Use of the Ornithine Decarboxylase Promoter to Achieve N-MYC-Mediated Overexpression of a Rabbit Carboxylesterase to Sensitize Neuroblastoma Cells to CPT-11 Cynthia A

doi:10.1006/mthe.2000.0064, available online at http://www.idealibrary.com on IDEAL ARTICLE

Use of the Ornithine Decarboxylase Promoter to Achieve N-MYC-Mediated Overexpression of a Rabbit Carboxylesterase to Sensitize Neuroblastoma Cells to CPT-11 Cynthia A. Pawlik, Rekha V. Iyengar, Erik J. Krull, Stephen E. Mason, Ruchi Khanna, Linda C. Harris, Philip M. Potter, Mary K. Danks,1 and Sylvie M. Guichard2

Department of Molecular Pharmacology, St. Jude Children’s Research Hospital, Memphis, Tennessee 38105

Received for publication February 7, 2000, and accepted in revised form March 30, 2000

Overexpression of specific transcription factors by tumor cells can be exploited to regulate expression of proteins that induce apoptosis or activate prodrugs, thereby producing tumor- selective toxicity. A majority of advanced-stage neuroblastomas overexpress the transcription factor N-MYC, and this overexpression is associated with poor prognosis. This study describes regulation of expression by N-MYC, via the ornithine decarboxylase (ODC) promoter, of a rabbit liver carboxylesterase (CE) that activates the prodrug CPT-11. Chloramphenicol acetyltransferase reporter assays and CE activity assays in transiently transfected neuroblastoma cell lines (SJNB-1, SJNB-4, NB-1691) and rhabdomyosarcoma cell lines (JR1neo20, JR1Nmyc6, JR1Nmyc9) support this approach as a potential method for sensitizing tumor cells to CPT-11. Clonogenic assays with IMR32 human neuroblastoma cells which express N-MYC and that had been stably transfected with a plasmid containing an ODC promoter/CE cassette corroborated results of enzyme activity assays. Specifically, IMR32.ODC.CE cells expressed approximately eightfold more CE activity than IMR32.CMV.neo cells; and 5 µM CPT-11 reduced the clonogenic potential of IMR32.ODC.CE cells to zero, while 50 µM CPT-11 was required to produce the same effect with IMR32.CMV.neo cells. Current experiments focus on adenoviral delivery of an ODC promoter/CE cDNA cassette for potential virus-directed enzyme prodrug therapy applications.

Key Words: irinotecan; gene therapy; prodrug; VDEPT; esterase; ornithine decarboxylase promoter; N-MYC.

INTRODUCTION tive therapy. Since viruses transduce both normal and tumor cells, selective expression of CE might be achieved We recently reported the isolation of a cDNA encoding a by a transcription factor/promoter combination unique rabbit liver carboxylesterase (CE) that efficiently con- to tumor cells. Protooncogene/transcription factor N- verts the prodrug CPT-11 to SN-38 and sensitizes human MYC and the ornithine decarboxylase (ODC) promoter tumor cells to CPT-11 both in vitro and in a preclinical represent such a combination. xenograft model (1–3). The level of expression of rabbit N-MYC is a transcription factor overexpressed in CE produces a parallel increase in sensitivity to CPT-11 ~40% of Grade III/IV neuroblastomas, and this overex- (1, 2). Our long-range goal is to use the rabbit CE in com- pression correlates with rapid tumor progression and bination with CPT-11 in a virus-directed enzyme pro- poor prognosis (4, 5). MYC family proteins (c-, N-, L-) drug therapy (VDEPT) approach to achieve tumor-selec- dimerize with MAX and bind to the consensus CACGTG E-box sequence (6–8), thereby activating several promoters including those that control expression of ODC, α- prothymosin, and p53 (9–12). Use of the ODC promoter 1To whom correspondence should be addressed at Department of Molecular Pharmacology, St. Jude Children’s Research Hospital, 332 to achieve tumor-specific expression of CE and activa- North Lauderdale, Memphis, TN 38105. Fax: (901) 521-1668. E-mail: tion of CPT-11 in neuroblastoma cells that overexpress [email protected]. N-MYC is the focus of this study. No studies of this type 2Present address: Institut Claudius Regaud, Laboratoire de Pharmacologie, 20-24, rue du pont St Pierre 31052, Toulouse cedex, have been done with the ODC promoter, but several France. studies have been published that investigate the effect of

MOLECULAR THERAPY Vol. 1, No. 5, May 2000, Part 1 of 2 Parts 457 Copyright The American Society of Gene Therapy ARTICLE

MYC-responsive E-box enhancer sequences (CACGTG) in combination with Simian virus-40 (SV40) or Herpes simplex virus thymidine kinase (HSVtk) promoters to upregulate expression of various prodrug-activating enzymes (13, 14). In those studies some degree of specificity was achieved, but promoter/enhancer activity was dependent on the interaction of different transcription factors with the promoter and the enhancer sequences. The purpose of this study was to demonstrate specific ODC promoter-regulated expression of the CAT reporter gene and rabbit CE in several human tumor cell lines that overexpress N-MYC and to show that increased sensitivity to CPT-11 results from ODC promoter-regulated CE expression.

