Skip to main content
NCBI home page
Molecular Therapy logoLink to Molecular Therapy
. 2009 May 26;17(8):1373–1380. doi: 10.1038/mt.2009.114

Abrogation of Microsatellite-instable Tumors Using a Highly Selective Suicide Gene/Prodrug Combination

Cristina Ferrás 1,2, Joachim AF Oude Vrielink 1, Johan WA Verspuy 1, Hein te Riele 3, Anastasia Tsaalbi-Shtylik 1, Niels de Wind 1
PMCID: PMC2835239  PMID: 19471249

Abstract

A substantial fraction of sporadic and inherited colorectal and endometrial cancers in humans is deficient in DNA mismatch repair (MMR). These cancers are characterized by length alterations in ubiquitous simple sequence repeats, a phenotype called microsatellite instability. Here we have exploited this phenotype by developing a novel approach for the highly selective gene therapy of MMR-deficient tumors. To achieve this selectivity, we mutated the VP22FCU1 suicide gene by inserting an out-of-frame microsatellite within its coding region. We show that in a significant fraction of microsatellite-instable (MSI) cells carrying the mutated suicide gene, full-length protein becomes expressed within a few cell doublings, presumably resulting from a reverting frameshift within the inserted microsatellite. Treatment of these cells with the innocuous prodrug 5-fluorocytosine (5-FC) induces strong cytotoxicity and we demonstrate that this owes to multiple bystander effects conferred by the suicide gene/prodrug combination. In a mouse model, MMR-deficient tumors that contained the out-of-frame VP22FCU1 gene displayed strong remission after treatment with 5-FC, without any obvious adverse systemic effects to the mouse. By virtue of its high selectivity and potency, this conditional enzyme/prodrug combination may hold promise for the treatment or prevention of MMR-deficient cancer in humans.

Introduction

The common hereditary cancer predisposition Lynch syndrome (also called hereditary nonpolyposis colorectal cancer) is caused by a defective allele of one of four DNA mismatch repair (MMR) genes.1,2 Somatic loss of the remaining wild-type allele results in MMR deficiency and the consequent spontaneous mutator phenotype underlies the rapid development of colorectal, stomach, endometrial, and other visceral cancers. Also a significant fraction of sporadic colorectal cancers has lost MMR, consequent to methylation of the MLH1 promoter.3 Destabilization of ubiquitous monotonous sequence repeats (“microsatellites”), called the microsatellite-instable (MSI) phenotype, is a phenotypic hallmark of MMR-deficient cancer.1,2

Currently, the chemotherapeutic drug 5-fluorouracil (5-FU) is the standard drug used in treatment of colorectal cancer.4 However, MSI colorectal cancer and cultured MSI cells appear relatively refractory to, and may be hypermutable by, 5-FU-based chemotherapy.5 This likely is a consequence of the loss of cytotoxic, MMR-dependent, processing of (fluoro)deoxyribonucleotides after their incorporation in DNA4,6 (Figure 1a).

Figure 1.

Figure 1

Open in a new tab

Suicide gene therapy constructs and genetic modification of embryonic stem (ES) cells. (a) Metabolism of 5-fluorocytosine (5-FC). Conversions indicated by dotted arrows are performed by cellular enzymes, solid arrows represent activities of FCU1, consisting of cytosine deaminase (CD) and uracil phosphoribosyltransferase (UPRT) moieties. TS, thymidylate synthetase. (b) VP22FCU1 constructs. Top: diagram of the triple chimeric VP22(C14/15)FCU1 genes. (C)n, C14 (in frame) or C15 (out-of-frame) microsatellite; FLAG, FLAG epitope; P, promoter; P(A), poly(A) addition signal. Middle: VP22 expressed from VP22(C15)FCU1, truncated downstream of the C15 microsatellite. Bottom: full-length VP22FCU1, expressed from VP22(C14)FCU1. (c) Targeting of VP22(C15)FCU1 at Rosa26 in Msh2+/− ES cells. I: Genomic Rosa26 locus. EcoRI sites and the location of the probe, used for S blotting, are indicated. II: Targeting vector pR3PCR-VP22(C15)FCU1, containing a promoterless Zeo cassette and VP22(C15)FCU1, flanked by cloned Rosa26 sequences. III: Genomic Rosa26 locus after correct targeting with pR3PCR-VP22(C15)FCU1. (d) Southern blot of the wild-type (WT) and targeted (T) Rosa26 alleles. 15.5 kb: EcoRI fragment, specific for the wild-type locus; 6 kb: EcoRI fragment, specific for the targeted locus.

Suicide gene therapy has emerged as an attractive strategy in the treatment of cancer. Usually, suicide genes encode nonmammalian enzymes that convert toxicologically inert agents into highly cytotoxic metabolites.7,8,9 The enzyme FCU1 (or superCD) and the antimycotic drug 5-fluorocytosine (5-FC) is such a combination.10,11,12,13 FCU1 is composed of fused cytosine deaminase (CD) and uracil phosphoribosyltransferase (UPRT) moieties derived from baker's yeast Saccharomyces cerevisiae. FCU1 deaminates 5-FC to 5-FU (by the CD moiety) followed by conversion to 5-fluorouridine-5′-monophosphate, by the UPRT moiety (Figure 1a). After further phosphorylation by cellular enzymes, the resulting 5-fluorouridine-5′-triphosphate is incorporated into RNA and thereby perturbs RNA processing, a highly cytotoxic event.7,8,9 The cytotoxicity of the FCU1/5-FC combination is enhanced by bystander effects mediated by diffusion of 5-FU from FCU1- expressing tumor cells.10,11 The transport of FCU1 to neighboring cells by fusing it to the “transporter” protein VP22 (refs. 14,15) further enhances bystander effects in vitro.16,17

A major challenge in suicide cancer therapy is the lack of selectivity, which relates to the inability of the transfer vehicle carrying the suicide gene to effectively discriminate between normal and malignant cells.18 We argued that the MSI phenotype provides a unique opportunity to restrict suicide gene expression to MMR-deficient cancer cells. Thus, by inserting an out-of-frame microsatellite in the 5′ coding region of the suicide gene, its activity will depend on reversion of the frameshifting microsatellite, an event that is restricted to the MSI cancer cells. Here we have explored the feasibility of this approach and have constructed a VP22FCU1 derivative that contains an out-of-frame microsatellite. This conditionally active suicide gene indeed mediates selective and highly efficient killing of cultured MSI cells and of MSI tumors. In addition, the RNA-directed toxicity induced by the VP22FCU1/5-FC combination bypasses the tolerance of MMR-deficient tumors to treatment with 5-FU. Thus, the out-of-frame VP22FCU1/5-FC combination is a promising novel approach for the selective gene therapy of MSI cancer.

