Bach1 Is Critical for the Transformation of Mouse Embryonic Fibroblasts by Rasv12 and Maintains ERK Signaling

ORIGINAL ARTICLE Bach1 is critical for the transformation of mouse embryonic ﬁbroblasts by RasV12 and maintains ERK signaling

A Nakanome1,2, A Brydun1,3, M Matsumoto1,4,KOta1, R Funayama5,6, K Nakayama5, M Ono7, K Shiga2, T Kobayashi2 and K Igarashi1,3,6

Reactive oxygen species (ROS), by-products of aerobic respiration, promote genetic instability and contribute to the malignant transformation of cells. Among the genes related to ROS metabolism, Bach1 is a repressor of the oxidative stress response, and a negative regulator of ROS-induced cellular senescence directed by p53 in higher eukaryotes. While ROS are intimately involved in carcinogenesis, it is not clear whether Bach1 is involved in this process. We found that senescent Bach1-deficient mouse embryonic fibroblasts (MEFs) underwent spontaneous immortalization the same as did the wild-type cells. When transduced with constitutively active Ras (H-RasV12), the proliferation and colony formation of these cells in vitro were markedly reduced. When transplanted into athymic nude mice, the growth and vascularization of tumors derived from Bach1-deficient cells were also decreased. Gene expression profiling of the MEFs revealed a new H-RasV12 signature, which was distinct from the previously reported signatures in epithelial tumors, and was partly dependent on Bach1. The Bach1-deficient cells showed diminished phosphorylation of MEK and ERK1/2 in response to H-RasV12, which was consistent with the alterations in the gene expression profile, including phosphatase genes. Finally, Bach1-deficient mice were less susceptible to 4-nitroquinoline-1-oxidide (4-NQO)- induced tongue carcinoma than wild-type mice. Our data provide evidence for a critical role of Bach1 in cell transformation and tumor growth induced by activated H-RasV12.

Oncogene (2013) 32, 3231–3245; doi:10.1038/onc.2012.336; published online 30 July 2012 Keywords: Bach1; ROS; cancer; Ras; ERK

INTRODUCTION reduced in Bach1-deficient mice compared with wild-type Reactive oxygen species (ROS) are inevitable by-products of controls,9,10 suggesting that Bach1 determines the ROS levels by aerobic respiration. ROS have conventionally been regarded as fine-tuning the expression of oxidative stress-response genes, having carcinogenic potential and profound associations with including Hmox1. tumor promotion.1 First, ROS are one of the major causes of DNA Intriguingly, Bach1 represses the oxidative stress-induced damage in vivo and increase mutations in DNA.2 Second, the cellular senescence directed by p53.11 Senescence, induced by production of ROS is induced in response to growth factors and the activation of tumor suppressors, inhibits the development of oncogene activation, and can modulate critical intracellular cancer by arresting the proliferation of damaged or stressed cells signaling molecules by affecting their redox regulation.3 Third, that are at risk for malignant transformation.12,13 Bach1 inhibits ROS affect heterotypic cell–cell interactions within tumor tissues p53-mediated cellular senescence by forming a complex with p53, and also affect angiogenesis.4 Therefore, changes in ROS histone deacetylase 1 and nuclear receptor co-repressor, thereby metabolism potentially affect both the prevention and repressing the transcriptional activity of p53.11 Although oxidative promotion of cancer. stress is one of the major causes of the senescence of mouse In higher eukaryotes, the transcriptional responses toward embryonic fibroblasts (MEFs),14 oncogenic Ras with activating oxidative stress are regulated by two major transcription factor mutations also induces cellular senescence.15 Importantly, families, the NF-kB/Rel family and the AP-1 superfamily, which although ROS are essential for Ras-induced cell transformation,16 includes Jun.5 Bach1 is a heme-regulated transcriptional repressor little is known about the involvement of the oxidative stress found in vertebrates and is a member of the AP-1 superfamily.6 response in Ras-driven tumorigeneisis. The regulation of ROS Bach1 forms heterodimers with small Maf oncoproteins (MafK, metabolism and cellular senescence by Bach1 suggest that it may MafF, MafG), binds to the Maf recognition element, which be relevant to tumorigenesis. encompasses an AP-1-binding site,7 and represses the Activating mutations in Ras family proto-oncogenes are very expression of oxidative stress-responsive genes, including heme common in human cancer, and modulators of aberrant Ras oxygenase-1 (Hmox1).8 These genes are activated by Nrf2, which signaling have an important role in tumorigenesis.17 Recently, also belongs to the AP-1 superfamily and forms heterodimers with there have been several reports describing that there is a large the small Maf proteins. Atherosclerosis and ischemic reperfusion group of genes/proteins that are not oncogenes, but if targeted, injury of the heart, both involving ROS-mediated damage, are can cause reduced proliferation of transformed cells.18 These

1Department of Biochemistry, Tohoku University Graduate School of Medicine, Sendai, Japan; 2Department of Otolaryngology-Head and Neck Surgery, Tohoku University Graduate School of Medicine, Sendai, Japan; 3Center for Regulatory Epigenome and Diseases, Tohoku University Graduate School of Medicine, Sendai, Japan; 4Department of Obstetrics and Gynecology, Tohoku University Graduate School of Medicine, Sendai, Japan; 5Division of Cell Proliferation, Tohoku University Graduate School of Medicine, Sendai, Japan; 6CREST, Japan Science and Technology Agency, Sendai, Japan and 7Department of Pathology, Tohoku University Graduate School of Medicine, Sendai, Japan. Correspondence: Professor K Igarashi, Department of Biochemistry, Tohoku University Graduate School of Medicine, Seiryo-machi 2–1, Sendai, Japan. E-mail: [email protected] Received 3 September 2011; revised 7 June 2012; accepted 20 June 2012; published online 30 July 2012 Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3232 phenomena have been interpreted in such a way that activated deﬁciency did not prevent the immortalization of MEFs despite oncogenes rely on genes that are themselves not oncogenes, thus their enhanced propensity for cellular senescence. However, leading to the so-called ‘non-oncogene addiction’.19 For instance, Bach1 was nearly indispensable for the transformation of the Ras family is dependent on the heat shock response in human immortalized cells with activated Ras (H-RasV12) in vitro, and cancer cells.18 enhanced the tumor formation in vivo. Bach1 deﬁciency reduced In this report, we examined the roles of Bach1 in the ROS levels, and profoundly affected the transcriptional response proliferation and ROS metabolism of transformed cells. Bach1 to H-RasV12, including the negative feedback loop of extracellular

Figure 1. Immortalization of Bach1-deficient MEFs. (a) Wild-type and Bach1-deficient primary cell cultures sequentially acquired morphologies typical of senescent and immortal cells (scale bar ¼ 100 mm). (b) The rates of spontaneous immortalization of 25 wild-type and 16 Bach1- deficient independent MEF cultures were monitored. (c) Top, the protein levels of Bach1 in primary and immortalized MEFs. Tubulin was used as loading control. N ¼ 3, the error bars indicate the s.d. Bottom, the result of a typical Western blot analysis is shown. (d) The expression levels of p19ARF mRNA were examined in the primary and two independently immortalized wild-type (black bars) and Bach1-deficient (gray bars) MEF cultures (N ¼ 3, the error bars indicate the s.d.). (e) The expression levels of p53 mRNA were compared as in (d). (f) The proliferation rates of immortalized Bach1-deficient and wild-type cells with high or low p19ARF expression were compared. The data represent typical results of three independent experiments.

Oncogene (2013) 3231 – 3245 & 2013 Macmillan Publishers Limited Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3233 signal-regulated kinase (ERK). These results suggest that Bach1 is a regulate the inhibition of proliferation induced by cell-to-cell critical factor for Ras-induced transformation and may represent a contact. Because the inactivation of p19ARF appeared to be a target gene of non-oncogene addiction by Ras. random event in immortalization, we selected only the i-MEF cultures that had lost p19ARF expression for the further analyses.

