CD20/TNFR1 dual-targeting antibody enhances lysosome rupture-mediated cell death in B cell lymphoma

Obinutuzumab is a therapeutic antibody for B cell non-Hodgkin’s Lymphoma (BNHL), which is a glyco-engineered anti-CD20 antibody with enhanced antibody-dependent cellular cytotoxicity (ADCC) and causes binding-induced direct cell death (DCD) through lysosome membrane permeabilization (LMP). Tumour necrosis factor receptor 1 (TNFR1), a pro-inflammatory death receptor, also evokes cell death, partly through lysosomal rupture. As both obinutuzumab- and TNFR1-induced cell deaths are mediated by LMP and combining TNFR1 and obinutuzumab can amplify LMP-mediated cell death, we made dual-targeting antibody for CD20 and TNFR1 to enhance DCD of obinutuzumab. Obinutuzumab treatment-induced CD20 and TNFR1 colocalisation, and TNFR1-overexpressing cells showed increased obinutuzumab-induced DCD. Two targeting modes, anti-CD20/TNFR1 bispecific antibodies (bsAbs), and obinutuzumab-TNFα fusion proteins (OBI-TNFαWT and OBI-TNFαMUT), were designed to cluster CD20 and TNFR1 on the plasma membrane. OBI-TNFαWT and OBI-TNFαMUT showed significantly enhanced LMP, DCD, and ADCC compared with that induced by obinutuzumab. TNFR1 expression is upregulated in many BNHL subtypes compared to that in normal B cells; OBI-TNFαMUT specifically increased DCD and ADCC in a B cell lymphoma cell line overexpressing TNFR1. Further, OBI-TNFαMUT blocked NF-κB activation in the presence of TNF-α, implying that it can antagonise the proliferative role of TNF-α in cancers. Our study suggests that dual targeting of CD20 and TNFR1 can be a new therapeutic strategy for improving BNHL treatment. The OBI-TNFαMUT fusion protein enhances DCD and ADCC and prevents the proliferating effect of TNFα signalling; therefore, it may provide precision treatment for patients with BNHL, especially those with upregulated TNFR1 expression.


