Page 43 - Haematologica - Vol. 105 n. 6 - June 2020

CD20: functions and regulation
reported to upregulate CD20 and improve rituximab effi- cacy both in vitro and in a mouse model (Figure 2, Table 1).81 Additionally, Bobrowicz et al. identified HDAC6 as a novel repressor of CD20 expression in B-cell lines and pri- mary CLL cells (Figure 2, Table 1).82 HDAC6 was shown to be overexpressed in CLL cells and its inhibition aug- mented the efficacy of anti-CD20 monoclonal antibodies in vitro and improved survival of complement-, NK cell- and macrophage-competent mice (SCID Fox Chase mice) injected with Raji cells and treated with rituximab.82 However, it seems that HDAC6 inhibitor does not induce MS4A1 transcription, but only increases MS4A1 mRNA translation. Its potential clinical use is, therefore, likely limited to malignancies that have highly active MS4A1 transcription but escape anti-CD20 antibodies by pre- venting its translation. This is not likely in most CLL cases, although, one report has suggested that MS4A1 mRNA might be repressed post-transcriptionally by microRNAs in CLL.83
Regulation of CD20 levels on the cell-surface during therapy
Besides transcriptional and epigenetic regulation, sever- al studies have demonstrated CD20 downmodulation on the B-cell surface in response to anti-CD20 therapy (Figure 1). One of these mechanisms is called “shaving”. Monocytes and macrophages recognize rituximab bind- ing to CD20 and remove this complex from the B-cell sur- face via the FcγRI-dependent process of endocytosis called trogocytosis53 and this was observed in CLL patients treated with rituximab in vivo (Figure 1).53,84,85 Alternatively, the acute change in CD20 levels in the malignant B-cell population after rituximab infusion might be partially due to elimination of those cells with the highest CD20 levels. Indeed, we have shown that after rituximab infusion in vivo, the antibody primarily targets and eliminates a subpopulation of CLL cells with the highest levels of CD20 via CDC, whereas many CLL cells with pre-therapy low CD20 levels survive.34 Importantly, the CLL cells with the highest cell-surface CD20 levels are also those with the highest BCR signaling propensity and also represent the vast majority of Ki67- positive cells in peripheral blood.34 This “targeting” of the most aggressive intraclonal CLL cell subpopulation at least partially explains the good clinical efficacy of ritux- imab. It remains to be determined whether rituximab is also targeting specific intraclonal cell subpopulations in diseases such as follicular lymphoma and DLBCL, in which malignant cells have relatively homogenously higher CD20 cell-surface levels.
Another mechanism reducing CD20 expression on B cells in response to anti-CD20 therapy is known as anti- genic modulation (Figure 1).86 This refers to the active internalization and subsequent degradation of CD20/monoclonal antibody complexes demanding ener- gy and cytoskeleton remodeling. Importantly, only type I anti-CD20 monoclonal antibodies induce marked anti- genic modulation. These anti-CD20 monoclonal antibod- ies (e.g. rituximab and ofatumumab) recognize and bind CD20 epitope in a different orientation than type II anti- bodies (obinutuzumab) and are able to redistribute CD20 into lipid rafts on the plasma membrane.87,88 Type I anti- CD20 monoclonal antibodies also have an approximately two-fold higher capacity to bind CD20 epitope, which makes them prone to internalization and proteolytic
degradation.86,89 Moreover, the extent of antigenic modu- lation depends on the type of B-cell malignancy. The most rapid internalization can be seen in CLL cells, fol- lowed by mantle cell lymphoma cells, while follicular lymphoma and DLBCL cells show relatively lower rates of antigen internalization.86 Lim et al. suggested that dif- ferent rates of internalization in B-cell malignancies are due to different levels of inhibitory FcγRIIb on B cells (predominantly expressed on CLL and mantle cell lym- phoma cells).90 Rituximab was proposed to crosslink CD20 and FcγRIIb on the same B cell, resulting in FcγRIIb phosphorylation, and internalization of these complexes into lysosomes for their degradation.
The selection pressure caused by rituximab therapy can also lead to the emergence of malignant B-cell clones that are relatively or fully negative for cell-surface CD20 expression (Figure 1). In some DLBCL patients, mutations in the MS4A1 coding sequence were identified; however, mutations involving rituximab epitope are extremely rare.91 Terui et al. analyzed CD20 mutations in samples obtained from patients with previously untreated or relapsed/refractory B-NHL and found MS4A1 mutations in 11 out of 50 patients (22%).92 Importantly, in four cases (8%), such mutations resulted in a C-terminal truncated form of CD20 protein and reduced its cell-surface expres- sion. Nakamaki et al. also reported a case of a relapsed DLBCL patient with a homozygous MS4A1 gene deletion after rituximab-based therapy.93
Notably, some recurrent genetic mutations in patients with B-cell malignancies might affect CD20 levels, and be favored during therapy. In a clinical trial comparing flu- darabine and cyclophosphamide treatment with fludara- bine, cyclophosphamide and rituximab treatment, it was found that NOTCH1 mutations are associated with a rel- ative resistance to the anti-CD20 therapy.94 Pozzo et al. showed that NOTCH1-mutated CLL cells are character- ized by a lower CD20 expression in comparison to that of NOTCH1-wildtype CLL cells.95 Mutations in NOTCH1 intracellular domain (NICD) result in dysregulation of HDAC-mediated epigenetic repression of CD20 through interactions with the RBPJ transcription factor. RBPJ acts as a negative regulator when forming a complex with HDAC1/2; however, accumulation of mutated NOTCH1 in the nucleus results in the preferential formation of NICD-RBPJ activating complex and higher HDAC1/2 lev- els available for interactions with an MS4A1 promoter.
Recently, microenvironmental interactions in various B- cell malignancies were brought into focus as these pro- vide essential pro-proliferative and pro-survival signals and promote drug resistance (reviewed by Seda & Mraz48). Interactions between mesenchymal stromal cells and CLL cells were shown to protect the leukemic cells from rituximab-induced CDC96 and direct apoptosis,97 and this can be therapeutically targeted by integrin inhi- bition (Figure 1).97 These observations led to the coining of the term “cell adhesion-mediated antibody resistance” as an analogy to the long-known “cell adhesion-mediated drug resistance”, which refers to resistance to classical chemotherapy. CD20 down-modulation in response to microenvironmental stimuli might be a theoretical expla- nation for cell adhesion-mediated antibody resistance. This is supported by the observation that stimulating nor- mal B cells by co-culture with CD40L-expressing fibrob- lasts results in rapid CD20 endocytosis and thus reduces the cell-surface levels of CD20.98 Additionally, Kawabata
haematologica | 2020; 105(6)
1501