A Genetically Retargeted Adenoviral Vector Enhances Viral Transduction in Esophageal Carcinoma Cell Lines and Primary Cultured Esophageal Resection Specimens

ORIGINAL ARTICLES

A Genetically Retargeted Adenoviral Vector Enhances ViralTransduction in Esophageal Carcinoma Cell Lines and

Primary Cultured Esophageal Resection Specimens

Christianne J. Buskens, MD,*† Willem A. Marsman, MD, John G. Wesseling, PhD,†G. Johan A. Offerhaus, MD,§ Masato Yamamoto, PhD,� David T. Curiel, MD, PhD,�

Piter J. Bosma, PhD,† and J. Jan B. van Lanschot, MD*

Objective: To evaluate if an integrin-retargeted adenoviral vectorcould establish a more efficient and tumor-specific gene transfer inesophageal carcinoma cells.Summary Background Data: Although preclinical data indicatedthat adenoviral gene therapy could be a promising novel treatmentmodality for various malignancies, clinical results are often disap-pointing. An important problem is the decreased tumoral expressionof the Coxsackie and adenovirus receptor (CAR), which mediatesadenoviral entry. Retargeting the adenoviral vector to other cellularreceptors, by inserting an arginine-glycine-aspartate (RGD) tripep-tide in the fiber knob, might overcome this problem.Methods: Four esophageal carcinoma cell lines and 10 fresh surgi-cal resection specimens were cultured. All were infected with thenative adenovirus (Ad) and the retargeted adenovirus (AdRGD),encoding for the reporter genes luciferase or Green FluorescentProtein to analyze gene transfer efficiency.Results: In all cell lines, an increase in viral expression per cell andan increase in the percentage of transduced cells were seen with theretargeted adenovirus. Also, in the primary cultures of carcinomacells, a more efficient gene transfer was seen when the retargetedvector was used. This phenomenon was less pronounced in normalcells, indicating that the RGD virus transduces tumor cells moreefficiently than normal cells.Conclusions: This study demonstrates that an RGD retargetedadenovirus infects human esophageal carcinoma cells with enhancedefficiency, while in normal esophageal cells this effect is lesspronounced. Therefore, this retargeted vector is expected to have a

better performance in vivo, when compared with nonretargetedvectors used for cancer gene therapy so far.

(Ann Surg 2003;238: 815–826)

Esophageal cancer is a highly aggressive disease with poorlong-term outcome. So far, surgical resection is still the

treatment of choice when curation is sought. However, thistreatment is associated with high morbidity and mortality,and even after potentially curative surgery, 5-year survivalrates rarely exceed 25%.1 Therefore, new approaches in thetreatment of this malignancy are considered. Gene therapywith adenoviral vectors seems a promising novel treatmentmodality for various malignancies in preclinical studies,2,3

but so far clinical results are often disappointing. This is atleast partly due to the following two important problems.

First, there is a significant discrepancy between theadenoviral vector efficacies observed in vitro using estab-lished cell lines and the tumor transduction rates achievablein in vivo delivery.4 Although carcinoma cell lines are valu-able tools for investigating various aspects of transductionefficiencies, only specific cell types can be studied. In addi-tion, these cells consist of a monolayer of monoclonal tumorcells, which are transformed due to multiple passages withlimited similarity to the heterogeneous carcinoma cell popu-lation in vivo. Therefore, it would be desirable to have aculture system of primary cells to study adenoviral transduc-tion in cells more comparable with the in vivo situation.

A second important limitation of the in vivo applicationof the current adenoviral cancer gene therapy is the resistanceof carcinoma cells to adenoviral infection. Adenoviral entryinto target cells is the rate-limiting step of gene delivery. Theinitial binding of an adenovirus to the cell surface is areceptor-mediated process; therefore, efforts have been madeto characterize the cellular receptors of the adenovirus. Re-cently, one of the adenoviral receptors on the surface of a host

From the Departments of *Surgery, †Liver Center, ‡Gastro-enterology, and§Pathology, Academic Medical Center/ University of Amsterdam, TheNetherlands; and �Division of Human Gene Therapy Center, Depart-ments of Medicine, Pathology, and Surgery, and the Gene TherapyCenter, University of Alabama, Birmingham, AL.

Supported by grants from the National Institutes of Health (R01 HL67962,P50 CA89019, R01 CA86881, R01 AG021875, R01 CA090547).

Reprints: C.J. Buskens, MD, Academic Medical Center, Deptartment ofSurgery, Suite G4-130, Meibergdreef 9, 1105 AZ Amsterdam, TheNetherlands. E-mail: [email protected].

Copyright © 2003 by Lippincott Williams & Wilkins0003-4932/03/23806-0815DOI: 10.1097/01.sla.0000098622.47909.c0

Annals of Surgery • Volume 238, Number 6, December 2003 815

cell was identified as the Coxsackie and adenoviral receptor(CAR), which is a protein that is also involved in maintainingtight junction integrity.5,6 Upon binding of the knob domainof the viral fiber protein to CAR, the virion enters the cellthrough the interaction of its penton base with the �v�3 andthe �v�5 integrins on the host cell surface.7 This is followedby the internalization of the virus within a clathrin-coatedendosome. Finally, the virus escapes the endosome, translo-cates to the nuclear pore complex, and releases its genomeinto the nucleoplasm where subsequent steps of viral repli-cation take place (Fig. 1).

