[CANCER RESEARCH 58. 3873-3882. Scplcmber I. 1998]

Superior Cytotoxicity with Ganciclovir Compared with Acyclovir and 1-/3-D-Arabinofuranosylthymine in Herpes Simplex Virus-ThymidineKinase-expressing Cells: A Novel Paradigm for Cell Killing1

Laura Z. Rubsatn, Beverly L. Davidson, and Donna S. Shewach2

Department of Pharmacology. University of Michigan Medical Center, Ann Arbor. Michigan 48109-0504 ¡LZ. R., D. S. S.]. ami Department of Internal Medicine ¡R.L D.I,

University of Iowa, Iowa City, Iowa 52242

ABSTRACT

Enzyme-prodrug therapy using ganciclovir and herpes simplex virus-thymidine kinase (HSV-TK) has demonstrated excellent antitumor activ

ity in many different types of malignant cells. Previously, we noted thatganciclovir was substantially more cytotoxic than other HSV-TK sub

strates. Therefore, we embarked on a study to determine the basis for thesuperior cytotoxicity of ganciciovir. In I 25Ilk human glioblastoma cellsthat stably express HSV-TK, ganciclovir elicited a >4 log cell kill insteadof the <1.S log cell kill mediated by two other HSV-TK substrates,1-ß-D-arabinofuranosylthymine (araT) and acyclovir. Study of the metabolism of these drugs demonstrated that acyclovir was poorly phosphoryl-

ated to its active triphosphate with DNA incorporation below the limit ofdetection, which may explain the < 1 log cell kill in these cells. Lower levelsof ganciclovir triphosphate accumulated compared with araT triphosphate (araTTP) under conditions that induced al log cell kill (67 versus1235 pmol/107 cells, respectively), and the half-life for the triphosphate of

ganciclovir was shorter than that of araT (terminal half-lives of 15 and

41 h, respectively). Incorporation of ganciclovir monophosphate into DNAwas less than that of araT monophosphate, and both analogues wereretained in DNA for a48 h. Thus, the superior cytotoxicity of ganciclovirwas not due to enhanced metabolism to active forms. Highly cytotoxicconcentrations of ganciclovir produced only weak inhibition of DNAsynthesis. This allowed cells to proceed through S and ( ;, M phases duringand after drug exposure, resulting in a doubling of cell number by 48 hafter drug washout. As they attempted to progress through the cell cyclea second time, ganciclovir-treated cells accumulated in early S-phase and

remained there until cell death, suggesting that ganciclovir incorporationin the DNA template was important for cytotoxicity. In contrast, stronginhibition of DNA synthesis by araTTP prevented cells from traversingthe cell cycle for at least 12 h after drug washout, when the activemetabolite was largely degraded. araT-treated cells were unable to divide

for at least 72 h after drug exposure, at which point the surviving cellsdisplayed a normal cell cycle distribution pattern. Based on the resultspresented here, we propose a novel paradigm in which the ability ofganciclovir to incorporate into DNA without inhibiting progressionthrough S-phase, combined with high cytotoxicity for incorporated gan

ciclovir monophosphate, produces multilog cytotoxicity.

INTRODUCTION

Tumor cells that have been engineered to express HSV-TK3 can

activate antiviral drugs such as ganciclovir to a toxic metabolite (1).Many studies have demonstrated that this strategy results in cytotoxicity for a variety of tumor cell types in vitro and has resulted in

Received 2/4/98: accepted 7/1/98.The costs of publication of this article were defrayed in part by the payment of page-

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

1This work was supported in part by Grants CA46452, CA7658I. and Training Grant

GM-07767 from the NIH.2 To whom requests for reprints should be addressed, at 4713 Upjohn Center. Depart

ment of Pharmacology. University of Michigan Medical Center. 1310 East Catherine. AnnArbor. Ml 48109-0504. Phone: (313)763-5810; Fax: (313)763-3438: E-mail:dshewach@umich.edu.

1The abbreviations used are: HSV-TK. herpes simplex virus type I thymidine kinase;araT, l-ß-n-arabinofuranosyllhymine: araTMP. l-|3-n-arabinofuranosylthymine 5'-mono-phosphate; araTTP. l-/3-D-arabinofuranosylthymine 5'-lriphosphale: GCVMP. ganciclo

vir monophosphate: GCVTP, ganciclovir triphosphate.

dramatic tumor regressions in several animal models (2-10). Thesepromising preclinical results have prompted clinical trials of HSV-TK/ganciclovir gene therapy in patients with end-stage cancers in

cluding brain tumors (11). Although this novel approach has thepotential for high impact in cancer treatment, the mechanism by whichHSV-TK/ganciclovir elicits this potent antitumor response is not wellunderstood. Cells expressing the HSV-TK gene are able to phospho-

rylate ganciclovir, an acyclic deoxyguanosine analogue, to a toxictriphosphate metabolite that is thought to interfere with cellular DNAsynthesis (12-15). Accordingly, other nucleoside analogues selectively phosphorylated by HSV-TK, such as araT and acyclovir, shouldalso elicit cytotoxicity in this model through a similar DNA-directedmechanism (16-18). However, previous results from our laboratory

demonstrated that ganciclovir was substantially more cytotoxic thaneither araT or acyclovir in C6 rat glioma cells that expressed HSV-TK

(19). Most impressive was that ganciclovir elicited a >4 log cell killcompared with s 1.5 log cell kill mediated by the other analoguesafter a 24-h incubation. Other nucleoside analogues, such as difluoro-deoxycytidine (20), typically produce only a 1- to 2-log decrease in

viability in vitro under similar conditions. Thus, the ability of ganciclovir to produce multilog cell killing is unusual for an antimetaboliteand likely contributes to the efficacy of HSV-TK/ganciclovir in ani

mal tumor models.Although a wealth of reports exist in the literature demonstrating

the antitumor activity of HSV-TK/ganciclovir, information on the

metabolism and cytotoxic mechanism is largely lacking. Mechanisticstudies with other nucleoside analogues, such as l -ß-D-arabinofurano-sylcytosine (21-24), have resulted in improved clinical protocols that

have optimized response and decreased toxicity (24, 25). These studies have demonstrated that for nucleoside analogues, the ability oftumor cells to retain the active triphosphate metabolite and incorporate it into DNA determines the cytotoxic potential of the parent drug.With the superior cytotoxicity of ganciclovir compared with araT andacyclovir, we hypothesized that the triphosphate derivative of ganciclovir would achieve higher concentrations, be retained for longerperiods of time, or be incorporated in DNA to a greater extent.Therefore, we initiated a study of the metabolism and cell cycleeffects of these three agents in a single cell line that stably expressesHSV-TK in an effort to elucidate the mechanism that may confer

multilog cytotoxicity. The results demonstrated that enhanced metabolism did not account for the superior cytotoxicity of ganciclovir.However, the drugs differed markedly with respect to progressionthrough the cell cycle. Based on these results, we propose a novelparadigm for the multilog cell killing observed with ganciclovir.

MATERIALS AND METHODS

Chemicals. [8-'H]Ganciclovir (12.4 Ci/mmol). [methyl-^rKN^lhymine-l-ß-D-arabinofuranoside (2.9 Ci/mmol). [8-'H]acyclovir (15.0 Ci/mmol), and|5-'H]uridine (20.0 Ci/mmol) were obtained from Moravek Biochemicals. Inc.(Brea, CA). [nit'f/iv/-"'H]Thymidine (64 Ci/mmol) was purchased from 1CN

Biomedicals. Inc. (Irvine, CA). Ganc'iclovir (Cytovene) was a generous gift

from Syntex (Palo Alto. CA). Propidium iodide. thymine-l-/3-D-arabino-

3873

on April 3, 2020. © 1998 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

MULTIUSO CYTOTOXICITY MEDIATED BY GANCICLOVIR

t'uranoside, acyclovir, 9-erythro-(2-hydroxy-3-nonyl) adenine, and other nu-

clcoside and nucleotide standards were purchased from the Sigma ChemicalCo. (St. Louis, MO). HPLC grade ammonium acetate and ammonium phosphate were purchased from J. T. Baker, Inc. (Phillipsburg, NJ). Ammoniumphosphate was also obtained from Fisher Scientific (Pittsburgh, PA). All otherchemicals were of the highest purity available.

Enzymes. RNase A (EC 3.1.27.5; 50 units/mg), spleen phosphodiesterase(EC 3.1.16.1: 2 units/mg). alkaline phosphatase from calf intestine (EC 3.1.3.1;2000 units/mg), proteinase K (EC 3.4.21.14; 20 units/mg), and nuclease fromStaphylncoccus aureus (micrococcal nuclease: EC 3.1.31.1: 15.000 units/mg)were obtained from Boehringer Mannheim (Indianapolis, IN). RNase Tl (EC3.1.27.3; 4618 units/mg) was obtained from Life Technologies, Inc. (GrandIsland. NY).