Results

A modified VP22FCU1 gene induces generalized cytotoxicity

We fused VP22 to the N terminus of FCU1 in an attempt to enhance bystander effects of the suicide protein. In addition, the VP22 moiety was engineered to contain an in-frame C14 repeat, that does not alter the VP22 protein sequence, and an N-terminal FLAG epitope (Figure 1b). Polyclonal stable transfectants were generated from facultatively MMR-deficient 293T Lα cells and these were called VF293T Lα [transfected with VP22(C14)FCU1], and F293T Lα (transfected with FCU1), respectively. Compared with 293T Lα cells, F293T Lα cells showed enhanced sensitivity to 5-FU, in support of FCU1-mediated toxicity toward RNA. 5-FU toxicity was further enhanced when VP22 was fused to FCU1 (Figure 2a, Supplementary Table S1). Compared to 293T Lα cells, toxicity of 5-FC was increased ~1,000- and 2,500-fold in F293T Lα and VF293T Lα cells, respectively (Figure 2b, Supplementary Table S1). In conclusion, the FCU1/5-FC combination mediates strong RNA-based toxicity and the toxicity is further enhanced by VP22.

Figure 2.

Figure 2

Open in a new tab

Responses of 293T Lα cells transfected with the pCI-neoFCU1 and pCI-neoVP22(C14)FCU1 constructs. (a) Sensitivity of 293T Lα (WT), VF293T Lα (VP22FCU1), and F293T Lα (FCU1) cells to 5-fluorouracil (5-FU), cultured in the absence or presence of Dox [mismatch repair (MMR)-proficient and MMR-deficient conditions, respectively]. Survival was measured after 6 days of treatment using the vital dye resazurin. Error bars indicate the SD. (b) Sensitivity of 293T Lα (WT), VF293T Lα (VP22FCU1), and F293T Lα (FCU1) cells to 5-fluorocytosine (5-FC), cultured in the absence or presence of Dox. Survival was measured after 6 days using resazurin. Error bars indicate the SD. WT, wild type.

Consistent with previous results,19 we did not find an influence of the MMR status of 293T Lα cells on sensitivity to a single dose of 5-FU (Figure 2a, Supplementary Table S1). The 5-FU or 5-FC sensitivity of the F293T Lα cells and VF293T Lα cells was also independent of their MMR status (Figure 2a,b, Supplementary Table S1).

Multiple bystander effects induced by FCU1 derivatives

To investigate bystander effect(s) mediated by the FCU1 derivatives, we mixed (naive) 293T Lα cells with either F293T Lα cells or VF293T Lα cells, prior to the addition of 5-FC. A significant bystander effect was already seen when F293T Lα cells were mixed with naive cells (Figure 3a, Supplementary Table S2). This bystander effect was exacerbated slightly, but significantly, when VP22 was fused to FCU1 (VF293T cells; Figure 3a, Supplementary Table S2). To define the nature of the bystander effect, we prepared conditioned media that were added to naive cells. None of the conditioned media displayed cytotoxicity in the absence of 5-FC (Figure 3b-1). Also 293T Lα–conditioned medium conferred no cytotoxicity in the presence of 5-FC (Figure 3b-2, white bar). Strong cytotoxicity was induced by media, conditioned by F293T Lα or VF293T Lα cells, after 5-FC addition, indicating that both FCU1 and VP22FCU1 are excreted (Figure 3b-2, gray bars). Incubation of the conditioned media with 5-FC prior to their addition to naive cells further enhanced cytotoxicity (Figure 3b-3) suggesting deamination of 5-FC into cytotoxic 5-FU in the medium by the excreted suicide proteins. Preincubation of F293T Lα- or VF293T Lα–conditioned media for 72 hours at 37 °C prior to the addition to naive cells still induced considerable cytotoxicity after adding 5-FC, indicating that excreted suicide proteins were stable (Figure 3b-4). Under all conditions, VP22FCU1-conditioned media conferred higher toxicity than FCU1-conditioned media. In addition, uniquely the VP22FCU1-containing medium conferred significant cytotoxicity after brief incubation with naive cells followed by washing and addition of 5-FC (Figure 3b-5). These results are consistent with reuptake of VP22FCU1 from the medium into naive cells. We could not detect VP22FCU1 in these cells by western blotting or by immunoprecipitation from the medium (data not shown), indicating that the protein concentration is low. Based on these data, we concluded that the VP22FCU1/5-FC combination is highly toxic, mediating multiple bystander effects. These properties are essential for cancer gene therapy when only a few tumor cells express the suicide enzyme.

Figure 3.

Figure 3

Open in a new tab

Bystander effects induced by VP22FCU1 and FCU1. (a) F293T Lα or VF293T Lα cells were mixed with 293T Lα cells in different ratios followed by culture in the presence of 1 mmol/l 5-fluorocytosine (5-FC). After 6 days, viability was assessed using resazurin. Error bars indicate the SD. (b) Conditioned media were prepared from 293T Lα (white bars), F293T Lα (light gray bars), and VF293T Lα (dark gray bars) cells and added to naive 293T Lα cells. Viability was measured after 6 days using resazurin. Error bars indicate the SD. 1: Freshly conditioned media added to 293T Lα cells in the absence of 5-FC. 2: Freshly conditioned media added to 293T Lα cells together with 5-FC. 3: Conditioned media preincubated for 72 hours at 37 °C with 5-FC prior to their addition to 293T Lα cells. 4: Conditioned media were added together with fresh 5-FC to 293T Lα cells after preincubation for 72 hours at 37 °C in the absence of 5-FC. 5: Freshly conditioned media were added to 293T Lα cells for 4 hours after which cells were washed and incubated in fresh medium +5-FC.

