD20. Our study has also shown that the augmentive effect was not due to the additivity of two separate pro-apoptotic signalling pathways following simultaneous co-ligation of CD19 and CD20 by their cognate antibodies on the target cell surface.

A major goal of the present study was to determine which mechanisms are operative and the relative contribution that each might make to the improved therapeutic effect achieved with this combination immunotherapy. Various workers have shown that CDC (Harjunpaa et al, 2000; Bellosillo et al, 2001; Di Gaetano et al, 2003; Manches et al, 2003), ADCC (Voso et al, 2002; Manches et al, 2003) and pro-apoptotic direct cell signalling pathways (Pedersen et al, 2002) may all contribute to rituximab killing of lymphoma cells. The complement inhibitory molecules CD55 (DAF) and CD59 have been shown to protect lymphoma cells from complement-mediated lysis with rituximab (Golay et al, 2000; Harjunpaa et al, 2000) and patients with high expression levels of one or both of these molecules appear to be refractory to rituximab therapy (Manches et al, 2003). Approximately 15% of Ramos cells used in the present study expressed either molecule and the remaining 85% should therefore have remained fully susceptible to complement-mediated lysis. It is of particular interest to note that both bovine and murine complement were efficiently fixed by rituximab and were subsequently lytic for Ramos cells; other workers were unable to show this in a different model system (Hernandez-Ilizaliturri et al, 2003). Confocal microscopy also confirmed this, showing that rituximab was capable of fixing SCID mouse C1q on the cell surface leading to the eventual deposition of C9 during formation of the membrane attack complex. BU12 antibody did not fix C1q on the Ramos cell surface but did, however, give some deposition of C9. The significance of this C9 deposition in the absence of C1q fixation by BU12 antibody is unknown and may represent an artefact as it appeared to play no apparent role in bringing about the lysis of these cells. The demonstration that SCID mouse complement could provide a source of complement that could also be fixed by rituximab is an important observation as this validates the relevance of in vivo complement-mediated effects for the SCID mouse experiments described here. However, our results showed that complement did not appear to play any role in rituximab augmentation of the protein synthesis inhibitory activity of BU12-SAPORIN or on the increased apoptotic effect of combination treatment of Ramos cells in vitro.

Through the series of experiments described here we have partially delineated the mechanisms that are responsible for the augmentive effect that rituximab exerts on BU12-SAPORIN therapeutic activity. Our data indicate that there is probably more than one mechanism operating, possibly involving both direct cell signalling via CD20 and classical pathway complement-mediated effects. A further possibility that is not excluded by our data is that cross-linking of CD20 molecules to which rituximab is bound on the Ramos cell surface may be achieved via SCID mouse effector cells expressing Fcγ R and this may lead to an increased pro-apoptotic signalling event in vivo as suggested by the work of others (Shan et al, 1998; Mathas et al, 2000). Equally, some have shown that rituximab exerts its anti-tumour effect largely through complement-mediated mechanisms (Harjunpaa et al, 2000) rather than ADCC, although there is also some available data to suggest the opposite (Cartron et al, 2002; Golay et al, 2003). While CDC mediated by rituximab against B-cell tumours shows classical features of apoptosis, it appears to be a caspase-independent phenomenon (Bellosillo et al, 2001; Chan et al, 2003). In this regard it has been shown that segregation of the CD20 molecule into lipid rafts following ligation by cognate antibody is an essential step for complement-mediated lysis (Cragg et al, 2003) and that this process may be determined by the epitope specificity of the ligating antibody (Cragg & Glennie, 2004).

