RNF8 is responsible for ATRA resistance in variant acute promyelocytic leukemia with GTF2I/RARA fusion, and inhibition of the ubiquitin–proteasome pathway contributes to the reversion of ATRA resistance

To generate cells stably expressing GTF2I-RARA-FLAG or control-FLAG, lentiviral vectors (GTF2I-RARA-FLAG) were transfected into HL60 cells. For details on the transfection conditions, please refer to the Additional file 1.

After cell culturing, cells were harvested and lysed in 1 × RIPA buffer supplemented with a protease inhibitor for western blotting. For immunofluorescence staining, fluorescent signals were acquired using a confocal microscope (Carl Zeiss AG, Oberkochen, Germany). For details on the operating steps, please refer to the Additional file 1.

Cell viability was analyzed with methylthiazolyldiphenyl-tetrazolium bromide (MTT) assays. Briefly, cells were seeded into 96-well plates followed by the administration of ATRA treatments for 24 h, 48 h, and 72 h. After this point, 20 μl of MTT solution was transferred to each well. After incubation for 4 h, cell viability assays were performed.

A coimmunoprecipitation (CoIP) experiment was performed as per the manufacturer’s instructions (Thermo Fisher Scientific, Waltham, MA, USA). Briefly, cell pellets were collected and lysed for 30 min on ice. Soluble lysates were incubated with an antibody coupled with resin at 4 °C overnight, and the proteins were eluted by boiling in 1 × SDS sample buffer before SDS-PAGE. The precipitated proteins were subsequently subjected to SDS-PAGE and blotted with specific antibodies.

NB4 cells and GTF2I-RARA-positive HL60 cells were treated with 1 μM of ATRA for 72 h. The cells were then incubated with PE-conjugated anti-human CD11b antibody (BioLegend, San Diego, CA, USA) and examined by flow cytometry. The percentage of CD11b-positive cells was analyzed using the FlowJo software (FlowJo LLC, Ashland, OR, USA). For morphological analysis, cytospin slides of each sample were stained with Wright-Giemsa staining and observed under an optical microscope.

ChIP was essentially performed on FLAG-GTF2I-RARA-HL60 cells using ANTI-FLAG as per the manufacturer’s instructions (Thermo Fisher Scientific, Waltham, MA, USA). Raw data and experimental methods are available in the Additional file 1.

Total RNA was extracted from cells using TRIzol reagent (Takara Bio, Inc., Otsu, Japan), according to the manufacturer’s protocols. Primer sequences for VRK2, Notch2, WDR26, ANAPC7, RNF8, EMP2, HDAC9, PTPN1, RARA, RXRA, and β-actin messenger RNA (mRNA) are listed in the Additional file 1.

Ubiquitination of proteins requires the covalent attachment of 8.6-kDa ubiquitin (Ub) to multiple lysine residues, forming poly-Ub chains bound to target proteins, and can be seen as a ladder of high-molecular-mass species on SDS–polyacrylamide gels. For details on the experimental design, please refer to the Additional file 1.

Transcriptional activity of RARA was assessed by a luciferase assay. For details on the experimental design, please refer to the Additional file 1.

Western blot confirmed that the GTF2I-RARA protein is stably expressed in GTF2I-RARA-positive HL60 cells (Fig. 1a). When ATRA is exposed to the GTF2I-RARA-positive HL60 and NB4 for 48 h, the PML-RARA protein expression in NB4 cells decreases in a dose-dependent manner, while there is no change regarding the effect in GTF2I-RARA-positive HL60 cells (Fig. 1b). MTT results show that ATRA might inhibit NB4 cell proliferation in a dose- and time-dependent fashion, while the GTF2I-RARA-positive HL60 cells display resistance to ATRA-mediated growth inhibition, especially when exposed to high doses of ATRA (1–2 μM) (Fig. 1c).

To check whether GTF2I-RARA impacts ATRA-induced leukemic cell differentiation, the myeloid lineage differential marker CD11b was detected by flow cytometry. Results showed that the percentage of CD11b-positive cells significantly increases in a dose-dependent manner on NB4 cells under ATRA treatment. However, GTF2I-RARA-positive cells exhibit insensitivity to ATRA-induced differentiation (Fig. 1d). Consistent with the findings of CD11b, morphological features (Wright-Giemsa staining) confirmed GTF2I-RARA-positive cells show attenuated differentiation even with ATRA treatment (Fig. 1e).