Introduction
B cell non-Hodgkin's lymphomas (BNHLs) are clonal tumours of mature and immature B cells, which constitute the majority (80-85%) of non-Hodgkin's lymphomas (NHLs) [1,2]. According to the World Cancer Research Fund International, BNHL occurs in more than half a million people worldwide annually [3]. The development of anti-CD20 antibody rituximab in 1997 initiated the introduction of targeted immunotherapy for BNHL treatment and greatly improved the prognosis of patients with these diseases [4,5]. However, relapse and development of resistance to rituximab are frequent in almost all BNHL subtypes, including follicular lymphoma (FL), chronic lymphocytic leukaemia (CLL), and diffuse large B cell lymphoma (DLBCL) [6][7][8]. To address these issues, several generations of anti-CD20 antibodies and combination treatments with chemotherapy have been developed [9].
Obinutuzumab is a glyco-engineered type II anti-CD20 antibody that increases antibody-dependent cellular cytotoxicity (ADCC) by enhancing binding affinity to FcγRIII on immune effector cells [10][11][12]. Obinutuzumab has a modified elbow-hinge region with a distinct epitope compared with that of rituximab, which results in spatial alteration of the CD20-mAb complex [13,14]. Thus, it induces direct cell death (DCD) characterised by homotypic adhesion, lysosomal swelling and rupture, and plasma membrane damage. However, the molecular mechanism of DCD remains elusive [15,16]. While the involved molecular pathways remain unknown, lysosome membrane permeabilisation (LMP) is known to be a prerequisite event for obinutuzumab-triggered DCD and the release of lysosomal proteases, cathepsins, which play an essential role in cell death [15]. Various lysosomotropic agents, such as O-methyl-serine dodecyl amide hydrochloride, Leu-Leu-OMe, lipid metabolites (e.g. sphingosine and phosphatidic acid), and ROS, can induce LMP [17]. Interestingly, co-treatment with lysosomotropic reagents and therapeutic antibodies that trigger lysosomal cell death was found to significantly enhance CLL cell death in vitro, indicating that amplification of LMP-mediated cell death could markedly improve the efficacy of cancer immunotherapy [18,19]. As LMP-mediated cell death has the potential to bypass resistance mechanisms and eliminate malignant cells refractory to conventional chemotherapy or immunotherapy, many studies are currently underway to promote LMP-induced DCD for managing several cancer types, including BNHL [17,18].
Tumour necrosis factor receptor 1 (TNFR1) is a multifunctional receptor that induces inflammation, proliferation, and cell death in cancer cells [20,21]. Membrane TNFR1 expression is increased in the active lymph nodes of patients with CLL and stimulation of TNFR1 with TNF α enhanced NF-κB activity and viability of CLL cells, suggesting its involvement in aggressive CLL progression [22]. Further, a poorer prognosis has been reported in patients with DLBCL showing high TNFα and TNFR1 expression [23,24]. Previous studies have shown that the early steps of TNFR1-induced LMP are mediated by receptor clustering and endocytosis [25,26]. During lysosomal cell death via TNFR1-TNFα signalling, lysosomal ceramide is increased and subsequently converted to sphingosine, a lysosomotropic detergent, which induces lysosomal rupture and release of cathepsins into the cytoplasm, eventually resulting in cell death [27][28][29]. Interestingly, obinutuzumab-induced DCD is affected by TNFR1, as CD20 and TNFR1 are colocalised after obinutuzumab treatment, and obinutuzumab-induced DCD is decreased by TNFR1 knockdown [30]. This implies that artificial ligation of CD20 and TNFR1 can amplify obinutuzumab-induced DCD through LMP.
Dual-targeting approaches using bispecific antibodies (bsAbs) and fusion proteins have been applied to cluster membrane receptors to amplify the efficacy of monoclonal antibodies. An anti-PD1-GITR-L bispecific agonist induces PD1-dependent and FcγR-independent GITR clustering, resulting in enhanced activation, proliferation, and memory differentiation of primed antigen-specific GITR + PD1 + T cells [31]. Further, surrogate cytokine agonists can modulate receptor dimerisation geometries and ligand-receptor affinities to promote proximity and artificially activate WNT, IL-2/15, type I IFN, and IL-10 receptors [32,33]. Likewise, CD20 and TNFR1 dual-targeting proteins can cluster with these membrane proteins to amplify obinutuzumab-induced DCD. Dual-targeting antibodies have been developed by fusing TNFα with antigen-specific antibodies to activate TNFR1 or 2 in target cells with enhanced specificity. Sm3E-TNFα generated by the fusion of murine TNFα with scFv S6m3E, which targets carcinoembryonic antigen, preferentially accumulates in tumours with excellent cytotoxicity [34]. The risk of unintended effects by the administration of cytokines, such as TNF-α, could be countered by using the mutant form, mutant TNF-α, which successfully inhibits TNFα-mediated activation of TNFR1 as well as inhibited hepatic injury in an experimental hepatitis model [35]. Based on these results, dual targeting of TNFR1 and CD20 can be beneficial for BNHL treatment because of the additive effects of LMP induced by simultaneous binding to both antigens. Furthermore, dual-targeting proteins have increased selectivity for cancerous cells with high TNFR1 expression compared to that in normal B cells, thus reducing cross-reactivity.
In this study, we generated two types of novel fusion proteins: obinutuzumab-anti-TNFR1 bispecific antibody and obinutuzumab-cytokine proteins, wild-type TNFα or mutant TNFα, attached to the Fc region of obinutuzumab (OBI-TNFα WT or -TNFα MUT ) and examined their effect on DCD and ADCC in BNHL cells compared to those induced by obinutuzumab alone.