MATERIALS AND METHODS

Drugs and chemicals. CPT-11 was a generous gift from Dr. J. P. McGovren p∆ODC.CAT were repeated with pCAT3-Basic and pC3B.ODC.CAT. Both (Pharmacia–Upjohn Co., Kalmazoo, MI). A 10 mM drug stock was pre- sets of plasmids gave the same ratio after normalizing to the appropriate − pared in methanol and stored at 20°C. Drug was further diluted in water negative control. Importantly, CAT activities (raw data) for all cell lines immediately before use. transfected with pCAT3-Basic and for MYC-negative SJ-G3 cells transfect- Cell lines. The neuroblastoma cell lines SJNB-1 and SJNB-4 were ed with pC3B.ODC.CAT were in the 500 dpm range, equivalent to mock derived from patients at St. Jude Children’s Research Hospital (SJCRH) transfectants. (Memphis, TN). Tumor samples were obtained after informed consent Plasmids for CE activity. The ODC promoter and rabbit liver CE were was given, in accordance with the guidelines of the St. Jude Children’s ligated into the pCIneo plasmid (Promega) as follows: pCIneo was digest- Research Hospital Institutional Review Board (IRB). The embryonal rhab- ed with BglII and I-PpoI to remove the CMV promoter, leaving the domyosarcoma cell line JR1 was provided by Dr. P. Dias (Imgenex, San chimeric intron intact. Blunt ends were created at the BglII and I-PpoI Diego, CA). NB-1691 cells were obtained from the Pediatric Oncology sites. The ODC promoter was released from pODC.CAT by digestion with Group. These cell lines were grown in RPMI 1640 (BioWhittaker, SmaI and inserted into pCIneo at this site by blunt-end ligation. CE was Walkersville, MD) supplemented with 10% fetal calf serum (Hyclone, released from pIRESrabbit (1) by digestion with EcoRI and ligated into this Logan, UT) and 2 mM glutamine. The neuroblastoma cell line IMR32 was site of the plasmid to generate pCI.ODC.CE (Table 1). acquired from American Type Tissue Culture Collection (Rockville, MD) and grown in DMEM (BioWhittaker), 10% fetal calf serum, and 2 mM Establishment of stably transfected cell lines. JR1 cells were transfected glutamine. Multiple cell lines were used in this study to ensure that N- with pIRES.CMV.N-myc plasmid DNA by the calcium phosphate method MYC/ODC promoter-regulated protein expression was not unique to a (Stratagene, La Jolla, CA). Stable clones expressing different levels of N- single cell line. Pediatric glioblastoma cell lines SJ-G2 and SJ-G3 were MYC were selected in 200 µg/ml of geneticin (Gibco BRL, Gaithersburg, derived from tumors of patients at SJCRH. Tumor samples were obtained MD). Two clones expressing different levels of N-myc RNA were expand- in accordance with the guidelines of the SJCRH IRB. The cell lines were ed and used in this study: JR1Nmyc6 and JR1Nmyc9. The control maintained in DMEM, 15% fetal calf serum, and 2 mM glutamine. SJ-G2 JR1neo20 cell line was generated by transfecting JR1 cells with the parent cells express c-MYC but not N-MYC; SJ-G3 cells express neither c-MYC pIRESneo plasmid. In addition, IMR32 neuroblastoma cells were trans- nor N-MYC and were used as controls for specific experiments, as indi- fected by electroporation with either pIRESneo or pIRES.ODC.CE using cated in the text. conditions previously published (2), and stable transfectants were selected in 500 µg of geneticin/ml. IMR32.CMV.neo and IMR32.ODC.CE cell Plasmids for CAT reporter activity. The N-myc cDNA was ligated into lines were not cloned following transfection and geneticin selection, but the bicistronic pIRESneo plasmid (Clontech, Palo Alto, CA) in the fol- include all cells that express sufficient neomycin resistance protein to sur- lowing manner: The N-myc cDNA was excised from pCN64RX and insert- vive the selecting concentration of geneticin. Results with these IMR32 ed into the EcoRV and EcoRI sites of pIRESneo. This plasmid is referred to cell lines, therefore, represent the entire transfected population and are as pIRES.CMV.N-myc. For clarity, names of plasmids in this paper contain not the properties of a single clone. the name of the promoter (ODC, CMV) and the gene or cDNA (CAT, CE, N-myc, neo) in each construct (Table 1). Western analysis. For each sample, one flask of cells in log-phase The pODC∆CAT and pODC∆CAT∆S plasmids were provided by Drs. growth was washed twice with ice-cold PBS, and the cells were harvested J. L. Cleveland (St. Jude Children’s Research Hospital) and P. Coffino by scraping into 0.6 ml of ice-cold PBS–ST (1% SDS, 1% Triton X-100) (University of California, San Francisco, CA) and have been described in containing 100 µg/ml of each of the following protease inhibitors: apro- detail previously (9). These constructs are referred to in the text as tinin, leupeptin, and antipain (Sigma, St. Louis, MO). Loading buffer (0.2 pODC.CAT and p∆ODC.CAT, respectively. pODC.CAT contains the ml of 4) was added to 600 µl of the cell solution, and the cell lysate was murine ODC promoter which regulates transcription of the reporter gene passed through a 23-gauge needle 10 times to shear the DNA. The lysates chloramphenicol acetyltransferase (CAT). pODC.CAT contains two MYC- were then boiled for 5 min and centrifuged briefly. Because of the insta- responsive E-box sequences (CACGTG) within the ODC promoter. bility of the N-MYC protein, all manipulations were done in less than 10 p∆ODC.CAT contains the ODC promoter but with a point mutation in min. Proteins were separated using SDS–PAGE and transferred by semidry each of the MYC binding sites (CACCTG). The p∆ODC.CAT plasmid was blotting (Millipore, Bedford, MA) to Immobilon-P polyvinylidene difluo- used as a negative control. ride membranes (Millipore). N-MYC was detected with the C-19 N-MYC Because some “leakage” of CAT activity was seen with MYC-negative antibody (Santa Cruz Biotechnology, Santa Cruz, CA) using the method control cell lines transfected with p∆ODC.CAT, a second set of plasmids reported previously (15). The images were scanned with a was constructed with a commercially available vector. The ODC promot- Hewlett–Packard ScanJet scanner and the pixel density and area of each er was removed from pODC.CAT and ligated into the pCAT3-Basic vector band used to quantitate relative protein levels using Imagequant software (Promega, Madison, WI) in the XmaI site of the multiple cloning region (Molecular Dynamics, Sunnyvale, CA). Blots were stripped and reprobed (pC3B.ODC.CAT), and experiments done with pODC.CAT and with anti-α-tubulin (ICN, Costa Mesa, CA) as a loading control.

