Research Article

Gene Therapy (2003) 10, 1519–1527. doi:10.1038/

Directed apoptosis in Cox-2-overexpressing cancer cells through expression-targeted gene delivery

W T Godbey1 and Anthony Atala1

1Laboratory for Cellular Therapeutics and Tissue Engineering, Harvard Medical School, The Children's Hospital, Boston, MA, USA

Correspondence: Dr A Atala, Laboratory for Cellular Therapeutics and Tissue Engineering, Harvard Medical School, The Children's Hospital, 300 Longwood Ave., Enders Bldg. #461, Boston, MA 02115, USA

Received 23 September 2002; Accepted 15 January 2003.



The principle of promoter-targeted gene delivery was used to direct the expression of reporter genes and inducible caspases to Cox-2-overexpressing cancer cells. The polycation poly(ethylenimine) was used in unmodified form to nonvirally deliver genes into cells, and targeting was achieved at the transcriptional level. Results demonstrated that reporter expression was reduced by an average of 89.8% in normal cells and cell lines not overexpressing Cox-2 when the strong cytomegalovirus promoter was replaced with the human Cox-2 promoter in delivered plasmids. Cocultures of normal and Cox-2-overexpressing cancer cells showed less than 0.5% reporter expression in normal fibroblast cells but over 35% reporter expression in PC3 prostate cancer cells. This targeting method was then used to direct the expression of inducible forms of caspases 3 and 9 to Cox-2-overexpressing cancer cells of the bladder and prostate. Following activation of the resulting caspase pro-forms, cells underwent apoptosis as evidenced by DNA fragmentation and cytoskeletal degradation. This result was also observed in cells resistant to apoptosis in terms of TNF-alpha initiation. Such directed apoptosis could eventually serve as a treatment for an entire class of Cox-2-overexpressing carcinomas.


expression targeting, apoptosis, caspase, Cox-2, transfection, polyethylenimine



An idealistic goal for cancer therapy would be a treatment that causes cancer cells to disappear, leaving behind only healthy, untransformed tissue. One approach to this oversimplified goal utilizes the technology of gene transfer, whereby cancer cells themselves produce an ultimately fatal protein that causes their own demise. Simply having tumor cells produce a fatal toxin is not sufficient to achieve the stated goal, however, because of lingering effects upon neighboring cells following cancer cell death. For example, the release of the toxic gene product into the extracellular space following necrotic cell death would cause a severe bystander effect resulting in harm to neighboring tissue. A more desirable process resulting in cell death is apoptosis, where cells package their own degradation products to prevent the bystander effect.

Gene delivery has been used in the past in an attempt to treat various cancers.1,2 However, independent of bystander effects, the use of many gene delivery methods adversely affects healthy, untransformed cells because of indiscriminant transfection of all cell types in a given area. A method to selectively target cancer cells for treatment is desirable, but the similarity of transformed cells to normal somatic cells makes this objective extremely difficult to achieve. Although the attachment of ligands to gene delivery complexes is a method that has yielded some progress in targeted gene delivery to normal tissues,3 this method has not produced much direct success with cancer cells because, in part, of the similarities between the receptors expressed by tumor cells and the tissues from which they originated. Commonalities between tumor cells yet also unique to tumor cells are continually being sought.

While the discovery of a cancer-specific receptor or transmembrane protein would be an invaluable boon for oncology research, the potential for tumor-targeted gene therapy is not limited to complexes modified with cell-specific ligands or antibodies for increased complex uptake. The expression of any unique protein potentially could be utilized to target a given cell type. If the protein is exocytosed, it could serve as a chemoattractant. If the protein is cytoplasmic, gene delivery complexes that react with the given protein could be manufactured to cause vectors to release their carried DNA only within the cytoplasms of the targeted cells. As well, expression of a protein can often be traced back to the transcriptional level. If a distinct promoter or enhancer controlled the gene for the given protein, and this upstream binding element was active within the tumor cell, then plasmids that contain the same element could be engineered for expression in such cells. This is the principle behind promoter-based gene targeting. While several cell types might endocytose the gene delivery complexes and translocate the delivered oligonucleotides into their nuclei, if the delivered exon is under the control of a promoter that is utilized by only one cell type, then transcription of the exon will theoretically only occur within that cell type.

Many cancer cells do display similar behaviors. One such phenotype, the constitutive overexpression of cyclooxygenase 2 (Cox-2), is linked with cellular resistance to senescence and therefore resistance to apoptosis,4 a hallmark of cancerous cells. Constitutive Cox-2 overexpression is also implicated as a component of tumorigenesis. Cox-2 is an inducible component in the prostaglandin synthesis cascade, and is inducible in normal cells by many cytokines, mitogens, and proinflammatory factors. However, in the absence of these signals, normal cells do not express Cox-2. This fact makes the Cox-2 promoter a desirable candidate for targeting tumor cells at the transcriptional level.

