Repression of PML Nuclear Body-Associated Transcription by Oxidative Stress-Activated Bach2

Abstract

Several lines of evidence suggest that gene expression is regulated not only by the interaction between transcription factors and DNA but also by the higher-order architecture of the cell nucleus. PML bodies are one of the most prominent nuclear substructures which have been implicated in transcription regulation during apoptosis and stress responses. Bach2 is a member of the BTB-basic region leucine zipper factor family and represses transcription activity directed by the 12-O-tetradecanoylphorbol-13-acetate response element, the Maf recognition element, and the antioxidant-responsive element. Bach2 forms nuclear foci associated with PML bodies upon oxidative stress. Here, we demonstrate that transcription activity associated with PML bodies is selectively repressed by the recruitment of Bach2 around PML bodies. Fluorescence recovery after photobleaching experiments revealed that Bach2 showed rapid turnover in the nuclear foci. The Bach2 N-terminal region including the BTB domain is essential for the focus formation. Sumoylation of Bach2 is required for the recruitment of the protein around PML bodies. These observations represent the first example of modulation of transcription activity associated with PML bodies by a sequence-specific transcription factor upon oxidative stress.

The PML gene was identified from the breakpoint of the t(15;17) chromosomal translocation found in acute promyelocytic leukemia (APL) (8, 19). This chromosomal translocation results in a fusion of PML with the retinoic acid receptor alpha (RARα) gene in APL cells. PML is typically found concentrated in discrete nuclear speckles named PML nuclear bodies together with other proteins such as SUMO-1, pRB, p53, BLM, and Sp100 (3, 12, 43, 56). PML is dislocated in APL together with other nuclear body components into aberrant nuclear structures (11, 23, 51). Retinoic acid induces degradation of the PML-RARα fusion protein and subsequent relocalization of the various protein components into normal nuclear bodies, a process that results in differentiation and apoptosis of APL cells (13, 31). Hence, the critical functions of the PML bodies in tumor suppression can be impaired in APL, giving the leukemic cells a proliferative and survival advantage (58). PML is essential for the formation of the nuclear bodies (15). Furthermore, PML needs to be covalently bound to SUMO-1, or sumoylated, in order to be localized in the nuclear body (34, 43, 57). Studies of the function of PML revealed that this protein plays several physiological roles, such as a mediator of interferon function (28, 40), a proapoptotic factor (39, 49), and a tumor suppressor (29, 33). Furthermore, PML might work as a transcription regulator by interacting with p53 and pRB (3, 12). Therefore, transcription regulation within and around PML bodies might play important roles in the various functions of PML, such as the regulation of apoptosis (58).

The mammalian transcription activator Nrf2 is a member of the cap'n'collar (CNC)-related basic region-leucine zipper (bZip) family. It forms a heterodimer with the small Maf oncoproteins, playing a critical role in inducing oxidative stress response genes by binding to their regulatory DNA sequence Maf recognition elements (MAREs) (16, 17). Bach2 was identified as a partner of small Maf and is distantly related to the CNC family (37). In contrast to other dimers, the Bach heterodimers function as transcription repressors (14, 36, 37). In chronic myeloid leukemia (CML) cells characterized by the BCR-ABL oncoprotein, BACH2 is induced after selective inhibition of the BCR-ABL kinase activity by STI571 (47). Since STI571 causes inhibition of cell proliferation and induction of apoptosis, these observations suggest Bach2 as a regulator of proliferation and/or apoptosis. STI571 has additive or synergetic effects when used in combination with cytarabine, doxorubicin, and etoposide in CML cell lines (21). These anticancer drugs are oxidative stressors and induce nuclear accumulation of BACH2 (20). Along this line, it was reported that Bach2 exerts an inhibitory effect on cell proliferation and induces apoptosis upon mild oxidative stress when overexpressed in NIH 3T3 and Raji B lymphoid cells (36). These findings suggest that Bach2 functions as a proapoptotic factor by inhibiting MARE-dependent gene expression upon oxidative stress.

Dynamic redistribution of transcription factors within a cell or nucleus is one of the key mechanisms of transcription regulation in cellular responses to various types of stimulation (6, 32). Several lines of evidence suggest that gene expression is regulated not only by the interaction between transcription factors and DNA but also by the higher-order architecture of chromatin and/or the position of the genes within the nucleus (6, 32). However, little is known about the relationships between higher-order nuclear architecture and transcription regulation. Bach2 is localized in the cytoplasm through its C-terminal evolutionarily conserved cytoplasmic localization signal (CLS) (14). CLS directs leptomycin B-sensitive and Crm1/Exportin 1-dependent nuclear export. Oxidative stressors abort the CLS activity and induce nuclear accumulation of Bach2. Interestingly, while both oxidative stress and LMB treatment translocate Bach2 into nuclei, the LMB-induced nuclear accumulation of Bach2 is not sufficient to induce apoptosis (36). These observations suggest that there is another layer of regulation of Bach2 activity in addition to the shuttling between the cytoplasm and nucleus. A candidate mediator of such regulation is a bric-a-brac, tramtrack, and broad complex (BTB) domain (also known as POZ for pox and zinc finger) of Bach2, which is involved in protein-protein interactions (2). Since proteins with the BTB/POZ domain are known to form nuclear foci (1, 9, 10, 41), subnuclear localization of Bach2 may be important for the transcriptional regulation by Bach2. PLZF, which was cloned from the breakpoint of a variant type of APL chromosomal translocation, carries the BTB domain and forms nuclear foci showing a tight spatial relationship with PML bodies (41). Recently, it was also found that Bach2 forms nuclear foci which show a close association with PML bodies (36). However, the mechanistic and biological significance of the spatial relationship between PML bodies and proteins with the BTB domain is still unclear. For example, we do not know whether these proteins regulate transcription in PML bodies.