458 MOLECULAR THERAPY Vol. 1, No. 5, May 2000, Part 1 of 2 Parts Copyright The American Society of Gene Therapy ARTICLE

Southern and Northern analyses. Genomic DNA was isolated using standard methods (16) and digested with EcoRI prior to electrophoresis in an agarose gel. Total RNA was prepared using an Ambion RNA kit (Austin, TX). Southern and Northern blottings were carried out by standard methods (16). A probe for glyceraldehyde 3-phosphate (G3PDH; Clontech, Palo Alto, CA) was used as control for equal sample loading for both DNA and RNA blots.

CAT assays. To investigate ODC promoter-regulated reporter gene expression, two different pairs of plasmids were used for transfection. The first set was pODC.CAT and p∆ODC.CAT. The latter plasmid contains a mutated MYC binding sequence and was used as a negative control. The second pair of plasmids was pCAT3-Basic and pC3B.ODC.CAT, each of which contains the CAT gene, but the Basic vector contains no promoter while pC3B.ODC.CAT contains the ODC promoter (Table 1). Cells (107) were electroporated with 10 µg of DNA of one of the above FIG. 1. Western (left), Northern (center), and Southern (right) analyses for plasmids and 15 µg of pCMVβ vector (Clontech). CAT activity in heat- N-MYC in SJNB-1, SJNB-4, and NB-1691 neuroblastoma cell lines. (Left) Cell treated (65°C for 10 min) cell extracts was determined 72 h after trans- lysates were separated by SDS–PAGE and immunoblotting was done for N- fection using the Quant-T-CAT kit (Amersham, Arlington Heights, IL). MYC protein. Membranes were stripped and reprobed with anti-α-tubulin as CAT activity was expressed as dpm of 3H-labeled acetylated chloram- a loading control. After normalizing to the tubulin signal, relative amounts of phenicol produced/mg protein/h. CAT activity was normalized to β- N-MYC in the SJNB-1, SJNB-4, and NB-1691 cells were quantitated to be 1.3, galactosidase (β-gal) activity (β-Galactosidase Enzyme Assay System; 2.5, and 1.0, respectively. (Center) Total cellular RNA was extracted, separat- Promega) to correct for transfection efficiency. β-Gal activity was ed by electrophoresis, and probed with a full-length cDNA for N-MYC.

expressed as A420 of nitrophenol produced/mg protein/h. Mock transfec- G3PDH was used as a loading control (not shown). After levels of N-MYC RNA tions served as additional negative controls. Data points were obtained were corrected for loading, relative amounts of N-myc RNA in SJNB-1, SJNB- in the linear range of each enzymatic assay. 4, and NB-1691 cells were determined to be 1, 123, and 36, respectively. (Right) Genomic DNA was extracted, digested with EcoRI, and separated by Carboxylesterase activity. Whole-cell sonicates were prepared and CE electrophoresis. The membrane was probed with a full-length N-myc cDNA activity was determined with a spectrophotometric assay using o-nitro- and relative gene copy number determined by assigning the signal for SJNB- phenylacetate as substrate (1). CE activity was expressed in µmol o- 1 cells a value of 1. SJNB-4 and NB-1691 cells were found to contain 16 nitrophenol produced/min/mg protein. Protein concentrations were copies of the N-myc gene. All of the above methods are detailed under determined using the Bio-Rad protein assay reagent (Hercules, CA) with Materials and Methods. bovine serum albumin as standard.

Clonogenic assays. Cells in log-phase growth were plated in 35-mm tissue culture dishes and allowed to adhere overnight. Cells were then exposed to various concentrations of CPT-11 for 2 h, when the medium levels of N-MYC protein from SJNB-1, SJNB-4, and NB-1691 was replaced with drug-free medium, and cells were allowed to grow for cell lines did not correlate with RNA levels. For example, SJNB- a time equivalent to five doublings of the untreated cells. 1 cells had the lowest N-MYC RNA level (Fig. 1, center), but Colonies were stained with crystal violet and quantitated using an the N-MYC protein level of this cell line was intermediate AlphaImager 2000 (Alpha Innotech Corp., San Leandro, CA) colony counter with minimum area settings to exclude clusters of fewer than between the other two lines. Further, Southern analysis indi-

