Rituximab induces Interleukin-6 production by human B cells

Abstract

Objective

Rituximab (RTX), an anti-CD20 monoclonal antibody, is highly effective in the treatment of several autoimmune diseases. The mechanism by which RTX treatment improves Rheumatoid Arthritis and ANCA-Associated Vasculitis is not easily related to B cell depletion. We have shown that RTX mediates a rapid stripping of CD20 and CD19 from the human B cell through a process known as trogocytosis. We hypothesized that changes in B cell phenotype resulting from trogocytosis would diminish the ability of B cells to promote autoimmune disease.

Methods

Human PBMC were cultured with RTX under conditions that permitted trogocytosis. Changes in B cell phenotype and cytokine production were measured under basal and activated (IL-4/anti-CD40) conditions. The effects of RTX were characterized for their requirements for FcγR and Fc-dependent interactions.

Results

Trogocytosis induced a marked loss of surface CD19, IgD, CD40 and BR3, but did not alter induction of CD86 expression on purified B cells by IL-4/anti-CD40 treatment. Unexpectedly, RTX-dependent trogocytosis of normal human B cells in vitro led to a rapid upregulation of IL-6 production, with no effect on TNFα, IL-1β, INFγ, or IL-10 production. This effect was Fc-dependent and required the presence of an FcγR bearing cell. This effect involved the release of pre-formed intracellular IL-6 protein as well as marked increases in IL-6 mRNA levels.

Conclusion

RTX mediated trogocytosis of B cells in vitro results in acute production and release of IL-6. The nature of this effect and its relationship to acute infusion reactions seen with RTX administration remain to be determined.

Rituximab (RTX) is a chimeric monoclonal antibody specific for human CD20. The utility of CD20 targeting by RTX has revolutionized the treatment of B cell malignancies (1-3). In addition, RTX has found wide application in many autoimmune disorders, such as anti-neutrophil cytoplasmic antibody (ANCA) associated vasculitis, autoimmune bullous diseases, and rheumatoid arthritis (RA) (4). RTX results in B cell depletion through several effector mechanisms such as antibody-dependent cellular cytotoxicity (ADCC), complement-dependent cytotoxicity (CDC), and induction of apoptosis (5). Interestingly, none of these mechanisms has been able to account for the incomplete relationship between B cell depletion and clinical response in RA (6-8). Thus, it seems reasonable to consider alternative effector functions of RTX that are not associated with B cell depletion (9-12).

In this regard, there is increasing appreciation of an alternative effector function of RTX, that of trogocytosis (13, 14). Trogocytosis refers to the process when one cell ‘gnaws’ or ‘shaves’ a portion of another cell's membrane following the creation of an immunologic synapse. This phenomenon has been described to occur in many cellular interactions, including antibody/FcγR engagement, T cell receptor/antigen presenting cell engagement, and B cell receptor/cell surface peptide engagement (13, 15-17). RTX mediated trogocytosis occurs when an FcγR bearing cell such as a monocyte engages the Fc portion of B cell-bound RTX and removes the RTX-CD20 complex, as well as a portion of the B cell membrane, without altering B cell viability (13, 14, 18, 19). The clinical significance of RTX mediated trogocytosis still is not fully understood, but holds considerable relevance as there is increased recognition of trogocytosis by several other antibody-based therapeutics, such as epratuzumab, daclizumab, trastuzumab, and cetuximab (20-22).

We previously identified that RTX exhibits little or no complement dependent cytotoxicity on normal human B cells (14). Instead, in the presence of RTX, phagocytes mediate rapid trogocytosis of CD19 and CD20 in the absence of cell death. Since the absence of CD19 and CD20 from B cells is associated with immunodeficiency (23, 24), this raised the possibility that RTX-dependent trogocytosis might alter the phenotype and function of B cells. In this model, following trogocytosis, functionally impaired B cells might result from their CD19- or CD20-deficiency states. We tested this hypothesis by characterizing phenotypic and functional responses of B cells following RTX-dependent trogocytosis. Instead of seeing functional impairment of B cells, RTX mediated trogocytosis of human B cells results in a selective and marked increase in production and release of IL-6. As we describe, the effect of RTX on IL-6 production was surprisingly specific both in terms of the cytokines being affected but also in its requirement for trogocytosis. In addition to identifying a novel pathway that triggers IL-6 production by B cells, this observation may be relevant to acute and chronic effects of RTX in shaping the immune response including infusion reactions (6, 7).