Here, we provide evidence that, upon oxidative stress, Bach2 is recruited around PML bodies, leading to selective repression of the transcription activity associated with PML bodies. The Bach2 BTB domain is essential but not sufficient for the focus formation, and sumoylation is required for the recruitment of the protein around PML bodies.

MATERIALS AND METHODS

Cell culture and transcription labeling with BrUTP.

GM02063, a simian virus 40-transformed human fibroblast cell line, and 293T cells were cultured in Dulbecco's modified Eagle's medium with 10% fetal calf serum (FCS). NAMALWA cells were maintained in RPMI 1640 medium with 10% FCS. For transcription labeling, 5-bromouridine 5′-triphosphate (BrUTP; Sigma) was added to the culture medium at a final concentration of 1 mM. Cells were scratched with a 27-gauge hypodermic needle and incubated for 5 min at 37°C before fixation with 4% paraformaldehyde in 1× phosphate-buffered saline (PBS). Cells were treated with diethyl maleate (DEM) at the indicated concentrations.

Plasmids.

FLAG epitope-tagged Bach2, Bach2ΔBTB, B1B2(A), B2B1(B), and B1B2(C) were described previously (36). The HcRed-Bach2 fusion plasmid was constructed by inserting the BamHI fragment from pcDNAFLAGBach2 (36) into the BamHI site of pHcRedC1. Bach2KR was constructed as follows. Lys-to-Arg substitutions in the putative sumoylation sites were introduced into the mouse Bach2 cDNA by using Altered Sites II in vitro mutagenesis systems (Promega). The mutagenic oligonucleotides used were 5′-CCAGTAGCAGAGAGAGAAGAAGCCCTG-3′ (for K201R), 5′-ATGGGACAGATTAGAAGTGAGCCACCC-3′ (for K275R), 5′-GAAACTCTCTGTAGGCAGGAGGGAGAG-3′ (for K420R), and 5′-CGACCCCAGATTAGATGTGAGCAGTCT-3′ (for K579R) (the mutation is underlined in each sequence). Mutated cDNAs were verified by sequencing, isolated as a BamHI DNA, and ligated with BamHI-digested pcDNA3.1BFLAG, resulting in an expression plasmid, pcDNABach2KR, that encodes a FLAG-tagged Bach2KR. BTB-bZip-CLS cDNA was constructed by ligating the KpnI-HindIII fragment from B1B2(A), encoding the Bach2 bZip and downstream region, and the BamHI-EcoRI fragment from B1B2(B), encoding the BTB domain of Bach2. The hemagglutinin (HA)-SUMO-1 (52) was kindly provided by A. Kikuchi, and green fluorescent protein-SUMO-1 (38) and HA-ubiquitin (44) plasmids were gifts from J. J. Palvimo and L. Poellinger, respectively. Cells were transfected with each expression vector with a FuGENE6 transfection reagent (Roche).

Detection of Bach1, Bach2, Bach2 mutants, PML, CREB-binding protein (CBP), p300, SC35, and transcription-labeled BrUTP.

Cells were fixed with 4% paraformaldehyde in 1× PBS. Next, nuclei were permeabilized with 0.1% sodium dodecyl sulfate (SDS)-0.5% Triton X-100-1× PBS for 5 min. For the detection of BrUTP, cells were incubated for 30 min at 37°C with rat anti-bromodeoxyuridine (Harlan) diluted at a 1:200 ratio in 1% bovine serum albumin-1× PBS. For the detection of PML, CBP, p300, and SC35, fixed cells were incubated with rabbit anti-PML (1:200; Santa Cruz), mouse anti-PML (1:200; Santa Cruz), anti-CBP (1:100; Santa Cruz), anti-p300 (1:100; Santa Cruz), and anti-SC35 (1:200; Sigma) antibodies, respectively, diluted in 1% bovine serum albumin-1× PBS. Bach2 and Bach2 mutants were detected with rabbit anti-Bach2 antiserum (F69-2, 1:200) (37) or mouse anti-FLAG (M2, 1:200; Sigma) antibodies. Fluorescein isothiocyanate- or Cy3-conjugated sheep anti-mouse (Dianova), Cy3-conjugated goat anti-rat (Rockland), and fluorescein isothiocyanate- or biotin-conjugated goat anti-rabbit (Biosource) were used as secondary antibodies. Avidin-Cy5 (Dianova) was used for the detection of biotin-conjugated goat anti-rabbit antibody. All antibody incubations were performed at 37°C for 30 min. Nuclei were stained with 10 μM Hoechst 33342.

Image acquisition.

Samples were examined with a Zeiss LSM510 confocal laser scanning microscope or a Leica DMRE epifluorescence microscope equipped with a charge-coupled device camera controlled by QFluoro software (Leica). For the three-dimensional (3D) reconstruction of the confocal sections, Amira (TGS) was used. Adobe Photoshop was used for the presentation of images.

Photobleaching experiments.

Fluorescence recovery after photobleaching (FRAP) was performed with a Zeiss LSM510 confocal laser scanning microscope with a 63×/1.4 plan-apochromat objective. Cells on round coverslips were transferred to a live-cell chamber FCS2 (Bioptecs), mounted on the microscope stage, and kept at 37°C. The objective was operated with an objective heater as part of the FCS2 system. The fluorescence recovery of HcRed-Bach2, bleached five times with a 543-nm HeNe-laser (0.5 mW) at 100% power, was monitored at 1% power, with time intervals as indicated in Fig. 2. Relative intensities in the bleached area were measured and normalized by using the average intensity before bleaching.