eight cells. Results are reported as IC99 which is defined as the concen- cated that SJNB-4 and NB-1691 cells had equal copy numbers tration of CPT-11 at which no colonies were detected either by automat- of the N-myc gene (~16 copies), but NB-1691 cells had the low- ed colony counter or microscopically. Data were graphed using est N-MYC protein level of the three cell lines (Fig. 1, right). It GraphPad Prism software (GraphPad Software Inc., San Diego, CA). is not known which of the above parameters, if any, correlates with MYC function. RESULTS Western blot analysis of an additional pair of IMR32 neuroblastoma cell lines that had been transfected with pIRESneo N-MYC Expression in Neuroblastoma Cell Lines or pIRES.ODC.CE showed that each of the transfectants expressed an equal level of N-MYC protein (Fig. 2A). A nega- We first analyzed a panel of neuroblastoma cell lines to deter- tive control cell line, the SJ-G3 pediatric glioblastoma cell line, mine relative levels of N-MYC protein. The predicted 57- and had no detectable N-MYC protein under the conditions of the 54-kDa N-MYC doublet was detected in SJNB-1, SJNB-4, and assay (data not shown). NB-1691 cell lines, and the relative levels of N-MYC were quantitated (Fig. 1, left). Levels of N-MYC were normalized to N-MYC Western Analysis of JR1 Rhabdomyosarcoma tubulin controls, and the amount of N-MYC expressed in NB- Cell Lines 1691 cells was set equal to 1. SJNB-1 and SJNB-4 cells expressed 1.3- and 2.5-fold more N-MYC, respectively, than Since many cellular factors might influence N-MYC func- NB-1691 cells. In addition to the 57- and 54-kDa N-MYC pro- tion, we also included in this study a series of isogenic rhab- teins, two smaller immunoreactive bands were detected. domyosarcoma JR1 cell lines in which the only difference These bands are likely the ∆MYC proteins described by Hann between the cell lines was the level of expression of N-MYC. (17). Similar to several other neuroblastoma cell lines, untransfect- Since N-MYC is a labile protein, it was also of interest to ed (not shown) and control JR1 cells transfected with determine whether N-MYC protein levels correlated with lev- pIRESneo plasmid express endogenous N-MYC protein (Fig. els of mRNA transcript; therefore, we performed Northern 2B). Two clonal JR1 cell lines that had been transfected with analysis on the above neuroblastoma cell lines. The relative pIRES.CMV.N-myc were also analyzed for N-MYC levels by

MOLECULAR THERAPY Vol. 1, No. 5, May 2000, Part 1 of 2 Parts 459 Copyright The American Society of Gene Therapy ARTICLE

FIG. 2. (A) Immunoblot for N-MYC in IMR32pIRES.CMV.neo and IMR32pIRES.ODC.CE cell lines. (B) Immunoblot and Northern blot for N-MYC in JR1neo20, JR1Nmyc9, and JR1Nmyc6 cell lines. Tubulin was used as a loading control. See the legend to Fig. 1 and Materials and Methods for details.

immunoblot. JR1Nmyc9 cells expressed the same level of N- JR1neo20 cells with pCAT3-Basic or pC3B.ODC.CAT. CAT MYC protein as JR1neo20 control cells. JR1Nmyc6 cells activity in JR1neo20 cells was ~2-fold higher in cells transfect- expressed readily detectable levels of the 57-kDa form of N- ed with the plasmid containing the ODC promoter MYC as well as the 54-kDa form seen in JR1neo20 and (pC3B.ODC.CAT) compared to no promoter (pCAT3-Basic). JR1Nmyc9 cells. Similar to comparisons of levels of expression Activation of the ODC promoter in JR1neo20 cells is attrib- of N-MYC protein and RNA in neuroblastoma cells (Fig. 1), uted to the basal level of endogenous MYC protein expressed the level of N-MYC protein could not be predicted by level of by these cells (Fig. 2A). The level of CAT activity obtained in RNA in these stably transfected JR1 rhabdomyosarcoma cell JR1neo20 cells transfected with pC3B.ODC.CAT was set equal lines (Fig. 2B). to 1. JR1Nmyc9 cells (N-MYC level equal to JR1neo20 control cells) had the same level of CAT activity (0.96 ± 0.91) as N-MYC/ODC Promoter-Regulated CAT Activity in JR1neo 20 cells (Fig. 4). In contrast, JR1Nmyc6 cells (elevated SJNB-1, SJNB-4, and NB-1691 Neuroblastoma Cell Lines To assess the ability of N-MYC to activate the ODC promoter, SJNB-1, SJNB-4, and NB-1691 neuroblastoma cell lines were cotransfected with either pODC.CAT or pODC.CAT and with pCMV to control for transfection efficiency (Fig. 3). The ratio of CAT activity/-gal activity after transfection with pODC.CAT for each cell line was set equal to 1, and a ratio of CAT activity obtained with pODC.CAT/pODC.CAT constructs was calculated. The cell line with the highest amount of N-MYC protein, SJNB-4, had the highest relative CAT activity, 6.7 ± 0.2-fold over pODC.CAT control transfections. Cell lines with similar levels of N-MYC protein, SJNB-1 and NB- 1691, had similar CAT activities of 2.0 ± 0.7- and 1.7 ± 0.2-fold greater than control, respectively. SJ-G3 cells which express no N-MYC protein showed no increase in CAT activity following transfection with plasmids in which CAT expression was regulated by the ODC promoter (not shown). Similar results were FIG. 3. Relative CAT reporter activity in SJNB-1, SJNB-4, and NB-1691 cell obtained with pCAT3-Basic and pC3B.ODC.CAT plasmids lines. Cell lines were cotransfected with either pODC.CAT or p∆ODC.CAT and pCMVβ plasmid DNA. Following correction for transfection efficiency, CAT (data not shown). In agreement with published studies (9), activity in cells transfected with p∆ODC.CAT was arbitrarily set equal to 1.0, these data suggest that N-MYC activates transcription and results obtained with pODC.CAT were compared to this control. Each bar through the ODC promoter and also that the level of N-MYC represents the average CAT activity (±SD) from four to six independent trans- protein correlates with the level of reporter activity. fections obtained at two different time points during the linear time course of the CAT assay. The raw data (number of dpm of 3H-labeled acetylated chloramphenicol produced/mg protein/h) ranged from ~1000 to 200,000 for cells ODC Promoter-Mediated CAT Activity in JR1 Cell transfected with pODC.CAT. Transfections with a second set of plasmids Lines with Exogenous Expression of N-MYC cDNA (pCAT3-Basic and pC3B.ODC.CAT) gave equivalent results (not shown). Additional control transfections of N-MYC-negative SJ-G3 cells with To show more directly the relationship between the levels pODC.CAT or pC3B.ODC.CAT produced levels of acetylated chloramphenicol of N-MYC protein and CAT activity, we first transfected equal to no DNA mock-transfected cells (not shown).