Several investigations have revealed a decrease in ex-pression of the CAR protein in neoplastic cells, which mayexplain the low transduction efficiency of carcinoma cellsseen in vivo.7,8 To overcome the limitations imposed by the

CAR dependence of adenoviral infection, targeting can beused by genetically modifying the fiber knob, enabling thevirus to attach and infect via another cell surface receptor.Since tumor cells express integrins abundantly, a retargetedadenoviral vector with an arginine-glycine-aspartate (RGD)peptide inserted into the HI-loop of the fiber knob might solvethe problem of poor transduction of tumor cells by allowinga CAR-independent gene transfer directly through integrins.9

The first aim of this study was to establish a culturesystem from fresh surgical resection specimens to studyadenoviral transduction in cells more comparable with the invivo situation. An additional advantage of such a system isthat apart from analyzing infection percentages in (heteroge-neous) carcinoma cells of various patients, the transductionefficiencies of adenoviral vectors in carcinoma cells candirectly be compared with normal squamous mucosal cells ofthe same patient.

The secondary aim of this study was to analyze theefficiency of the genetically retargeted vector for esophagealcarcinoma in comparison with the native adenovirus. TheRGD-retargeted adenovirus was tested on four establishedhuman esophageal carcinoma cell lines and on 10 primarycell cultures from fresh surgical esophageal resection mate-rial. Because primary cultures of human material will containvarious cell types (especially fibroblasts, lymphocytes, andhematopoietic cells) in addition to cancer cells,10–12 wesubsequently characterized the origin of the cells that wereinfected by both vectors.

MATERIALS AND METHODS

Established Esophageal Carcinoma Cell LinesThe human esophageal adenocarcinoma cell lines

OE19 and OE33 were purchased from the European Collec-tion of Cell Cultures (Salisbury, UK). The squamous cellcarcinoma cell lines TE1 and TE2 have been isolated andcharacterized at Tohoku University (Sendai, Japan).13 Thesecell lines were maintained in Dulbecco modified Eagle me-dium (DMEM) supplemented with 10% heat-inactivated fetalcalf serum, 300 �g/mL L-glutamine, 100 U/mL penicillin,and 100 �g/mL streptomycin (DMEM/ 10% FCS/ L-glu/ Pen/Strep), at 37°C in a humidified, 10% CO2 atm.

Primary Cell Cultures of Esophageal ResectionSpecimens

This study was performed in accordance with theguidelines of the local ethics committee. In The Netherlands,a written informed consent is not required to perform addi-tional experiments on primary resected material.

In patients who underwent subtotal esophageal resec-tion and proximal gastrectomy with curative intent for ade-nocarcinoma of the esophagus or gastroesophageal junctionor squamous cell carcinoma of the esophagus, the left gastricartery was preserved until the end of the resectional phase of

FIGURE 1. A: Entry of native adenoviral vector by binding theviral fiber knob (closed circles) to the Coxsackie and adenoviralreceptor (CAR). B: Genetical targeting of the adenoviral vectorby inserting an arginine-glycine-aspartate (RGD) peptide intothe HI loop of the fiber knob (open circles). C: The retargetedadenoviral vector can directly bind to the integrins, allowing aCAR-independent gene transfer.

Buskens et al Annals of Surgery • Volume 238, Number 6, December 2003

© 2003 Lippincott Williams & Wilkins816

the operation to maintain maximal viability of the removedspecimen. Samples of normal squamous esophageal mucosawithout the underlying submucosal connective tissue andproper muscle, and vital tumor (assigned by experienced GIpathologists G.J.A.O. or F.T.K.) were collected, washed inphosphate-buffered saline (PBS), and minced into smallpieces. The fragments of mucosa or tumor were resuspendedin 10 mL Liver Digest Medium (Life Technologies, Breda,The Netherlands) and floated in a shaking waterbath at 37°C.After 2 hours, the supernatant containing dissociated cellswas decanted, supplemented with 50% fetal calf serum, andcentrifuged (1500 rpm/min for 5 minutes at 4°C). The singlecells were resuspended in cold culture medium (DMEM/ 10%FCS/ L-glu/ Pen/ Strep) and stored on ice until use. Thedissociation step was repeated with fresh Liver Digest Me-dium three times. All dissociated cells obtained per tissuesample were pooled and applied to tissue culture platescoated with collagen type IV.

Two X 105 cells in a well of a 24-well plate or iniso-concentration in 6-well plates were cultured for 24 hoursin 0.5 mL DMEM/ 10% FCS/ L-glu/ Pen/ Strep supplementedwith 0.1 mg/mL Fungizone, in a humidified atmospherecontaining 10% CO2 at 37°C.

Adenoviral ConstructsThe native adenoviral vectors (Ad), derived from hu-

man adenovirus type 5, were obtained from R.D. Gerard(University of Leuven, Leuven, Belgium). For adenoviraltransduction analysis, the adenovirus was made replicationdeficient by deleting the E1 region. Into this deleted E1region, the reporter genes luciferase (Adluc) and Green Flu-orescent Protein (AdGFP) were cloned and constructed to bedriven by the human cytomegalovirus promoter. These re-porter genes were used to analyze viral transduction levels bymeasuring the amount of luciferase and GFP transcription.