FIG. 2.

FRAP analyses of Bach2 in GM02063 cells. Cells expressing HcRed-Bach2 were subjected to a local bleach pulse, and the kinetics of fluorescence recovery in the bleached area was determined. (A to D) Examples of the primary data obtained with the photobleaching protocol. Arrows indicate photobleached Bach2 nuclear foci. HcRed-Bach2 signals before photobleaching (A), just after photobleaching (B), and 2 min after photobleaching (C) are shown. (D) A merged image of the HcRed signal and a transmission image are shown. Bar, 5 μm. (E) Quantitative FRAP analysis of HcRed-Bach2 nuclear foci. Recovery of fluorescence was measured at the indicated time points after the bleach pulse. All data points represent the mean of data from five different measurements, and the error bars indicate twice the standard error.

Coimmunoprecipitation assays.

Detection of sumoylated and ubiquitinated Bach2 protein was carried out as previously described (44). Briefly, we transiently expressed FLAG-tagged Bach2 or Bach2KR with HA-tagged SUMO-1 (4 μg each) or HA-tagged ubiquitin (4 μg each) in 293T cells in 10-cm-diameter dishes. After 12 h of incubation, cells were treated with 1 mM DEM, 5 μM MG132 (Peptide Institute Inc.), or vehicle alone (100% ethanol) for 2 h before harvesting the cells. The cell pellet was resuspended in 200 μl of whole-cell extract buffer (25 mM HEPES, 100 mM NaCl, 5 mM EDTA, 20 mM β-glycerophosphate, 20 mM para-nitrophenyl phosphate, 0.5% Triton X-100, 100 μM sodium orthovanadate) supplemented with 20 μM N-ethylmaleimide and then centrifuged for 30 min at 600 × g. Eight hundred micrograms of total cell protein was incubated with 4 μl of anti-FLAG M2 (Sigma) at room temperature for 30 min. Subsequently, 40 μl of a 50% slurry of protein G-Sepharose in TEG buffer (20 mM Tris-HCl [pH 7.4], 1 mM EDTA, 10% glycerol, 1 mM dithiothreitol) containing 150 mM NaCl and 0.1% Triton X-100 was added to the reaction mixtures and incubated for 12 h at 4°C under rotation. After rapid centrifugation, the resulting Sepharose pellets were washed four times with supplemented TEG buffer, and coimmunoprecipitated proteins were analyzed by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and then immunoblotted with anti-FLAG or anti-HA (Santa Cruz) as the primary antibodies, diluted at a 1:500 ratio, and anti-mouse immunoglobulin (Ig)-horseradish peroxidase conjugate (Amersham Life Science) as a secondary antibody, diluted at a 1:1,000 ratio in Tris-buffered saline (TBS) containing 0.1% Tween 20 (TBS-T) and 1% nonfat milk for 1 h. After extensive washing with TBS-T buffer, immunocomplexes were visualized by enhanced chemiluminescence (Amersham Pharmacia Biotech) according to the manufacturer's instructions.

For the detection of sumoylated endogenous Bach2 in BAL17 cells, 10⁸ cells were collected, and 1 mg of whole-cell extract was subjected to immunoprecipitation. Immunoprecipitation with anti-Sentrin (SUMO-1; Alexis) or anti-HA and immunoblotting with anti-Bach2 antiserum as a primary antibody and anti-rabbit Ig-horseradish peroxidase conjugate (Amersham Life Science) as a secondary antibody were carried out as described above.

RESULTS

Spatial association of Bach2 foci with PML bodies upon oxidative stress.

A previous study demonstrated close spatial association of Bach2 foci with PML bodies upon oxidative stress in 293T cells (36). To study the dynamics of Bach2 in detail, the distribution of Bach2 in human fibroblast cell line GM02063 cells after transfection of a FLAG-Bach2 expression plasmid was examined. Distribution of Bach2 in GM02063 cells was similar to that in 293T cells (36); while Bach2 was predominantly distributed in the cytoplasm under normal conditions, it showed translocation from the cytoplasm to the nucleus and formed nuclear foci upon oxidative stress (Fig. 1A and B). We noticed the presence of three different types of Bach2 foci with regard to the association with PML bodies. One type showed a distribution that was independent from PML bodies and was found in roughly 25% of cells treated with DEM (Fig. 1C; for a compilation, see Fig. 5E). This type was also found in a small fraction of Bach2-expressing cells without oxidative stress. The second type showed adjacent localization with PML bodies and was found in 25% of Bach2 foci upon oxidative stress (Fig. 1D and 5E). The third type, which represented 51% of the foci observed upon oxidative stress, showed complete envelopment of PML bodies (Fig. 1E to G and 5E). It should be noted, however, that we found no Bach2 foci that showed complete colocalization with PML bodies in spite of their close spatial association. Thus, even in Bach2 foci that surrounded PML bodies, the intensity of the Bach2 signal was lower at the center of the foci where the PML signal showed peak intensity (Fig. 1F). These results suggest that Bach2 is recruited specifically around PML bodies upon oxidative stress and forms nuclear foci distinct from PML bodies.

FIG. 1.