460 MOLECULAR THERAPY Vol. 1, No. 5, May 2000, Part 1 of 2 Parts Copyright The American Society of Gene Therapy ARTICLE

ODC Promoter-Mediated CE Activity Following Transfections of JR1neo20, JR1Nmyc6, JR1Nmyc9, and IMR32 Cell Lines Although reporter assays are a simple, reliable method to compare transcription factor/promoter interactions in vitro, our long-range goal is to upregulate expression of CE in situ. Therefore, JR1 clones were transfected transiently with either pCIneo or pCI.ODC.CE and CE activity measured 72 h following transfection. Figure 5 shows that CE activity was detected in each cell line transfected with pCI.ODC.CE, but only background levels of CE activity (12 µmol/mg/min) were seen in untransfected or pCIneo-transfected JR1neo20, JR1Nmyc9, and JR1Nmyc6 cells. In these experiments, buffer requirements of CE and β-gal prohibited normalization of FIG. 4. CAT activity in JR1 isogenic cell lines that express different levels of results for transfection efficiency; therefore, the experiments N-MYC. Untransfected control cells or cells transiently transfected with a were done five times. Similar to in vitro reporter activity assays, β pCAT3-Basic or pC3B.ODC.CAT and pCMV were analyzed for CAT activity. CE activity generated in situ paralleled the level of N-MYC pro- The background range of dpm values of acetylated chloramphenicol generated was routinely ~500, similar to values seen with mock-transfected con- tein in these isogenic cell lines. trols. See the legend to Fig. 3 and Materials and Methods for details. Data We next confirmed the finding of in situ production and shown are representative of a minimum of three experiments. activity of ODC-mediated CE activity in a stably transfected neuroblastoma cell line. Transfection of IMR32 with pIRESneo or pIRES.ODC.CE and selection of transfected populations with geneticin generated a cell line with 7.9-fold more CE level of N-MYC) had 7.6 ± 2.4-fold more CAT activity than activity than pIRESneo-transfected controls (Table 2). JR1neo20 cells. Similar results were obtained with Additional control transfections with pIRESneo plasmids that pODC.CAT/p∆ODC.CAT plasmids (data not shown). The contained neither the CMV nor the ODC promoter, but did data indicate that the level of N-MYC-mediated CAT reporter contain the cDNA encoding the rabbit CE and the neo resist- activity in these isogenic lines is related to N-MYC protein ance gene, repeatedly yielded, appropriately, no viable levels. colonies. Data in Table 2 show that the ODC promoter mediates CE expression in situ in an N-MYC-expressing cell line.

Sensitivity of IMR32.ODC.CE Cells to CPT-11 We next determined whether the increased level of CE activity in IMR32.ODC.CE cells conferred an increase in sensitivity to CPT-11. IMR32 transfectants were analyzed for CPT- 11 sensitivity by clonogenic assay. Because the long-range goal of these studies is neuroblastoma cell-specific cell kill, results are expressed as concentrations of CPT-11 at which no

viable colonies could be detected (IC99). IMR32.CMV.neo and IMR32.ODC.CE cells were exposed to a range of CPT-11 concentrations for 2 h. Dishes were scored for colonies of greater than eight cells after five doublings of untreated control cells. Notably, 5 µM CPT-11 com- pletely eliminated the clonogenic potential of IMR32.ODC.CE cells, while 50 µM CPT-11 was required to produce the same effect in IMR32.CMV.neo cells (Fig. 6). The 10-fold increase in sensitivity to CPT-11 was concomitant

FIG. 5. CE activity in JR1 isogenic cell lines transiently transfected with either the pCI.ODC.CE or pCIneo plasmid. CE activity was determined with a spectrophotometric assay by measuring the conversion of o-nitrophenol acetate to nitrophenol at 420 nm. Because of technical limitations, results could not be normalized for transfection efficiency; therefore, the experiment was done five times. Each bar represents the average ± SD of five independent experiments. Specific procedures are outlined under Materials and Methods.

