ILF2 enhances the DNA cytosine deaminase activity of tumor mutator APOBEC3B in multiple myeloma cells

DNA cytosine deaminase APOBEC3B (A3B) is an endogenous source of mutations in many human cancers, including multiple myeloma. A3B proteins form catalytically inactive high molecular mass (HMM) complexes in nuclei, however, the regulatory mechanisms of A3B deaminase activity in HMM complexes are still unclear. Here, we performed mass spectrometry analysis of A3B-interacting proteins from nuclear extracts of myeloma cell lines and identified 30 putative interacting proteins. These proteins are involved in RNA metabolism, including RNA binding, mRNA splicing, translation, and regulation of gene expression. Except for SAFB, these proteins interact with A3B in an RNA-dependent manner. Most of these interacting proteins are detected in A3B HMM complexes by density gradient sedimentation assays. We focused on two interacting proteins, ILF2 and SAFB. We found that overexpressed ILF2 enhanced the deaminase activity of A3B by 30%, while SAFB did not. Additionally, siRNA-mediated knockdown of ILF2 suppressed A3B deaminase activity by 30% in HEK293T cell lysates. Based on these findings, we conclude that ILF2 can interact with A3B and enhance its deaminase activity in HMM complexes.


A3B-interacting protein candidates are clustered largely into ribonucleoprotein complexes.
To characterize the A3B-interacting protein candidates obtained by mass spectrometry, we employed the STRING interaction database (http:// www. string-db. org/ 27 ). We extracted the protein-protein interaction profiles of the 30 candidates that were common between AMO1 and RPMI8226 cells. Our analysis revealed that A3B-interacting protein candidates clustered largely into ribonucleoprotein complexes (Fig. 1). GO enrichment analysis demonstrated that a large proportion of them were associated with terms related to RNA metabolism, including RNA binding, mRNA splicing, translation, and regulation of gene expression (Supplementary dataset file 3). KEGG pathway analysis revealed that these interacting protein candidates belong to the spliceosome and ribosome complexes ( Supplementary Fig. S1a,b).
Most of the interacting proteins bind to A3B in an RNA-dependent manner. We performed co-immunoprecipitation assays in myeloma cell lines to confirm the binding between A3B and the following representative A3B-interacting proteins: SAFB, SRSF7, Matrin-3, interleukin enhancer-binding factor 2 (ILF2), DHX9, RBMX, hnRNP A1, PTBP1, hnRNP K, hnRNP A3 and hnRNP C. Myeloma cell lysates were subjected to co-immunoprecipitation assays using anti-FLAG M2 affinity gel with or without RNase A. The anti-FLAG antibody specifically co-immunoprecipitated the interacting proteins that were detected by immunoblotting ( Fig. 2 and Supplementary Fig. S6). We observed that in most cases the interaction was abolished in the presence of RNase A suggesting that the interaction was RNA-dependent. Among the 11 proteins tested in Fig. 2, SAFB was still detected after RNase A treatment, suggesting an RNA-independent interaction with A3B.
Most of the interacting proteins are components of HMM complexes. We also tested whether these proteins form HMM complexes with A3B in AMO1-KI and RPMI8226-KI cells by using density gradient sedimentation assays (Fig. 3, Supplementary Figs. S7, S8). We detected SAFB, SRSF7, Matrin-3, ILF2, DHX9, and RBMX in the high-density fractions, but not hnRNP A1, PTBP1, hnRNP K, hnRNP A3, or hnRNP C. After RNase A treatment, most of the putative interacting proteins were observed in the low-density fractions, but not in the high-density fractions. However, SAFB, SRSF7, and Matrin-3 remained in the high-density fractions even after RNase A treatment. Taken together, components of A3B HMM complexes are SAFB, SRSF7, and Matrin-3 via RNA-independent interactions, and ILF2, DHX9, and RBMX via RNA-dependent interactions.
We performed DNA deaminase activity assays with each fraction with or without RNase to evaluate the impact of these interactions on A3B functionality ( Supplementary Fig. S9). Consistent with previous reports, RNase treatment increased deaminase activity. Interestingly, higher density fractions showed strongest activity after RNase treatment. It is possible that after RNA depletion, A3B and other DNA-binding proteins form RNAindependent high molecular masses.  www.nature.com/scientificreports/ ILF2 co-localizes with A3B in the nucleus and enhances the cytidine deaminase activity of A3B. We further characterized two A3B interacting proteins: SAFB, which is a nuclear matrix protein known to form molecular assemblies 28 , and which interacted with A3B in an RNA-independent manner (shown in Fig. 2), and ILF2, which interacted with A3B in an RNA-dependent manner and which is the only protein shown to contribute to multiple myeloma pathogenesis 29 among the interacting proteins listed so far. First, we checked the intracellular localization of these proteins in AMO1-KI cells by immunofluorescence microscopy. We observed that both SAFB and ILF2 localized in the nucleus. Taking into consideration also the results of co-IP and density gradient sedimentation analyses, these proteins co-localize with A3B in the nucleus (Fig. 4a,b). We confirmed the interactions in the nucleus using co-IP and reciprocal co-IP with nuclear lysates from HEK293T cells ( Supplementary Fig. S10a,b).
