The FASEB Journal • Research Communication

Keratin 7 promoter selectively targets transgene expression to normal and neoplastic pancreatic ductal cells in vitro and in vivo

Judit Pujal,*,1 Meritxell Huch,†,§,1 Anabel Jose´,† Ibane Abasolo,* Annie Rodolosse,*,‡ ʈ ʈ Alba Duch,*,‡ Luis Sa´nchez-Palazo´n,† Frances J. D. Smith, W. H. Irwin McLean, Cristina Fillat,†,§,2 and Francisco X. Real*,‡,¶,2 *Unitat de Biologia Cel.lular i Molecular, Institut Municipal d’Investigacio´Me`dica, Parc de Recerca Biome`dica de Barcelona, Barcelona, Spain; †Centre de Regulacio´ Geno`mica and ‡Departament de Cie`ncies Experimentals i de la Salut, Universitat Pompeu Fabra, Barcelona, Spain; §Centro de ʈ Investigacio´n Biome´dica en Red de Enfermedades Raras, Barcelona, Spain; Epithelial Genetics Group, Division of Molecular Medicine, Colleges of Life Sciences and Medicine, Dentistry, and Nursing, University of Dundee, Dundee, UK; and ¶Programa de Patología Molecular, Centro Nacional de Investigaciones Oncolo´gicas, Madrid, Spain

ABSTRACT 7 is expressed in simple epithelia homodimers result from the tissue-specific expression but is expressed at low or undetectable levels in gastroin- of each family member, for example, K8/K18 in simple testinal epithelial cells. In the pancreas, it is present in epithelia (1, 2). Much evidence on keratin function has ductal but not in acinar cells. K7 mRNA is overexpressed come from the analysis of genetically modified mice (3, in pancreatic cancers. Here we use luciferase reporter 4) and from the study of mutations in human keratin assays to analyze the tissue-specific regulatory elements of , which are associated with inherited diseases, murine keratin 7 (Krt7) promoter in vitro and in vivo. All such as cryptogenic liver disease (5, 6). elements required for appropriate cell and tissue speci- K7 and K19 are expressed in a subset of simple ؊ ficity in reporter assays are present within the Krt7 234 epithelia; their function has not been well elucidated bp sequence. This fragment appears more selective to (3, 7). These polypeptides are of particular interest in pancreatic ductal cells than the Krt19 promoter. GC-rich the study of gastrointestinal and hepatic biology: K7/ sequences corresponding to putative Sp1, AP-2 binding K19 are expressed in a subset of cells in the pancreas, sites are essential for in vitro activity. Krt7-LacZ transgenic bile duct, kidney, breast, and bladder (1, 8). In the mice were generated to analyze in vivo activity. Sequences pancreas they are present in ductal and centroacinar located 1.5 or 0.25 kb upstream of the transcription cells but are absent from acinar and endocrine cells initiation site drive reporter expression to ductal, but not (8); in the liver they are restricted to the biliary acinar, cells in transgenic mice. LacZ mRNA was detected and are absent from hepatocytes (9). In in the pancreas as well as in additional epithelial tissues— general, K7 and K19 are coexpressed (8, 9); a remark- such as the intestine and the lung—using both promoter constructs. An AdK7Luc adenovirus was generated to able exception is the gastrointestinal epithelium, where assess targeting selectivity in vivo by intravenous injection K19 is present in all epithelial cells, whereas K7 is to immunocompetent mice and in a xenograft model of undetectable (1, 9). K7 is also regulated in the context . The ؊0.25 kb region showed pancreatic of differentiation in pancreatic ductal adenocarcinoma selectivity, high activity in pancreatic cancers, and sustained (PDAC) (10). transgene expression in xenografts. In conclusion, the krt7 The murine Krt19 promoter is suitable to drive promoter is useful to target pancreatic ductal adenocarci- transgene expression in pancreatic ductal cells as well noma cells in vitro and in vivo.—Pujal, J., Huch, M., Jose´, A., as in other tissues (11, 12), and it has been used to Abasolo, I., Rodolosse, A., Duch, A., Sa´nchez-Palazo´n, L., develop novel animal models of PDAC (13). By con- Smith, F. J. D., McLean, W. H. I., Fillat, C., Real, F. X. trast, the promoter of the coding for K7 has not Keratin 7 promoter selectively targets transgene expression to normal and neoplastic pancreatic ductal cells in vitro and 1 in vivo. FASEB J. 23, 1366–1375 (2009) These authors contributed equally to this work. 2 Correspondence: F.X.R., Programa de Patología Molecu- lar, Centro Nacional de Investigaciones Oncolo´gicas, Calle Key Words: ⅐ adenovirus ⅐ cystic fibrosis ⅐ biolu- Melchor Ferna´ndez Almagro 3, 28029-Madrid, Spain. E-mail: minescent imaging [email protected]; C.F., Centre de Regulacio´ Geno`mica-CRG, Edifici Parc de Recerca Biome`dica de Barcelona, Carrer del Dr. Aiguader, 88, 08003-Barcelona, Spain. E-mail: cristina. (k) are selectively expressed in epithelial fi[email protected] cells. In vivo, tetramers formed by pairs of 2 classes of doi: 10.1096/fj.08-115576