It was our goal to instigate apoptosis within cancer cells using a nonviral method to deliver caspase genes. The gene delivery vehicle selected for this work was poly(ethylenimine) (PEI), a polycation chosen for its branched structure to aid in DNA protection5 as well as its potential for a high degree of protonation because of a high amine concentration within each molecule. Owing to an abundance of primary amines within this polymer, the attachment of ligands or signaling molecules to PEI is not a chemical challenge. However, one goal of these investigations was to determine whether PEI can be used to deliver genes to cells in a targeted fashion without chemical modification.

The present investigation seeks to answer the question of whether targeted gene therapy can be achieved at the DNA level, through the use of appropriate promoters, in such a way that entire classes of cells can be treated with one transfection. Specifically, this work is aimed at determining whether the phenomenon of Cox-2 overexpression that is seen in numerous carcinomas can be exploited to target the expression of delivered plasmids to such tumors. As a check to this system, healthy, untransformed cells are included to determine whether these cells, which do not normally express Cox-2 in unstressed situations, will be excluded from the effects of delivered Cox-2-driven plasmids.

After the establishment of a system whereby Cox-2-overexpressing cancer cells are easily targeted, the focus of this work shifts to a modified system with potential clinical significance. The question of whether tumor cells can be destroyed through the use of targeted, artificially induced apoptosis in an in vitro application is addressed. This question includes cells that are often classified as apoptosis resistant to demonstrate the potential power of this novel application.



A total of eight cell types were used for demonstrating whether Cox-2-overexpression could be used to target certain tumor types. The cells chosen originated from one of the two groups: normal, untransformed cells, or cancerous cells. The untransformed cells were either human foreskin fibroblasts (HFF) or prostate epithelial cells (PrEC). The types of carcinomas selected represent three affected tissues types – prostate (PC3 and LNCaP cell lines), bladder (HTB5 and HTB1), and breast (MB231 and MCF7) – and within these subgroups two cell lines, one that constitutively overexpresses Cox-2 and one that does not, were investigated to verify the principle of expression targeting. The level of Cox-2 expression in these cell types was verified by RT-PCR (Figure 1a). Of the cell types examined, only PC3, HTB5, and MB231 overexpressed Cox-2.

Figure 1.
Figure 1 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact or the author

Cox-2 transcription levels correlate with Cox-2-driven expression of delivered genes. (a) Cox-2 transcription levels as determined by RT-PCR. (b) Normalized comparison of targeted versus untargeted transfections. When adjusted for the positive control (CMV-driven) transfection efficiency for each cell type, only the three Cox-2 overexpressing cell lines expressed the GFP reporter plasmid. *=Significantly different from the combined data sets for all non-Cox-2-overexpressing cells (ngreater than or equal to4, P<0.05).

Full figure and legend (70K)

An enhanced green fluorescent protein (Gfp) reporter gene driven by the human Cox-2 promoter (Cox2-Gfp) was delivered to cells via the polycation PEI. The results showed that Gfp expression was reduced by an average of 89.8% in normal cells and cell lines not overexpressing Cox-2 when the strong cytomegalovirus (CMV) promoter was replaced with the human Cox-2 promoter in delivered plasmids (not shown). Since the efficiency of gene transfer differs between cell types, transfection efficiencies for each cell type were normalized to their positive controls, where the delivered gene was the Gfp gene driven by the strong promoter CMVie (Cmv-Gfp). This normalization is expressed as a ratio of the transfection efficiencies of Cox2-Gfp to Cmv-Gfp, where transfection efficiency is defined as the percentage of cells expressing GFP out of the entire population. Using this methodology, it was found that only the three Cox-2-overexpressing cell lines – PC3, HTB5, and MB231 – expressed GFP under the Cox-2 promoter control system at levels significantly higher than untransformed control cells (Figure 1b).

To further demonstrate the applicability of Cox-2-driven gene expression for tumor cell targeting, co-cultures of transformed and normal cells were grown and transfected with the Cox2–Gfp reporter. To help distinguish between the two cell types within the coculture, a monoculture of one cell type was fluorescently labeled prior to combining it with a monoculture of a different cell type. Figure 2 is an example of this procedure, where untransformed HFF cells were labeled red before mixing them with a culture of immortal PC3 cells. After combining the two cell types to create an easily distinguishable coculture, the cells were incubated overnight prior to transfection to facilitate gene delivery. Transfection results from delivery of the nonspecific Cmv-Gfp control plasmid showed a roughly equivalent percentage of HFF and PC3 cells expressing the green reporter. (Red-labeled cells expressing the GFP reporter often showed up as yellow on computer-generated overlay images.) However, when the Cox2-Gfp plasmid was delivered to cocultures, the untransformed cells failed to express the reporter in appreciable amounts. (The transfection efficiency of the untransformed cells using this system was less than 0.5%.) However, PC3 and HTB5 cells transfected in the HFF coculture setups expressed the GFP reporter at roughly 25–35% transfection efficiency in the Cox2-driven systems. Figure 2 shows the results of such a transfection in an HFF/PC3 coculture with red-labeled fibroblasts, wherein a lack of red-labeled cells expressing the green reporter is noted in panel b. The converse experiment, which utilized red-labeled cancer cells and unlabeled fibroblasts, was also performed and produced similar transfection efficiencies (not shown). Figure 2 also shows the monoculture transfection efficiencies of these cell types when transfected with each of the two types of promoter-driven plasmids.