Generation of Stably Transduced U251 Cell Lines. U251 human glio-

blastoma cells were maintained in exponential growth in RPMI 1640 supplemented with 10% calf serum and 2 ITIMglutamine (Life Technologies, Inc.). Togenerate U251 clonal cell sublines that stably expressed the HSV-TK (U25ltk)or /3-galactosidase (U251ßg) cDNA, cells were transduced with retrovirus

vectors that use the retrovirus long terminal repeat sequence as a promoter andcontain the cDNA for neomycin resistance. Transduced cells were selectedwith 400 fig/ml G418: individual colonies were expanded and subsequentlymaintained in medium containing 200 /ig/ml G418. Expression of HSV-TK

was confirmed by assaying cell lysates for ganciclovir phosphorylation (19),and ß-galactosidase expression was verified by staining with the substrate5-bromo-4-chloro-3-indolyl-ß-D-galactopyranoside.

Cytotoxicity Studies. Cytotoxicity was assessed in the parental U251.U251ßg(ß-galactosidase-expressing). and U251tk (expressing HSV-TK) cell

lines by a standard colony formation assay as described previously (19).Exponentially growing cells were treated with drug for up to 24 h, followed byplating 10-50 viable cells/well. Cells were allowed to grow for 7-9 days, at

which time they were stained with crystal violet, and colonies of at least 50cells were enumerated.

Analysis of Ganciclovir, AraT, and Acyclovir Metabolism. After incubation with radiolabeled ganciclovir. araT, or acyclovir, cells were harvested,and nucleotides were extracted with perchloric acid and neutralized as described (20). The phosphorylated derivatives of ganciclovir and araT wereseparated from endogenous nucleotides and quantitated by strong anión exchange HPLC using a Waters (Milford. MA) gradient system composed of twoModel 501 pumps, a U6K injector module, and a Model 996 photodiode arraydetector, and the system was controlled by Millenium 2010 software. Beforeinjection, each sample was spun at 12.(XX)X g for 5 min to remove paniculatematter, and the pH was adjusted to 2.8 to match the starting elution buffer pH.Samples were loaded onto a 5 ¡IMPartisphere 4.6 X 250-mm SAX column

(Whatman, Hillsboro, OR), and nucleoside triphosphates were eluted with alinear gradient of ammonium phosphate buffer ranging from 0.15 M. pH 2.8, to0.6 M (pH 3.7 or pH 2.8). To separate nucleoside mono-, di-, and triphosphates.

ammonium phosphate buffer ranged from 0.01 (pH 2.8) to 0.7 M (pH 3.7 or2.8). Analogue nucleotides were resolved from the endogenous nucleotides bythese procedures. Nucleotides were identified based on their UV spectra overthe range of 200-355 nm and coelution with authentic standards. Cellular

nucleotides were quantitated by comparison of their peak areas with that of aknown amount of the appropriate standard at wavelengths 254 and 281.Fractions containing radiolabeled ganciclovir and araT nucleotides were collected and quantitated based on the known specific activity of the parenttritiated nucleoside. To confirm the identity of radiolabeled ganciclovir andaraT nucleotides. samples were degraded to nucleosides with alkaline phosphatase and analyzed by reversed phase HPLC. Greater then 95% of totaltritium comigrated with authentic standards for ganciclovir or araT.

DNA Incorporation Studies. To isolate DNA from cells, we initiallyextracted nucleic acids with phenol-chloroform, followed by digestion of RNA

with RNase A according to standard procedures (26). However, HPLC analysisof the digested DNA demonstrated that with these methods, at least 10% of theRNA remained in the DNA preparation. Therefore, we developed an improvedprocedure to degrade the RNA completely, which required the use of RNaseTl in addition to RNase A. as we have recently described in detail (27). Foranalysis of the amount of analogue nucleotide incorporated into DNA, cellswere incubated with tritiated ganciclovir, araT. or acyclovir, and harvested;DNA was isolated according to our improved procedure. The purified DNApreparations were digested to internally incorporated 3'-monophosphate nu

cleotides and terminal nucleosides with micrococcal nuclease and spleenphosphodiesterase as described previously (27). The products were analyzedby reversed-phase HPLC. DNA digests were loaded onto a Partisphere C18

column (Whatman, Hillsboro. OR), and samples were eluted with 0.1 Mammonium acetate (pH 5.4) isocratically for 8 min, followed by a lineargradient that added methanol/0. l M ammonium acetate (pH 5.4) at a rate of0.9% per minute for 24 min. One-min fractions were collected and assayed for

radioactivity. The nucleotides and nucleosides were quantitated by comparisonof their peak areas with that of a known amount of the appropriate standard atwavelengths of 254 and 281 nm. The 3'-monophosphate derivatives of gan

ciclovir and araT were identified and quantitated after digestion to nucleosides.As expected, ganciclovir and araT incorporation in nucleic acids was

associated exclusively with DNA. Therefore, in some studies the amount ofdrug incorporated into DNA was determined from the amount of tritium in theacid-insoluble fraction of perchloric acid cell extracts.

Measurement of DNA Synthesis. To measure the effect of ganciclovir andaraT on DNA synthesis, cells were treated with [mi>r/iv/-3H]thymidine (0.016

(XM)for the last 10 min of each drug incubation. After cellular nucleotides wereextracted with perchloric acid, the acid-insoluble fraction was washed once inice-cold 0.4 N perchloric acid and assayed for radioactivity. The acid-soluble

nucleotides were separated and quantitated by HPLC as described above todetermine the specific activity of the ['H]dTTP precursor pool. The incorpo

ration of tritium into DNA was then calculated as cpm incorporated divided bycpm/nmol ['HldTTP. Values are expressed as a percentage of the control (no

drug) values.Cell Growth and Cell Cycle Analysis. The effect of ganciclovir and araT

on cell growth and cell cycle distribution was measured periodically during a24-h drug incubation and for 3 days after drug removal. At each time point,cells were harvested, counted, and fixed by the addition of ice-cold 70%

ethanol. Fixed cells were washed once in PBS and then incubated with 40fjLj>/m\RNase A and 18 /ig/ml propidium iodide for at least 2 h at 4°Cto stain

the DNA. In some experiments, trout erythrocytes were included as an internalstandard for (low cytometry. DNA content was determined using a CoulterEPICS C flow cytometer (Hialeah. FL) measuring propidium iodide fluorescence at 630 nm. Cell cycle distribution was analyzed using the MultiCyclecomputer program (Phoenix Flow Systems, San Diego. CA).

RESULTS

Cytotoxicity of Ganciclovir, AraT, and Acyclovir. The effects ofganciclovir. araT, and acyclovir were evaluated on the clonogenicsurvival of U251 cells that were transduced to stably express HSV-TK(U251tk). U251 cells expressing the ß-galactosidase gene (U251ßg)

were used as controls for retroviral transduction. All three compoundswere more cytotoxic to cells that expressed HSV-TK (U251tk) than incells lacking HSV-TK (U251ßg;Fig. 1), consistent with reports that

o.ooi0.01 0.1 l 10 100

\iM Drug1000

Fig. 1. Comparison of ganciclovir, araT. and acyclovir cytoloxicity in U25ltk cells.U251tk (cinseJ symhuls) and U251ßg(O) cells were incubated wilh ganciclovir (•,O),araT (•).or acyclovir (A) for 24 h and then assayed for clonogenic survival. Valuesrepresent the average ot"triplicate determinations; barx, SE. AraT and acyclovir ( 100 /AM)

did noi decrease the survival of U25lßgcells (dala noi shownl.

3874

on April 3, 2020. © 1998 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

MULTILOG CYTOTOXICITY MEDIATED BY GANCICLOVIR

'Eaco

¡eU 0.1 -

0.01O 4 8 12 16 20 24Hr after Ganciclovir Addition

Fig. 2. Effect of length of incubation on cytotoxicity of ganciclovir. U251 tk cells wereincubated with ganciclovir at concentrations of 0.1 (•),1.0(*), 10 (•),or 100 (A) /¿Mfor the indicated time periods. Survival was determined by a colony formation assay.

these analogues are selectively phosphorylated by HSV-TK (14-18).During a 24-h incubation, ganciclovir (IC5()= 0.12 /AM)was substantially more cytotoxic than either araT (IC5()= 11 /AM)or acyclovir(IC50 = 73 /AM)in U251tk cells (Fig. 1), similar to previous studies inC6 rat glioma cells that expressed HSV-TK (19). Most impressive wasthat ganciclovir elicited a >4 log cell kill at 3 /AM,whereas 1000 /AMaraT or acyclovir resulted in s 1.5 log cell kill. This profound decrease in cell survival was observed during shorter drug incubationperiods as well, where a 2-h exposure to 10 or 100 /AMganciclovirdecreased cell survival by 2 and 4 logs, respectively (Fig. 2). Ganciclovir was also more toxic than acyclovir or araT in U251/3g cells,although these cells were 1000-fold less sensitive to ganciclovir thanU251tk cells (not shown).

