Cytosine deaminase expressing human mesenchymal stem cells mediated tumour regression in melanoma bearing mice


  • Lucia Kucerova,

    Corresponding author
    1. Laboratories of Molecular Oncology, Cancer Research Institute of Slovak Academy of Sciences, Bratislava, Slovakia
    • Laboratory of Molecular Oncology, Cancer Research Institute, Slovak Academy of Sciences, Vlarska 7, 833 91 Bratislava, Slovakia.
    Search for more papers by this author
    • Both investigators contributed equally and should be considered as senior authors.

  • Miroslava Matuskova,

    1. Laboratories of Molecular Oncology, Cancer Research Institute of Slovak Academy of Sciences, Bratislava, Slovakia
    Search for more papers by this author
    • Both investigators contributed equally and should be considered as senior authors.

  • Andrea Pastorakova,

    1. Laboratories of Molecular Oncology, Cancer Research Institute of Slovak Academy of Sciences, Bratislava, Slovakia
    Search for more papers by this author
  • Silvia Tyciakova,

    1. Laboratories of Molecular Oncology, Cancer Research Institute of Slovak Academy of Sciences, Bratislava, Slovakia
    Search for more papers by this author
  • Jana Jakubikova,

    1. Laboratories of Immunology, Cancer Research Institute of Slovak Academy of Sciences, Bratislava, Slovakia
    Search for more papers by this author
  • Roman Bohovic,

    1. Laboratories of Molecular Oncology, Cancer Research Institute of Slovak Academy of Sciences, Bratislava, Slovakia
    Search for more papers by this author
  • Veronika Altanerova,

    1. Laboratories of Molecular Oncology, Cancer Research Institute of Slovak Academy of Sciences, Bratislava, Slovakia
    Search for more papers by this author
  • Cestmir Altaner

    1. Laboratories of Molecular Oncology, Cancer Research Institute of Slovak Academy of Sciences, Bratislava, Slovakia
    Search for more papers by this author



Previously, we validated capability of human adipose tissue-derived mesenchymal stem cells (AT-MSC) to serve as cellular vehicles for gene-directed enzyme prodrug molecular chemotherapy. Yeast fusion cytosine deaminase : uracil phosphoribosyltransferase expressing AT-MSC (CDy-AT-MSC) combined with systemic 5-fluorocytosine (5FC) significantly inhibited growth of human colon cancer xenografts. We aimed to determine the cytotoxic efficiency to other tumour cells both in vitro and in vivo.


CDy-AT-MSC/5FC-mediated proliferation inhibition against a panel of human tumour cells lines was evaluated in direct and indirect cocultures in vitro. Antitumour effect was tested on immunodeficient mouse model in vivo.


Although culture expansion of CDy-AT-MSC sensitized these cells to 5FC mediated suicide effect, expanded CDy-AT-MSC/5FC still exhibited strong bystander cytotoxic effect towards human melanoma, glioblastoma, colon, breast and bladder carcinoma in vitro. Most efficient inhibition (91%) was observed in melanoma A375 cell line when directly cocultured with 2% of therapeutic cells CDy-AT-MSC/5FC. The therapeutic paradigm of the CDy-AT-MSC/5FC system was further evaluated on melanoma A375 xenografts on nude mice in vivo. Complete regression in 89% of tumours was achieved when 20% CDy-AT-MSC/5FC were co-injected along with tumour cells. More importantly, systemic CDy-AT-MSC administration resulted in therapeutic cell homing into subcutaneous melanoma and mediated tumour growth inhibition.


CDy-AT-MSC capability of targeting subcutaneous melanoma offers a possibility to selectively produce cytotoxic agent in situ. Our data further demonstrate beneficial biological properties of AT-MSC as a cellular vehicle for enzyme/prodrug therapy approach to molecular chemotherapy. Copyright © 2008 John Wiley & Sons, Ltd.