RESULTS Bach1 facilitates H-RasV12-induced proliferation and Bach1-deficient MEFs bypass senescence in vitro transformation Senescent MEFs eventually resume proliferation because of the Unlike primary MEFs, i-MEFs can be transformed by single inactivation of p53 and/or p19ARF, resulting in their immortaliza- oncogenes, such as H-RasV12.21 We therefore compared the tion.20 To examine whether Bach1 deficiency would affect this transformation of wild-type and Bach1-deficient i-MEFs by process beyond the premature entry into senescence, we compared H-RasV12. When transduced with H-RasV12,theBach1-deficient the kinetics of immortalization using senescent wild-type and Bach1- i-MEFs produced smaller numbers of colonies in mono-layered deficient MEFs. While senescent cells developed typical phenotypes, culture (Figure 2a). When reconstituted with a Bach1-expressing including flattened morphology and cell cycle arrest, immortalized plasmid, Bach1-deficient i-MEFs formed more colonies (Figure 2a). cells (i-MEFs) regained their typical fibroblastic morphology and Therefore, the observed defect was a direct effect of the ablation of proliferation capacity irrespective of the Bach1 genotype (Figure 1a). Bach1. Similar results were obtained in experiments using p19ARF- Although Bach1-deficient MEFs became senescent earlier than wild- positive cells (data not shown). The overexpression of Bach1 itself in type cells, both of the cell types resumed proliferation with the same wild-type i-MEFs did not induce transformation (data not shown), rate between 30 and 110 days in culture (Figure 1b). Immortalization indicating that Bach1 is not a classical oncogene. The proliferation of did not affect expression level of Bach1 (Figure 1c). Although the Bach1-deficient cells expressing H-RasV12 was slower than that of protein and mRNA levels of p19ARF were higher in Bach1-deficient their wild-type counterparts (Figure 2b), thus indicating that Bach1 than in wild-type cells in primary cultures, there was no correlation was also required for the vigorous proliferation driven by between the p19ARF expression and Bach1 genotype in i-MEFs constitutively active H-RasV12. However, the absence of Bach1 did (Figure 1d, data not shown). The mRNA levels of p53 did not differ not affect the expression of H-RasV12 upon retroviral transduction of significantly between the primary or immortalized wild-type or immortalized cells, and transformation itself did not change the Bach1-deficient cell cultures (Figure 1e). expression levels of Bach1 (Figure 2c). The above results indicated that the senescent phenotype was To examine the Bach1 function under more physiologically reversible in wild-type and Bach1-deficient MEFs, and that this relevant conditions, we transplanted the H-RasV12-transduced cells process involved, at least partly, the inactivation of p19ARF. subcutaneously into the dorsal flanks of athymic nude mice. Interestingly, the p19ARF-negative i-MEFs showed higher satura- Tumors derived from wild-type cells were consistently larger tion densities than those expressing p19ARF, irrespective of the compared with their Bach1-deficient counterparts (Figures 3a–c). Bach1 genotype (Figure 1f), thus suggesting that p19ARF might Moreover, Bach1-deficient tumors showed notable cellular

Figure 2. Bach1-dependent transformation of immortalized MEFs by activated Ras. (a) The results of the colony formation assay in Bach1- deﬁcient or wild-type i-MEFs expressing H-RasV12. p19ARF-negative immortalized cells were used. The effect of Bach1 reconstitution is also shown. (b) The proliferation of wild-type and Bach1-deﬁcient i-MEFs expressing H-RasV12 was compared. The data represent the means and s.d. from three independent experiments (N ¼ 6, *Po0.05). (c) The expression levels of Bach1 and H-Ras in two independent immortalized MEF cultures. A typical result of three independent experiments is shown.

& 2013 Macmillan Publishers Limited Oncogene (2013) 3231 – 3245 Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3234 polymorphism and dyskariosis (Figure 3d), suggesting that Genetic targeting of Bach1 results in a modified cell response to depletion of Bach1 might have induced mitotic stress. Interest- oncogenic RAS ingly, the vascularization was more prominent in tumors derived Based on the observation that Bach1 was essential for the from wild-type cells than from Bach1-deficient cells (Figure 3e). transformed growth of i-MEFs in vitro and in vivo, we hypothesized Taken together, our findings indicate that the cells lacking Bach1 that it might modulate a Ras-dependent transcriptional program were less adapted for cancerous growth in vitro and in vivo. The relevant to cell proliferation. To obtain a global view of the genes overall results suggested that Bach1 is a critical component of that were regulated by Bach1 and dependent on H-RasV12, H-RasV12-mediated transformation. we carried out gene expression profiling of wild-type and

1.2 * 1 0.8 0.6

Weight (g) 0.4 0.2 0 WT Bach1-/-

2.5

) * 3 2

1.5

WT Bach1-/- WT Bach1-/- WT Bach1-/- 0.5 Volume (1000 mm 0 WT Bach1-/-

HE x200

HE x1000

WT Bach1-/-

* 20

m) 15 µ

CD31 10

5 Length/cell (

WT Bach1-/- 0 WT Bach1-/- Figure 3. Bach1-dependent tumor formation by Ras-transduced MEFs. (a) Wild-type and Bach1-deficient i-MEFs were transduced with H-RasV12 and subcutaneously transplanted into the dorsal flanks of athymic nude mice (left, wild-type; right, Bach1-deficient). The weights and volumes (b, c) of the tumors are shown (asterisks indicate Po0.05 and Po0.01, respectively, N ¼ 6). (d) Hematoxylin and eosin staining of tumor tissues. The cells in Bach1-deficient tumors showed notable cellular polymorphism and dyskariosis (scale bar ¼ 50 mm). (e) Left, endothelial cells were detected by CD31 staining (brown). Right, the relative ratios of the blood vessel length to count of nucleated cells (N ¼ 9, *Po0.01, scale bar ¼ 50 mm).

Oncogene (2013) 3231 – 3245 & 2013 Macmillan Publishers Limited Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3235 Bach1-deficient i-MEFs transduced with H-RasV12 or an empty Bach1 is required to maintain signaling via the ERK signal vector. By comparing the gene expression profiles of the transduction pathway immortalized cells with the cells transfected with the empty We found that in Bach1-deficient cells, the transduction of vector, we identified a Bach1 knockout signature comprised of H-RasV12 resulted in the upregulation of a subset of phosphopro- 81 entities that were differentially expressed by at least twofold tein phosphatase genes compared with wild-type i-MEFs (33 upregulated and 48 downregulated entities) in Bach1- (Figure 5a). Only two of these genes (Ptp4a1 and Ppm1b) possess deficient i-MEFs compared with wild-type cells (Figure 4a and a Maf recognition element, and might be direct transcriptional Table 1). Only two genes among those upregulated by Bach1 targets of Bach1. Phosphatases have previously been reported to knockout had Bach1-binding sequences according to the be an important part of a negative feedback loop impending the SABiosciences’ DECODE database (SABiosciences, Frederick, MD, activity of the mitogen-activated protein kinase (MAPK) transduc- USA), thus suggesting that their regulation was mostly indirect. tion pathway. In particular, Dusp1, Dusp9 and Ppp2cb are potent Although Bach1 is known to have an important role in the inhibitors of MAPK activation.30–34 The substrate of Cdc14b is oxidative stress response and heme metabolism, a gene ontology Cdc25c, and its overexpression induces cell cycle arrest by analysis revealed no enrichment of such genes regulating these activating mitotic check point.35 By validating the microarray metabolic processes in i-MEFs. Instead, several genes known to results by qPCR, we confirmed the upregulation of the inhibit cell proliferation (Dusp16/MKP-7, Tcf7l1, Esrp2)22–24 and phosphatase genes, which might contribute to the phenotype angiogenesis (Serpinf1, Tnmd)25,26 were expressed in Bach1- of H-Ras-transformed Bach1-defifient i-MEFs in our experimental deficient cells at higher levels compared with wild-type cells. model (Figure 5b). As the Ras-induced proliferation and cellular Among the targets that were suppressed in Bach1-deficient i-MEFs, transformation of murine cell lines are dependent on the we identified two genes that might directly affect the signal activation of the MAPK cascade,36 we examined the transduction downstream and upstream of Ras, such as H-Ras phosphorylation of ERK1/2. The phosphorylation of ERK1/2 was palmitoyltransferase (Zdhhc18) and GEF-protein Vav1.27,28 clearly induced by H-RasV12 in the wild-type i-MEFs, but not in the The expression of oncogenic H-RasV12 in wild-type i-MEFs Bach1-deficient i-MEFs (Figure 5c). resulted in a dramatic change in the gene expression landscape. We carried out a kinetic analysis of ERK1/2 activation in We identified a group of genes whose expression was induced or response to serum stimulation after starvation. The phosphoryla- suppressed by H-RasV12 in wild-type but not in Bach1-defifient tion of ERK1/2 was markedly induced between 10 and 30 min after i-MEFs (Figure 4b and Table 2). Among a total of 96 genes (8 serum stimulation in wild-type cells. In contrast, it was not obvious upregulated and 88 downregulated), we observed no clear in Bach1-deficient cells (Figure 6a). The phosphorylation of MEK, a overlap with the previously published Ras signature obtained kinase upstream of ERK, showed similar patterns. In contrast, the from an analysis of epithelial tumors.29 Interestingly, in contrast to phosphorylation of Akt, a downstream effector of Ras, was not the wild-type i-MEFs, transformation by H-RasV12 failed to induce induced, and did not differ between wild-type and Bach1-deficient the suppression of many genes involved in regulating the cells (Figure 6a). Interestingly, we observed no differences in the transcription, translation and RNA processing in Bach1-deficient phosphorylation of ERK1/2 and MEK between wild-type and cells (Table 2). These observations suggested that Bach1 modified Bach1-deficient i-MEFs in the absence of H-RasV12 (Figure 6b). The the gene expression profile induced by H-RasV12. phosphorylation of ERK and MEK was clearly observed in Bach1-

Figure 4. Changes in gene expression in response to oncogenic Ras in Bach1-deficient i-MEFs. (a) Heat map visualizations of the 82 genes that were more than twofold differentially expressed genes in Bach1-deficient i-MEFs compared with WT i-MEFs. The data are normalized and represented as median-centered log-transformed values, using average linkage clustering on entities and conditions. Red and blue correspond to increased and decreased expression, respectively, compared with the experiment-wide median. (b) A heat map of the 96 genes expression of which was changed more than twofold in WT but not in Bach1-deficient i-MEFs after transduction with H-RasV12. The data are expression ratios of the normalized gene expression levels with and without H-RasV12, and are presented as median-centered log-space data. Clustering and visualization were performed as described above.