Phosphorylation and Retention of Ganciclovir, AraT, and Acyclovir. Initially we hypothesized that the multilog cytotoxicity ofganciclovir was due to accumulation of higher levels of phosphorylated ganciclovir compared with the less effective araT. Therefore, wecompared the amounts of analogue mono-, di- and triphosphateformed from each nucleoside after a 24-h incubation. With 0.1 /AMganciclovir (25% cell survival) or 1 /AMaraT (90% cell survival), theamount of GCVTP that accumulated was at least 20-fold lower thanaraTTP over the 24-h incubation period (Fig. 3). AraTTP achieved itspeak value (1588 pmol/107 cells) within 40 min of drug addition andmaintained that level over the 24-h time course. GCVTP accumulatedlinearly for 4 h and then accumulated more slowly over the next 20 hto 44 pmol/107 cells. This low level of GCVTP is similar to the

concentration of the cellular deoxynucleotides. For both ganciclovirand araT, the triphosphate was the major metabolite (70-95% of totalradioactivity), compared with the di- and monophosphate forms.Ganciclovir diphosphate represented a slightly higher proportion of

the ganciclovir nucleotides compared with araT diphosphate, whichmay reflect better phosphorylation of araT diphosphate in mammaliancells. No other radiolabeled metabolites were observed for gancicloviror araT.

Analogue triphosphate formation was also evaluated at higherconcentrations of ganciclovir and araT over a 4-h incubation period.At concentrations ranging from 0.1 to 10 /AM,the amount of araTTPexceeded that of GCVTP by 2-20-fold (Fig. 4). Only at a concentration of 100 LIMdid the amount of GCVTP exceed that of araTTP.However, this concentration is 20-fold higher than the amount ofganciclovir that reduced cell survival by >99%. Similar studies conducted in cells treated with 10 or 100 /AMacyclovir for 24 h resultedin acyclovir triphosphate levels that were 7-30-fold lower thanGCVTP or araTTP levels after 4 h (Fig. 4). Thus, the weak cytotoxicity of acyclovir may be due to low triphosphate accumulation.

Because the superior cytotoxicity of ganciclovir could not be explained by higher accumulation of its triphosphate, we hypothesizedthat lengthy retention of GCVTP compared with araTTP accountedfor the multilog cytotoxicity. Cells were incubated for 4 h with 1or 10/AMganciclovir or 10 or 100 /AMaraT to measure the rate of degradation of the triphosphates (Fig. 5). Elimination of ganciclovirtriphosphate appeared monophasic with iln values of 11 and 15 hafter incubation with 1 and 10 /AMganciclovir, respectively. Theelimination of araTTP was biphasic at 10 and 100 /AM,with tl/2 valuesof 1.4 and 1.9 h, and tl/2l} values of 35 and 41 h, respectively.Although the initial elimination of araTTP was more rapid than that ofGCVTP, the terminal half-life of araTTP was substantially longer andresulted in 5-fold higher levels of araTTP 2 days after drug washout(Fig. 5).

0.1 1 10 100 1000Drug Concentration (\iM)

Fig. 4. Accumulation of triphosphates of ganciclovir. araT. and acyclovir. U25ltk cellswere incubated with tritium-labeled ganciclovir (•)or araT (•)for 4 h or acyclovir (A)for 24 h. Nucleotides were analyzed by HPLC. Values represent the average of duplicatedeterminations: bars. SD.

Fig. 3. Phosphorylation of ganciclovir and araT. U251tkcells were incubated with ['HJganciclovir (O.I /IM: A) or['HJAraT ( 1 /XM;Bl for 24 h. Nucleotides were extracted

with perchloric acid, neutralized, and analyzed by HPLC.Values represent the average of duplicate determinations;bars, SD.

B

0.010

GCVTP

GCVDP

GCVMP

1(MMM)E"

AraTTP

10 "—'-1

12 18 24 12 18 24

Hr after Ganciclovir Addition

3875

Hr after AraT Addition

on April 3, 2020. © 1998 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

MULTILOG CYTOTOXICITY MEDIATED BY GANC1CLOVIR

Fig. 5. Comparison of GCVTP and araTTP elimination. U251tkcells were treated with [3H|GCV at concentrations of 1 (•)or 10 ¿LM<•)or (3H]araT at concentrations of 10 (•)or 100 (•)/IM for 4 h.

Alter drug washout. GCVTP and araTTP were extracted from cellsand analy/cd by HPLC. Values represent the average of duplicatedclerminalions; htir\, SD.

MMIO KMHHI

10 T

0

2«

1000

12 24 36 48

Hr After Ganciclovir Washout

O 12 24 36 48

Hr After AraT Washout

Incorporation of Ganciclovir, AraT, and Acyclovir into DNA.All three analogues have been reported to incorporate into cellularDNA (13-15, 17. 18. 28). Because DNA incorporation has been

correlated with cell death for several other nucleoside analogues (29),the extent of incorporation was studied with ganciclovir, araT, andacyclovir. After incubation with tritiated ganciclovir or araT. >90%of the radiolabel was recovered in the DNA fraction, similar to therecovery of ('H]thymidine, with background levels of radioactivity

associated with RNA. These results demonstrate that GCVMP andaraTMP incorporate exclusively into DNA in these cells. Similarexperiments conducted with radiolabeled acyclovir resulted in unde-

tectable levels of tritium incorporated into nucleic acid, which likelyreflects the low levels of acyclovir triphosphate that accumulated.Because more sensitive analysis of acyclovir incorporation into DNArequired prohibitively expensive levels of ['Hjacyclovir, these studies

were not pursued further.To determine whether GCVMP and araTMP incorporated into

DNA at internuclcotide or terminal positions. DNA was degraded to3'-monophosphates and terminal nucleosides with the successive ac

tions of micrococcal nuclcase and spleen phosphodiesterase as described previously (27). Analysis of DNA isolated from cells treatedwith 0. l /IM ganciclovir for 6 h revealed that less than 5% of the drugwas incorporated as the terminal nucleoside, whereas >90% appearedto elute at a position consistent with the ganciclovir derivative of a3'-monophosphate (Fig. 6. A and B). To confirm that the radiolabeled

compound was ganciclovir monophosphate, for which a standard wasnot available, samples were degraded to ganciclovir with alkalinephosphatase and analy/cd by reversed-phase HPLC. As shown in Fig.

6, C and D, >95% of total tritium comigrated with internal standardsfor ganciclovir. Similar studies confirmed the identity of araT 3'-

monophosphate. Increasing the dose of ganciclovir to l /IM or theincubation time up to 48 h did not increase the percentage of terminally incorporated drug (data not shown). Similar studies conducted incells treated with 10 and 1(X)/IM araT for 24 h demonstrated that>90% of the araTMP also incorporated into DNA at internal positions(data not shown).

In some ganciclovir-treatcd samples, a small level (5-10%) oftritium elutcd from the column 1-10 min after the internal standard for

ganciclovir (Fig. 6. A and C). The amount of these products decreasedas the amount of micrococcal nuclease was increased. Also, theamount of these products varied depending upon the commercialsource of the micrococcal nuclease. This suggested that the late-eluting products were due to the incomplete digestion of small oligo-

nucleotidcs containing ganciclovir, similar to results reported byCheng el al. (13). In DNA digests from cells labeled with tritiatedaraT or thymidine. all of the radioactivity could be accounted for bythe amount of the corresponding 3'-monophosphate and nucleoside

(data not shown).

3876

These studies demonstrated that both ganciclovir and araT areincorporated at internucleotide positions in cellular DNA. Next wewished to determine whether the superior cytotoxicity of ganciclovirwas due to higher DNA incorporation during a 24-h drug incubation.