Figure 2.
Figure 2 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact or the author

Transfection of labeled cocultures. Fibroblasts are labeled with red dye, PC3 cancer cells are unlabeled. The coculture cells were transfected with a Gfp gene driven by one of the two promoters. (a) Transfection with Cmv-Gfp plasmid. Both fibroblasts and PC3 cells express the GFP reporter. (b) Transfection with Cox2-Gfp plasmid. Only PC3 cells express the reporter. The transfection efficiency of Cox-2-driven expression was <0.05% for fibroblasts in coculture. F='fibroblast,' CA='cancer cell,'+='expresses reporter, '-='not expressing reporter' (c) Percentage of cells in monoculture expressing GFP when transcription was directed by the Cmv or the Cox2 promoter. CMV-driven transfection efficiencies were typically in the range of 25–35%. *=significantly different from HFF transfection efficiency (ngreater than or equal to6, P<0.00001): filled square=HFF cells, square=PC3 cells, Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact or the author HBT5 cells.

Full figure and legend (348K)

Functional plasmids under the control of the Cox-2 promoter were also constructed. These plasmids coded for modified forms of proteins involved with apoptosis. The basic proteins, caspases 3 and 9, are members of the apoptosis cascade, with caspase 9 being an initiator molecule and caspase 3 being an executer molecule within the cascade.6 The modifications applied to these proteins resulted in pro-forms of the molecules that require homodimerization for their activation. The homodimerization is achieved through the addition of AP20187 (ARIAD, Cambridge, MA, USA). The inducibility of the gene products gives an additional level of control to the gene delivery system at the post-translational level, a potentially important feature for when the system moves to an in vivo setting.

Phase contrast images show that many Cox-2-overexpressing cells transfected with the Cox2-iCasp3 plasmids condensed within 8 h after the addition of the AP20187 activator (Figure 3). In addition, many cells within these cultures demonstrated a more granular cytoplasm, and there appeared to be a greater number of vesicular structures in the supernatants as compared to controls – transfected cells not receiving AP20187 and HFF cell AP20137 (not shown). These findings were also seen in Cox2-iCasp9-transfected cells that received the AP20187 activator. These qualitative findings are consistent with apoptosis.

Figure 3.
Figure 3 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact or the author

Morphological changes in cells receiving inducible caspase gene. Cells transfected with a gene coding for inducible caspase 3 condensed following administration of the activator AP20187, as shown by phase contrast images. The same was true for inducible caspase 9 (not shown): (a) HTB5 cells and (b) PC3 cells.

Full figure and legend (187K)

On a molecular level, apoptosis is characterized, in part, by fragmentation of chromosomal DNA into nucleosomes. To show that the morphological changes seen in Cox-2-overexpressing cells were caused by apoptosis, ELISA was used to detect and quantify these histone-bound DNA fragments in HTB5 cells 8 h after the addition of activator. This method also allowed for the distinction between apoptosis and necrosis, based on the fact that necrosis is characterized by rapid plasma membrane permeablization and would have been completed much sooner than 8 h postactivation. Removing cell supernatants prior to the ELISA procedure removed any DNA fragments produced by necrotic processes. Results from these experiments demonstrate a significant increase (P<0.05) in the number of histone-bound DNA fragments 8 h after the addition of AP20187 in Cox2-iCasp3- and Cox2-iCasp9-transfected cells (Figure 4). ANOVA analysis of these data demonstrates no significant difference between groups for the set containing all controls plus iCasp transfections that not only included the AP20187 activator (ANOVA, P>0.775), but also included either of the iCasp + AP20187 data sets resulting in significant differences between groups (P<0.0005).

Figure 4.
Figure 4 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact or the author

DNA fragmentation in response to caspase production and activation. DNA fragmentation was detected via ELISA against nucleosome-bound DNA. A405 results for all samples were normalized to those of sham transfections. CID=chemical inducer of dimerization (AP20187 – activator of delivered caspases), iCasp3 mutant=inactive variation of inducible caspase 3, no prom=transfections that utilized a promoterless iCasp3 plasmid, *=significantly different from sham-treated data (P<0.05), ###=no difference between groups (ANOVA, P>0.775).

