A Guanylate Kinase/HSV-1 Thymidine Kinase Fusion Protein Enhances Prodrug-Mediated Cell Killing

Home , Thymidine

Gene Therapy (2006) 13, 1309–1312 & 2006 Nature Publishing Group All rights reserved 0969-7128/06 $30.00 www.nature.com/gt SHORT COMMUNICATION A guanylate kinase/HSV-1 thymidine kinase fusion protein enhances prodrug-mediated cell killing

CL Willmon, E Krabbenhoft and ME Black Department of Pharmaceutical Sciences, Washington State University, Pullman, WA, USA

Herpes simplex virus thymidine kinase (HSVTK) with the enzyme in the GCV activation pathway, guanylate kinase guanosine analog ganciclovir (GCV) is currently the most (GMK). As a means to overcome the bottleneck of prodrug widely used suicide gene/prodrug system for gene therapy activation from the monophosphate to the diphosphate, of cancer. Despite the broad application of the HSVTK/GCV we sought to combine both the critical HSVTK and GMK approach, phosphorylation of GCV to its active state is activities together. In this report we describe the construction inefficient such that high, myelosuppressive doses of GCV of a fusion or chimeric protein of HSVTK and guanylate are needed to observe an antitumor effect. One strategy kinase, show data that demonstrate it confers a B175-fold used to overcome the poor substrate specificity of HSVTK decrease in IC50 compared to HSVTK alone in response to towards GCV (Km ¼ 45 mM) has been to create novel forms of ganciclovir treatment in stably transfected C6 glioma cells TK with altered substrate preferences. Such mutant TKs and finally, we present biochemical evidence of a kinetic have shown benefit and are currently in clinical use. We basis for this improved cell killing. describe here a second strategy to increase the amount of Gene Therapy (2006) 13, 1309–1312. doi:10.1038/ intracellular triphosphorylated GCV by involving the second sj.gt.3302794; published online 29 June 2006

