A Novel MASH1 Enhancer with N-Myc and CREB-Binding Sites Is Active In

ORIGINAL ARTICLE A novel MASH1 enhancer with N-myc and CREB-binding sites is active in neuroblastoma F Watt1,2, R Watanabe2, W Yang2,NA˚ gren2,3, Y Arvidsson2,4 and K Funa2 1Children’s Cancer Institute Australia for Medical Research, Randwick, New South Wales, Australia; 2Department of Medical Chemistry and Cell Biology, Institute of Biomedicine, Sahlgrenska Academy, Go¨teborg University, Gothenburg, Sweden; 3Department of Medical Cell Biology, Biomedical Center, Uppsala University, Uppsala, Sweden and 4Lundberg Laboratory for Cancer Research, Sahlgrenska Academy, Go¨teborg University, Gothenburg, Sweden

Neuroblastoma is one of the most common solid tumors in childhood. With the aim of developing a targeting vector for neuroblastoma, we cloned and characterized an enhancer in the 50-flanking regions of the MASH1 gene by a random-trap method from a 36 kb cosmid DNA. The enhancer-containing clone was identified by the expression of GFP when transfected into neuroblastoma cell lines. The enhancer-luciferase activity is higher in neuroblastoma cell lines, IMR32, BE2 and SH-SY5Y, compared with those in non-neuroblastoma cell lines, U1242 glioma, N417 small cell lung cancer and EOMA hemangioma. The core enhancer was determined within a 0.2 kb fragment, yielding three- to fourfold higher activity than that of the MASH1 promoter alone in IMR32 and BE2. This area possesses GATA- and CREB-binding sites, as well as the E-box. EMSA on this area demonstrated that CREB/ATF could bind the DNA. Chromatin immunoprecipitation assay revealed that N-myc, CREB, and co-activators CBP and PCAF, but not HDAC1, are bound to the core enhancer at the same time as the co-activators and N-myc bind to the promoter. This supports the idea that the commonly overexpressed genes HASH1 and N-myc are regulated in concert, confirming their importance as prognostic markers or targets for therapy. Cancer Gene Therapy (2007) 14, 287–296. doi:10.1038/sj.cgt.7701012; published online 24 November 2006