Second, to examine whether these interacting proteins affect A3B function, we measured the deaminase activity of A3B in vitro using purified proteins produced in a wheat germ cell-free expression system (Supplementary Fig. S11a-d). A3B-C terminal domain (CTD), ILF2 and oligonucleotides were mixed and rotated for 30 min at 37 °C. We used the fluorescent protein Venus as a negative control because it doesn't interact with A3B. ILF2 enhanced A3B deaminase activity compared to Venus in a dose-dependent manner ( Supplementary  Fig. S12a,b). We could not test SAFB in this assay because we were not able to purify in vitro-translated SAFB proteins by a standard method.
Third, we measured the deaminase activity of A3B using HEK293T cell lysates with overexpressed A3B and interacting proteins as previously described 12  www.nature.com/scientificreports/ by 30%, but of SAFB did not (Fig. 5a,b). Additionally, siRNA-mediated knockdown of ILF2 suppressed A3B deaminase activity by 30% (Fig. 5c,d). Taken together, these data indicate that ILF2 positively regulates the deaminase activity of A3B.

Discussion
RNA-mediated formation of HMM ribonucleoprotein complexes plays a crucial role in APOBEC deaminase activity. Mishra et al. reported that mass spectrometry-based proteomics using HEK293T cells with overexpressed A3B identified A3B-interacting proteins, however, the precise role of these proteins has not been elucidated 24 .
In this study, we have performed for the first time a proteomic analysis of the A3B interactome in a multiple myeloma context. We used myeloma cells that endogenously express A3B and identified 30 putative interacting proteins. Most of the putative interacting proteins were RNA-binding proteins including hnRNPs, ribosomal proteins and splicing factors. Co-IP experiments showed that, except for SAFB, the interactions of these proteins with A3B were RNA-dependent. We also showed that these interacting proteins exist in HMM complexes and most of them move to low-molecular fractions when treated with RNase A. Additionally, immunofluorescence microscopy studies showed that ILF2 and SAFB co-localize with A3B in the nucleus. Lastly, functional assays showed that ILF2 enhanced the deaminase activity of A3B.
Our results indicate that the components of A3B HMM complexes may play important roles in regulating its deaminase activity. In terms of other APOBEC family proteins, components of the complexes they are part of have been reported to modulate their catalytic activity. For example, activation-induced cytidine deaminase (AID) monomers interact with hnRNP K, whereas AID dimers interact with hnRNP L 30 . hnRNP K is involved in DNA cleavage events associated with somatic hypermutation, while hnRNP L is involved in DNA recombination events associated with class switch recombination 31 , suggesting that different interacting proteins play different roles in the function of APOBEC-protein complexes. However, little is known about how protein-protein interactions affect the deaminase activity of APOBEC-protein complexes. Four tyrosine mutations and W127A in the N-terminal domain of A3B impair HMM assembly and A3B deaminase activity 23 . Other components of the HMM complexes also bind to ssDNA and might provide substrates to adjacent A3B, facilitating deamination. RNase A treatment of the complexes drastically elevates deaminase activity 18,23 although it disrupts RNAdependent interaction with activating co-factors including ILF2. Additionally, our density sedimentation analysis showed that the fraction with the highest deaminase activity shifted to higher molecular weight fractions after RNase treatment. It is possible that RNA-mediated A3B inhibition might be more important than co-factor mediated activation. Another possibility is that A3B HMM may be converted into RNase-resistant complexes with unknown activating co-factors.  www.nature.com/scientificreports/ ILF2 is overexpressed in various tumor types, including multiple myeloma 29 . Of note, ILF2 is located in the chromosomal region 1q21. 1q21 amplification occurs in approximately 30 to 50% of newly diagnosed and 50 to 80% of relapsed/refractory multiple myeloma cases, and is associated with disease progression and drug resistance [32][33][34][35][36] . The candidate genes which participate in the pathogenicity of 1q21 amplification include MUC1 37 , MCL1 38 , PDZK1 39 , IL6R 40 , BCL9 41 , CKS1B 42 , PSMD4 43 and ILF2 29 , but the crucial driver oncogenes have not been identified yet. ILF2 induces drug resistance in myeloma cells via mRNA processing and the stabilization of transcripts involved in homologous recombination in response to DNA damage 29 . In this study, we found that ILF2 interacts with A3B and enhances A3B deaminase activity, suggesting that ILF2 may contribute to clonal evolution or drug resistance not only by enhancing mRNA splicing of DNA damage response proteins, but also by increasing A3B deaminase activity.  www.nature.com/scientificreports/ In summary, we demonstrate that, in myeloma cells, A3B HMM complexes are comprised of multiple RNAbinding proteins and some of them can affect A3B deaminase activity. However, it remains to be established whether these interacting proteins regulate A3B deaminase activity in myeloma cells in vivo. It is possible that other co-factors interact with A3B and play a different role in certain conditions such as under replication stress or after viral infection. Further investigation is required to elucidate which interactions are meaningful in physiological settings.