& 2013 Macmillan Publishers Limited Oncogene (2013) 3231 – 3245 Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3236 Table 1. List of more than twofold differentially expressed entities in Bach1-deﬁcient i-MEFs compared with wild-type cells

Probe ID Gene symbol Accession number Description

Upregulated entities A_51_P441745 N/A AK021396 0 day neonate eyeball cDNA, RIKEN full-length enriched library, clone: E130119H09 A_52_P225584 Abhd15 NM_026185 Abhydrolase domain-containing 15 A_52_P166275 Gpr82 NM_175669 G protein-coupled receptor 82 A_51_P513764 Iqcf4 NM_026090 IQ motif-containing F4 A_51_P223656 Lamb3 NM_008484 Laminin, beta 3 A_52_P426740 Rab27a NM_023635 RAB27A, member RAS oncogene family A_52_P134639 Tgoln2 NM_009444 Trans-golgi network protein 2 A_51_P318830 Syt10 NM_018803 Synaptotagmin X

Downregulated entities A_51_P460153 N/A AK031247 13 days embryo forelimb cDNA, RIKEN full-length enriched library, clone:5930437B06 A_51_P287698 N/A AK035950 16 days neonate cerebellum cDNA, RIKEN full-length enriched library, clone:9630020E20 A_52_P358913 N/A AK054418 2 days pregnant adult female ovary cDNA, RIKEN full-length enriched library, clone:E330023J15 A_51_P408729 Phgdh NM_016966 3-phosphoglycerate dehydrogenase A_52_P661606 Phgdh NM_016966 3-phosphoglycerate dehydrogenase A_51_P408732 Phgdh NM_016966 3-phosphoglycerate dehydrogenase A_51_P178319 Arl5b NM_029466 ADP-ribosylation factor-like 5B A_52_P104885 Arl5b NM_029466 ADP-ribosylation factor-like 5B A_52_P43423 N/A AK034063 Adult male diencephalon cDNA, RIKEN full-length enriched library, clone:9330153B03 A_51_P209965 Alkbh4 NM_028070 alkB alkylation repair homolog 4 (E. coli) A_51_P133345 Apln NM_013912 Apelin A_52_P571591 Ash2l NM_011791 ash2 (absent, small or homeotic)-like (Drosophila), transcript variant 1 A_52_P328078 Atp5b NM_016774 ATP synthase, H þ transporting mitochondrial F1 complex, beta subunit A_51_P215815 Atp11c NM_001037863 ATPase, class VI, type 11C, transcript variant 1 A_51_P254541 Bax NM_007527 BCL2-associated X protein A_52_P377703 Brms1l NM_001037756 Breast cancer metastasis-suppressor 1-like A_52_P652104 Brd3 NM_023336 Bromodomain containing 3, transcript variant 3 A_52_P304281 N/A NM_145591 cDNA sequence BC003267 A_51_P144783 Cct5 NM_007637 Chaperonin containing Tcp1, subunit 5 (epsilon) A_52_P504068 Cdk8 NM_153599 cyclin-dependent kinase 8 A_52_P471502 Dcaf13 NM_198606 DDB1- and CUL4-associated factor 13 A_52_P362603 Dscc1 NM_183089 Defective in sister chromatid cohesion 1 homolog (S. cerevisiae) A_51_P386549 Dync1i2 NM_010064 Dynein cytoplasmic 1 intermediate chain 2 A_52_P583458 E2f3 NM_010093 E2F transcription factor 3 A_51_P126437 Enc1 NM_007930 Ectodermal-neural cortex 1 A_51_P223666 Erh NM_007951 Enhancer of rudimentary homolog (Drosophila) A_51_P173086 Eif3l NM_145139 Eukaryotic translation initiation factor 3, subunit L A_51_P141308 Eif3m NM_145380 Eukaryotic translation initiation factor 3, subunit M A_52_P38639 Fermt3 NM_153795 Fermitin family homolog 3 (Drosophila) A_52_P476935 Flrt3 NM_178382 Fibronectin leucine-rich transmembrane protein 3, transcript variant 2 A_51_P204564 Gpaa1 U27838 Glycosyl-phosphatidyl-inositol-anchored protein homolog (S. cerevisiae) A_51_P224445 Gpn3 NM_024216 GPN-loop GTPase 3 A_51_P443482 Hnrnpa3 NM_146130 Heterogeneous nuclear ribonucleoprotein A3, transcript variant b A_52_P431965 Hmgxb3 NM_134134 HMG box domain-containing 3, transcript variant 2 A_51_P209135 Jtb AB016490 Jumping translocation breakpoint A_51_P408071 Kntc1 NM_001042421 Kinetochore-associated 1 A_51_P100181 Med14 NM_001048208 Mediator complex subunit 14, transcript variant 1 A_51_P208355 Mcm8 NM_025676 Minichromosome maintenance deﬁcient 8 (S. cerevisiae) A_51_P178592 Mrpl1 NM_053158 Mitochondrial ribosomal protein L1, transcript variant 1 A_51_P423625 Mrpl21 NM_172252 Mitochondrial ribosomal protein L21 A_51_P493270 Mrps5 NM_029963 Mitochondrial ribosomal protein S5 A_52_P650288 Mrps5 NM_029963 mitochondrial ribosomal protein S5 A_51_P298514 Mapkap1 NM_177345 Mitogen-activated protein kinase associated protein 1 A_51_P144014 Gdap5 Y17854 mRNA for ganglioside-induced differentiation associated protein 5 A_51_P262871 Nans NM_053179 N-acetylneuraminic acid synthase (sialic acid synthase) A_52_P633201 Ncoa4 NM_001033988 nuclear receptor coactivator 4, transcript variant 2 A_51_P163305 Nudt5 NM_016918 nudix (nucleoside diphosphate linked moiety X)-type motif 5 A_51_P330226 Polr3k NM_025901 Polymerase (RNA) III (DNA directed) polypeptide K A_52_P240453 Polr3k NM_025901 Polymerase (RNA) III (DNA directed) polypeptide K A_52_P94150 Gm8096 NR_033590 predicted gene 8096, non-coding RNA A_52_P90124 Prep NM_011156 Prolyl endopeptidase A_52_P271555 Psmf1 NM_212446 Proteasome (prosome, macropain) inhibitor subunit 1 A_51_P270478 Pin4 NM_027181 protein (peptidyl-prolyl cis/trans isomerase) NIMA-interacting, 4 (parvulin) A_52_P360085 Ppp2r3c NM_021529 protein phosphatase 2, regulatory subunit B00, gamma A_52_P585448 Ptp4a2 NM_008974 protein tyrosine phosphatase 4a2, transcript variant 1

Oncogene (2013) 3231 – 3245 & 2013 Macmillan Publishers Limited Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3237 Table 1 (Continued )

Probe ID Gene symbol Accession number Description

A_52_P33498 Rab14 NM_026697 RAB14, member RAS oncogene family A_51_P409496 Arhgap11a NM_181416 Rho GTPase activating protein 11A A_51_P506674 Rpp38 NM_001013376 Ribonuclease P/MRP 38 subunit (human) A_51_P163389 Rpe NM_025683 Ribulose-5-phosphate-3-epimerase A_52_P598634 1190007I07Rik NM_001135567 RIKEN cDNA 1190007I07 gene, transcript variant 1 A_51_P332392 2010305A19Rik NM_027250 RIKEN cDNA 2010305A19 gene A_51_P234888 2310003L22Rik NM_027093 RIKEN cDNA 2310003L22 gene A_52_P549927 5730455P16Rik NM_027472 RIKEN cDNA 5730455P16 gene A_52_P106789 Rbm18 NM_026434 RNA-binding motif protein 18, transcript variant 1 A_51_P437786 Rbm28 NM_133925 RNA-binding motif protein 28 A_52_P281620 Rbm4b NM_025717 RNA-binding motif protein 4B A_52_P277854 Snrpb NM_009225 Small nuclear ribonucleoprotein B A_51_P395743 Spopl NM_029773 Speckle-type POZ protein-like, transcript variant 1 A_51_P297165 Sf3a3 NM_029157 Splicing factor 3a, subunit 3 A_52_P198916 St7l NM_153091 Suppression of tumorigenicity 7-like A_51_P280532 Supt16h NM_033618 suppressor of Ty 16 homolog (S. cerevisiae) A_52_P667215 Ttc19 NM_028360 Tetratricopeptide repeat domain 19, transcript variant 1 A_52_P397012 Tars NM_033074 Threonyl-tRNA synthetase A_51_P127635 Ttf2 NM_001013026 transcription termination factor, RNA polymerase II A_52_P127362 Tmem49 NM_029478 Transmembrane protein 49 A_51_P509229 Tnrc6a NM_144925 Trinucleotide repeat-containing 6a A_51_P378371 Tpm1 NM_001164255 Tropomyosin 1 alpha, transcript variant 9 A_51_P104933 Trub2 NM_145520 TruB pseudouridine (psi) synthase homolog 2 (E. coli), transcript variant 1 A_52_P297731 Ube2i NM_011665 Ubiquitin-conjugating enzyme E2I, transcript variant 1 A_51_P162253 Vdac2 NM_011695 Voltage-dependent anion channel 2 A_51_P211573 Wdr47 NM_181400 WD repeat domain 47 A_52_P134899 Zfp2 NM_001044698 Zinc ﬁnger protein 2, transcript variant 4 A_52_P496163 Gja5 ENSMUST00000107064 Gap junction protein, alpha 5 A_52_P303235 Ndufab1 ENSMUST00000033157 NADH dehydrogenase (ubiquinone) 1, alpha/beta subcomplex 1 A_52_P130899 Nmmhcb AK040822 Cellular myosin heavy chain, type B A_52_P33453 D13Mgi7 XR_001567 Pseudogene A_52_P542970 6720456H20Rik ENSMUSG00000036339 RIKEN cDNA 6720456H20 gene A_52_P60254 N/A NM_172690 Mus musculus expressed sequence N28178 Abbreviations: cDNA, complementary DNA; N/A, not available.