Stable cell line construction
Human MS4A1 and TNFRSF1A cDNA in pCNS vector were purchased from Korean human gene bank. The full cDNA sequences of MS4A1 and TNFRSF1A were cloned into the pLVX lentiviral vector with Xba1. TNFRSF1A Δ CD (lacking cytoplasmic domain (aa233-455)) was cloned into the pLVX lentiviral vector with Nhe1 and Not1. Primers CCC GCT AGC ATG GGC CTC TCC ACC and GGG CGG CCG CTT AGT AGA GCT TGG ACT TCC ACC were used. Lentiviruses were produced as previously described.
To generate HeLa cell stably overexpressing CD20 or TNFR1ΔCD and Raji cell stably overexpressing empty vector (EV) and TNFR1, HeLa and Raji cells were plated in 12-well plates with 10 μg/ml polybrene (TR-1003-G, Sigma-Aldrich) and incubated with virus-containing medium for 24 h. Viruses were removed and cells were supplemented with fresh medium containing 1 μg/ml puromycin (ant-pr-1, Invivogen) for 48 h. Protein expression was confirmed by flow cytometry and immunoblotting.

Cell binding assay
Ramos and HeLa-CD20, TNFR1 ΔCD cells were resuspended in PBS and incubated with the indicated dose of antibodies at 4 °C for 30 min. Cells were rinsed twice with PBS and incubated with a FITC-conjugated antihuman Ig Fc-specific secondary antibody (109-095-003, Jackson Laboratories) at 1:500 dilutions for 30 min. After being rinsed twice with cold PBS, a total of 10,000 cells are counted by flow cytometry (FACS, BD Biosciences, Franklin Lakes, NJ, USA) and analysed with FlowJo software.

Direct cell death assay
To measure antibody-induced direct cell death, 1 × 10 5 cells were resuspended in full media and then incubated with the indicated dose of antibodies (7, 21, and 70 nM) for 6 h. Then, cells were stained with PI (Propidium iodide) for 20 min in 37 °C. % of direct cell death PI + cells among 10,000 counted total cells were calculated by FACS Verse (FACS, BD Biosciences, Franklin Lakes, NJ, USA) and analysed by Flow Jo software.

Lysosomal membrane permeability assay
To measure lysosomal membrane permeability, cells were resuspended in full media and then incubated with the indicated dose of antibodies for 1 or 2 h. Then, cells were stained with lysotracker™ deep red (50uM) for 30 min. % of lysotracker positive cells among 10,000 counted total cells were calculated by FACS Verse (FACS, BD Biosciences, Franklin Lakes, NJ, USA) and analysed by Flow Jo software.

PBMC and primary B cell isolation
Peripheral blood mononuclear cells (PBMC) were purified from healthy donors who voluntarily participated in this study, for which informed consent was obtained to the study contents. All of these processes were conducted in accordance with the IRP procedure (#4-2016-0600) approved by the Yonsei University Institutional Review Committee. The blood is mixed with PBS in a 1:1 ratio and piled up on the tube where Ficoll (HISTOPAQUE-1077, Sigma, 10771) was put in advance. White blood cells and red blood cells were separated by centrifugation at 25 °C for 400 × g and 30 min, collected and transferred to a new tube. Cells were rinsed twice with PBS and centrifuged at 300 × g for 10 min at room temperature. This process was repeated twice to completely remove the platelets. Then, PBMC were resuspended in RPMI full media to be counted and used. Primary human B cells were isolated from PBMC using the B cell isolation kit II (130-091-151, Miltenyl Biotec) following the manufacturer's protocol.