Keywords: thymidine kinase; guanylate kinase; ganciclovir; fusion gene

Herpes simplex virus thymidine kinase (HSVTK) is complete ablation, myelosuppressive doses of GCV are widely used as a suicide gene in combination with required. Several approaches have been reported with ganciclovir (GCV) for the treatment of a variety of varying success to overcome high-dose treatments cancers. To date, over 70 clinical gene therapy trials using including the use of multiple gene copies to yield high HSVTK have been approved.1 Moolten first described expression,7 modulation of nucleoside-metabolizing the potential application of HSVTK for cancer treatment pathways using drugs,8 application of dual suicide based on its ability to phosphorylate the antiviral drug gene/dual prodrug approaches9 and the use of mutant GCV.2 Following delivery of the gene encoding HSVTK suicide genes with improved kinetic parameters towards to cancer cells, the prodrug is administered. Because the prodrug.10,11 HSVTK, but not the endogenous thymidine kinase (TK), In this report, we describe a novel approach taking is able to phosphorylate GCV, cytotoxicity is limited to advantage of the GCV phosphorylation pathway to the site of transfection. Once monophosphorylated, GCV overcome the limitations in the intracellular conversion is further phosphorylated to the triphosphate by the of GCV-monophosphate generated by HSVTK to the endogenously expressed enzymes, guanylate kinase toxic GCV-triphosphate. There is evidence that once (GMK) and nucleoside diphophokinases, respectively. the prodrug monophosphate is formed, a bottleneck In the triphosphorylated state, GCV competes with occurs, leading to accumulation of the ineffective deoxyguanosine triphosphate (dGTP) for incorporation intermediate product.12–14 Responsible for the second into the nascent DNA strand by DNA polymerase and phosphorylation step of GCV and its close relative once incorporated prevents chain elongation that subse- acyclovir (ACV), GMK is an essential enzyme whose 3,4 quently results in apoptosis. expression level and specificity is likely a key determi- A major factor in tumor ablation using the HSVTK/ nant in the overall efficacy of the HSVTK/GCV GCV approach involves the transfer of antimetabolites to approach.15,16 While thymidine kinase from herpes neighboring untransfected cells through gap junctions or simplex virus type I displays broad substrate specificity 5,6 via apoptotic vesicles. Implicit in the bystander effect and can phosphorylate pyrimidines and both pyrimidine is that sufficient phosphorylated GCV be transferred to and purine analogs including thymidine, deoxycytidine, neighboring cells to elicit cell killing. As such, to achieve azidothymidine, acyclovir and GCV, GMK has a restricted substrate specificity and phosphorylates only Correspondence: Dr ME Black, Department of Pharmaceutical the mono-phosphorylated forms of guanosine (GMP), Sciences, Washington State University, P.O. Box 646534, Pullman, deoxyguanosine (dGMP) and guanosine analogs such as WA 99164-6534, USA. 15–18 E-mail: [email protected] GCV and ACV (GCV-MP, ACV-MP). From an enzyme Received 1 December 2005; revised 5 March 2006; accepted 23 kinetic perspective, the high Km of GMK for GCV-MP B March 2006; published online 29 June 2006 (Km ¼ 42–54 mM) compared to the Km for GMP ( 25 mM) Guanylate kinase/thymidine kinase fusion protein CL Willmon et al 1310 supports the notion that low endogenous GMK activity a may limit production of the pharmacologically active form of GCV.15,19 As a means to overcome the bottleneck of prodrug activation at the mono- to diphosphate step, we combined both the critical HSVTK and GMK pET23d pET23d: activities together in a single fusion or chimeric protein HSVTK and assessed the ability of pathway engineering to reduce to the level of GCV required for effective cell pET23d: pET23d: killing. mgmk/TK mgmk To construct the gmk/HSVTK fusion gene, the mouse gmk (mgmk) gene was isolated from pET23d:mgmk as an BglII/MscI fragment and ligated to BglII/MluI(blunt- ended)-digested pET23d:HSVTK.10,19 The resulting plas- b mid, designated pET23d:mgmk/TK, was sequenced to confirm in-frame fusion of the two genes. As a result of the cloning sites used, the first nine amino acids of HSVTK and the last two amino acids of MGMK were deleted. To rapidly assess the functionality of the fusion construct with respect to both HSVTK and GMK activities, genetic complementation