Full figure and legend (57K)

The existence of apoptosis was further confirmed at the molecular level through fluorescent immunostaining. Fluorescein-conjugated antibodies were used to label a specific cleavage site of cytokeratin 18 (CK18), an intermediate filament protein. Cells transfected with Cox2-iCasp3 which also received the activator stained positively for cleaved CK18, while sham-transfected cells and cells transfected with the Cox2-iCasp3 plasmids but not receiving the activator did not stain for cleaved CK18 (Figure 5). With cytoskeleton degradation being one of the results of apoptosis, the positive staining of cleaved CK18 seen in transfected cells supports the claim that the delivered, activated caspases cause apoptotic death in affected cells.

Figure 5.
Figure 5 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact or the author

Cytoskeletal degradation in response to caspase production and activation. Immunostaining of HTB5 cells with a fluorescein-conjugated antibody to a specific cleavage product of CK18 demonstrates cytoskeletal breakdown in cells transfected with Cox2-iCasp3 following addition of AP20187 activator: (a) Sham-transfected cells, (b) Cox2-iCasp transfection, without activator, (c) Cox2-iCasp transfection, with activator.

Full figure and legend (352K)



Our study shows that the polycation PEI can be used to deliver genes into cells in a targeted fashion without modification of the polymer. We also demonstrated that cells that constitutively overexpress Cox-2 will readily transcribe delivered plasmids that are under the control of the Cox-2 promoter. In addition, we proved that inducible forms of both initiator and executor caspases are effective in such a controlled system on a variety of tumor types, including those resistant to apoptosis because of a lack of certain receptors.

The polycation PEI was selected for transfection for several reasons. A nonviral delivery vehicle was desirable because of issues related to a host immunological response to virus administration, in addition to problems concerning virulence caused by possible interactions with wild-type viruses. These considerations are important in light of the potential clinical relevance of the described treatment system. Among nonviral gene delivery methods, PEI has been shown to yield relatively high transfection efficiencies, in part because PEI offers protection to the DNA it carries following endocytosis5 as the delivered DNA is trafficked into cell nuclei.7 This is in contrast to liposomal methods, which deliver unprotected DNA into cell cytoplasms.8 Although no mechanism for DNA release from PEI following endocytosis has yet been shown, there exists evidence that such release is not necessarily required for transcription of the delivered genes.9

The concept behind the majority of gene delivery targeting regimens involves the conjugation of ligands to gene delivery complexes. This idea was first used prior to gene delivery work with the attachment of ligands to liposomes for targeted delivery of drugs or proteins to specific cell types.10,11 However, problems still exist with ligand attachment methods in the in vivo setting caused by possible recognition of ligands by immune cells or potential clearance in the liver because of either properties of the ligands themselves or the binding of serum proteins to ligands that mark the complexes for clearance. Partially in response to these complications, another form of targeted gene delivery has evolved that does not utilize cell-specific ligands; targeting is achieved at the transcriptional level through the use of upstream binding elements that are specific to certain cell types. Examples of using cell-specific promoters for gene delivery targeting include dendritic (Langerhans) cells,12 ovarian cells,13 and B-lymphoid cells,14 among others.

The work presented herein sought to utilize promoter targeting for targeting gene delivery to cancer cells. While there has yet to be demonstrated a single gene expressed universally, and uniquely, by all cancer cells, certain classes of cancers do demonstrate common expression characteristics. It has been observed that COX-2 is constitutively overexpressed in many carcinomas, including those of intestinal epithelia,15 esophageal squamous cells,16 pancreatic cells,17 various colorectal tumors,4 adenocarcinoma,18 plus specific mammary,19 prostate,20 and bladder21 cancers. Although Cox-2 is an inducible gene in untransformed cells, it is not normally expressed in healthy, unstressed tissues, including breast,22 prostate,20 and bladder23. This makes the Cox-2 promoter an interesting focus for cancer cell-targeted gene delivery.

As shown in Figure 1 and Figure 2, Cox-2-driven reporter expression was only seen in Cox-2-overexpressing cells. Of note is the lack of expression in normal cells, especially in coculture conditions. This specificity of expression serves as one component of the safety of a potential Cox-2-driven treatment of cancer cells. In addition, Cox-2 overexpression is thought to be one of the preliminary events in cell transformation,16,24 so this targeting method could also be useful as an early stage treatment, affecting precancerous cells that might go otherwise undetected.

Following the establishment of the expression-targeted system for singling out Cox-2-overexpressing cancer cells, Cox-2-driven plasmids were modified to code for a functional gene. Although the ultimate goal of these investigations is to destroy cancer cells, delivering a toxic protein that will result in tumor death via cell necrosis is undesirable because lethal proteins would be released from transfected cells following lysis. A bystander effect would result that would potentially kill healthy, untransformed cells in the area. It is for this reason that a gene product involved with apoptosis is a desirable candidate for bringing about cell death, and that caspases were selected as the functional product of our Cox-2-targeted transfections, because apoptotic cells package their own degradation products to prevent a bystander effect.