Methods
Cell lines and cell culture. Two human myeloma cell lines, AMO1 and RPMI8226 cells were maintained in RPMI1640 (Nacalai Tesque) containing 20% FBS and 1% PSG (Invitrogen). HEK293T cells were maintained in DMEM (Nacalai Tesque) containing 10% FBS and 1% PSG. A3B reporter cell lines that contain a 3 × FLAG tag at the C-terminus of the A3B gene: AMO1-KI and RPMI8226-KI cells were generated as previously reported 26 .

Co-immunoprecipitation assays for mass spectrometry analysis. For co-immunoprecipitation
with an anti-FLAG antibody for mass spectrometry analysis, nuclear extracts were prepared as previously described 44 . 5 × 10 7 cells were briefly washed once with phosphate-buffered saline and the cell pellet were suspended in Hypobuffer (10 mM HEPES-KOH pH 7.9, 10 mM KCl). After incubating for 15 min on ice, Tri-tonX-100 was added. The lysed cellular suspension was briefly vortexed and microcentrifuged for 30 min at 4 °C. The supernatant was discarded as cytoplasmic extract. The pellet was resuspended in high salt radioimmunoprecipitation (RIPA) buffer (50 mM Tris-HCl pH 8.0, 450 mM NaCl, 1% (v/v) Triton X-100, 0.1% (v/v) SDS, 0.1% (v/v) sodium deoxycholate) and agitated for 30 min at 4° C. After centrifugation, the supernatant was mixed with 2 volumes of no salt RIPA buffer (50 mM Tris-HCl pH 8.0, 1% (v/v) Triton X-100, 0.1% (v/v) SDS, 0.1% (v/v) sodium deoxycholate, 5% Glycerol). After pre-clearing with 10 μL of FG beads (Tamagawa Seiki), the supernatant was incubated with antibody-coupled FG beads for 2 h at 4° C. FG beads were conjugated with the anti-FLAG antibody according to the manufacturer's instructions and then used for immunoprecipitation. The beads were washed five times with low salt RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% (v/v) Triton X-100, 0.1% (v/v) SDS, 0.1% (v/v) sodium deoxycholate) and then incubated for 30 min after adding FLAG elution buffer (TBS supplemented with 150 μg/mL of 3 × FLAG peptide). The protease inhibitor cocktails (Nacalai Tesque) and PhosSTOP (Roche) were added to Hypobuffer and RIPA buffer just before use. The eluted samples were analyzed by silver staining using the Silver Staining MS kit (Wako, Osaka, Japan) according to the manufacturer's instructions.
Label-free quantification of the relative protein abundance. The acquired datasets were analyzed using ProteinPilot 5.0.1 (SCIEX) with the UniProtKB/Swiss-Prot human database (May 2018) appended with known common contaminants (SCIEX). The quality of the database search was confirmed by the false discovery rate analysis in which the reversed amino acid sequences were used as decoy. The reliability of protein identifications was evaluated by the number of identified peptides with confidence of at least 95%, and Unused ProtScores that were calculated by the Pro Group algorithm (SCIEX). Relative abundances of the identified proteins were estimated on the platform of Progenesis QI for Proteomics 4.1 (Nonlinear Dynamics, Newcastle upon Tyne, UK). All raw data files in wiff format (SCIEX) were imported to generate aggregates, and the peptide identification results by ProteinPilot with confidence of at least 95% were used for assignment. Label-free quantification of proteins was performed by relative quantitation using the Hi-N method (Nonlinear Dynamics).

Functional classification and network of interactions. Protein-protein interaction networks were
constructed with the STRING database 27 . GO annotation and KEGG pathway enrichment analysis was conducted using DAVID Functional Annotation Tool 45-49 . Antibodies. Antibodies used for protein analysis were as follows: Anti-FLAG M2 antibody (  Co-immunoprecipitation assays for immunoblotting. Myeloma cells were lysed with low salt RIPA buffer containing complete Protease Inhibitor Cocktail (Roche) and PhosSTOP (Roche). The lysates were immunoprecipitated using the ANTI-FLAG M2 Affinity Gel (A2220 Sigma-Aldrich) at 4 °C for 2 h in the presence or absence of RNase A, DNase-free, followed by immunoblotting.