Keywords: MASH1; HASH1; N-myc; CREB; enhancer; neuroblastoma

Introduction MASH1 is also widely expressed in the early stages of CNS development,1 and a CNS-specific enhancer has 7 In the developing nervous systema genetic program been isolated and shown to control MASH1 expression. specifies the correct order for genes to be activated, both It is assumed that there exist control elements in this gene, spatially and temporally. Proneural genes encode basic which are specific for the peripheral autonomic nervous helix-loop-helix (bHLH) transcription factors, essential in system. Isolating such elements may enable us to track the early neurogenesis.1 Vertebrate homologs of proneural regulation of gene expression as stemcells differentiate genes are known as the atonal-related neurogenins2 and into autonomic neurons. These elements could also be the achaete-scute-related MASH1.3 MASH1 is the best used to target a therapy-gene expression in various neural characterized member of the vertebrate proneural genes, crest-derived tumors. and is sufficient to induce differentiation of neurons.2 One example is neuroblastoma – one of the most MASH1 is expressed in neural crest cells after they common extracranial solid tumors in childhood, being colonize the anlage of the sympathetic ganglia, adrenal derived fromsympatheticderivatives of neural crest cells. medulla, subsets of enteric precursors, and in parasympa- Many primary neuroblastoma cells and cell lines possess thetic neurons which surround major blood vessels.4 the immature phenotype, and express high levels of 8 MASH1À/À mice have severe developmental defects in the human homolog of the MASH1, that is HASH1. sympathetic ganglia5 and chromaffin cells,6 indicating its Using an enhancer sequence of MASH1 gene, which is importance for the maturating autonomic nervous system. specifically active in the sympathetic neurons, allows the targeting of a therapy gene in neuroblastoma expressing HASH1. The MASH1-promoter alone has been shown to Correspondence: Professor K Funa, Department of Medical provide cell type-specific expression and cytotoxic effects Chemistry and Cell Biology, Institute of Biomedicine, Sahlgrenska of the therapy gene (Escherichia coli purine nucleoside Academy, Go¨ teborg University, Box 420, SE-405 30 Gothenburg, 9 Sweden. phosphorylase) in neuroblastoma cell lines. This ap- E-mail: [email protected] proach should increase the efficacy of in vivo tumor- Received 6 June 2006; revised 22 July 2006; accepted 29 August specific expression of exogenous gene in tumor cells 10 2006; published online 24 November 2006 without giving unwanted effects on normal tissues. In MASH1-enhancer active in neuroblastoma F Watt et al 288 most reported gene therapy trials, therapy genes are with strongest expression were subcloned into pGL3 transduced by viral enhancer/promoters, often giving rise vector for quantitative analysis in both neuroblastoma to strong expression but lacking in tissue-specificity. and non-neuroblastoma cell lines. Mapping of enhancer Nevertheless, other studies have demonstrated specific clones was performed by the appropriate restriction and long-lasting expression of a therapy gene in target enzymes and Southern transfer, using several probes cells, under the control of tissue-specific enhancer/ made from the positive clone. promoters.11 In the present study, we attempt to isolate specific cis- Promoter reporter assay acting elements operating in HASH1-expressing neuro- Cells were seeded in 24-well plates at a density of blastoma cell lines. We isolate and characterize an 2 Â 104 cells/well in 10% serum. The following day, the enhancer located upstreamof the MASH1-coding se- cells were transiently transfected with 100 fmole of quence, froma 36 kb cosmid(McosI) containing both each enhancer construct using 60 ml Fugene6 (Roche, upstreamand downstreamelementsof the entire MASH1 Mannheim, Germany) or Lipofectamine (Invitrogen, gene.11 We then compare its activity in various Carlsbad, CA, USA) according to the vendor’s instruc- neuroblastoma and non-neuroblastoma cell lines. The tions. After 48 h cells were lysed with 100 ml/well of a transcription factors binding to this region, CREB and reporter lysis buffer (Promega, Madison, WI, USA) and N-myc, are identified by EMSA and Chromatin Immuno- luciferase activity was measured. Standardization was precipitation (Chip) assays. done by co-transfection with b-galactosidase reporter plasmid and results were normalized to the activity. The values were presented as means and s.e. of three independent transfections, each carried out in triplicate. Materials and methods Comparisons using t-tests were made between the Cell cultures MASH1 promoter alone and the values obtained by The human neuroblastoma cell lines SH-SY5Y, BE2, different enhancer constructs within the respective cell IMR32 and mouse neuroblastoma C1300 were cultured, lines. Differences were considered significant at *P 0.05, 9 p as described previously. IMR32, BE2 and C1300 contain **Pp0.01 or ***Pp0.001. Only increased values were amplified N-myc and show increased HASH1 or MASH1 marked. expression. The small-cell lung cancer (SCLC) NIH- N417, the breast cancer MCF7, and the mouse heman- Electrophoresis mobility shift assay gioma EOMA lines were cultured in DMEM, and the The 55-mer oligonucleotide (oligo) and the 82 bp HinfI malignant glioma U1242 line was cultured in a-MEM. fragment in the core enhancer HindIII/EcoRV were used as probes (Figure 6). The sequences of the oligos that were Preparation of the enhancer-cloning vector used for competition are: GATA2/3, 50-TTGATAACA An enhancer trap vector was designed as described GAAAGTGATAACTCT-30; AP2, 50-GATCGAACTGA elsewhere.11 The vector consists of Green Fluorescent CCGCCCGCGGOCCGT-30; AP1, 50-CGCTTGATGAC Protein (GFP) reporter gene driven by 0.3 kb MASH1 TCAGCCGGAA-30; ATF-EIIa, 50-TGGAGATGACGT gene promoter (base 2-386 including 314 bp upstream AGTTT-30; CREB-SC, 50-AGAGATTGCCTGACGTC of the transcription start site and 73 bp of the 50 AGAGAGCTAG-30; CREB-SCm, 50-AGAGATTGCCT untranslated region, accession number U68534). The GTGGTCAGAGAGCTAG-30, NF-1, 50-AGAGATTG length of the promoter was determined by our in vitro CCTGTGGTCAGAGAGCTAG-30. The probes were data obtained after comparison with activities of various end-labeled with g-32P-ATP (Amersham, Buckingham- length of the promoter cloned into pGL3b luciferase shire, UK) and T4 polynucleotide kinase and then vector (Promega).9 The multiple cloning sites for DNA purified, and nuclear extracts (NE) were made from fragment insertion were placed upstream of the promoter. BE2, IMR, MCF-7 cells. Electrophoresis mobility shift As negative background controls were used p0.3pGFP, assay (EMSA) was performed as described previously.13 containing only the 0.3 kb MASH1 promoter, and a For the competition experiments or supershift assay, basic vector. preincubation was performed in the presence of 10- and 100-times unlabeled competitor oligo. Cloning of enhancer that is active in neuroblastoma cell line IMR32 Chromatin immunoprecipitation assay For enhancer screening, a library was prepared by Chip assays were performed as described previously14 by inserting into the trap vector partially digested DNA using the mouse neuroblastoma cell line C1300. The fragments from an approximately 36 kb MASH1 cosmid antibodies directed against CREB, CBP, PCAF, USF1 (pMI cos) clone, provided by Dr J Johnson.12 Briefly, we and N-myc, or mouse IgG were purchased from Santa cleaved the cosmid DNA to generate partially digested Cruz Biotechnology, Santa Cruz, CA, USA. All the products, which then were cloned into a vector to make antibodies were shown to work adequately for Chip a library. The digested cosmid DNA containing any assay. Immune complexes were collected by incubation enhancer can be identified by the expression of GFP when with protein G-agarose (Santa Cruz) for 1 h at 41C. transfected into IMR32 cells. Eight plates of colonies in Aliquots of the purified DNA were diluted and analyzed 7 Â 7 matrix were screened in three rounds, and clones by PCR with the appropriate primer pairs. The primers