1366 0892-6638/09/0023-1366 © FASEB yet been characterized. A potential advantage of using GAATCTTCTTGTGA 3Ј; LacZ forward GACGTCTCGTTGCT- K7 regulatory sequences to target the expression of GCATAA, LacZ reverse CAGCAGCAGACCATTTTCAA; Krt19 Ј Ј transgenes is the fact that it has a more restricted tissue forward 5 CCTCCCGAGATTACAACCACT, Krt19 reverse 5 GGCGAGCATTGTCAATCTGT; Krt7 forward CACGAACAAG- distribution while maintaining a selective expression in GTGGAGTTGGA, Krt7 reverse TGTCTGAGATCTGCGACT- ductal cells in the pancreas. Because there is a dearth of GCA; Hprt forward GGC CAG ACT TTG TTG GAT TTG, Hprt knowledge on promoters that can target transgenes to reverse TGC GCT CAT CTT AGG CTT TGT. Total RNA was pancreatic ductal cells, the study of the K7 promoter is normalized using QuantumRNA 18S Internal Standards at a 1:4 a worthwhile endeavor. Here we analyze the murine ratio (Ambion), as indicated in the text. Krt7 promoter and identify regions required for pan- creatic ductal cell-restricted expression in vitro and Site-directed mutagenesis in vivo. In addition, we have addressed its ability to selectively drive transgene expression to the pancreas For site-directed mutagenesis, the Quick ChangeTM Site- by systemic adenoviral gene transfer (AdK7Luc) and to Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) was test the potency of this adenovirus in PDAC mouse used (Supplemental Information). PCR products were di- gested with DpnI, purified, and used for bacterial transfor- models. The regulatory sequences of murine Krt7 merit mation. Plasmids were verified by sequencing both DNA being used in additional studies to target transgenes to strands. pancreatic ductal cells, and possibly to other simple epithelial cell types. This work may contribute to im- Reporter assays proved genetic therapeutic strategies in conditions such as cystic fibrosis and PDAC. Cells were seeded in 24-well plates; 24 h later, cells were transfected using Gene Porter reagent (Gene Therapy Sys- tems, San Diego, CA, USA). pGL3-reporter plasmids (100 ng) MATERIALS AND METHODS were cotransfected with a plasmid encoding Renilla reniformes luciferase to normalize for transfection efficiency. Luciferase activity in total cell lysates, determined 48 h later, was Cell culture expressed as fold activity relative to the control plasmid after correction for transfection efficiency; in some cases results Tumor cell lines (see Supplemental Information) were main- were expressed per mg cellular . Experiments were tained as described elsewhere (10, 14). Human pancreatic performed in triplicate, and results were verified at least twice duct epithelial (HPDE) cells were maintained as described independently. The total amount of DNA transfected was elsewhere (15). always normalized using the corresponding empty vector. For in vitro adenoviral transduction studies, cells were Plasmid constructs seeded in triplicate at a density of 2 ϫ 104 cells/well in 96-well plates. After 24 h, cells were infected at 104 viral particles (vp) A 4.5 kb fragment of Krt7 isolated from a P1-derived artificial per cell; 72 h later, luciferase was assayed as described above (PAC) 129/Sv library (16) was digested to obtain and expressed per microgram of cellular protein. fragments ranging from Ϫ3723/ϩ32 to Ϫ55/ϩ32, which were cloned into pGL3 Basic Luciferase Reporter (Promega, Madi- Electrophoretic mobility shift assays (EMSAs) son, WI, USA). A Krt19 Ϫ1970/ϩ46 luciferase reporter (12) was generated similarly. Plasmid sequences were verified in both Nuclear extracts were prepared as described (17). Double- directions. stranded oligonucleotides (Supplemental Information) were radiolabeled with 32P-dATP; EMSAs were performed with 5 Immunocytochemistry ␮g of protein in a final volume of 20 ␮l binding buffer. Mixtures were incubated with 150 fmol of labeled double- K7 expression in pancreas was assayed as described previously stranded oligonucleotides for 30 min at 4°C. For competition (8). Cells were fixed with methanol:acetone (1:1), washed experiments, a 50-fold excess of unlabeled oligonucleotides with PBS, and incubated for 1 h with RCK105 (1:5 dilution of was added for 15 min at 4°C prior to radiolabeled probes. To hybridoma supernatant) (8), a kind gift of F. C. S. Ramaekers supershift, antibodies detecting Sp1 (PEP2-G; Santa Cruz (University of Maastricht, Maastricht,The Netherlands). After Biotechnology, Santa Cruz, CA, USA), Sp3 (D-20-G, Santa washing, cells were incubated with fluorescein-labeled anti- Cruz), or AP-2 (H-79, Santa Cruz) were added for 20 min at mouse Ig (Dako, Glostrup, Denmark), washed, and mounted. 4°C, prior to addition of radiolabeled probes. DNA-protein complexes were electrophoresed, and dried gels were ex- Reverse transcriptase-polymerase chain reaction posed to phosphorimager screens. analysis (RT-PCR) Generation of transgenic reporter mice using the RNA was isolated from cultured cells or mouse tissues using the Krt7 promoter RNeasy Extraction Kit (Qiagen, Hilden, Germany), treated with DNaseI, reverse transcribed using Moloney leukemia virus re- All animal experiments were approved by Departament verse transcriptase (Ambion, Austin, TX, USA), and amplified d’Agricultura Ramaderia i Pesca (DARP), Generalitat de for 32–40 cycles according to transcript abundance. Following Catalunya, Spain, and performed following recommenda- denaturation (30 s at 95°C), annealing (30 s at 60°C) and tions for proper care of laboratory animals. Krt7 1.5 (Ϫ1474)- extension (30 s at 72°C) were performed in a thermal cycler Luc and Krt7 0.25 (Ϫ234)-Luc fragments were cloned at the (GeneAmp PCR System 9700; Applied Biosystems, Foster City, BamHI restriction site of plasmid Krt19-␤-Gal, a kind gift of CA, USA). The following primers were used: K7 forward 5Ј Dr. A. Rustgi (University of Pennsylvania, Philadelphia, PA, CAGGAACTCATGAGCGTGAA 3Ј, K7 reverse 5Ј GGGTGG- USA) (12). Next, ϳ7.0 kb and ϳ5.75 kb transgenes contain-

THE KRT7 PROMOTER TARGETS DUCTAL PANCREATIC CELLS IN VIVO 1367 ing a lacZ reporter gene under the control of Krt7 1.5 or Krt7 mals were sacrificed, and liver and pancreas were fixed and 0.25 promoter were excised from each construct with NotI, gel embedded in paraffin. Serum aspartate aminotransferase (AST) purified using the GenClean spin kit (Q-Biogene, Carlsbad, and alanine transaminase (ALT) levels were assayed in an CA, USA), and microinjected (2–5 ␮g/ml) into B6CBAF2 Olympus AU400 Analyzer (Olympus, Tokyo, Japan). zygotes, which were transferred to the oviducts of 0.5 days postcoitus (dpc) pseudopregnant CD1 females. Plasmid DNA Adenoviral DNA content in liver was linearized, purified, and microinjected into B6CBA F2 fertilized oocytes, which were transferred to pseudopregnant females. Genotyping of founder mice was performed by DNA was isolated by overnight incubation of frozen tissue at Southern blotting with a 425 bp LacZ probe; genotyping of 55°C in buffer containing 0.2 mg/ml RNaseA and 0.1 mg/ml the progeny was performed by PCR using LacZ-specific prim- protease; adenoviral DNA content was determined using ers. Three founder mice were identified. Transgenic mice real-time PCR (100 ng total DNA) with SYBER Green I Master were maintained by crossing with B6CBA F1 mice. Plus mix (Roche, Indianapolis, IN, USA) and previously reported primers (21). Adenovirus copy number was quanti- fied using a standard curve consisting of adenovirus DNA X-Gal histochemistry (10–107 copies) in a background of mouse genomic DNA. Samples and standards were amplified in triplicate, and the Tissues were fixed overnight with 0.2% paraformaldehyde, average number of copies was normalized to copies/cell cryoprotected with 30% sucrose-PBS, embedded in OCT, and based on the input DNA weight amount and a genome size of frozen in isopentane at Ϫ80°C. Cryosections were fixed with 6 ϫ 109 bp/cell. Results are expressed as vp/100 cells. 0.25% glutaraldehyde for 10 min, preincubated with 0.5% Triton X-100, 5 mM potassium ferrocyanide, 5 mM potassium ferricyanide, and 2 mM MgCl2; reactions were developed with X-gal at room temperature. RESULTS