deficient cells. Taken together, these results indicate that Bach1 is phosphorylation,39 these results suggested that enhanced required for the activation of the ERK signal transduction pathway negative feedback regulation might operate upstream of Raf. by constitutively active H-RasV12 but was dispensable for this V12 event in the absence of H-Ras . Bach1 facilitates the ROS accumulation induced by H-RasV12 Ras-induced transformation can lead to the production of ROS Enhanced feedback regulation of ERK in the absence of Bach1 through pathways involving flavoprotein and Rac1.40 Several A strong activation of the Ras-MAPK cascade is known to result in reports have shown that ROS are essential for transformation by H-RasV12 and contribute to various aspects of malignancy, its negative feedback inhibition as a result of the induction of 41,42 phosphatases.37 Therefore, the increased expression of including metastasis and angiogenesis. As Bach1 inhibits the expression of HO-1, the activity of which is critical for ROS phosphatases (Figures 5a and b) may be responsible for the 8 alterations in ERK activation induced in Bach1-deficient cells by elimination, we investigated whether the ROS levels might be H-RasV12. To examine this possibility, we treated cells with okadaic affected by the Bach1 deficiency. We found that the ROS levels were much higher in wild-type cells than in Bach1-deficient cells in acid, an inhibitor of serine–threonine phosphatases (Figure 6c), V12 and orthovanadate, an inhibitor of tyrosine phosphatases the presence of H-Ras (Figure 7a). The ROS levels increased in (Figure 6d). Similar levels of ERK1/2 phosphorylation were Bach1-deficient i-MEFs when Bach1 expression was restored by retroviral transduction (Figure 7b). Reciprocally, the ROS levels in observed in both types of cells in the presence of a higher V12 concentration of the inhibitors. These results suggested that the wild-type cells expressing H-Ras decreased upon knockdown of protein kinases upstream of ERK1/2 were functional, and that the Bach1 (Figure 7c). Taken together, these results suggested that reduced ERK1/2 phosphorylation was, at least in part, due to Bach1 facilitated the process of transformation at least in part by increased phosphatase activity. sustaining higher ROS levels. To further localize the target of the negative feedback inhibition of the MAPK pathway in Bach1-deficient cells, we expressed a Carcinogen-induced tumor formation is inhibited in Bach1- constitutively active Raf mutant fused to the ligand-binding deficient mice domain of the estrogen receptor (Raf-ER) in i-MEFs.38 When Raf-ER Chronic exposure of mice to 4-nitroquinoline-1-oxide (4-NQO) was activated by treatment with tamoxifen, the phosphorylation causes DNA damage and induces invasive squamous cell of ERK1/2 occurred in wild-type cells. Similar levels of ERK1/2 carcinoma in the oral cavity and esophagus.43 The 4-NQO- phosphorylation were observed in Bach1-deficient cells induced squamous cell carcinoma in rats often exhibits (Figure 6e). Considering that wild-type Raf is regulated by auto- activating mutations in H-Ras.44 To test the hypothesis that

& 2013 Macmillan Publishers Limited Oncogene (2013) 3231 – 3245 Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3238 Table 2. List of entities in which expression was induced or suppressed by H-RasV12 in wild-type but not in Bach1-deﬁﬁent i-MEFs

Probe name Gene symbol Accession number Description

Upregulated entities A_51_P517075 Serpinf1 NM_011340 Serine (or cysteine) peptidase inhibitor, clade F, member 1 (Serpinf1) A_51_P242634 Dusp16 AF345951 Map kinase phosphatase-M A1 isoform A_51_P191865 Lama2 U12147 Laminin-2 alpha2 chain A_52_P64356 Sparcl1 NM_010097 SPARC-like 1 A_52_P28651 Pvrl1 NM_021424 Poliovirus receptor-related 1 A_52_P167249 Add3 NM_013758 Adducin 3 (gamma) transcript variant 2 A_51_P376934 Bace2 NM_019517 Beta-site APP-cleaving enzyme 2 A_51_P477414 Echdc2 NM_026728 Enoyl coenzyme A hydratase domain-containing 2 A_51_P172502 Cxcl12 NM_001012477 Chemokine (C-X-C motif) ligand 12 transcript variant 3 A_52_P652859 Lama2 NM_008481 Laminin, alpha 2 A_51_P153053 Smpdl3a NM_020561 Sphingomyelin phosphodiesterase, acid-like 3A A_51_P156547 Tcf7l1 NM_001079822 Transcription factor 7-like 1 (T-cell speciﬁc, HMG box) transcript variant 1 A_51_P261051 Dlx5 NM_010056 Distal-less homeobox 5, transcript variant 1 A_52_P486279 Popdc3 NM_024286 Popeye domain-containing 3 A_52_P634090 Jag1 NM_013822 Jagged 1 A_51_P123724 N/A AI842957 Unknown A_52_P182298 Rln1 NM_011272 Relaxin A_52_P514352 Kcnk5 NM_021542 Potassium channel, subfamily K, member 5 A_52_P431872 Ptcd3 NM_027275 Pentatricopeptide repeat domain 3 A_51_P344399 Rilpl2 NM_030259 Rab interacting lysosomal protein-like 2 A_51_P382849 Emb NM_010330 Embigin A_52_P597791 Robo2 ENSMUST00000116586 Roundabout homolog 2 (Drosophila) A_52_P322301 Esrp2 NM_176838 Epithelial splicing regulatory protein 2 A_52_P61735 Flnc NM_176838 Filamin C, gamma A_52_P569375 Fgf5 NM_010203 Fibroblast growth factor 5 A_52_P489295 Adamts1 NM_009621 Disintegrin-like and metallopeptidase (reprolysin type) with thrombospondin type 1 motif A_52_P429876 Tbx20 NM_194263 T-box 20 (Tbx20), transcript variant 1 A_51_P345867 Hsf4 NM_011939 Heat shock transcription factor 4 A_51_P398191 Auts2 NM_177047 Autism susceptibility candidate 2 A_51_P108183 Tnmd NM_022322 Tenomodulin (Tnmd) A_52_P820923 1700001L05Rik NR_027980 Mus musculus RIKEN cDNA 1700001L05 gene, non-coding RNA A_52_P234127 Odz3 NM_011857 Mus musculus odd Oz/ten-m homolog 3 (Drosophila) A_52_P148069 Cap2 NM_026056 Mus musculus CAP, adenylate cyclase-associated protein, 2 (yeast)