Worldwide clinical studies with cytosine deaminase (CD) in combination with prodrug 5-fluorocytosine (5FC) have shown the safety of this approach, but limited therapeutic benefit on the other hand. Therefore there is still a need for more efficient and targeted delivery of therapeutic suicide gene(s) for successful cancer gene therapy 1. There have been several attempts to overcome disadvantages of direct intravenous or intratumoural virus injection. The use of cell-based delivery approach was reported to have significant antitumour efficiency in vivo. It was shown that endothelial progenitor cells producing CD could efficiently inhibit tumour growth in a model of lung and melanoma brain metastasis in vitro and in vivo2, 3. Moreover, Kim et al.4 demonstrated a significant inhibition of tumour growth in a medulloblastoma model with therapeutic cells derived from engineered embryonic neural progenitor cells producing CD. A different enzyme/prodrug combination such as secreted rabbit carboxylesterase/CPT-11 delivered by neural stem and progenitor cells into neuroblastoma led even to long-term disease-free survival in a model of disseminated neuroblastoma 5, 6. These cells pose a valuable stem cell type due to their capacity to differentiate into various neural cell types, thus having potential supportive role in brain tissue regeneration; however, there are serious considerations with respect to the adverse effects of cells of embryonic origin, such as tumourigenicity.

Recently, it was shown that the mesenchymal stem cells derived from bone marrow are capable of directed migration towards the tumours of various types and origin, not only within the brain, when injected intracarotidally, but also upon systemic administration into various types of subcutaneously induced tumour of neural and non-neural origin 7–11. The inherent tumour-targeting capability of these cells is a prerequisite for the cells to become suitable candidate for tumour site restricted production of cytotoxic compound, thus sparing the healthy tissue from damage. Therefore, we and others hypothesized that mesenchymal stem cells could be more suitable vehicle for gene-directed enzyme prodrug therapy (GDEPT) of cancer compared to other stem cells sources 7, 12–14. Furthermore, it has been shown that there are sources of mesenchymal stem cells other than bone marrow, such as adipose tissue, which makes these cells accessible for autologous use without any ethical concerns. Human mesenchymal stromal cells derived from adipose tissue (AT-MSC) were described not to differ significantly from the better characterized bone marrow-derived MSC; however, they can be obtained with less risk and discomfort to the donor, and are expandable to the desired amount even repeatedly if needed 15–18. AT-MSC were anticipated to be of a great clinical interest in regenerative medicine for the treatment of many diseases such as skeletal tissue repair 19, ischaemic disease 20 or neurological disorders 21. Nevertheless, AT-MSC have been tested as gene delivery vehicles for GDEPT of cancer to a limited extent both in vitro22, 23 and in vivo14.

In the present study, we aimed to extend our pilot study 14 to other tumour cell types and evaluate any potential therapeutic effect of yeast fusion CDy expressing human mesenchymal stromal cells derived from adipose tissue (CDy-AT-MSC). Bystander cytotoxicity was mediated towards many different human cancer cell lines by CDy-AT-MSC in the presence of prodrug 5FC in vitro, even though the culture expanded therapeutic CDy-AT-MSC cells were rather sensitive to a suicide CDy gene effect. Moreover, melanoma growth inhibition in vivo highlights the potential of expanded AT-MSC to serve as a therapeutic tumour-targeted self-inactivating delivery system of suicide gene(s) in the molecular chemotherapy of cancer.

Materials and methods

Cell lines and chemicals

Human colon adenocarcinoma cells HT-29 were purchased from ECACC (Porton Down, UK) (ECACC no. 91072201). HT-29, Human breast adenocarcinoma cells MDA-MB-231 (kindly provided by Dr J. Sedlak, CRI SAS, Bratislava, Slovakia), MDA-MB-361 (kindly provided by Professor J.Kovarik, MMCI, Brno, Czech Republic), colon carcinoma HCT 116, breast tumour T47D, brain astrocytoma MOG-G-UVW, bladder carcinoma EJ138 (kindly provided by Dr S.Squires, Zoology Department, Cambridge University, Cambridge, UK), and glioblastoma multiforme 8-MG-BA (kindly provided by Dr A.Perzelova, Medical Faculty, Comenius University, Bratislava, Slovakia), A375, M4Beu and human fibroblasts (kindly provided by Dr J. Bizik, CRI SAS, Bratislava, Slovakia), mouse fibroblast NIH-3T3 cells, retroviral packaging mouse cell lines GP + E-86 and GP + envAM12 (kindly provided by Dr J. Bies NCI NIH, Bethesda, MD, USA) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% FCS and Antibiotic-Antimycotic mix (Gibco BRL, Gaithesburg, MD, USA). Cells were kept in humidified atmosphere and 5% CO2 at 37 °C.

All chemicals were purchased from Sigma (St Louis, MO, USA) if not stated otherwise.