Gene-expression data sets
The whole expression data set consisting of 873 biopsy specimens studied by means of Affymetrix HG-U133A GeneChip microarrays was collected as part of the German MMML consortium (Molecular Mechanisms in Malignant Lymphoma). The complete normalized data and follow-up information are available from 'The Leipzig Health Atlas' repository under accession number 7VT47TM4GV-1 (https:// www. health-atlas. de/ index. php/ en/ lha/ 7VT47 TM4GV-1). They divide into reference samples (tumour cell lines, sorted B cells, and tonsils), mature B cell lymphomas and other tumours collected in the study. One of the lymphoma specimens was measured twice on two arrays. Tumours were diagnosed in panel meetings of the MMML pathology group. Normal B cells were isolated from peripheral blood samples of healthy individuals. For isolation of the B cell subsets, FACS sorting employing antibodies against CD19 and IgD (naïve B cells), and CD19 and CD27 (post-GC memory B cells) was used.

Antibody-dependent cellular cytotoxicity assay
Ramos and Raji cells were washed with PBS and stained with 0.25 μM calcein-AM. Cells were incubated for 30 min and rinsed twice with RPMI full media. Purified PBMC (effector: target = 5: 1) were added and cells were treated with the indicated dose of antibodies (7, 70 nM) and incubated at 37 °C in 5% CO2 for 4 h. The % of FITC + cells among total 60,000 PBMC was calculated by FACS Verse (FACS, BD Biosciences, Franklin Lakes, NJ, USA) and analysed by FlowJo software. Normalization was done by ADCC % = 100 -(% of Treated/Untreated live cell population) *100.

TNFR1 signalling reporter assay
NF-κB promoter-dependent GFP reporter HEK293 cells were gifted from Prof. Jinu Lee (Yonsei University College of Pharmacy, Korea). Reporter cells were rinsed once with PBS, seeded 5 × 10 4 cells per well in a 96-well plate and incubated for 24 h. In the antibody dose-dependent TNFαinduced NF-κB signalling activation assay, cells were incubated with the indicated dose of antibodies (0, 0.1, 0.3, 1, 3, 10 nM) for 24 h. In the antibody-dependent TNFα-induced NF-κB signalling blockage assay, cells were pre-treated with the indicated dose of antibodies (70 nM) before 30 min, and then incubated with 20 ng/ml TNFα at 37 °C for 24 h. Cells were rinsed once with PBS and detached with Versene solution (15,040,066, Gibco Life Technologies™). Cells were rinsed with PBS and plated on a 96-well opaque plate. The fluorescence intensity of GFP per well was measured using a microplate reader (Flexstation 3, Molecular Devices) with an excitation/emission of 488 nm/520 nm. For normalization, cells were lysed with Triton X 2% and stained with PI staining. The fluorescence intensity of PI per well was measured using a microplate reader (Flexstation 3, Molecular Devices) with an excitation/emission of 518 nm/620 nm.

Statistical analysis
Statistical analyses were performed using Prism v9 (Graph-Pad Software). Data are presented as the means ± standard deviation as indicated in the figure legends. Statistical significance was assessed by Student's multiple unpaired t test. For Fig 4A, B Statistics were calculated using one-way ANOVA followed by Dunnett's multiple comparisons tests. Significance is indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.
Materials used in this study and methods related to each construct making were moved to the supplementary materials for limit the length of article.