Viruses Keratin 7 expression in epithelial cancer cells

Replication-defective adenoviruses AdCMVGFPLuc and K7 is expressed in normal pancreatic ducts (Supple- AdK7Luc express firefly luciferase under the control of the mental Fig. 1A) and in most cell lines derived from cytomegalovirus (CMV) or Krt7 0.25 promoter, respectively. pancreatic, bladder, and breast tumors, but it is unde- AdCMVGFPLuc was kindly provided by R. Alemany (ICO, tectable in all colorectal cancer lines examined. Repre- Barcelona, Spain) (18). The AdK7Luc virus was generated by sentative findings are shown in Supplemental Fig. 1B. inserting a 0.25 kb Krt7 promoter fragment and the luciferase gene into pAdTrack. Next, homologous recombination with KRT7 mRNA was analyzed using RT-PCR (Supplemen- the adenoviral genome was carried out (19). Viruses were tal Fig. 1C): high transcript levels were detected in generated and propagated in HEK293 cells and purified by pancreas cancer cells, in immortalized HPDE cells, and CsCl banding (20). Physical particle concentration was deter- in normal pancreas cultures displaying a ductal pheno- mined by optical density reading (OD250). type (ref. 22; not shown); transcripts were undetectable in SW480 (colon cancer) and IMR-90 (fibroblasts), in Bioluminescence and quantification agreement with immunocytochemistry. Among rodent pancreatic cells, Krt7 mRNA was detected in DSL6A Xenografts were generated as described elsewhere (14). cells displaying a ductal phenotype (22) but not among Mice were anesthetized, and D-Firefly-Luciferin (Xenogen, those displaying acinar (AR42J, 266-6) or endocrine Alameda, CA, USA) was administered intraperitoneally (i.p., 16 mg/kg). Luciferase activity was visualized and quantified (Ins1, Min6) phenotypes (not shown). using an in vivo bioluminescent system (IVIS, Xenogen) and the Living Image 2.20.1 software (Xenogen) overlay on Igor Analysis of the transcriptional activity of Krt7 Pro4.06A software (Wavematrics, Seattle, WA, USA) as de- proximal sequences using promoter scribed elsewhere (14). Regions of interest (ROIs) were deletion constructs drawn over tumor or liver areas, and total photons captured were calculated. ROIs in all images were kept at a constant area. Results are expressed as photons/second. RWP-1 (K7-positive) and SW480 (K7-negative) cells were used because they can be transfected at high Biodistribution studies efficiency. High levels of luciferase activity were de- tected in RWP-1 cells after transfection with the Virus (2ϫ1010 vp) was injected via tail vein into Balb/C male Ϫ3723/ϩ43 Krt7 promoter fragment, whereas low lev- mice; 3 and 5 d later, luciferase activity was visualized and els were detected in SW480 cells. The Ϫ1474 and quantified. At d 5, animals were sacrificed, and major organs Ϫ1195 bp fragments had an activity similar to that of were collected and frozen. When stated, frozen tissues were the Ϫ3723 fragment. Lower activity levels were detected mechanically homogenized, and 100 mg was used for protein using fragments Ϫ482 and Ϫ353, but the Ϫ234 frag- extraction using the Cell Culture Lysis Reagent (Promega) ␮ ment displayed high levels of activity (Fig. 1A). Shorter during 15 min at 25°C, and 10 l was assayed for luciferase Ϫ Ϫ Ϫ activity. promoter fragments ( 91, 55, 11) showed moder- ate or undetectable activity. All constructs lacked re- Assessment of liver toxicity porter activity in SW480 cells. Therefore, the Ϫ234 bp fragment contains sufficient information to confer Five days after viral administration, blood samples were obtained both transcriptional activity and tissue selectivity. Addi- by intracardiac injection under anesthesia. Subsequently, ani- tional positive regulatory elements are present in the

1368 Vol. 23 May 2009 The FASEB Journal PUJAL ET AL. Figure 1. Cell-type specific functional analysis of the 5Ј Krt7 sequence using luciferase reporter assays. Comparison with Krt19 promoter activity. A) Deletion reporter constructs of the 5Ј Krt7 promoter were tested by cotransfection with a plasmid encoding Renilla luciferase. Results are normalized for transfection efficiency. B) Reporter construct activity was assayed in a panel of cells derived from tissues with high, medium, or no KTR7 expression. C) Same constructs were assayed in rodent pancreatic tumor cells displaying an acinar (AR42J and 266-6) or undifferentiated phenotype (ARIP). Luciferase activity in B and C is normalized for transfection efficiency and expressed as percentage of RWP-1 activity. D) Krt7-luciferase and Krt19-luciferase reporter plasmids were independently transfected to RWP-1 and SW480 cells, and activity was compared. The Ϫ1.5 kb and Ϫ0.25 Krt7 promoters are 60- and 24-fold more active in pancreatic than in intestinal cells, whereas the Krt19 promoter is 2.7-fold more active in pancreatic cells than in intestinal cells. Therefore, the Ϫ1.5 kb and Ϫ0.25 Krt7 promoters are 22- and 9-fold more pancreas selective than the Krt19 promoter in vitro. Representative results from 3 independent assays are shown. sequences spanning from Ϫ482 to Ϫ1195 and from quence. Putative transcription factor (TF) binding sites Ϫ234 to Ϫ91; negative regulatory elements occur in the were identified using the Tess program (http://www. region encompassing nucleotides Ϫ353 and Ϫ234. cbil.upenn.edu/cgi-bin/tess). High conservation was present in the 500 bp immediately upstream of exon 1 The Krt7 promoter shows tissue-specific activity in between mouse and human, with very high conserva- epithelial cells tion between Ϫ123 and Ϫ74 bp. No TATA box was found in the vicinity of the transcriptional start site. The activity of reporter constructs Ϫ3723/ϩ34 (Ϫ3.7 Ϫ ϩ Ϫ Ϫ ϩ Ϫ Several putative TF binding sites were identified kb), 1474/ 34 ( 1.5 kb), and 234/ 34 ( 0.25 (Supplemental Fig. 2). Further analyses were based on kb) was analyzed in human and rodent lines originating prior information on factors regulating the expression from K7-positive or K7-negative tissues and compared of other keratins, including Sp1, AP-2, and GKLF. In to RWP-1 cells. RWP-1 cells, Sp1 increased—and AP-2 and Elk1 de- High promoter activity levels were detected in RWP-1 creased—the activity of the Ϫ234 bp promoter-reporter (Fig. 1B), MiaPaCa-2, and AsPC-1 (not shown) pancre- construct in a dose-dependent manner; a plasmid cod- atic K7-expressing cells. Moderate activity was detected ing for GKLF had no effect (Fig. 2A). By contrast, both in cells originating from other K7-expressing epithelia, Sp1 and GKLF increased promoter activity in cells such as breast and bladder, and lower levels were found expressing low (VMCUB-2 and 639V) or undetectable in kidney cancers (Fig. 1B). Overall, activity was much (SW480) K7 levels (not shown). Therefore, Sp1 is able higher in pancreatic cells than in other cell types. to increase the activity of the Krt7 promoter regardless These findings suggest that the main sequences provid- of the endogenous levels of K7 expression, unlike AP-2, ing high levels of promoter activity, while maintaining Elk-1, and GKLF. tissue specificity, are present within the Ϫ234 fragment. To determine the role of Sp1 sites, mutant promoter- K7 is expressed at high levels in ductal and centroaci- reporter plasmids were generated by replacing 2 con- nar cells in the pancreas, but not in acinar cells (8). secutive G:C base pairs with A:T bp, in sites Sp1a Promoter activity in 2 acinar cell lines (AR42J and Ϫ Ϫ 266-6), and in ARIP cells (Fig. 1C), was ϳ100-fold lower ( 153), Sp1b ( 82), or both. In RWP-1 cells, Sp1a and than in RWP-1 ductal cells. Sp1b mutations led to 50% and 80% decrease in Ϫ Ϫ reporter activity, respectively. Mutation of both sites The 1.5 kb and 0.25 kb Krt7 promoter reporters Ϫ were 22-fold and 9-fold more pancreas selective than completely abolished the activity of the 234 bp frag- the Krt19 promoter reporter using RWP-1 and SW480 ment (Fig. 2B). Mutation of the Sp1c/AP-2 site almost cells in transient transfection assays (Fig. 1D). completely abolished reporter activity (Fig. 2B). Muta- tion of the distal GKLF site led to a 38% decrease in Identification of cis-regulatory elements within the activity (Fig. 2B). Sp1 overexpression increased Krt7 Krt7 promoter reporter activity when mutated at the Sp1a, Sp1b, or both binding sites: absolute activity levels were substan- To identify relevant sequences within Ϫ234 bp of Krt7 tially lower in mutant constructs, but the fold increase promoter, we compared this region to the KRT7 se- was greater (Fig. 2C and data not shown). Cotransfec-