Topological association of Bach2 foci with the PML body. (A to J) GM02063 cells were transiently transfected with a FLAG-Bach2 expression plasmid. Bach2 is visualized in green, and PML (A to G), CBP (H), p300 (I), and SC35 (J) are visualized in red. (A) Bach2 in GM02063 cells showed a predominantly cytoplasmic distribution. (B) Bach2 formed nuclear foci that showed close association with PML bodies after treatment with 1 mM DEM for 2 h. (C) Bach2 foci showed no association with PML bodies. (D) After DEM treatment for 2 h, Bach2 foci showed localization adjacent to PML bodies. (E to G) PML bodies completely surrounded by Bach2 are shown. (F) Two Bach2 foci (subpanels a and b in panel E) are further enlarged. a and b, merged images; c and d, PML bodies; e and f, Bach2. These Bach2 foci show ring-like structures with PML bodies in the center of the foci. (G) 3D reconstruction from light optical sections of the same nucleus. (H and I) Bach2 enveloped CBP (H; red) and p300 (I; red) after treatment with 1 mM DEM for 2 h. (J) Bach2 nuclear foci under normal conditions showed no association with nuclear speckles. (K) NAMALWA cells were treated with 800 μM DEM for 2 h and fixed with 4% paraformaldehyde. Bach2 and PML were detected by indirect immunofluorescence staining and visualized in green and red, respectively. Arrows indicate Bach2 foci showing close spatial association with PML bodies. Bars, 10 μm.

FIG. 5.

The N-terminal regions of Bach2 are essential for focus formation. (A) Schematic representation of chimeric and mutant proteins. Percentages of amino acid identity between subregions are shown between Bach1 and Bach2. (B) GM02063 cells were transfected with either FLAG-Bach2 (Bach2)-, B1B2(A)-, B2B1(B)-, B1B2(C)-, FLAG-Bach1 (Bach1)-, BTB-bZip-CLS-, or FLAG-Bach2ΔBTB (Bach2ΔBTB)-expressing vectors and treated with 1 mM DEM for 2 h. The distribution of these proteins was studied by indirect immunofluorescence staining. Bars, 10 μm. (C) Bar plots of the percentages of cells expressing the indicated proteins show the different types of protein distribution: predominantly cytoplasmic distribution (cytoplasmic), diffuse nuclear distribution (diffuse nuclear), and nuclear foci showing independent localization with PML bodies [PML(−)] or showing close spatial association with PML bodies [PML(+)].

To confirm the association of Bach2 nuclear foci with PML bodies, we examined the colocalization of the transcriptional coactivators CBP and p300, both shown to be recruited to PML bodies (48), with Bach2 nuclear foci by double-staining experiments. As shown in Fig. 1H and I, endogenous CBP and p300 were enveloped by Bach2 after DEM treatment. These results indicate that Bach2 nuclear foci are formed in the nuclear compartment associated with PML bodies but are distinct from nuclear speckles upon oxidative stress.

Next, we examined the topological relationships between nuclear speckles and Bach2 nuclear foci in GM02063 cells. Nuclear speckles are known as splicing factories containing splicing factor SC35 and are distinct from PML bodies. In contrast to CBP and p300, we observed no obvious overlap or colocalization of Bach2 nuclear foci with nuclear speckles before and after DEM treatment (Fig. 1J and data not shown).

Expression of Bach2 is restricted to B-lineage cells (35). We examined the dynamics of intrinsic Bach2 upon oxidative stress in NAMALWA cells, a human B-lineage cell line, by indirect immunofluorescence staining. Bach2 showed a predominantly cytoplasmic distribution with some nuclear foci under normal culture conditions (data not shown). After DEM treatment, cytoplasmic staining diminished and Bach2 formed intense nuclear foci. The endogenous Bach2 foci in NAMALWA cells also showed close spatial association with PML bodies (Fig. 1K).

Next, we addressed the nature of the Bach2 nuclear foci. To study the dynamics of Bach2 in nuclear foci of living cells, we constructed an HcRed-tagged Bach2 expression vector. We confirmed that HcRed-Bach2 showed a predominantly cytoplasmic distribution under normal culture conditions and formed nuclear foci after DEM treatment in GM02063 cells. After DEM treatment, Bach2 started to form small nuclear foci. Then, we analyzed the dynamics of Bach2 in nuclear foci after DEM treatment with FRAP (30, 32). Interestingly, photobleached Bach2 in nuclear foci showed quick recovery of the fluorescence intensity (Fig. 2A to D, arrows). Half the original fluorescence intensity of Bach2 in the nuclear foci was recovered within 60 s. Quantitative data sets are shown in Fig. 2E. These results clearly indicate that Bach2 nuclear foci are dynamic nuclear structures in which Bach2 shows rapid turnover upon oxidative stress.

Transcription repression by Bach2 around PML bodies after DEM treatment.

Several lines of evidence suggest the importance of PML bodies in the regulation of transcription (58). First, newly synthesized RNA is associated with PML bodies (5, 26). Second, various transcription regulators, including AP-1, CBP, RARα, p53, TIF1a, and pRB show biochemical interaction with PML (3, 12, 26, 46, 55). Third, CBP, pRB, and p53 are found in PML bodies (3, 12, 26). In addition, multiple corepressors (c-Ski, N-CoR, and mSin3A) have been shown recently to interact with PML, and this interaction is required for transcriptional repression mediated by the Mad tumor suppressor (22). Among these corepressors, c-Ski and N-CoR form nuclear domains which colocalize with PML bodies. However, it is still not clear whether these transcription factors and cofactors regulate the transcription reaction within PML bodies. For example, it is possible that the PML-dependent regulation is not necessarily taking place within PML bodies, since a significant portion of PML is found outside the nuclear bodies (58). To address this issue, we took advantage of the oxidative stress-induced envelopment of PML bodies by Bach2 foci. Thus, we compared transcriptional activities associated with PML nuclear bodies before and after DEM treatment in cells expressing Bach2.

