Update your browser to view this website correctly. Update my browser now
The antibody rituximab, which binds to the protein CD20 on the surface of B-cells, has been used to treat B-cell malignancies for several years. However, the molecular mechanisms underlying this treatment are not yet fully understood. One well established rituximab-induced mechanism, natural killer (NK) cell-mediated antibody-dependent cellular cytotoxicity (ADCC), has recently been described to involve the polarisation of bound rituximab and CD20 to one side of the B-cell. B-cells polarised this way were cleared more efficiently by NK-cells, which led us to further investigate the cellular events involved in the polarisation process. Using optical microscopy on rituximab-treated cells, we have found that the rituximab/CD20-rich, polarised side accumulated mitochondria and actin, whereas the nucleus was reorganised to the opposite side of the cell. Depleting actin via different methods correlated with a decrease in rituximab, mitochondria, and nucleus polarisation, suggesting polarisation to be actin-dependent, active process that triggers intracellular rearrangement. The influence of these intracellular rearrangements on the efficiency of NK-cell-mediated clearance of B-cell malignancies remains open for future investigation.
Rituximab was the first Food and Drug administration (FDA) approved monoclonal antibody for use in cancer therapy and is now used to treat some non-Hodgkin lymphomas and rheumatoid arthritis. Further off-label use in systemic lupus erythematosus, multiple sclerosis, autoimmune haemolytic anaemia, and graft versus host disease exemplifies the importance of rituximab in current medicine. Despite its long-standing use, its mechanism of action is not fully understood. Rituximab is binding to the protein CD20 on the surface of B-cells, and this seems to induce a combination of complement- dependent cellular cytotoxicity (CDCC), “direct signalling” induced apoptosis, and antibody-dependent cellular cytotoxicity (ADCC) to deplete malignant or autoreactive B-cells. Recent research into the molecular basis of one of these mechanisms, ADCC, found that rituximab and CD20 polarised to one side of the target B-cell upon rituximab binding. In contrast to other forms of antibody capping, this effect was found to be cross-linking and Fc-receptor-independent. Interestingly, malignant B-cells polarised in this way were more likely to be cleared by NK cells than B-cells with homogeneously bound rituximab.
Building on these findings, we set out to further investigate cellular changes upon rituximab polarisation since further understanding the underlying processes will ultimately allow harnessing this polarisation effect to increase rituximab treatment efficiency or to screen other monoclonal antibodies for triggering ADCC in various situations. We here aim to elucidate: (1) organelle repositioning upon rituximab-induced polarisation, and (2) involvement of the cortical actin cytoskeleton in rituximab-induced polarisation.
Rituximab polarisation induces asymmetric nucleus repositioning
We first validated rituximab-induced polarisation of B-cells by taking confocal images of Raji cells (an EBV-transformed Burkitt’s lymphoma B-cell line) treated with AlexaFluor 647-labelled rituximab (Rtx-Alexa647) (Fig. 1A). Nucleus staining with NucBlue showed that this polarisation in turn triggered asymmetric positioning of the nucleus (Fig. 1B). While 62% of Rtx-Alexa647- polarised Raji cells (n = 172) showed nuclear asymmetry, on Rtx-Alexa647 bound but non-polarised Raji cells only 20% (n = 168) showed nuclear asymmetry. Rtx-Alexa-647-untreated control cells showed nucleus polarisation only in 9% of all cases (n = 539) (Fig. 1B). Nuclear asymmetry is therefore linked to rituximab-triggered polarisation, rather than just rituximab binding.
Rituximab polarisation actively reorganises mitochondria
Nuclear movement likely affects other organelles. Since mitochondria plays a key role in apoptosis-related phenomena, we have investigated potential repositioning of this organelle. 3–color confocal imaging of Raji cells incubated with Rtx-Alexa647, NucBlue for nuclei staining, and MitroTracker Orange for mitochondria staining showed a clearly correlated reorganisation (Fig. 1C). While the nucleus polarised to the opposite side of the Rtx-Alexa647 cap, the mitochondria localised towards the Rtx-Alexa647 cap. A detailed analysis showed that 90±5% (n = 106) of rituximab-binding cells in which the nucleus was polarised also had their mitochondria polarised towards the rituximab-rich cap. In contrast, rituximab-polarised cells without asymmetric positioning of the nucleus were found to polarise their mitochondria in only 14±6% of all cases (n = 66). The few cells that were not polarised by rituximab, yet showing asymmetric nuclear positioning, showed a similar likelihood of mitochondrial polarisation as did cells with polarised organisation of rituximab and nucleus, i.e. polarisation of mitochondria is observed whenever the cell nucleus polarises, independent of rituximab capping. This suggests that rituximab triggers asymmetric positioning of the nucleus, and this subsequently induces polarisation of mitochondria. We further questioned whether Raji cells would show nucleus and mitochondrial prepolarisation which could favor rituximab binding and polarisation. Time-lapse 3–color confocal microscopy of mitochondria (MitoTracker) and nucleus (NucBlue) in Raji cells upon addition of Rtx-Alexa647 showed that rituximab first bound uniformly, then polarised towards one side of the target cell which subsequently led to the reorganisation of nucleus and mitochondria (Fig. 1D). This sequence of events, taking place over an hour, indicates that rituximab actively triggers the rearrangement of cellular organelles upon polarisation.