Materials and Methods

Cells

Blood was obtained from healthy volunteer donors following informed consent and PBMC purified by discontinuous gradient isolation using Ficoll-Paque PLUS (GE Healthcare Biosciences). Neutrophils (PMN) were isolated by dextran sedimentation from the blood pellet and erythrocytes lysed using BD PharmLyse RBC lysing buffer (BD Biosciences). After washing, PBMC and neutrophils were re-suspended in RPMI plus 10% heat-inactivated FBS serum (Hyclone). B cells were isolated by negative selection using Invitrogen's Untouched BCell Isolation Kit (Catalog #113.51d) as specified below. This protocol involves the addition of biotin-labeled antibodies targeting all non-B cells, followed by a wash, then addition of avidinbound magnetic beads. The non-B cells are then removed with a magnet, consistently resulting in B cell enrichment of >95%.

Antibodies & Reagents

FITC anti-CD45, APC anti-CD19, FITC anti-IgD and PE anti-CD27 were purchased from BD Biosciences. PE anti-IL-6, FITC anti-CD80, PE anti-CD86, PE anti-BR3, and PE anti-CD40 were purchased from eBiosciences. LEAF (low endotoxin, azide-free) anti-CD40 was purchased from BioLegend. Rituximab (Genentech) was obtained through the hospital pharmacy. Rituximab F(ab’)2 fragments were created by treating RTX with immobilized pepsin (Thermo Scientific) per manufacturer's instructions, followed by removal of Fc fragments using protein A Sepharose beads (GE Healthcare), and purity (>99%) was confirmed by SDSPAGE. Recombinant human IL-4 was purchased from eBiosciences. Imiquimod and CpG (ODN 2006) were purchased from InvivoGen. Carboxyflourescein succinimidyl ester (CellTrace CFSE) was purchased from Life Technologies.

RTX trogocytosis

The trogocytosis protocol was performed as previously described (14). For preparation of cells for FACS analysis, B cell isolation, and culture, the following trogocytosis protocol was utilized: PBMC (2 million cells/ml) in RPMI + 10% FBS were left untreated, or were treated with RTX or RTX F(ab’)2 10 μg/ml for 30 minutes at RT, followed by a wash. For FACS, the cells were washed in ice cold PBS/BSA/azide, pelleted (1500 rpm for 5 minutes at 4°C), incubated on ice for 30 minutes with fluorochrome labeled antibodies, then washed and fixed using 1% paraformaldehyde. Flow cytometry was performed using FACSCalibur (BD Biosciences) with analysis by FlowJo software (Tree Star). Initial gating of lymphocytes was based on side scatter and forward scatter properties, with subsequent quadrant or histogram analysis based on fluorochrome positivity. In one experiment, instead of FBS, 10% human serum was utilized, obtained after informed consent from healthy donors, seronegative RA patients, or high-rheumatoid factor seropositive RA patients.

Activation of B cells

Following the trogocytosis protocol (loss of CD19 was seen in each experiment) as noted above, B cells were isolated (>95%) by negative selection (total of 3 washes) and cultured (2×10⁶ /ml; RPMI-10% FBS) for specified times under basal or activated conditions: anti-CD40 (10 μg/ml) and IL-4 (1 ng/ml), imiquimod (5 μg/ml), or CpG (5μM). Activation was assessed by the induction of CD86 expression at 48 hours and demonstrated the superiority of anti-CD40/IL-4, hence its continued use.

Cytokine expression following trogocytosis

Following the trogocytosis protocol using RTX or RTX F(ab')2 as noted above, B cells were isolated by negative selection and cultured under basal or activated conditions. Supernatants were collected at the specified time, clarified by centrifugation (5’, 1500 rpm), and frozen at -80°C. Cytokine levels were measured by the Meso Scale Discovery (MSD) electrochemiluminescence platform using the MSD Human ProInflammatory 4-Plex I (IFN-γ, IL-1β, IL-6, and TNF-α), IL-6 Singleplex, or IL-10 Singleplex 96 well plates and processed according to the manufacturer's instructions. In confirmatory experiments, B cells were first isolated from PBMC using negative selection, labeled with CellTrace CFSE per manufacturer's instructions, then re-combined with PBMCs. Following trogocytosis, the B cells were isolated by cell sorting using FACSAria, then were cultured as above under activating conditions.

