Obinutuzumab (GA101) vs. rituximab significantly enhances cell death, antibody-dependent cytotoxicity and improves overall survival against CD20+ primary mediastinal B-cell lymphoma (PMBL) in a xenograft NOD-scid IL2Rgnull (NSG) mouse model: a potential targeted agent in the treatment of PMBL

Primary mediastinal large B-cell lymphoma (PMBL), a distinct mature B-cell lymphoma, expresses CD20 and has recently been successfully treated with the combination of a type I anti-CD20 monoclonal antibody, rituximab, with multiple combination chemotherapy regimens. Obinutuzumab is a glycoengineered type II anti-CD20 monoclonal antibody (mAb), recognizing a unique CD20 extracellular membrane epitope with enhanced antibody dependent cellular cytotoxicity (ADCC) vs rituximab. We hypothesize that obinutuzumab vs rituximab will significantly enhance in-vitro and in-vivo cytotoxicity against PMBL. PMBL cells were treated with equal dose of obinutuzumab and rituximab for 24 hours (1–100 μg/ml). ADCC were performed with ex-vivo expanded natural killer cells at 10:1 E: T ratio. Mice were xenografted with intravenous injections of luciferase expressing Karpas1106P cells and treated every 7 days for 8 weeks. Tumor burden was monitored by IVIS spectrum system. Compared with rituximab, obinutuzumab significantly inhibited PMBL cell proliferation (p = 0.01), promoted apoptosis (p = 0.05) and enhanced ADCC (p = 0.0002) against PMBL. Similarly, in PMBL xenografted NOD scid gamma mice, obinutuzumab significantly enhanced survival than rituximab when treated with equal doses (p = 0.05). Taken together our results suggest that obinutuzumab significantly enhanced natural killer cytotoxicity, reduced PMBL proliferation and prolonged the overall survival in humanized PMBL xenografted NOD scid gamma mice.


INTRODUCTION
Primary mediastinal large B-cell lymphoma (PMBL) is a rare form of non-Hodgkin lymphoma representing < 5% of mature B-cell NHL (B-NHL) [1,2]. The immunophenotypic features and molecular characteristics distinguish PMBL from other types of aggressive B-NHL and large B cell lymphoma [3,4]. PMBL shares one third of its genes with Hodgkin lymphoma [3]. PMBL is classified as midway in the biologic spectrum between diffuse large B-cell lymphoma (DLBCL) and classical Hodgkin lymphoma [3] with overexpression of genes in signaling pathways of NF-kB and Janus kinase (JAK)/signal transducer and transcription (STAT), high expression of CD20, downregulation of major histocompatibility class I and II molecules and overexpression of programmed death ligands [2,[5][6][7][8][9].
Obinutuzumab is a glycoengineered humanized monoclonal antibody (mAb) recognizing a unique CD20 type II epitope [10] with significant antibody dependent cellular cytotoxicity (ADCC) enhanced as compared to rituximab in-vitro and in-vivo [11][12][13][14]. The anti-tumor effects of obinutuzumab alone or in combination with other agents were further investigated in clinical trials. The safety and efficacy of obinutuzumab was compared with rituximab in relapsed indolent lymphoma in the randomized phase II trial (GAUSS) (NCT00576758) [15]. Among patients with follicular lymphoma (FL), obinutuzumab demonstrated a higher overall response rate than rituximab (44.6% v 33.3%; P = .08) but with no difference in progression-free survival (PFS) between the two arms [15]. The phase III GALLIUM trial (NCT01332968) and GADOLIN trial (NCT01059630) were conducted to treat previously untreated FL patients or patients with rituximab-refractory indolent non-Hodgkin lymphoma utilizing obinutuzumab combined with chemotherapy [16,17]. Obinutuzumab-based therapy significantly reduced the risk of progression or death and prolonged overall survival (OS) as compared to rituximab-based therapy or chemotherapy [16,17]. Obinutuzumab plus chlorambucil prolonged the OS or PFS and resulted in higher rates of complete response in patients with chronic lymphocytic leukemia (CLL) or coexisting conditions as compared to chlorambucil alone, or rituximab plus chlorambucil, respectively in the CLL11 clinical trial (NCT01010061b) [18]. Furthermore, the phase III iLLUMINATE trial (NCT02264574) demonstrated that obinutuzumab plus Ibrutinib is an efficacious combination therapy for previously untreated patients with CLL or small lymphocytic lymphoma [19]. Based on these exciting results, obinutuzumab in combination with chemotherapy has been approved for the treatment of untreated and rituximab refractory FL [16,17] and CLL [19]. Unfortunately, the clinical results of obinutuzumab for patients with DLBCL were not promising. Obinutuzumab was not superior to rituximab when combined with chemotherapies in patients with DLBCL shown in the phase III GOYA trial (NCT01287741) and the GAINED trial (NCT01659099) [20][21][22]. Additionally, the pre-clinical and clinical efficacy of obinutuzumab compared to rituximab in patients with PMBL is currently unknown.
We hypothesize that obinutuzumab would be a superior anti-CD20 antibody in the treatment of PMBL by inducing targeted programmed cell death and enhancing immune cell mediated ADCC compared to rituximab. In this study, we report the in-vitro and in-vivo efficacy of obinutuzumab against PMBL cell lines and in human PMBL xenografted immunodeficient NOD scid gamma (NSG) mouse model compared to rituximab.