for both activities was independently evaluated. We previously described the establishment of an Escherichia coli strain À (TS202A(DE3)) that is deficient in TK activity (tdk ) + arabinose –arabinose and is conditionally gmk deficient.20 Briefly, the mouse gmk gene was integrated into the arabinose operon of Figure 1 Functional complementation assays in E. coli À E. coli KY895 (ilvÀ tdkÀ) to provide induction of GMK TS202A(DE3). E. coli TS202A(DE3) (LAM , tdk-1, IN(rrnD-rrnE)1, ilvÀ276 kanR araÀ) was used for genetic complementation of GMK protein expression in the presence of arabinose. The and TK activities as well as for protein expression.20 For culturing bacterial gmk was disrupted by insertion of a kanamycin- purposes, cells were grown on 2 Â YT (16 g tryptone, 10 g yeast resistance gene into the bacterial gmk locus by recombi- extract, 5 g NaCl per liter) containing 0.2% L-arabinose and nation. Because GMK is an essential enzyme, E. coli 50 mg/ml kanamycin. Carbenicillin was added to 50 mg/ml to TS202A(DE3) requires the presence of arabinose. In the media for growth of vector-transformed cells. For complementation 10 absence of arabinose, no gmk is expressed and the cells of the tdk deficiency, TK selection medium plates supplemented with arabinose and kanamycin were used. For gmk complementa- are nonviable. TS202A(DE3) cells were transformed with tion, M9 minimal medium plates containing kanamycin with or pET23d, pET23d:HSVTK, pET23d:mgmk or the fusion without arabinose were used.20 (a) Diagram of the plating pattern construct, pET23d:mgmk/TK. Cultures were grown in used (left) and TS202A(DE3) cells harboring pET23d, M9 minimal medium in the presence of arabinose to pET23d:mgmk, pET23d:HSVTK and pET23d:mgmk/TK on TK allow the expression of endogenous mouse GMK at selection plates after overnight incubation at 37oC (right). (b) Same 371C overnight before streaking onto TK selection plates10 strains as in (a) after incubation on M9 plus arabinose (left) or M9 containing arabinose, M9 plates plus arabinose or M9 minus arabinose plates (right). plates without arabinose.20 As anticipated, cells harboring pET23d and pET23d:mgmk were not able to complement the TK deficiency of TS202A(DE3), whereas against HSVTK or MGMK were performed on transfec- cells harboring pET23d:HSVTK and pET23d:mgmk/TK tant lysates and show the presence of HSVTK at around were viable (Figure 1a). Similarly, only pET23d:mgmk- 45 kDa when the anti-TK serum is used, and MGMK at and pET23d:mgmk/TK-transformed TS202A(DE3) were around 23 kDa when anti-MGMK serum is used (data viable on M9 plates in the absence of arabinose (Figure not shown). These results are in accord with the proteins’ 1b). These results indicate that the fusion protein is respective predicted molecular masses. The fusion expressed and maintains both TK and GMK activities. protein cross-reacted with both anti-sera at the predicted All cultures grew on the control M9 with arabinose molecular mass of approximately 68 kDa. Relatively plates. equivalent protein levels were also observed. With confirmation of both TK and GMK activities in To determine the cytotoxicity of GCV, transfected the fusion protein, we next sought to evaluate the ability pools were plated in eight replicates and treated with of the fusion gene to sensitize cancer cells to GCV. a wide range of GCV concentrations (0–100 mM). Cell Toward this end, the HSVTK, mgmk and mgmk/TK genes survival was assessed after 7 days using the redox were subcloned into the pREP8D7 vector as described in indicator Alamar Blue and is expressed as a percentage Kokoris et al.21 Rat C6 glioma cells were transfected using of control wells that did not receive GCV (0 mM). Figure 2 electroporation with the vector control (pREP8D7:dual- shows a representative GCV-sensitivity curve of the rat GFP) and vectors containing HSVTK, mgmk or mgmk/ C6 glioma cells transfected with pREP8D7:dual-GFP TK. Insertion of HSVTK, mgmk or the fusion gene into (vector alone), pREP8D7:TK-GFP, pREP8D7:mgmk-GFP this vector allows expression from an RSV LTR promoter. and pREP8D7:mgmk/TK-GFP. Vector control and mgmk- Following initial selection on histidinol, GFP expressed transfected cells were unaffected by GCV treatment. from a constituitive metallothionein promoter enabled Cells expressing the fusion protein MGMK/TK were cell sorting based on GFP expression to enrich transfec- sensitive to GCV concentrations two orders of magnitude

