Transcription Factor BACH1 Is Recruited to the Nucleus by Its Novel Alternative Spliced Isoform*

The transcription factor Bach1 is a member of a novel family of broad complex, tramtrack,bric-a-brac/poxvirus and zinc finger (BTB/POZ) basic region leucine zipper factors. Bach1 forms a heterodimer with MafK, a member of the small Maf protein family (MafF, MafG, and MafK), which recognizes the NF-E2/Maf recognition element, a cis-regulatory motif containing a 12-O-tetradecanoylphorbol-13-acetate-responsive element. Here we describe the gene structure of human BACH1, including a newly identified promoter and an alternatively RNA-spliced truncated form of BACH1, designated BACH1t, abundantly transcribed in human testis. The alternate splicing originated from the usage of a novel exon located 5.6 kilobase pairs downstream of the exon encoding the leucine zipper domain, and produced a protein that contained the conserved BTB/POZ, Cap'n collar, and basic region domains, but lacked the leucine zipper domain essential for NF-E2/Maf recognition element binding. Subcellular localization studies using green fluorescent protein as a reporter showed that full-length BACH1 localized to the cytoplasm, whereas BACH1t accumulated in the nucleus. Interestingly, coexpression of BACH1 and BACH1t demonstrated interaction between the molecules and the induction of nuclear import of BACH1. These results suggested that BACH1t recruits BACH1 to the nucleus through BTB domain-mediated interaction.

Unlike other CNC members, Bach1 and Bach2 posses BTB/ POZ (broad complex, tramtrack, bric-a-brac/poxvirus and zinc finger) domains (23), which have been implicated in transcription repression by transcription factors BCL6 and PLZF, involved in cases of non-Hodgkin lymphoma and acute promyelocytic leukemia, respectively (24,25). These factors interact with the corepressors N-CoR and SMRT via the BTB/POZ domain (26). Transcription factors that contain BTB/POZ domains are of particular interest as they are thought to play a variety of structural and organizational roles (27). For example, Drosophila Mod (mdg4) and GAGA factors are involved in chromatin organization and transcription regulation through the establishment of higher order domains (28 -30). Bach1-MafK heterodimers generate higher order complexes through the Bach1 BTB domains (31), and the resultant complex binds target DNA sequences consisting of multiple MAREs, generating DNA loops (32).
The BTB/POZ domain acts as a specific protein-protein interaction domain, with most factors forming homo-and/or hetero-oligomers via BTB/POZ domain binding (24,33). The Drosophila pipsqueak (psq) gene encodes a BTB/POZ-containing protein required for early oogenesis (34). This gene is large and complex, encoding multiple transcripts and protein isoforms. The structure of the gene product (PsqA) and the nature of the truncated form (psq PZ fusion protein) suggest that these proteins interact directly with other proteins, including Psq isoforms, through BTB domains. However, no BTB/POZ gene family members that encode multiple protein isoforms have been reported in mammals.
In the present study, we identified an alternatively spliced isoform of human BACH1 (designated BACH1t). The BACH1t isoform encoded a truncated form of BACH1 protein that lacked the leucine zipper domain, but retained the BTB domain, CNC domain, and basic region. Although both BACH1 and BACH1t transcripts were expressed in all tissues examined, both were most abundant in testis. In transfected cells, full-length BACH1 was localized to the cytoplasm, whereas BACH1t accumulated in the nucleus. BACH1t formed heterooligomers with BACH1 through the BTB/POZ domains, and induced nuclear accumulation of BACH1. Thus, the subcellular localization of BACH1 is regulated by its non-DNA-binding isoform BACH1t.

