Pevonedistat, a NEDD8‐activating enzyme inhibitor, induces apoptosis and augments efficacy of chemotherapy and small molecule inhibitors in pre‐clinical models of diffuse large B‐cell lymphoma

Abstract We studied the biological activity of pevonedistat, a first‐in‐class NEDD8‐activating enzyme (NAE) inhibitor, in combination with various cytotoxic chemotherapy agents and small molecule inhibitors in lymphoma preclinical models. Pevonedistat induced cell death in activated B‐cell (ABC) diffuse large B‐cell lymphoma (DLBCL) cell lines and to a lesser degree in germinal center B‐cell (GCB) DLBCL cell lines. In pevonedistat sensitive cells, we observed inhibition of NF‐κB activity by p65 co‐localization studies, decreased expression of BCL‐2/Bcl‐XL, and upregulation of BAK levels. Pevonedistat enhanced the activity of cytarabine, cisplatin, doxorubicin, and etoposide in ABC‐, but not in the GCB‐DLBCL cell lines. It also exhibited synergy with ibrutinib, selinexor, venetoclax, and A‐1331852 (a novel BCL‐XL inhibitor). In vivo, the combination of pevonedistat and ibrutinib or pevonedistat and cytarabine prolonged survival in SCID mice xenograft models when compared with monotherapy controls. Our data suggest that targeting the neddylation pathway in DLBCL is a viable therapeutic strategy and support further clinical studies of pevonedistat as a single agent or in combination with chemotherapy or novel targeted agents.

MCL-1, and BCL-XL levels) or by degrading pro-apoptotic BCL-2related proteins (such as BAK) [5,6]. Inhibition of the UPS is an attractive therapeutic approach in lymphoid malignancies, including DLBCL. Proteasome inhibitors have been used in the treatment of lymphoma; however, their use is limited due to adverse effects [7,8].
Proteasomal degradation of cellular proteins is a multistep process that requires "tagging" of targeted proteins with poly-ubiquitin chains.
The ubiquitin ligases (E3s) are responsible for selectively recognizing substrates [9] and their diversity presents an attractive target for more selective inhibition of the UPS that may potentially be more clinically effective and less toxic. Activation of cullin-based RING-ubiquitin ligase (CRLs) requires neddylation of the cullin subunit, which disrupts its inhibitory binding to the cullin-associated NEDD8-dissociated protein 1 (CAND1) [10]. Neddylation is a posttranslational modification that involves the addition of the ubiquitin-like protein, NEDD8, to a target protein (E3s). Pevonedistat is a first-in-class, NEDD8-activatingenzyme (NAE) inhibitor that selectively prevents the activation of CRLs and alters the ubiquitination and proteasomal degradation of cellular proteins. Inhibition of NAE led to cell death in various cancer models [11][12][13][14]. In vitro exposure of cancer cell lines to pevonedistat was shown to induce apoptosis, cellular senescence, or autophagy [11,14,15,16]. We characterized the preclinical activity of pevonedistat as a single agent and in combination with chemotherapy and other novel agents in DLBCL.

Cell lines and primary tumor cells
RL (GCB-DLBCL) and U2932 (ABC-DLBCL) cell lines were obtained from American Type Culture Collection (Rockville, MD). Rituximabresistant cell lines (RL 4RH and U2932 4RH) were created as described previously [17].

In vitro effects of pevonedistat as a single agent
DLBCL cell lines (0.25 × 10 6 cells/ml) were exposed in vitro to pevonedi-    derived from patients to pevonedistat lead to greater than 20% decrease in viability in two of three ABC-DLBCL samples whereas no effect was noted in a GBC-DLBCL sample. (E) Clinical and pathological characteristics from which primary tumor cells derived from DLBCL patients were obtained. Experiments were repeated three separate times and were reported as the median with standard deviation error bars (SE). Pevonedistat or vehicle control was utilized at 500 nM in patients with DLBCL. Cell death was determined by Cell Titer Glo luminescence assay. *P < .05

Changes in BCL-2 family members and key regulatory apoptotic proteins in DLBCL cell lines after exposure to pevonedistat
To study the mechanisms responsible for pevonedistat antitumor activity in DLBCL, cells were exposed to pevonedistat [U2932 (50 and

In vitro effects of pevonedistat on nuclear factor kappa B activity in DLBCL
To assess the effects of NAE inhibition in NF-B activity, DLBCL cell lines, SUDHL4, TMD8, U2932, and OCI-LY2 (2 × 10 6 viable cells), were exposed to pevonedistat (0.5-5 µM) for 1 and 4 h. NF-B activity was determined using a nuclear translocation assay using ImageStream technology, as previously described [19]. PMA/ionomycin was added to fully activate NF-kB as a positive control.