4000

30003'GCVMP

CPM2000

1000

B

AU

4000

3000

CPM2000

1000

GCV

10 15 20 25 30 35 40

AU

«* .u™

20Tim«(min)

Fig. 6. HPLC analysis of DNA digests from U25ltk cells treated with ganciclovir.U25ltk cells were treated with O.I JXM['Hjganciclovir lor 6 h, followed by isolation ofDNA and degradation to 3'-monophosphates and terminal nucleosides with micrococcal

nuclease and spleen phosphodiesterase (A and B). To verify the identity of the peakidentified as ganciclovir 3'-monophosphate, an aliquot of the digest was degraded further

to nucleosides with alkaline phosphatase (C and D). Both samples were collected andassayed in I-min fractions (A and O, and ahsorbance of the nucleotides and nucleosideswas measured at 254 nm (B and /)). One nmol of GCV was included in the digests as aninternal standard. Shown are representative chromatograms from more than 12 experiments.

on April 3, 2020. © 1998 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

MULTILOG CYTOTOXICITY MEDIATED BY GANCICLOVIR

Table 1 incorporation of nucleotide analogues into DNA

U251tk cells were incubated with ganciclovir or araT for 24 h. DNA from cells wasextracted and digested to 3'-monophosphates and terminal nucleosides as described in"Materials and Methods." Greater than 90% of the incorporated nucleotides were present

in intemucleotide linkages. Results are expressed as the total amount of analogue in DNA(sum of 3'-monophosphate and nucleoside incorporation). Each value represents the

average ±SD of duplicate determinations.

Length ofConcentration incubation

Drug (UM)(h)Ganciclovir

0.1612241.0

61224AraT

100 24pmol

nucleotide per100 pig of DNA(average ±SD)1.042.999.304.0415.1947.950.660.460.401.513.4835.16261.01

15.98

10000

1000r-o

Oja"o

iHr After Drug Washout

Fig. 7. Retention of nucleotide analogues in DNA. U251tk cells were incubated withtritium-labeled ganciclovir (0.1 U.M) (•)or araT (IO (ÌM)(•)for 4 h, followed byperchloric acid precipitation of the nucleic acid fraction. The acid-insoluble pellet wassolubilized, and tritium was measured by scintillation spectromelry. The results representthe average of duplicate determinations from a representative experiment: hars, SD.

DNA incorporation increased for both araT and ganciclovir throughout the incubation period. As indicated in Table 1, the amount ofaraTMP exceeded that of GCVMP by a factor of at least 5-fold aftera 24-h incubation, although survival decreased by more than 3 logs

with 1 /AMganciclovir compared with <1 log with 100 /AMaraT. Thus,despite its superior incorporation into DNA, araT was not as effectiveas ganciclovir at decreasing cell survival.

Although GCVMP was incorporated into DNA to a lesser extentthan araTMP, it seemed possible that lengthy retention of GCVMPmight account for its multilog cytotoxicity. DNA retention ofGCVMP and araTMP was assessed for 24 h after a 4-h incubation

with 0.1 /AMganciclovir or 10 /AMaraT (Fig. 7). After drug washout,the amount of each analogue in DNA increased for 4 h. No significantdecrease in amount of araTMP or GCVMP in DNA was observed forat least 24 h after drug washout. Small decreases in GCVMP in DNAat 48 h after drug washout reflected larger numbers of cells in thosecultures, because the cell number nearly doubled over this timeperiod. Thus, these analogues appeared similar in that both were wellretained in DNA in the absence of exogenous drug.

Effects on DNA Synthesis and Cell Cycle Progression. Duringthe course of these studies, we noted that although ganciclovir wasmuch more cytotoxic, it was not a potent inhibitor of cell growth. Wehypothesized that this was due to a weak effect of ganciclovir on DNAsynthesis. Therefore, the ability of each drug to inhibit DNA synthesiswas measured in intact cells using [3H]thymidine incorporation into

cellular DNA after a 1-h drug exposure. Results were corrected for thespecific activity of [3H]dTTP in the precursor pool. Treatment with 1

/AMganciclovir inhibited DNA synthesis by only 50%, whereas amuch less cytotoxic treatment with 100 /AMaraT resulted in 90%

inhibition (Fig. 8). Longer incubation with 1 /AMganciclovir for up to24 h did not result in a more complete block (60% maximum inhibition; data not shown). Greater inhibition of DNA synthesis (80%) withganciclovir required concentrations 10-fold (älO JU.M)higher than

necessary to kill 99.99% of the population. This weak inhibition ofDNA synthesis by ganciclovir corresponded to lower levels oftriphosphate compared with araT (Figs. 3 and 4).

To verify that potential competition between the analogues andthymidine for HSV-TK did not interfere with the accuracy of the

DNA synthesis assay, we measured DNA synthesis inhibition usingaphidicolin in U251tk and wild-type U251 cells using the ['Hlthymi-

dine incorporation assay. Aphidicolin blocked DNA synthesis to asimilar extent in both cell lines (data not shown), indicating that thisis an appropriate measure of DNA synthesis in cells that expressHSV-TK.

In view of the weak inhibition of DNA synthesis elicited byganciclovir, we extended these studies to compare the effects of thesedrugs on cell growth and progression through the cell cycle. Theseparameters were measured during a 24-h incubation with I /AMgan

ciclovir or 100 /AMaraT and for 6 days after drug washout (Fig. 9).AraT had a greater inhibitory effect on cell growth compared withGCV during the 24-h drug exposure period (89% versus 55% growthinhibition, respectively). After drug washout, araT-treated cells re

mained growth inhibited for at least 2 days and then increased innumber. Although there was a decrease in cell number between 5 and6 days after araT washout, this trend must have reversed at a later timebecause as much as 24% of the population survived, according to acolony formation assay. The results discussed below also indicate thatthe decrease in araT cell number represented the elimination of deadcells, leaving primarily viable cells in the culture. In cultures treatedwith ganciclovir, initially the cells were able to replicate, and theycompleted one doubling between 12 and 48 h after drug removal.However, the cultures then displayed a continuous decline in cellnumber for the next 4 days, and the majority of the cells lifted off theplate within 5 days after drug removal. This was consistent with thekilling of >99.99% of the cells observed in a colony formation assayconducted at the same time.

In the study displayed in Fig. 9, cell cycle progression was measured by flow cytometry after cells were fixed and the DNA stainedwith propidium iodide. During drug incubation, araT-treated cells

accumulated in late G,/early S phase and appeared to be blocked thereuntil after drug washout (Fig. 10). After drug washout, araT-treated

cells progressed synchronously through S phase and then through

0.01 0.1 1 10 100Drug Concentration (|i \1)

Fig. 8. Effect of ganciclovir and araT on DNA synthesis. U251tk cells were incubatedwith the indicated concentrations of ganciclovir (•)or araT (•)for I h. Each culture waspulsed with [3H]thymidine for the last IO min of the incubation. Cells were then extracted

with perchloric acid, and the amount of tritium in the acid-insoluble material wasmeasured. The ['HJdTTP levels were measured by HPLC to correct for alterations in the

specific activity of this nueleotide. Values represent the average of duplicate determinations; bars. SD.

3877

on April 3, 2020. © 1998 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

MULTILOO CYTOTOXICITY MEDIATED BY GANCICLOVIR

5x10

Control

AraT

+Drug Hr After Drug Washout

Fig. 9. Effect of ganciclovir and araT on cell growth. U251tk cells were incubated withno drug (•),1 /¿Mganciclovir (•).or 100 /¿MaraT (A) for 24 h. at which time thedrug-containing medium was washed out of each culture and replaced with fresh medium.Cells were harvested by trypsinization at the indicated times, and cell number wasmeasured for 6 days after drug washout. Values represent the average of duplicatedeterminations.

G-.-M over a 48-h period (Fig. 11). Cells continued to progress

through the cell cycle, and by 6 days after drug removal, the flowcytogram exhibited a normal distribution. In contrast, ganciclovir-

treated cells were able to progress through mid and late S phase duringthe drug exposure period (Fig. 10). Within 48 h after ganciclovirwashout, cells had completed one cell cycle, at which point the cellnumber had doubled (Fig. 11) and the cell cycle distribution patternappeared similar to that of control cells. However, as the cells attempted to initiate another round of DNA synthesis during the next24 h, all cells accumulated in early to mid S-phase and remained there

until cell death.Induction of Apoptosis. With previous reports indicating that

ganciclovir can induce apoptosis in some cell types (30-32), we

wished to determine whether the difference in cytotoxicity betweenaraT and ganciclovir was reflected in the onset or percentage of cells

Control

Fig. 10. Cell cycle progression in U25ltk cells during ganciclovir oraraT exposure. Cell cycle progression was assayed at three time pointsduring a 24-h ganciclovir (1 JIM)or araT (100 ¿IM)incubation. After theindicated drug incubation periods, cells were harvested, fixed, andstained with propidium iodide for analysis by flow cytometry as described in "Materials and Methods." This set of DNA histograms is

representative of data derived from three similar experiments. Data fromchannels 1-40. the sub-G, fraction representing mainly apoptotic cells.were omitted for clarity and are not included in the cell cycle analysis.