To visualize transcription activity, we applied the scratch transcription labeling (STL) method. Instead of microinjection (50), we scratched cells with needles in a way similar to the scratch replication labeling method (42) to introduce BrUTP into cell nuclei. BrUTP was added to cell culture medium at a final concentration of 1 mM. Just after the addition of BrUTP, a series of random scratches were made with the tip of a 27-gauge hypodermic needle to facilitate the penetration of BrUTP into cells. After 5 min, cells were fixed with 4% paraformaldehyde. By using this method, nascent RNA synthesized during the 5 min before fixation could be detected (Fig. 3A). We observed strong labeling of the nucleoplasmic region with BrUTP. As reported previously with microinjection of BrUTP, labeling in the nucleoli was weak (50). In the presence of 1 μg of α-amanitin/ml, a concentration that inhibits specifically RNA polymerase II activity, the staining of BrUTP in nucleoplasm was obliterated, and only nucleolar labeling was observed (Fig. 3B). These results indicate that the BrUTP labeling in the nucleoplasm detected mRNA synthesis by the RNA polymerase II. Using STL, we first examined the topological relationships between PML bodies, Bach2 foci, and nascent RNA in GM02063 cells. In each experiment, light optical sections of whole nuclei were taken from 20 cells, including 10 cells not expressing Bach2 on the same coverslip as a control. In GM02063 cells not expressing Bach2, nascent RNA molecules were detected throughout the nucleoplasm, and some of them were clearly overlapped by PML bodies (Fig. 4A to D). The percentages of PML bodies that overlapped the signals of newly synthesized RNA were 91 and 97% in Bach2-expressing and control cells, respectively. The peak signal intensity for PML and nascent RNA showed a good correlation in a profile plot analysis (Fig. 4D and E). The 3D reconstruction of the optical nuclear sections demonstrated the close spatial association of high transcription activity with nearly all PML bodies (Fig. 4F). We observed associations of newly synthesized RNA with the periphery of PML bodies. This finding is consistent with a previous report using electron spectroscopic imaging (5). As described above, Bach2 foci enveloped PML nuclear bodies after DEM treatment (Fig. 4G to J). The profile plot through the nuclei clearly demonstrated the segregation of nascent RNA synthesis and PML bodies upon the accumulation of Bach2 (Fig. 4J and K). No transcription activity was observed in nearly all PML bodies surrounded by Bach2, as shown in Fig. 4L. In contrast, more than 90% of the PML bodies in control cells not expressing Bach2 on the same coverslip showed active transcription in PML bodies (Fig. 4M). We observed no significant difference in the distribution of nascent RNA outside of PML bodies in Bach2-expressing cells with or without DEM treatment. These results strongly suggest that the transcription activity around PML bodies is selectively repressed by the recruitment of Bach2 upon oxidative stress.

FIG. 3.

STL of GM02063 cells. Transcription labeling for 5 min at 37°C with 1 mM BrUTP by STL was carried out with GM02063 cells. After detection of incorporated BrUTP by indirect immunofluorescence staining, DNA was stained with Hoechst 33342. BrUTP and DNA are visualized in red and blue, respectively. (A) A merged image of BrUTP and DNA is shown. (B) An RNA polymerase II inhibitor, α-amanitin (1 μg/ml), was added to culture medium together with 1 mM BrUTP before the scratch. Nucleolar labeling was apparent after RNA polymerase II inhibition. Bars, 10 μm.

FIG. 4.

PML body-specific repression of transcription by Bach2. Transcription labeling for 5 min with 1 mM BrUTP by STL was carried out with GM02063 cells. (A to F) GM02063 cell without oxidative stress. (G to L) GM02063 cell expressing Bach2 after 1 mM DEM treatment for 2 h. Raw images of PML bodies (A and G), Bach2 (B and H), and newly synthesized RNA (C and I) noted in confocal nuclear sections are shown. (D and J) Merged images of PML bodies, Bach2 foci, and newly synthesized RNA in red, green, and blue, respectively. (E and K) Profile plots of signal intensity along the arrows in panels D and J. (F and L) 3D reconstructions and projection from the series of confocal nuclear sections of the same nuclei. (M) Merged image of PML, Bach2, and nascent RNA in a cell not expressing Bach2 on the same coverslip carrying the cell shown in panels G to L. The green signal (Bach2) on the left side of the nucleus is from a neighboring cell. Bars, 5 μm.

The region required for the focus formation of Bach2.

Bach2 is characterized by the presence of two functional domains, the bZip and BTB domains in the C- and N-terminal regions, respectively (37) (Fig. 5A). The bZip domain of Bach2 is required for the dimer formation with the small Maf family factors and is thus essential for the target gene selection. The BTB domain, found in more than 100 human proteins (27), mediates oligomer formation (4, 9, 10), DNA loop generation (54), and nuclear focus formation (1, 9, 10, 41). Bach1, which is evolutionarily related to Bach2, also carries the bZip and BTB domains. However, only Bach2 can induce apoptosis in response to oxidative stress (36). Whereas the bZip domains of Bach1 and Bach2 are interchangeable for the induction of apoptosis upon oxidative stress, the BTB domains are not (36). As described above, we found that Bach2 forms nuclear foci upon oxidative stress. In contrast, Bach1 did not form nuclear foci, showing a predominantly cytoplasmic distribution even after DEM treatment (Fig. 5B and C). The following experiments were designed to determine which region of Bach2 was essential for the focus formation upon oxidative stress.

