Evi-1 Raises AP-1 Activity and Stimulates c-fos Promoter Transactivation with Dependence on the Second Zinc Finger Domain*

Evi-1 is a gene, encoding a zinc finger protein, associ-Recently, the distribution of Evi-1 expression was reported in ated with a common viral integration site in murine leu-adult and embryonic mouse tissues (12). It was demonstrated kemias. It is suggested that Evi-1 Plays important roles that Evi-1 is expressed at high levels in several embryonic tis- in embWogenesis and transformation of myeloid cells. sues, but at low levels in most adult tissues. Regions that exhibit TO elucidate mechanisms by which Evi-1 induces such high-level expression in the embryo include the urinary system, biolo&al effects, we analyzed the relationship between bronchial epithelium of the lung, developing limbs and Some Evi-l and which rem1ate Polifera-regions of the nasal cavity and heart. The restricted pattern of tion and differentiation- When Evi-l was expressed, Evi-1 expression in embryonic tissues suggests that Evi-1 plays transactivation through a 12-o-tetradecanoy1phorbo1 an important role in cellular proliferation and differentiation in and Pl9 cells. Eui-1-transfected P19 Staining-P19 cells (1.5 x lo5) were transfected by the calcium-phosphate precipitation method, with pSV2neo (50 ng) and either pME-Evi-1 (3.2 pg) or equivalent-molar pME18S DNA, then selected by G418 (200 pg/ml) for 6 days. Procedures for the fixation of cells and incubation with the first and second antibodies have been described previously (36). The first antibodies used in this study were anti-SSEA-1 (36) (mouse monoclonal IgM, AH-6) and anti-Hsp47 (37) (rat monoclonal IgG, 7C4). The second antibodies were fluorescein-conjugated goat antibodies against mouse IgM and rat IgG (TAGO, Inc.).

Plasmid Constructions-The cDNA of human Eui-1 was derived from AMLl /Eui-1 fusion cDNA obtained from the S M -I cell line, in which the truncated AMLl gene was revealed to fuse to the entire coding region of the Eui-1 gene (22). The sequence of this Eui-1 cDNA was confirmed to be essentially identical to the cDNA which Morishita et al.
(3) obtained. For construction of Evi-1 expression plasmids, a n EcoRI site was produced by the method of site-directed mutagenesis (23) at position 33 base pairs upstream from the Eui-1 translation initiation site. Eui-1 cDNA was inserted into the EcoRI site of pME18S, an SRa promoter-driven expression plasmid (24), to give pME-Evi-1 and pME-Evi-R in the sense and antisense orientation, respectively. Ligation of the Eui-1 cDNA to EcoRI-digested pEF-BOS (29, an elongation factor promoter-driven expression plasmid, produced pEF-Evi-1 and pEF-Evi-R in the sense and antisense orientation, respectively. The pBLLuc plasmid was produced by replacing the XhoI-HpaI fragment of pBLCAT2 (26) with the HindIII-HpaI fragment of luciferase cDNA using a HindIII linker. For construction of a p(wi1d TREIx3-tk-Luc reporter plasmid, in which three tandemly repeated TPA-responsive elements (TRE) were inserted immediately upstream of the herpes simplex virus thymidine kinase promoter followed by firefly luciferase cDNA (27), oligonucleotides (5"GATCGTGAGTCAGCGCGGTGAGT-CGCGCTGACTCACCGCGCTGACTCAC-3') were annealed and inserted into the BamHI site of pBLLuc. For construction of a p(mut. TRE)xS-tk-Luc reporter plasmid, in which two nucleotides were replaced in each TRE (28,29), oligonucleotides (5"GATCGGGAGA-CAGCGCGGGGAGACAGCGCGGGGAGACAGCGCG-3' and 5'-GATC-CGCGCTGTCTCCCCGCGCTGTCTCCCCGCGCTGTCTCCC-3') were annealed and inserted into the BamHI site of pBLLuc. Loss of TPAresponsiveness of the p(mut. TRE)x3-tk-Luc reporter plasmid was confirmed by estimating luciferase activity in NIH3T3 cells transfected with p(mut. TRE)x3-tk-Luc plasmid DNA and stimulated by TPA.