MOLECULAR THERAPY Vol. 1, No. 5, May 2000, Part 1 of 2 Parts 461 Copyright The American Society of Gene Therapy ARTICLE

viously been shown to enhance the activity of the constitu- tive SV40 and HSVtk promoters (13, 14). Potentially, the approach described here may have greater tumor specificity. Other promoters have also been used to achieve tumor-specific expression of various cDNAs for appropriate tumors. These promoters include α-fetoprotein for liver carcinoma (20), carcinoembryonic antigen promoter for lung adenocarcinoma (21), c-erbB-2 for adenocarcinoma of the stomach, breast, colon, and ovary (22), and DF3 for myeloma (23) or tyrosinase promoter for melanoma (24). Importantly, Teoh et al. (23) reported successful ex vivo purging of multiple myeloma cells from human hematopoietic progenitor cells by producing tumor cell-specific expression of a drug-activating enzyme in combination with ganciclovir. In that study, bone marrow mononuclear cells contaminated with multiple FIG. 6. Sensitivity of IMR32pIRES.CMV.neo and IMR32pIRES.ODC.CE cells myeloma cells were transduced with a replication-deficient lines to CPT-11, determined by clonogenic assay. Each cell line was exposed adenovirus encoding the DF3 promoter regulating expression to the indicated concentration of CPT-11 for 2 h at which time cells were placed in drug-free medium and allowed to grow for a time equal to five dou- of HSVtk. Subsequent exposure of the contaminated progen- blings of untreated control cells. Dose–response curves are shown, and the itor cells to ganciclovir produced myeloma cell-specific DF3 concentration of drug required to decrease colony number to zero (IC99) is promoter-mediated toxicity. This VDEPT approach was pro- listed in Table 2. Values shown are means ± SD of duplicate determinations posed as a method for purging marrow or peripheral stem from each of two independent determinations. cells prior to autologous transplant. In VDEPT approaches, the most efficacious combination of enzyme/prodrug for any given tumor may ultimately with the increase in CE activity (Table 2). The sensitivity to SN- depend on the properties of individual tumor types as well as 38 of transfectants that expressed the rabbit CE was identical on characteristics of the drugs and enzymes themselves. to that of transfectants that did not express this enzyme (data Relevant to our study is the finding that neuroblastoma cell not shown). Clonogenic assays with IMR32.ODC.CE cells lines and xenografts are relatively sensitive to CPT-11. Even indicate that ODC-regulated expression of CE and subsequent more important may be the observation that as little as a 2- sensitization to CPT-11 are achievable in neuroblastoma cells. fold difference in dose of CPT-11 administered to mice bear- ing human neuroblastoma xenografts results in complete DISCUSSION regressions, compared to a low frequency of complete responses with regrowth of all tumors by week 10 at the lower The novel finding in this study is that the ODC promoter can dose (25). In the study presented here, neuroblastoma cells be used to regulate expression of a drug-activating enzyme via were sensitized to CPT-11 by 10-fold, concomitant with a 7.9- the interaction of N-MYC with this promoter. Specifically, we fold increase in CE activity. Preclinical studies (25) suggest that demonstrated that upregulation of a rabbit liver CE in N- this degree of sensitization may be sufficient to alter the ther- MYC-overexpressing neuroblastoma cells increased CE activi- apeutic outcome of CPT-11 administration. Alternatively, it ty in situ and sensitized tumor cells to the prodrug CPT-11. The may be possible to modify the ODC promoter to increase the potential application of this approach to achieve tumor-spe- potency of the promoter and still retain its specificity to cific toxicity of tumors such as neuroblastoma, glioblastoma, MYC/MAX heterodimer activation (Iyengar and Danks, man- and lung and colon carcinomas that overexpress MYC family uscript in preparation). It is also likely that delivery of the transcription factors is under investigation. ODC/CE cassette to neuroblastoma cells using adenoviral vec- In situ, both c-MYC and N-MYC have been reported to acti- tors will achieve levels of CE not obtainable by simple trans- vate transcription of ODC, as measured by RNA levels (9, 18). fection experiments. Our data are the first to show that an N-MYC/ODC promoter- It may be noteworthy that transient upregulation of ODC mediated increase in transcription produces a concomitant promoter-mediated expression of CAT reporter activity was increase in the target protein. The two- to fourfold increase in easily achieved in all N-MYC-expressing cell lines in this N-MYC/ODC promoter-mediated CAT activity in SJNB-1, study. In contrast, repeated efforts were required to establish SJNB-4, and NB-1691 neuroblastoma cell lines we report here the stable IMR32.ODC.CE transfectant. These observations is similar to the approximately threefold increase in ODC suggest that perhaps ODC-mediated expression of the mRNA in SH-EP neuroblastoma cells following induction of neomycin resistance gene was inadequate to confer resistance N-MYC (18). Interestingly, a correlation between the expres- to the selecting concentration of geneticin, although this is sion of c-myc mRNA and ODC mRNA has also been reported unlikely since transient ODC promoter-mediated expression in primary breast carcinoma tissue (19). was readily detected. Alternatively, high levels of expression of Data presented here are also the first to show that the ODC CE, even in the absence of CPT-11, may be toxic. promoter can be used to regulate expression of drug-activating In addition to the ODC/CE/CPT-11 combination on enzymes. However, MYC-responsive E-box elements have pre- which our study focused, various other