THE KRT7 PROMOTER TARGETS DUCTAL PANCREATIC CELLS IN VIVO 1369 Figure 2. Identification of relevant sequence motifs involved in Krt7 and promoter activity in RWP-1 cells. A) Activity of the 0.25 kb reporter plasmid was tested on transfection with empty plasmid (gray bars) or with the same plasmid containing the cDNA of Sp1, AP-2, GKLF, or Elk-1. Increasing amounts of relevant plasmids (50 and 100 ng) were tested. Total amount of plasmid DNA transfected was the same under all experimental conditions. Results are shown after normalization for transfection efficiency with a plasmid encoding Renilla luciferase. B) Analysis of the activity of putative Sp1, AP-2, and GFLK binding sites on site-directed mutagenesis. Results are shown as activity relative to that of the wild-type sequence. C) Activity of mutant reporter plasmids on cotransfection of increasing amounts of an Sp1 coding plasmid. D) Activity of GKLF-site mutant reporter plasmid on transfection of increasing amounts of a plasmid coding for GKLF. In C and D, results are shown after normalization for the activity of transfection in the absence of a plasmid coding for Sp1 or GKLF, respectively. tion of the Sp1c/AP-2 mutated reporter and AP-2 K7, the constructs were also active in normal biliary plasmids failed to increase luciferase activity (not duct epithelium (Fig. 3B). Endogenous ␤-galactosidase shown). Transfection of GKLF cDNA did not consis- activity hindered assessing LacZ expression in the in- tently affect reporter activity (Fig. 2D), suggesting that testinal epithelium. Because anti-␤-galactosidase anti- this site is not crucial for K7 expression. bodies are recognized not to always be effective for These results indicate that both Sp1-like sites and the detection of reporter protein from tissue-specific pro- Sp1c/AP-2-like site are fundamental for Krt7 promoter moters, we used RT-PCR. As shown in Fig. 3C, LacZ regulation in RWP-1 cells. transcripts were detected in normal pancreas from mice harboring the Ϫ1.5 kb transgene. In addition, EMSA analysis of nuclear complexes binding to the LacZ mRNA was detected in other simple epithelial Sp1-like sites in the proximal Krt7 promoter tissues, such as the liver, salivary gland, lung, and kidney; expression was also detected in the small bowel To characterize the binding features of the Sp1a, Sp1b, but not in the stomach and only very weakly in the and Sp1c/AP-2 oligonucleotides, we performed EMSA colon. LacZ was not detected in the spleen, a nonepi- analysis with nuclear extracts from K7-positive (RWP-1) thelial tissue. Similar results were obtained with the 3 and K7-negative (3T3 and SW480) cells. Using a 32P-Sp1b lines generated using the Ϫ1.5 kb promoter and with probe, retarded complexes were demonstrated with the 3 lines harboring the Ϫ0.25 kb promoter (not extracts from the 3 cell types; complexes were inhibited shown). In all cases variable levels of transgene expres- with unlabeled Sp1b, Sp1a, or Sp1c/AP-2 oligonucleo- sion in the small bowel were detectable. We therefore tides but not by unrelated oligonucleotides (Supple- examined endogenous Krt7 and Krt19 expression in the mental Fig. 3), which suggests recognition by the same same tissues using RT-PCR (Fig. 3D). Indeed, Krt7 . Similar results were obtained with a labeled mRNA was expressed at relatively higher level than Sp1a probe. The 3 cell types yielded similar EMSA Krt19 in the pancreas. In contrast, the reverse pattern patterns, indicating that differential binding of nuclear was observed in the colon and small bowel. These proteins to Sp1-like sequences does not account for cell findings indicate that LacZ expression in the intestine type-specific expression. of transgenic mice reflects a low level of expression of Krt7 mRNA in this tissue type. The Krt7 promoter is active in ductal pancreatic cells in the murine pancreas The Krt7 promoter in Ad context has enhanced activity in the pancreas We generated transgenic mice carrying the LacZ gene under the control of the Ϫ1.5 kb or the Ϫ0.25 kb Krt7 To assess the ability of the Krt7 promoter to drive promoter and analyzed transgene expression using transgene expression specifically in pancreatic ductal X-gal staining. Similar results were obtained with both cells, we generated a replication-defective reporter ad- promoter sequences. X-gal activity could be demon- enovirus in which the luciferase gene was controlled by strated in normal pancreatic ducts—but not in acinar the 0.25 fragment of the Krt7 promoter, AdK7Luc. or centroacinar cells—using either construct (Fig. 3A). AdK7Luc and AdCMVGFPLuc (as control) were in- As expected, based on the endogenous expression of jected i.v. into mice, and biodistribution was studied