Cancer Gene Therapy MASH1-enhancer active in neuroblastoma F Watt et al 289 used for the MASH1 enhancer and promoter were Enhancer activity of deletion mutants of MASH1 #4 50-GGGATATCTCGTTCTTGCCTTCAC-30 and 50-CT enhancer in various cell lines ACAGGGTAATCTTGATTAAGCT-30 as well as 50-TT We optimized the activity of enhancers by screening ATTCAGCCGGGAGTCCGG-30 and 50-TGCGGCCG deletion constructs to find a shortest but sufficient length CCTTTTCAATGG-30. The primers for a coding region of the fragment giving a strong expression of the reporter used for the control were described previously.9 Ampli- gene. Various sizes of fragments from the #4en were fication was performed for a predetermined optimal prepared by restriction mapping, and cloned into the number of cycles. PCR products were separated by p0.3pGL3 plasmid containing the 0.3 kb MASH1 pro- electrophoresis on 2% agarose gels, and stained with moter upstream of the luciferase gene. They were ethidiumbromide. analyzed by using reporter assays by transfecting the constructs into the IMR32 and BE2 neuroblastoma cell lines (Figure 2). These fragments were cloned in both sense and antisense directions to the promoter Results (Figure 2a). The strongest activity was obtained with Cloning and characterization of MASH1 enhancer the XbaI/EcoRI (clone 12; Figure 2b). The deletion The cloning method has been described in Watt et al.11 constructs devoid of the HindIII/HindIII fragment, Restriction fragments randomly cleaved from the pMI cos NcoI/EcoRI D(H3-H3) (cl. 9) and PstI/HindIII (cl. 8), were inserted in the pMASH0.3pEnTrap vector in the resulted in a dramatic decrease, indicating the presence of upstreamof the MASH1 promoterand the GFP reporter an active site in the HindIII/HindIII. The antisense XbaI/ gene. We screened their activities by counting GFP EcoRI (cl. 12) exhibited a stronger activity than the sense- expressing cells at 12, 24 and 48 h after transfection of oriented cl. 11 and 14. Further deletion constructs were the clones inserted in the trap vector in IMR32 cell line. made based on this sequence as shown in Figure 3a, and Three rounds of screenings resulted in isolation of 11 their activities were compared (Figure 3b). positive clones, and the best clone was chosen for further Four of the deletion constructs shown in Figure 3 analysis. In order to map the trapped fragment in this demonstrated higher activities: the XbaI/EcoRI (cl. 12), clone, BamHI fragments from pMIcos was subcloned its deletion mutants (cl. 18, 21), and the HindIII/HindIII and a restriction map was made. In addition, patterns of (cl. 24). It is obvious that the core enhancer activity restriction digestion were compared between the enhancer resides in the 0.5 kb HindIII/HindIII fragment. These fragments cloned into Bluescript. To determine where constructs yielded similar and stable activities in both the cloned enhancer lies within the genome, Southern IMR32 and BE2 cell lines. This 0.5 kb fragment was transfer analysis with a variety of restriction enzymes was shown to contain a stronger enhancer activity when the conducted with probes made from the positive clone sequence was ligated in the antisense orientation to the indicated in Figure 1. promoter. In BE2 cells, the activity of cl. 12, 18, 21 and 24 The most active clone, #4en, was approximately 2 kb was about threefold higher than that of the promoter long and mapped by hybridization to the 50 flanking alone, p0.3GL3. region. Position of the clone was found to be upstream In order to ensure the cell-type specificity, we compared within 4 kb proximity of the JIB,7 as depicted in Figure 1, activity of the enhancer constructs in several cell lines, being not in the region where the CNS-specific enhancer including the SH-SY5Y, a relatively differentiated neuro- was identified.7 The 2 kb enhancer containing NcoI/ blastoma cell line without N-myc amplification, as well EcoRI fragment was sequenced. The sequence was found as the neuroendocrine SCLC N417. They were compared to lie on Ch10 at location around 42.7 kb upstreamof the with the glioma cell line, U1242, and the hemangioma cell ATG initiation site of Mus musculus MASH1 gene. The line, EOMA. As shown in Figure 4, the promoter active 0.5 kb HindIII/HindIII site carried a spliced stretch p0.3pGL3b showed slight but clearly higher activity in of totally 243 bp homology with Rattus norvegicus IMR32 and SH-SY5Y compared with other three non- sequence at the corresponding area on Ch7. However, neuroblastoma cell lines. This confirms that the promoter only a few stretches of 20–30 bp homologous sequences also has some cell-type specific activity as reported.9 By were found on human Ch12 where HASH1 resides. contrast, the SV40 promoter (cl. 30) did not show any

pMIco s N X X X Sp N

#4en CNSen Mash1 exon s 1000 bps J1B Figure 1 Location of the cloned enhancer, #4en, in the pMIcos relative to the previously reorted J1B cosmid (XbaI–SphI) containing the MASH1 sequence. The trapped #4en (Tsp5091–EcoRI) was located 50 to the J1B, upstream of the previously cloned CNS enhancer (CNSen).

Cancer Gene Therapy MASH1-enhancer active in neuroblastoma F Watt et al 290 H H a P N M Ap Xb RV K Pm S Bx RI

b 140 IMR-32 BE(2) 120 1. pGL3b 2. p0.3pGL3b ∗∗∗

∗∗∗ 3. p2.1en0.3pGL3b (PRI) 100 4. p#4en0.3pGL3b (NRI) 5. p1.8en0.3pGL3b (PBx) 6. p1.4en0.3pGL3b (PH) 80 ∗

∗ 7. p1.2en0.3pGL3b (PK)

∗∗ 8. p0.9en0.3pGL3b (PH) 9. p1.5en0.3pGL3b d(H-H) 60 ∗∗ ∗∗ 10. p1.6en0.3pGL3b (MRI) ∗ 11. p1.3en0.3pGL3b (ApRI) 40 12. p1.2en0.3pGL3b (XbRI)

Relative Luc Activate 13. p1.4en0.3pGL3b (MBx) 14. p1.1.en0.3pGL3b (ApBx) 20 15. p0.8 en0.3pGL3b (ApS)

0 1 23456789101112131415 Figure 2 Enhancer activity of the various deletion mutants of #4en measured by luciferase in two neuroblastoma cell lines. (a) Illustration of the deletion mutants analyzed. The constructs cl. 2–15 contains the 0.3 kb MASH1 promoter. The cl. 1 is the basic pGL3 and the cl. 2 contains only the MASH1 promoter. The orientation of the enhancer in cl. 3, 5–8 are not determined. N, NcoI; M, MunI; Ap, ApoI; X, XbaI; H; HindIII; RV; EcoRV; K, KpnI; Pm, PmlI; S, SacI; Bx, BstXI; RI, EcoRI. (b) The deletion of H-H (cl. 9) reduced the activity of the #4en (5) as seen when compared with cl. 10–15. Relative luciferase activities are shown as comparison with the MASH promoter vector (cl. 2). Differences were considered significant at *Pp0.05, **Pp0.01 or ***Pp0.001. Only increased values were marked.