The resulting condensation of mitochondria and enhanced proximity between mitochondria and the plasma membrane may favor Fas-ligand or granzyme-mediated apoptosis. Such hypothesis is in line with reports of rituximab-polarised B-cells being cleared more efficiently by NK cells. This is open for future research which may test the dependency between the efficiency and time-course of NK-cell or complement-mediated lysis and mitochondria reorganisation.
Rituximab polarisation is actin-dependent and leads to actin reorganisation
The involvement of the actin cytoskeleton in these polarisation events is very likely, since actin is known to interact with both the plasma membrane and intracellular organelles. To investigate a possible correlation between rituximab binding and the actin cytoskeleton, we used two different actin polymerisation inhibitors, Latrunculin B (LatB) and Cytochalasin D (CytD). Confocal microscopy of Raji-cells treated with either LatB or CytD in the presence of Rtx-Alexa647 decreased both (i) rituximab binding (>6–fold decrease in cell-bound intensity, Fig. 1E) and (ii) rituximab polarisation (reduction from around 43% (n = 420) Rtx-polarised cells to 12±8% (n = 479) and 18±5% (n = 557) after LatB or CytD treatment, respectively; Fig. 1F). To confirm the role of actin in rituximab-induced polarisation, we moved to giant plasma membrane vesicles (GPMVs), which are derived from living cells through vesiculation agents. While GPMVs have similar membrane composition as their parent cells, they differ in lacking an intact actin cytoskeleton and cellular organelles. Confocal images of GPMVs derived from Raji-cells and treated with Rtx-Alexa647 showed binding of rituximab to the vesicles but no polarisation in any of the investigated vesicles (n = 589) (Fig. 1G). While these observations confirm the correlation between rituximab-induced polarisation of CD20 and functional actin cytoskeleton, one has to keep in mind that GPMVs differ from their parent cells in multiple aspects in addition to the absence of an actin cytoskeleton. Most importantly, GPMVs are in thermal equilibrium, which may alter membrane properties and subsequently the conformation of CD20 and thus binding kinetics of rituximab. We therefore investigated the reorganisation of the lipid PIP2 upon rituximab binding. Plasma membrane-integrated PIP2 is a strong binding partner of actin; it is known to play a major role in actin polymerisation as well as in tethering the cytoskeleton to the membrane; and it is therefore a candidate to orchestrate actin polarisation during CD20 capping. Dual–color confocal images of Rtx-Alexa647 and PH-PLC-GFP (a marker for PIP2) highlighted that PIP2 was enriched in the rituximab-rich side of the Raji cells, which suggests that actin is also presumably enriched in that region. Our observations suggest that rituximab-mediated CD20 polarisation is actively supported by the actin cytoskeleton and not a passive event within the plasma membrane. Polarisation may, for example, be mediated by actin-dependent signalling or actin-induced changes of the membrane organisation. Our findings are important in the light of recent chemotherapy approaches that target the actin cytoskeleton, especially tropomyosin fibres that form actin filaments. Disruption of the actin cytoskeleton in this therapy may thus counteract the impact of rituximab treatments, since our findings suggest that actin disruption correlates with decreased rituximab polarisation and consequently decreased NK-cell killing efficiency.
CD20-mediated, rituximab-induced polarisation of CD20 in B-cells leads to a repositioning of their nuclei towards the CD20-depleted and of mitochondria and actin towards the CD20-rich side. This active actin-dependent process augments effective NK-cell killing.
Experiments above were carried out using B-cell lines. The use of primary B-cells may provide further insights.
Here, we do not report the exact order of the polarisation of mitochondria and nucleus. They are clearly dependent, however, which one triggers the other cannot be seen from our observations. It is probable that the Rtx polarisation process is energy dependent and it triggers a closer positioning of mitochondria to the capped region of membrane. This in turns, might result in the asymmetric positioning of nucleus.
Given the role of mitochondria in apoptosis, it is plausible that mitochondrial accumulation towards one side of the cell may accelerate apoptosis. This is open for future research which may look at the efficiency and time-course of NK-cell or complement-mediated lysis of mitochondria in polarised versus non-polarised B-cells. Related antibodies may be screened for similar properties. Further research into actin organisation may provide insights on how to improve rituximab polarisation which may in turn improve clinical treatments.
100 µg rituximab (Invivogen, anti-hCD20-hIgG1) was labelled using Alexa Fluor® 647 Monoclonal Antibody Labelling Kit (Invitrogen) following the protocol provided. After that, labelled rituximab was added to Raji cells (ATCC CCL-86) and Jurkat cells separately to test for CD20 specificity. Different concentrations of rituximab were tested to determine the minimum concentration required to saturate the cell surface with rituximab via comparison of geometrical means of different fluorescence intensities.