Of note is the fact that many cancers that constitutively overexpress Cox-2 are also apoptosis resistant,4,15,18 which is why some carcinomas display resistance to proapoptotic drugs.25 This feature is the driving force behind Cox-2 inhibition therapies against cancer, such as NSAID delivery,26 which are designed to lower Cox-2 expression within cancer cells to render them more susceptible to senescence and apoptosis. The investigations we report here do not claim to alter Cox-2 expression, so it might be of concern that the cells we are targeting might still be resistant to apoptosis. In the case of the bladder tumor cell line HTB5, we first verified that cells were apoptosis resistant through administration of tumor necrosis factor-alpha (TNF-alpha). This molecule was selected because it has been shown experimentally to be an initiator of apoptosis.27 TNF-alpha is a type II membrane protein that acts as a ligand to TNF receptors to bring about the activation of pro-forms of initiator caspases, in much the same way that the Fas ligand interacts with Fas receptors to spark the apoptotic cascade.6 We observed a lack of apoptotic response to TNF-alpha, which served to verify that the cells used were apoptosis resistant. However, the results shown in Figure 3-5 indicate that such cells are not necessarily immune to apoptosis. Lacking the appropriate receptors does make a cell resistant to many of the signals that commonly initiate the apoptotic cascade, but the data show that such cells are still susceptible to apoptosis if a signal is provided for a downstream event in the cascade.

Previously published work addresses the phenomenon of heightened COX-2 expression in cancer cells, specifically in the area of cancer treatment via Cox-2 inhibition.26 Cox-2 inhibitors are already in the market to combat inflammation, sold under the tradenames Vioxx™ and Celebrex™. However, the work presented here is distinct from these examples in that it is not our goal to alter COX-2 levels, but merely to use heightened Cox-2 expression levels as a means for guidance of gene expression in transfected cells.

Other published work relevant to the present investigation includes the delivery of genes coding for caspases for apoptosis induction.28,29 In those investigations, adenovirus-mediated gene delivery complexes were delivered to prostate tumors through direct injection into tumor sites. Part of the novelty of the system we present here is that potential immunogenicity is lowered through the use of nonviral gene delivery complexes. The Cox-2-driven system is also novel in terms of greater robustness because of its potential for applicability to all Cox-2-overexpressing carcinomas. The targeting feature of the system could eliminate the need for directly injecting tumors for complex delivery.

There is also mention in the literature regarding the use of the Cox-2 promoter in gene delivery constructs.30 The work dealt with the use of the Cox-2 promoter to reduce the level of toxicity caused by thymidine kinase gene delivery in the liver. While the work demonstrated success in lowering hepatic toxicity, the referenced investigations were not focused on targeting specific tumor cells. Our studies were directed at proving the validity and robustness of Cox-2 promoter-driven gene expression in several tumor lines. Cox-2-overexpressing tumor cells were forced into apoptosis, while neighboring cells remained unaffected by the gene delivery or bystander effects.

The system described in this paper could potentially be applicable in the clinical setting for Cox-2-overexpressing tumors, possibly via direct tumor injection or systemic administration of transfection complexes. A potential weakness of the Cox-2-driven apoptosis induction system in this setting lies in the possibility that untransformed cells that happen to be overexpressing Cox-2 will be forced into apoptosis. For example, Cox-2 is expressed by cells as a part of the inflammatory process, so the presence of inflammation could result in widespread cell death following activation of the Cox-2-driven transfection system. Local inflammation is occasionally associated with some tumor types. If an extratumor inflammatory process were to be present, Cox-2 overexpression could be minimized with a short course of anti-inflammatory agents prior to initializing the tumor treatment. The presence of limited Cox-2 expression does not result in apoptosis in our system: RT-PCR data from Figure 1 show that the normal cell types tested did express Cox-2 to a limited extent in the experimental environments in which they were kept, but the transfection data show that this limited amount of Cox-2 expression was not enough to yield reporter expression or apoptosis using the Cox-2-driven transfection system. Late-stage cancer may be associated with systemic inflammation, and in such cases the use of the Cox-2-driven transfection system may not be applicable. Clinically, the described system may include a pretreatment tumor analysis in order to confirm the presence of Cox-2 overexpression. Despite the known Cox-2 expression characteristics present in major tumor classes, Cox-2 expression variability may occur between individual patients.



The results reported here indicate that entire classes of cells can be treated via gene therapy that employs the judicious use of appropriate upstream binding elements for transcriptional control. It was shown that the phenomenon of Cox-2 overexpression seen in numerous tumor types can be exploited to target the expression of delivered plasmids to such tumors, as was demonstrated in experiments utilizing cells of bladder, prostate, and mammary cancers. Healthy, untransformed cells do not normally express Cox-2, and did not express delivered Cox-2-driven plasmids.

This work also served to further the concept of Cox-2 promoter-targeting for cancer cells through the use of the targeting system in an in vitro application designed to bring about selective cell death through apoptosis. The nonviral polycation PEI was sufficient for gene delivery into cells and yielded transfection efficiencies typically in the range of 25–35% of cells expressing reporter. The use of inducible caspase technology served to demonstrate that either an initiator or an executor caspase is sufficient by itself to trigger apoptotic cell death, even in certain apoptosis-resistant cancer cells. Such selective cell death is potentially applicable to precancerous cells, which are thought to overexpress Cox-2 prior to their transformation.