The effect of TNFR1 in obinutuzumab-induced direct cell death
To assess the targetability of TNFR1 in combination with obinutuzumab, we first examined CD20 and TNFR1 localisation on the plasma membrane after obinutuzumab treatment. Immunocytochemistry was utilised to observe the movement of CD20 and TNFR1 on the cell membrane after obinutuzumab treatment. Before fixation, obinutuzumab-treated cells showed colocalisation of CD20 and TNFR1, but this colocalisation was not detected after fixation, implying that obinutuzumab binding enhances the proximity of CD20 and TNFR1 (Fig. 1A). Then, the total and surface expression patterns of TNFR1 in B cell lymphoma cell lines, Ramos and Raji, were significantly different, as detected by immunoblotting and flow cytometry, respectively ( Fig. 1B and  C). As we detected a different TNFR1 expression pattern in B cell lymphoma cell lines, we hypothesised that TNFR1 expression levels may affect obinutuzumab-induced DCD. To address this issue, we generated Raji cells that stably express an empty vector (EV) and TNFR1 using lentiviruses and confirmed that overexpression of TNFR1 does not affect CD20 expression in Raji cells by immunoblotting and flow cytometry ( Fig. 1D and E). Next, we treated Raji stable cell lines with obinutuzumab and observed increased obinutuzumab-induced DCD in TNFR1-overexpressing Raji cells (Fig. 1F). However, co-treatment with obinutuzumab and sTNFα did not induce additional DCD in TNFR1-overexpressing Raji cells, indicating that the presence of TNFα is not sufficient to enhance obinutuzumab-induced DCD (Figure S1). Overall, dual targeting of CD20 and TNFR1 by engineering obinutuzumab could bring the targets in proximity and, in turn, enhance obinutuzumab-induced DCD.

Generation of CD20 and TNFR1 dual-targeting bsAb, OBI-TNFα WT , and OBI-TNFα MUT
To bring CD20 and TNFR1 closer by targeting both antigens simultaneously, we first generated a bispecific antibody (bsAb) comprising obinutuzumab and anti-TNFR1 antibody using the 'knobs into holes' [36] and 'orthogonal Fab interfaces' technologies [37] (Fig. 2A, see supplementary information for mutations in detail). However, the bsAb was bound to CD20 to a lesser extent than that of the divalent obinutuzumab, most likely owing to the monovalent nature of bsAb [38] (Fig. 3B). To overcome the limitation of affinity to CD20, we generated OBI-TNFα WT , a fusion antibody with soluble TNFα WT attached to the Fc region of obinutuzumab. However, considering as co-treatment with TNFα did not enhance obinutuzumab-induced DCD ( Figure S1) and because TNFα can cause unwanted systemic side effects, we conjugated obinutuzumab to a soluble TNFα mutant (a TNFR1-selective antagonist, see supplementary information for mutations in detail) that specifically binds TNFR1 rather than TNFR2 and inhibits TNFα-mediated NF-κB signalling [35] (Fig. 2A). All engineered antibodies were produced from CHO-K1 cells and subjected to sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) to verify their stability and purity against rituximab (Mabthera), a commercial monoclonal antibody (mAb) control (Fig. 2B). To validate that these engineered antibodies bind to both targets, we evaluated the affinity of these antibodies using HeLa cells that stably overexpress CD20 and TNFR1ΔCD. OBI-TNFα WT and OBI-TNFα MUT were bound to HeLa-CD20 stable cells in a similar affinity to that of obinutuzumab, whereas the bsAb showed reduced affinity (Fig. 2C). Further, OBI-TNFα WT showed binding affinity comparable to that of anti-TNFR1 when bound to TNFR1-overexpressing HeLa cells, whereas bsAb and OBI-TNFα MUT showed slightly reduced affinity (Fig. 2D). Overall, these results confirmed that all the CD20 and TNFR1 dual-targeting antibodies were stable and showed specific binding to each target.