The MSI phenotype enables selective gene activation

As embryonic stem (ES) cells mimic tumor stem cells and allow controlling the copy number and genomic integration site of a transfected construct, these cells are a suited model to study relative rates of microsatellite instability and the efficacy of a conditional suicide gene/prodrug combination. We wanted to investigate the feasibility of using a suicide gene containing an out-of-frame microsatellite to selectively target its expression to MSI cells. To this aim, we generated a reporter construct consisting of a neo gene that is disrupted by an out-of-frame G10 microsatellite within its coding region. The G10neo reporter gene was inserted at the chromosomal Rb locus in isogenic ES cell lines, deficient for the Msh2 or Msh6 MMR genes, and in wild-type cells. We investigated the stability of the G10 microsatellite by growing the ES cell lines for a determined number of doublings followed by plating the cell lines in the presence of G418. In Msh2−/− and Msh6−/− ES cells, the microsatellite was destabilized 6,000- and 1,850-fold, respectively (Table 1). By PCR, we determined that in virtually all G418-resistant colonies, a −1 frameshift within the G10 microsatellite had rendered neo in frame (data not shown). These results demonstrate the feasibility of using an out-of-frame microsatellite to restrict gene expression to MSI cells, largely independent of their specific MMR gene defect.

Table 1.

Frameshift reversion rates in mismatch repair–mutant ES cells

graphic file with name mt2009114t1.jpg

Open in a new tab

Selective killing of MSI ES cells containing a frameshifted VP22FCU1 gene

Based on the results with the out-of-frame neo reporter, we constructed a derivative of VP22(C14)FCU1 containing an additional deoxycytosine residue within the artificial microsatellite in VP22, rendering the downstream moiety out-of-frame (Figure 1b). A single copy of the gene, named VP22(C15)FCU1, was inserted at the Rosa26 locus in microsatellite-stable (MSS) Msh2+/− ES cells (Figure 1c). Correct gene targeting was demonstrated by the appearance of a 6-kb band after probing a genomic EcoRI digest (Figure 1c,d). The resulting (MSS) ES cell line was called V(+1)FMsh2+/− and used for the derivation of (MSI) V(+1)FMsh2−/− ES cell lines 1–3.

To measure the reversion of VP22(C15)FCU1 to an in-frame VP22FCU1 gene the cells were cultured in medium without 5-FC. At fixed intervals, a sample from the culture was seeded in medium containing 5-FC and viability of the cells was measured after 5 days. No sensitivity to 5-FC emerged after the 16 days of culture of V(+1)FMsh2+/− cells (Figure 4a) indicating stability of the C15 out-of-frame microsatellite in MSS cells. In sharp contrast, the three MSI progeny lines displayed a progressive 5-FC sensitivity (Figure 4a). This was already apparent at the start of the culture, which probably is caused by reversion during the 5-day treatment with 5-FC. In agreement with the viability assays, analysis of clonal survival demonstrated major loss of viability in cells sampled at day 4 of culture and virtually complete loss at day 16 (Figure 4b). We used western blotting to investigate whether the rapid conversion to 5-FC sensitivity reflects the accumulation of VP22FCU1-expressing cells. In the MSI ES cells, a VP22FCU1-specific band was indeed detected, starting at day 4 of culture and increasing over the 16-day culture period. As expected, the MSS cells did not express VP22FCU1 protein at the end of the 16-day culture period (Figure 4c). In order to better quantify the rate of reversion of the VP22(C15)FCU1 gene in MSI ES cells, V(+1)FMsh2−/− ES line 1 was seeded at clonal density after 30 doublings in medium without 5-FC. One hundred clones were analyzed for expression of VP22FCU1 by western blotting. A bimodal distribution of VP22FCU1 expression levels was seen in these clones, with 20 clones displaying a high level of expression (Figure 4d). We infer that the latter clones originated from cells expressing the protein at the time of seeding, resulting from reversion during the 30 doublings in the absence of 5-FC. In contrast, in the remaining, poorly expressing, clones a fraction of the cells likely has acquired VP22FCU1 expression after seeding at clonal density. This result implies a very high reversion rate for VP22(C15)FCU1, of ~13 × 10−3.

Figure 4.

Figure 4

Open in a new tab

Rapid elimination of (MSI) V(+1)FMsh2−/− embryonic stem (ES) cells, but not of (MSS) V(+1)FMsh2+/− cells, by 5-fluorocytosine (5-FC). (a) V(+1)FMsh2+/− line and V(+1)FMsh2−/− ES lines 1–3 were purged from pre-existing VP22FCU1-expressing cells and then cultured for 16 days (~24 doublings) in the absence of 5-FC. At the indicated time points, aliquots from the culture were seeded in the presence of 5-FC. Viability was assessed by counting adherent cells after culturing for 5 days in the presence of 5-FC. Error bars indicate the SD. (b) Viability assessed by clonal survival analysis of adherent cells, sampled at days 4 and 16 (see Figure 4a), after a 5-day treatment with 5-FC. (c) Expression of VP22FCU1 in microsatellite-stable (MSS) and microsatellite-instable (MSI) ES cells. Lane 1: Msh2−/− control cells. Lane 2: Lysate of V(+1)FMsh2+/− cells cultured for 16 days (24 doublings) in the absence of 5-FC. Lane 3: VP22FCU1-expressing reverted subline of V(+1)FMsh2−/− ES line 1 (see Figure 4d). Lanes 4–9: V(+1)FMsh2−/− ES line 1 after culture for different periods in the absence of 5-FC. Increasing expression of VP22FCU1 is detected from day 4 of culture onward. (d) A bimodal distribution of VP22FCU1 expression is found after 30 doublings of V(+1)FMsh2−/− ES cells in the absence of 5-FC followed by seeding at clonal dilution. +: clones expressing a high level of VP22FCU1 resulting from reversion prior to clonal seeding. −: clones expressing a low level of VP22FCU1 resulting from reversion after clonal seeding.

Selective abrogation of MSI tumors

We wanted to investigate whether the out-of-frame VP22(C15)FCU1 gene enables the selective elimination of MSI tumors. To this aim, MSI V(+1)FMsh2−/− ES cell line 1 was injected subcutaneously in one flank of 26 athymic mice whereas the other flank was injected with MSS V(+1)FMsh2+/− cells (Figure 5a). This protocol results in the outgrowth of subcutaneous teratomas.20 Starting at day 7 after injection, 5-FC was administered daily to a randomly chosen cohort, comprising half of the mice (Figure 5a). The remaining mice were mock treated with phosphate-buffered saline (PBS) and growth of MSS and MSI teratomas was measured daily.