Raji cell culture
Raji cells were maintained in 90% RPMI media, 10% FCS (fetal calf serum) supplemented with 1% L-glutamine at 37°C, 5% CO2. Cells were split every 2–3 days to keep them at appropriate confluence.
Labelling of cells
15 µL of cells (around 7×106 cells/mL) were put in 75 µL cell-complete media (described above). 10 µL (0.1 mg/mL) of labelled rituximab was added onto the cells in an Eppendorf tube. This mix was then incubated for 1 h at 37°C, 5% CO2. Simultaneously, µ-Slide 8 well glass-bottom imaging chambers (Ibidi) were coated with 1 mg/mL BSA for 30 min and washed twice with 200 µL phenol red-free L15 medium, leaving 200 µL L15 in each well. Following the 1 h antibody incubation, 500 µL L15 was added to dilute unbound antibody. Cells were subsequently spun down (2000 rpm, 5 min, “Eppendorf mini-spin” centrifuge), supernatant was discarded, and the pelleted cells taken up in either (i) 300 µL PBS and transferred into the coated microscopy well (removing the L15 from the well) for imaging of labelled antibody only; or (ii) 100 µL L15 for other experiments. In case (ii), depending on the experiment, one or more of the following reagents were added: nuclear staining- 1 drop NucBlue® Live cell stain (Life Technologies) and incubation at RT for 5 min; mitochondrial staining- 0.5 µL (1 µM) MitoTracker® Orange CMTMRos (Thermofisher, #M7510) during antibody incubation and incubation for 30 min at 37°C. Thereafter, the mix was transferred into the imaging chamber containing 200 µL L15 media. Subsequent imaging was carried out on a Zeiss LSM 780 confocal microscope. Note: L15 serum-free media may be used to minimise cross-reactivity of antibody with serum. However, this comes at the cost of cell health and increased cell death. We hence recommend using cell media. Note: BSA coating prevents cell attachment to the well bottom. However, cell attachment in BSA-uncoated wells did not influence the number of polarised cells.
Actin disruption with Latrunculin B or Cytochalasin D
Before labelling, 1 µL (0.1 mM) (final concentration 1 µM) of Latrunculin B or Cytochalasin D was added to 75 µL complete-cell media and rituximab was subsequently added right away or after 30 min at 37°C, 5% CO2.
Visualization of PIP2
In order to visualise PIP2, the PH-PLCD1 domain from the PH-PLCD1-GFP plasmid (Addgene plasmid # 51407) was cloned into the standard lentiviral vector Phr. PH-PLCD1 domain integrity was verified by sequencing. Subsequently lentivirus was obtained by transfecting HEK293T cells with pQ8.91 0.5 µg, pMD-G 0.5 µg, pHR-PH-GFP 0.5 µg, Genejuice (Merck) 4.5 µL, DMEM (Sigma-Aldrich) 150 µL, milliQ water 20 µL. 48 h after transfection, supernatant was harvested and spun down to remove particulates. 0.5 mL of viral suspension was then added to 1.5 mL of 70% confluent Raji suspension. After 3 days, localisation of PIP2 was imaged after labeling with Rtx-Alexa647, as described previously, using the Zeiss LSM 780 confocal microscope.
Preparation of GPMVs
Raji cells were washed twice with 1 mL of hypotonic GPMV buffer (50 mM NaCl, 2 mM CaCl2, 10 mM HEPES; adjusted to pH 7.4 with HCl or NaOH). Then, 1 mL isotonic GPMV buffer (150 mM NaCl, 2 mM CaCl2, 10 mM HEPES; adjusted to pH 7.4 with HCl or NaOH) containing the vesiculation agent N-ethylmalemide (NEM, 2 mM final concentration) was added to cells. The cells were transferred to a 35 mm petri dish and incubated at 37°C for 1 h. Most of the dead cells attach to the bottom of the dish, making the isolation of floating GPMVs easy. The buffer containing GPMVs was gently transferred to an Eppendorf tube. To get rid of the remaining dead cells, the solution was spun down at 1000 rpm for 2 min in an “Eppendorf minispin” centrifuge. The GPMV-enriched supernatant was then used for further experiments, similar to cells in earlier described experiments.
Presented column statistics was done using GraphPad Prism, comparing measurements for two conditions through unpaired t-test analysis. Error bars in graphs represent standard deviations.
Financial support was provided by the Wolfson Foundation, the Medical Research Council (MRC, grant number MC_UU_12010/unit programmes G0902418 and MC_UU_12025), MRC/BBSRC/ESPRC (grant number MR/K01577X/1), the Wellcome Trust (grant ref 104924/14/Z/14) and Oxford internal funding (John-Fell-Fund and EP Abraham Cephalosporin Trust Fund). ES is supported by EMBO Long Term (ALTF 636-2013) and Marie Curie Intra-European Fellowships (MEMBRANE DYNAMICS). TK and SB were funded by the MSc in Integrated Immunology program of the University of Oxford.
We would like to thank Enzo Cerundolo and Simon Davis for useful discussions and the Wolfson Imaging Centre (Christoffer Lagerholm, Esther Garcia and Dominic Waithe) for microscopy support.