OBI-TNFα WT and OBI-TNFα MUT enhance DCD through LMP and augment ADCC in a TNFR1-dependent manner
The functional efficacy of the three antibodies against BNHL was evaluated by assessing their homotypic adhesion, binding affinity, DCD, and ADCC in Ramos cells expressing endogenous CD20 and TNFR1. As shown in Fig. 3A, similar levels of homotypic adhesion were detected among obinutuzumab, OBI-TNFα WT , and OBI-TNFα MUT . In cell binding experiments using various antibody concentrations, rituximab was bound to almost twice as much as obinutuzumab, consistent with previous data [14]. In the absence of CD20, Both cells were treated with antibodies (7 and 70 nM) for 6 h, stained with propidium iodide (PI), and analysed using flow cytometry. Data were collected from three independent experiments and plotted as mean ± SD. Statistical significance was assessed using Student's unpaired t test *p < 0.05, **p < 0.01, **p < 0.001 OBI-TNFα MUT showed lower binding affinity than that of OBI-TNFα WT (Fig. 2D). However, in the presence of excess CD20, both OBI-TNFα WT and OBI-TNFα MUT were bound to CD20 with equivalent affinity, implying that the affinity of the fusion proteins for their targets was wholly dependent on CD20 expression (Fig. 3B). We then tested antibody-induced DCD and found that bsAbs induced less DCD than that induced by obinutuzumab. OBI-TNFα WT and OBI-TNFα MUT induced greater DCD compared to that with obinutuzumab, implying that cell death was enhanced by targeting TNFR1, as the TNFα mutant can only bind TNFR1 and not TNFR2. Considering that obinutuzumab-induced DCD is LMP-dependent, we tested whether OBI-TNFα WT and OBI-TNFα MUT could enhance LMP. We treated Ramos cells with OBI-TNFα WT and OBI-TNFα MUT and stained them with Lysotracker deep red, an intracellular dye that accumulates in acidic organelles. Cells treated with OBI-TNFα WT and OBI-TNFα MUT showed decreased LysoTracker fluorescence, indicating an increased LMP (Fig. 3D, E). Next, we incubated Ramos cells with isolated human PBMCs at a ratio of 1:5 and treated them with engineered antibodies. Strikingly, ADCC induced by OBI-TNFα MUT was almost twofold higher than that induced by obinutuzumab (Fig. 3F). These results indicate that OBI-TNFα MUT induces greater LMP-dependent

OBI-TNFα MUT functions specifically in TNFR1-expressing B lymphoma
Based on previous results, we anticipated that OBI-TNFα MUT could improve the treatment of BNHL subtypes with high TNFR1 expression. We analysed TNFRSF1A and MS4A1 expression in BNHL subtypes compared to those in normal B cells using a gene-expression data set comprising 913 samples in Affymetrix HG-U133A Gene-Chip microarrays collected by the German MMML consortium (molecular mechanisms in malignant lymphoma) [39]. In these RNA expression data, most BNHL subtypes expressed higher levels of TNFRSF1A than those in normal B cells (Fig. 4A). In contrast, the B cell marker MS4A1 was highly expressed in most BNHL, including normal B cells, except for in multiple myeloma, which is known as the reduced CD20 expression [40] (Fig. 4B). We tested whether OBI-TNFα MUT specifically increased DCD, LMP, and ADCC in Raji cells that stably overexpressed TNFR1 (Fig. 1D and E). OBI-TNFα WT and OBI-TNFα MUT fusion proteins induced higher DCD than that induced by obinutuzumab alone in TNFR1 Raji cells (Fig. 4C). Interestingly, OBI-TNFα MUT induced higher LMP in TNFR1 Raji cells than in EV Raji cells (Fig. 4D),  (Fig. 1F). This suggests that a high therapeutic effect can be expected, even in aggressive BNHL with few immune cells, by DCD through LMP. Similar to the higher ADCC in Ramos cells (Fig. 3C), OBI-TNFα MUT significantly induced higher ADCC than that induced by obinutuzumab (Fig. 4E). Overall, along with enhanced DCD and ADCC, increased expression of TNFR1 in BNHL suggests that OBI-TNFα MUT is a well-suited and superior therapeutic agent for BHNL.