1370 Vol. 23 May 2009 The FASEB Journal PUJAL ET AL. Liver damage was evaluated by histological examining and by measuring serum transaminase levels at d 5, a time at which a peak of toxicity occurs (23). In mice injected with AdCMVGFPLuc, hepatocyte swelling, lympho- cyte infiltration, and necrosis were observed, as has been reported (18). However, livers from mice re- ceiving the AdK7Luc virus showed only mild lympho- cytic infiltrates (Fig. 5B). In agreement with these observations, transaminase levels were higher in mice receiv- ing control virus than in those receiving AdK7Luc: ([AST (U/L)] AdCMVGFPLuc 523Ϯ112 vs. AdK7Luc 195Ϯ64; [ALT (U/L)] AdCMVGFPLuc 203Ϯ83 vs. AdK7Luc 124Ϯ11). Notably, AST levels in AdK7Luc-injected mice were within the normal range (54-298 U/L; http://www. ahc.umn.edu/rar/refvalues.html). Viral particle determina- tion by qPCR was used to rule out that reduced damage did not result from reduced viral transduction: an equiv- alent number of viral particles were detected in both conditions (AdCMVGFPLuc 20.00Ϯ1.21 vp/100 cells vs. AdK7Luc 19.31Ϯ2.8 vp/100 cells), indicating true re- duced toxicity of AdK7Luc.

Figure 3. Transgene expression in Ϫ1.5 kb Krt7-LacZ trans- Activity of the Krt7 promoter in PDAC xenograft genic mice. Transgenic mice in which the LacZ gene was models expressed under the control of the Ϫ1.5 or Ϫ0.25 kb Krt7 promoter were generated. A, B)Arrows show X-gal activity in ductal pancreatic (A) and biliary duct (B) epithelial cells in To test Krt7 promoter selectivity for cancer cells, tumor pancreas and liver, respectively. There was no staining in and nontumor cell lines were transduced with AdK7Luc or control nontransgenic littermates. Similar results were obtained AdCMVGFPLuc at 104 vp/cell, and the relative strength in 3 independent transgenic lines analyzed using both pro- ϫ ϫ of the Krt7 and CMV promoters was determined. The moter fragments. Original views: 400 (A); 200. C) RT-PCR highest promoter activity was detected in PDAC cells analysis of LacZ and Hprt mRNAs in different tissues from one of the founder lines using the Ϫ1.5 kb promoter. followed by immortalized HPDE cells; very weak activity Transgene expression is detected in the pancreas and in was recorded in IMR-90 fibroblasts (Fig. 6A). These results other epithelia, including low-level expression in the small indicate that the Krt7 promoter retains selectivity for bowel and the colon. There was no detectable LacZ expres- pancreatic ductal cells in the context of an E1A-deleted sion in the stomach or in the spleen. Similar results were adenoviral genome with increased activity in tumoral Ϫ obtained with the 3 transgenic founders of the 1.5 kb cells. reporter. D) RT-PCR analysis of Krt7 and Krt9 mRNAs in the To determine whether Krt7 promoter activity was main- samples analyzed in C. Expression of endogenous Krt7 is detected at variable levels in the epithelial tissues tested; tained in vivo, the levels and duration of luciferase expres- relative Krt7 mRNA levels are higher in the pancreas than in sion were assessed after delivery of AdCMVGFPLuc and intestinal tissues. St, stomach; SmB, small bowel; Lu, lung; Li, AdK7Luc viruses intratumorally into BxPC-3 subcutane- liver; K, kidney; Sg, salivary gland; Co, colon; Sp, spleen; P, ous tumors. Bioluminescent imaging showed a peak of pancreas. luciferase expression in both groups at d 3. In mice receiving control virus, luciferase expression decreased measuring luciferase by noninvasive optical imaging 3 at later time points, becoming significantly lower at d and 5 d later. Major bioluminescence signal was present 11 (Fig. 6B). By contrast, in mice receiving AdK7Luc, in the abdomen; quantification analysis showed a 10-fold luciferase expression persisted up to d 11 (Fig. 6B), reduction in luciferase expression in mice injected with indicating improved transgene expression of AdK7Luc AdK7Luc at both time points (Fig. 4A). At d 5, activity in in tumors. selected nonrelevant tissue extracts (i.e., heart, lung, kid- ney, testis, spleen, and liver) was weaker with the Krt7 promoter than with CMV promoter. Similar activity was DISCUSSION observed in adipose tissue, intestine, and stomach. By contrast, the Krt7 promoter showed a 2.5-fold increased The cell-type specific distribution of keratins has proven activity in pancreas (Fig. 4B), indicating selectivity for this useful for a wide variety of studies, including the targeting organ. This biodistribution pattern prompted us to of transgenes to selected cell types. Here we focus on the study whether enhanced targeting was associated study of the gene coding for K7, a protein with a distinct with enhanced toxicity. There were no major histo- distribution among simple epithelia that is restricted to logical changes in the pancreas of mice injected with ductal pancreatic and biliary epithelium in the hepato- either virus, indicating that increased activity of gastrointestinal system and that is overexpressed in a wide AdK7Luc did not lead to tissue damage (Fig. 5A). variety of cancers.

THE KRT7 PROMOTER TARGETS DUCTAL PANCREATIC CELLS IN VIVO 1371 Figure 4. Biodistribution studies of AdK7Luc (Ϫ0.25 kb promoter) and AdCMVGFPLuc. Here 2 ϫ 1010 vp/mouse of either AdK7Luc (nϭ5), AdCMVGFPLuc (nϭ5), or saline (nϭ5) was injected i.v. into mice.A) Luciferase activity was measured at d 3 and 5. Color-coded images: red indicates highest amount of emitted light (250,000 photons/s); blue indicates lowest (50,000 photons/s). Luciferase activity is expressed as photons per second. *P ϭ 0.04, **P ϭ 0.02; Mann-Whitney nonparametric test. B) Luciferase activity was measured in protein lysates at d 5 with a luminometer. For quantification, background levels from saline-injected mice were subtracted. Results are expressed as relative light units (LU) per milligram protein. *P Ͻ 0.05.

