(SUMO)-wrestling with rituximab

In this issue of Blood, Nakamura et al show that the SUMOylation inhibitor subasumstat (TAK-981) augments the ability of rituximab to evoke macrophage phagocytosis and natural killer (NK) cell cytotoxicity and delivers improved antitumor activity in murine models (see figure).¹ Anti-CD20 monoclonal antibodies (mAbs) have become the mainstay of treatment for B-cell disorders ranging from non-Hodgkin lymphoma to rheumatoid arthritis and multiple sclerosis, with next-generation mAbs building on the success of rituximab.² Nevertheless, treatment success is varied, and the challenge is to overcome resistance and improve outcomes for patients.³ Nakamura et al present one such approach.

SUMOylation⁴ represents a posttranslational protein modification akin to ubiquitination that controls multiple cellular and organismal processes, including inflammation, whereupon a small ubiquitin-related modifier (SUMO) is added and removed dynamically from relevant target proteins to modulate their activity.

The study by Nakamura et al uses the SUMOylation inhibitor subasumstat to promote the innate antitumor immune response through the induction of type I interferons (IFN1s). By using human and mouse macrophages, they convincingly show the induction of IFN1s after treatment with TAK-981 alongside other inflammatory mediators. This activity was sufficient to elicit inflammatory M1-type macrophage skewing and was largely dependent on IFN1s. TAK-981 also impaired anti-inflammatory M2-type polarization, suppressing interleukin-4 (IL-4)–dependent M2 induction either when TAK-981 was added simultaneously or when IL-4 signaling had already been initiated. This suggests that TAK-981 can repolarize macrophages that have already been exposed to anti-inflammatory stimuli such as might be present in the tumor, potentially overcoming a means of resistance to anticancer treatment.

These observations recapitulate those reported for other inducers of IFN1s, such as stimulator of interferon genes (STING) agonists.⁵ Of interest, the clinical development of STING agonists has been hampered by dose-limiting toxicities for systemic treatment, leading to the adoption of intratumoral administration or other means of regulating systemic exposure. Here, combining TAK-981 with rituximab or the anti-CD38 daratumumab did not result in any overt toxicities in the immunodeficient mouse models used (judged by the lack of weight loss). So it will be interesting to see whether it elicits toxicity in the clinic or whether IFN1 production through the modulation of SUMOylation differs from that induced by other means. If successful, these studies might also provide impetus for exploring combinations of mAbs and TAK-981 in other diseases in which suboptimal deletion of target cells is associated with lack of therapeutic efficacy of mAbs in rheumatoid arthritis and systemic lupus erythematosus,⁶ although care will be needed in diseases in which IFN1s may exacerbate existing autoimmune inflammatory environments.

In mouse models at least, phagocytosis through myeloid cells is the critical effector mechanism of direct targeting mAbs such as anti-CD20. This process is entirely dependent upon activating the fragment crystallizable (Fc) gamma receptor (FCGR) expression in macrophages and is impaired by expression of the sole inhibitory FCGR2(B) in both mice and humans.⁷ Therefore, an important observation was that treatment with TAK-981 elevated the expression of activating FCGR1 and FCGR3A in humans and FCGR4 in mice, both transcriptionally and at the cell surface. Notably, in the Nakamura et al study, expression of activating FCGR2A was unchanged, whereas FCGR2A and FCGR3A were previously upregulated on macrophages by both STING agonists and IFN1 but FCGR1 was unchanged.⁵ This presumably reflects differences in the ground state of the macrophages in each study. The specific changes in FCGR in human macrophages also raises questions about the potential impact of TAK-981 on next-generation anti-CD20 mAbs such as obinutuzumab. Obinutuzumab is a type II glycoengineered anti-CD20 mAb that exhibits reduced cell surface internalization, greater direct cell killing, and augmented affinity for FCGR3, resulting in greater efficacy in multiple models of B-cell deletion.^2,6,8 Whether TAK981 combined with obinutuzumab will also deliver improved deletion of target cells is therefore of interest, given the higher initial depleting effect of obinutuzumab and the already boosted interaction with FCGR3.

NK cells, the main mediator of antibody-dependent cellular cytotoxicity–type responses, at least in human blood, were also activated by TAK-981, although this did not lead to increased expression of FCGR, which supports the differential regulation of FCGR in different cell types. Surprisingly, TAK-981 killed mouse NK cells, which perhaps indicates the sensitivity of these cells to either SUMO regulation, IFN1 production, or both. It also raises questions regarding the additional impacts of TAK-981–mediated SUMO modulation above those deriving from IFN1 production. Comprehensive transcriptomic and proteomic studies may shed light on the shared and divergent pathways and indicate the possible additional utility of SUMO inhibition.

Despite differences among species, the association between upregulated activating FCGR and improved phagocytosis remained and was coincident with improved therapeutic efficacy in a number of xenograft models in which TAK-981 was combined with rituximab or daratumumab. Subsequent mechanistic experiments showed that the therapeutic effect of the combination of rituximab and TAK-981 was entirely dependent upon the rituximab Fc as a P329G, L234A, L235A (PG-LALA) mutant (which obviates binding to FCGRs as well as complement component 1q [C1q]) was entirely inert.⁹ However, the data for the key cell type involved was less clear, with both clodronate (removing macrophages) and anti-asialo GM1 (ASGM1) (removing NK cells) showing a relatively limited effect. However, at least in the case of macrophages, this may reflect the inability to fully deplete the tumor-associated macrophages, perhaps because depletion was not initiated until after tumors had formed.

One limitation of the Nakamura et al study was the adoption of human lymphoma cell lines as xenografts and the lack of evidence for the TAK-981:anti-CD20 effect in syngeneic and orthotopic models. The authors have already documented impacts of TAK-981 in immunocompetent mice.¹⁰ In those earlier studies, TAK-981 was shown to activate dendritic cells and T cells, improve antigen cross-presentation and T-cell priming in vivo, and inhibit growth of syngeneic mouse tumors, potentiating the impact of anti–programmed cell death protein 1 (anti-PD-1) and anti-CTLA4 mAbs. Therefore, although the Nakamura et al study permitted the assessment of the impact of TAK-981 in the absence of these contributors, it precluded holistic understanding of the net impact across all responding cell types in terms of both efficacy and toxicity.

Notwithstanding these gaps, we await the ongoing clinical studies with enthusiasm. TAK-981 presents an exciting new member of our armory of small molecules that can be combined with large biomolecule therapeutics to overcome mechanisms of treatment resistance and improve clinical responses for patients.