Surface and intracellular IL-6 were measured by flow cytometry using PE-labeled anti-IL-6. Intracellular IL-6 levels were assessed by light (10’, RT) fixation with PBS-0.01% paraformaldehyde, followed by permeabilization (15’, RT) in 0.5% Tween-20 in PBS, then incubation (30’, RT) with anti-IL-6 PE in PBS-0.1% Tween-20, followed by washing and fixation in 1% paraformaldehyde (protocol adapted from http://www.abcam.com /index.html?pageconfig=resource&rid=11448). In separate experiments, purified B cells and neutrophils were cultured (24h) in a 1:1 ratio in the presence or absence of RTX 10 μg/ml and assayed for IL-6 expression by flow cytometry with neutrophils excluded by side scatter gating.

IL-6 mRNA expression

Following specified treatments, total cellular RNA was extracted from enriched B cells as described by Gough (25). RNA was quantified using the NanoDrop spectrophotometer (NanoDrop Technologies) and reverse-transcribed (0.5 μg) using the SuperScript III Reverse Transcriptase (Invitrogen) with quantitative real-time PCR performed using an iCycler iQ Real-Time PCR detection system, using iQ SYBR Green Supermix (BioRad). Expression levels of IL-6 were quantified relative to the expression of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and actin housekeeping genes (26). IL-6 primers utilized (Eurofins MWG Operon):

forward 5’ CACAAGCGCCTTCGGTCCAGTTG 3’

reverse 5’ TGTCTGTGTGGGGCGGCTACATC 3’

Immobilized rituximab

In tests exploring the IL-6 inducing effects of CD20 cross-linking, RTX and RTX F(ab’)2, 10 μg/ml in PBS, were incubated (RT-2h) to immobilize to a 96 well flatbottom ELISA plate (BD Falcon), followed by extensive washing and blocking of non-specific binding with RPMI-10% FBS. RTX immobilization to ELISA wells was maximal under these conditions (Data not shown). Highly enriched B cells (without prior RTX treatment) were added to these wells for 24 hours at 37°C and 5% CO2, followed by supernatant measurement of IL-6 using the MSD IL-6 Singleplex.

Statistical analysis

Analysis was performed using STATA software version 12.1 (StataCorp). Student's t-tests were performed to calculate statistical differences between the means of the different variables. Two-tailed p values <0.05 were considered significant.

Results

Effect of RTX trogocytosis on B cell phenotype

In previous work, we have identified that treatment of human PBMC with RTX rapidly results in transfer of CD19 from the B cell to the FcγR bearing cell (monocyte or neutrophil) without resulting in B cell death (14). Using F(ab’)2 fragments, it was demonstrated that RTX mediated trogocytosis requires its Fc domain as well as an FcγR–expressing cell. As these experiments have been carried out in the presence of fetal bovine serum, we also evaluated the ability of monocytes to mediate trogocytosis of B cells in the presence of 10% human serum (Figure 1B shows representative results, with similar results obtained using cells from additional donors). Using identical conditions (30’ RT, RTX 10 μg/ml), we identified that neither the presence of IgG, nor the presence of rheumatoid factor impaired trogocytosis. We did find that one seronegative RA patient serum had decreased trogocytosis compared to the others for unclear reasons, though some degree of trogocytosis still occurred. We then surveyed for changes in functionally important membrane proteins found on B cells that might be removed by trogocytosis in addition to CD20 and CD19 (Figure 1A, C). Surface expression of CD40, BR3 (BAFF receptor), and IgD all declined significantly (>50%, p <0.05) along with previously described changes in CD19. In contrast, the expression of CD80 and CD86 was unaffected. These results imply that uniform loss of B cell surface proteins does not occur with RTX-mediated trogocytosis, but that several surface proteins are removed in addition to the RTX-CD20 complex.

Figure 1.