neuroblastoma-specificity. Again, the 0.5 kb antisense (CNS/PNSen) was tested in the same manner (Figure 4, cl HindIII/HindIII construct yielded the highest activity in 27). This has been shown to give a broader expression in IMR32 (cl. 24). The p1.6en0.3pGL3 (cl. 10) yielded about the NS.7 None of these constructs showed any clear the same activity as the MASH1 promoter alone, activity in the tested cell lines when compared with the however, the activity in SH-SY5Y was lower. The N417 0.3 kb MASH promoter vector. SCLC showed a significantly higher activity of the enhancer (cl. 24, 28), compared with that of the promoter Core enhancer sequence alone. In order to identify the most active core enhancer area, With all the constructs examined, the strongest further deletion mutants were made in the p1.2enXbaI/ response was seen with the HindIII/HindIII in IMR32, EcoRI containing the 0.5 kb HindIII/HindIII and the more than twofold of the activity seen in other cell lines. activities examined in IMR32 and BE2 (Figure 5a). In the A weak enhancer activity was also found in the proximity of the 30 HindIII toward the downstream neuroendocrine cell line N417 (cl. 24). Cell lines lacking EcoRI, there is a GAGA-stretch, which has been shown the HASH1 expression, U1242 and EOMA, showed no to link an enhancer to its cognate promoter through POZ- clear specificity to the MASH1 promoter/enhancer. The domain-mediated GAGA oligomerization.15 As shown in MASH1 promoter activity was comparable in both Figure 5b, the deletion of SacI/BstXI fragment, contain- IMR32 and SH-SY5Y but the enhancer activity was not ing the GAGA-stretch (cl. 18) or a slightly larger deletion appreciable in SH-SY5Y. of PmI/BstXI (cl. 21) containing also an E-box, did not The previously reported 1158-bp 50-MASH1 CNS change the activity of the original XbaI/EcoRI (cl. 12), enhancer7 was inserted upstreamof the 0.3 bp MASH1 and the activity is comparable to the 0.5 kb antisense promoter in the pGL3 vector (CNSen) and the luciferase HindIII/HindIII (cl. 24). The results are concordant with activity was measured (Figure 4, cl. 26). The deletion of those shown in Figure 3b. This fragment showed always the 118 bp sequence at the 30-end of this CNS enhancer better activity in the antisense orientation (cl. 23, 24), but

Cancer Gene Therapy MASH1-enhancer active in neuroblastoma F Watt et al 291 H H a P N M Ap Xb RV K Pm S Bx RI

10 11 12 16 17 18 19 20 21 22 23 24

b 180 1. pGL3b IMR-32 2. p0.3pGL3b 160 BE(2) ∗∗ 10. p1.6en0.3pGL3b (MRI)

∗∗∗ 11. p1.3en0.3pGL3b (ApRI)

140 ∗∗∗ 12. p1.2en0.3pGL3b (XbRI) ∗∗∗ ∗∗∗ 16. p1.4en0.3pGL3b (MRI) d(SBx) 120 ∗∗∗ ∗∗∗ ∗∗∗ 17. p1.1en0.3pGL3b (ApRI) d(SBx)

100 ∗∗ 18. p0.9en0.3pGL3b (XbRI) d(SBx) ∗

∗∗ 19. p1.5en0.3pGL3b (MRI) d(PmS) ∗∗ ∗ ∗∗ ∗∗ ∗∗ 80 ∗∗ 20. p1.2en0.3pGL3b (ApRI) d(PmS) ∗ 21. p0.8en0.3pGL3b (XbRI) d(PmBx) ∗∗ 60 22. p1.2en0.3pGL3b (ApRI) d(RVK) 23. p0.5en0.3pGL3b (H-H)(+) Relative Luc Activities 40 24. p0.5en0.3pGL3b (H-H)(-) 25. pGL3-cont vector (SVpen) 20

0 1210 11 12 16 17 18 19 20 21 22 23 24 25 Figure 3 Enhancer functions of further deletion mutants of #4en (MunI–EcoRI) analyzed in IMR32 and BE2. (a) All the constructs except cl. 1 and cl. 25 contain 0.3 kb MASH1 promoter upstream of the various enhancer fragments as indicated. The cl 25 contains SV40 promoter/ enhancer. (b) The shortest fragments covering the 0.5 kb H-H in the antisense direction (cl. 18, 21, 24) exhibited the highest activity. Relative luciferase activities are shown as comparison with the MASH promoter vector. Significances are indicated as for Figure 2.

120 1. pGL3b. IMR-32* 2. p0.3GL3b ** 26. p1.1E0.3pGL3b(+) (CNSen) SH-SY5Y 100 27. p1.0E0.3pGL3b(+) (CNS/PNSen) N417 10. p1.6en0.3pGL3b(+) (M-RI)(-) 23. p0.5en0.3pGL3b (H-H)(+) U-1242 24. p0.5en0.3pGL3b (H-H)(-) EOMA 80 28. PSVpGL3b-1.6en (M-RI)(+) 29. PSVpGL3b-1.6en (M-RI)(-) 30. PSVpGL3b (promoter vect) 60 *

* **

Relative Luc Activities Luc Relative 40

0 1 2 26 27 10 23 24 28 29 30 Figure 4 Activities of the best enhancer fragments of #4en (cl. 23, 24) compared with the MASH1CNSen (cl. 26), CNS/PNS enhancer (CNS/ PNSen, cl. 27) placed upstream of the 0.3 kb MASHI promoter. The 1.6 kb (M-RI) #4en was also tested in combination with the heterologous SV40 promoter (cl. 28, 29). IMR32 and SHSY5Y and non-neuroblastoma, N417 SCLC, U1242 glioma, and EOMA hemangioma cell lines were used. 1, the basic pGL3, 2, 0.3 kb MASH1 promoter, 10, and pSV40 promoter alone. The constructs 2–7 contain 0.3 kb MASH1 promoter. Relative luciferase activities compares with the MASH promoter vector.