HEK293T cells were transfected with expression vectors for A3B-FLAG using the XtremeGENE HP DNA Transfection Reagent. The nuclear extract was collected as described for the sample preparation for mass spectrometry and immunoprecipitated using the anti-FLAG, anti-ILF2 or anti-SAFB antibody along with Protein G beads at 4 °C, followed by immunoblotting.
Density gradient sedimentation. Nine layers of 4 to 36% sucrose were prepared in GST lysis buffer (150 mM NaCl, 25 mM HEPES pH 7.4, 0.5% Triton-X100, 1 mM MgCl 2 , 1 mM ZnCl 2 , 10% Glycerol). The whole cell extract, which was made with GST lysis buffer containing protease inhibitor cocktail (Roche), was loaded on top of the gradient and ultracentrifuged for 14 h at 25,000 rpm (CP65; Hitachi Koki) in a SW41Ti rotor. After ultracentrifugation, 10 fractions were collected from the top of the gradient and subjected to immunoblotting and gel-based cytidine deaminase activity assay.
Immunofluorescence assays. Myeloma cells were air-dried and fixed in 3.7% formaldehyde in phosphate-buffered saline (PBS) for 10 min at room temperature on glass slides using Shandon cytospin 2 (Thermo Fisher Scientific). Fixed cells were permeabilized, reduced and denatured for 30 min at room temperature in PBS buffer containing 0.5% SDS, 5% β-mercaptoethanol and 10% FBS. Then, cells were washed three times with PBS containing 4% FBS and 0.1% Triton X-100 (PET buffer) and incubated with the anti-FLAG antibody (1:200 dilution) and either rabbit anti-ILF2 (1:200 dilution), or anti-SAFB (1:200 dilution) antibody for 1 h. Cells were then washed three times with PET buffer and incubated with Alexa Fluor 555 and Alexa Fluor 647 labeled secondary antibodies (A11078; A11037 Cell Signaling Technology) for 1 h in the dark. All antibodies were diluted with 3% BSA and 0.5% Tween in PBS. Slides were mounted in VECTASHIELD with DAPI (Vector Labs) and observed with a confocal laser scanning microscope (BZ-X800, KEYENCE).
Protein expression and purification using a wheat germ cell-free expression system. Histagged A3B (C-terminal domain) was expressed in a wheat germ cell-free expression system using the WEPRO7240H Expression Kit (CellFree Sciences, Matsuyama, Japan). Brij-35 at a final concentration of 0.04% and zinc acetate at a final concentration of 1 µM were added to the translation reaction solution. Expressed proteins were purified by affinity chromatography with nickel-Sepharose.
Flag/GST-tagged ILF2, hnRNP A1, and Venus were expressed in a wheat germ cell-free expression system using the WEPRO7240G Expression Kit (CellFree Sciences, Matsuyama, Japan). Brij-35 at a final concentration of 0.04% was added to the translation reaction solution. Expressed proteins were purified by affinity chromatography with glutathione-Sepharose.
Gel-based cytidine deaminase activity assays. To prepare cell lysates with overexpressed A3B and interacting proteins, HEK293T cells were transfected with expression vectors for A3B and for each interactor using XtremeGENE HP DNA Transfection Reagent (Roche). The cells were harvested 48 h later, lysed with 120 µl of GST lysis buffer/well and used for gel-based cytidine deaminase activity (CDA) assays. To prepare interactor-depleted lysates, HEK293T cells (5 × 10 5 ) were transfected with 60 nM interactor-specific or scrambled siRNA. At 24 h, the cells were transfected with expression vectors for A3B. At 72 h, the cells were harvested, lysed and assayed. 5 µl of HEK293T cell lysate and 1 pmol of single-stranded DNA oligonucleotide (ATT ATT ATT ATT CAA ATG GAT TTA TTT ATT TAT TTA TTT ATT T) with 5'-attached FAM were incubated with 0.005 units of UDG and 3.75 µl of reaction buffer in 10 µl reaction volume for 2 h at 37 °C. Subsequently, the oligo products were incubated in 100 mM NaOH for 30 min at 37 °C, stained with loading dye, and denatured, followed by electrophoresis in 20% Tris/urea-acrylamide gel, which was then visualized with Image Quant LAS 500. Purified A3B-CTD, ILF2, and Venus were produced using the wheat germ cell-free expression system 50,51 and were provided by Cell Free Sciences ( Supplementary Fig. S11a-d). For the gel-based CDA assays using purified proteins, we used purified A3B-CTD instead of lysate. Before the reaction, 1 pmol of A3B-CTD and 2, 4, or 8 pmol of each interactor were mixed and rotated for 30 min at 37 °C. For the assays using the gradient fractions from myeloma cell lines, each lysate fraction, DNA substrates and UDG were mixed and incubated for 6 h at 37 °C before the assay.