MATERIALS AND METHODS
Isolation of Human BACH1t cDNA and BACH1 Gene-A human testis cDNA library (CLONTECH) was plated on 150 mm Petri dishes at a density of 5 ϫ 10 4 plaque-forming units/plate, and 1 ϫ 10 6 plaqueforming units screened with the 1.5-kb PstI fragment of human BACH1 cDNA. Duplicate filters were made from each plate and hybridized to the probe in 6ϫ SSC and 0.25% skim milk at 55°C overnight. Following washing under high stringency conditions, positive clones were identified and purified by two additional rounds of plaque hybridization. All isolated clones were directly sequenced using primers gt11ϩ (5Ј-CAGCGAATTCGGTGGCGACGACTCCTGGA GC-3Ј) and gt11Ϫ (5Ј-CAGCGAATTCTTGACACCAGACCAACTGGTA-3Ј), the longest positive clone insert (clone 97) subcloned into the EcoRI site of pBluescript KS (ϩ) (Stratagene) and both strands sequenced using an ABI PRISM cycle sequencing kit (Applied Biosystems). Genomic clones encoding BACH1 were isolated by screening a commercial P1 phage human genomic library (Genome inc.) by polymerase chain reaction (PCR) analysis using primers B1genoϩ (5Ј-AGGATGCTGCTCTGGCCTT GC-3Ј) and B1genoϪ (5Ј-CTACTATCTTCCCTGGTGCCC-3Ј). Following mapping of the restriction enzyme sites of the positive clone, exon-intron boundaries were determined by Southern blot analysis and DNA sequencing.
Plasmid Construction-Eukaryotic expression plasmids, GFP fusion protein expression plasmids, and FLAG-tagged expression plasmids were generated by ligating BACH1 or BACH1t cDNA into BamHI and XhoI sites of pcDNAI/Neo (Invitrogene), EcoRI and BamHI sites of pEGFP-NI (CLONTECH), and EcoRI and SalI sites of pCMV2 (Sigma), respectively. To construct prokaryotic expression plasmids, BACH1 and BACH1t cDNA fragments encoding the CNC-basic region were inserted into pMAL-c2 (New England Biolabs). The resulting plasmids encoded the C-terminal 187 and 74 amino acids of BACH1 and BACH1t, respectively.
5Ј-Rapid Amplification of cDNA Ends (5Ј-RACE)-5Ј-RACE was carried out using human testis marathon-ready cDNA (CLONTECH) as template and primers designed to amplify the 5Ј-end of BACH1t (adapter primer 1, 5Ј-CCATCCTAATACGACTCACT ATAGGGC-3Ј; adapter primer 2, 5Ј-ACTCACTATAGGGCTCGGCGGC-3Ј; T-GSP-A; and T-GSP-A2, 5Ј-TGCAACACTACTATCTTCCCTGGTGCCC-3Ј). PCR was carried out according to the supplier's recommendations. The first reaction was performed using adapter primer 1 and T-GSP-A, and incubated at 94°C for 30 s, followed by 5 cycles at 94°C for 5 s and 72°C for 4 min; 5 cycles at 94°C for 5 s and 70°C for 4 min; 20 cycles at 94°C for 5 s and 72°C for 4 min. Nested PCR was then performed using adapter primer 2 and T-GSP-A2 and incubated at 94°C for 30 s, followed by 5 cycles at 94°C for 5 s and 72°C for 4 min, and 20 cycles at 94°C for 5 s and 70°C for 4 min. Nested PCR products were subcloned to pCRII (Invitrogene) plasmids and DNA sequenced.
Northern Hybridization-Northern blots containing multiple human tissue RNA samples were purchased from CLONTECH. Each tissue sample contained 2 g of poly (A) ϩ RNA. Blots were hybridized with 32 P-labeled human BACH1 or rat glyceraldehyde-3-phosphate dehydrogenase cDNA probes, and subsequent washings carried out as described previously (17).
Transient Transfection-QT6 cells (36) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and seeded in six-well dishes 24 h before transfection. Following replacement of the cell culture medium with Opti-MEM (Life Technologies, Inc.), cells were transfected with GFP fusion, Flag-tagged protein, or nontagged protein expression vectors by lipofection using LipofectAMINE (Life Technologies, Inc.) according to the supplier's recommendations.