Downregulated entities A_52_P225158 N/A N/A Unknown A_52_P415116 Trem1 NM_021406 Triggering receptor expressed on myeloid cells 1 A_51_P331288 Akr1b7 NM_009731 Aldo-keto reductase family 1, member B7 A_52_P55951 N/A NP060941 Unknown A_51_P448479 Slc10a4 NM_173403 Solute carrier family 10 (sodium/bile acid cotransporter family), member 4 A_51_P415369 Zbtb32 NM_021397 Zinc finger and BTB domain-containing 32 A_51_P246962 Ap3s1 NM_009681 Adapter-related protein complex 3, sigma 1 subunit (Ap3s1), mRNA [NM_009681] A_51_P191448 Slc8a1 NM_011406 Solute carrier family 8 (sodium/calcium exchanger), member 1 A_51_P286878 Ttll11 NM_029774 Tubulin tyrosine ligase-like family, member 11, transcript variant 1 A_52_P391095 Crem NM_013498 cAMP-responsive element modulator, transcript variant 3 A_52_P346231 Azi2 NM_013727 5-azacytidine-induced gene 2, transcript variant 1 A_52_P226489 N/A CJ042244 CJ042244 RIKEN full-length enriched mouse cDNA library A_51_P198076 1810046J19Rik NM_025559 RIKEN cDNA 1810046J19 gene A_52_P14321 Whsc1 NM_001177884 Wolf-Hirschhorn syndrome candidate 1 (human), transcript variant 3 A_52_P164578 Cdh2 AK084821 Cadherin 2 A_52_P963966 N/A AK048646 16 days embryo head cDNA A_52_P112203 Spire1 NM_176832 Cpire homolog 1 (Drosophila), transcript variant 2 A_51_P161682 Coq7 NM_009940 Coenzyme Q7 homolog, ubiquinone (yeast) A_52_P665742 Srebf2 BC069989 Sterol regulatory element binding factor 2 A_51_P304879 Ndfip1 ENSMUST00000025293 Nedd4 family interacting protein 1 A_52_P317820 Fbxo41 NM_001001160 F-box protein 41 A_52_P261843 Zdhhc18 NM_001017968 Zinc finger, DHHC domain-containing 18 A_52_P157704 3OST3A1 ENSMUST00000106476 Heparan sulfate (glucosamine) 3-O-sulfotransferase 2 A_51_P203474 Tsg101 NM_021884 Tumor susceptibility gene 101 A_51_P379428 Syp NM_009305 Synaptophysin A_52_P60194 C4bp NM_007576 Complement component 4 binding protein A_52_P671029 Col11a1 ENSMUSG00000027966 Collagen alpha-1(XI) chain A_52_P149438 Pdlim5 NM_019809 PDZ and LIM domain 5, transcript variant 2 A_52_P686701 Nfam1 NM_028728 Nfat activating molecule with ITAM motif 1 A_52_P572178 D130043K22Rik NM_001081051 RIKEN cDNA D130043K22 gene A_52_P221057 N/A 2210408K08 Unknown A_52_P1089323 Rbm44 NM_001033408 RNA-binding motif protein 44 A_52_P173726 Vav1 NM_011691 Vav 1 oncogene, transcript variant 1

Oncogene (2013) 3231 – 3245 & 2013 Macmillan Publishers Limited Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3239 Table 2 (Continued )

Probe name Gene symbol Accession number Description

A_51_P517215 Grb14 NM_016719 Mus musculus growth factor receptor-bound protein 14 A_51_P137236 Olfm1 NM_019498 Olfactomedin 1, transcript variant 1 A_52_P67794 Itsn2 NM_011365 Intersectin 2 A_51_P350033 Lrat NM_023624 Lecithin-retinol acyltransferase (phosphatidylcholine-retinol-O- acyltransferase) A_52_P562624 Pkhd1l1 NM_138674 Polycystic kidney and hepatic disease 1-like 1 A_51_P228768 Slfn3 NM_011409 Schlafen 3 A_51_P273213 Grhl1 NM_145890 Grainyhead-like 1 (Drosophila), transcript variant 2 A_52_P642955 Erc2 NM_177814 ELKS/RAB6-interacting/CAST family member 2 A_52_P656800 BC046404 NM_198861 Unknown A_51_P154983 Zfp296 NM_022409 Zinc ﬁnger protein 296 A_51_P308362 Ugt1a6b NM_201410 UDP glucuronosyltransferase 1 family, polypeptide A6B A_52_P505968 EG436081 XM_001004069 Predicted gene, EG436081, mRNA [XM_001004069] A_51_P450007 5730585A16Rik AK019974 RIKEN full-length enriched library, clone:5730585A16 A_51_P181097 Pfas NM_001159519 Phosphoribosylformylglycinamidine synthase (FGAR amidotransferase) A_51_P108072 4930557B06Rik AK016157 RIKEN full-length enriched library, clone:4930557B06 Abbreviations: cDNA, complementary DNA; N/A, not available.

Figure 5. Bach1-dependent inactivation of phosphatases genes and activation of ERK by Ras. (a) A heat map of a subset of 24 phosphatase genes, which were derepressed in Bach1-deficient, H-RasV12-expressing i-MEFs. Clustering and visualization were performed as described above. (b) Thirteen phosphatase genes in panel a were validated by real-time RT–PCR. The heat map represents relative ratios of genes expression levels with and without H-RasV12.(c) The phosphorylation of ERK1/2 was detected by fluorescent immunostaining. Wild-type or Bach1-deficient i-MEFs were transduced with H-RasV12. DNA was stained with Hoechst 33258. Magnification, Â 400.

Bach1 could facilitate carcinogen-induced tumor formation in and polymorphic. We observed papillomas, as well as endophytic mice, we exposed wild-type and Bach1-deficient mice to 4-NQO in tumors invading through the basal membrane (Figure 8c). The their drinking water. All but two wild-type mice survived the 16- total number of tumors per animal was higher in wild-type than in week treatment period and for the 8 weeks post-treatment. There Bach1-deficient mice (Figure 8d). The number of the tongue were no differences in the body weight loss between the wild- lesions was also significantly higher in wild-type mice. The type and Bach1-deficient survivors (Figure 8a). Multiple precancer- difference in the numbers of esophageal lesions was non- ous and cancerous lesions of the tongues and esophagi significant (Figure 8d). We also observed no difference in the developed in all mice (N ¼ 36) regardless of their genotype during total tumor volume between the mice of different genotypes the 8 weeks following carcinogen treatment (Figure 8b). The (Figure 8e). There were no malignant lesions in the lungs, livers or carcinogenic process in the tongues and esophagi was multifocal stomachs, irrespective of the genotype of the mice (N ¼ 6). These

& 2013 Macmillan Publishers Limited Oncogene (2013) 3231 – 3245 Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3240

Figure 6. The defective ERK activation in Bach1-deficient cells involved phosphatase activities. (a, b) The results of the kinetic analysis of ERK1/ 2, MEK and Akt activation in response to serum stimulation in H-RasV12-transduced i-MEFs (a) and primary MEFs (b) of the indicated genotypes. (c, d) The effects of treatment with okadaic acid (OA) (c) and orthovanadate (OV) (d) on ERK1/2 phosphorylation in wild-type and Bach1-deficient cells with H-RASV12.(e) Wild-type and Bach1-deficient i-MEFs were transduced with a Raf-ER-expressing retrovirus. Cells were treated with tamoxifen or a vehicle, and the levels of ERK1/2 phosphorylation were compared.

results suggest that Bach1 was required in vivo for cellular as a kind of resistor that tunes the engagement of the negative transformation, rather than for tumor growth and progression. feedback loop (Figure 9). When the activity of Bach1 is reduced, the negative feedback loop would then become hyper-inducible in response to H-RasV12. The changes in the 4-NQO model in the DISCUSSION absence of Bach1 may also reflect altered ERK signaling. As Bach1 In this study, we showed that Bach1 was necessary for effective itself is regulated by oxidative stress and heme,7,8 the present transformation of mouse fibroblasts by H-RasV12 in vitro and observations suggest a mechanism by which oxidative stress in vivo, as well as for the squamous cell carcinoma formation affects ERK signaling and its output, such as cell proliferation. It is induced by 4-NQO in mice. These two alterations caused by the not clear at present whether Bach1 directly regulates the Bach1 deficiency may be related in terms of the underlying phosphatase genes, and further studies are necessary. molecular mechanisms. The results described here suggest The second possibility is that the enhanced expression of possible interconnected mechanisms regulating cell transforma- antioxidant and detoxification enzymes driven by Nrf246 resulted tion (Figure 9), and are discussed below. in inefficient transformation of Bach1-deficient i-MEFs and First, the reduced ERK signaling in the absence of Bach1 may reduced the damage caused by 4-NQO. This possibility is cause the two phenotypic changes. It was especially important supported by the finding that H-RasV12 increased the ROS levels that Bach1 deficiency blunted the ERK signaling only in the in wild-type cells, but not in Bach1-deficient cells. Some of the Nrf2 presence of the H-RasV12 oncogene. The steady state of ERK target genes are repressed by Bach1, which binds to their phosphorylation in the cell is maintained by negative feedback enhancers in competition with Nrf2.8,47,48 For instance, HO-1 regulation: strong ERK signaling induces a number of protein expression is induced in the absence of Bach1,8 and it reduces the phosphatases, which abrogate the signaling by removing the ROS levels.49 This is the reason why 4-NQO-induced damage may activating phosphorylation from Raf, MEK and ERK.45 In the be alleviated by the reduction in ROS. Recently, it was reported absence of Bach1, the expression of multiple phosphatases that activated Ras increases the ROS levels only when it is involved in the feedback loop was strongly induced in response overexpressed. Rather, endogenously encoded, activated Ras to H-RasV12. Bach1 may function within the ERK signaling system induces Nrf2 transcription, promoting ROS detoxification and

Oncogene (2013) 3231 – 3245 & 2013 Macmillan Publishers Limited Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3241 stress responses and transformation by H-RasV12. The proposal 250 250 that tumor cells can modify their oxidative stress response via Bach1-/- + H-RasV12 Bach1-/- -/- Bach1 may explain the ability of the incipient cancer cells to adapt V12 Bach1 + Bach1 200 WT + H-Ras 200 to the insufﬁcient vascularization, hypoxia, oxidative stress and 150 150 mitotic stress associated with the development and progression of cancer. Therapeutic targeting of Bach1 may result in the selective 100 100 elimination of the cancer cells by exploiting its metabolic effects on the adaptation of cells to an inappropriate microenvironment 50 50 or by reinforcing conventional genotoxic stress-based therapies.