Cancer Gene Therapy MASH1-enhancer active in neuroblastoma F Watt et al 292 a HindIII HindIII Xbal Eco RV Kpnl Pmll Sacl Bstxl EcoRI

p1.2en(Xb - RI) 12.

p0.9en(XbRI) d(Sc-Bx) d(GAGA) 18.

p0.8 en(XbRI) d(Pm-Bx) d(GA/Eb) 21.

p0.5en(Ha-Hb) 23, 24.

p0.4en(Ha-Hb) d(RV-Kp) 31, 32.

p0.3en(Ha-Kp) 33.

p0.2en(Ha-RV) 34.

p0.3en(Rv-Hb) 35. p0.2en(Kp-Hb) 36.

4.5 b * IMR-32 * 4.0 BE(2) 3.5 2. p0.3GL3b 12. p1.2en0.3pGL3b (XbRI) 3.0 18. p0.9en0.3pGL3b (XbRI) d (GAGA) 21. p0.8en0.3pGL3b (XbRI) d (GA/Eb) 2.5 23. p0.5en0.3pGL3b (H3a-H3b) (+) 24. p0.5en0.3pGL3b (H3b-H3a) (-) 2.0 34. p0.2en0.3pGL3b (H3a-RV) (+) 33. p0.3en0.3pGL3b (H3a-Kpn) (+) 1.5 31. p0.4en0.3pGL3b (H3a-H3b) d (RV-Kpn) (+)

Relative Luc Activities 32. p0.4en0.3pGL3b (H3b-H3a) d (Kpn-RV) (-) 1.0 36. p0.2en0.3pGL3b (H3b-Kpn) (-) 35. p0.3en0.3pGL3b (H3b-RV) (-) 0.5 25. pGL3 cont vector (SVpen)

0.0 2 12 18 21 23 24 34 33 3132 36 35 25 Figure 5 (a) The p1.2 kb XbaI/EcoRI (XbRI) and the p0.5 kb HindIII/HindIII (H3a-H3b, the upstream HindIII is denoted as H3a and the downstream H3b) were used to determine the activity of the core enhancer fragment. The 50–30 orientation is indicated as þ and the 30–50 as –. The p0.9en(XbRI) lacks the GAGA-sequence containing SacI/BstXI, and the p0.8en(XbRI) lacks the PmII/BstXI that contains GATA-binding sites and E-box. (b) The luciferase activities were compared in IMR32 and BE2 cell lines. The shortest fragment containing the sufficient enhancer activity was the 0.2 kb H3a–RV (cl. 34). Relative luciferase activities are shown as comparison with the MASH promoter vector.

a deletion of the middle segment EcoRV/KpnI reduced the EcoRV area. Transcription binding motifs found in the repression seen in the sense orientation. The 30 deletion 0.5 kb HindIII/HindIII is schematically depicted in constructs 0.2 kb HindIII/EcoRV (cl. 34) and 0.3 kb Figure 6a, containing several binding sites for AP-1, HindIII/KpnI (cl. 33), both in the sense orientation, GATA2/3, Nkx-2, Oct1, CREB/CRE-BP, BMP respon- showed slightly higher activities than the 0.5 kb antisense sive elements and USF. HindIII/HindIII (cl. 23, 24). In contrast, the 30 part of the In order to determine which proteins bind to the HindIII/HindIII, the 0.2 kb KpnI/HindIII (cl. 36) as well enhancer, two probes were made around the recognition as the 0.3 kb EcoRV/HindIII (cl. 35), both in the antisense sites for the above-mentioned transcription factors. One orientation, yielding a lower activity than the promoter of the probes is a 55-mer oligo containing putative alone. Taken together, the 0.2 kb HindIII/EcoRV is binding sites for CREB, CRE-BP, GATA2/3, AP1, AP2, sufficient for the activity, independent of its orientation. ATF-EIIa and NF-1 (Figure 6a). The EMSA with NE of BE2 showed two major bands, one upper weak and the Transcription factors binding the core enhancer lower stronger one (Figure 6b). The self-consensus oligo Analysis of the various constructs indicates that the efficiently blocked the upper band and decreased most of enhancer core is located around the 0.2 kb HindIII/ the lower bands. None of the GATA2/3, AP2, AP1 or

Cancer Gene Therapy MASH1-enhancer active in neuroblastoma F Watt et al 293 a HindIII EcoRV KpnI HindIII

55-mer Oligo

82bp Hinf frag.

USF, Myc CREB, CRE-BP AP-1 GATAs

Oct-1 or BMP Nkx-2 its related

55-mer oligo CREB-SC CREB-m bc55-mer #4en oligo m m

EB-SC EB-SC EB-SC EB-SCm REB-SC R R -2/3 -SC -2/3 -SC self C C CR C CR TA TA EB

A P-2 REB A P-2 R + +

– self G A AP-1 ATF-EIIaC NF-1 G A AP-1 ATF-EIIaC NF-1

+ – – + + + + + – + 1234567891011 12 13 14 1234567891011

Figure 6 (a) The candidate transcription factors binding to the #4en core. The probes used for EMSA are indicated as black bars. (b) The 55- mer oligo was labeled and incubated with NE of IMR32 alone or with excess amounts of different consensus oligos described on each lane 2–8. The each consensus oligo used for blocking was labeled and incubated with the NE for EMSA (lane 9–14). (c) The 55-mer oligo was labeled and incubated with 100 Â self oligo, 10 Â ( þ ), 100 Â ( þþ) of CREB consensus oligo or 10 Â of a mutated CREB oligo with NE. The CREB consensus and the mutated CREB oligos were labeled and used for EMSA alone or with CREB or the mutated consensus oligo.