Determination of the DNA-binding Consensus Sequence-Binding site selection and PCR amplification were performed as described previously (37) with some modifications. Briefly, 69-base pair singlestranded synthetic oligonucleotides were prepared with the sequences 5Ј-GCGGATCCTGCAGCTCGAC(n 30 )GTCGACAAGCTTCTAGAGCA-3Ј, where n 30 represents a stretch of 30 random base pairs. Double-stranded oligonucleotides were prepared with specific forward (5Ј-GCG-GATCCTGCAGCTCGAG-3Ј) and reverse primers (5Ј-TGCTCTAGAAG-CTTGTCGAC-3Ј) and the double-stranded oligonucleotide pool then incubated with recombinant maltose-binding protein (MBP)-BACH1, MBP-BACH1t, or MBP-LacZ fusion proteins bound to amylose resin in binding reaction buffer (25 mM HEPES, 100 mM KCl, 1 mM EDTA, 10 mM MgCl 2 , 5% glycerol, 1 mM dithiothreitol, 0.2 mg/ml poly(dI-dC), 0.2 mg/ml bovine serum albumin) on ice for 30 min. The resins were washed three times in binding reaction buffer and boiled to release bound oligonucleotides into the supernatant. The recovered oligonucleotides were amplified by PCR with the same forward and reverse primers. After this procedure had been repeated five times, the resultant PCR products were cloned into the pCRII (Invitrogen) plasmid vector, and nucleotide sequences of the inserts determined.
Immunofluorescence and GFP Fusion Protein Analysis-Cells transfected with GFP-tagged protein or other expression vectors were incubated for 16 h on coverslips. Cells were then fixed in PBS containing 4% formaldehyde for 30 min, followed by 0.1% Triton X-100 treatment for 15 min at room temperature. Following treatment with blocking reagent (Dako) for 5 min, cells were stained with 10 g/ml biotin-labeled M2 anti-FLAG monoclonal antibody (Sigma) for 30 min at room temperature, washed five times in Tris-buffered saline containing 1 mM CaCl 2 (TBS/Ca), and stained with streptavidin-conjugated R-phycoerythrin (Dako) diluted 1:20 for 10 min at room temperature. After five washes with TBS/Ca, fluorochrome-labeled cells expressing GFP fusion proteins were visualized using a microscope equipped with fluorescence optics (Nikon) and images recorded using SensiaII Fujichrome film.

Molecular Cloning of the BACH1t and BACH1
Genes-In an attempt to clone the human homologue of murine Bach1, we screened a human endothelial cell cDNA library under nonstringent hybridization conditions with a 1.4-kb NcoI fragment of murine Bach1 cDNA as a probe, and isolated seven positive clones. Upon Southern blotting and sequence analysis, three clones (63, 67, and 71) were found to encode human BACH1, based on sequence similarity to murine BACH1. Clone 63 con- contained longer 3Ј-and 5Ј-noncoding regions than clone 63, coding region sequences were completely identical. Comparison with cDNA sequences isolated by Blouin et al. showed eight coding region mismatches, resulting in two amino acid differences at positions 158 and 171 (Thr 3 Ser and Gly 3 Glu, respectively). These differences may be due to allelic variation.
Northern blot analysis using a multitissue blot (CLON-TECH) and a 1.5-kb PstI fragment of clone 63, including CNC-bZip region sequence, as a probe revealed the existence of a smaller sized transcript in testis, as reported by the previous studies (data not shown) (38,39). To clone this molecule, we screened a human testis cDNA library using the PstI fragment of clone 63 as a probe. Southern blot and DNA sequence analysis revealed that the eight isolated clones lacked sequences encoding the C-terminal region that contained the leucine zipper domain and instead possessed a novel sequence (Fig. 1, A  and B). We designated the truncated sequence as BACH1t. Eleven other isolated clones obtained by the same screening method included the leucine zipper domain. As none of the isolated clones were full-length, we performed 5Ј-RACE using the Marathon cDNA amplification kit (CLONTECH) and primer T-GSP-A, derived from BACH1t-specific sequence, to obtain the complete BACH1t ORF. As shown in Fig. 1C, the sequence of the 5Ј-noncoding region from the isolated BACH1t clones was different to previously reported sequences (38,39).