The in vivo studies reported here show that decomplemented SCID-Ramos mice have a reduced therapeutic outcome with the rituximab + BU12-SAPORIN combination to a level that is intermediate between the response for the combination and rituximab alone. Although this result was not significant, it is tempting to suggest that, while complement fixation by rituximab makes a contribution to the augmentation of BU12-SAPORIN in SCID-Ramos mice, there must be a least one other mechanism operating that is complement-independent. Our data do not convincingly demonstrate a contribution of NK cell-mediated ADCC to the therapeutic effect of rituximab or combination treatment of SCID-RAMOS mice. ADCC does positively contribute significantly to anti-CD7 antibody or immunotoxin therapy in a SCID mouse model of human T-ALL (Flavell et al, 1998) though this does not appear to be the case in the present SCID-Ramos model. It has recently been shown that SCID mouse neutrophils can mediate ADCC with rituximab in vivo against xenografts of the human lymphoma cell line Raji (Hernandez-Ilizaliturri et al, 2003). In the present study we only eliminated NK cells and so it is plausible that neutrophil-mediated ADCC was still active in our SCID-Ramos model and this might explain the failure to significantly influence the therapeutic effects of rituximab or combination treatment just by NK cell depletion. Paradoxically, in the present study, NK cell depletion resulted in an increased survival time for SCID-Ramos mice treated with rituximab compared with identically treated NK cell intact animals. This is probably explained by an additivity of rituximab treatment with the increased survival effect that anti-asialo GM1 antibody treatment has on SCID-Ramos mice (as seen in control animals).

In the present study, treatment of Ramos cells with rituximab or BU12-SAPORIN resulted in activation of caspase 3 and cleavage of PARP. The combination of rituximab + BU12-SAPORIN resulted in considerably more PARP cleavage in Ramos cells to that seen with either rituximab or BU12-SAPORIN used individually. This molecular finding is in accord with the significantly greater degree of apoptosis that is induced in these cells by combination treatment. The broad-spectrum caspase inhibitor zVAD completely inhibited caspase 3 activation and PARP cleavage following rituximab treatment, whilst almost completely inhibiting PARP cleavage due to BU12-SAPORIN. However, zVAD only partially inhibited PARP cleavage as a result of combination treatment. In accord with this, zVAD completely inhibited apoptosis of Ramos cells treated with rituximab or BU12-SAPORIN individually but only partially inhibited apoptosis in cells treated with the combination of both immunotherapeutics. Although these results should be considered in light of the poor cell permeability of small molecule caspase inhibitors, such as zVAD, these findings may indicate that apoptosis elicited by BU12-SAPORIN immunotoxin is entirely caspase-dependent while that due to combination treatment is only partially so. In a previous study, Keppler-Hafkemeyer et al (1998) showed that a pseudomonas toxin-based immunotoxin kills target MCF-7 breast cancer cells in a partially caspase-dependent manner and that following caspase inhibition, cell death still occurs, presumably due to direct PSI. Our results, however, are not completely in accord with this as BU12-SAPORIN–induced apoptosis appears to be completely inhibitable by zVAD. To resolve this issue, further studies will be needed over a longer time course to determine the survival levels of BU12-SAPORIN-treated cells in the presence of caspase inhibitors or by using gene silencing approaches (e.g. siRNA) to interfere with the expression of specific caspases.

It is now clear that ribosome-inactivating proteins, such as saporin, kill cells not just by inhibiting protein synthesis but also through caspase-dependent and -independent mechanisms initiated by the so called ribotoxic stress response that initiates various apoptotic pathways (Johnson et al, 2003; Narayanan et al, 2004, 2005 ). We hypothesise that diverse apoptotic pathways, activated independently by BU12-SAPORIN and rituximab in Ramos cells, act co-operatively on common and/or different death substrates with the net result of achieving a bigger cell kill and a consequent better therapeutic effect, as seen in our SCID-Ramos model. The fact that combination treatment appears to be only additive with respect to apoptosis induction but supra-additive with respect to PSI argues that these two phenomena are independent but that both probably contribute to the demise of the targeted cell. Whilst this study has provided an insight into some of the operative mechanisms, only further studies will uncover the precise molecular pathways involved and the ways in which these interact to exert their terminal effects in the target cell.


The authors thank the children's leukaemia research charity, Leukaemia Busters ( for supporting this work.