Ganciclovir

AraT

undergoing apoptosis. In the experiment described in Figs. 9-11,apoptotic cells were identified by sub-G, DNA content as analyzed by

flow cytometry. In control cultures, no more than 6% of the cellpopulation was undergoing apoptosis at any time during the study(Fig. 12). Both araT- and ganciclovir-treated cultures contained ap

optotic cells within 48 h after drug removal. The percentage ofapoptotic cells in each culture increased for 4 days after drug removal,when at least 60% of the cells treated with araT or ganciclovir wereundergoing apoptosis. At 6 days after drug removal, the culturesdiverged in which araT-treated cells showed a dramatic decrease inapoptotic cells, whereas ganciclovir-treated cells continued to exhibit>60% apoptotic cells. This corresponded to the time at which a

normal cell cycle distribution pattern was observed with araT,whereas ganciclovir cells remained blocked in early S phase. It wasnoted at 5 days after araT and ganciclovir washout that a substantialnumber of cells were not attached to the flasks. Analysis of thesefloating cells by flow cytometry demonstrated that a90% were apoptotic at 5 days after drug incubation. On day 6, the floating cellsfrom the ganciclovir-treated culture were 90% apoptotic, whereas thearaT-treated culture showed a decrease to only 60% of floating cells

as apoptotic. Taken together, these results demonstrate that cellsexposed to ganciclovir undergo apoptosis for a longer period of timecompared with araT, reflecting the multilog cytotoxicity uniquelyassociated with ganciclovir.

DISCUSSION

Although many groups have reported impressive antitumor activityof ganciclovir in cells transduced to express the HSV-TK gene (1-10),

there are few reports on the cytotoxic mechanism of ganciclovirnucleotides. Such information becomes more important in view of ourrecent report that ganciclovir is a superior cytotoxic agent than otherHSV-TK substrates tested, eliciting a multilog cell kill compared with•¿�^1.5 log cell kill with araT or acyclovir (19). It was expected that the

potent antitumor activity would be related to accumulation of highlevels of ganciclovir nucleotides or their lengthy retention, propertiesshown to be important for clinically effective nucleoside analogues

i ^^ i

r~^*1^^^^^\*f~~\ r"1i^^^^^il**n r~"^*T^^^^^^^"~i

[—^^ i r"7^ i r*^^ i0 200 400 600 800 0 200 400 600 800

DNA Content0 200 400 600 800

12 18 24

Hr After Drug Addition3878

on April 3, 2020. © 1998 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Mri.Tll.Od CYTOTOXICHY MEDIATED BY (ÃŒANCIC1.OVIR

Ganciclovir

AraT

r^^T^^^i^n r^T^i^in rT^i^'n r^T^r^í"1"!

L* .JU iJLl ' l ' ! M ' l l ' ¡ " ! ! l ' l l ' M l ' l ' l

r"*"r^^^T"l~i r'^^^^T*"! Mi^^^^'n r^^^r^^"! r^^^^^T^i r^T^^^in0 200400600800 0 200400600800 0 200400600800 0 200400600800 0 200400600800 0 200400600800

12

DNA Content

24 48 72

Hr After Drug Washout96 144

Fig.medium.

1A1 SVIICI .LSIUg

11. Cell cycle progression in U251lk cells after 24 h of drug exposure. Following the drug exposure described in Fig. 10, drug-containing medium was replaced with fresh

i. and cells were harvested periodically and processed for cell cycle analysis by flow cytometry.

(21, 22, 25). The studies described here demonstrate that ganciclovirdiffered markedly from classic nucleoside analogues in which massive cytotoxicity with ganciclovir was associated with low nucleotidelevels and modest effects on DNA synthesis, suggesting a novelmechanism for multilog cell killing.

Of the three analogues evaluated here, acyclovir was poorly phos-

phorylated, and incorporation into DNA was below the limit ofdetection. Thus, for this analogue, low activation and incorporationinto DNA is consistent with its low cytotoxicity. In contrast, comparison of the cellular metabolism of ganciclovir and araT showed thatGCVTP at levels similar to that of the endogenous deoxynucleotidesresulted in a 5 log cell kill, whereas araTTP levels 300 times higherelicited no more than 90% cytotoxicity. Lengthy retention of lowlevels of ganciclovir nucleotides could not explain the superior cytotoxicity because at all concentrations tested, the terminal half-life of

70

60

50

40

30

20

10

0

o•¿�B.oa.

S?

Il12 24 36 48 72 96 120

Hr After Drug WashoutFig. 12. Apoptosis after drug exposure. U251tk cells were incubated with either no

drug (^). ganciclovir (S). or araT (•)as described in the legend to Fig. 10. Apoptosiswas analyzed as the percentage of cells with sub-G, DNA content as measured by flow

cytometry.

GCVTP was shorter than that of araTTP. Furthermore, there was noevidence that a metabolite other than GCVTP was the cytotoxic agentbecause this nucleotide was the major form of ganciclovir in U251tkcells, and no unidentified metabolites were observed. Taken together,these results demonstrate that the superior cytotoxicity of ganciclovircompared with araT is not due to higher accumulation or a longerhalf-life of its toxic triphosphate form. Rather, GCVTP appears to be

a more potent cytotoxic agent.A comparison of the metabolism of ganciclovir from previous

studies using either murine cells transduced to express HSV-TK (14,33) or herpes virus-infected cells (12, 34, 35) demonstrates that theaccumulation of ganciclovir mono-, di-, and triphosphate derivativeswas similar to that observed in U251tk cells. However, in HSV-

infected Vero cells (12, 34) or primary rabbit kidney cells (35), thehalf-life of GCVTP was longer than that in U251tk cells. Smee et al.(34) showed that 60-70% of total GCVTP was retained in HSV-

infected Vero cells 24 h after drug removal, compared with only 20%retention observed here in the U251 tk cells. These differences may beaccounted for by the different cell types used in these studies, orpossibly that viral infection slows elimination of GCVTP from mammalian cells.

Although nucleotide accumulation was lower from ganciclovircompared with araT, it seemed possible that higher incorporation ofGCVMP into DNA could explain the superior cytotoxicity. However,more araTMP was observed in DNA compared with GCVMP at allconcentrations tested. High incorporation of araTMP into DNA waslikely due to the excellent affinity of replicative DNA polymerases foraraTTP (36) and the fact that araTTP levels exceeded those of theendogenous competitor dTTP by at least 10-fold. In contrast, GCVTPlevels were usually 10-fold lower than those of araTTP and similar to

the amount of the competing dGTP. Greater cytotoxicity with ganciclovir, despite lower incorporation into DNA, demonstrates the potency of the incorporated nucleotide.

The unique cytotoxicity of ganciclovir does not involve DNA chain

3879

on April 3, 2020. © 1998 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

MULTILOG CYTOTOXICITY MEDIATED BY OANCICLOVIR

termination because DNA digestion studies showed that GCVMPincorporated into DNA at internal positions to a similar extent asaruTMP. Other groups have observed internal incorporation ofGCVMP in DNA in studies with virally infected or transformed cells(13. 15). This is consistent with the fact that ganciclovir has ahydroxyl group on its sugar ring analogous to a 3'-hydroxyl on dGMP

that can support further DNA elongation. In contrast, investigatorsdemonstrated that acyclovir, which lacks an equivalent 3'-hydroxyl

group, acts as a chain terminator (17, 37). In cell-free systems,

investigators have reported that GCVMP polymerization onto a synthetic template primer inhibits further DNA extension, either immediately after GCVMP incorporation or after the polymerization of anadditional nucleotidc (38-40). Although the method for measuring

internucleotidc DNA incorporation described here cannot distinguishbetween GCVMP incorporated in the penultimate position on a DNAstrand versus elongation into mature DNA, the cell cycle studiesclearly indicated that the GCVMP-containing DNA was elongated to

full length. In contrast, studies with purified polymerases on a definedtemplate demonstrated that GCVMP is a strong inhibitor of elongationafter the addition of one nucleotide (38, 40). Reardon (38) demonstrated that there was a 7-fold increase in Km values for polymerase a

for addition of the next two nucleotides after GCVMP was incorporated. There are two possible explanations for this apparent discrepancy between cell-free and intact cell results: (a) the strong inhibition

of polymerization beyond one or two nucleotides after GCVMPobserved in cell-free studies was done in a short-term assay system. If

it were possible to extend the incubation time in this system, it mayshow that further elongation was slowed but possible. In intact cells,DNA synthesis was slowed by ganciclovir, and it is possible that thekinetics of elongation were slowed but the extended incubation timeallowed cells to complete DNA replication: and (b) there may becritical factors absent in cell-free experiments but present in intact

cells that allow the polymerases to elongate DNA containingGCVMP. Current models for DNA replication implicate polymeraseS in leading strand synthesis with participation in lagging strandsynthesis after priming by polymerase a and primase (41). Althoughthe specific role lor polymerase €¿�is not clear, it is a highly processivepolymerase, as is polymerase 8. It is possible that in the case that thehigher concentrations of GCVTP can inhibit polymerase 6, its activitymay be assumed by the less sensitive polymerase e to continue leadingand lagging strand synthesis.