To eliminate the possibility that the smaller transcript was an artifact of DNA manipulation or PCR amplification, we isolated a genomic clone encoding BACH1 by PCR using the primers B1genoϩ and B1genoϪ (Genome inc.). The gene structure and exon-intron boundaries of BACH1 were determined by Southern blot hybridization and DNA sequence analysis (Fig.  1, D and E) and revealed that all spliced sites conformed to the GT/AG rule (40). Analysis of the structure of the 3Ј-region BACH1t-specific exons revealed four exons (exons VI, VII, VIII, and IX) located downstream of exon V, which encoded the leucine zipper domain (Fig. 1D). The 5Ј-noncoding region of the BACH1t clones was located 16 kb upstream of the exon that contained the translation initiation codon. A data base search for related sequences identified a human genomic sequence (GenBank accession no. AF124731) that covered a 108-kb region. We identified three segments within this sequence that were identical to previously reported 5Ј-noncoding sequences of BACH1 (38,39) and to BACH1t, indicating that the segments were alternative exon I or noncoding exon sequences for the BACH1 gene. We designated these exons described by Ohira et al. (38), Blouin et al. (39), and by this present study, as Ia, Ib, and Ic, respectively (Fig. 1D).
BACH1t Is Ubiquitously Transcribed-To examine the expression patterns of BACH1 and BACH1t in various tissues, we performed RT-PCR analysis using combinations of a primer located in exon IV (primer B1-E4) with primers in exons V (primer F-GSP-A) or VI (primer T-GSP-A), using a multitissue cDNA panel (CLONTECH) as template. Although both BACH1 and BACH1t were transcribed in all tissues examined, expression profiles varied depended on tissue type ( Fig. 2A). Both BACH1 and BACH1t mRNAs were most abundantly expressed in the testes.
Because BACH1 cDNAs have three different 5Ј-noncoding sequences, we analyzed the tissue specificity of these putative first exons by RT-PCR with each isomer-specific primer and a primer common to all isomers located in exon V (Fig. 2B). Results showed that, although the alternate first exons were all transcribed, tissue-specific usage remained unclear.
Northern analysis showed that the short mRNA form was highly abundant in testis, but undetectable in other tissues (38,39). In contrast, the long form was of similar abundance in several tissues, but higher in spleen and leukocytes than in testis. Thus, these results appeared to be disconcordant with the PCR results (see Fig. 2A). To clarify whether BACH1 and BACH1t accounted for the long and short forms, respectively, we performed Northern blot hybridization on multitissue blots (CLONTECH) probed with an ϳ300-base pair PstI/EcoRI fragment of BACH1 cDNA (clone 63), corresponding to exon V, as a BACH1-specific probe. Unexpectedly, the results showed that the short form as well as the long form was detectable in testis (Fig. 2C). Northern blot hybridization was performed using BACH1t-specific probes, but specific bands were not detected due to high background. Although BACH1t could not be distinguished from BACH1 by Northern blot analysis, results suggested that the long RNA form contained both BACH1 and BACH1t sequence.
BACH1t Does Not Recognize NF-E2/MARE-To examine the DNA binding properties of BACH1t, we performed gel retardation analysis using labeled double-stranded oligonucleotides containing NF-E2 motifs and proteins extracted from QT6 fibroblasts transfected by various combinations of BACH1, BACH1t, MAFK, and MAFG expression constructs. As shown in Fig. 3, the BACH1 and small MAF heterodimer recognized the NF-E2 motif (Fig. 3, lanes 7 and 8). In contrast to BACH1, no interaction between BACH1t and NF-E2 sites was observed, irrespective of the presence of small MAFs (Fig. 3, lanes 9 and  10).
To investigate the possibility that BACH1t recognized a unique motif, we then attempted to determine the optimal binding site for BACH1t by PCR-assisted selection of binding sites. Fusion proteins of the MBP-C-terminal fragment containing the basic region of BACH1t, or the MBP-C-terminal region containing the bZip domain of BACH1 were used for five rounds of selection. Although BACH1 exhibited binding preferences toward oligonucleotides containing the TRE consensus sequence, BACH1t showed no obvious binding activity to specific DNA sequences (Fig. 4). These results suggest that BACH1t is a non-DNA-binding isoform.