OBI-TNFα MUT does not activate NF-κB and blocks signalling by TNFα
As TNFα-mediated NF-κB signalling is known to have many inflammatory side effects, we evaluated NF-κB signalling activation using three antibodies [41]. The NF-kB reporter cell system driven by NF-κB promoter-dependent c, e GFP fluorescence in HEK293 reporter cells induced by TNF-α and antibodies was measured using a microplate reader with an excitation/emission wavelength of 488 nm/520 nm. For normalisation, the cells were lysed with 2% Triton X and stained with PI. The fluorescence intensity of PI was measured using a microplate reader at excitation/emission wavelengths of 518 nm/620 nm. Normalisation was performed using the mean GFP fluorescence intensity/PI mean fluorescence intensity of lysed cells. Statistical significance was assessed using Student's unpaired t test; *p < 0.05, **p < 0.01, ***p < 0.001 GFP fluorescence in HEK293 cells was used with TNF-α as a positive control (Fig. 5A). OBI-TNFα WT activated NF-κB signalling more than twice as much as that observed using TNFα because OBI-TNFα WT has a twofold higher amount of TNFα at the same molar concentration. In contrast, OBI-TNFα MUT did not activate NF-κB signalling at ≤ 3 nM but showed only slight activation at 10 nM. Other antibodies did not activate NF-κB signalling at concentrations up to 10 nM (Fig. 5B, C). In addition to the enhanced DCD and ADCC with OBI-TNFα MUT , we could expect additional benefits by blocking NF-κB signalling through TNFR1, as TNFα also facilitates B lymphoma proliferation [42,43]. Whether the antibodies could block NF-κB signalling by TNFα was also assessed using a reporter system (Fig. 5D). After pre-treatment with the antibodies, TNFα was added to HEK reporter cells. Enbrel was used as a positive condition to block TNFα and was found to completely block the signalling. Further, bsAb blocked NF-κB signalling less than anti-TNFR1, whereas OBI-TNFα MUT showed blocking comparable to that of anti-TNFR1 antibody (Fig. 5E, F). Other antibodies could not block NF-κB signalling activated by TNFα. Taken together, these results indicate that OBI-TNFα MUT does not induce NF-κB signalling and can block TNFα-mediated NF-κB signalling, implying fewer side effects and additional advantages.