To exploit this restricted pattern distribution, we have Sp1b, Sp1c/AP-2, and Sp1a putative binding se- analyzed the regions of the mouse Krt7 promoter re- quences are crucial for Krt7 promoter activity in RWP-1 quired for cell type-specific activation of reporter genes cells because mutations therein lead to a markedly in vitro and in vivo. As in other genes coding for keratins reduced promoter activity. Sp1 overexpression led to (11, 12, 24), the proximal region of Krt7 promoter increased reporter activity in cells displaying a wide contained sufficient information to reproduce, at least in range of endogenous KRT7 mRNA and Krt7 reporter part, the pattern of gene activity observed in vivo. Se- levels. However, we have found no evidence that Sp1- quences encompassing 0.25 kb and 1.5 kb upstream from like factors are involved in the tissue selectivity of the the transcription initiation site activated reporter expres- Krt7 promoter. A similar diversity in the control of sion in cells derived from K7-expressing tissues, either at activity vs. cellular specificity has been described for high (pancreatic cells) or moderate levels (bladder and KRT5 (24). Sp1 participates in the regulation of expres- breast). By contrast, low or undetectable reporter activity sion of several genes with tissue- and cell type-specific was found in intestinal cells, originating from K7-negative patterns, including those coding for keratins (24–26) epithelia. The proximal 500 bp of the promoter of the and mucins under both constitutive and regulated genes coding for mouse, human, and marsupial K7 are conditions (27). Overexpression and aberrant expres- conserved, and important regulatory elements lie therein sion of keratin mRNAs—including KRT7—have also (16). The fragment encompassing Ϫ234 bp upstream been reported in tumors (28–30), and Sp1 has been from the transcription initiation site contains GC-rich proposed to be involved in these events (31). sequences that constitute putative binding sites for Sp1 Most chronic diseases of the exocrine pancreas affect and/or AP-2 and contains the most important sequence ductal cell function. Chronic pancreatitis is character- information to confer tissue selectivity in vitro. Unfortu- ized by the replacement of the acinar parenchyma by nately our study does not provide clues as to the regula- ductal complexes; the latter are characterized by the tory elements involved in repressing K7 expression in extinction of the acinar differentiation program and intestinal cells. increased cell proliferation (32), as well as increased

Figure 5. Histological analysis of pancreas and liver after systemic AdK7Luc and AdCMVGFPLuc administration. Here 2 ϫ 1010 vp/mouse of either AdK7Luc (nϭ5), AdCMVGFPLuc (nϭ5), or saline (nϭ5) was injected i.v. into mice. At d 5, animals were sacrificed, and pancreas (A) and liver (B) sections were stained by H&E. Arrows indicate lymphocytic infiltrate. Arrowheads indicate necrotic foci. Views: ϫ100; ϫ200.

1372 Vol. 23 May 2009 The FASEB Journal PUJAL ET AL. Figure 6. AdK7Luc activity in pancreatic cells in vitro and in vivo. A) Cells were seeded in 96-well plates and infected with AdK7Luc or AdCMVGFPLuc (104 vp/cell). Luciferase activity was measured 72 h later. Results are represented as percentage of relative promoter activity. Data are expressed as means Ϯ se of 3 independent experiments. B) Tumors were established by s.c. injection of 3 ϫ 106 BxPC-3 PDAC cells into nude mice. When tumors reached 100 mm3 volume, 2 ϫ 1010 vp/tumor of either AdK7Luc (nϭ7) or AdCMVGFPLuc (nϭ7) was injected. Bioluminescent images were taken 3, 7, and 11 d later. Luciferase activity recorded from the tumors was quantified by measuring total amount of emitted light captured by the camera. Results are expressed as photons per second. *P Ͻ 0.05; Mann-Whitney test. risk of progression to cancer (33). The majority of bowel, in comparison to the pancreas. However, it should human pancreatic tumors display a ductal phenotype. be considered that the fraction of the pancreas in which Furthermore, cystic fibrosis results from aberrant cystic the Krt7 promoter is active is a small one, as ductal cells fibrosis transmembrane conductance regulator-depen- represent at most 10% of all pancreatic cells. Of note, dent bicarbonate secretion in pancreatic ducts (34). expression was much lower in the colon than in the small Nevertheless, there is a dearth of knowledge on pan- bowel. Therefore, the quest to identify even more selec- creatic ductal cell biology, and there is a need to tive pancreatic ductal promoters remains important. identify novel pancreatic ductal-restricted promoters. Tissue-specific promoters are good candidates in gene Genetic models of pancreas cancer in mice are mainly therapy to transcriptionally regulate adenovirus armed based on transgenes targeted to acinar cells or to with therapeutic genes. Target cell tropism is a key multipotent precursors (35). The Krt19 promoter is determinant factor for safe and effective treatments. unique by targeting transgenes selectively to ductal cells Our results provide evidence for the selectivity and (11, 12), thus driving expression in the cell type from safety of the Krt7 promoter in the context of an which most ductal adenocarcinomas are thought to adenoviral vector. Notably, after systemic administra- arise (36). As K19, K7 is expressed in normal ductal tion of AdK7Luc the pancreas was the only tissue with cells, whereas it is undetectable in acinar and endo- enhanced luciferase activity, whereas, in all other or- crine cells (8). Targeting activated K-ras with the Krt19 gans, there was a generalized reduced luciferase expres- promoter led to transgene expression in the stomach sion as compared to AdCMVGFPLuc. This was accom- and to gastric cell hyperplasia and preneoplastic lesions panied by lack of signs of tissue damage in the pancreas (13). In agreement with this observation we found that and mild liver inflammation with normal AST levels on Krt19 mRNA is expressed at relatively high levels in AdK7Luc administration. Enhanced pancreatic target- gastrointestinal epithelia. By contrast, Krt7 mRNA is ing with low toxicity is crucial to support the develop- expressed at relatively lower levels in these cell types, ment of Krt7-based therapeutic adenoviruses for dis- therefore being potentially advantageous for selective eases affecting pancreatic ductal cells. Among them, genetic targeting of ductal pancreatic cells. LacZ expres- PDAC is of major interest because there is a dramatic sion pattern from the Ϫ1.5 kb promoter is highly similar lack of effective treatment. We investigated the capacity to that of the expression pattern of endogenous tran- of AdK7Luc adenovirus to target these cells. Several scripts, supporting the notion that the major regulatory studies have shown an up-regulation of KRT7 mRNA sequences involved in Krt7 regulation are contained expression in PDAC (28–30). Infection with AdK7Luc therein. However, we cannot rule out the participation of virus showed higher luciferase activity in all PDAC cells additional upstream sequences. The discrepancy between compared to immortalized HPDE cells, indicating in- the mRNA expression data and the extensive evidence on creased cancer specificity by incorporating the Krt7 low or absent K7 protein in gastrointestinal epithelia in promoter fragment in the adenovirus. The favorable the literature suggests that K7 expression is regulated at targeting with AdK7Luc might be explained, in part, by the posttranscriptional level. the up-regulation and improved access of relevant Using a LacZ reporter we show that Krt7 promoter transcription factors—such as Sp1—to Krt7 sequences activity is restricted to ductal cells in the murine exocrine present in an episomal form rather than in the cellular pancreas. As expected, limited expression of the reporter genome, in which epigenetic modifications might limit was found in other simple epithelial tissues using RT-PCR transcription factor binding (37). for LacZ. The RT-PCR experiments seem to indicate a Regarding potency, the Krt7 promoter was 15–28% considerable expression of the transgene in the small of that of the CMV promoter in PDAC cells. It is well

