Page 41 - Haematologica - Vol. 105 n. 6 - June 2020

CD20: functions and regulation
known positive regulatory elements present in the MS4A1 promoter include an E-box motif (binding μE3- specific transcription factors such as USF and TFE3), “PU.1/PiP” binding site and a BAT box (Figure 3, Table 1).58,59 The BAT box is a sequence element present in the most proximal region and serves as a binding site for the transcription factors OCT1 and OCT2 with a B-cell restricted co-activator BOB (Figure 3, Table 1).58,60 The BAT element is important for the high constitutive expression of CD20 in mature B cells and the induction of CD20 in pre-B cells.58 The “PU.1/PiP” binding site is a putative site for transcription factors belonging to the ETS family (e.g. PU.1) and protein PiP (IRF4) (Figure 3, Table 1). PiP is recruited to this DNA binding site indirectly by phosphorylated PU.159 and a “PU.1/PiP” binding site seems to be critical for CD20 expression as it is occupied only in CD20-positive B cells. Additionally, PU.1/Pip are downregulated during plasma cell differentiation,61 and mutations in this binding site nearly completely abol- ished the promoter activity of MS4A1.59 Moreover, tran- scriptional CD20 activation in primary CLL and non- Hodgkin lymphoma (NHL) B cells was associated with increased PU.1 and OCT2 binding to the MS4A1 promot- er in response to farnesyltransferase inhibition.62 Downregulating PU.1 expression by overexpression of its negative regulator, namely FLT3, also led to lower CD20 expression in CLL cells and vice versa.63
Several transcription factors from the ETS family, such as ELK1 and ETS1, were observed to be activated in an ERK-dependent manner and enhance CD20 cell-surface expression in B-NHL cell lines and primary CLL cells after bryostatin-1 treatment in vitro (which activates the MEK1/ERK-1/2 pathway via PKC) (Figure 2, Table 1).64 Furthermore, it was proposed that NFκB might positively regulate CD20 expression62,65 and gemcitabine treatment of DLBCL cell lines augmented CD20 expression together with NFκB signaling activation (Figures 2 and 3, Table 1).66 Chromatin immunoprecipitation sequencing analysis also revealed MS4A1 as a direct MYC target gene in Burkitt lymphoma cell lines (Table 1), and MYC silencing resulted in CD20 upregulation.67 The repression of CD20 by MYC is surprising and remains to be confirmed in other lymphoma cell types, since B-cell activation (also leading to MYC expression) is generally known to induce CD20 expression in B cells.68 To reveal other factors regu- lating CD20 expression, Slabicki et al.69 performed a genome-wide RNA interference screening using a library of small hairpin RNAs delivered into Raji cells (Burkitt lymphoma cell line) by lentiviral vectors. They identified 37 potential CD20 repressors and 51 activators, among them CHD4 and MBD2 as novel MS4A1 inducers (Table 1). Both CHD4 and MBD2 are members of the nucleo- some remodeling deacetylase complex, which plays an important role in the regulation of gene transcription. This screening also revealed CREM as the top candidate for CD20 repression, and the presence of three half- cAMP response elements in MS4A1 promoter sites (TGACG) led to the notion that cAMP-mediated signal transduction plays a role in CD20 transcriptional repres- sion. Most recently, FOXO1 transcription factor was described as a negative MS4A1 transcription regulator in lymphoma B cells (Figures 2 and 3, Table 1).70 This is in agreement with the observation that DLBCL patients with activating FOXO1 mutations have shorter overall survival upon rituximab-based therapy.71 As the exact
localization of the putative FOXO1 binding site in the MS4A1 promoter was not determined, it is believed that FOXO1 binds indirectly to the DNA-binding element between -182 and -88 bp (Figure 3).70 These data (and our unpublished data) suggest that FOXO1 inhibitors might theoretically be combined with anti-CD20 antibodies to induce CD20 expression and potentiate the effect of the monoclonal antibodies. Similarly, other groups have pro- posed that inhibiting aurora kinase A/B could also lead to upregulation of CD20 and potentiation of rituximab’s clinical efficacy.72,73
It is not surprising that several recent studies suggested that CD20 is at least partially regulated by epigenetic mechanisms. Tomita et al. demonstrated that treating a CD20-negative B-cell line with the histone deacetylase (HDAC) inhibitor trichostatin A resulted in robust upreg- ulation of CD20 mRNA and protein.74 In vitro treatment of primary cells obtained from relapsed CD20-negative B- NHL patients using the DNA methyl-transferase (DNMT) inhibitor 5-aza-2-deoxycytidine also led to the stimula- tion of MS4A1 mRNA and cell-surface expression within 3 days, and restoration of rituximab sensitivity.75 Despite the fact that CD20 stimulation by DNMT inhibitors was described both in vitro75,76 and in vivo in patients with B-cell malignancies,77 CD20 is less likely to be regulated by CpG (de)methylation as its promoter region does not contain any CpG islands up to ~5 kb upstream from the transcrip- tion start site.76 However, it is plausible that DNMT inhi- bition regulates the methylation status of transcription factors critical for MS4A1 transcription, or some more distant genomic regions (enhancers) are involved in MS4A1 transcription. Furthermore, it was reported that a Sin3A-HDAC1 co-repressor complex is recruited to the MSA41 promoter in CD20-negative B-cell lines (Figure 2, Table 1).76 This complex dissociates from the promoter with 5-aza-2-deoxycytidine and trichostatin A treatment, resulting in histone acetylation and partial restoration of CD20 expression. Shimizu et al. showed that HDAC inhibitors (valproic acid or romidepsin) are able to induce CD20 expression in B-cell lines through MS4A1 promoter hyperacetylation and recruit the SP1 transcription factor within 48 hours (Figures 2 and 3, Table 1).65 At the moment, several ongoing clinical trials are evaluating the efficacy of epigenetic modulators in combination with rit- uximab (Table 2). In the VALFRID study, pretreatment with valproic acid before first-line therapy with CHOP plus rituximab in DLBCL patients resulted in histone acetylation, CD20 upregulation at the mRNA and cell- surface levels78 and improved overall survival.79 In con- trast, the analysis of three CLL patients from the PREVAIL study showed no CD20 induction upon pretreatment with valproic acid.80 A plausible explanation might be that valproic acid induces a bivalent MS4A1 promoter status in primary CLL cells in vivo as it induces histone acetyla- tion, but also transient recruitment of the transcriptional repressor EZH2 to the MS4A1 promoter (Figure 2, Table 1). Administering a DNMT inhibitor and pan-HDAC inhibitor (valproic acid, romidepsin, trichostatin A, SAHA) stimulates CD20 expression and might improve anti-CD20 therapy in vivo, at least in some patients with B-NHL. However, the clinical use of pan-HDAC inhibitors is hindered by adverse effects77,79 and thus the involvement of individual HDAC molecules and selective HDAC inhibitors are undergoing pre-clinical studies. Recently, entinostat, a selective HDAC1/4 inhibitor, was
haematologica | 2020; 105(6)
1499