The RARA chimeric proteins commonly exhibit increased affinity for transcriptional corepressors, N-CoR/SMRT/HDAC3, which are recruited to RARE, resulting in repression of the RARA target gene [16]. To assess whether GTF2I-RARA could also recruit these transcriptional corepressors, we performed FLAG-tagged GTF2I-RARA pcDNA3.1 plasmid transfection and coimmunoprecipitation assays. Consistent with other RARA fusion proteins, GTF2I-RARA is able to be coimmunoprecipitated with N-CoR, SMRT, and HDAC3 (Fig. 2a). Immunofluorescence localization revealed that FLAG-tagged-GTF2I-RARA colocalizes with HDAC3 mainly in the nucleus and with N-CoR mainly in the cytoplasm, while it colocalizes with SMRT in both the nucleus and cytoplasm in 293T cells (Fig. 2c).

It has been shown that pharmacological doses of ATRA may precipitate the dissociation of RARA fusion proteins from transcriptional corepressors and re-recruit transcriptional activators [17‐19]. As shown in Fig. 2b, our results indicate that a low dose of ATRA does not dissociate the GTF2I-RARA/HDAC3 complex but could impel the release of SMRT and N-CoR from the GTF2I-RARA/SMRT/N-CoR complexes. Consistent with coimmunoprecipitation, the intracellular distribution of GTF2I-RARA and corepressors changed, in that N-CoR and SMRT migrate to the cytoplasm and separate from GTF2I-RARA. Meanwhile, the GTF2I-RARA/HDAC3 complex fails to disassociate and remains colocalized in the nucleus, suggesting that HDAC3 plays a role to a certain extent in GTF2I-RARA-driven transcriptional regulation (Fig. 2c).

To map genome-wide binding sites of GTF2I-RARA, we carried out a ChIP-seq experiment. We identified 221 binding sites of GTF2I-RARA with significant enrichment (Fig. 3a). Using Multiple EM for Motif Elicitation software, motif analysis was performed. The rank 1 de novo motif of GTF2I-RARA was detected in 9% of the DNA regions (Fig. 3c), which is not in agreement with the classic RARE. The raw data files have been uploaded to NCBI (GEO accession number: GSE122354).

While analyzing the 221 predicted peaks in the gene body, most of the peaks were found to be dispersed in introns (37.7%) and intergenic regions (42.3%), with only a few binding to exonic areas (Fig. 3b). The number of genes proximal to the GTF2I-RARA binding sites [within 5 kb upstream from the transcription start site (TSS), the gene body, and 5 kb downstream from the TTS] was 123. Known RARA target genes, such as RARb2, ASB2, IL8, CEBPE, and PRAM1 [20‐23], were not included in the 123 genes. To identify reliable novel target genes for GTF2I-RARA, we reviewed the biological function of the 123 annotated genes and focused on VRK2, ANAPC7, PTPN1, EMP2, WDR26, RNF8, NOTCH2, and HDAC9 genes, which are involved in cell proliferation, differentiation, and apoptosis [24‐31]. We compared the target gene expression between GTF2I-RARA-positive HL60 and control HL60 cells. Among the eight genes mentioned above, only RNF8 expression is significantly upregulated at both the mRNA and protein level (Fig. 3d, e). Because GTF2I-RARA can bind both exonic and intronic regions of RNF8 and functional RNF8 is known to act as a key Ub ligase (E3), we speculate that RNF8 might play an important role in the pathogenesis of the variant APL and confer ATRA resistance.