Rituximab modulates B cell phenotype. Healthy donor PBMC were untreated or treated with RTX for 30 min (hereafter referred to as the trogocytosis protocol – see text) followed by evaluation for loss of surface proteins. A, Representative flow cytometry histogram of gate on the lymphocyte population, showing change in CD19 and BR3 surface expression due to RTX mediated trogocytosis (black= no RTX, gray= post-RTX). B, Comparison of the use of human serum to fetal calf serum in the trogocytosis protocol. The presence of human IgG and/or RF did not affect ability of monocytes to mediate trogocytosis. Data shown representative of results obtained using cells from different donors. C, Compiled results evaluating change in surface expression of several B cell surface proteins (n=6; IgD measured in separate experiments, n=8). All results determined by FACS with initial gating on lymphocyte population using FSC/SSC, followed by quantification of specific fluorochrome positivity per 10,000 events. Asterisks indicate significance (p<0.05). Error bars represent SEM.

Effect of RTX on B cell activation and cytokine production

Based on the effect of RTXtrogocytosis on CD40 expression, responsiveness of purified B cells to stimulation with anti-CD40/IL-4 was analyzed. Purified B cells were obtained by negative selection from PBMC following incubation without or with RTX. RTX was removed by washing prior to negative selection. CD86 induction was used as a marker of activation since RTX treatment had no effects on its expression through shaving. While anti-CD40/IL-4 increased CD86 expression in untreated B cells, treatment with RTX did not alter that effect despite reductions in CD19, CD20 and CD40 expression by ∼50% (Figure 2A).

Figure 2.

RTX trogocytosis promotes B cell production of IL-6. A, Following incubation with or without RTX (10 ug/ml, 30’ RT) (e.g. ‘trogocytosis protocol’, see Figure 1) B cells were purified from PBMC by negative selection and cultured for 48h. Using CD86 as a marker of activation, anti-CD40/IL-4 shows significant change over unstimulated (p=0.007, n=3); no effect of RTX was seen under either set of conditions. B, The role of an intact Fc domain was obtained using the trogocytosis protocol incubating PBMC without or with RTX or RTX F(ab’)2. Purified B cells were cultured under resting and activated conditions for 24h and supernatants analyzed for IL-6 levels (n=6). Marked differences were seen with RTX relative to controls or treatment with RTX F(ab’)2 under resting (p=0.03) or activated conditions (p=0.02). RTX treatment had no effect on IL-1β, IFNg, TNFa, and IL-10 levels (data not shown). C, A similar experiment was performed by separating CFSE-labeled B cells by flow-cytometric sorting (instead of by negative selection) following trogocytosis, with nearly identical results (n=3, p=0.04). Error bars represent SEM.

Analysis of purified B cell supernatants at 24 hours demonstrated that anti-CD40/IL-4 activation increased the levels of IL-6, IFNγ, IL-1β, TNFα, and IL-10 2-6-fold relative to basal conditions.. Conditions under which RTX-dependent trogocytosis occurred had no effect on the levels of IFNγ, IL-1β, TNFα, and IL-10 (data not shown). In contrast, IL-6 levels were markedly elevated in supernatants of B cells purified from PBMC treated with RTX for 30’. This effect was evident under both basal and activated conditions (Figure 2B, p = 0.03 and 0.02, respectively). This effect of RTX required an intact antibody, as incubation of PBMC with RTX F(ab’)2 fragments had no effect (p = 0.97 with resting conditions). To ensure that the technical aspects of negative selection of B cells following trogocytosis did not confound the results, we performed additional experiments wherein B cells were first isolated, labeled with CellTrace CFSE and then recombined with PBMC and treated with RTX. Following trogocytosis, CFSE+ B cells were isolated by cell sorting using FACSAria, and cultured for 24 hours under activating conditions. This alternate form of B cell isolation following trogocytosis gave nearly identical increases in IL-6 production (Figure 2C).

Characterization of the mechanism of increased IL-6 release by human B cells

Flow cytometry analysis showed that a majority of healthy human B cells have intracellular stores of IL-6. In contrast, only ∼5% circulating B cells expressed surface IL-6, presumably representing either IL-6 bound to its receptor or in the process of being secreted at the time of paraformaldehyde fixation (27, 28). Treatment of PBMC with RTX but not RTX F(ab’)2 fragments, followed by B cell purification decreased intracellular IL-6 by approximately 40% and increased surface IL-6 expression by roughly 4.5-fold (Figure 3, n=6). Time course studies demonstrated that the maximal effect on surface and intracellular IL-6 occurred at 4 hours following trogocytosis and remained constant for 24 hours (data not shown).