RESULTS
Expression of CD20 mRNA and protein in obinutuzumab treated PMBL CD20 mRNA and protein expression in Karpas-1106P were measured by real time quantitative reverse transcription polymerase chain reaction and immunoblotting prior to any anti-CD20 treatment. Karpas-1106P showed a significant increase in the expression of both CD20 mRNA and protein ( Figure 1A and 1B) compared to Burkitt lymphoma (BL) (Raji) and Hodgkin lymphoma (HDLM-2) cell lines.

Obinutuzumab effects on the STAT6/NF-kB/ PI3K/AKT pathway
Constitutive STAT6 activation is identified as a key feature of PMBL from other lymphoma and the NF-κB signaling pathway and the PI3K/AKT pathway are also reported to be activated in PMBL [2,23]. Based on this known information, we firstly examined if obinutuzumab could potentially inhibit these pathways by decreasing the phosphorylations of some key members of these pathways. Karpas-1106P cells were treated with obinutuzumab (10 μg/ml) for 48 hours and the phosphorylation levels of AKT, STAT6 and IκBα were measured by Western blotting. There was a significant decrease of p-STAT6 (treated vs control: 0.725 ± 0.0065, p < 0.05) in the JAK/STAT pathway, p-AKT (treated vs control: 0.69 ± 0.0028, p < 0.005) in the PI3K/AKT pathway and p-IκBα (treated vs control: 0.706 ± 0.009, p < 0.005) in the NF-κB pathway in obinutuzumab treated Karpas 1106P cells compared to PBS controls ( Figure 4A and 4B). There were no significant differences of total STAT6 and AKT proteins between the treated and the controls (treated vs control: 0.978 ± 0.0071, p > 0.05; 0.976 ± 0.011, p > 0.05; respectively) ( Figure 4B). Total IκBα protein level was significantly reduced in obinutuzumab treated Karpas-1106P cells compared to PBS controls (treated vs control: 0.682 ± 0.0066, p < 0.005) ( Figure 4B).

Gene expression changes in obinutuzumab treated PMBL cells
To identify other biological processes affected by obinutuzumab, we investigated gene expression changes in obinutuzumab treated PMBL cells. Genomics employing Illumina microarray technology in obinutuzumab treated Karpas-1106P PMBL cells were compared to PBS or IgG isotype control. A total of 65 differentially expressed genes (0.3% of all genes) were identified (>1.5-fold) from 21,381 human genes ( Figure 5). Among the 65 filtered genes, 38 genes (58.5%) were up-regulated and 27 (41.5%) genes including ID3 and RAB6B were down-regulated. When we performed gene set enrichment analysis (GSEA) to identify the biological processes affected by the obinutuzumab treatment, we observed 19 gene sets were enriched to the obinutuzumab treatment while 8 were enriched to the IgG treatment (Supplementary Table 1) and 9 were enriched after PBS treatment (Supplementary Table 2), respectively. Gene sets enriched in the obinutuzumab treatment group included IL6-JAK-STAT6 signaling, Hedgehog signaling, interferon gamma signaling, apoptosis and MTORc1 signaling pathways ( Supplementary Tables 1 and 2).