Figure 5.

Figure 5

Open in a new tab

Abrogation of microsatellite-instable (MSI) tumors, but not of microsatellite-stable (MSS) tumors, by 5-fluorocytosine (5-FC). (a) Schematic representation of the experiment. See the main text for details. (b) Cumulative V(+1)FMsh2+/− (MSS) and V(+1)FMsh2−/− (MSI) tumor incidences showing selective abrogation of MSI tumor development by treatment with 5-FC. (c) Growth curves of individual MSS and MSI tumors showing abrogation of MSI tumorigenesis when treated with 5-FC at an early stage and reduction in tumor size when treated at a late stage. MSS tumors are not affected by the 5-FC treatment. Numbers refer to the identity of each mouse. (d) Cytoplasmic expression of the FLAG epitope in only a small fraction of cells in both MSS and MSI tumors (arrows) indicates large-scale epigenetic silencing of the suicide gene. Nuclei were stained with 4′,6-diamidino-2-phenylindole. ES, embryonic stem; PBS, phosphate-buffered saline.

After 9 days, in the mock-treated mice, the incidence and sizes (Figure 5b,c) of MSI and MSS tumors were indistinguishable. In addition, no difference was detected in incidence or sizes of MSS tumors between the mock-treated control and 5-FC-treated groups (Figure 5b,c). This is in agreement with the undetectable reversion frequency of VP22(C15)FCU1 in MSS ES cells. In sharp contrast, MSI tumors had not appeared in the mice that were treated with 5-FC (Figure 5b). We then inverted the treatment, proceeding with daily 5-FC injections in the mice that were hitherto treated with PBS, and vice versa. MSI tumors did not reappear in mice that were initially treated with 5-FC and subsequently with PBS (Figure 5b,c). Moreover, all MSI tumors that had emerged in mice that were mock treated with PBS until day 16 of the experiment had significantly regressed, or even disappeared, after 2 weeks of subsequent treatment with 5-FC (Figure 5c).

We anticipated that MSI tumors regressed due to the high reversion frequency of VP22(C15)FCU1, resulting in toxicity to the entire tumor owing to bystander effects. In contrast to our expectation, we were unable to detect full-length VP22FCU1 protein in MSI tumors from mice that were mock treated by western blotting (data not shown). Sections of these mock-treated MSI and MSS tumors were analyzed for expression of the FLAG epitope by immunohistochemical staining. This enabled to detect individual cells that express an N-terminal portion of the protein, irrespective of frameshifting within the downstream-located microsatellite. Both within MSI and MSS tumors only a small, but very similar, minority of cells displayed cytoplasmic expression of the FLAG epitope, indicating generalized loss of gene expression (Figure 5d). Also in an MSI ES cell line of which most cells produce in-frame VP22FCU1, the initial level of the protein was reduced approximately tenfold after 1 week of culture in the absence of the selective agent zeocin (data not shown). Together, these results indicate that both the in-frame and out-of-frame VP22FCU1 genes are prone to epigenetic silencing in both MSS and MSI tumors, and in cultured cells. Nevertheless, our results demonstrate that even a very low number of in-frame VP22FCU1-expressing cells within a MSI tumor is sufficient to induce significant remission, in line with the strong bystander effects of the suicide gene in cultured cells.

During the course of the experiment, no adverse effects were seen of the 5-FC treatment on weight, behavior, and overall health of the mice, and no obvious visceral abnormalities or anemia were observed. In addition, in 5-FC-treated mice, the size of MSS tumors was not affected by the simultaneous presence of MSI tumors (Figure 5c, mice 33, 34, 41, and 42).

Discussion

Here we have investigated the feasibility of using a conditionally active suicide gene/prodrug combination, VP22(C15)FCU1/5-FC, for the treatment of MSI cancer. We predicted that this combination might have several advantages over the current 5-FU-based chemotherapy of MSI cancer. First, the MSI phenotype enables targeting of the toxic effects of the enzyme/prodrug combination selectively to cancer cells and we indeed demonstrate very highly selective toxicity toward MSI cells. Second, owing to the UPRT moiety of FCU1, the major 5-FC metabolite is 5-fluorouridine-5′-triphosphate that derives its toxicity from incorporation into RNA (Figure 1a). Indeed, we confirmed that FCU1 confers RNA-mediated cytotoxicity (Figure 2, Supplementary Table S1). We anticipate that this mitigates the relative insensitivity and hypermutability of MMR-deficient cancer cells to 5-FU-based chemotherapy.4 We detected no MMR-dependent tolerance to 5-FU in 293T Lα cells, in contrast to results from others,4,21 a discrepancy that is likely related to differences in treatment regimens.19,22 Third, the strong bystander effects, induced by the VP22FCU1/5-FC combination were anticipated to enhance toxicity toward the tumor. Indeed, VP22FCU1 mediates generalized toxicity by at least four mechanisms: (i) direct production of 5-fluorouridine-5′-monophosphate in expressing cells, affecting RNA processing; (ii) diffusion of 5-FU from FCU1- and VP22FCU1-expressing cells into the medium (ref. 10 and data not shown); (iii) release of the suicide protein to the culture medium followed by deamination of 5-FC to 5-FU in the medium (this work); and (iv) import of VP22FCU1 into naive cells where it converts 5-FC into 5-fluorouridine-5′-monophosphate (ref. 17 and this work).

Msh6 defects specifically predispose to endometrial cancer, in addition to colon cancer.23 A G10 microsatellite, embedded within the neo gene integrated at the Rb locus, was destabilized 6,000- and 1,850-fold in Msh2−/− and Msh6−/− ES cells, respectively, and −1 frameshifting was a predominant event. The slightly lower instability of the microsatellite in Msh6−/− ES cells is likely a consequence of the redundant repair of slipped intermediates by the alternative Msh2/Msh3 MMR heterodimer.1,2 This result resembles data obtained in Msh2 and Msh6-deficient S. cerevisiae.24 In apparent contrast, others have found that a C12 microsatellite in a reporter gene was only 20-fold less stable in MMR deficient, compared to MMR proficient, 293T Lα cells, owing to a relatively high reversion frequency in the MMR-proficient cells.25 Possibly this is caused by the presence of multiple copies of the reporter gene that may revert by unequal gene conversion, or by spontaneous loss of expression of the transactivator that is required for expression of the MLH1 MMR gene.