Figure 3.

Rituximab, but not RTX F(ab’)2, modulates intracellular and surface IL-6 expression. IL-6 was measured at 24 hours by flow cytometry both (A) intracellularly and (B) on the surface of resting B cells purified by negative selection from PBMC following treatment with either RTX or RTX F(ab’)2 (10 ug/ml, 30’ RT) or without RTX as control (n=6). Data is shown as (A) percent of control or (B) fold-change over control. Intracellular IL-6 levels decreased by nearly 40% while there was a 4.5 fold increase in surface IL-6 expression. Error bars represent SEM.

RTX upregulates IL-6 mRNA levels

To further define this observation and its requirement for trogocytosis, PBMC were incubated with RTX or RTX F(ab’)2 fragments for 30’, the B cells purified by negative selection and cultured for 24 hours under basal and stimulated conditions. Total cellular RNA was isolated and IL-6 levels measured by qRT-PCR. As seen with flow cytometry, the cells treated with RTX, but not RTX F(ab’)2, induced IL-6 mRNA expression over that seen under basal conditions, although the magnitude varied between donors (20-440-fold increase) (Figure 4A). Strikingly, this effect persisted even under activated (anti-CD40/IL-4) conditions which increased the levels of IL-6 mRNA from 43-160-fold in the untreated controls, but as much as >1000-fold higher with RTX treatment (Figure 4B).

Figure 4.

Rituximab induces a rise in total cellular IL-6 mRNA transcripts. Following the trogocytosis protocol with RTX F(ab’)2 or RTX, B cells were isolated and cultured (24h) under (A) resting or (B) activated (IL-4/anti-CD40) conditions followed by RNA isolation. Quantification of IL-6 mRNA transcript levels was by qRT-PCR. Data points represent IL-6 mRNA fold difference over resting, untreated cells; lines link values for individual donors. Consistent increases in IL-6 mRNA levels as a function of RTX, but not RTX F(ab’)2, treatment were seen (n=7). Activation increased the levels of IL-6 transcripts 22-160 fold over that seen under basal conditions in untreated and RTX F(ab’)2, while a much greater increase was seen with RTX treatment, with 20-440 and 89-1209 fold increase over control in resting and activated conditions, respectively.

RTX effect on IL-6 release requires an intact Fc portion and FcγR bearing cells

The ability of a brief (30’ RT) incubation of PBMC with RTX, but not RTX F(ab’)2 fragments, to increase IL-6 mRNA and protein release by B cells indicates that CD20 crosslinking alone does not induce this effect. These data instead suggest a requirement for the Fc portion of RTX, perhaps to enhance CD20 cross-linking of RTX through engagement of FcγR. Alternatively, engagement of RTX by FcγR-bearing cells in PBMC (monocytes, NK cells) might facilitate an intercellular interaction that results in IL-6 release by B cells. To address these issues, we first observed that soluble RTX had no effect on IL-6 production when added directly to purified B cells (data not shown). Subsequently, we demonstrated that incubation of purified B cells with immobilized RTX or RTX F(ab’)2 also lacked this activity (Figure 5A). Thus, immobilized RTX antibody cross-linking of CD20 alone (and therefore enhanced CD20 clustering) was insufficient to induce IL-6 production. Finally, we examined the effect of a distinct FcγR-bearing cell not found in PBMC, the neutrophil, which can mediate trogocytosis (14), to modulate IL-6 production in a RTX-dependent manner. Purified B cells and neutrophils were combined in a 1:1 ratio in 10% FBS and were treated ±RTX 10 μg/ml (24h) (Figure 5B). Changes in B cell expression of IL-6 were only seen in the presence of RTX and neutrophils, supporting the notion that trogocytosis was the primary requirement for inducing IL-6.

Figure 5.