The genetically retargeted adenoviral vectors containingthe recombinant RGD (Arg-Gly-Asp) peptide in the HI-loopof the fiber knob were generated by transfection of 293 cellswith PacI-digested pVK703,14 and constructed to encode foreither the luciferase or GFP reporter gene (AdlucRGD andAdGFPRGD, respectively).9 The viral vectors were propa-gated on the permissive cell line 293 and purified by doublecesium chloride gradient centrifugation.15 All virus prepara-tions were dialyzed against PBS, aliquoted, and stored at-80°C until use. Adenovirus titers in plaque forming units(pfu) were determined in parallel by a plaque forming assayusing 293 cells, and the number of viral particles (vp) wasmeasured by optical density-based physical titration. Thevp/pfu ratios of the native vectors were 10 to 20 and those ofthe RGD-modified vectors were 40 to 50.

Adenoviral Gene TransferThe adenoviral transduction experiments on the four

cell lines were performed in triplicate. Cell lines were grown

overnight at a count of 1 X 106 cells/well in 6-well plates toallow adherence. The monolayers of cells were then washedwith PBS, and incubated for 1 hour with the different adeno-viral constructs at various multiplicity of infection of 1, 10,and 100 pfu/cell. Then, complete fresh medium was added,and after 24 hours of incubation, the luciferase and GFPexpression was analyzed for the different viral constructs. Forall experiments, a sample of cells in which no virus wasadded was used as a negative control.

After cell lysis, quantitative levels of viral luciferaseproduction in the cells were measured in a luminometer(Berthold Detection System, Pforzheim, Germany) using aluciferase assay system (Promega, Madison, WI). The proteinconcentration of the lysates was determined with a proteinassay (Pierce Biotechnology, Rockford, IL).

Infection percentages were analyzed by determining thepercentage of GFP expressing cells by fluorescence activatedcell scanning (FACS; BD Biosciences, San Jose, CA). Cul-tured cells were trypsinized, washed in PBS, and fixated in 1mL 2% paraformaldehyde. After centrifuging the cells at1000 rpm for 5 minutes, the cells were resuspended in PBSwith 1% bovine serum albumin. The percentage of trans-duced cells was determined by gating the right-hand tail ofthe distribution of the negative control sample for eachindividual cell line at a 1% positivity (based on the negativecontrol results), which was considered as autofluorescence ofthe cell line.

To analyze adenoviral transduction levels in primarycells, the single cells were cultured for 24 hours after isola-tion. Then, cultures were rinsed with PBS and vital cells (ie,cells attached to the collagen coating) were exposed to thedifferent adenoviral vectors as described above. Dependenton the amount of primary cells available, the experimentswere performed at least in duplicate at various multiplicitiesof infection. For quantitative viral expression levels, cellswere infected with Adluc or AdlucRGD and lysed after 24hours for luciferase measurement. The transduction ratio (ie,luciferase activity per sample infected with AdlucRGD di-vided by the luciferase activity per sample infected withAdluc) was used to analyze the differences between genetransfer increase in carcinoma cells and normal epithelialcells from the same patient.

To analyze if the quantitative increase in viral expressionlevels was due to an increase in the number of transduced cellsor an increase in the viral expression per cell, the primarycultures were also infected with AdGFP and AdGFPRGD andanalyzed by fluorescence microscopy.

Characterization of Primary CulturesThree of the 10 primary cultures infected with AdGFP

and AdGFPRGD were used to characterize the origin of thedifferent primary cells and to determine the infection percent-ages of the various cell types by two-color flow-cytometric

Annals of Surgery • Volume 238, Number 6, December 2003 Gene Therapy for Esophageal Carcinoma

© 2003 Lippincott Williams & Wilkins 817

analysis. Cells were trypsinized, washed, and resuspended inPBS-Tween (0.2%) to make the cells permeable for thedetection of intracellular cytokeratin proteins by FACS anal-ysis. Cytokeratins are part of the cytoskeleton of epithelialcells and are not present in fibroblasts and other stromal cells.All cells were incubated for 60 minutes with saturatingconcentration of the MNF116 mouse-antihuman cytokeratinmonoclonal antibody (DAKO, Glostrup, Denmark),16 whichreacts with human epithelial tissue and malignant epitheliallesions. Cells were then washed 3 times in PBS and incubatedfor another 60 minutes with the secondary fluorescein-iso-thiocyanate-labeled goat antimouse immunoglobulin G anti-body (Jackson, Westgrove, PA). After 3 more washing steps,the cells were resuspended in 1 mL 2% paraformaldehyde. Asa negative control, an aliquot of cells from each patient wasused in which control IgG was added to the cells withoutprevious labeling with MNF116. The percentage of trans-duced epithelial cells was analyzed by determining the per-centage GFP positivity and cytokeratin positivity per primarycultured tissue by quadrant statistics after two-color flowcytometry.

Statistical AnalysisTo analyze if the increase in luciferase activity with the

retargeted adenoviral vector was significantly higher in car-cinoma cells than in normal squamous epithelial cells, aparametric Student t test was used to compare the transduc-tion ratios of the primary cells. A P value of �0.05 wasconsidered statistically significant. The statistical analysiswas performed using the Statistical Software Package version11.5 (SPSS Inc., Chicago, IL).