First, to study the role of the BTB domain of Bach2 in the dynamics of Bach2, the distribution patterns of Bach2ΔBTB and BTB-bZip-CLS, which lack the BTB domain and the region between the BTB and bZip domains (Fig. 5A), respectively, were examined in GM02063 cells. Bach2ΔBTB showed diffuse nuclear localization, indicating that the BTB domain was required for the focus formation of Bach2 (Fig. 5B and C). BTB-bZip-CLS also showed a diffuse predominantly nuclear distribution after DEM treatment (Fig. 5B and C). Taken together, Bach2 requires both BTB and the region between BTB and bZip for the focus formation upon oxidative stress.

Second, to test whether these regions of Bach2 are interchangeable for the focus formation with the corresponding regions of Bach1, we analyzed the dynamics of three Bach2/Bach1 chimeric proteins, shown in Fig. 5A, after DEM treatment in GM02063 cells. The chimeric proteins B1B2(A) and B1B2(C) are the fusion proteins containing the Bach1 N-terminal region with the BTB domain and the Bach2 C-terminal region including the bZip domain (36). The region between the BTB and bZip domains in B1B2(A) is from Bach1, and that region in B1B2(C) is from Bach2. In contrast, B2B1(B) carries the N terminus of Bach2 including the BTB domain and the C terminus of Bach1 including the bZip domain. Sixty percent of cells expressing B1B2(A) showed diffuse nuclear distribution in the absence of DEM (Fig. 5C). After DEM treatment, the percentage of cells with the predominantly nuclear distribution pattern of the protein rose to 90% (Fig. 5C); the distribution was entirely diffuse, with no focus formation (Fig. 5B and C). The chimeric protein B1B2(C) showed mostly predominantly cytoplasmic distribution even after DEM treatment (Fig. 5B). While a small fraction of cells accumulated B1B2(C) in nuclei, the protein did not form nuclear foci. B2B1(B) showed predominantly cytoplasmic distribution in the absence of oxidative stress (Fig. 5C). After DEM treatment, 40% of the cells showed focus formation, and half of them showed close association with PML bodies similar to the patterns observed with wild-type Bach2 (Fig. 5B and C). These results confirmed that the BTB as well as the region between the BTB and the bZip domains are involved in the formation of nuclear foci upon oxidative stress.

Sumoylation of Bach2 is required for the envelopment of PML bodies.

The ubiquitin-like protein SUMO-1 is covalently conjugated with PML and shows colocalization with PML bodies (43, 57). A PML mutant that cannot be modified by SUMO-1 fails to recruit Sp100 and Daxx to nuclear bodies (57). These observations suggest that sumoylation plays an important role in the formation of PML bodies (57).

Mouse Bach2 possesses potential sumoylation sites within the region between the BTB and bZip domains that are conserved in human BACH2 (Fig. 6A). As shown in Fig. 6B, Bach2 nuclear foci showed colocalization with SUMO-1 after DEM treatment. We investigated whether Bach2 is sumoylated in vivo. A fraction of Bach2 was sumoylated when Bach2 and HA-tagged SUMO-1 were coexpressed in 293T cells (Fig. 6C). Since there were at least three bands that were reactive with anti-HA antibody, Bach2 could be sumoylated at several sites. Sumoylation of Bach2 was not affected by the presence or absence of oxidative stress. To determine whether Bach2 was subject to sumoylation at endogenous expression levels, proteins of BAL17, a mature B-cell line expressing Bach2 (35), were immunoprecipitated with anti-SUMO-1 antibody. When immunoprecipitates were analyzed by SDS-PAGE and Western blotted with anti-Bach2 antiserum, bands corresponding to SUMO-1-conjugated Bach2 were visible (Fig. 6D). These bands were not detected when proteins were immunoprecipitated with a control antibody (anti-HA). Taken together, these results indicate that Bach2 is a substrate for sumoylation.

FIG. 6.

SUMO-1 and ubiquitin modification of Bach2 protein. (A) Bach2 possesses four sequences that conform to SUMO consensus. Bach2KR was generated by mutating all four lysine residues in the putative sumoylation sites to arginine. (B) a to d, GM02063 cells were transiently transfected with FLAG-Bach2 and GFP-SUMO-1 expression vectors. After treatment with 1 mM DEM for 2 h, Bach2 formed nuclear foci showing colocalization with SUMO-1. PML (a), Bach2 (b), and SUMO-1 (c) are visualized in red, green, and blue, respectively, in panel d. Bars, 5 μm. e to g, GM02063 cells were transiently transfected with FLAG-Bach2KR. After treatment with 1 mM DEM for 2 h, Bach2KR formed nuclear foci, most of which showed adjacent localization with PML bodies. PML (e) and Bach2KR (f) are visualized in red and green, respectively, in panel g. Bars, 5 μm. (C) FLAG epitope, FLAG-tagged Bach2, FLAG-tagged Bach2KR, and HA-tagged SUMO-1 were expressed in 293T cells in the indicated combinations. a, FLAG-Bach2 and FLAG-Bach2KR (KR) were expressed at similar levels 1 day after transfection; b, cells were treated with 1 mM DEM for 2 h. After immunoprecipitation of whole-cell extracts by anti-FLAG, sumoylated forms of Bach2 (arrowheads) were detected by anti-FLAG immunoblotting (right panel). Shifted bands were confirmed as HA-SUMO-1-conjugated FLAG-Bach2 by reprobing with anti-HA (left panel). Molecular weight markers (in thousands) are shown on the right-hand side of the blots. (D) Proteins extracted from BAL17 cells were precipitated by an anti-SUMO-1 antibody and Western blotted with anti-HA (αHA; lane 1) or anti-Bach2 (αSUMO-1; lane 2) antibodies. Sumoylated forms of endogenous Bach2 are indicated by arrowheads. Lane 3, whole-cell extracts (WCE) blotted with anti-Bach2 antiserum. (E) Average numbers of Bach2/Bach2KR foci showing indicated types of association with PML bodies in a GM02063 cell expressing FLAG-Bach2 (Bach2) or FLAG-Bach2KR (Bach2KR) treated with 1 mM DEM for 2 h. Twenty nuclei were analyzed in each experiment. (F) HA epitope, FLAG-tagged Bach2 (FLAG-Bach2), FLAG-tagged Bach2KR (FLAG-Bach2KR), and HA-tagged ubiquitin (HA-Ubiquitin) were expressed in 293T cells in the indicated combinations. Cells were treated with either 1 mM DEM or 5 μM MG132 for 2 h. After immunoprecipitation of whole-cell extracts by anti-FLAG, ubiquitinated Bach2 was detected by anti-HA immunoblotting. Molecular weight markers (in thousands) are shown on the right-hand side of the blots.