DISCUSSION
We have demonstrated significant in-vivo and in-vitro effects of obinutuzumab against PMBL. Specifically, we describe superior efficacy of obinutuzumab vs. rituximab (10 μg/ml) against Karpas 1106-P PMBL cells, including significantly decreased cell viability, increased cell death and enhanced ADCC. In addition, PMBL xenografted NSG mice treated with obinutuzumab showed a significant increase in survival compared with rituximab.
Obinutuzumab has been reported to have increased cytotoxicity causing B-cell depletion with greater activity against CD20+ tumor cells resulting in improved overall survival [11,24]. Treatment in adults with relapsed indolent CD20+ B-NHL has demonstrated a higher overall response rate compared to rituximab [15]. As mentioned earlier, studies of obinutuzumab vs rituximab in patients with newly diagnosed or aggressive DLBCL in the upfront setting have not shown the same survival advantage [21,22]. The reasons for the lack of advantage with obinutuzumab in DLBCL were unclear. This could be the result of differences in biologic and clinical profiles as well as differences in the immune microenvironment [22].
In contrast to the Type I anti-CD20 antibody rituximab, the tumor-killing mechanisms of obinutuzumab are reported to associate with enhanced direct cell death and ADCC, but reduced CDC [25]. Treatment of Karpas- Figure 3: Obinutuzumab is significantly more effective in inducing NK cell mediated ADCC activity compared to rituximab. Karpas-1106P cells were treated with obinutuzumab, rituximab and IgG isotype control (10 µg/ml) and incubated with expanded and activated NK cells for additional 4 hours at 37°C at a 5:1 & 10:1 effector: targets ratio. Cell lysis was determined by manufacturer's protocol using DELFIA cell cytotoxicity assays. Data shown are mean ± SD (n = 4) obinutuzumab vs rituximab, (p = 0.0009 and p = 0.004). www.oncotarget.com 1106P PMBL cells with obinutuzumab showed significant decreased cell viability irrespective of administered dose ( Figure 2). This report is also consistent with increased cell death by increased ADCC in BL and pre B-cell acute lymphoblastic leukemia cell lines [11], and by increased activation of polymorphonuclear neutrophils resulting in increase in phagocytosis in chronic lymphocytic lymphoma cell lines [26]. Consistently with previous reports that obinutuzumab induced superior ADCC with effector cells using peripheral blood mononuclear cells [24], we found that obinutuzumab significantly enhanced ADCC mediated by ex vivo expanded NK cells ( Figure  3). Constitutive activation of STAT6 pathway in PMBL has been previously reported by other investigators [27]. Differential regulation of AKT phosphorylation by obinutuzumab has also been reported in a 3D NHL model [28,29]. STAT6, NF-κB and PI3K/AKT pathways were reported to be activated in PMBL [2,23]. Our group recently demonstrated the difference in differential phosphorylation of 978 unique phosphoproteins between rituximab sensitive and resistant BL cell lines after obinutuzumab vs. rituximab treatments [12]. In this study, we found that obinutuzumab significantly reduces the phosphorylations of STAT6, AKT and IκBα in PMBL compared to controls (Figure 4). The treatment of a PMBL cell line with obinutuzumab differentially phosphorylated the PI3K/AKT pathway. This, in turn, affected phospho-IκBα and phospho-STAT6 signaling and NF-kB, JAK/ STAT signaling pathways. Our results indicate that obinutuzumab inhibits PMBL proliferation partially by reducing the activation of the PI3K/AKT, NF-kB, JAK/ STAT pathways.
Obinutuzumab enhances ADCC through several mechanisms. The prominent one is involved in enhancing the binding affinity to the FcγRIII receptor on immune effector cells such as NK cells through the post-translational glycoengineering [25,30]. The second mechanism is that obinutuzumab overcomes the inhibition from inhibitory killer cell Ig-like receptors (KIRs) on NK cells [31]. Another mechanism is related to CD32B mediated internalization. CD32B is the inhibitor Fcγ receptor IIB (FcγRIIB) with low-affinity for IgG [32]. It promotes rituximab internalization from B cells and abrogates ADCC, which contributes to rituximab resistance and is correlated with poor clinical responsiveness [33,34]. Obinutuzumab was found to be less sensitive to the internalization, partially due to their ineffectiveness to redistribute CD20 into lipid rafts [35] and less ability to interact with and activate CD32B [36]. Obinutuzumab significantly enhanced expanded NK mediated ADCC than rituximab (Figure 3) possibly through all of the mechanisms.
Gene expression analysis uncovered very small number of genes (0.3%) exhibiting differential expression between obinutuzumab treated and control treated cell lines. GSEA clearly revealed the upregulation of genes related to apoptosis and mitogen-activated protein kinase signaling pathway and protracted downregulation of genes related to cell cycle and cell proliferation including ID3 and RAB6B though the reason for this is not well understood [28,37,38].
We demonstrated that obinutuzumab showed improved survival and inhibition of tumor progression in PMBL xenografted NSG mice. This data is consistent with superior activity of obinutuzumab in other B-cell lymphoma xenografts when used alone or in combination with other agents [1,[39][40][41][42][43]. In conclusion, this study reveals that using obinutuzumab as a single agent is significantly more effective in PMBL cells both in-vitro and in-vivo utilizing PMBL xenografted NSG mice compared to equal doses of rituximab. This novel anti-CD20 type II antibody should be considered as an alternative treatment for PMBL utilizing chemoimmunotherapy regimens.