IL-6 induction by B cells requires trogocytosis. A, B cells were purified from PBMC by negative selection and cultured for 24 hours alone (none) or in the presence of either immobilized (‘imm’) RTX or RTX F(ab’)2 under resting or activated (IL-4/anti-CD40) conditions. Addition of RTX to purified B cells had no effect (Data not shown). B cells purified from RTX-treated PBMC (‘RTX’) are used as a positive control; supernatant IL-6 production was measured (n=3). B, Isolated B cells were combined with isolated neutrophils and treated ±RTX 10 μg/ml. After 24 hours incubation, surface and intracellular IL-6 of B cells was measured by FACS, with gating excluding neutrophils based on side scatter. Representative histogram showing an increase in surface IL-6 and a decrease in intracellular IL-6 in RTX treated (gray) compared to untreated (black) cells, indicating trogocytosis as a likely inducer of IL-6 (n=3).

Discussion

We report the completely unexpected finding that a brief exposure of a human B cell to RTX in the presence of an FcγR-bearing cell triggers IL-6 release and biosynthesis. This induction of IL-6 release is remarkable not just for its apparent requirement for trogocytosis, but also its rapidity (occurring in <4 hours) and cytokine selectivity (no change in IFNγ, IL-1β, TNFα, or IL-10 production). Finally, the magnitude of this effect is substantial if not robust; it did not require continued exposure to RTX and was equally apparent under both basal and activated (IL-4/anti-CD40) conditions.

Admittedly, an extensive measure of the effect of RTX exposure on all cytokines potentially elaborated by human B cells was not pursued. Rather, we focused on trying to understand the mechanism by which this effect might be occurring. Neither soluble nor immobilized RTX to purified B cells was capable of triggering this biologic activity. Thus, higher order crosslinking of CD20 is insufficient as is the presence of a purified phagocyte population in the absence of RTX (Figure 5). These latter experiments establish that the removal of RTX from the culture is not required for this effect to be seen. We conclude that IL-6 production and release result from the interaction of the Fc portion of RTX with a functionally competent FcγR-bearing cell while bound to CD20 on B cells. The rapidity and reproducibility of this response suggest that trogocytosis itself results in induction of IL-6 release, although it does not establish whether ‘shaving’ of CD20 (or another surface protein including CD19) is specifically responsible for the IL-6 release. Delineating this latter issue is technically formidable using normal human B cells and serves as direction for future research.

Nevertheless, these findings prompt consideration of CD20 trogocytosis as the key event in IL-6 induction by RTX. The exact function of CD20 is unknown, though it is thought to play a role in optimizing B cell immune response in a T-cell independent manner (29). When engaged by RTX, CD20 is rapidly incorporated into membrane lipid rafts, where it is thought to contribute to cell signaling by inducing a flux of intracellular cytosolic calcium (30, 31). A study by Walshe et al. suggests that CD20 induced calcium change is dependent on BCR signaling and is halted with inhibition of Syk, Src, and PI3K kinases (32). As stated above, whereas we cannot unequivocally state that this effect is directly mediated by CD20 trogocytosis by RTX, the specificity for IL-6 relative to IFNγ, IL-1β, TNFα, or IL-10 is intriguing.

The clinical relevance of these findings is uncertain. We have identified that RTX mediated trogocytosis occurs in the presence of human serum in vitro (Figure 1B), but it remains to be identified if trogocytosis occurs in vivo. While it is clear that there is widespread B cell death following RTX infusion, this does not eliminate the potential for trogocytosis also to occur. There is in vivo evidence of RTX mediated trogocytosis occurring both in mice and in patients with CLL (33-35). Further, kinetic studies have identified that trogocytosis occurs more rapidly than cell mediated cytotoxicity, indicating that trogocytosis may precede B cell death in patients receiving RTX for autoimmune diseases (36).

In additional support for the notion that RTX mediated trogocytosis leads to acute IL-6 release in vivo, Agarwal et al. evaluated cytokine expression before and one hour after rituximab infusion in patients with renal failure, and found that the only consistent change in cytokine levels was an acute rise in IL-6 in eight of nine patients (37). Interestingly, the kinetics of this effect on serum IL-6 levels parallel those seen for trogocytosis following RTX administration in vivo (38). This selectivity may be different from the cytokine release syndrome seen with RTX administration to patients with B cell malignancies in which tumor cell lysis is presumed to play a role (39). Further studies to evaluate this acute effect of RTX on IL-6 levels and its relationship to infusion reactions in vivo are on-going. Understanding RTX-dependent trogocytosis is important not only for treatment of autoimmune disease but also for the basic biology of B cell function.