Discussion
Antibody-mediated B cell depletion is one of the most effective therapies for BNHL. Although rituximab treatment has dramatically improved patient survival and prognosis, rituximab resistance and relapsed/refractory B cell lymphoma remain major concerns. To prevent the current and emerging issues, a novel anti-CD20 monoclonal antibody, obinutuzumab, was developed and exhibited superior efficacy compared to that of rituximab in affecting the B cell depletion rate when combined with chemotherapy [11,44]. The distinct features that enhance the efficacy of obinutuzumab include DCD triggered by LMP and an increase in ADCC by glyco-engineering [12]. As LMP-mediated cell death has the potential to bypass resistance mechanisms and eliminate malignant cells refractory to conventional treatments, including chemotherapy and immunotherapy, studies are underway to enhance LMP-induced DCD for improved efficacy in treating BNHL. We successfully developed a novel fusion protein, OBI-TNFα MUT , which amplified obinutuzumab-induced DCD through LMP and blocked NF-κB signalling induced by TNFα. As TNFR1 expression is upregulated in many BNHL subtypes compared to that in normal B cells (Fig. 4A), our OBI-TNFα MUT can exert improved therapeutic effects in some subsets of BNHL patients.
Understanding the mechanism of increased LMP and DCD in OBI-TNF TNFα MUT may suggest the application of TNFα MUT fusion to various antibodies. TNFR1 overexpression, but not TNFα itself, increased obinutuzumab-induced DCD (Fig. 1F, S1). OBI-TNFα MUT enhanced LMP, DCD, and ADCC as much as OBI-TNFα WT in Ramos and TNFR1overexpressing cells (Fig. 3, 4). These results imply that the degree of clustering in CD20 and TNFR1 by OBI-TNFα MUT , not NF-κB signalling, contributed to the increased LMP and DCD. The increased co-localization of CD20 and TNFR1 after obinutuzumab treatment shown in Fig. 1A also suggests the possibility that LMP-induced cell death could be enhanced by the aggregation of these proteins. When two proteins are artificially assembled, internalization of the two proteins can occur, which is one of the main mechanisms of action of bispecific antibodies [45,46]. However, endocytosis does not appear to be an LMP potentiation mechanism of OBI-TNFα MUT . To compare the amount and rate of endocytosis of each antibody, the degree of endocytosis according to the time after binding of each antibody was compared ( Figure S2). Contrary to our expectations, the three antibodies showed the same degree of fluorescence increase over time as for Ramos cells expressing endogenous TNFR1 or Raji cells overexpressing TNFR1, which indicates that the endocytosis degree of the three antibodies is almost the same. The mechanism by which OBI-TNFα MUT represents increased LMP is currently unclear from this study alone. However, considering the phenomenon that OBI-TNFα MUT is internalized to the same extent as obinutuzumab, it is presumed that this phenomenon is entirely due to the internalization of TNFR1. According to the previously reported, LMP is caused by the internalization of TNFR1 [25,28], and the sphingomyelinase (SMase) domain in the intracellular region of TNFR is important for LMP [25]. In addition, sphingosine, a metabolite of sphingomyelin (SM) abundant in the plasma membrane, but not in the lysosome [47], was identified as an LMP trigger induced by obinutuzumab [30]. In this context, since OBI-TNFα MUT forms a relatively largesized endosome derived from PM, a greater amount of SM can be introduced into the lysosome. In addition, the SMase region of TNFR1 delivered to the lysosome has the potential to use the introduced SM to produce more sphingosine. Further studies are needed to precisely elucidate the mechanism of enhanced LMP by OBI-TNFα MUT . This may provide a method to increase the therapeutic effect by accompanying LMP as a dual target of TNFR1 with other targets.
Based on the RNA expression data from 913 patients with BNHL, TNFR1 expression was higher in BNHL than in normal B cells (Fig. 4A). The remarkable incidence of high TNFR1 expression in BNHL, particularly poor prognosis, and aggressive progression, suggests that TNFR1 is 1 3 likely involved in the cancer progression [22,24]. Moreover, OBI-TNFα MUT enhanced cell death more than obinutuzumab only in malignant B cells (Fig. 4), but not in normal B cells ( Figure S3). This result suggests that OBI-TNFα MUT is superior to obinutuzumab in terms of efficiency and specificity in treating TNFR1 expressing BNHL. Furthermore, OBI-TNFα MUT showed enhanced DCD, as well as NF-κB signalling activation by TNFα (Fig. 5). Considering TNFα expression in DLBCL correlated with poorer overall survival and progression-free survival and that TNFα-induced stimulation of TNFR1 enhances NF-κB activity and CLL cell viability, inhibition of TNFα signals by OBI-TNFα MUT can provide additional therapeutic benefits in inhibiting cancer progression [22,23]. In particular, administering OBI-TNFα MUT to TNFR1-or TNFα-overexpressing patients can provide the advantage of precision medicine with high efficacy and fewer side effects. Currently, rituximab is used in the treatment of autoimmune diseases, and obinutuzumab has been assessed in clinical trials [48,49]. Given that the comparable DCD activity of OBI-TNFα MUT to that of obinutuzumab in purified primary B cells, OBI-TNFα MUT can be a suitable treatment option for autoimmune diseases as it provides the additional benefit regard to the blocking NF-κB signalling-mediated immune activation.
A new anti-cancer agent with stronger cancer cell killing effects without adverse effects is strongly needed for treating patients with cancer. Overall, this study presents a new type II anti-CD20 antibody fused with TNFα MUT , which shows improved cancer cell killing ability and suppresses cancer cell proliferation by inhibiting NF-κB signalling induced by TNFα. Moreover, it can be a suitable customised treatment for patients with BNHL showing increased TNFR1 expression, which is known to have a poor prognosis.

Conclusion
Our results suggest that dual targeting of CD20 and TNFR1 can be an effective therapeutic option for improving BNHL treatment and that the OBI-TNFα MUT fusion protein with enhanced DCD and ADCC can provide improved therapeutic options for patients with BNHL.