Materials and methods


Both normal and cancerous tissues were represented by a variety of cell types. Normal tissue was represented by prostate epithelial cells (PrEC) and human foreskin fibroblasts (HFF). Transformed cells were used to represent cancers of the prostate (PC3, LNCaP), bladder (HTB1, HTB5), and breast (MCF7, MB231)(Table 1).


Engineered plasmids were manufactured using pEGFP-N1 (Clontech, Palo Alto, CA, USA) as a starting vector. The existing CMVie promoter was excised from pEGFP-N1 using AseI and BglII restriction enzymes, followed by blunt ending with the Klenow fragment and religation using T4 ligase. The promoterless plasmid was then cut with XhoI and HindIII, followed by insertion of the human Cox-2 promoter (-891 to +9)31 (courtesy of KK Wu, The University of Texas Health Science Center at Houston, Houston, TX, USA) excised to contain sticky ends complementary to the plasmid being engineered. The resulting plasmid coded for an enhanced green fluorescent protein (EGFP) under the control of the human Cox-2 promoter. Additional constructs were produced through modification of this reporter plasmid through replacement of the EGFP exon with the exon for inducible caspase 3 (iCasp3) or iCasp9 (both generous gifts from D Spencer, Baylor College of Medicine, Houston, TX, USA) using SacII and BamHI restriction sites. Because of an existing BamHI site within the coding region of the caspase plasmids, partial digestion of the donated iCasp plasmids was required and achieved through the use of suboptimal BamHI digestion conditions. Appropriately sized fragments were excised after agarose gel electrophoresis, and purified using a Gel Extraction Kit (Qiagen).

Activation of inducible caspases

The proform caspases were made active through addition of AP21087, a component of the ARIAD homodimerization kit (ARIAD, Cambridge, MA, USA). The molecule was delivered 2 days post-transfection at a concentration of 1.0 nM in growth medium.

Analysis of transcription

Transcription levels of Cox-2 and the housekeeping gene GAPDH were analyzed via RT-PCR of total RNA. Cells were grown in 10 cm dishes to approximately 80% confluence, followed by scraping and RNA isolation via a commercially available kit (cat. #74104, Qiagen, Valencia, CA, USA). Cox-2 cDNA (304 bp) was amplified using the following primers: sense TTCAAATGAGA TTGTGGGAAAATTGCT, antisense AGATCATCTCTG CCTGAGTATCTT. GAPDH cDNA was amplified using the following primers: sense TCACCATCTTCCAG GAGCG, antisense CTGCTTCACCACCTTCTTGA. RT-PCR was performed using 34 cycles of denaturation at 95°C for 1 h 30 min, annealing at 58°C for 2 h, and extension at 73°C for 40 sec.

RT-PCR products were loaded onto a 1.5% agarose gel and run at 85 V for approximately 1 h. Photoimaging was performed with a Stratagene Eagle Eye II, and printed using Eagle Sight version 3.2 software. Densitometry was then performed using NIH Image 1.61 software.


Gene delivery was performed as previously published32 using the nonviral gene delivery vehicle PEI. Cells used for in vitro PEI-mediated transfections were plated at approximately 1.04 times 104 cells/cm2 (100 000 cells per 35 mm dish) and allowed to grow approximately 16 h prior to transfection. Specifications of the PEI-mediated transfections include a 7.5:1 PEI amine to DNA phosphate ratio based upon 3.6 mug of DNA per 35 mm plate transfection. The PEI used (Sigma-Aldrich, St Louis, MO, USA, cat. #40,872-7) had a weight-average molecular weight (Mw) of approximately 25 000 Da. Transfections were allowed to proceed for 2 h at 37°C, after which the transfection medium was aspirated and replaced by 2 ml of growth medium.

Analyses of transfection were performed either visually using an Olympus IX70 inverted microscope with an IXFLA fluorescence attachment, or via FACS using a FACScalibur (Becton Dickinson, San Jose, CA, USA) set to a flow rate of 1 mul/s with Cell Quest version 3.3 software. Transfection efficiencies for GFP-transfected cells are defined as the percentage of cells expressing GFP per total number of cells counted.

Coculture experiments

For certain cells to be used in coculture experiments, the red fluorescent marker 5 (and 6)-chloromethyl SNARF®-1 acetate (cat. # C-6826, Molecular Probes, Eugene, OR, USA) was used to prelabel cultures. For a given pair of cell types, each was grown to late log stage in T-25 flasks. The cell type to be labeled was washed once with phosphate-buffered saline (PBS)/EDTA, and incubated at 37°C for 30 min in the appropriate serum-free medium containing 20 muM of the cell label. The staining medium was removed and the cells were washed once with PBS/EDTA followed by trypsinization for plating of the cocultures. Cocultures were carried out in six-well plates with approximately 50 000 cells of each cell type (100 000 cells total) plated 1 day prior to transfection.