The inhibition of ganciclovir-treated cells in early S phase after one

cell division suggested that the presence of GCVMP in the DNAtemplate forms a critical lesion for ganciclovir cytotoxicity. Severalobservations described here indicate that GCVMP can incorporateinto the DNA template: (a) incorporated GCVMP formed internucle-

otide linkages; (b) GCVMP was retained in DNA after the cellnumber had doubled: and (c) most cells divided after cytotoxic ganciclovir treatments. Thust et al. (42) have also hypothesized thatGCVMP incorporation in the DNA template is a critical lesion basedon studies showing that ganciclovir-induced chromosomal aberrationsin Chinese hamster V79-E cells were much greater 24 h after drug

removal (after cells had divided) than during drug treatment. Themechanism by which GCVMP in the template strand may producemultilog cytotoxicity is not clear. It is possible that GCVMP is anunsuitable substrate in the DNA template for replicative polymerases,or alternatively, the acyclic moiety may alter the secondary structureof DNA, interfering with the binding of enzymes involved in DNAreplication. Indeed, it has been demonstrated that ganciclovir inducessignificant distortion in a decamer duplex of DNA (43). It is not clearwhether araTMP is present in the DNA template strand or whetherthis lesion is repaired prior to the next round of DNA synthesis. Thestudies of araTMP retention in DNA lasted for 48 h after drug

exposure, and the cells exposed to araT did not divide during thistime. However, in view of the lengthy retention of araTMP in DNA,it seems likely that this analogue would be present to some extent inthe DNA template. If so, then it would be important to compare theeffects of araTMP and GCVMP in the template on DNA replication.

These studies demonstrated that ganciclovir produced only modestinhibition of DNA synthesis. The two major factors that result ininhibition of DNA synthesis are: (a) direct competition by GCVTPwith dGTP for DNA polymerases; and (b) slowing of DNA elongationafter incorporation of GCVMP (38-40). Regarding direct inhibition

of DNA polymerases, in vitro studies have demonstrated that GCVTPis a selective inhibitor of DNA polymerase ô(K{ = 2 /J.M), with40-70-fold less potent inhibition of DNA polymerases a and e (40).

The studies shown here demonstrate that, even at highly cytotoxicconcentrations, intracellular concentrations of GCVTP ranged fromonly 5-20 /J.Mduring a 24-h incubation period. Because the cellularconcentration of the endogenous competitor, dGTP, is ~4 JU.Min U251

cells, potent inhibition of polymerase ôwould be expected only athigh concentrations and long incubation periods. Regarding the effectof incorporated GCVMP on DNA synthesis, the data presented heresuggest that GCVMP in DNA is not a potent inhibitor of chainelongation in intact cells. Taken together, the weak inhibition of DNApolymerases and poor inhibition of elongation by ganciclovir resultedin weak inhibition of DNA synthesis in intact cells.

The strong inhibition of DNA synthesis observed with agents suchas araTTP may in fact protect cells from cytotoxicity. The dataindicated that the majority of the incorporation of araTMP into DNAoccurred during drug exposure, when cells were blocked at the G,-S

interface, and up to 24 h after drug washout when most cells progressed only into mid-S phase. Therefore, araTMP was incorporated

mainly into early replicating regions of DNA. In contrast, incorporation of GCVMP occurred throughout the drug exposure period and upto 12 h after drug washout, as cells progressed slowly through early,mid, and late S-phase. Thus, GCVMP was incorporated throughout

the genome, whereas araTMP incorporated primarily into early andperhaps some mid-S replicating regions.

Based on these results, we propose a novel paradigm for multilogcell killing by ganciclovir. The data suggest that there are two majorfactors that contribute to the multilog cytotoxicity observed withganciclovir: (a) ability to be incorporated into DNA without stronginhibition of DNA synthesis; and (b) potent cytotoxic effect of theincorporated GCVMP. This then would allow incorporation of thehighly toxic GCVMP into DNA throughout the genome, which wepropose is more cytotoxic than incorporation into early replicatingregions. Clearly, the chemical nature of the analogue plays a rolebecause bromodeoxyuridine, a thymidine analogue, can be incorporated throughout the genome but is not as cytotoxic, likely due to thestructural similarity of this analogue to endogenous thymidine (44).There are several mechanisms by which incorporation throughout thegenome could elicit multilog cytotoxicity. Incorporated GCVMP isknown to cause distortions in the structure of duplex DNA (43).Multiple incorporations may cause greater distortion in the DNAconformation so that, during the subsequent S-phase, DNA synthesis

is initiated but halted rapidly. Alternatively, it is possible thatGCVMP in the template strand is unsuitable as a substrate for DNApolymerases and that when such a lesion is encountered, DNA replication is terminated.

An important finding of these studies was that the ability of nucle-

oside analogues to inhibit cell growth did not predict relative cytotoxicity for the three analogues studied here. Compared with ganciclovir, araT was a superior inhibitor of DNA synthesis and cell growthduring a 24-h drug exposure and for 72 h afterward. By these meas

ures, araT would have been judged the more potent of these agents.

3880

on April 3, 2020. © 1998 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

MULTILOG CYTOTOXICITY MEDIATED BY OANCICLOVIR

10.

However, the colony formation assay demonstrated that only ganci -

clovir elicited a 4 log cell kill at concentrations that produced littlegrowth or DNA synthesis inhibition over a 72-h period. Commonly,

growth inhibition has been used as a predictor of cytotoxicity forganciclovir as well as numerous other antitumor agents. The data here 6'

demonstrate that, at least for ganciclovir, growth inhibition grosslyunderestimates the cytotoxicity of the drug and is an inappropriate 7-

measure of drug efficacy. The weak effect on cell growth at highlycytotoxic concentrations of ganciclovir is not unique to the U251 tk 8.cell line because it has occurred in other cell types that we haveexamined, including rat glioma cells (19) and two colon carcinoma 9cell lines (45). Although we note some variation between cell lines insensitivity to ganciclovir, multilog cell killing has been demonstratedin all cell types we have examined to date. Thus, these studiesdemonstrate that, due to the delayed nature of the cytotoxicity produced by ganciclovir, longer-term survival assays such as colony- "•

forming ability should be used in assessing the antitumor activity for ¡3.this drug.

The potency of GCVTP reported here has important implicationsfor the susceptibility of HSV-TK-negative cells to ganciclovir cyto- 13toxicity when in the presence of HSV-TK-positive cells (termed thebystander effect). A major reason that HSV-TK/ganciclovir therapy

has been successful in reducing tumor size in animal studies is that 14.expression of HSV-TK in as few as 10% of the cells in the tumor is

sufficient to produce complete tumor regression after ganciclovirtreatment (5-7). Although there may be an important immune com- is.ponent that contributes to the efficacy of HSV-TK/ganciclovir therapy

in vivo (46), bystander cell killing has been demonstrated in vitro in [6the absence of immune effects (1, 33, 45, 47-53). Although the

mechanism of this bystander effect in vitro is not entirely understood,it has been proposed that it is due to passage of toxic ganciclovirnucleotides from cells that express HSV-TK expression to bystandercells that lack HSV-TK through gap junctional intercellular commu- '8-

nication (47, 48, 51, 53-55) or engulfment of apoptotic vesicles (56).

Both mechanisms should also facilitate transfer of other nucleotide 19.analogues such as araTTP. However, based on the results describedhere in cells that express HSV-TK, low levels of GCVTP comparedwith araTTP will elicit substantially more cytotoxicity in bystander 20.

cells. Because present gene transfer techniques typically result inslO% transduction of tumors in animal models (11), only minimal 21.transfer of phosphorylated nucleotides to HSV-TK-negative cells ispossible. Consequently, HSV-TK-mediated therapy requires highly 22

potent agents such as GCVTP for effective tumor reduction.In summary, these studies have demonstrated that the superior

cytotoxicity of ganciclovir compared with araT did not result fromenhanced metabolism or higher DNA incorporation. Instead, theseresults provide evidence that incorporation of GCVMP into the DNAtemplate is a highly potent lesion. The multilog cell killing displayeduniquely by ganciclovir occurs through a delayed mechanism thatpermits cells to complete one cell division, after which they appear to 25'

permanently arrest in early S-phase. Based on these findings, theeffect of template incorporated GCVMP on DNA replication merits 26.further investigation.