In vivo effects of pevonedistat as a single agent or in combination with rituximab, chemotherapeutic agents, or small-molecule inhibitors in DLBCL murine models
In vivo studies utilized a disseminated human lymphoma-bearing SCID

Statistics
All experiments were performed in triplicates on three separate occasions. Data were plotted and analyzed using SPSS 21.0 software. For in vitro and ex vivo studies, statistical differences between treatment groups and controls were determined by Student's t-test. For in vitro studies combining pevonedistat and chemotherapy drugs, the coefficient of synergy was calculated using the CalcuSyn software. In addition, differences in survival between treatment groups were calculated using Kaplan-Meier curves. P-value of less than .05 was defined as having statistical significance. assay. Experiments were performed in triplicates. All four chemotherapeutic agents showed synergy with pevonedistat in TMD8 and U2932. Cisplatin, doxorubicin and etoposide also showed synergy with pevonedistat pretreatment in OCI-LY2 cells, but to a lesser extent DLBCL patient samples (both ABC-DLBCL subtype) showed cell death ( Figure 1D).

Pevonedistat in vitro exposure altered the balance of several BCL-2 family member proteins favoring the induction of apoptosis in DLBCL
We selected DLBCL cell lines with high (U2932 and OCI-LY2) or low (TMD8 and SUDHL4) IC 50 values. TMD8 and U2932 repre-

Pevonedistat decreased NF-B activity in DLBCL cell lines
NF-B activity is known to be tightly regulated by the UPS, and translocation of p65 into the nucleus leads to the increased transcription of anti-apoptotic Bcl-2 family members. Similar to observations with other UPS inhibitors (i.e., bortezomib), pevonedistat decreased NF-B activity in the cell lines tested, as demonstrated by p65 co-localization studies ( Figure 3A). NF-B p65 expression was decreased in U2932 and DHL-4 by Western blotting ( Figure 3B).  Figure 5D) in all fourcell lines tested, suggesting a cell-of-origin agnostic mechanism of synergy.

Pevonedistat displayed synergy with chemotherapeutic agents and small molecule inhibitors in vitro and was highly active as a single agent and in combination in vivo
In vivo, pevonedistat alone or in combination with cytarabine or ibrutinib improved the median survival of lymphoma-bearing SCID mice (not reached) when compared to animals treated with cytarabine (41 days) or ibrutinib (47 days) monotherapy (P < .001) ( Figure 6A,B).

Effect of Pevonedistat on anti-CD20 monoclonal antibody activity in vitro and in vivo
Previous investigators have demonstrated that ADCC is an important mechanism of action of anti-CD20 monoclonal antibodies in vitro and in vivo [20,21]. Pre-incubation of DLBCL cell lines with pevonedistat for 48 h prior to rituximab exposure dampened anti-CD20 monoclonal antibody-mediated CMC for TMD-8 and U2932 but enhanced CMC activity in SUDHL4 (Supporting Information Figure 1B). Results for ADCC were more consistent with pre-incubation with pevonedistat for 48 h prior to rituximab exposure decreasing anti-CD20 monoclonal antibody-mediated ADCC in all 4 cell lines tested (Supporting Information Figure 1C). Additive effects were noted on apoptosis when pevonedistat was combined with rituximab in ABC-DLBCL (Supporting Information Figure 1A). However, more importantly, additive activity was observed in vivo when pevonedistat was combined with rituximab in a lymphoma mouse model. TMD8-bearing SCID mice treated with pevonedistat and rituximab resulted in a 100% long-term disease control when compared with animals treated with pevonedistat (P = .048).
However, there was no improvement in survival when the combination was compared to rituximab monotherapy (P = .118) ( Figure 6C).