RESULTS

Adenoviral Gene Transfer in EstablishedEsophageal Carcinoma Cell Lines

To compare the efficiency in gene transfer of theparental virus to the genetically modified virus, the fouresophageal carcinoma cell lines were infected with Adluc andits retargeted variant AdlucRGD as described in “Materialsand Methods.” A dramatic augmentation of luciferase activitywas seen with the vector containing the Arg-Gly-Asp (RGD)peptide inserted into the HI-loop of the fiber knob (Fig. 2).With a multiplicity of infection of 1 and 10 pfu/cell,AdlucRGD demonstrated in the esophageal carcinoma celllines OE19, OE33, TE1, and TE2 an approximately 250-,30-, 300-, and 700-fold increase, respectively, in luciferaseactivity, compared with the cells infected with the parentalvirus. With a multiplicity of infection of 100, this increasewas less pronounced, but there was still an 80-, 10-, 150-, and400-fold enhancement in viral expression for the carcinomacell lines with AdlucRGD.

To analyze whether this enhancement was due to anincrease in luciferase expression per cell or an increase innumber of cells infected, the experiments were repeated withthe viruses encoding GFP. With all experiments, an increasein the percentage of infected cells with the AdGFPRGDvector was seen for all four cell lines (Fig. 3). In addition,there was also an increase in the intensity of fluorescence percell, as was seen by the shifting of the bulk of the cells to theright. This indicates that the increase in luciferase expressionseen with AdlucRGD is caused by both an increased expres-sion per cell as well as a higher percentage of infected cells.

Adenoviral Gene Transfer in Primary CellCultures of Esophageal Resection Specimens

Promising results in established cell lines do not alwaysreliably predict efficient adenoviral transduction of carcinomacells in vivo. Therefore, 10 cultures of both normal esopha-geal epithelium and carcinoma cells were established asdescribed in “Materials and Methods.” This primary materialwas derived from 6 men and 4 women with a median age of62 years (range 46–82 years). Three patients had a distalesophageal adenocarcinoma developed in a Barrett segment,4 patients had an adenocarcinoma of the cardia, and 3 patientshad a squamous cell carcinoma. The tumor characteristics aredescribed in Table 1.

All 10 primary cultures were infected with the fourviral constructs expressing either luciferase (Adluc andAdlucRGD) or GFP (AdGFP and AdGFPRGD). The increasein viral luciferase expression of the retargeted viral vectorversus the parental vector is shown for both the carcinomacells and the normal squamous epithelial cells in Figure 4Aand B. In all patients, an increase in viral expression was seenwith AdlucRGD versus the native adenoviral vector, varyingfrom a 5- to almost a 100-fold increase. This transductionratio for both normal squamous epithelial cells and carcinomacells is shown in Figure 4C. The increase in gene transfer wassignificantly less pronounced in normal cells (P � 0.03). Incombination with the fact that the absolute luciferase expres-sion of AdRGD was higher in carcinoma cells than in normalcells, this suggests a more specific transduction of esophagealcarcinoma cells by the RGD retargeted adenoviral vector.

To analyze whether this increase in viral expressionrepresented more transduced cells or an increase in viralexpression per cell, all primary cultures were also infectedwith both the parental AdGFP and the genetically retargetedvariant AdGFPRGD, and subsequently analyzed by fluores-cence microscopy. It was demonstrated that with the retar-geted virus both the number of infected cells, and the fluo-rescence intensity per cell increased in all cultures (Fig. 5,upper panels). However, the cultures of primary tumor ma-terial seemed to contain two cell populations with differentmorphology. About 50% of the cells revealed a fibroblast-like



morphology, and most likely are stromal cells. The remainingcells showed a more epithelial cell-like shape and probablyare cancer cells (Fig. 5, lower panels). These data thereforesuggest that the increased transduction of the primary culturesmight result from a better transduction of stromal cells, cancercells, or both.

Characterization of Infected Cell TypesTo discriminate between tumor cells and stromal cells,

the primary cultures of patients 1, 3, and 6, were analyzed byflow cytometry for the epithelial cytokeratin marker MNF116.As expected, the primary cultures were indeed a mixture ofcells from different origin. To analyze differences in thetransduction efficiency of AdGFP and AdGFPRGD in bothtumor cells and stromal cells, a two-color FACS analysis for

GFP positivity and cytokeratin positivity was performed forthe three primary cultures (Fig. 6). The percentages of unin-fected epithelial cells (GFP-negative and MNF116-positive),infected epithelial cells (GFP-positive and MNF116-posi-tive), uninfected stromal cells (GFP-negative and MNF116-negative), and infected stromal cells (GFP-positive andMNF116-negative) per culture were analyzed by quadrantstatistics. The increase in percentage of transduction withAdGFPRGD was determined for the different cell popula-tions. In these three cultures, an increase was seen of 13.8%,28.7%, and 4.5%, respectively, in the percentage of trans-duced epithelial cells, and an increase of 30.0%, 23.3%, and8.1%, respectively, in the percentage of transduced stromalcells when compared with infection with AdGFP. Theseresults indicate that the RGD retargeted adenovirus infected

FIGURE 2. Adenovirus-mediated gene transfer of the parental adenoviral vector and the integrin retargeted adenoviral vector,both encoding luciferase (Adluc and AdlucRGD, respectively) to human esophageal carcinoma cell lines (OE19, OE33, TE1, andTE2). Cells were infected with either virus at a multiplicity of infection of 1, 10, or 100. Luciferase levels were corrected for theprotein concentration and are shown in light units (LU)/mg protein. Background luciferase activity in the samples in which no viruswas added (negative control) is also shown.



both primary epithelial cells and stromal cells more effi-ciently.