462 MOLECULAR THERAPY Vol. 1, No. 5, May 2000, Part 1 of 2 Parts Copyright The American Society of Gene Therapy ARTICLE

promoter/enzyme/prodrug combinations have been investi- C. A., Houghton, P. J., and Potter, P. M. (1999). Comparison of activation of CPT-11 by rabbit and human carboxylesterases for use in enzyme/prodrug therapy. Clin. Cancer Res. 5: 917–924. gated for their potential use for VDEPT (26). The promoter, 4Matthay, K. (1997). Neuroblastoma: Biology and therapy. Oncology 11: 1857–1875. obviously, must be specific for the targeted tumor. It may be 5Brodeur, G. M., and Castleberry, R. P. (1997). Neuroblastoma. In Principles and Practice of equally important that the enzyme/prodrug combination be Pediatric Oncology (P. A. Pizzo and D. G. Poplack, Eds.), 3rd. ed., pp. 761–797, Lippincott–Raven, Philadelphia. chosen for the antitumor potential of the activated drug with 6Blackwell, T. K., Kretzner, L., Blackwood, E. M., Eisenman, R. N., and Weintraub, H. (1990). respect to a given tumor type, as well as for efficient activation Sequence-specific DNA binding by the c-Myc protein. Science 250: 1149–1151. 7 of drug in situ. Blackwood, E. M., and Eisenman, R. N. (1991). Max: A helix–loop–helix zipper protein that forms a sequence-specific DNA-binding complex with Myc. Science 251: 1211–1217. The two enzyme/prodrug combinations that have been 8Halazonetis, T. D., and Kandil, A. N. (1991). Determination of the c-MYC DNA-binding site. most extensively characterized are HSVtk/ganciclovir and Proc. Natl. Acad. Sci. USA 88: 6162–6166. 9Bello-Fernandez, C., Packham, G., and Cleveland, J. L. (1993). The ornithine decarboxylase cytosine deaminase/5FC. The CE/CPT-11 combination differs gene is a transcriptional target of c-Myc. Proc. Natl. Acad. Sci. USA 90: 7804–7808. from these in several respects: (1) Early clinical data suggest 10Gaubatz, S., Meichle, A., and Eilers, M. (1994). An E-box element localized in the first that the spectrum of tumors against which CPT-11 is active is intron mediates regulation of the prothymosin alpha gene by c-myc. Mol. Cell. Biol. 14: 3853–3862. unique (27), thereby making it a useful addition to the list of 11Ronen, D., Rotter, V., and Reisman, D. (1991). Expression from the murine p53 promoter available prodrugs. (2) Unlike ganciclovir and 5FC, CPT-11 is is mediated by factor binding to a downstream helix–loop–helix recognition motif. Proc. Natl. converted to its active form (SN-38) by human enzymes, as Acad. Sci. USA 88: 4128–4132. 12Reisman, D., Elkind, N. B., Roy, B., Beamon, J., and Rotter, V. (1993). c-Myc trans-activates shown by the activation of ~0.5 to 5.0% of CPT-11 adminis- the p53 promoter through a required downstream CACGTG motif. Cell Growth Differ. 4: 57–65. tered to patients (28). It is not known whether this inefficient 13Sugaya, S., Fujita, K., Kikuchi, A., Ueda, H., Takakuwa, K., Kodama, S., and Tanaka, K. (1996). Inhibition of tumor growth by direct intratumoral gene transfer of herpes simplex virus activation of CPT-11 by human enzymes is a consequence of thymidine kinase gene with DNA–liposome complexes. Hum. Gene Ther. 7: 223–230. low levels of expression of a single efficient enzyme (29) or 14Kumagai, T., Tanio, Y., Osaki, T., Hosoe, S., Tachibana, I., Ueno, K., Kijima, T., Horai, T., and whether CPT-11 is a relatively poor or inaccessible substrate Kishimoto, T. (1996). Eradication of Myc-overexpressing small cell lung cancer cells transfected with herpes simplex virus thymidine kinase gene containing Myc–Max response elements. for several enzymes (3, 29). However, both in vitro (1–3) and in Cancer Res. 56: 354–358. vivo (3) data show that overexpression of the rabbit CE sensi- 15Potter, P. M., Wolverton, J. S., Morton, C. L., Wierdl, M., and Danks, M. K. (1998). Cellular tizes cells to CPT-11, with lower levels of CPT-11 toxic to localization of a rabbit and a human carboxylesterase: Influence on irinotecan (CPT-11) metab- olism by the rabbit enzyme. Cancer Res. 58: 3627–3632. tumor cells, likely reducing the potential for activation of 16Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: A Laboratory CPT-11 by endogenous enzymes to even lower levels. (3) A Manual, Cold Springs Harbor Laboratory Press, Cold Springs Harbor, NY. 17Hann, S. R. (1995). Methionine deprivation regulates the translation of functionally-dis- third difference among the activated forms of CPT-11, ganci- tinct c-MYC proteins. In Diet and Cancer, pp. 107–116, Plenum Press, New York. clovir, and 5FC is the potential for a bystander effect (30, 31). 18Lutz, W., Stöhr, M., Schürmann, J., Wenzel, A., Löhr, A., and Schwab, M. (1996). Unlike the activated forms of ganciclovir and 5FC, SN-38 Conditional expression of N-myc in human neuroblastoma cells increases expression of a-prothymosin and ornithine decarboxylase and accelerates progression into S-phase early after mito- freely diffuses across cell membranes. Initially, we considered genic stimulation of quiescent cells. Oncogene 13: 803–812. that this diffusion would be advantageous in the treatment of 19Mimori, K., Mori, M., Shiraishi, T., Tanaka, S., Haraguchi, M., Ueo, H., Shirasaka, C., and residual disease and disadvantageous for in vitro purging appli- Akiyoshi, T. (1998). Expression of the ornithine decarboxylase mRNA and c-myc mRNA in breast tumors. Int. J. Oncol. 