Immunoblotting
Whole-cell extracts were resolved on sodium dodecyl sulfate polyacrylamide gel electrophoresis, transferred to nitrocellulose membrane, probed with appropriate antibodies and blocked in 5% milk for 1 hour, subsequently incubated with primary antibodies at 4°C for overnight and developed using Enhanced Chemiluminescence Reagent (GE Healthcare Bio-Sciences, Piscataway, NJ, USA) [44]. Band intensities on sodium dodecyl sulfate polyacrylamide gel electrophoresis were measured using ImageJ software.

Quantitative reverse-transcription polymerase chain reaction
Total RNA was prepared using Trizol reagent (Invitrogen) according to the manufacturer's manual. The cDNA was then synthesized using qScript™ cDNA Synthesis Kit (Quantas). Relative quantification (ddCt) of mRNA expression was determined by normalizing to the housekeeping gene (GAPDH).

Cell apoptosis assays
Cell death was assayed by flow cytometric analysis (Beckman Coulter, Brea, CA, USA) of Annexin V, and propidium iodide (BD Biosciences, San Jose, CA, USA). 3 × 10 5 cells were either untreated (control) or treated with 10 μg/ml of obinutuzumab, rituximab or humanimmunoglobulin G for 24 to 72 hours. Annexin V/PI staining was performed according to the manufacturer instructions.

NK cell isolation and expansion
Peripheral blood was obtained from normal healthy adult donors after informed consent from the New York Blood Center and mononuclear cells were isolated by Ficoll-Paque (Amersham Biosciences, Piscataway, NJ, USA) density gradient separation as we have previously described [45].

Antibody dependent cell mediated cytotoxicity assay
ADCC was determined using DELFIA EuTDA cytotoxicity assay as we have previously described [45]. 1 × 10 6 target cells were labeled with BATDA labeling ligand for 30 minutes. Tumor target cells were washed and incubated with 10 μg/ml obinutuzumab, rituximab or polyclonal IgG at 37°C and incubated with expanded NK effector cells, at 10:1 and 5:1 E: T target ratio for 4 hours at 37°C. Plates were centrifuged and the supernatant was collected into a flat-bottom plate and Europium solution was added and incubated for 15 minutes at room temperature. Fluorescence was measured with a timeresolved fluorometer (Perkin Elmer). The percentage of cytotoxicity was calculated using following equation: % specific release = (E-S/(M-S) × 100) E = experimental release, S = spontaneous release (target cells without effector cells), M = maximum release (lysed target cells).

Gene expression microarray
Gene expression experiments were performed using Illumina Human-HT12 v4.0 array at the Rockefeller University Genomics Facility (New York, NY, USA). Biotinlabeled RNA was prepared using 200 μg of total RNA by MessageAmp™ Premier RNA Amplification Kit (Applied Biosystems, Foster City, CA, USA). Antisense RNA (750 ng) was mixed with hybridisation reagents and heated at 65°C for 5 minutes. After it was cooled to room temperature, hybridization solution was applied to Illumina HumanHT-12 v4 array. The array was scanned using Illumina BeadArray Reader. The scanning was done using standard DirectHyb Gene Expression protocol. The raw data was extracted using Illumina BeadStudio software without normalization. Data has been deposited in the Gene Expression Omnibus database (accession number GSE75605).
GSEA with MSigMB v5.2 software was used to investigate the biological processes and pathways differentially implicated between obinutuzumab treated and PBS or IgG treated PBML cells [47,48]. Genes with > 1.5 fold or < 1.5 fold expression ratio were considered to be significantly differentially expressed genes as we and others have previously discussed [48,49].