NF1 consensus oligos was able to diminish any of the EIIa, suggesting their identities. Furthermore, EMSA was bands. ATF-EIIa-oligo completely blocked the upper performed to examine whether the binding protein is band, and the lower band was decreased, yet still visible. CREB or not, by competition with the consensus or a CREB-oligo abrogated the upper band and the middle mutated CREB oligo. The upper two bands were band, but not the lowest band (Figure 6c). competed out at the similar efficiency with the self and The presence of these transcription factors in the NE the CREB consensus oligos but not with the mutated one. was confirmed by using consensus oligos as probes. When the CREB consensus oligo was used as a probe, a Although these probes are smaller than the 55-mer probe similar pattern was seen, which again disappeared by a similar binding pattern shown by the 55-mer probe was adding the consensus but not the mutated oligo. As detected with the consensus probes of CREB and ATF- expected, a labeled mutated CREB oligo did not build

Cancer Gene Therapy MASH1-enhancer active in neuroblastoma F Watt et al 294 any complex. Repeated EMSA with NE from IMR32 specific differentiation of stemcells or progenitors. They showed similar results as compared with BE2 (results not can be employed clinically to obtain a tissue-specific gene shown). expression for diagnosis and therapy in patients with Another probe used was a HinfI fragment that overlaps deficiencies of vital proteins, or tumors that must be with the 30 half of the 55-mer probe extending to the eliminated. This can be achieved by an optimal combina- further downstream, including an E-box but not the tion of enhancers/promoters placed upstream of a marker CREB site (Figure 6a). NE fromBE2 and IMR32 as well or therapy gene. Such an approach could be used for as the breast cancer cell line MCF7 were used. A series of advanced neuroblastomas that are resistant against EMSA indicates that USF is present in all the cell lines conventional therapies, and which presently lack effective examined, and the protein binding the enhancer probe is therapy. not a single protein (not shown). The DNA-binding The human homolog of MASH1, HASH1, is expressed protein complex seems to contain different proteins in in 66% of primary neuroblastomas17 and in six out of the neuroblastoma cell lines from those in MCF7 (not six neuroblastoma cell lines derived from the sympathetic shown). nervous system.18 These reports also described that In order to see whether the above-mentioned transcrip- HASH1 transcripts became downregulated in several tion factors would bind to the core enhancer region in cell lines during induced differentiation. HASH1 is also vivo, Chip assays were performed with mouse neuro- expressed in other neuronal and neuroendocrine malig- blastoma C1300 cells. As expected, CREB and the co- nancies such as SCLC and medullary thyroid carcinoma, activator CBP were found to bind chromatin within the which are highly metastatic tumors with lethal out- enhancer region. In addition, binding of N-myc but not comes.19 Many other neuroendocrine tumors, such as USF to chromatin was evident (Figure 7). The co- pheochromocytomas and carcinoids, also express activator PCAF was also detected to bind, but the HASH1,19,20 while the expression is completely absent corepressor HDAC1 did not. Interestingly, N-myc as in non-neuroendocrine tumor cells or adult normal well as both of the co-activators bound the 0.3 kb cells.19,21 In these neuroendocrine tumors, HASH1 MASH1 promoter region. No binding was seen with expression is strongly associated with the neuroendocrine any of the antibodies on a MASH1 coding area. These properties.20 An SCLC cell line, N417, yielded a weak results confirman active status of the enhancer chromatin but specific enhancer activity in our study. Whether this in this cell line. Immunoblotting confirmed that CREB enhancer can be also used for a targeted gene therapy of is expressed in SHSY5Y and BE2 cells, and N-myc is SCLC and other neuroendocrine tumors deserves further overexpressed in IMR32, BE2 and C1300 cells (not study. shown).16 We used the murine MASH1 0.3 kb promoter to trap its enhancer. This promoter is sufficient to yield a promoter activity when compared with longer promoters, that is, 0.8 and 1.2 kb promoters upstream of the Discussion initiation site in the 50-flanking region.9 The smaller the regulatory region, the easier it is to obtain a flexible Cell-type specific enhancers have been widely applied for therapy gene cassette. It has been reported that this 0.3 kb various purposes, that is, to create transgenic or knockout promoter, including repeated Sp1-consensus sites, is mice targeted in a specific cell type, and to mark a lineage- conserved in mouse and human.19,22 In addition, the a-USF1 a-HDAC1 a-CBP a-PCAF a-N-myc a-CREB1 Input mouse IgG

MASH1 Enhancer

MASH1 promoter a-CREB1 a-USF1 a-HDAC1 a-PCAF a-CBP a-N-myc Input mouse IgG

MASH1 coding region

Figure 7 Chip assay over the area covering the MASH1 core enhancer and promoter. Chromatin extracted from C1300 cells was immunoprecipitated with antibodies against CREB, USF, HDAC1, PCAF, CBP, N-myc or mouse IgG. DNA was purified and analyzed by PCR together with whole cell lysate (input), using primers specific for the enhancer and promoter of MASH1 with the coding region as a negative control.