0 0 100 101 102 103 104 100 101 102 103 104 MATERIALS AND METHODS 250 Animal studies WT + Control si WT + Bach1 si All experimental protocols using mice were approved by the Institutional 200 Animal Care and Use Committee of Tohoku University. Bach1-deﬁcient mice have been described previously.8 Heterozygous Bach1 þ / À mice were 150 intercrossed to obtain littermates of the desired genotypes. Athymic nude 100 mice (BALB/cAJc1-nu/nu) were purchased from CLEA Japan (Tokyo, Japan).

50 Plasmids V12 0 The pBabe-puro-H-Ras expression vector was kindly provided by 100 101 102 103 104 Nobuyuki Tanaka from Nippon Medical University. The pcDNA3.1-Bach1 Cell count DCF-DA and pBabe-puro-Bach1 expression vectors were constructed by digesting the pBSA1 plasmid7 with SalI, and ligation of the resulting fragments into Figure 7. The Bach1-dependent Ras-induced increase of ROS levels. the cloning sites of the pcDNA3.1 and pBabe-puro vectors, respectively. (a) The ROS levels in wild-type and Bach1-deficient i-MEFs Raf-ER, a constitutively active form of Raf fused to the ligand-binding V12 transduced with H-Ras were compared using a FACS analysis of domain of the estrogen receptor, was kindly provided by Fuyuki Ishikawa DCF-DA-stained cells. (b) The effects of Bach1 reconstitution on the at Kyoto University.38 ROS levels in Bach1-deficient i-MEFs. (c) The effects of Bach1 knockdown on the ROS levels in wild-type i-MEFs expressing H-RasV12. The data represent typical results of at least two Isolation and immortalization of MEFs independent experiments. MEFs were prepared and maintained at 37 1C in Dulbecco’s modified Eagle medium (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS) (JRH Bioscience, Levexa, KS, USA), 0.1 mM nonessential amino acids (Invitrogen), penicillin/streptomycin (10 000 U/ml each, tumorigenesis.50 Therefore, ROS appear to be maintained within a Invitrogen) and 55 mM 2-mercaptoethanol (Wako Pure Chemicals, Osaka, narrow range during transformation based on the balance Japan), as previously described.11 For immortalization, primary MEFs were between Bach1 and Nrf2. The regulation of many known Nrf2 cultured under normal conditions. When proliferation ceased, the MEFs target genes is less dependent on Bach1 in human were maintained without passage until signs of regrowth became keratinocytes.51 Therefore, the genome-wide identification of the apparent. Then the cultures were replated at a density of 3 Â 105 cells Bach1 and Nrf2 target genes is necessary. per 60 mm culture dish every 3 days.53 Stable immortalized cultures Interestingly, neoangiogenesis was less robust in the tumors (i-MEFs) were analyzed for their Bach1, p19ARF and p53 expression status 54 derived from H-RasV12-transduced Bach1-deficient i-MEFs than in by an immunoblotting analysis. the wild-type tumors. The elevated levels of ROS in the presence of Bach1 may induce tumor angiogenesis, because neoangiogen- Cell treatment esis in tumors driven by activated Ras involves ROS as a To induce MAPK activation, the cells were deprived of serum (0.05% FBS) messenger.4 It is also possible that Bach1 directly or indirectly for 48 h and then treated with 30% FBS for the indicated periods of time. affects the expression of angiogenic genes, such as VEGF, or To inhibit the phosphatase activity, the cells were treated with the inhibitors of angiogenesis. H-RasV12 is known to induce ROS indicated concentrations of okadaic acid (Calbiochem, Darmstadt, Ger- production dependent on NADPH oxidases.29 In the present study, many) or orthovanadate (Sigma, St Louis, MO, USA) for 30 min. Raf-ER was we found that H-RasV12 increased the ROS levels in wild-type cells, activated by 10 nM of 5-hydroxytamoxifen (Sigma) for 12 h. but not in Bach1-deficient cells. As several protein phosphatases, such as LMW-PTP, are regulated by the redox status of critical Colony formation assay cysteine residues.52 it is worth examining whether the activity of The pBabe-puro-H-RasV12 vector was transfected into i-MEFs by electro- LMW-PTP or related enzymes is increased in Bach1-deficient cells poration (Amaxa, Lonza Bio, Basel, Switzerland). After transfection, 1 Â 106 owing to the lower ROS levels. cells were seeded on 100 mm cell culture dishes and selected by Further consideration of the present results reminds us of the puromycin (10 mg/ml) for 24 h. The medium was changed every 3 days. emerging concept that oncogene-directed transformation is often After 3 weeks, the colonies were visualized by Giemsa staining. dependent on the cellular signal transduction and stress response pathways.19 Particular oncogenes may heavily depend on certain In vivo tumor growth assay and histological analysis proteins for their function (non-oncogene addiction).19 Our study The pBabe-puro-H-RasV12 vector was used to integrate mutant Ras into the has implications in the concept of non-oncogene addiction, and cell genome utilizing a retroviral system.55 Suspensions of H-RasV12- 6 may represent a target for exploitation. Although Bach1 appears transduced i-MEFs (1 Â 10 cells in 100 ml of serum-free medium) were to be largely nonessential for the normal life of mice under mixed with 100 ml of 12 mg/ml Matrigel (Becton Dickinson, Franklin Lakes, controlled laboratory conditions (manuscript in preparation), our NJ, USA) and injected subcutaneously into the dorsal flanks of 8-week-old V12 female athymic nude mice. The animals were euthanized 14 days after current results indicate that H-Ras appears to confer upon cells transplantation. The tumors were then excised and measured externally an increased dependency on Bach1. Bach1 is therefore a new with a caliper in two dimensions. The tumor volumes were calculated using addition to the vast and unexplored landscape of non-oncogene the equation V ¼ (L Â W 2) Â 0.5, where L was the length and W was the addiction. Like one of the regulators of the heat shock response, width. Tissue sections were stained with hematoxylin and eosin or a heat shock factor 1 (HSF1),18 Bach1 has important roles in both polyclonal goat anti-CD31 (PECAM) antibody (Santa Cruz Biotech, Santa

& 2013 Macmillan Publishers Limited Oncogene (2013) 3231 – 3245 Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3242

Figure 8. Bach1 is involved in chemical carcinogenesis in mice. (a) Wild-type and Bach1-deﬁcient mice were exposed to 4-NQO in drinking water. Changes in body weight were monitored for 24 weeks. (b) A typical view of the tumors of the tongue and esophagus (scale bar ¼ 5 mm). (c) Morphological variants of tumors in the wild-type and Bach1-deﬁcient mice (scale bar ¼ 200 mm). (d) The total numbers of tumors per animal and numbers of tongue and esophageal tumors are shown (*Po0.05, **Po0.01). (e) The total tumor volumes were compared between the indicated genotypes.

Cruz, CA, USA). The signal for CD31 was visualized by an subjected to FACS (FACS Calibur, Beckton Dickinson) using the Cell Quest immunoperoxidase method using diaminobentizin as a substrate. The software program for acquisition and analysis. specimens were examined by an experienced pathologist. Immunoblotting analysis ROS measurement Whole cell extracts were prepared as described previously.11 Lysates were The cells incubated with DCF-DA (5 mg/ml for 30 min, Molecular Probes, resolved on 7.5–15% SDS–PAGE gels and transferred to PVDF membranes Carlsbad, CA, USA) were washed with phosphate-buffered saline and (Millipore, Billerica, MA, USA). The antibodies for detection of GAPDH,

Oncogene (2013) 3231 – 3245 & 2013 Macmillan Publishers Limited Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3243 tubulin (B-7) and H-Ras (C-20) were purchased from Santa Cruz Biotech addition, a technical replicate using a mixture of RNAs isolated from the (Santa Cruz, CA, USA). The antibodies for the detection of MAPK and two Bach1 À / À cells in a independent experiment was also carried out. related proteins were purchased from CST Japan (Tokyo, Japan): anti- phospho-MEK1/2 (no. 9121), anti-MEK1/2 (no. 4694), anti-phospho-Erk1/2 (no. 9101), anti-Erk1/2 (no. 9102), anti-phospho-Akt (no. 9271) and anti-Akt Real-time qPCR (no. 9272). An anti-Bach1 antibody was described previously.8 PCR was performed using a Light Cycler instrument in the SYBR green Immunoreactive proteins were detected using ECL blotting reagents (GE format (Roche Diagnostics, Mannheim, Germany). b-Actin was used as an Healthcare, Little Chalfont, UK). internal control. The primer sequences for b-actin,56 p5357 and Bach148 were published previously. The primers sequences for other examined Gene expression profiling genes are summarized in the Table 3. The qPCR for Dusp9 was performed in the TaqMan format with commercial gene expression kit (Life All equipment and reagents used for the gene expression profiling were Technologies, Tokyo, Japan). To compare effects of H-RasV12 upon gene purchased from Agilent Technologies (Santa Clara, CA, USA). Total RNAs expression levels, values were corrected for beta actin and then normal- were prepared from cells using the Total RNA Isolation minikit. RNA ized (z-score, in R software using package: genefilter). samples were amplified using a Low RNA Input Fluorescent Linear Amplification Kit following the manufacturer’s protocol, and labeled with cyanine-3 dye. Labeled samples were incubated with Whole Mouse Immunofluorescent analysis Genome Array (4 Â 44 K, G4122F) slides for 17 h. The slides were scanned, Immunostaining was performed exactly as described previously.58 Briefly, and the gene expression analysis was performed using the Genespring GX the cells were fixed in 10% neutral formalin solution and permeabilized software program version 11. The analyses were carried out in biological with 0.1% SDS and 0.5% Triton-X in phosphate-buffered saline. After triplicates using i-MEFs derived from three wild-type and biological incubation with an anti-phospho-ERK1/2 antibody (1:1000, CST Japan) for duplicates using two Bach1-deficient independent mouse embryos. In 30 min at 371 C, an anti-rabbit IgG FITC-conjugated secondary antibody was used to detect the phoshpo-ERK1/2 signal. Nuclei were stained with 10 mM Hoechst 33258 (Sigma). The images were acquired using a Leica FW4000 fluorescent microscopy system (Leica Microsystems, Wetzlar, Germany) and were processed by a 2D deconvolution algorithm.