tant pools. Immunoblots using polyclonal serum raised lower than those cells expressing HSVTK (IC50 ¼ 70 and

Gene Therapy Guanylate kinase/thymidine kinase fusion protein CL Willmon et al 1311 130 0.4 mM, respectively). As such, the approximately 175- 120 fold lower IC50 value observed with the fusion construct may be highly relevant in clinical settings where high 110 doses of GCV are not well tolerated by patients. Erbs 100 et al.22 report similar findings when cells were trans- 90 duced with an adenoviral vector expressing a yeast cytosine deaminase:uridine phosphoryltransferase fu- 80 sion protein compared to cytosine deaminase. 70 Although earlier experiments co-expressing HSVTK 60 and MGMK as separate proteins did not demonstrate more than twofold alteration in GCV sensitivity (data not 50 shown and Akyu¨ rek et al.23), the substantial improve- Percent survival Percent 40 ment observed with the fusion protein prompted us to 30 investigate whether functional changes towards GCV are a result of fusing the two proteins together. Towards 20 this end, MGMK, HSVTK and the fusion protein were 10 purified to 495% homogeneity using affinity chromato- 0 graphy and their respective kinetic parameters determined (Table 1). For all substrates examined, the kcat is 0.1 1 10 100 1.8–2.9- fold higher (Po0.002) with the fusion enzyme GCV ( M) versus the respective single enzymes. This indicates that Figure 2 GCV sensitivity assays. In all, 10 mgofeachDNA(pREP8D7: the fusion enzyme has an overall improved turnover rate dual-GFP (vector alone), pREP8D7:mgmk-GFP, pREP8D7:TK-GFP and for all substrates. Interestingly, whereas MGMK/TK Km pREP8D7:mgmk/TK-GFP) was used to transfect 1 Â106 rat C6 glioma values for the normal substrates (GMP and thymidine) cells by electroporation. Cells were grown in Dulbecco’s modified are 2.4–3.8-fold higher (Po0.008 for both substrates), Eagle’s medium plus supplements (5% fetal bovine serum, 1 mM respectively, the Km for GCV is over threefold lower sodium pyruvate, 10 mM HEPES, 100 mM nonessential amino acids, ¼ 100 U/ml penicillin G, 10 mg/ml streptomycin sulfate, 292 mg/ml L- (P 0.006) and is suggestive of an increased relative glutamine, 100 mM sodium citrate and 0.0014% NaCl). After 2 days of preference for binding GCV. Perhaps most revealing is incubation, the medium was replaced with DMEM minus histidine that the kcat/Km value of MGMK/TK for GCV is ninefold plus supplements containing 0.25 mM histidinol. After 2 weeks, better (Po0.002) than the catalytic efficiency displayed histidinol-resistant clones (100–500) were collected and pooled. by wild-type HSVTK towards GCV. These data provide, Approximately 200 000 pooled transfectants were bulk sorted using a at least in part, a kinetic explanation for the improved MoFlo cell sorter (Cytomation, Fort Collins, CO, USA) based on photoactivation and emission of green fluorescent protein (GFP). Data sensitivity of mgmk/TK-transfected tumor cells to GCV. was analyzed using Cyclops software (Cytomation). To determine the Another possibility is that the fusion protein localizes in cytotoxicity of GCV, pools of transfectants were transferred to 96-well transfected cells in a way that provides a more efficient microtiter plates at an initial density of 600 cells per well in DMEM plus cell killing effect. supplements.11 After cell adherence overnight, GCV (0–100 mM)was Many nucleoside analogs require activation by several added in sets of eight wells for each concentration tested. The plates different enzymes before they can exert their cytotoxic were incubated for 6 days at which time the redox indicator dye Alamar Blue was added. Cell survival was determined several hours effects. Evidence suggests that not all enzymes involved later as according to the manufacturer’s instructions and is plotted with in a pathway act efficiently and thus can lead to an standard deviation bar. Legend: pREP8D7:dual-GFP (vector alone; &), accumulation of potentially toxic intermediate products, pREP8D7:mgmk-GFP (’), pREP8D7:TK-GFP (K)andpREP8D7: for example, AZT-MP. Several groups now recognize this mgmk/TK-GFP (m). as a rate-limiting step for antiviral, antineoplastic and

Table 1 Kinetic parameters of puriﬁed enzymes

GMP Thymidine GCV

À1 À1 À1 À1 À1 À1 Km (mM) kcat (s ) kcat/Km (s /mM) Km (mM) kcat (s ) kcat/Km (s /mM) Km (mM) kcat (s ) kcat/Km (s /mM)

MGMK 25 426.6 17.1 ND* ND ND ND MGMK/TK 95a 779.4a 8.3a 2.5b 92.2b 36.9 14.6b 162.6b 11.1b HSVTK ND ND 1.1 40.3 38.0 45.5 55.3 1.2