THE KRT7 PROMOTER TARGETS DUCTAL PANCREATIC CELLS IN VIVO 1373 established that tissue-specific promoters have reduced 4. Tamai, Y., Ishikawa, T., Bosl, M. R., Mori, M., Nozaki, M., activity. In fact, these values are relatively high when Baribault, H., Oshima, R. G., and Taketo, M. M. (2000) Cyto- keratins 8 and 19 in the mouse placental development. J. Cell compared to those of other PDAC-specific promoters Biol. 151, 563–572 such as CCKAR, which has an activity of Ͻ0.3% that of 5. Porter, R. M., and Lane, E. B. (2003) Phenotypes, genotypes and CMV (38). Notably, Krt7 promoter activity in PDAC their contribution to understanding keratin function. Trends Genet. 19, 278–285 xenografts was sustained while CMV promoter activity 6. Smith, F. (2003) The molecular genetics of keratin disorders. decreased very rapidly after intratumoral injections. Am. J. Clin. Dermatol. 4, 347–364 The different behavior of these promoters could be 7. Owens, D. W., and Lane, E. B. (2003) The quest for the function explained by at least 2 independent mechanisms. First, of simple epithelial keratins. Bioessays 25, 748–758 8. Schussler, M. H., Skoudy, A., Ramaekers, F., and Real, F. X. the CMV promoter—while robust—is susceptible to (1992) Intermediate filaments as differentiation markers of epigenetic inactivation (39), while these effects may be normal pancreas and pancreas cancer. Am. J. Pathol. 140, less relevant when using endogenous promoters. Sec- 559–568 ond, we can speculate that the elevated intratumoral 9. Osborn, M., van Lessen, G., Weber, K., Kloppel, G., and Altmannsberger, M. (1986) Differential diagnosis of gastrointes- Krt7 activity can also result from the presence of tinal carcinomas by using monoclonal antibodies specific for microenvironmental factors (i.e., stroma) positively in- individual keratin polypeptides. Lab. Invest. 55, 497–504 fluencing Krt7 promoter activity in vivo. 10. Vila, M. R., Lloreta, J., Schussler, M. H., Berrozpe, G., Welt, S., and Real, F. X. (1995) New pancreas cancers cell lines that Altogether, our findings support the notion that the represent distinct stages of ductal differentiation. Lab. Invest. 72, Krt7 promoter should be useful in further investigating 395–404 pancreatic ductal cell biology and might contribute to 11. Brembeck, F. H., and Rustgi, A. K. (2000) The tissue-dependent developing improved therapeutic strategies for pancre- gene transcription is regulated by GKLF/KLF4 and Sp1. J. Biol. Chem. 275, 28230–28239 atic diseases affecting ductal cells. The genetic target- 12. Brembeck, F. H., Moffett, J., Wang, T. C., and Rustgi, A. K. ing of ductal cells through selective delivery or selective (2001) The keratin 19 promoter is potent for cell-specific activation of cellular mechanisms may provide strate- targeting of genes in transgenic mice. Gastroenterology 120, gies to improve outcomes in cystic fibrosis, chronic 1720–1728 13. Brembeck, F. H., Schreiber, F. S., Deramaudt, T. B., Craig, L., pancreatitis, and pancreatic cancer. Rhoades, B., Swain, G., Grippo, P., Stoffers, D. A., Silberg, D. G., and Rustgi, A. K. (2003) The mutant K-ras oncogene causes We thank all investigators mentioned in the main text and pancreatic periductal lymphocytic infiltration and gastric mu- Supplemental Information for providing cells and reagents; cous neck cell hyperplasia in transgenic mice. Cancer Res. 63, members of the Unitat de Biologia Cel.lular i Molecular, IMIM, 2005–2009 for valuable contributions; Dr. C. Ciudad (Universitat de Barce- 14. Huch, M., Abate-Daga, D., Roig, J. M., Gonzalez, J. R., Fabregat, J., Sosnowski, B., Mazo, A., and Fillat, C. (2006) Targeting the lona, Barcelona, Spain) for valuable discussions and reagents; CYP2B 1/cyclophosphamide suicide system to fibroblast growth and Drs. M. I. Herna´ndez-Mun˜oz, A. Skoudy, X. Mayol, and A. factor receptors results in a potent antitumoral response in Merlos for critical reading of the manuscript. We also thank the pancreatic cancer models. Hum. Gene. Ther. 17, 1187–1200 Clinical Biochemistry Service of the Veterinary Faculty of the 15. Liu, N., Furukawa, T., Kobari, M., and Tsao, M. S. (1998) Autonomous University of Barcelona for the determination of Comparative phenotypic studies of duct epithelial cell lines serum transaminases. This work was supported, in part, by grants derived from normal human pancreas and pancreatic carci- from Plan Nacional de IϩD (SAF2001-0420, SAF2004-01137, noma. Am. J. Pathol. 153, 263–269 and SAF2007-60860), Accio´n Especial de Geno´mica (GEN2001- 16. Smith, F. J., Porter, R. M., Corden, L. D., Lunny, D. P., Lane, 4748-c05-01), CIRIT (Generalitat de Catalunya) (SGR-00410), E. B., and McLean, W. H. (2002) Cloning of human, murine, and marsupial keratin 7 and a survey of K7 expression in the Marato´ de TV-3, and Biomed Program (QLG-CT-2002-01196) to mouse. Biochem. Biophys. Res. Commun. 297, 818–827 F.X.R., and grants BIO2005-08682-C03-02 from the Spanish 17. Schreiber, E., Matthias, P., Muller, M. M., and Schaffner, W. Ministry of Education and Science and SGR0500008 from the (1989) Rapid detection of octamer binding proteins with ‘mini- Generalitat de Catalunya to C.F. The group of C.F. is also extracts,’ prepared from a small number of cells. Nucleic Acids partially funded by CIBERER, Instituto de Salud Carlos Res. 17, 6419 III. W.H.I.M. and F.J.D.S. were supported by a Wellcome Trust 18. Alemany, R., and Curiel, D. T. (2001) CAR-binding ablation Senior Research Fellowship in Basic Biomedical Science (to does not change biodistribution and toxicity of adenoviral W.H.I.M.). M.H. was a predoctoral fellow (BEFI) at the Instituto vectors. Gene Ther. 8, 1347–1353 de Salud Carlos III. A.J. was supported by a predoctoral fellow- 19. He, T. C., Zhou, S., da Costa, L. T., Yu, J., Kinzler, K. W., and Vogelstein, B. (1998) A simplified system for generating ship (FPU) granted by the Spanish Ministry of Education and recombinant adenoviruses. Proc. Natl. Acad. Sci. U. S. A. 95, Science. I.A. was supported by a postdoctoral fellowship from 2509–2514 the Basque government. The authors declare no conflict of 20. Becker, T. C., Noel, R. J., Coats, W. S., Gomez-Foix, A. M., Alam, interest. T., Gerard, R. D., and Newgard, C. B. (1994) Use of recombi- nant adenovirus for metabolic engineering of mammalian cells. Methods Cell Biol. 43(Pt. A), 161–189 21. Cascante, A., Abate-Daga, D., Garcia-Rodriguez, L., Gonzalez, REFERENCES J. R., Alemany, R., and Fillat, C. (2007) GCV modulates the antitumoural efficacy of a replicative adenovirus expressing the 1. Moll, R., Franke, W. W., Schiller, D. L., Geiger, B., and Krepler, Tat8-TK as a late gene in a pancreatic tumour model. Gene Ther. R. (1982) The catalog of human cytokeratins: patterns of 14, 1471–1480 expression in normal epithelia, tumors and cultured cells. Cell 22. Adell, T., Gomez-Cuadrado, A., Skoudy, A., Pettengill, O. S., 31, 11–24 Longnecker, D. S., and Real, F. X. (2000) Role of the basic 2. Strelkov, S. V., Herrmann, H., and Aebi, U. (2003) Molecular helix-loop-helix transcription factor p48 in the differentiation architecture of intermediate filaments. Bioessays 25, 243–251 phenotype of exocrine pancreas cancer cells. Cell Growth Differ. 3. Ku, N. O., Michie, S. A., Soetikno, R. M., Resurreccion, E. Z., 11, 137–147 Broome, R. L., Oshima, R. G., and Omary, M. B. (1996) Suscepti- 23. Christ, M., Louis, B., Stoeckel, F., Dieterle, A., Grave, L., Dreyer, bility to hepatotoxicity in transgenic mice that express a dominant- D., Kintz, J., Ali Hadji, D., Lusky, M., and Mehtali, M. (2000) negative human mutant. J. Clin. Invest. 98, 1034–1046 Modulation of the inflammatory properties and hepatotoxicity