Chemical carcinogenesis The experiments were carried out under controlled conditions with a 12-h light/dark cycle. The male and female 8-week-old wild-type or Bach1- deﬁcient C57BL/6 J mice were treated with 4-NQO (Sigma) in their drinking water (100 mg/ml) for 16 weeks. The control group received equivalent volume of DMSO only. The mice were euthanized after 8 weeks post- treatment. The specimens of the tongues and esophagi were stained with hematoxylin and eosin. The tumors were counted and measured externally as described above. The lesions were examined and graded by an experienced pathologist.

CONFLICT OF INTEREST The authors declare no conﬂict of interest.

ACKNOWLEDGEMENTS We are grateful to Nobuyuki Tanaka from Nippon Medical University and Fuyuki Ishikawa from Kyoto University for kindly providing plasmids. This work was Figure 9. Dependence of oncogenic Ras-transformed cells on Bach1. supported by Grants-in-aid and the Network Medicine Global-COE Program from the Bach1 buffers the excessive activity of p53 target genes and Ministry of Education, Culture, Sports, Science and Technology of Japan. Additional oxidative stress-response genes, such as Hmox1, in normal cells (gray initial support was provided by the Uehara Foundation, the Takeda Foundation and lines), setting a threshold for senescence. In transformed cells, the the Astelas Foundation for Research on Metabolic Disorders. Restoration of laboratory adaptive function of Bach1 shifts to the suppression of the negative damage from the 2011 Tohoku earthquake was provided in part by the Astelas feedback induced by activated oncogenes and to maintain the ROS Foundation for Research on Metabolic Disorders, the Banyu Foundation, the Naito levels (black lines), facilitating transformation and tumorigenesis. Foundation, A Miyazaki and A Iida.

Table 3. Primers for real-time RT–PCR Gene symbol Accession number Forward primer Reverse Primer

Acp1 NM_021330 GAGGATAGACAGTGCGGCTAC TAATCTGCCGTGCAATGTGCT Cdc14b NM_172587 AGCAGACCAAAGAGTGCAACA CCTCAGCATTGTAATGGACTTGA Cdc25c NM_009860 AAAATGCAGCGTTCCTGCTTC CTTGGGGTCCTAGTGCCTC Dusp1 NM_013642 GCGCTCCACTCAAGTCT TGCACTGTCAGGCACACTA p19 NM_009877 GCTCTGGCTTTCGTGAACA TCGAATCTGCACCGTAGTTG Pgam5 NM_001163538 ATCTGGAGAAGACGAGTTGACA CCTGTTCCCGACCTAATGGT Ppm1b NM_001159496 GAGGAAGCCGTGAAGAGAGATT CATGGGCAAGATCAGGCATT Ppp1cb NM_172707 GATGTCGTCCAGGAAAGATTGT TCAGTGGTGCTTCCAATTCCA Ppp2cb NM_017374 CAGGCTGCTATCATGGAATTA ACTTCCACATACAAAGACAGGT Ppp5c NM_011155 CGGACTGAGTGTGCTGAGAC CCTTGGCTTTGAAGTAGTCGT Ptp4a2 NM_008974 CCACCAATGCGACTCTCAACA CCATCATCAAACGGCCAATCT Ptpn14 NM_008976 GAAGGAGCGGGTCAAGAAAGA GGGCTTCCTACCGTCAGTG Styx NM_019637 AGAAGTGCTGCCTTTGTCATT AGGACCTTTCTATTTGGAGTGGT