27.

REFERENCES28.

1. Moolten. F. L. Tumor chemosensitivity conferred by inserted herpes thymidine kinasegenes: paradigm for a prospective cancer control strategy. Cancer Res.. 46: 5276-5281. 1986. 29.

2. Moolten. F. L. Drug sensitivity ("suicide") genes for selective cancer chemotherapy.Cancer Gene Ther.. /.- 279-287, 1994. 30.

3. Moolten, F. L., Wells, J. M.. Heyrnan. R. A., and Evans, R. M. Lymphoma regressioninduced by ganciclovir in mice bearing a herpes virus thymidine kinase transgene. 31.Human Gene Ther.. /.- 125-134. 19W.

4. Ezzeddine. Z. D.. Martuza. R. L.. Platika. D.. Short. M. P.. Malick. A.. Choi. B.. andBreakefield. X. O. Selective killing of glioma cells in culture and in vivo by relrovirus

3881

24.

transfer of the herpes simplex virus thymidine kinase gene. New Biol.. 3: 608-614,1991.Culver, K. W., Ram. Z.. Walbridge, S.. Ishii, H., Oldfield, E. H., and Blaese, R. M.In vivo gene transfer with retroviral vector producer cells for treatment of experimental brain tumors. Science (Washington DC). 256: 1550-1552, 1992.

Ram. Z., Culver, K. W.. Walbridge. S.. Blaese. R. M.. and Oldfield. E. H. In xiiiiretroviral-medialed gene transfer for the treatment of brain tumors in rats. CancerRes., 53: 83-88. 1993.Caruso. M.. Pañis,Y., Gagandeep. S.. Houssin. D.. Salzmann, J-L.. and Klatzmann,

D. Regression of established macroscopic liver métastasesafter in silu transduction ofa suicide gene. Proc. Nati. Acad. Sci. USA. 90: 7024-7028. 1993.

Takamiya. Y.. Short. M. P., Moolten, F. L., Fleet, C. Mineta. T.. Breakefield, X. O.,and Martuza. R. L. An experimental model of retrovirus gene therapy for malignantbrain tumors. J. Neurosurg., 79: 104-110, 1993.Chen. S-H.. Shine. H. D.. Goodman. J. C.. Grossman. R. G., and Woo, S. L. C. Genetherapy for brain tumors: regression of experimental glioma* by adenovirus-medialedgene transfer in vira. Proc. Nati. Acad. Sci. USA. 91: 3054-3057, 1994.O'Malley. B. W.. Jr.. Chen. S-H.. Schwartz. M. R.. and Woo, S. L. C. Adenovirus-

mediated gene therapy for human head and neck squamous cell cancer in a nudemouse model. Cancer Res.. 55: 1080-1085, 1995.

Roth. J. A., and Cristiano. R. J. Gene therapy for cancer: what have we done andwhere are we going? J. Nail. Cancer Insl.. 89: 21-39. 1997.St. Clair. M. H., Miller, W. H.. Miller. R. L.. Lambe. C. U., and Furman, P. A.Inhibition of cellular a DNA polymerase and herpes simplex virus-induced DNApolymerases by the triphosphate of BW759U. Antimicrob. Agents Chemolher.. 25:191-194, 1984.

Cheng. Y.. Grill, S. P.. Dutschman, G. E.. Nakayama. K.. and Bastow, K. F.Metabolism of 9-(1.3-dihydroxy-2-propoxymethyl)guanine. a new anti-herpes viruscompound, in herpes simplex virus-infected cells. J. Biol. Chem.. 20: 12460-12464.

1983.Balzarini. J.. Bohman. C.. and DeClercq. E. Differential metabolism of cylostalic of(E)-5-(2-bromovinyl)-2'-deoxyuridine, 9-(l.3-dihydroxy-2-propoxymethyl)guanine.

and other antiherpetic drugs on tumor cells transfected by the thymidine kinase geneof herpes simplex virus type I or type 2. J. Biol. Chem.. 269: 6332-6337, 1993.St. Clair, M. H.. Lambe, C. U., and Furman. P. A. Inhibition by ganciclovir of cellgrowth and DNA synthesis of cells biochemically transformed with herpesvirusgenetic information. Antimicrob. Agents Chemother., 31: 844-849. 1987.

Keller. P. M., Fyfe, J. A.. Beauchamp. L.. and Spector. T. Enzymatic phosphorylalionof acyclic nucleoside analogs and correlations with antiherpetic activities. Biochcm.Pharmacol., 30: 3071-3077. 1981.

Furman. P. A.. McGuirt. P. V.. Keller. P. M.. Fyfe. J. A., and Elion. G. B. Inhibitionby acyclovir of cell growth and DNA synthesis of cells biochemically transformedwith herpesvirus genetic information. Virology. 102: 420-430. 1980.

Aswell. J. F.. Allen. G. P.. Jamieson. A. T.. Campbell. D. E., and Gentry. G. A.Antiviral activity of arabinosylthymine in herpes viral replication: mechanism ofaction in vivo and in vitru. Antimicrob. Agents Chcmother.. 12: 243-254. 1977.Shewach. D. S.. Zcrbe. L. K.. Hughes. T. L.. Roessler, B. J.. Breakefield. X. O.. andDavidson, B. L. Enhanced cytotoxicity of antiviral drugs mediated by adenovirusdirected transfer of the herpes simplex virus thymidine kinase gene in rat glioma cells.Cancer Gene Ther.. /: 107-112. 1994.Shewach. D. S., Hahn. T. M., Chang. E.. Hertel. L. W.. and Lawrence. T. S.Metabolism of 2'.2'-difluoro-2'-deoxycylidine and radiation sensitization of human

colon carcinoma cells. Cancer Res.. 54: 3218-3223. 1994.

Kufe. D. W., Major. P. P.. Egan. E. M., and Beardsley, G. P. Correlation ofcytotoxicity with incorporation of ara-C into DNA. J. Biol. Chem.. 255: 8997-9000,

1980.Rustum. Y. M.. and Prcisler. H. D. Correlation between leukemic cell retention ofl-ß-D-arabinofuranosylcylosine 5'-triphosphate and response to therapy. Cancer Res..

j9; 42-49, 1979.

Plunkett, W., Liliemark. J. O.. Adams. T. M.. Nowak. B.. Estey. E.. Kantarjian. H..and Keating. M. J. Saturation of l-/3-i>-arabinofuranosylcytosine 5'-triphosphate

accumulation in leukemia cells during high-dose l -ß-n-arabinofuranosylcytosinetherapy. Cancer Res., 47: 3005-3011. 1987.

Heinemann. V.. and Jehn. U. Rationales for a pharmacologically optimized treatmentof acute nonlymphocytic leukemia with l-ß-D-arabinofuranosylcytosine. Leukemia(Baltimore), 4: 790-796. 1990.Plunkett. W.. Liliemark. J. O.. Estey. E.. and Keating. M. J. Saturation of ara-CTPaccumulation during high-dose ara-C therapy: pharmacologie rationale for intermediate-dose ara-C. Semin. Oncol., 14: 159-166, 1987.Shewach. D. S.. Ellero, J.. Mancini. W. R.. and Ensminger. W. D. Decrease in TTPpools mediated by 5-bromo-2'-deoxyuridine exposure in a human glioblastoma cell

line. Biochem. Pharmacol...«: 1579-1585. 1992.Rubsam. L. Z.. and Shewach. D. S. Improved method to prepare RNA-free DNA frommammalian cells. J. Chromalogr.. 702: 61-68, 1997.

Dahmen, N.. Kowalzick. L.. Spadari. S.. and Koch. G. Arabinose furanosyl thymidine: uptake, phosphorylation and incorporation into DNA of mammalian cells. InVivo, 4: 25-32, 1990.

Plunkett. W.. and Saunders. P. P. Metabolism and action of purine nucleosideanalogs. Pharmacol. Ther.. 49: 239-268. 1991.Samejima, Y., and Mcruelo. D. "Bystander killing" induces apoplosis and is inhibited

by forskolin. Gene Ther.. 2: 50-58. 1995.