Cancer Gene Therapy MASH1-enhancer active in neuroblastoma F Watt et al 295 proximal region between positions À305 and À234 is a region. A closer inspection of the promoter region binding site for HES1, considered to be the tissue-specific revealed a non-canonical Myc-binding motif CACGCG, repressor.19 In fact, HES1 overexpression directly re- located at À243 fromthe transcription start site 27 and presses HASH1 expression, and this regulatory mechan- being distal to the C-site known to bind HES. Even closer ismhas been shown to be involved in the normal to the initiation site there are CpG islands, possibly development of the nervous system.22,23 For gene target- facilitating the binding of N-myc to the CACGCG ing, inclusion of this sequence could add the tissue motif.9,28 However, transfection of N-myc into the specificity by excluding non-neural tissues that contain a neuroblastoma cell lines could not augment the MASH1 high level of HES1. Hence, it can contribute to the enhancer-promoter luciferase activity (result not shown). specificity in neuronal cells over other cell types. Recently, This unresponsiveness might be due to the abundant it has been reported that HES1 together with the endogenous N-myc expressed in neuroblastoma. These corepressor TLE1 is bound to the MASH1-promoter cell lines constitutively express HASH1, and N-myc is one and silences it. HES1 can be phosphorylated by calmo- of the best markers associating with the aggressive dulin kinase IId, forming a complex with a co-activator progression of advanced neuroblastoma.29 The activity on the promoter in order to act as an activator in of this enhancer therefore might serve as a good indicator proneural cells.24 Thus, HES1 bound on this promoter for endogenous expression of both HASH1 and N-myc can alter its activity depending on the stimulus, providing proteins in neuroblastoma cells. By analogy, Yazawa a possibility for broader cell type-specific responses. et al.30 demonstrated that N-myc regulates HASH1, and We identified a MASH1 enhancer/promoter combina- both are bound to the E-box in class II transactivator, tion that provided a tissue-specific expression of a which is strongly downregulated in SCLC and neuro- reporter gene. Furthermore, the best MASH1 enhancer/ blastoma. It is thus possible that these transcription promoter combination containing the core enhancer factors, being overexpressed and/or amplified specifically region was more active in N-myc amplified neuroblastoma in such neuroendocrine tumors, are functional for cell lines, correlating with the HASH expression. This maintenance of their phenotypes. suggests that normal neuroendocrine cells and tissues may Johnson et al.7 previously cloned an enhancer of not be affected in vivo, ensuring the safety of the gene MASH1 by using a transgenic model, which confers therapy using this targeting cassette. CNS-specific MASH1 expression. As expected, this The core enhancer region was narrowed to 0.2 kb of the enhancer could not show any activity in the cell lines enhancer, and the transcription factors binding to this derived fromthe PNS, which we tested here. The region were examined in order to understand the redundancy of MASH1 enhancers spread in this gene is mechanism how the MASH1 enhancer confers it activity. concordant with its transient but broad expression in the The factors that were found to bind the sequence were autonomic nervous system in the PNS, as well as in CREB/ATF2. However, other factors that were thought neuronal progenitor cells in the CNS. to bind to this area, AP1, AP2 or GATA2/3, fail to do so. In conclusion, we have cloned and characterized a Members of the CREB/CREM/ATF family of transcrip- MASH1 enhancer that is active in neuroblastoma cell tion factors either enhance or repress transcription after lines. This enhancer might be employed to target a binding to the cAMP response elements (CREs) of therapy-gene expression in neuroblastoma and other numerous genes.25 CREB-mediated gene expression is neural crest-derived tumors in vivo tumor models. particularly important in the nervous system, regulating neuronal survival, memory, biological clock and synapto- genesis. The broad spectrumof its activity mightdepend Acknowledgements on the fact that CREB target genes frequently encode transcription factors such as Egr1, Nurr1 and herein We are grateful to Dr J Johnson for the McosI cosmid MASH1. ATF is considered to be a preferentially and her valuable comment on the manuscript, as well as expressed CREB family member in the nervous system. to Drs O Nilsson, S Pa˚ hlman, RA Reisfeld, L Welsh, and ATF is induced during neuronal differentiation, and B Westermark for cell lines. We thank K Bjo¨ rklund and when neuronal cells are injured ATF protects the cells T Olsson for excellent technical assistance. This work was by inducing various genes.26 However, overexpression of supported by grants fromthe Swedish Science Council, CREB but not of ATF upregulated the MASH1 the Swedish Cancer Society, the Swedish Children’s enhancer/promoter activity in BE2 cells (result not Cancer Society, STINT, the Swedish Institute, the shown). The co-activator CBP was also found on the Hjalmar Svensson Fund, and the Medical Faculty at chromatin, binding CREB to enhance the transcription Go¨ teborg University. FW was partly supported by the activity. Swedish Institute. Slightly downstreamof the CREB-binding site there is an E-box, where a certain protein was found to bind in our EMSA. This protein was present in the NE fromthe References two neuroblastoma cell lines, but not that of the MCF7 breast cancer cell line (results not shown). Chip assay in 1 Guillemot F, Joyner AL. Dynamic expression of the murine C1300 cells revealed that N-Myc was strongly bound to Achaete-Scute homologue Mash-1 in the developing nervous this enhancer chromatin and also the proximal promoter system. Mech Dev 1993; 42: 171–175.