For construction of pFOS(-403)Luc, pFOS(-l42)Luc and pFOS-(-1OO)Luc plasmids, the HindIII-HindIII, Eco473-HindII1, and BssHIE HindIII fragments of the FC4 plasmid (30) were, respectively, blunted by Klenow fragment and inserted into the SmaI site of pUCOOLuc, which was obtained by replacing the CAT coding region of pUCOOCAT (31) with luciferase cDNA. To construct pFOS(-222)Luc and pFOS(-88)Luc, ApaI-HindIII, and BstXI-Hind111 DNA fragments of the FC4 plasmid were blunted by T4 DNA polymerase and inserted into the SmaI site of pUCOOLuc, respectively. Site-directed mutagenesis was performed to make Sal1 sites at -334 or -298 base pairs from transcription initiation of the human c-fos promoter and each SalI-Hind111 DNA fragment was blunted by Klenow fragment and inserted into the SmaI site of pUCOOLuc to construct pFOS(-334)Luc and pFOS(-298)Luc, respectively. The HindIII-Hind111 DNA fragment of the FC4 plasmid, in which Sal1 site was created a t -298 in the c-fos promoter by sitedirected mutagenesis, was blunted by Klenow fragment and inserted into the SmaI site of pUCOOLuc to construct pFOS(mAP1)Luc.
Luciferase Assay-Reporter and expression plasmids were transfected into NIH3T3 cells or P19 cells by the calcium-phosphate precipitation method (32). For the analysis of luciferase activities observed in cotransfection with several expression plasmids, the equivalent-molar plasmid DNAs were transfected. To keep the transfection efficiency as constant as possible among the samples to be compared, the total amount of DNA in terms of weight was adjusted to be equal by adding the plasmid pUC13. As an internal control of transfection efficiency, a plasmid expressing p-galactosidase driven by SRa promoter (1 pg) was cotransfected. NIH3T3 cells or P I 9 cells were washed twice by phosphate-buffered saline after 10-12 h of transfection, cultured for 30-36 h in DMEM containing 0.5% calf serum or DMEM (high glucose formu-CAGCGCGGTGAGTCAGCGCG-3' and 5"GATCCGCGCTGACTCAC- The values (means * standard deviations of three independent experiments) represent luciferase activity expressed relative to the levels when control vector pME18S was cotransfected, instead of pME-Evi-1 or pME-Evi-R, with the corresponding reporter plasmid. in the presence of isopropylthio-p-o-galactoside. The product (fusion protein of maltose-binding protein and partial Evi-1) was purified using an amylose resin column, according to manufacturer's protocol (New England Biolabs) and used to immunize a rabbit. Western Blots-Cells were lysed on ice by the lysis buffer (50 mM Tris-HC1, pH 7.4, 150 mM NaCl, 0.05% SDS, 1% deoxycholate, 1% Triton X-lOO,lO unitslml approtinin and 2 mM phenylmethylsulfonyl fluoride). Protein concentration of cell lysates was quantified using protein assay dye (Bio-Rad). Cell lysates were subjected to SDS-polyacrylamide gel electrophoresis (PAGE) and electrotransferred onto polyvinylidene difluoride filters (Millipore), then reacted with anti-cJun antibody (OPA 13/1 purchased from Medac Molecular Biology) (341, anti-c-Fos antibody (PC05 purchased from Oncogene Science) (35), anti-a-tubulin (DMlA, gift from N. Hirokawa (University of of Tokyo)) or anti-Evi-1 serum (described above). The blots were visualized by Protoblot system (Promega). Immunofluorescent Staining-P19 cells (1.5 x lo5) were transfected by the calcium-phosphate precipitation method, with pSV2neo (50 ng) and either pME-Evi-1 (3.2 pg) or equivalent-molar pME18S DNA, then selected by G418 (200 pg/ml) for 6 days. Procedures for the fixation of cells and incubation with the first and second antibodies have been described previously (36). The first antibodies used in this study were anti-SSEA-1 (36) (mouse monoclonal IgM, AH-6) and anti-Hsp47 (37) (rat monoclonal IgG, 7C4). The second antibodies were fluoresceinconjugated goat antibodies against mouse IgM and rat IgG (TAGO, Inc.).