To examine the role of sumoylation of Bach2 in oxidative stress, we mutated all four lysine residues in Bach2, which might be targets of sumoylation (Fig. 6A). The resulting Bach2KR was expressed as a FLAG epitope-tagged protein. Compared to wild-type Bach2, sumoylation of Bach2KR was significantly reduced in 293T cells (Fig. 6C). Having established that Bach2 is sumoylated through at least one of these lysine residues, GM02063 cells were transfected with Bach2KR-expressing plasmids. After DEM treatment, Bach2KR formed nuclear foci like wild-type Bach2 (Fig. 6B). However, the percentage of Bach2KR foci showing complete envelopment of PML bodies was less than 10% (Fig. 6E). More than 90% of Bach2KR foci showed no association or only adjacent localization with PML bodies upon oxidative stress (Fig. 6E). In contrast, 50% of wild-type Bach2 foci showed complete envelopment of PML bodies under the same experimental conditions. These results indicate that Bach2 requires sumoylation to surround PML nuclear domains.

Modification of proteins by the covalent attachment of ubiquitin is known to target them for degradation by proteasomes. In several cases, ubiquitination and sumoylation compete for the same lysine residues (18). We thus examined whether Bach2 is ubiquitinated or not in vivo by cotransfecting FLAG-tagged Bach2 and HA-tagged ubiquitin expression vectors into 293T cells. As shown in Fig. 6F, a fraction of Bach2 was ubiquitinated. Ubiquitination of Bach2 was strongly enhanced by the treatment of cells not only with MG132, a proteasome inhibitor, but also with DEM, indicating that oxidative stress induced ubiquitination of Bach2. Bach2KR, carrying mutations of all four lysine residues which were required for the sumoylation, still showed ubiquitination similar to that of wild-type Bach2 (Fig. 6F). Taken together, Bach2 is ubiquitinated upon oxidative stress but the modification site(s) is different from the sumoylation sites.

Role of sumoylation and the BTB domain of Bach2 in transcription regulation associated with PML bodies.

The results described above indicate that Bach2 represses the transcription activity around PML bodies upon oxidative stress, and the Bach2 BTB domain and sumoylation of the protein play important roles in the recruitment of the protein around PML bodies. To study the role of sumoylation and the BTB domain of Bach2 in transcription regulation around PML bodies, we examined the transcription activity around these structures in cells expressing Bach2KR and Bach2ΔBTB. We performed STL using BrUTP to detect transcription sites in GM02063 cells expressing Bach2KR or Bach2ΔBTB. As described above, Bach2KR nuclear foci failed to envelop PML bodies upon oxidative stress. The profile plot of the signal intensities through the nuclei clearly demonstrated nascent RNA in PML bodies but not in Bach2KR foci (Fig. 7A). More than 90% of PML bodies showed colocalization with nascent RNA, indicating almost no modification of the transcriptional activity associated with the nuclear domain by Bach2KR. Bach2ΔBTB showed only diffuse nuclear distribution and also failed to repress transcription activity associated with PML bodies upon oxidative stress (Fig. 7B). These results demonstrated the importance of sumoylation and the BTB domain of Bach2 for the selective transcription repression around PML bodies upon oxidative stress.

FIG. 7.

Role of sumoylation and the BTB domain of Bach2 in repression of transcription activity around PML bodies upon oxidative stress. Transcription labeling for 5 min with 1 mM BrUTP by STL was carried out in FLAG-Bach2KR- or FLAG-Bach2ΔBTB-expressing GM02063 cells after 1 mM DEM treatment for 2 h. Profile plots of signal intensity in a cell expressing Bach2KR (A) or Bach2ΔBTB (B) along the arrows are shown. Bach2KR (A), Bach2ΔBTB (B), PML, and transcription sites are visualized in green, red, and blue, respectively. Bars, 5 μm.