DISCUSSIONIn this study, the utility of a retargeted adenoviral

vector containing an RGD peptide in the HI-loop of the fiberknob was examined for gene therapy for esophageal carci-noma. It was shown that the retargeted virus had a 5 to 800times more efficient gene transfer than the native virus inestablished esophageal cancer cell lines and in primary cul-

tured esophageal carcinoma cells. Since the transductionincrease was significantly less pronounced in normal cellswith AdRGD and the adenoviral transgene levels were higherin carcinoma cells, the retargeted adenoviral vector was alsodemonstrated to be more efficient for carcinoma cells than theparental virus. The observation that a higher percentage ofcarcinoma cells is transduced with a more specific genetransfer is of clinical importance because the therapeuticeffect of gene therapy will depend on efficient gene transferto carcinoma cells and with fewer viral particles needed for

FIGURE 3. An example of flow-cytometric analysis showing the percentage of infected cells from four esophageal carcinoma celllines (OE19, OE33, TE1, and TE2) comparing the parental to the retargeted virus encoding GFP (AdGFP and AdGFPRGD,respectively). On the y-axis, the number of counted cells is shown (counts), and on the x-axis, the intensity of the GreenFluorescent Protein expression (FL1-height) is depicted.



the same therapeutic effect, the vector-related toxicity islikely to be decreased.

Gene therapy, although originally developed for cor-rection of genetic deficiencies of inherited disorders, hasbecome a promising therapeutic entity for various carcino-mas.3 In cancer gene therapy, therapeutic genes are intro-duced into tumor cells aiming at arrest of tumor growth. Tocompete with conventional therapeutic modalities, cancergene therapy should be both safe and effective. Targeting ofvectors can be used to increase both the selective transductionand the transduction efficiency. Genetic targeting by insertingthe tripeptide RGD into the HI-loop of an adenoviral vectorwas previously demonstrated to enhance infection of ovarian,pancreatic, and head-and-neck carcinomas with promising invitro results.17–20 Despite promising adenoviral cancer genetherapy results in preexisting cell lines, so far only 2 of themore than 500 approved clinical trials are being evaluated ina phase III clinical trial due to disappointing results invivo.21,22 To be able to analyze gene therapy vectors in cellsmore resembling the in vivo situation, a culture system wassuccessfully established from fresh surgical resection speci-mens. Culturing gastrointestinal mucosa has been proven tobe extremely difficult.23 We also encountered several diffi-culties of this primary culture system. It was noted thatpreserving the left gastric artery intact until the end of theresectional phase of the operation was important to maintainmaximal viability of the esophageal cells. Sometimes a fun-gal overgrowth developed (especially in specimens frompatients with an obstructing tumor or candida esophagitis), orthe viability of the cultures was limited (especially in speci-mens from patients with an ulcerating tumor). However,

ultimately our established culture system was reproducible andunique in allowing a comparison of gene transfer in primarycarcinoma cells and normal squamous epithelial cells.

Apart from epithelial cells, primary cultures establishedfrom fresh surgical resection specimens were demonstrated tocontain stromal cells that are also more efficiently transducedwith the retargeted virus. Due to the artificial culturingconditions, these primary cultures will probably contain arelatively high percentage of fibroblasts, which are suscepti-ble for viral infection.12 However, fibroblasts will also bepresent in the stroma of a primary tumor. Thus, transductionof stromal cells may also play a role in vivo. On the one hand,transduction of stromal cells might have a positive therapeu-tic effect because the vitality of a tumor is at least partlydependent on its stroma, and attacking these stromal cellsmay result in a more efficient gene therapy, but on the otherhand, tumoral stroma is considered a barrier limiting tumorexpansion,24 and tumors lacking stromal cells have beenreported to be more aggressive and to induce more metasta-ses.25 This phenomenon therefore implies that transduction ofstromal cells may counteract the therapeutic possibilities ofnonreplicating cytotoxic adenoviral vectors in cancer genetherapy.

For treating neoplastic diseases, carcinoma cells shouldbe efficiently transduced and killed. In previous studies of ourgroup, it was demonstrated that the absolute luciferase ex-pression after viral transduction is representative for theamount of cell kill with a suicide gene.17 However, nonrep-licating adenoviral vectors encoding suicide genes were dem-onstrated to be not the most efficient vectors to be used in aclinical setting. Nowadays, conditionally replicating adeno-

TABLE 1. Characteristics and Histopathologic Findings of 10 Tumors Used to Establish Primary Cultures

Patient No. Operation Tumor Type

TumorLength

(cm)Infiltration

DepthLymph Node

StatusDistant

MetastasisDifferentiation

Grade Radicality

1 THE Barrett* 6.4 T3 N1 M0 Poor R0‡

2 THE Barrett 7.1 T3 N1 M0 Poor R03 THE Barrett 3.0 T1 N0 M0 Good R04 TTE AC cardia† 4.5 T2 N1 M0 Moderate R05 THE AC cardia 6.2 T3 N0 M0 Poor R06 THE AC cardia 7.0 T3 N1 M0 Moderate R07 THE AC cardia 4.5 T1 N1 M0 Moderate R08 TTE SSC 8.0 T3 N1 M1 Poor R09 THE SSC 11.0 T3 N1 M0 Poor R2§

10 TTE SSC 5.4 T3 N0 M0 Moderate R0

THE, transhiatal esophageal resection; TTE, transthoracic esophageal resection; SCC, squamous cell carcinoma of the esophagus.*Adenocarcinoma of the distal esophagus developed in a Barrett segment.†Adenocarcinoma of the gastric cardia.‡Microscopically radical resection.§Macroscopically irradical resection.