Colombo. B. M.. Benedetti. S.. Ottolenghi. S.. Mora. M.. Pollo. B.. Poli, G.. andFinocchiaro. G. The. "bystander effect": association of U-87 cell death with ganci-

clovir-mediated apoptosis of nearby cells and lack of effect in athymic mice. HumanGene Ther.. 6: 763-772. 1995.

on April 3, 2020. © 1998 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

MULTILOG CYTOTOXICITY MEDIATED BY GANCICLOVIR

32. Wallace, H., Clarke, A. R., Harrison, D. J., Hooper. M. L., and Bishop. J. O.Ganciclovir-induced ablation of non-proliferating thyrocytes expressing herpesvirusthymidine kinase occurs by p53-independent apoptosis. Oncogene. 13: 55-61. 1996.

33. Ishii-Morita. H.. Agbaria. R.. Mullen. C. A.. Hirano. H.. Koeplin. D. A.. Ram, Z..Oldfield. E. H.. Johns. D. G.. and Blaese. R. M. Mechanism of "bystander effect"

killing in the herpes simplex thymidine kinase gene therapy model of cancer treatment. Gene Ther.. 4: 244-251. 1997.

34. Smee. D. F.. Boehme. R.. Chernow. M., Binko. B. P.. and Matthews. T. R. Intracel-lular metabolism and en/.ymatic phosphorylation of 9-(l,3-dihydroxy-2-propoxym-elhyllguanine and acyclovir in herpes simplex virus-infected and uninfected cells.Biochcm. Pharmacol., 34: 1049-1056. 1985.

35. Germershausen, J.. Bostedor. R., Field. A. K.. Perry, H.. Liou, R.. Bull, H.. Tolman.R. L.. and Karkas, J. D. A comparison of the antiviral agents 2'-nor-2'-deox-

yguanosine and acyclovir: uptake and phosphorylation in tissue culture and kineticsof in vitrti inhibition of viral and cellular DNA polymerases by their respectivetriphosphates. Biochem. Biophys. Res. Commun., 31: 360-367. 1983.

36. Matsukage. A.. Ono. K.. Ohashi. A.. Takahashi. T.. Nakayama. C., and Saneyoshi. M.Inhibitory effect of 1-ß-D-arabinofuranosylthymine 5'-triphosphate and l-ß-D-ar-abinofuranosylcytosine 5'-triphosphale on DNA polymerases from murine cells and

oncomavirus. Cancer Res., 38: 3076-3079, 1978.

37. Reardon. J. E.. and Speclor. T. Herpes simplex virus type 1 DNA polymerase:mechanism of inhibition by acyclovir triphosphate. J. Biol. Chem.. 264: 7405-7411,

1989.38. Reardon, J. E. Herpes simplex virus type 1 and human DNA polymerase interactions

with 2'-deoxyguanosine 5'-triphosphate analogues. J. Biol. Chem., 264: 19039-

19044, 1989.39. Reid. R.. Mar. E-C. Huang. E-S.. and Topal, M. D. Insertion and extension of acyclic,

dideoxy. and ara nucleolides by herpesviridae. human or and human ßpolymerases.J. Biol. Chem., 263: 3898-3904. 1988.

40. llsley. D. D.. Lee. S-H.. Plunkett, W.. Miller. W.. and Kuchla, R. D. Inhibition of

eucaryotic DNA polymerases and DNA primase by acyclic guanosine analogs:ganciclovir Iriphosphate is a potent and specific DNA polymerase 8 inhibitor. Biochemistry. 34: 2504-2510. 1995.

41. Bambara. R. A.. Murante. R. S.. and Henricksen, L. A. Enzymes and reactions at thecukaryotic DNA replication fork. J. Biol. Chem.. 272.- 4647-4650. 1997.

42. Thust, R.. Schackc. M.. and Wutzler. P. Cytogenetic genotoxicity of antiherpesvirostatics in Chinese hamster V79-E cells. I. Purine nucleoside analogues. AntiviralRes., 31: 105-113, 1996.

43. Foli. M., Marshalko. S., Schurter, E.. Kumar, S.. Beardsley, G. P.. and Schweitzer.B. I. Solution structure ol a DNA dccamer containing the antiviral drug ganciclovir:combined use of NMR. restrained molecular dynamics, and full relaxation matrixrefinement. Biochemistry, 36: 5336-5345. 1997.

44. Prusoff. W. H., and Goz, B. Halogenated pyrimidine deoxyribonucleosides. In: A. C.Salterelli and D. G. Johns (eds.). Handbook of Experimental Pharmacology, pp.272-347. New York: Springer-Verlag. 1975.

45. Boucher, P. D., Ruch. R. J.. and Shewach. D. S. Differential ganciclovir-mediatedcylotoxicity and bystander killing in human colon carcinoma cell lines expressingherpes simplex virus thymidine kinase. Hum. Gene Ther.. 9: 801-814. 1998.

46. Freeman. S. M.. Whartenby, K. A., Freeman. J. L.. Abboud. C. N.. and Marrogi. A. J.In situ use of suicide genes for cancer therapy. Semin. Oncol.. 23: 31-45. 1996.

47. Bi. W. L.. Parysek. L. M.. Warnick. R.. and Stambrook, P. J. In vitro evidence thatmetabolic cooperation is responsible for the bystander effect observed with HSVtkretroviral gene therapy. Hum. Gene Ther.. 4: 725-731, 1993.

48. Dilber. M. S.. Abedi. M. R.. Christensson. B.. Bjorkstrand, B., Kidder, G. M., Naus.C. C. G., Gahrton, G., and Smith, C. I. E. Gap junctions promote the bystander effectof herpes simplex virus thymidine kinase in vivo. Cancer Res., 57: 1523-1528. 1997.

49. Ramesh. R.. Marrogi, A. J.. Munshi. A.. Abboud, C. N.. and Freeman. S. M. In vivoanalysis of the "bystander effect": a cytokine cascade. Exp. Hematol., 24: 829-838,

1996.50. Kaneko. Y.. and Tsukamolo, A. Gene therapy of hepatoma: bystander effects and

non-apoptotic cell death induced by thymidine kinase and ganciclovir. Cancer Lett.,96: 105-110, 1995.

51. Pick. J.. Barker. F. G.. II. Dazin. P.. Weslphale. E. M.. Beyer, E. C., and Israel. M. A.The extent of heterocellular communication mediated by gap junctions is predictiveof bystander cytotoxicity in vitro. Proc. Nati. Acad. Sci. USA. 92: 11071-11075,

1995.52. Mesnil. M.. Piccoli. C.. Tiraby. G.. Willecke, K., and Yamasaki, H. Bystander killing

of cancer cells by herpes simplex virus thymidine kinase gene is mediated byconnexins. Proc. Nail. Acad. Sci. USA, 93: 1831-1835, 1996.

53. 1 I-li.um. A. A.. Saavedra. A.. Zhang. H.. Kucharczuk. J. C.. Spray, D. C., Fishman,G. I., Amin. K. M.. Kaiser. L. R., and Albelda, S. M. Gap junctions play a role in the"bystander effect" of the herpes simplex virus thymidine kinase/ganciclovir system in

vitro. Gene Ther., 3: 85-92, 1996.

54. Wygoda. M. R.. Wilson, M. R., Davis, M. A.. Trosko, J. E., Rehemtulla, A., andLawrence, T. S. Protection of herpes simplex virus thymidine kinase-transduced cellsfrom ganciclovir-mediated cytotoxicity by bystander cells: the Good Samaritan effect.Cancer Res., 57: 1699-1703. 1997.

55. Mesnil, M., Piccoli. C.. Klein. J-L.. Morand. !.. and Yamasaki. H. Lack of correlation

between the gap junctional communication capacity of human colon cancer cell linesand expression of the DCC gene, a homologue of a cell adhesion molecule (N-CAM).Jpn. J. Cancer Res.. 84: 742-747. 1993.

56. Freeman, S. M.. Abboud. C. N.. Whartenby. K. A., Packman, C. H.. Koeplin. D. S..Moollen. F. L.. and Abraham. G. N. The. "bystander effect": tumor regression when

a fraction of the tumor mass is genetically modified. Cancer Res.. 53: 5274-5283.

1993.

3882

on April 3, 2020. © 1998 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1998;58:3873-3882. Cancer Res   Laura Z. Rubsam, Beverly L. Davidson and Donna S. Shewach  Cell Killing

forVirus-Thymidine Kinase-expressing Cells: A Novel Paradigm -d-Arabinofuranosylthymine in Herpes Simplexβand 1-

Superior Cytotoxicity with Ganciclovir Compared with Acyclovir

  Updated version

  http://cancerres.aacrjournals.org/content/58/17/3873

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/58/17/3873To request permission to re-use all or part of this article, use this link

on April 3, 2020. © 1998 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from