DISCUSSION

There have been many reports suggesting relationships between higher-order nuclear architecture and nuclear function (7, 25). For example, nuclear compartmentalization was shown to contribute to gene regulation (6). Several transcription regulators involved in tumor suppression, apoptosis, cell growth control, and immune response have been identified as components of PML bodies (58). Transcription coactivators and corepressors have also been found in PML bodies. Furthermore, transcription activity inside or on the periphery of PML bodies has been shown (5, 26). Therefore, PML bodies might be the nuclear domains where a cluster or set of genes is regulated. Considering the involvement of PML bodies in various biological responses, the status of transcription activity topologically associated with PML bodies is important in these responses and may be under dynamic control. However, this hypothesis has not been examined experimentally. Identification of the conditional association of Bach2 and PML bodies allowed us to directly address this issue. Most importantly, upon oxidative stress, the transcription repressor Bach2 was dynamically recruited around PML bodies to selectively suppress the transcription activity around the nuclear domain. To our knowledge, this is the first evidence that a sequence-specific transcription factor regulates the PML body-associated transcription activity.

The Bach2 N-terminal region including the BTB domain is required for focus formation. Especially, we conclude that the BTB domain of Bach2 plays an important role in focus formation since Bach2ΔBTB failed to do so. This conclusion is consistent with observations regarding PLZF. PLZF also carries the BTB domain and forms nuclear foci closely associated with PML bodies (41). At present, however, we cannot generalize the involvement of BTB domains in recruiting these proteins around PML bodies, since Bach1 did not form any nuclear foci (Fig. 5). Furthermore, B1B2(A) and B1B2(C) chimeric proteins which carry the BTB domain of Bach1 failed to form foci as well (Fig. 5). Therefore, it is possible that there are at least two types of BTB domains: one that can mediate nuclear focus formation and another that cannot. Transcription regulation by the former type of factors like Bach2 involves two layers of specificities: one is determined by the specific DNA binding sites (MAREs for Bach2), and the other is the subnuclear addresses of target genes (i.e., the subnuclear positioning). On the other hand, the latter type of factors like Bach1 may regulate gene expression in a more conventional way (i.e., in a way independent of nuclear structure).

In this study, we found that sumoylation of Bach2 is required for the recruitment of the protein around PML bodies upon oxidative stress (Fig. 6). Sumoylation of PML has been proposed to play an important role in the recruitment of nuclear proteins, like Sp100, Daxx, and PML itself, into the nuclear domain (34, 57). These proteins show colocalization with PML bodies. In contrast, Bach2 is recruited around but not inside PML bodies upon oxidative stress. Bach2 surrounded PML bodies and formed ring-like nuclear foci (Fig. 1). Accumulation of Bach2 around preexisting PML bodies may contribute to the formation of such a ring-like structure. Interestingly, PML bodies themselves are also known to have a ring-like structure. A study using electron microscopy revealed that the core of PML bodies is a dense, protein-based structure that does not contain detectable nucleic acid (5). These observations suggest that the nuclear regions where PML and Bach2 accumulate are particular nuclear compartments containing a common scaffold for sumoylated proteins. PML could directly or indirectly bind to the scaffold in these compartments without any stress. Under various conditions, specific regulatory proteins might be recruited to modify the transcription activity of specific genes in the compartments. In the case of oxidative stress, Bach2 is recruited to these compartments and can compete in the scaffold with PML dynamically, resulting in the modification of the transcription activity of the genes in the compartments. Sumoylation of Bach2, as well as PML, might play an important role in the recruitment of not only Bach2 but also other nuclear proteins which bind to Bach2. To clarify the role of these higher-order nuclear structures in nuclear function, further characterization of proteins associated with PML bodies and Bach2 foci and their dynamics will be required. We identified three types of Bach2 foci with regard to their association with PML bodies. These different types of foci may be interconvertible. The other possibility is that they were formed independently. Bach2 in the nuclear foci showed rapid turnover in living cells (Fig. 2), indicating that the nuclear foci are dynamic nuclear structures. This finding supports the former model. The evidence, however, is circumstantial and needs to be corroborated in future studies.

Bach2 is abundantly expressed in the early stages of B-cell differentiation and is suppressed in terminally differentiated plasma cells (35). Upon oxidative stress, Bach2 translocation into nuclei has been observed not only in 293T cells and GM02063 cells, as shown in Fig. 1, but also in NIH 3T3 cells (36) and primary mouse B cells (D. Ikebe, unpublished observation). In yeast cells, the bZip transcription factors yAp-1 and Pap1, which are related to the vertebrate AP-1, play key roles in the oxidative stress responses, and activities of both proteins are also regulated by a conditional nuclear export by Crm1/exportin 1 through their nuclear export signal like Bach2 (24, 45, 53). These findings raise an interesting possibility in the regulation of B-cell development. Positive and negative selections (i.e., induction of apoptosis) of B cells are essential for their proper differentiation. B cells may utilize such an evolutionarily conserved common protein sorting system in oxidative stress response for the effective elimination of cells during differentiation, since Bach2 has the proapoptotic function (36). The other possibility is that Bach2 might be recruited around PML bodies to regulate a specific group of genes involved in B-cell differentiation together with other proteins, such as Bcl-6, also recruited around PML bodies. Further studies on the dynamics of Bach2 within the cell will open new opportunities to examine the relationships between transcription, nuclear structure, and differentiation of B cells.

BACH2 is induced after STI571 treatment in lymphoid CML cells, resulting in the induction of apoptosis of leukemia cells (47). Nuclear translocation of BACH2 leads to increased chemosensitivity to anticancer drugs producing reactive oxygen species in lymphoid leukemia cell lines (20). In this study, we found that Bach2 nuclear foci in NAMALWA cells, a human lymphoid cell line, showed close spatial association with PML bodies after DEM treatment. These results suggest that the transcription modulation associated with PML bodies by BACH2 upon oxidative stress might play an important role in the cytotoxic effect of these drugs on lymphoid leukemia cells. Targeting the transcriptional activity associated with PML bodies might be a novel rationale in developing a new anticancer therapy.