viruses (CRAds) represent a promising and novel way forcancer gene therapy.26 These agents are designed to replicatespecifically in tumor cells, followed by the spread of the viralprogeny to neighboring cancer cells. This specific replication

will at the same time prevent the adverse effects that could becaused by infection of stromal cells. An example is theONYX-015 virus, which has been modified in the earlyregulatory protein E1b that normally allows the virus to bind

FIGURE 4. A: Luciferase levels in LU/mg protein representing the adenoviral gene transfer of Adluc and AdlucRGD in primarycultured normal squamous epithelial cells from resection material of 10 patients. B: Luciferase levels (LU/mg protein) in carcinomacells of the same 10 patients. C: Transduction ratio (ie, increase in luciferase expression) in normal squamous epithelial cellscompared with carcinoma cells. The increase in gene transfer was significantly lower in normal cells (P � 0.03).



and inactivate the host p53 gene to promote its own replica-tion. The mutated adenovirus, however, only replicates in andlysis human cells with a defective p53 pathway.27,28 Thisvirus is expected to be especially useful in esophageal carci-nomas since for this malignancy a p53 mutation is the mostfrequent alteration identified (75–100%).29 It has also beendemonstrated that p53 mutations are an early event during the

malignant degeneration of esophageal cells.29 Therefore, thisvirus could even be used for the treatment of premalignantstages. Targeting such CRAds with the RGD tripeptide couldimprove the utility of this adenoviral vector by creating aCAR-independent infection capability for tumor cells.

To our knowledge, this is one of the first reportedstudies that has investigated the possibilities for adenoviral

FIGURE 6. Example of two-color flow-cytometric analysis to determine the percentage of transduced epithelial and stromal cells.The cultured carcinoma cells were infected with AdGFP and AdGFPRGD, and incubated with the cytokeratin antibody MNF116.The percentage of GFP-positive cells (FL1-height) and cytokeratin-positive cells (FL-2 height) was analyzed by quadrant statisticsfor the upper left (uninfected epithelial cells), upper right (infected epithelial cells), lower left (uninfected stromal cells), and lowerright quadrant (infected stromal cells) (UL, UR, LL, LR, respectively).

FIGURE 5. Example of fluorescence microscopy analysis of primary esophageal carcinoma cells transduced with the native andretargeted virus encoding GFP (AdGFP and AdGFPRGD, respectively). The infected cells (green) are shown in the upper panels.With light microscopy of the same material, the cultures of primary tumor material seemed to contain two cell populations withdifferent morphology (lower panels). About 50% of the cells had a fibroblast-like morphology (black arrow); the remaining cellswere smaller and had a more epithelial cell-like shape (white arrow). Original magnification: x10 objective.



gene therapy to treat a gastrointestinal malignancy in primarytumor cells. However, we acknowledge that although thisculture system is more resembling the in vivo situation thanthe established monoclonal cell lines, it is still an artificialmonolayer test system. Because the complex three-dimen-sional structure of a tumor might influence the transductionefficiency of an adenoviral vector, the development of asystem using cultured biopsies to test viral infection is cur-rently under investigation.

It should also be noted that using the RGD virus willonly partly solve the current difficulties with cancer genetherapy. Also, with this retargeted adenoviral vector it seemsimpossible to transduce 100% of the tumor cells, which isnecessary to create efficient cancer treatment, and althoughthis vector did establish a more efficient gene transfer incarcinoma cells in comparison to normal squamous epithelialcells, it did not create a selective viral transduction. There-fore, further research has to be focused on more selectivetargeting motifs before gene therapy can be incorporated indaily clinical practice.30

CONCLUSIONThis study demonstrates that adenoviral entry via the

primary adenoviral receptor CAR is limited in esophagealcarcinoma cells. Adenoviral entry was increased when anintegrin retargeted adenoviral vector was used. The increasein viral expression was not only due to an increase in thepercentage of infected cells, but also to an increased geneexpression per cell. Therefore, genetical targeting of theadenoviral vectors with an RGD tripeptide seems a promisingtreatment strategy to optimize cancer gene therapy.

ACKNOWLEDGMENTSThe authors thank Dr. J.B. Reitsma his help with the

statistical analyses and Dr. W.N. Dinjens for his assistancesetting up cell cultures.

REFERENCES1. Daly JM, Fry WA, Little AG, et al. Esophageal cancer: results of an

American College of Surgeons Patient Care Evaluation Study. J Am CollSurg. 2000;190:562–572.

2. Kovesdi I, Brough DE, Bruder JT, et al. Adenoviral vectors for genetransfer. Curr Opin Biotechnol. 1997;8:583–589.

3. Gomez-Navarro J, Curiel DT, Douglas JT. Gene therapy for cancer. EurJ Cancer. 1999;35:2039–2057.

4. Mountain A. Gene therapy: the first decade. Trends Biotechnol. 2000;18:119–128.

5. Bergelson JM, Cunningham JA, Droguett G, et al. Isolation of acommon receptor for Coxsackie B viruses and adenoviruses 2 and 5.Science. 1997;275:1320–1323.

6. Walters RW, Freimuth P, Moninger TO, et al. Adenovirus fiber disruptsCAR-mediated intercellular adhesion allowing virus escape. Cell. 2002;110:789–799.