DISCUSSION
In pevonedistat restored sensitivity to death receptor-mediated apoptosis through the mitochondrial pathway [24]. In CLL, it also induced DNA damage and checkpoint activation by deregulation of Cdt1, a DNA replication licensing factor, and cell cycle inhibitors p21 and p27 [25]. It exhibited synergy with alkylating agents such as bendamustine and chlorambucil [25]. In our study, while significant in vitro synergy was noted with a variety of chemotherapeutic agents and small molecule inhibitors, we could not replicate similar results in vivo as our model incorporating TMD8 ABC-DLBCL cell lines into SCID mice was quite sensitive to pevonedistat, even as a single agent. Further experiments at varying doses will be needed to demonstrate synergy between these agents in in vivo mouse models. It is also interesting to note the variability of responses between different cell lines. The next line of investigation will be to identify factors causing primary resistance to pevonedistat in cell lines and translate that data to identify patients who stand to benefit the most from this therapy and find strategies to overcome both primary and acquired resistance.
The molecular events observed in lymphoma or CLL pre-clinical models using pevonedistat are similar to what has been demonstrated with reversible (bortezomib or ixazomib) or irreversible (carfilzomib) proteasome inhibitors [26,27]. In DLBCL, the anti-tumor activity of proteasome inhibitors appears to be limited to the ABC-DLBCL subtype [7]. Clinically, the activity of proteasome inhibitors in lymphoid malignancies has been limited by toxicities. Early clinical trials evaluating bortezomib in combination with systemic chemotherapy in patients with relapsed DLBCL demonstrated anti-tumor activity in ABC-DLBCL [7]. In contrast, a large randomized clinical trial comparing rituximab and CHOP chemotherapy vs. rituximab, cyclophosphamide, doxorubicin, bortezomib, and prednisone in previously untreated ABC-DLBCL failed to meet its primary endpoint [28]. The phase 3 REMoDL-B study F I G U R E 6 Effect of Pevonedistat on the anti-tumor activity of cytarabine, ibrutinib and rituximab in vivo. In vivo, pevonedistat in combination with cytarabine or ibrutinib improved the median survival (not reached at the time of sacking the TMD8-bearing SCID mice on 214th day post-treatment) compared to cytarabine (41 days) (A) or ibrutinib (47 days) (B) monotherapy (P < .001). The combination of pevonedistat with rituximab did not improve survival compared to rituximab alone (C). Survival differences between groups were compared using log rank analysis. P values are of combination compared to single agent treatment. Experiments were repeated three separate times. (n/a = Not reached) also demonstrated no differences in outcomes with the addition of bortezomib to R-CHOP in patients with either ABC-or GCB-DLBCL [29]. While studies evaluating the addition of carfilzomib to various chemotherapy backbones in DLBCL are ongoing [8,30], there is a clinical need to develop novel and less toxic agents targeting the UPS system such as pevonedistat. Preclinical data, including our current work, support the development of NAE inhibitors in DLBCL.
Pevonedistat has been studied as a single agent in a phase 1 trial in patients with relapsed/refractory multiple myeloma and lymphoma [31]. The drug was well tolerated with modest activity; dose-limiting toxicities included febrile neutropenia, transaminase elevations, muscle cramps, and thrombocytopenia. It is now being studied in a host of solid and hematological malignancies in various combinations. Based on the preclinical synergy between pevonedistat and ibrutinib in CLL, MCL, and now in DLBCL [22,32], a phase 1 trial combining the two drugs in patients with relapsed/refractory CLL or NHL including ABC-DLBCL (NCT03479268), is currently ongoing and actively recruiting patients at our institution.
In the future, combinations of pevonedistat and second generation BTK inhibitors such as acalabrutinib, zanubrutinib and LOXO-305 merit investigation due to their enhanced safety profile. While pevonedistat did demonstrate in vitro synergy with small molecule inhibitors such as the SINE compound, selinexor, BCL-XL inhibitor, A-1331852 and BCL2 inhibitor, venetoclax, the mechanisms of synergy between these agents have to be further elucidated before these combinations can be studied in patients. While pevonedistat is active in many disease settings, relapsed DLBCL is a promising area of interest as adding pevonedistat to cytarabine-containing salvage regimens may lead to higher response rates without compromising safety.

ACKNOWLEGEMENTS
The authors thank the patients and their families for providing clinical samples, their clinical colleagues for supporting their research and the funding agencies for financial support.

FUNDING INFORMATION
This work was supported, in part, by grants from the National Can-