DNA fragmentation analysis

ELISA was used to detect histone-bound DNA fragments. The Cell Death Detection ELISAPlus (Roche, Mannheim, Germany) kit employed anti-histone–biotin antibodies to bind the histone component of the nucleosomes from cell lysates to streptavidin-coated 96-well plates. Anti-DNA-POD was used to bind the developing agent to the DNA component of the bound nucleosomes. Cells were grown in 24-well polystyrene plates, and cell lysis was achieved by using 150 mul of the supplied lysis buffer per well, followed by incubation for 5 min at 37°C. All subsequent steps were performed according to the manufacturer's recommendations.

Cytoskeletal degradation analysis

Immunofluorescence staining was used to detect cytoskeletal degradation, which is a marker of apoptosis. M30 CytoDEATH*, Fluorescein (Roche) is a fluorescein-conjugated mouse IgG specific to a caspase-cleaved epitope of the human cytokeratin 18 cytoskeletal protein. Following fixation in cold methanol for 1 h at 4°C, cells (grown on a 24-well culture plate) were washed twice with PBS containing 0.1% Tween 20. The M30 CytoDEATH* antibody was diluted 1:120 in a solution of PBS containing 1.0% bovine serum albumin and 0.1% Tween 20, and 100 mul was added to each well of prepared cells. The reactions were allowed to proceed on a shaker in the dark for 1 h at room temperature. Following the incubation, cells were mounted with one drop of Vectashield that contained DAPI (for labeling nuclei).


All phase contrast and fluorescent cell images were captured using an Olympus IX70 inverted microscope and Magnafire imaging software.


Groups of data were analyzed by single-factor ANOVA. For pairwise comparisons, the F-test was used to determine whether a given pair of population variances was equal (alpha=0.05). This information was then used in designating the appropriate t-tests (typically heteroscedastic) to perform for comparing the means of population pairs. Significantly different pairs were defined as having P<0.05.