BACH1t Associates with BACH1-MafK Heterodimer through the BTB Domain-Previous studies have shown that Bach1 BTB domains mediate protein-protein interactions (31). To clarify interactions between BACH1t and the BACH1/small MAF heterodimer, we performed coimmunoprecipitation analysis using fibroblasts transfected with FLAG-labeled BACH1t expression construct. Cotransfections were performed using various combinations of FLAG-labeled BACH1t, murine Bach1-MafK fusion protein (B1K), and Bach1 lacking BTB domain-MafK fusion protein (⌬BTB) expression vectors into fibroblasts by lipofection. When cell lysates derived from fibroblasts expressing FLAG-labeled BACH1t and B1K were immunoprecipitated by anti-FLAG antibody, anti-Bach antibody (F69-1) recognized the immunoprecipitated B1K (Fig. 5, lane  3). However, after cotransfection with ⌬BTB, the fusion protein was not detected (Fig. 5, lane 5). These results suggested that interaction occurred between BACH1t and the BACH1-MafK heterodimer through the BTB domains.
BACH1t Mediates Nuclear Import of BACH1-To examine the interaction between BACH1t and BACH1 at the subcellular level, GFP-labeled BACH1 or GFP-BACH1t, BACH1t expression vectors were transfected into fibroblasts and immunofluorescence analysis performed. Unexpectedly, BACH1 and BACH1t were distributed separately in the transfected QT6 cells, with the former cytoplasmically localized and excluded from the nucleus, and the latter accumulated in the nuclei (Fig.  6A). However, when the FLAG-MAFK expression vector was transfected to QT6, MAFK was accumulated in the nucleus as expected.
To further examine the interaction between BACH1t and BACH1, we cotransfected both expression constructs into QT6 cells. As shown in Fig. 6B (a and b), although BACH1t local-ization was not affected, significant accumulation of GFP-labeled BACH1 was observed in the nucleus. To exclude the possibility that this was due to the presence of the GFP label, we performed the same transfection assay using FLAG-tagged BACH1 and GFP-labeled BACH1t, and GFP-labeled BACH1 and nontagged BACH1t, with the same changes in BACH1 localization observed in each experiment (data not shown). We then performed cotransfection of expression vectors for GFPlabeled BACH1 and FLAG-labeled BACH1t lacking BTB domain (FLAG-⌬BTB BACH1t) to examine whether nuclear accumulation of BACH1 depended on the interaction through the BACH1 and BACH1t BTB domains. As expected, the nuclear accumulation of BACH1 alone was markedly reduced compared with cotransfection of GFP-labeled BACH1 and FLAGlabeled BACH1t expression vectors (Fig. 6B (c and d)). Nevertheless, a small amount of BACH1 still accumulated in nuclei and therefore we could not exclude the possibility that BACH1 and ⌬BTB BACH1t interacted with each other outside the BTB domains. Indeed, we have previously found a weak BACH1 interaction mediated through a region between the BTB and bZip domains. A, BACH1t did not show any obvious binding preference to specific DNA sequences after five rounds of selection, although BACH1 exhibited binding preferences toward oligonucleotides containing TRE consensus. B, sequence alignment of 27 independent clones such that the centers of the palindromic sequences are occupied by a purine. The consensus site, below, was derived by selecting nucleotides at each position that were present in more than 70% (uppercase) or 30% (lowercase) of clones.
Because the DNA binding activity of BACH1 requires the presence of small MAF proteins, we examined the subcellular localization of BACH1 in the presence of MAFK. When GFPlabeled BACH1 and FLAG-tagged MAFK were coexpressed in QT-6 fibroblasts, a small amount of BACH1 accumulated in the nucleus (Fig. 6B, e and f). However, MAFK protein, localized in the nucleus in the absence of BACH1, showed very similar distribution to BACH1 when coexpressed, such that MAFK was also more abundant in the cytoplasm. These results suggest that interactions between small MAF and BACH1 influence subcellular localization. These observations were consistent with the results obtained by electrophoretic mobility shift assays, which showed that significant NF-E2 binding activities existed in nuclear extracts when small MAF and BACH1 expression vectors were cotransfected (Fig. 3). However, we cannot exclude the possibility that more NF-E2 binding activity existed in the cytoplasm of QT-6 cells. DISCUSSION In the present study, we described the isolation and characterization of an alternatively spliced BACH1 isoform, BACH1t, that originated from the usage of a novel exon located downstream of the exon encoding the leucine zipper domain. The BACH1t isoform encoded a truncated form of BACH1 protein that lacked a leucine zipper domain. Although both BACH1 and BACH1t transcripts were abundantly expressed in human testis, expression was observed in all tissues tested. In transfected cells, BACH1 was localized to the cytoplasm, whereas BACH1t accumulated in the nucleus. BACH1t formed heterooligomers with BACH1 through the BTB/POZ domains and induced the nuclear accumulation of BACH1. These results suggested that BACH1t may play an important role in the control of the subcellular localization of BACH1.