Overexpression of TK, Mgmk and mgmk/TK from pET vector constructs was as previously reported.11,19 Mgmk was expressed from pETHT which fuses a histidine tag to MGMK and the enzyme purified using nickel affinity chromatography.19 Purification of TK and MGMK/TK was performed by affinity chromatography using CH-Sepharose 4B-coupled p-aminophenylthymidine 3’-phosphate resin as previously described.11 Phosphorylation of radiolabeled substrates (thymidine and GCV) was detected by a filter-binding assay11 and GMK assays (GMP) were carried out as described in Brady et al.19 Radioisotopes for kinetic assays: [methyl-3H]thymidine (specific activity, 85 Ci/mmol) was purchased from Amersham and [8-3H]ganciclovir (specific activity, 17.6 Ci/mmol) was purchased from Moravek Biochemicals (Brea, CA, USA). Data were plotted as the double reciprocal of velocity (min/mmole Â 104) versus substrate concentration (mMÀ1) and the intercepts values used to determine Km and Vmax. These values were used to calculate the kcat from the double reciprocal plot data as per active site (one active site per monomer). Kinetic values were compared using Student’s unpaired t-test, using KaleidaGraph 4.01 software: a or b indicates a P-value o0.015 compared to MGMK or TK, respectively. ND, not determined.