7. Pearson AS, Koch PE, Atkinson N, et al. Factors limiting adenovirus-mediated gene transfer into human lung and pancreatic cancer cell lines.Clin Cancer Res. 1999;5:4208–4213.

8. Dmitriev I, Krasnykh V, Miller CR, et al. An adenovirus vector with

genetically modified fibers demonstrates expanded tropism via utiliza-tion of a coxsackievirus and adenovirus receptor-independent cell entrymechanism. J Virol. 1998;72:9706–9713.

9. Krashnykh VN, Mikheeva GV, Douglas JT, et al. Generation of recom-binant adenovirus vectors with modified fibers for altering viral tropism.J Virol. 1996;70:6839–6846.

10. Siggelkow H, Hilmes D, Rebenstorff K, et al. Analysis of human primarybone cells by fluorescence activated cell scanning: methodological problemsand preliminary results. Clin Chim Acta. 1998;272:111–125.

11. Trifillis AL, Cui X, Jacobs S, et al. Culture and characterization ofnormal epithelium from cystoscopic biopsies of human bladder. In VitroCell Dev Biol. 1993;29:908–911.

12. Hoganson DK, Sosnowski BA, Pierce GF, et al. Uptake of adenoviralvectors via fibroblast growth factor receptors involves intracellularpathways that differ from the targeting ligand. Mol Ther. 2001;3:105–112.

13. Nishihira T, Kasai M, Mori S, et al. Characteristics of two cell lines(TE-1 and TE-2) derived from human squamous cell carcinoma of theesophagus. Gann. 1979;70:575–584.

14. Reynolds PN, Dmitriev I, Curiel DT. Insertion of an RGD motif into theHI loop of adenovirus fiber protein alters the distribution of transgeneexpression of the systemically administered vector. Gene Ther. 1999;6:1336–1339.

15. Graham FL, Prevec L. Methods for construction of adenovirus vectors.Mol Biotechnol. 1995;3:207–220.

16. Goddard MJ, Wilson B, Grant JW. Comparison of commercially avail-able cytokeratin antibodies in normal and neoplastic adult epithelial andnon-epithelial tissues. J Clin Pathol. 1991;44:660–663.

17. Vanderkwaak TJ, Wang M, Gomez-Navarro J, et al. An advancedgeneration of adenoviral vectors selectively enhances gene transfer forovarian cancer gene therapy approaches. Gynecol Oncol. 1999;74:227–234.

18. Kanerva A, Wang M, Bauerschmitz GJ, et al. Gene transfer to ovariancancer versus normal tissues with fiber-modified adenoviruses. MolTher. 2002;5:695–704.

19. Wesseling JG, Bosma PJ, Krasnykh V, et al. Improved gene transferefficiency to primary and established human pancreatic carcinoma tar-getcells via epidermal growth factor receptor and integrin-targetedadenoviral vectors. Gene Ther. 2001;8:969–976.

20. Dehari H, Ito Y, Nakamura T, et al. Enhanced antitumor effect of RGDfiber-modified adenovirus for gene therapy of oral cancer. Cancer GeneTher. 2003;10:75–85.

21. Vorburger SA, Hunt KK. Adenoviral gene therapy. Oncologist. 2002;27:746–759.

22. Post LE. Selectively replicating adenoviruses for cancer gene therapy: an updateon clinical development. Curr Opin Invest Drugs. 2002;3:1768–1772.

23. Sheridan M, Reid IM, Mothersill C, et al. Culture of human esophagealendoscopic biopsies. In Vitro Cell Dev Biol Anim. 1995;31:338–339.

24. Van den Hooff A. Connective tissue as an active participant in theprocess of malignant growth. Anticancer Res. 1986;6:775–780.

25. Van den Hooff A. Stromal involvement in malignant growth. AdvCancer Res. 1988;50:159–196.

26. Suzuki K, Alemany R, Yamamoto M, et al. The presence of theadenovirus E3 region improves the oncolytic potency of conditionallyreplicative adenoviruses. Clin Cancer Res. 2002;8:3348–3359.

27. Heise C, Sampson-Johannes A, Williams A, et al. ONYX-015, an E1Bgene-attenuated adenovirus, causes tumor-specific cytolysis and antitu-moral efficacy that can be augmented by standard chemotherapeuticagents. Nat Med. 1997;3:639–645.

28. Lowe S. Progress of the smart bomb cancer virus. Nat Med. 1997;3:606–608.

29. Neshat K, Sanchez CA, Galipeau PC, et al. p53 mutations in Barrett’sadenocarcinoma and high-grade dysplasia. Gastroenterology. 1994;106:1589–1595.

30. Kruyt FA, Curiel DT. Toward a new generation of conditionally repli-cating adenoviruses: pairing tumor selectivity with maximal oncolysis.Hum Gene Ther. 2002;13:485–495.



DiscussionDR. J.R. IZBICKI: I would like to thank the Association

for the privilege to discuss this very fine paper. You all knowthat gene therapy, especially esophageal cancer, is a hot topic.One of the main obstacles is that in gene therapy with viraltransduction you can only reach a certain percentage of tumorcells. And this is probably the main reason for the therapeuticfailures which we have seen in the past and I think that Dr.Buskens and her group have to be congratulated to tackle thisissue.

I have some questions.The first is according the life span of the cell lines:

when you consider the life span of the primary cell lines, thatare typically reduced in contrast with the survival of theestablished cell lines, because these established cell lines areoften genetically altered. What were the results and evalua-tion of the infection rate and death rate of the primary celllines? In our studies, we have had problems with recurrentearly infections of the primary cell cultures.