Characterization of human BACH1 demonstrated a complex gene structure, with multiple differentially spliced transcripts and protein isoforms. The present results and data base information indicated that the BACH1 gene has three alternate exon I or noncoding exon sequences (Fig. 1D), and RT-PCR analysis showed that all three alternate exons were transcribed (Fig. 2B). Such complex structures of genes encoding BTB/POZ proteins have not been described previously in mammals.
Recently, Hoshino et al. found that Bach2 was mainly localized in the cytoplasm through its evolutionarily-conserved Cterminal cytoplasmic localization signal (CLS) (41). The CLS directs leptomycin B-sensitive nuclear export of reporter proteins through the binding of leptomycin B to the nuclear exporter carrier protein Crm1/exportin 1. This inhibits a nuclear export signal, leading to Crm1/exportin 1-dependent nuclear export. As oxidative stressors diminished CLS activity and induced nuclear accumulation of Bach2, the resulting rapid nuclear accumulation may allow Bach2 to regulate gene expression in response to oxidative stress. The functional significance of the CLS is reinforced by the fact that this region is highly conserved between Bach1 and Bach2, which suggests that Bach1 activity may also be regulated by a similar mechanism. In the present study, we found that BACH1 was localized cytoplasmically in transfected cells, while the naturally occurring truncated form of BACH1, which lacks the CLS, was found to be localized in the nucleus. Furthermore, our results demonstrated that BACH1t recruited BACH1 to the nucleus. It is possible that the subcellular localization of BACH1 is determined by the net balance of nuclear localization signal (NLS) and CLS activities. Thus, if one molecule of BACH1 associated with one molecule of BACH1t through their BTB domains, the resulting heterodimer would have two NLS and one CLS, such that the NLS activity would be stronger than that of the CLS. The present data also showed that small MAF protein promoted the accumulation of BACH1 in nuclei, but to a far lesser extent than compared with BACH1t (Fig. 6). Interestingly, BACH1 influenced the subcellular localization of MAFK but not of BACH1t, suggesting that there are differences in NLS activity between BACH1t and MAFK. These results indicated that there may be at least two different pathways in the control of BACH1 subcellular localization: an oxidative stress pathway and a BACH1t pathway. To our knowledge, the control of subcellular localization by heterodimer formation with an alternate isoform has not been described for other CNC family members.
Recently, it was reported that the BACH1 gene localized to an ϳ400-kb region on 21q22.1 (38,39), within the potential Down's syndrome-associated gene region proposed by Korenberg et al. (42). Furthermore, the BACH1 gene is part of the APP-SOD1 region involved in some of the features associated with monosomy 21. Down's syndrome, the most common birth defect causing mental retardation, is characterized by specific phenotypes including subfertility or sterility and hypogonadism in males. In contrast, several women with Down's syndrome have given birth to offspring. It has been proposed that the effect of trisomy 21 on spermatogenesis and fertility is a consequence of the behavior of the extra chromosome during the meiotic prophase (43). However, an abundant expression of BACH1 and BACH1t in testis suggests that these molecules may have an important role in this tissue and raises the possibility that overexpression of BACH1 and BACH1t caused by the presence of three gene copies may lead to some aspects of the Down's phenotype, including hypogonadism or sterility. Analysis of transgenic mice with an extra copy of BACH1 or a targeted BACH1 gene knockout may provide further information to elucidate its function and involvement in Down's syndrome and monosomy 21.