RNF8 is able to bind RXRA through the N-terminal regions of both proteins [32]. The confirmed fact that RARA binds to RXRA as a heterodimer, in turn, results in the transcriptional activation of RA target genes [33]. We presume that RNF8 likely interacts with RARA and participates in RARA ubiquitination. Our preliminary results showed that there is downregulated expression for RARA in GTF2I-RARA-positive HL60 cells; however, there is no change in the RXRA level (Fig. 4a). Coimmunoprecipitation results confirmed RNF8 not only is bound to RXRA but also interacts with RARA (Fig. 4b), suggesting that RNF8 can form a heterodimer with RARA or RXRA and regulate transcriptional activation or inhibition of RA targeting genes. The intracellular localization results for RNF8, RXRA, and RARA proteins revealed that RNF8 is predominantly distributed in the cytoplasm, while RXRA and RARA are localized both in the nucleus and cytoplasm (Fig. 4c, left panel). Furthermore, in GTF2I-RARA-positive HL60 cells, RNF8 displays overt co-localization with RXRA or RARA mainly in the cytoplasm (Fig. 4c, right panel), suggesting that GTF2I-RARA mediates the transposition of RNF8 and the formation of RNF8/RXRA or RNF8/RARA heterodimers. To prove whether RNF8 disturbs RARA expression by downregulation, RNF8-overexpressed or knocked down HL60 cell models were constructed. Results showed that the protein level, not the mRNA level of RARA, is downregulated in RNF8-overexpressed HL60 cells. Meanwhile, RNF8-siRNA HL60 cells exhibit upregulated RARA protein expression and have no effect on RARA mRNA expression (Fig. 4d, e). These results confirm that RNF8 is a negative control factor for the RARA protein and might influence the post-translational modification of RARA. This could be responsible for GTF2I-RARA-driven ATRA resistance.

RNF8 plays a critical role in catalyzing both K48- and K63-linked Ub chains. While functions of many of these distinct Ub chains remain obscure, polyubiquitin chains composed of K48-linkages are generally associated with commitment for proteasomal degradation, whereas K63-linked polyubiquitylation plays established roles in DNA damage–repair, protein kinase activation, and receptor endocytosis [29, 34]. We presume that RNF8 might also be involved in RARA ubiquitination. To validate this assumption and further illustrate the ubiquitin landscape on RARA, we first used the proteasome inhibitor MG132 at different concentrations and applied it to HL60 cells expressing RNF8. Results showed that RNF8 overexpression could antagonize MG132-induced RARA expression (Fig. 5a). To investigate linkage-specific polyubiquitin conjugation on RARA, we further performed coimmunoprecipitation in 293T cells following co-overexpression of RNF8, RARA-Ha, GTF2I-RARA, wild type (WT)-Ub-His, or lysine-only mutant (K48, K63) Ub, in which all lysine residues except one were mutated to arginine. We found that overexpression of RNF8 or GTF2I-RARA leads to a marked increase in RARA Lys48-linkage Ub modification, as well as Ub modification, but it has a minimal effect on K63-linked ubiquitination of RARA (Fig. 5b). These results support the idea that RNF8 is responsible for K48-ubiquitin chain formation of RARA and has been associated with RARA proteasomal degradation. It is known that RARA plays a critical role in myeloid differentiation. Hence, we further assessed the effects of RNF8 on the transcriptional activity of RARA. Luciferase assay results indicated that RNF8 could significantly suppress the transcriptional activity of RARE (Fig. 5c), while knockdown of RNF8 by siRNF8 exhibits an ATRA-induced transcriptional activation in GTF2I-RARA-positive 293T cells (Fig. 5d). To uncover whether overexpression of RNF8 could counteract ATRA-induced cell differentiation, RNF8-transduced HL60 cells were treated with various concentrations of ATRA, and CD11b was detected by flow cytometry. The results showed that the proportion of CD11b-positive cells is found at a lower level in RNF8-transduced HL60 cells compared with the control group (Fig. 5e). Wright-Giemsa staining also confirmed that HL60 with RNF8 overexpression represents attenuated ATRA-induced differentiation (Fig. 5g). However, knockdown of RNF8 in GTF2I-RARA-positive HL60 cells increases the percentage of CD11b-positive cells and confers sensitivity of differentiation to ATRA (Fig. 5f). Cell morphological observation showed that downregulated RNF8 reverses the effects of differentiation arrest of GTF2I-RARA on ATRA (Fig. 5h).

Based on the results above, we envisage that inhibiting RARA degradation might help to restore sensitivity to ATRA in GTF2I-RARA-positive HL60 cells. Therefore, we treated GTF2I-RARA-positive HL60 cells with both MG132 (0–0.2 μM) and ATRA (1.0 μM) and observed cell differentiation status. Our results showed that the proportion of CD11b-positive cells is significantly increased in the MG132 + ATRA group compared with ATRA alone (Fig. 6a). Additionally, cell morphology revealed that GTF2I-RARA-positive cells possess differentiated characteristics following the progressive increases of MG132 concentration (Fig. 6b).