The second question: what are the rates of targetingnormal cells in your model and have you done some studiesabout the effect on these accidentally targeted normal cells?

The last question: it is not possible to infect 100% ofthese carcinoma cells with this retargeted adenovirus, whichseems to be crucial for the effectiveness of cancer genetherapy? What can we initially do to improve this?

DR. C.J. BUSKENS: Thank you for your questions. Yeswe did encounter some difficulties setting up the primary cellcultures. Sometimes, the cultures had to be terminated due tofungal overgrowth. This happened frequently when we usedresection material from patients with an obstructing tumor ora candida esophagitis. If this was not the case, it was possibleto maintain the cells in culture for several days. It was alsotechnically possible to trypsinize the cells after a few daysand culture them for several passages. However, then newproblems developed with fibroblast overgrowth, and since theaim of this study was to analyze gene transfer in primary cellsresembling the in vivo situation, we used for these experi-ments cells that were only cultured in vitro for 48 hours.

Your last question is very interesting. Even in cell lineswith a population of monoclonal cells, it was never possibleto transduce 100% of the cells. These experiments demon-strated that in heterogeneous primary cells the infectionpercentages were even lower, so indeed this is a problem. Forcurative gene therapy, it is essential to infect all tumor cells;therefore, it will definitely be necessary to either use areplication deficient adenoviral virus with a bystander effecton surrounding cells or a replication competent vector. How-ever, with the results of the present vectors, it is more realisticthat the current gene therapy for carcinomas will be used as

a (neo)adjuvant therapy or in a palliative setting. Theseexperiments demonstrate that an RGD-retargeted vector in-creases the transduction efficiency dramatically; therefore,we feel that using this virus will bring us one step further inoptimizing cancer gene therapy for clinical purposes.

DR. M. REED: Could you comment on the distribution ofthe RGD receptor in more normal tissues. You just looked atnormal epithelia, but it is expressed in vascular endotheliumas well; and if you were administering it systemically, thatwould be crucial in terms of targeting.

DR. C.J. BUSKENS: We did analyze the presence of theCAR receptor and integrins in esophageal carcinoma cellsand normal squamous epithelium of the esophagus, but not invascular endothelium or other normal epithelium. However,if you want to administer the adenoviral vectors systemically,the RGD-retargeted virus will not be selective enough. Thepurpose of the inserted tripeptide is predominantly to increasethe transduction efficiency. For a selective transduction oftumor cells, another targeting motif has to be used. Anexample is to target the vector with specific ligands (like abi-specific antibody to EpCAM) so that the virus will onlyinteract with receptors that are exclusively present on thesurface of the carcinoma cells. Another possibility to achievetumor-specific gene expression is with the use of tumor-specific promoters. Hereby, the virus is under the control ofregulatory regions from proteins only expressed in carcinomacells and not in normal epithelium (like cyclooxygenase-2),so that the therapeutic gene will not be transcribed in normalcells.

DR. P.A. CLAVIEN: Congratulations for your very nicestudy. I think what we would like is a little bit more thera-peutic or clinical aspects of what you have presented here.Your strategy to use a genetically retargeted adenovirus withmodified fiber knobs is a sophisticated approach to make viralentry more specific for tumor cells. My first question: is youradenovirus a replicating or nonreplicating vector and what arethe genes of interest? There is an adenovirus vector thatselectively replicates in p53 mutated cells. Could you com-ment on your plans to use your modified virus for genetherapy? The second question is directed to the route ofapplication. I am not sure if you plan to give the virus directlyinto the tumor, or do you plan to give it systematically? Couldyou speculate a little bit here what would be your further testthat we can see this application in patients and the effective-ness in killing tumor cells.

DR. C.J. BUSKENS: These are very interesting questions.I think, as I said before, that it is too early for systemicadministration of adenoviral vectors due to the hepatotropismof the virus and immune responses of the human body to the



adenovirus type 5. So far, the clinical application of cancergene therapy is limited to local administration intratumorally.This treatment with local injection of vectors into the tumoris preeminently suitable for the esophagus since this organ iseasily accessible via endoscopy. However, even with thisstrategy, it remains important to transduce only carcinomacells since normal mucosal cells neighboring the tumor arealso at risk with the procedure and to avoid complicationsresulting from leakage of viruses to other cells in the body.Therefore, our therapeutic genes of interest are predomi-nantly the conditionally replicating viruses like the ONYX-015 virus, which only replicates in p53-mutated cells. Cur-rently, we are also investigating a transcriptionally targeted

adenoviral vector that will only be transcribed in cyclooxy-genase-2-positive cells, since we demonstrated in a previousstudy that the cyclooxygenase-2 expression in esophagealcarcinoma cells is very high, whereas it is not expressed innormal squamous epithelial cells of the esophagus.

The last thing we are currently investigating to improvecancer gene therapy for clinical use is the use of otherserotypes of adenoviral vectors (like serotypes 16 and 41).Most adults have been exposed to the adenovirus type 2 and5; therefore, the use of other adenoviruses might prevent thesystemic immune response. This would bring us one stepcloser to the desirable systemic administration of the adeno-viral vector for curative cancer treatment.



Documents

A Genetically Retargeted Adenoviral Vector Enhances Viral Transduction in Esophageal Carcinoma Cell Lines and Primary Cultured Esophageal Resection Specimens