Gene Therapy Guanylate kinase/thymidine kinase fusion protein CL Willmon et al 1312 suicide gene therapeutic applications.12–14 We report here exhibit enhanced metabolic suicide and radiosensitivity. Hum the first attempt to overcome the bottleneck in the widely Gene Ther 1997; 8: 73–85. used HSVTK-driven GCV activation pathway using a 10 Black ME, Newcomb TG, Wilson HM, Loeb LA. Creation of kinetically favorable fusion protein. By providing effec- drug-specific herpes simplex virus type 1 thymidine kinase tive cell killing at low GCV concentrations, this novel mutants for gene therapy. Proc Natl Acad Sci USA 1996; 93: fusion protein could benefit many patients by improving 3525–3529. tumor ablation perhaps through an expanded bystander 11 Kokoris MS, Sabo P, Adman ET, Black ME. Enhancement of effect and by reducing the side effects associated with the tumor ablation by a selected HSV-1 thymidine kinase mutant. high doses of GCV currently administered. Gene Therapy 1999; 6: 1415–1426. 12 Fridland A, Connelly MC, Ashmun R. Relationship of deoxy- nucleotide changes to inhibition of DNA synthesis induced by Acknowledgements the antiretroviral agent 30-azido-30-deoxythymidine and release of its monophosphate by human lymphoid cells (CCRF-CEM). We are grateful to Mark Kokoris for initial cell culture Mol Pharmcol 1990; 37: 665–670. and kinetic studies. This project was supported by NIH 13 Johansson M, Amiri M, Karlsson A. Phosphorylation of 9-b-D- grant CA85939. arabinofuranosylguanine monophosphate by Drosophila melano- gaster guanylate kinase. Biochem Pharmacol 2005; 70: 987–992. 14 Ostermann N, Lavie A, Padiyar S, Brundiers R, Veit T, Reinstein References J et al. Potentiating AZT activation: structures of wild-type and 1 Edelstein ML, Abedi MR, Wixon J, Edelstein RM. Gene therapy mutant human thymidylate kinase suggest reasons for the clinical trials worldwide 1989–2004 – an overview. J Gene Med mutants’ improved kinetics with the HIV prodrug metabolite 2004; 6: 597–602. AZTMP. J Mol Biol 2000; 304: 43–53. 2 Moolten FL, Wells JM. Curability of tumors bearing herpes 15 Boehme RE. Phosphorylation of the antiviral precursor 9-(1,3- thymidine kinase genes transferred by retroviral vectors. J Natl dihydroxy-2-propoxymethyl)guanine monophosphate by gua- Cancer Inst 1990; 82: 297–300. nylate kinase isozymes. J Biol Chem 1984; 258: 12346–12349. 3 Field AK, Davies ME, DeWitt C, Perry HC, Liou R, Germer- 16 Miller WH, Miller RL. Phosphorylation of acyclovir (acyclogua- shausen J et al. 9-([2-hydroxy-1-(hydroxymethyl)ethoxy]methyl)- nosine) monophosphate by GMP kinase. J Biol Chem 1980; 255: guanine: a selective inhibitor of herpes group virus replication. 7204–7207. Proc Natl Acad Sci USA 1983; 80: 4139–4143. 17 Agarwal KC, Miech RP, Parks Jr RE. Guanylate kinases from 4 Reardon JE. Herpes simplex virus type 1 and DNA polymerase human erythrocytes, hog brain, and rat liver. Methods Enzymol interactions with 20-deoxyguanosine 50-triphosphate analogs. 1978; 51: 483–490. Kinetics of incorporation into DNA and induction of inhibition. 18 Gaidarov IO, Suslov ON, Abdulaev NG. Enzymes of the cyclic J Biol Chem 1989; 264: 19039–19044. GMP metabolism in bovine retina. I. Cloning and expression of 5 Bi WL, Parysek LM, Warnick R, Stambrook PJ. In vitro evidence the gene for guanylate kinase. FEBS Lett 1993; 335: 81–84. that metabolic cooperation is responsible for the bystander effect 19 Brady WA, Kokoris MS, Fitzgibbon M, Black ME. Cloning, observed with HSV tk retroviral gene therapy. Hum Gene Ther characterization, and modeling of mouse and human guanylate 1993; 4: 725–731. kinases. J Biol Chem 1996; 271: 16734–16740. 6 Freeman SM, Abboud CN, Whartenby KA, Packman CH, 20 Stolworthy TS, Krabbenhoft E, Black ME. A novel Escherichia coli Koeplin DS, Moolten FL et al. The ‘bystander effect’: tumor strain allows functional analysis of guanylate kinase drug regression when a fraction of the tumor mass is genetically resistance and sensitivity. Anal Biochem 2003; 322: 40–47. modified. Cancer Res 1993; 53: 5274–5283. 21 Kokoris MS, Sabo P, Black ME. In vitro evaluation of mutant 7 Kim YG, Bi W, Feliciano ES, Drake RR, Stambrook PJ. HSV-1 thymidine kinases for suicide gene therapy. Anticancer Res Ganciclovir-mediated cell killing and bystander effect is 2000; 20: 959–964. enhanced in cells with two copies of the herpes simplex virus 22 Erbs P, Regulier E, Kintz J, Leroy P, Poitevin Y, Exinger F et al. thymidine kinase gene. Cancer Gene Ther 2000; 7: 240–246. In vivo cancer gene therapy by adenovirus-mediated transfer of a 8 Boucher PD, Ostruszka LJ, Shewach DS. Synergistic enhance- bifunctional yeast cytosine deaminase/uracil phosphoribosyl- ment of herpes simplex virus thymidine kinase/ganciclovir- transferase fusion gene. Cancer Res 2000; 60: 3813–3822. mediated cytotoxcity by hydroxyurea. Cancer Res 2000; 60: 23 Akyu¨ rek LM, Nallamshetty S, Aoki K, San H, Yang Z-H, Nabel 1631–1636. GJ et al. Coexpression of guanylate kinase with thymidine kinase 9 Rogulski KR, Kim JH, Kim SH, Freytag SO. Glioma cells enhances prodrug cell killing in vitro and suppresses vascular transduced with an Escherichia coli CD/HSV-1 TK fusion gene smooth muscle cell proliferation. Mol Ther 2001; 3: 779–786.

Gene Therapy