& 2013 Macmillan Publishers Limited Oncogene (2013) 3231 – 3245 Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3244 REFERENCES 28 Gulbins E, Coggeshall KM, Langlet C, Baier G, Bonnefoy-Berard N, Burn P et al. 1 Jackson AL, Loeb LA. The contribution of endogenous sources of DNA damage to Activation of Ras in vitro and in intact fibroblasts by the Vav guanine nucleotide the multiple mutations in cancer. Mutat Res 2001; 477: 7–21. exchange protein. Mol Cell Biol 1994; 14: 906–913. 2 Cerutti PA. Prooxidant states and tumor promotion. Science 1985; 227: 375–381. 29 Loboda A, Nebozhyn M, Klinghoffer R, Frazier J, Chastain M, Arthur W et al. A gene 3 Weinberg F, Chandel NS. Reactive oxygen species-dependent signaling regulates expression signature of RAS pathway dependence predicts response to PI3K and cancer. Cell Mol Life Sci 2009; 66: 3663–3673. RAS pathway inhibitors and expands the population of RAS pathway activated 4 Xia C, Meng Q, Liu LZ, Rojanasakul Y, Wang XR, Jiang BH. Reactive oxygen species tumors. BMC Med Genom 2010; 3: 26. regulate angiogenesis and tumor growth through vascular endothelial growth 30 Friedberg I, Nika K, Tautz L, Saito K, Cerignoli F, Godzik A et al. Identification and factor. Cancer Res 2007; 67: 10823–10830. characterization of DUSP27, a novel dual-specific protein phosphatase. FEBS Lett 5 Karin M, Takahashi T, Kapahi P, Delhase M, Chen Y, Makris C et al. Oxidative stress 2007; 581: 2527–2533. and gene expression: the AP-1 and NF-kappaB connections. Biofactors 2001; 15: 31 Liu C, Shi Y, Du Y, Ning X, Liu N, Huang D et al. Dual-specificity phosphatase 87–89. DUSP1 protects overactivation of hypoxia-inducible factor 1 through inactivating 6 Igarashi K, Sun J. The heme-Bach1 pathway in the regulation of oxidative ERK MAPK. Exp Cell Res 2005; 309: 410–418. stress response and erythroid differentiation. Antioxid Redox Signal 2006; 8: 32 Muda M, Boschert U, Smith A, Antonsson B, Gillieron C, Chabert C et al. Molecular 107–118. cloning and functional characterization of a novel mitogen-activated protein 7 Oyake T, Itoh K, Motohashi H, Hayashi N, Hoshino H, Nishizawa M et al. Bach kinase phosphatase, MKP-4. J Biol Chem 1997; 272: 5141–5151. proteins belong to a novel family of BTB-basic leucine zipper transcription factors 33 Ritt DA, Monson DM, Specht SI, Morrison DK. Impact of feedback phosphorylation that interact with MafK and regulate transcription through the NF-E2 site. Mol Cell and Raf heterodimerization on normal and mutant B-Raf signaling. Mol Cell Biol Biol 1996; 16: 6083–6095. 2010; 30: 806–819. 8 Sun J, Hoshino H, Takaku K, Nakajima O, Muto A, Suzuki H et al. Hemoprotein 34 Yu W, Imoto I, Inoue J, Onda M, Emi M, Inazawa J. A novel amplification target, Bach1 regulates enhancer availability of heme oxygenase-1 gene. EMBO J 2002; DUSP26, promotes anaplastic thyroid cancer cell growth by inhibiting p38 MAPK 21: 5216–5224. activity. Oncogene 2007; 26: 1178–1187. 9 Omura S, Suzuki H, Toyofuku M, Ozono R, Kohno N, Igarashi K. Effects of genetic 35 Tumurbaatar I, Cizmecioglu O, Hoffmann I, Grummt I, Voit R. Human Cdc14B ablation of bach1 upon smooth muscle cell proliferation and atherosclerosis after promotes progression through mitosis by dephosphorylating Cdc25 and reg- cuff injury. Genes Cells 2005; 10: 277–285. ulating Cdk1/cyclin B activity. PLoS One 2011; 6: 14711. 10 Yano Y, Ozono R, Oishi Y, Kambe M, Yoshizumi M, Ishida T et al. Genetic ablation 36 Karnoub AE, Weinberg RA. Ras oncogenes: split personalities. Nat Rev Mol Cell Biol of the transcription repressor Bach1 leads to myocardial protection against 2008; 9: 517–531. ischemia/reperfusion in mice. Genes Cells 2006; 11: 791–803. 37 Pratilas CA, Taylor BS, Ye Q, Viale A, Sander C, Solit DB et al. (V600E)BRAF is 11 Dohi Y, Ikura T, Hoshikawa Y, Katoh Y, Ota K, Nakanome A et al. Bach1 inhibits associated with disabled feedback inhibition of RAF-MEK signaling and elevated oxidative stress-induced cellular senescence by impeding p53 function on chro- transcriptional output of the pathway. Proc Natl Acad Sci USA 2009; 106: 4519–4524. matin. Nat Struct Mol Biol 2008; 15: 1246–1254. 38 Iwasa H, Han J, Ishikawa F. Mitogen-activated protein kinase p38 defines the 12 Bartkova J, Horejsi Z, Koed K, Kramer A, Tort F, Zieger K et al. DNA damage common senescence-signalling pathway. Genes Cells 2003; 8: 131–144. response as a candidate anti-cancer barrier in early human tumorigenesis. Nature 39 Dougherty MK, Muller J, Ritt DA, Zhou M, Zhou XZ, Copeland TD et al. Regulation 2005; 434: 864–870. of Raf-1 by direct feedback phosphorylation. Mol Cell 2005; 17: 215–224. 13 Collado M, Gil J, Efeyan A, Guerra C, Schuhmacher AJ, Barradas M et al. Tumour 40 Moore KA, Sethi R, Doanes AM, Johnson TM, Pracyk JB, Kirby M et al. Rac1 is biology: senescence in premalignant tumours. Nature 2005; 436: 642. required for cell proliferation and G2/M progression. Biochem J 1997; 326(Pt 1): 14 Parrinello S, Samper E, Krtolica A, Goldstein J, Melov S, Campisi J. Oxygen sensi- 17–20. tivity severely limits the replicative lifespan of murine fibroblasts. Nat Cell Biol 41 Liu R, Li B, Qiu M. Elevated superoxide production by active H-ras enhances 2003; 5: 741–747. human lung WI-38 VA-13 cell proliferation, migration and resistance to TNF-alpha. 15 Serrano M, Lin AW, McCurrach ME, Beach D, Lowe SW. Oncogenic ras provokes Oncogene 2001; 20: 1486–1496. premature cell senescence associated with accumulation of p53 and p16INK4a. 42 Kim EY, Seo JM, Cho KJ, Kim JH. Ras-induced invasion and metastasis are regu- Cell 1997; 88: 593–602. lated by a leukotriene B4 receptor BLT2-linked pathway. Oncogene 2010; 29: 16 Alexandrova AY, Kopnin PB, Vasiliev JM, Kopnin BP. ROS up-regulation mediates 1167–1178. Ras-induced changes of cell morphology and motility. Exp Cell Res 2006; 312: 43 Tang XH, Knudsen B, Bemis D, Tickoo S, Gudas LJ. Oral cavity and esophageal 2066–2073. carcinogenesis modeled in carcinogen-treated mice. Clin Cancer Res 2004; 10: 17 Bos JL. The ras gene family and human carcinogenesis. Mutat Res 1988; 195: 301–313. 255–271. 44 Suzui M, Yoshimi N, Tanaka T, Mori H. Infrequent Ha-ras mutations and absence of 18 Dai C, Whitesell L, Rogers AB, Lindquist S. Heat shock factor 1 is a powerful Ki-ras, N-ras, and p53 mutations in 4-nitroquinoline 1-oxide-induced rat oral multifaceted modifier of carcinogenesis. Cell 2007; 130: 1005–1018. lesions. Mol Carcinog 1995; 14: 294–298. 19 Luo J, Solimini NL, Elledge SJ. Principles of cancer therapy: oncogene and non- 45 Owens DM, Keyse SM. Differential regulation of MAP kinase signalling by dual- oncogene addiction. Cell 2009; 136: 823–837. specificity protein phosphatases. Oncogene 2007; 26: 3203–3213. 20 Carnero A, Hudson JD, Price CM, Beach DH. p16INK4A and p19ARF act in over- 46 Itoh K, Chiba T, Takahashi S, Ishii T, Igarashi K, Katoh Y et al. An Nrf2/small Maf lapping pathways in cellular immortalization. Nat Cell Biol 2000; 2: 148–155. heterodimer mediates the induction of phase II detoxifying enzyme genes 21 Newbold RF, Overell RW. Fibroblast immortality is a prerequisite for transforma- through antioxidant response elements. Biochem Biophys Res Commun 1997; 236: tion by EJ c-Ha-ras oncogene. Nature 1983; 304: 648–651. 313–322. 22 Filleur S, Hirsch J, Wille A, Schon M, Sell C, Shearer MH et al. INTS6/DICE1 inhibits 47 Sun J, Brand M, Zenke Y, Tashiro S, Groudine M, Igarashi K. Heme regulates the growth of human androgen-independent prostate cancer cells by altering the cell dynamic exchange of Bach1 and NF-E2-related factors in the Maf transcription cycle profile and Wnt signaling. Cancer Cell Int 2009; 9:28. factor network. Proc Natl Acad Sci USA 2004; 101: 1461–1466. 23 Masuda K, Katagiri C, Nomura M, Sato M, Kakumoto K, Akagi T et al. MKP-7, a JNK 48 Igarashi K, Hoshino H, Muto A, Suwabe N, Nishikawa S, Nakauchi H et al. Multi- phosphatase, blocks ERK-dependent gene activation by anchoring phosphory- valent DNA binding complex generated by small Maf and Bach1 as a possible lated ERK in the cytoplasm. Biochem Biophys Res Commun 2010; 393: 201–206. biochemical basis for beta-globin locus control region complex. J Biol Chem 1998; 24 Warzecha CC, Sato TK, Nabet B, Hogenesch JB, Carstens RP. ESRP1and ESRP2 are 273: 11783–11790. epithelial cell-type-specific regulators of FGFR2 splicing. Mol Cell 2009; 33: 49 Willis D, Moore AR, Frederick R, Willoughby DA. Heme oxygenase: a novel target 591–601. for the modulation of the inflammatory response. Nat Med 1996; 2: 87–90. 25 Dawson DW, Volpert OV, Gillis P, Crawford SE, Xu H, Benedict W et al. Pigment 50 DeNicola GM, Karreth FA, Humpton TJ, Gopinathan A, Wei C, Frese K et al. epithelium-derived factor: a potent inhibitor of angiogenesis. Science 1999; 285: Oncogene-induced Nrf2 transcription promotes ROS detoxification and tumor- 245–248. igenesis. Nature 2011; 475: 106–109. 26 Oshima Y, Shukunami C, Honda J, Nishida K, Tashiro F, Miyazaki J et al. Expression 51 MacLeod AK, McMahon M, Plummer SM, Higgins LG, Penning TM, Igarashi K et al. and localization of tenomodulin, a transmembrane type chondromodulin-I-rela- Characterization of the cancer chemopreventive NRF2-dependent gene battery in ted angiogenesis inhibitor, in mouse eyes. Invest Ophthalmol Vis Sci 2003; 44: human keratinocytes: demonstration that the KEAP1-NRF2 pathway, and not the 1814–1823. BACH1-NRF2 pathway, controls cytoprotection against electrophiles as well as 27 Gregory SG, Barlow KF, McLay KE, Kaul R, Swarbreck D, Dunham A et al. The DNA redox-cycling compounds. Carcinogenesis 2009; 30: 1571–1580. sequence and biological annotation of human chromosome 1. Nature 2006; 441: 52 Chiarugi P. The redox regulation of LMW-PTP during cell proliferation or growth 315–321. inhibition. IUBMB Life 2001; 52: 55–59.

Oncogene (2013) 3231 – 3245 & 2013 Macmillan Publishers Limited Bach1 as a possible target of non-oncogene addiction A Nakanome et al 3245 53 Todaro GJ, Green H. Quantitative studies of the growth of mouse embryo 56 Kuroishi T, Endo Y, Muramoto K, Sugawara S. Biotin deficiency up-regulates cells in culture and their development into established lines. J Cell Biol 1963; 17: TNF-alpha production in murine macrophages. J Leukoc Biol 2008; 83: 299–313. 912–920. 54 Matheu A, Maraver A, Klatt P, Flores I, Garcia-Cao I, Borras C et al. Delayed 57 Ota K, Dohi Y, Brydun A, Nakanome A, Ito S, Igarashi K. Identification of senes- ageing through damage protection by the Arf/p53 pathway. Nature 2007; 448: cence-associated genes and their networks under oxidative stress by the analysis 375–379. of Bach1. Antioxid Redox Signal 2011; 14: 2441–2451. 55 Swift S, Lorens J, Achacoso P, Nolan GP. Rapid production of retroviruses for 58 Yamasaki C, Tashiro S, Nishito Y, Sueda T, Igarashi K. Dynamic cytoplasmic efficient gene delivery to mammalian cells using 293 T cell-based systems. Curr anchoring of the transcription factor Bach1 by intracellular hyaluronic acid Protoc Immunol 2001; Chapter 10: Unit 10 17C. binding protein IHABP. J Biochem 2005; 137: 287–296.

& 2013 Macmillan Publishers Limited Oncogene (2013) 3231 – 3245