Based on our results, a VP22FCU1 derivative was constructed that carries a C15 microsatellite, conferring a +1 frameshift, within the N terminus of the fusion protein. We expected that, because of its larger size, the C15 microsatellite would be significantly more destabilized in MSI cells than the G10 microsatellite used in the neo reporter. Indeed, the rate of reversion of the VP22(C15)FCU1 gene was approximately tenfold higher than that of the G10-containing neo gene. This rate was sufficiently high to induce massive cytotoxicity within a few cell doublings, mediated by the strong bystander effects of VP22FCU1.

We tested the in vivo selectivity and efficacy of the VP22(C15)FCU1 gene toward MSI tumors using a mouse tumor model. In this experiment, the development of MSI tumors was irreversibly abrogated when treated at an early stage whereas large MSI tumors showed continued regression upon 5-FC treatment. In contrast, isogenic MSS tumors were not affected, attesting to the high selectivity of the conditional suicide gene. Furthermore, no adverse systemic effects of the 5-FC treatment were seen in the mice, and growth of MSS tumors in 5-FC-treated mice was not affected by the presence of a MSI tumor in the same animals. Together, these data indicate that the toxic bystander effects predominantly remain restricted to the tumor. Surprisingly, we could not detect expression of VP22FCU1 in MSI tumors by western blotting and we found that within both MSS and MSI tumors the out-of-frame suicide gene is silenced at a large scale. Silencing was also found for expression of the (in-frame) VP22FCU1 protein in cultured MSI ES cells within a week of culture in the absence of the selective agent zeocin. It is likely that this silencing is caused by an epigenetic event. Indeed, the FCU1 gene contains many CpG dinucleotides that are potential targets for methylation. It is therefore conceivable that a modified FCU1 allele (InvivoGen, San Diego, CA) that contains optimized codons and is devoid of CpG dinucleotides will yet be more efficacious in vivo than the FCU1 allele that we used. Furthermore, integration at a more active locus or even episomal replication of the suicide gene may reduce epigenetic inactivation. Nevertheless, our data show that already a very low level of VP22FCU1 expression is sufficient to induce strong tumor regression.

The use of an out-of-frame repeat to target MSI cancer is not restricted to VP22FCU1 but can also be extended to other suicide genes like thymidine kinase and nitroreductase.8,25 A combination of multiple out-of-frame suicide genes in one construct may further enhance their efficacy without interfering with the very high selectivity toward MSI tumors. Alternatively, the introduction of multiple independent copies of the VP22(C15)FCU1 gene may also enhance their efficacy without significantly reducing their selectivity. Finally, oncolytic viruses26 may be constructed that carry an out-of-frame microsatellite in a gene essential for efficient replication. Because the essential gene will only revert in MSI tumor cells, this will restrict viral replication to the tumor.

Hitherto, suicide genes generally have to be applied topically into cancers, due to their relatively low selectivity. The high selectivity of out-of-frame suicide genes for MSI cells may allow their systemic administration in high titers. This may also target distant (micro) metastases of an MSI tumor. These advantages additionally aid in circumventing the problem of delivery into all cells of tumors, one of the major impediments to the gene therapy of cancer.27 Alternatively, the high selectivity of the suicide gene to MSI cells might enable the application of the gene in a preventive fashion. Thus, the introduction of VP22(C15)FCU1 in normal colonic stem cells of hereditary nonpolyposis colorectal cancer patients, followed by periodic 5-FC treatment, will not harm these MSS cells but may abrogate the development of MSI carcinomas at an early stage. The limited number of colonic stem cells28 would have the added advantage of minimizing the adverse risks associated with genomic integration of the suicide gene (i.e., activation of an oncogene or inactivation of a tumor suppressor gene). It should be noted, however, that to our knowledge no efficient gene delivery system currently exists for these cells. In a similar fashion, Lynch syndrome–related stomach cancer and endometrial cancer may be prevented or treated using VP22(C15)FCU1 or other out-of-frame suicide genes.

In conclusion, a very low expression of VP22(C15)FCU1, combined with 5-FC administration, induces strong and selective toxicity to mouse tumors expressing VP22FCU1, in the absence of systemic toxicity. For these reasons, out-of-frame enzyme/prodrug combinations hold promise for the targeted and highly potent therapy, or even prevention, of MSI cancer in humans.

Materials and Methods

Cell lines. 293T Lα cells that are facultatively MMR deficient, by culture in the presence of the tetracycline derivative doxycycline (Dox), were provided by Josef Jiricny (Institute of Molecular Cancer Research, University of Zürich) and cultured as described.25 129/OLA ES cells [Msh2+/−, Msh2−/− (ref. 29), and Msh6−/− (ref. 30)] were cultured on irradiated feeder cells as described.29

Construction of suicide vectors and expressing cell lines. The amino acid stretch 101–105 (APPPP) of VP22 is encoded by the DNA sequence GCGCCTCCGCCACCC. By PCR, we mutated this stretch to a microsatellite-containing GCCCCCCCCCCCCCC (C14) stretch without altering the encoded amino acid sequence. To the 5′ end of VP22 we fused a FLAG epitope that enables its detection. This VP22 derivative was fused (in frame) to the 5′ terminus of FCU1, cloned in expression vector pCI-neo (Promega, Leiden, the Netherlands). The resulting chimeric gene was called VP22(C14)FCU1 (Figure 1b).

293T Lα cells were transfected with the pCI-neoVP22(C14)FCU1 plasmid and its parent pCI-neoFCU1, together with a linear puromycin resistance marker, and puromycin (2 µg/ml)-resistant cells of each transfection were pooled, minimizing the risk of differences in expression between both suicide genes. The resulting polyclonal stable transfectants VF293T Lα and F293T Lα were cultured under continuous puromycin pressure (1 µg/ml).