1374 Vol. 23 May 2009 The FASEB Journal PUJAL ET AL. of recombinant adenovirus vectors by the viral E4 gene prod- site and the minimal promoter contribute to overexpression of ucts. Hum. Gene. Ther. 11, 415–427 the 18 gene in tumorigenic clones relative to that in 24. Kaufman, C. K., Sinha, S., Bolotin, D., Fan, J., and Fuchs, E. nontumorigenic clones of a human carcinoma cell line. Mol. (2002) Dissection of a complex enhancer element: mainte- Cell. Biol. 15, 2490–2499 nance of keratinocyte specificity but loss of differentiation 32. Rodolosse, A., Chalaux, E., Adell, T., Hagege, H., Skoudy, A., specificity. Mol. Cell. Biol. 22, 4293–4308 and Real, F. X. (2004) PTF1alpha/p48 transcription factor 25. Chen, T. T., Wu, R. L., Castro-Munozledo, F., and Sun, T. T. couples proliferation and differentiation in the exocrine pan- (1997) Regulation of K3 keratin gene transcription by Sp1 and creas (corrected). Gastroenterology 127, 937–949 AP-2 in differentiating rabbit corneal epithelial cells. Mol. Cell. 33. Herna´ndez-Mun˜oz, I., Skoudy, A., Real, F. X., and Navarro, P. Biol. 17, 3056–3064 (2008) Pancreatic ductal adenocarcinoma: cellular origin, sig- 26. Opitz, O. G., and Rustgi, A. K. (2000) Interaction between Sp1 nalling pathways and stroma contribution. Pancreatology 8, 462– and cell cycle regulatory proteins is important in transactivation 469 of a differentiation-related gene. Cancer Res. 60, 2825–2830 34. Steward, M. C., Ishiguro, H., and Case, R. M. (2005) Mecha- 27. Van Seuningen, I., Pigny, P., Perrais, M., Porchet, N., and nisms of bicarbonate secretion in the pancreatic duct. Annu. Aubert, J. P. (2001) Transcriptional regulation of the 11p15 Rev. Physiol. 67, 377–409 mucin genes. Towards new biological tools in human therapy, in 35. Leach, S. D. (2004) Mouse models of pancreatic cancer: the fur inflammatory diseases and cancer? Front. Biosci. 6, D1216– is finally flying!. Cancer Cell. 5, 7–11 D1234 36. Real, F. X. (2003) A “catastrophic hypothesis” for pancreas 28. Goldstein, N. S., and Bassi, D. (2001) Cytokeratins 7, 17, and 20 cancer progression. Gastroenterology 124, 1958–1964 reactivity in pancreatic and ampulla of vater adenocarcinomas. 37. Yuan, P., Wang, L., Wei, D., Zhang, J., Jia, Z., Li, Q., Le, X., Percentage of positivity and distribution is affected by the Wang, H., Yao, J., and Xie, K. (2007) Therapeutic inhibition of cut-point threshold. Am. J. Clin. Pathol. 115, 695–702 Sp1 expression in growing tumors by mithramycin a correlates 29. Iacobuzio-Donahue, C. A., Maitra, A., Shen-Ong, G. L., van directly with potent antiangiogenic effects on human pancreatic Heek, T., Ashfaq, R., Meyer, R., Walter, K., Berg, K., Holling- cancer. Cancer 110, 2682–2690 sworth, M. A., Cameron, J. L., Yeo, C. J., Kern, S. E., Goggins, M., 38. Xie, X., Xia, W., Li, Z., Kuo, H. P., Liu, Y., Ding, Q., Zhang, S., and Hruban, R. H. (2002) Discovery of novel tumor markers of Spohn, B., Yang, Y., Wei, Y., Lang, J. Y., Evans, D. B., Chiao, P. J., pancreatic cancer using global gene expression technology. Abbruzzese, J. L., and Hung, M. C. (2007) Targeted expression Am. J. Pathol. 160, 1239–1249 of BikDD eradicates pancreatic tumors in noninvasive imaging 30. Iacobuzio-Donahue, C. A., Ashfaq, R., Maitra, A., Adsay, N. V., models. Cancer Cell. 12, 52–65 Shen-Ong, G. L., Berg, K., Hollingsworth, M. A., Cameron, J. L., 39. Krishnan, M., Park, J. M., Cao, F., Wang, D., Paulmurugan, R., Yeo, C. J., Kern, S. E., Goggins, M., and Hruban, R. H. (2003) Tseng, J. R., Gonzalgo, M. L., Gambhir, S. S., and Wu, J. C. Highly expressed genes in pancreatic ductal adenocarcinomas: (2006) Effects of epigenetic modulation on reporter gene a comprehensive characterization and comparison of the tran- expression: implications for stem cell imaging. FASEB J. 20, scription profiles obtained from three major technologies. 106–108 Cancer Res. 63, 8614–8622 31. Gunther, M., Frebourg, T., Laithier, M., Fossar, N., Bouziane- Received for publication July 10, 2008. Ouartini, M., Lavialle, C., and Brison, O. (1995) An Sp1 binding Accepted for publication December 4, 2008.

THE KRT7 PROMOTER TARGETS DUCTAL PANCREATIC CELLS IN VIVO 1375