Eui-1 Raises
A P -I Activity in NIH3T3 Cells"NIH3T3 cell line is frequently used for studies of AP-1 activity because, in these cell lines, increased AP-1 activity by various mitogens is obviously detected as the response ofAP-1-dependent indicator genes (28). To clarify the effect of Evi-1 on AP-1 activity, we transfected the AP-1-dependent reporter gene (p(wild TRE)x3tk-Luc) and the Evi-1 expression plasmid (pME-Evi-1) or control plasmid (pME-Evi-R) to NIH3T3 cells, and measured luciferase activities. The expression of Evi-1 led to efficient transactivation (51.6-fold induction) of the reporter (Table I). This activation is dependent on the AP-1 recognition sequence because significant transactivation in the presence of Evi-1 was not observed with the mutated AP-1 recognition sequence (p(mut.TRE)x3-tk-Luc). No significant transactivation through t h e m -1 recognition sequence was observed in cotransfection of the control plasmid, pME-Evi-R, in which Eui-1 cDNA was inserted downstream of a promoter in the antisense orientation. In the construct of pME-Evi-l, Evi-l expression is driven by the SRa promoter (24). In order to confirm that such transactivation through TRE depends on Evi-1 expression, we constructed pEF-Evi-1 and pEF-Evi-R plasmids in which expression is driven by the elongation factor promoter (25). Expression of Evi-1 by this promoter resulted in the transactivation pattern of reporters, similar to that when Evi-1 was expressed by the SRa promoter (data not shown). These results show that Evi-1 expression raises AP-1 activity in NIH3T3 cells.

Evi-1 Activates AP-1 Depending on Second Zinc Finger Domain
Evi-1 Induces Differentiation and Increases c-Fos and c-Jun Expression in P19 Embryonal Carcinoma Cells-To explore biological responses to the AP-1 activity increased by Evi-1, we used the P19 embryonal carcinoma (EC) cell line, because ectopic expression of cjun or c-fos in EC cells leads to cellular differentiation (14,15). We investigated alteration of AP-1 activity and cellular differentiation in Evi-1-transfected P19 cells. The AP-1-dependent reporter gene was transactivated by Evi-1 in P19 cells, although the fold induction was lower than in NIH3T3 cells (Table I). This activation is dependent on the AP-1-recognition sequence, since no significant transactivation in the presence of Evi-1 was observed with the mutated AP-1 recognition sequence.
When P19 cells were selected for G418 resistance 5-7 days after cotransfection of pME-Evi-1 and pSV2ne0, surviving cells formed colonies and 85-94% of them showed morphological differences from control P19 cells which were transfected with pME18S and pSV2neo. Evi-1-transfected P19 cells showed differentiated phenotypes, characterized by a flattened and enlarged morphology (Fig. lA). These changes were indistinguishable from the morphological changes seen in P19 cells treated with retinoic acid or transfected with cjun (38, 39). To confirm the differentiation of P19 cells induced by Evi-1, stemcell and differentiation marker proteins were examined by the indirect immunofluorescence method (Fig. 1 B ) . The stage-specific embryonic antigen SSEA-1, known as a stem cell marker in EC cells, was detected in P19 cells without Evi-1 transfection, but not in Evi-1-transfected cells. In contrast to SSEA-1, the heat shock protein Hsp47 (37), known as a differentiation marker, was detected in Evi-1-transfected P19 cells, but not in the absence of Evi-1 transfection. In addition, we observed elevated expression of endogenous cdun and c-Fos in Evi-ltransfected P19 cells, compared with pME18S-transfected P19 cells (Fig. 2). The increased expression of endogenous c-Jun is also reported in EC cells transfected with exogenous cjun (39) or activated c-H-rus (36) and treated with retinoic acid (14). It has also been reported that expression of endogenous c-Fos is increased along with EC cell differentiation induced by retinoic acid or dimethyl sulfoxide (40, 41). The c-fos Promoter Is Dansactivated in the Presence of Evil-Transcriptional activation of the c-fos gene is one of the main mechanisms by which various growth factors and oncogenes raise AP-1 activity (42). We examined whether the c-fos promoter is transactivated in the presence of Evi-1. The human c-fos promoter sequence containing nucleotide base pairs between -403 and +43, relative to the start site of transcription, was positioned directly upstream of the firefly luciferase cDNA. This reporter plasmid, designated pFOS(-403)Luc, was cotransfected with the Evi-1-expression vector (pME-Evi-1) or a control plasmid (pME-Evi-R) into NIH3T3 and P19 cells. The reporter was transactivated 26.5-fold when Evi-1 was cotransfected into NIH3T3 cells (Table 11). Similarly, weak but significant transactivation of the c-fos promoter was observed in P19 cells when Evi-1 was cotransfected, which is consistent with elevated expression of c-Fos protein shown in Fig. 2.