Analysis of drug sensitivity and bystander effects. To determine drug sensitivities, 293T Lα–derived cells were precultured in either the presence or absence of Dox (MMR-deficient and -proficient conditions, respectively). Then 4 × 104 cells per well were plated in triplicate in a 24-well plate with or without Dox and 5-FC or 5-FU were added. Cellular viability was determined after 6 days by resazurin staining.

To measure direct bystander effects, different ratios of 293T Lα and either VF293T Lα or F293T Lα cells were prepared and 4 × 104 cells per well were plated in triplicate in a 24-well plate in medium containing 1 mmol/l 5-FC. Viability was assessed as described earlier.

Conditioned medium was prepared by seeding VF293T Lα, F293T Lα, or 293T Lα cells (1 × 106 per ml of medium). After overnight culture, medium was collected and filtered with a 0.22-µm filter. The media were supplemented with 1 mmol/l 5-FC and either added directly, or after preincubation, in triplicate to 2 × 104 of 293T Lα cells per well in a 24-well culture plate. Viability was assessed as described earlier.

Construction and gene targeting of ES cell lines. Using mutagenic oligonucleotides, out-of-frame derivatives of the pMC1neo gene were constructed in which a G10 microsatellite was inserted downstream of the start codon of the neo gene, The inactive neo allele was introduced at the genomic Rb locus as described (ref. 31 and M. Aarts, M. Dekker, S. de Vries, and H. te Riele, manuscript in preparation) in Msh2−/−, Msh6−/− and wild-type ES cell lines.

By PCR we generated a derivative of VP22(C14)FCU1, called VP22(C15)FCU1, that carries one additional deoxycytosine within the mono deoxycytosine microsatellite in the VP22(C14) moiety. This insertion renders the downstream moiety of the chimeric gene out-of-frame (Figure 1b). VP22(C15)FCU1 was cloned under control of the CAG promoter32 in the Rosa26 gene–targeting vector pR3PCR that carries an upstream zeocin resistance gene preceded by a splice acceptor site, allowing the enrichment for correct gene-targeting events33 (Figure 1c). Gene targeting and analysis of targeting events were performed as described33 (Figure 1c). The resulting ES cell line V(+1)FMsh2+/− ES contains a single copy of the pCAG-driven VP22(C15)FCU1 out-of-frame chimeric gene integrated at the Rosa26 locus.

Msh2-deficient derivatives of V(+1)FMsh2+/−, called V(+1)FMsh2−/− lines 1–3, were isolated by a brief 6-thioguanine treatment following an established procedure.34 Loss of heterozygosity at Msh2 was verified by a multiplex PCR.28 V(+1)FMsh2+/− and V(+1)FMsh2−/− ES cells were grown at a low density in the presence of zeocin (75 µg/ml) and of 0.01 mmol/l 5-FC, to purge the cells from VP22FCU1-expressing cells.

Microsatellite slippage rates in ES cells. Single cells were expanded to cultures of 106–108 cells. The number of G418-resistant cells that accumulated during culturing was determined by seeding between 105 and 2 × 106 cells in medium containing 400 µg/ml of G418. The mutation rate p was determined according to Luria and Delbrück fluctuation analysis using the formula: 0.6 × G418total/C = Nxpxlog(CxNxp), where G418total = the total number of G418-resistant cells in a culture expanded to N cells, C = the number of different cultures, and p = the number of new mutations per cell division.35

To investigate the production of VP22FCU1 in MSS and MSI ES cell lines, 2 × 106 cells of the V(+1)FMsh2+/− line and V(+1)FMsh2−/− ES lines 1–3 were cultured in the absence of 5-FC to allow the reversion of VP22(C15)FCU1. At 4-day intervals, 4 × 104 cells of each line were plated in triplicate per well in 24-well plate in medium containing 0.01 mmol/l of 5-FC. After 5 days, the number of adherent cells was counted. Additionally, viability of these adherent cells was measured by clonal survival.33 Furthermore, cell lysates were made at each time point for analysis of VP22FCU1 expression by western blotting, using FLAG epitope antibody M2 (1:1,000 dilution; Sigma-Aldrich, Zwijndrecht, the Netherlands).

To estimate the VP22(C15)FCU1 reversion rate, V(+1)Msh2−/− ES cell line 1 was purged with 5-FC, subsequently cultured 20 days in the presence of zeocin and then seeded at clonal density. Individual colonies were expanded briefly and analyzed for VP22FCU1 expression by western blotting. The reversion frequency was calculated by dividing the number of highly VP22FCU1-expressing colonies that have undergone reversion prior to plating by the total number of colonies analyzed. The reversion rate was estimated by dividing the reversion frequency by the estimated number of cellular doublings during the 20-day culture period × 0.5.35

Gene therapy of MSI tumors. V(+1)FMsh2−/− (MSI) ES cells were purged with 5-FC up to 2 days before administering 4 × 106 cells subcutaneously to one flank of 26 athymic CD-1-Foxn1nu mice (Charles River, Maastricht, the Netherlands). The other flank of the mice was injected with V(+1)FMsh2+/− ES cells that serve as an MSS control (day 0). The use of these immune-deficient mice prevents the development of a T-cell response induced by frameshifted neopeptides in the resulting MSI teratomas.36 Starting at 1 week after injection, 5-FC (in PBS, 500 mg/kg body weight) was administered daily by intraperitoneal injection to a randomly chosen half of the population whereas the other half was injected with PBS alone (see Figure 5). The development and volume of teratomas were measured daily and animals were killed when the volume of the tumor exceeded (approximately) 1 cm3. After 9 days of treatment, the 5-FC- and PBS-treated groups were inverted and treatment was continued until day 31, or when the size of a tumor exceeded a volume of (approximately) 1 cm3. After euthanasia, the mice were autopsied and the tumors were snap frozen. A permit (DEC 07194) was obtained for these experiments from the Leiden University Animal Ethics committee.

Immunohistochemical detection of the FLAG epitope was done on acetone-fixed cryosections. After pretreatment with saponin sections were blocked and incubated with anti-FLAG antibody (1:200, 1 hour at room temperature). Incubation with the secondary antibody [biotinylated rabbit anti-mouse IgG1, 1:100, 1 hour at room temperature; Zymed] was followed by visualization using fluorescein-conjugated streptavidin (1:400, 1 hour at room temperature; Jackson ImmunoResearch, Newmarket, UK) in PBS. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole.