12: 597–601. cations, However, no toxicity to CD34+ progenitor cells is 20Huber, B. E., Richards, C. A., and Krenitsky, T. A. (1991). Retroviral-mediated gene thera- observed even when CD34+ cells are contaminated with as py for the treatment of hepatocellular carcinoma: An innovative approach for cancer therapy. Proc. Natl. Acad. Sci. USA 88: 8039–8043. high as 10% tumor cells that express 800 units of CE activi- 21Osaki, T., Tanio, Y., Tachibana, I., Hosoe, S., Kumagai, T., Kawase, I., Oikawa, S., and ty and exposed to concentrations of CPT-11 that eradicate the Kishimoto, T. (1994). Gene therapy for carcinoembryonic antigen-producing human lung can- clonogenic potential of neuroblastoma cells (Danks and cer cells by cell type-specific expression of herpes simplex virus thymidine kinase gene. Cancer Res. 54: 5258–5261. Meck, unpublished observations). Since it is unlikely that 22Takakuwa, K., Fujita, K., Kikuchi, A., Sugaya, S., Yahata, T., Aida, H., Kurabayashi, T., hematopoietic cell preparations containing 10% tumor cells Hasegawa, I., and Tanaka, K. (1997). Direct intratumoral gene transfer of the herpes simplex would be used for stem-cell rescue, CPT-11 may be useful for virus thymidine kinase gene with DNA–liposome complexes: Growth inhibition of tumors and lack of localization in normal tissues. Jpn. J. Cancer Res. 88: 166–175. purging (Meck and Danks and Guichard and Danks, manu- 23Teoh, G., Chen, L., Urashima, M., Tai, Y.-T., Celi, L. A., Chen, D., Chauhan, D., Ogata, A., scripts in preparation), as well as for the treatment of residual Finberg, R. W., Webb, I. J., Kufe, D. W., and Anderson, K. C. (1998). Adenovirus vector-based purging of multiple myeloma cells. Blood 92: 4591–4601. disease. 24Vile, R. G., and Hart, I. R. (1993). Use of the tissue-specific expression of the herpes sim- In summary, our study supports the hypothesis that use of plex virus thymidine kinase gene to inhibit growth of established murine melanomas following MYC/ODC/CE and CPT-11 for gene therapy/chemotherapy direct intratumoral injection of DNA. Cancer Res. 53: 3860–3864. 25Thompson, J., Zamboni, W. C., Cheshire, P. J., Lutz, L., Luo, X., Li, Y., Houghton, J. A., compares favorably with other combinations and warrants Stewart, C. F., and Houghton, P. J. (1997). Efficacy of systemic administration of irinotecan further study. We are currently evaluating the potential of this against neuroblastoma xenografts. Clin. Cancer Res. 3: 423–431. 26 combination for purging and as an adjunct for treatment of Dachs, G. U., Dougherty, G. J., Stratford, I. J., and Chaplin, D. J. (1997). Targeting gene therapy to cancer: A review. Oncol. Res. 9: 313–325. localized residual disease. 27Furman, W. L., Stewart, C. F., Poquette, C. A., Pratt, C. B., Santana, V. M., Zamboni, W. C. Bowman, L. C., Ma, M. K., Hoffer, F. A., Meyer, W. H., Pappo, A. S., Walter, A. W., and Houghton, P. J. (1999). Direct translation of a protracted irinotecan schedule from a xenograft model to ACKNOWLEDGMENTS Phase I trial in children. J. Clin. Oncol. 17:1815–1824. 28Rivory, L. P., Haaz, R.-C., Canal, P., Lokiec, F., Armand, J.-P., and Robert, J. (1997). This study was supported by CA79763, CA23099, CA63512, CA66124, and Pharmacokinetic interrelationships of irinotecan (CPT-11) and its three major plasma metabo- CA21765 and by American Lebanese Syrian Associated Charities. lites in patients enrolled in Phase I/II trials. Clin. Cancer Res. 3:1261–1266. 29Morton, C. L., Wadkins, R. M., Danks, M. K., and Potter, P. M. (1999). the anticancer pro- REFERENCES drug CPT-11 is a potent inhibitor of acetylcholinesterase but is rapid catalyzed to SN-38 by butyrylcholinesterase. Cancer Res. 59:1458–1463. 1Danks, M. K., Morton, C. L., Pawlik, C. A., and Potter, P. M. (1998). Overexpression of a 30Dilber, M. S., Abedi, M. R., Christensson, B., Bjorkstrand, B., Kidder, G. M., Naus, C. C., rabbit liver carboxylesterase sensitizes human tumor cells to CPT-11. Cancer Res. 58: 20–22. Gahrton, G., and Smith, C. I. (1997). Gap junctions promote the bystander effect of herpes sim- 2Potter, P. M., Pawlik, C. A., Morton, C. L., Naeve, C. W., and Danks, M. K. (1998). Isolation plex virus thymidine kinase in vivo. Cancer Res. 58:2588–2593. and partial characterization of a cDNA encoding a rabbit liver carboxylesterase that activates the 31Laurence, T. S., Rehemtulla, A., Ng, E. Y., Wilson, M., Trosko, J. E., and Stetson, P. L. (1998). prodrug irinotecan (CPT-11). Cancer Res. 58: 2646–2651. Preferential cytotoxicity of cells transduced with cytosine deaminase compared to bystander 3Danks, M. K., Morton, C. L., Krull, E. J., Cheshire, P. J., Richmond, L. B., Naeve, C. W., Pawlik, cells after treatment with 5-flucytosine. Cancer Res. 58:2588–2593.

MOLECULAR THERAPY Vol. 1, No. 5, May 2000, Part 1 of 2 Parts 463 Copyright The American Society of Gene Therapy