  1. Rubin J et al. Phase I study of immunotherapy of hepatic metastases of colorectal carcinoma by direct gene transfer of an allogeneic histocompatibility antigen, HLA-B7. Gene Therapy 1997; 4: 419–425. | Article |
  2. Vogelzang NJ, Lestingi TM, Sudakoff G, Kradjian SA. Phase I study of immunotherapy of metastatic renal cell carcinoma by direct gene transfer into metastatic lesions. Hum Gene Ther 1994; 5: 1357–1370. | PubMed | ChemPort |
  3. Hood JD et al. Tumor regression by targeted gene delivery to the neovasculature. Science 2002; 296: 2404–2407. | Article | PubMed | ISI | ChemPort |
  4. Watson AJ. Chemopreventive effects of NSAIDs against colorectal cancer: regulation of apoptosis and mitosis by COX-1 and COX-2. Histol Histopathol 1998; 13: 591–597. | PubMed | ISI | ChemPort |
  5. Godbey WT et al. Poly(ethylenimine)-mediated transfection: a new paradigm for gene delivery. J Biomed Mater Res 2000; 51: 321–218. | Article | PubMed | ChemPort |
  6. Huppertz B, Frank HG, Kaufmann P. The apoptosis cascade – morphological and immunohistochemical methods for its visualization. Anat Embryol (Berlin) 1999; 200: 1–18. | Article | ChemPort |
  7. Godbey WT, Wu KK, Mikos AG. Tracking the intracellular path of poly(ethylenimine)/DNA complexes for gene delivery. Proc Natl Acad Sci USA 1999; 96: 5177–5181. | Article | PubMed | ChemPort |
  8. Zelphati O, Szoka Jr FC. Mechanism of oligonucleotide release from cationic liposomes. Proc Natl Acad Sci USA 1996; 93: 11493–11498. | Article | PubMed | ChemPort |
  9. Bieber T et al. Intracellular route and transcriptional competence of polyethylenimine–DNA complexes. J Control Release 2002; 82: 441.
  10. Davidenkova EF. Current status of the problem of the treatment and prevention of hereditary diseases. Zh Nevropatol Psikhiatr im SS Korsakova 1982; 82: 9–17.
  11. Matthay KK, Heath TD, Papahadjopoulos D. Specific enhancement of drug delivery to AKR lymphoma by antibody-targeted small unilamellar vesicles. Cancer Res 1984; 44: 1880–1886.
  12. Morita A et al. Development of a Langerhans cell-targeted gene therapy format using a dendritic cell-specific promoter. Gene Therapy 2001; 8: 1729–1737. | Article |
  13. Bao R, Selvakumaran M, Hamilton TC. Targeted gene therapy of ovarian cancer using an ovarian-specific promoter. Gynecol Oncol 2002; 84: 228–234. | Article | PubMed | ChemPort |
  14. Maxwell IH, Glode LM, Maxwell F. Expression of the diphtheria toxin A-chain coding sequence under the control of promoters and enhancers from immunoglobulin genes as a means of directing toxicity to B-lymphoid cells. Cancer Res 1991; 51: 4299–4304. | PubMed | ChemPort |
  15. DuBois RN et al. G1 delay in cells overexpressing prostaglandin endoperoxide synthase-2. Cancer Res 1996; 56: 733–737. | PubMed | ISI | ChemPort |
  16. Shamma A et al. Up-regulation of cyclooxygenase-2 in squamous carcinogenesis of the esophagus. Clin Cancer Res 2000; 6: 1229–1238. | PubMed | ISI | ChemPort |
  17. Molina MA et al. Increased cyclooxygenase-2 expression in human pancreatic carcinomas and cell lines: growth inhibition by nonsteroidal anti-inflammatory drugs. Cancer Res 1999; 59: 4356–4362. | PubMed | ISI | ChemPort |
  18. Battu S, Rigaud M, Beneytout JL. Resistance to apoptosis and cyclooxygenase-2 expression in a human adenocarcinoma cell line HT29 CL.19A. Anticancer Res 1998; 18: 3579–3583. | PubMed |
  19. Subbaramaiah K et al. Transcription of cyclooxygenase-2 is enhanced in transformed mammary epithelial cells. Cancer Res 1996; 56: 4424–4429. | PubMed | ISI | ChemPort |
  20. Kamijo T, Sato T, Nagatomi Y, Kitamura T. Induction of apoptosis by cyclooxygenase-2 inhibitors in prostate cancer cell lines. Int J Urol 2001; 8: S35–S39.
  21. Bostrom PJ et al. Expression of cyclooxygenase-1 and -2 in urinary bladder carcinomas in vivo and in vitro and prostaglandin E2 synthesis in cultured bladder cancer cells. Pathology 2001; 33: 469–474.
  22. Hwang D, Scollard D, Byrne J, Levine E. Expression of cyclooxygenase-1 and cyclooxygenase-2 in human breast cancer. J Natl Cancer Inst 1998; 90: 455–460. | Article | PubMed | ChemPort |
  23. Yoshimura R et al. Expression of cyclooxygenase-2 in patients with bladder carcinoma. J Urol 2001; 165: 1468–1472. | Article | PubMed | ISI | ChemPort |
  24. Howe LR, Subbaramaiah K, Brown AM, Dannenberg AJ. Cyclooxygenase-2: a target for the prevention and treatment of breast cancer. Endocr Relat Cancer 2001; 8: 97–114. | Article | PubMed | ISI | ChemPort |
  25. Haq R, Zanke B. Inhibition of apoptotic signaling pathways in cancer cells as a mechanism of chemotherapy resistance. Cancer Metast Rev 1998; 17: 233–239.
  26. Elder DJ, Halton DE, Hague A, Paraskeva C. Induction of apoptotic cell death in human colorectal carcinoma cell lines by a cyclooxygenase-2 (COX-2)-selective nonsteroidal anti-inflammatory drug: independence from COX-2 protein expression. Clin Cancer Res 1997; 3: 1679–1683. | PubMed | ISI | ChemPort |
  27. Katsen AD, Vollmar B, Mestres-Ventura P, Menger MD. Cell surface and nuclear changes during TNF-alpha-induced apoptosis in WEHI 164 murine fibrosarcoma cells. A correlative light, scanning, and transmission electron microscopical study. Virchows Arch 1998; 433: 75–83. | Article | PubMed | ChemPort |
  28. Xie X et al. Adenovirus-mediated tissue-targeted expression of a caspase-9-based artificial death switch for the treatment of prostate cancer. Cancer Res 2001; 61: 6795–6804. | PubMed | ISI | ChemPort |
  29. Shariat SF et al. Adenovirus-mediated transfer of inducible caspases: a novel 'death switch' gene therapeutic approach to prostate cancer. Cancer Res 2001; 61: 2562–2571. | PubMed | ISI | ChemPort |
  30. Yamamoto M et al. Characterization of the cyclooxygenase-2 promoter in an adenoviral vector and its application for the mitigation of toxicity in suicide gene therapy of gastrointestinal cancers. Mol Ther 2001; 3: 385–394. | Article | PubMed | ISI | ChemPort |
  31. Tazawa R, Xu XM, Wu KK, Wang LH. Characterization of the genomic structure, chromosomal location and promoter of human prostaglandin H synthase-2 gene. Biochem Biophys Res Commun 1994; 203: 190–199. | Article | PubMed | ISI | ChemPort |
  32. Godbey WT, Wu KK, Hirasaki GJ, Mikos AG. Improved packing of poly(ethylenimine)/DNA complexes increases transfection efficiency. Gene Therapy 1999; 6: 1380–1388. | Article | PubMed |


We thank Shay Soker, PhD for scientific discussions. Funding for this work was provided by the National Institutes of Health (R01-DK57260).