SUPPLEMENTARY MATERIALTable S1. IC50 of 5-FU and 5-FC for VF293T Lα and F293T Lα cells under mismatch repair-proficient and -deficient conditions.Table S2. Bystander effects induced by VP22FCU1 and FCU1.

Supplementary Material

Table S1.

IC50 of 5-FU and 5-FC for VF293T Lα and F293T Lα cells under mismatch repair-proficient and -deficient conditions.

Click here for supplemental data (259.5KB, doc)
Table S2.

Bystander effects induced by VP22FCU1 and FCU1.

Click here for supplemental data (34KB, doc)

Acknowledgments

Part of this work was granted by the European Union (QLG1-CT-2000-001230). C.F. was in part funded by the Portuguese Foundation for Science and Technology (SFRH/BPD/34633/2007). We thank Transgene (Strasbourg, France) for the FCU1-expression vector and J. Jiricny for helpful discussions. R. Hoeben, A. de Vries, and J. Jansen are acknowledged for critical comments on the manuscript.

REFERENCES

  1. Li GM. Mechanisms and functions of DNA mismatch repair. Cell Res. 2008;18:85–98. doi: 10.1038/cr.2007.115. [DOI] [PubMed] [Google Scholar]
  2. Hsieh P., and , Yamane K. DNA mismatch repair: molecular mechanism, cancer, and ageing. Mech Ageing Dev. 2008;129:391–407. doi: 10.1016/j.mad.2008.02.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Imai K., and , Yamamoto H. Carcinogenesis and microsatellite instability: the interrelationship between genetics and epigenetics. Carcinogenesis. 2008;29:673–680. doi: 10.1093/carcin/bgm228. [DOI] [PubMed] [Google Scholar]
  4. Meyers M, Hwang A, Wagner MW, Bruening AJ, Veigl ML, Sedwick WD, et al. A role for DNA mismatch repair in sensing and responding to fluoropyrimidine damage. Oncogene. 2003;22:7376–7388. doi: 10.1038/sj.onc.1206941. [DOI] [PubMed] [Google Scholar]
  5. Jover R, Zapater P, Castells A, Llor X, Andreu M, Cubiella J, et al. The efficacy of adjuvant chemotherapy with 5-fluorouracil in colorectal cancer depends on the mismatch repair status. Eur J Cancer. 2009;45:365–373. doi: 10.1016/j.ejca.2008.07.016. [DOI] [PubMed] [Google Scholar]
  6. Liu A, Yoshioka K, Salerno V., and , Hsieh P. The mismatch repair-mediated cell cycle checkpoint response to fluorodeoxyuridine. J Cell Biochem. 2008;105:245–254. doi: 10.1002/jcb.21824. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Greco O., and , Dachs GU. Gene directed enzyme/prodrug therapy of cancer: historical appraisal and future prospectives. J Cell Physiol. 2001;187:22–36. doi: 10.1002/1097-4652(2001)9999:9999<::AID-JCP1060>3.0.CO;2-H. [DOI] [PubMed] [Google Scholar]
  8. Young LS, Searle PF, Onion D., and , Mautner V. Viral gene therapy strategies: from basic science to clinical application. J Pathol. 2006;208:299–318. doi: 10.1002/path.1896. [DOI] [PubMed] [Google Scholar]
  9. Portsmouth D, Hlavaty J., and , Renner M. Suicide genes for cancer therapy. Mol Aspects Med. 2007;28:4–41. doi: 10.1016/j.mam.2006.12.001. [DOI] [PubMed] [Google Scholar]
  10. Erbs P, Regulier E, Kintz J, Leroy P, Poitevin Y, Exinger F, et al. In vivo cancer gene therapy by adenovirus-mediated transfer of a bifunctional yeast cytosine deaminase/uracil phosphoribosyltransferase fusion gene. Cancer Res. 2000;60:3813–3822. [PubMed] [Google Scholar]
  11. Chung-Faye GA, Chen MJ, Green NK, Burton A, Anderson D, Mautner V, et al. In vivo gene therapy for colon cancer using adenovirus-mediated, transfer of the fusion gene cytosine deaminase and uracil phosphoribosyltransferase. Gene Ther. 2001;8:1547–1554. doi: 10.1038/sj.gt.3301557. [DOI] [PubMed] [Google Scholar]
  12. Graepler F, Lemken ML, Wybranietz WA, Schmidt U, Smirnow I, Gross CD, et al. Bifunctional chimeric SuperCD suicide gene -YCD: YUPRT fusion is highly effective in a rat hepatoma model. World J Gastroenterol. 2005;11:6910–6919. doi: 10.3748/wjg.v11.i44.6910. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Khatri A, Zhang B, Doherty E, Chapman J, Ow K, Pwint H, et al. Combination of cytosine deaminase with uracil phosphoribosyl transferase leads to local and distant bystander effects against RM1 prostate cancer in mice. J Gene Med. 2006;8:1086–1096. doi: 10.1002/jgm.944. [DOI] [PubMed] [Google Scholar]
  14. Elliott GD., and , Meredith DM. The herpes simplex virus type 1 tegument protein VP22 is encoded by gene UL49. J Gen Virol. 1992;73:723–726. doi: 10.1099/0022-1317-73-3-723. [DOI] [PubMed] [Google Scholar]
  15. Lemken ML, Wolf C, Wybranietz WA, Schmidt U, Smirnow I, Bühring HJ, et al. Evidence for intercellular trafficking of VP22 in living cells. Mol Ther. 2007;15:310–319. doi: 10.1038/sj.mt.6300013. [DOI] [PubMed] [Google Scholar]
  16. Mori T, Mineta Y, Aoyama Y., and , Sera T. Efficient secretion of the herpes simplex virus tegument protein VP22 from living mammalian cells. Arch Virol. 2008;153:1191–1195. doi: 10.1007/s00705-008-0094-x. [DOI] [PubMed] [Google Scholar]
  17. Lemken ML, Graepler F, Wolf C, Wybranietz WA, Smirnow I, Schmidt U, et al. Fusion of HSV-1 VP22 to a bifunctional chimeric SuperCD suicide gene compensates for low suicide gene transduction efficiencies. Int J Oncol. 2007;30:1153–1161. [PubMed] [Google Scholar]