Cancer Gene Therapy MASH1-enhancer active in neuroblastoma F Watt et al 296 2 Ma Q, Kintner C, Anderson DJ. Identification of neurogen- 17 Ichimiya S, Nimura Y, Seki N, Ozaki T, Nagase T, in, a vertebrate neuronal determination gene. Cell 1996; 87: Nakagawara A. Downregulation of hASH1 is associated 43–52. with the retinoic acid-induced differentiation of human 3 Johnson JE, Birren SJ, Anderson DJ. Two rat homologues neuroblastoma cell lines. Med Pediatr Oncol 2001; 36: of Drosophila achaete-scute specifically expressed in neuronal 132–134. precursors. Nature 1990; 346: 858–861. 18 SoderholmH, Ortoft E, Johansson I, Ljungberg J, Larsson 4 Anderson DJ. Stemcells and transcription factors in the C, Axelson H et al. Human achaete-scute homologue 1 development of the mammalian neural crest. FASEBJ 1994; (HASH-1) is downregulated in differentiating neuroblastoma 8: 707–713. cells. Biochem Biophys Res Commun 1999; 256: 557–563. 5 Guillemot F, Lo LC, Johnson JE, Auerbach A, Anderson 19 Chen H, Biel MA, Borges MW, ThiagalingamA, Nelkin BD, DJ, Joyner AL. Mammalian achaete-scute homolog 1 is Baylin SB et al. Tissue-specific expression of human achaete- required for the early development of olfactory and scute homologue-1 in neuroendocrine tumors: transcrip- autonomic neurons. Cell 1993; 75: 463–476. tional regulation by dual inhibitory regions. Cell Growth 6 Huber K, Bruhl B, Guillemot F, Olson EN, Ernsberger U, Differ 1997; 8: 677. Unsicker K. Development of chromaffin cells depends on 20 Ball DW, Azzoli CG, Baylin SB, Chi D, Dou S, Donis-Keller MASH1 function. Development 2002; 129: 4729–4738. H et al. Identification of a human achaete-scute homolog 7 Verma-Kurvari S, Savage T, Gowan K, Johnson JE. highly expressed in neuroendocrine tumors. Proc Natl Acad Lineage-specific regulation of the neural differentiation gene Sci USA 1993; 90: 5648–5652. MASH1. Dev Biol 1996; 180: 605–617. 21 Borges M, Linnoila RI, van de Velde HJ, Chen H, Nelkin 8 GestblomC, Grynfeld A, Ora I, Ortoft E, Larsson C, BD, Mabry M et al. An achaete-scute homologue essential Axelson H et al. The basic helix-loop-helix transcription for neuroendocrine differentiation in the lung. Nature 1997; factor dHAND, a marker gene for the developing human 386: 852–855. sympathetic nervous system, is expressed in both high- and 22 Meredith A, Johnson JE. Negative autoregulation of low-stage neuroblastomas. Lab Invest 1999; 79: 67–79. Mash1 expression in CNS development. Dev Biol 2000; 9 Arvidsson Y, Sumantran V, Watt F, Uramoto H, Funa K. 222: 336–346. Neuroblastoma-specific cytotoxicity mediated by the Mash1- 23 Sasai Y, Kageyama R, Tagawa Y, Shigemoto R, Nakanishi promoter and E. coli purine nucleoside phosphorylase. S. Two mammalian helix-loop-helix factors structurally Pediatric Blood Cancer 2005; 44: 77–84. related to Drosophila hairy and Enhancer of split. Genes 10 Akyurek LM, Yang ZY, Aoki K, San H, Nabel GJ, Dev 1992; 6: 2620–2634. Parmacek MS et al. SM22alpha promoter targets gene 24 Ju BG, SolumD, Song EJ, Lee KJ, Rose DW, Glass CK expression to vascular smooth muscle cells in vitro and in et al. Activating the PARP-1 sensor component of the vivo. Mol Med 2000; 6: 983–991. groucho/TLE1 corepressor complex mediates a CaMKinase 11 Watt F, Martorana A, Brookes DE, Ho T, Kingsley E, IIdelta-dependent neurogenic gene activation pathway. Cell O’Keefe DS et al. A tissue-specific enhancer of the prostate- 2004; 119: 815–829. specific membrane antigen gene, FOLH1. Genomics 2001; 73: 25 Carlezon Jr WA, Duman RS, Nestler EJ. The many faces of 243–254. CREB. Trends Neurosci 2005; 28: 436–445. 12 Johnson JE, Birren SJ, Saito T, Anderson DJ. DNA binding 26 Zaman K, Ryu H, Hall D, O’Donovan K, Lin KI, Miller and transcriptional regulatory activity of mammalian MP et al. Protection fromoxidative stress-induced apoptosis achaete-scute homologous (MASH) proteins revealed by in cortical neuronal cultures by iron chelators is associated interaction with a muscle-specific enhancer. Proc Natl Acad with enhanced DNA binding of hypoxia-inducible factor-1 Sci USA 1992; 89: 3596–3600. and ATF-1/CREB and increased expression of glycolytic 13 Izumi H, Molander C, Penn LZ, Ishisaki A, Kohno K, Funa enzymes, p21(waf1/cip1), and erythropoietin. J Neurosci K. Mechanismfor the transcriptional repression by c-Myc 1999; 19: 9821–9830. on PDGF beta-receptor. J Cell Sci 2001; 114: 1533–1544. 27 Blackwell TK, Huang J, Ma A, Kretzner L, Alt FW, 14 Uramoto H, Wetterskog D, Hackzell A, Matsumoto Y, Eisenman RN et al. Binding of myc proteins to canonical Funa K. p73 competes with co-activators and recruits and noncanonical DNA sequences. Mol Cell Biol 1993; 13: histone deacetylase to NF-Y in the repression of PDGF 5216–5224. beta-receptor. J Cell Sci 2004; 117: 5323–5331. 28 Brenner C, Deplus R, Didelot C, Loriot A, Vire E, De Smet C 15 Mahmoudi T, Katsani KR, Verrijzer CP. GAGA can et al. Myc represses transcription through recruitment of DNA mediate enhancer function in trans by linking two separate methyltransferase corepressor. EMBO J 2005; 24: 336–346. DNA molecules. EMBO J 2002; 21: 1775–1781. 29 Brodeur GM. Molecular pathology of human neuroblasto- 16 Kato H, Okamura K, Kurosawa Y, Kishikawa T, mas. Semin Diagn Pathol 1994; 11: 118–125. Hashimoto K. Characterization of DNA rearrangements of 30 Yazawa T, Ito T, Kamma H, Suzuki T, Okudela K, Hayashi N-myc gene amplification in three neuroblastoma cell lines H et al. Complicated mechanisms of class II transactivator by pulsed-field gel electrophoresis. FEBS Lett 1989; 250: transcription deficiency in small cell lung cancer and 529–535. neuroblastoma. Am J Pathol 2002; 161: 291–300.

Cancer Gene Therapy