Many reports indicate that AP-1 positively regulates c-jun transcription through the AP-1 recognition site in the cjun promoter. On the other hand, AP-1 negatively regulates c-fos transcription through the serum response element in the c-fos promoter (13,(43)(44)(45). There is an AP-1 recognition site at the position of around -295 in the c-fos promoter (461, but this site does not contribute at all to the c-fos promoter transactivation in response to the serum stimulation or overexpressed c-fos (44). To clarify that c-fos promoter transactivation in the presence of Evi-1 is not a secondary effect of enhanced activity of AP-1, the AP-1 site of around -295 (CATCTGCGTCAGCAGG) in the c-fos promoter was converted into a totally different sequence (CATCGTCGACAGCAGG) from the AP-1 recognition consensus by the site-directed mutagenesis method. Even when this mutated reporter plasmid (pFOS(mAP1)Luc) was used, no change was observed in enhanced transactivation of the c-fos promoter in the presence of Evi-1 (Fig. 3). This result indicates that this AP-1 site has no regulatory role in the activation of the c-fos promoter when Evi-1 is expressed. We assume that the transactivation of c-fos promoter is not the result of a rise in AP-1 activity induced by Evi-1, but is a mechanism for it, because the c-fos promoter is reported to be negatively regulated by activated AP-1 (13,451 and in fact, the AP-1 site in the c-fos promoter does not contribute to up-regulation of the c-fos promoter by Evi-1, and because increased c-fos-expression leads to a rise in AP-1 activity (47).

Evi-1 Activates AP-1 Depending on Second Zinc Finger Domain
Fold Induction by Evi-1 Exp. 1 Exp. 2 Fold induction represents the luciferase activity when pME-Evi-1 was cotransfected with each reporter plasmid, relative to the activity in cotransfection of pME18S with the corresponding reporter. The data of two independent experiments are shown. It should be investigated whether Evi-1 also takes part in transcriptional control of c-jun and other members of the judfos family. We assume that the c-jun promoter can be transactivated in the presence of Evi-1 because c-jun transcription is positively regulated by AP-1 through the AP-1 binding site in the promoter (13,43). We should also analyze the relationship between Evi-1 and posttranslational control of AP-1 activity, such as phospholylation of cJun.
Several DNA-binding proteins are reported to contain two widely separated zinc finger motifs, for example, hunchback  (561, Ikaros (57), PRDII-BF1 (581, and Evi-1. To elucidate the mechanisms by which these zinc finger proteins transactivate their target genes, it is important to analyze functions of each zinc finger domain. In PRDII-BF1, each set of zinc fingers recognizes the same DNA sequence (58). As for Evi-1, on the other hand, each binds different sequences (7,18,19). Our experiments revealed the distinct function of each zinc finger domain in Evi-1. The second domain is essential for raising AP-1 activity and transactivation of the c-fos promoter, whereas the first domain is responsible for only part of their activation. Kreider et al. (7) suggested that the first zinc finger domain of Evi-1 inhibits GATA-1-mediated transactivation. It is possible that the primary role of each zinc finger domain is different; inhibition of GATA-binding proteins' functions may be the role of the first zinc finger domain and increasing AP-1 activity may be that of the second domain. In the first zinc finger domain, ZF5-7, ZF2-4, and ZF1 play important roles in activating AP-1 in the sequence given. This result coincides somewhat with Delwel et a2.k (18) report indicating that ZF1-3 in the first domain is not important for DNA binding. Although the acidic domain of Evi-1 was indicated as a putative transactivation domain (2), it was not essential for raising AP-1 activity and transactivation of the c-fos promoter in our experiments. There are two possible explanations of these results. One is that other regions play roles as transactivation domains. The candidates are proline-rich regions found between two separate zinc finger domains. It is known that proline-rich regions possibly function as transactivation domains (59). The other explanation is that transactivation domains are not required for raising AP-1 activity nor for transactivation of the c-fos promoter.We have elucidated that Evi-1 stimulates Ap-1 activity and the c-fos promoter transactivation. We could expect that Evi-1 manifests its roles in embryogenesis and transformation of myeloid cells, at least partly, owing to the AP-1 activation. Further studies would clarify those relationships. It has also been revealed that the second zinc finger domain of Evi-1 is essential for activating AP-1 and stimulating the c-fos promoter transactivation. The mechanism should be further investigated, by which the second zinc finger domain induces such effects.
for providing the pUCOOCAT plasmid, S. Nagata (Osaka Bioscience