  18. Waehler R, Russell SJ., and , Curiel DT. Engineering targeted viral vectors for gene therapy. Nat Rev Genet. 2007;8:573–587. doi: 10.1038/nrg2141. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Meyers M, Wagner MW, Hwang HS, Kinsella TJ., and , Boothman DA. Role of the hMLH1 DNA mismatch repair protein in fluoropyrimidine-mediated cell death and cell cycle responses. Cancer Res. 2001;61:5193–5201. [PubMed] [Google Scholar]
  20. Kielman MF, Rindapää M, Gaspar C, van Poppel N, Breukel C, van Leeuwen S, et al. Apc modulates embryonic stem-cell differentiation by controlling the dosage of beta-catenin signaling. Nat Genet. 2002;32:594–605. doi: 10.1038/ng1045. [DOI] [PubMed] [Google Scholar]
  21. Meyers M, Wagner MW, Mazurek A, Schmutte C, Fishel R., and , Boothman DA. DNA mismatch repair-dependent response to fluoropyrimidine-generated damage. J Biol Chem. 2005;280:5516–5526. doi: 10.1074/jbc.M412105200. [DOI] [PubMed] [Google Scholar]
  22. Fischer F, Baerenfaller K., and , Jiricny J. 5-Fluorouracil is efficiently removed from DNA by the base excision and mismatch repair systems. Gastroenterology. 2007;133:1858–1868. doi: 10.1053/j.gastro.2007.09.003. [DOI] [PubMed] [Google Scholar]
  23. Goodfellow PJ, Buttin BM, Herzog TJ, Rader JS, Gibb RK, Swisher E, et al. Prevalence of defective DNA mismatch repair and MSH6 mutation in an unselected series of endometrial cancers. Proc Natl Acad Sci USA. 2003;100:5908–5913. doi: 10.1073/pnas.1030231100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Sia EA, Kokoska RJ, Dominska M, Greenwell P., and , Petes TD. Microsatellite instability in yeast: dependence on repeat unit size and DNA mismatch repair genes. Mol Cell Biol. 1997;17:2851–2858. doi: 10.1128/mcb.17.5.2851. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Cejka P, Marra G, Hemmerle C, Cannavó E, Storchova Z., and , Jiricny J. Differential killing of mismatch repair-deficient and -proficient cells: towards the therapy of tumors with microsatellite instability. Cancer Res. 2003;63:8113–8117. [PubMed] [Google Scholar]
  26. Power AT., and , Bell JC. Taming the Trojan horse: optimizing dynamic carrier cell/oncolytic virus systems for cancer biotherapy. Gene Ther. 2008;15:772–779. doi: 10.1038/gt.2008.40. [DOI] [PubMed] [Google Scholar]
  27. Greco O, Scott SD, Marples B., and , Dachs GU. Cancer gene therapy: “delivery, delivery, delivery”. Front Biosci. 2002;7:d1516–d1524. doi: 10.2741/A731. [DOI] [PubMed] [Google Scholar]
  28. Barker N, van Es JH, Kuipers J, Kujala P, van den Born M, Cozijnsen M, et al. Identification of stem cells in small intestine and colon by marker gene Lgr5. Nature. 2007;449:1003–1007. doi: 10.1038/nature06196. [DOI] [PubMed] [Google Scholar]
  29. de Wind N, Dekker M, Berns A, Radman M., and , te Riele H. Inactivation of the mouse Msh2 gene results in mismatch repair deficiency, methylation tolerance, hyperrecombination, and predisposition to cancer. Cell. 1995;82:321–330. doi: 10.1016/0092-8674(95)90319-4. [DOI] [PubMed] [Google Scholar]
  30. de Wind N, Dekker M, Claij N, Jansen L, van Klink Y, Radman M, et al. HNPCC-like cancer predisposition in mice through simultaneous loss of Msh3 and Msh6 mismatch-repair protein functions. Nat Genet. 1999;23:359–362. doi: 10.1038/15544. [DOI] [PubMed] [Google Scholar]
  31. te Riele H, Maandag ER., and , Berns A. Highly efficient gene targeting in embryonic stem cells through homologous recombination with isogenic DNA constructs. Proc Natl Acad Sci USA. 1992;89:5128–5132. doi: 10.1073/pnas.89.11.5128. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Araki K, Imaizumi T, Okuyama K, Oike Y., and , Yamamura K. Efficiency of recombination by Cre transient expression in embryonic stem cells: comparison of various promoters. J Biochem. 1997;122:977–982. doi: 10.1093/oxfordjournals.jbchem.a021860. [DOI] [PubMed] [Google Scholar]
  33. Hendriks G, Calléja F, Vrieling H, Mullenders LH, Jansen JG., and , de Wind N. Gene transcription increases DNA damage-induced mutagenesis in mammalian stem cells. DNA Repair (Amst) 2008;7:1330–1339. doi: 10.1016/j.dnarep.2008.04.015. [DOI] [PubMed] [Google Scholar]
  34. Borgdorff V, van Hees-Stuivenberg S, Meijers CM., and , de Wind N. Spontaneous and mutagen-induced loss of DNA mismatch repair in Msh2-heterozygous mammalian cells. Mutat Res. 2005;574:50–57. doi: 10.1016/j.mrfmmm.2005.01.021. [DOI] [PubMed] [Google Scholar]
  35. Foster PL. Methods for determining spontaneous mutation rates. Meth Enzymol. 2006;409:195–213. doi: 10.1016/S0076-6879(05)09012-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Schwitalle Y, Kloor M, Eiermann S, Linnebacher M, Kienle P, Knaebel HP, et al. Immune response against frameshift-induced neopeptides in HNPCC patients and healthy HNPCC mutation carriers. Gastroenterology. 2008;134:988–997. doi: 10.1053/j.gastro.2008.01.015. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Table S1.

IC50 of 5-FU and 5-FC for VF293T Lα and F293T Lα cells under mismatch repair-proficient and -deficient conditions.

Click here for supplemental data (259.5KB, doc)
Table S2.

Bystander effects induced by VP22FCU1 and FCU1.

Click here for supplemental data (34KB, doc)

Articles from Molecular Therapy: the Journal of the American Society of Gene Therapy are provided here courtesy of The American Society of Gene & Cell Therapy

RESOURCES