Construction of retroviral vector containing yeast CDy and preparation of virus producing cells

Construction of recombinant retrovirus pST2 was performed by standard polymerase chain reaction (PCR) cloning techniques as described 14. GP + E-86 cells were transfected with 2 µg of pST2 plasmid using Effectene (Qiagen, Hilden, Germany). Titre of viral inoculum used for AT-MSC transduction was determined by titration on NIH-3T3 fibroblasts. 5 × 104 NIH-3T3 per well were plated in six-well plates 1 day before infection. Infection was performed with 0.5 ml of virus containing medium supplemented with 5 µg/ml protamine sulphate. Cells were cultivated in the presence of 0.5 mg/ml of G418 for 10 days. Macroscopic colonies were visualised by May–Grünwald staining and counted. Initial viral titres post-transfection in GP + E-86/pST2 cells were 4.2–7 × 104. To increase viral titre and produce amphotropic retrovirus particles, three rounds of repeated ping-pong transduction of GP + envAM12/pST2 and GP + E-86/pST2 were performed as previously described 14, 24, 25. The titre of GP + envAM12/pST2 increased by two orders of magnitude up to 1 × 106 to 2.8 × 106 CFU/ml.

Virus-containing medium for transduction was collected from 90% confluent cultures of GP + envAM12/ pST2 cells, filtered through 0.45-µm filters and used either fresh or kept frozen at − 80 °C.

AT-MSC culture and retrovirus transduction

AT-MSC cells from five different donors (aged 21–35 years) were isolated from lipoaspirate using a collagenase type VII digestion and plastic adherence technique as previously described 14, 26. Material was obtained from healthy individuals undergoing elective lipoaspiration, who provided an informed consent. Conclusions were drawn from the similar results obtained in the experiments performed as described below with each isolate if not specified otherwise.

To prepare mesenchymal stem cells expressing cytosine deaminase (CDy-AT-MSC), sub-confluent cultures of AT-MSC were transduced three times in three consecutive days with virus-containing media from GP + envAM12/pST2 cells supplemented with 5 µg/ml of protamine sulphate. Cells were cultivated in selection media containing 0.5 mg/ml of G418 for 10 days and expanded to obtain the same batch of cells in passages 15–20 for the experiments conducted in vitro and in vivo. Human fibroblasts expressing CDy (CDy-Fib) were prepared as described above.

Immunophenotyping and MSC differentiation

AT-MSC immunophenotype and differentiation capabilities were determined as previously described 14. Cells were labelled with antibodies against CD14, CD29, CD34, CD44, CD45, CD90 and CD105. Differentiation into osteogenic and adipogenic lineage was tested in each isolate and after long-term culture expansion of AT-MSC. Human fibroblasts were used as a negative control.

Cell proliferation assays and direct coculture of human tumour cells and CDy-AT-MSC in vitro

AT-MSC sensitivity to 5FC and 5FU and direct cocultures of tumour cells with CDy-AT-MSC were performed as previously described 14.

Fluorescent evaluation of the bystander effect

Adherent tumour cells were labelled with 5 µM carboxy-fluorescein diacetate, succinimidyl ester (CFDA-SE; Molecular Probes, Eugene, OR, USA) in a serum-free DMEM for 15 min at 37 °C. Medium was replaced for standard culture medium for overnight incubation. On the next day, triplicates of target CFDA-SE tumour cells (5 × 103 per well) were seeded with the same number of CDy-AT-MSC into 96-well plates. Culture medium with increasing 5FC concentrations was added to the mixtures 4 h later, incubated for 3 days and plates were washed with phosphate-buffered saline (PBS) three times. Fluorescence intensity was determined using a FluoStar Optima plate reader (BMG Labtechnologies, Offenburg, Germany). Results were expressed as means of relative fluorescence, where the fluorescence of the cell mixtures without 5FC was set to 1 by default.

Indirect cocultures

HT-29 cells were plated on six-well plate (5 × 104 cells per well). AT-MSC were plated on 0.4 µm inserts (5 × 104 cells per insert) (Nalge Nunc International, Rochester, NY, USA) after 24 h of cultivation, inserts with transduced CDy-AT-MSC or untransduced AT-MSC were transferred into wells with HT-29 cells. Medium was replaced and 5FC (100 µg/ml) was added to respective wells. Apoptosis and necrosis was evaluated after 24, 48 and 72 h of cultivation. Apoptotic cells were stained with Phycoerythrin-labelled Annexin V (Chemicon, Temecula, CA, USA); dead cells were detected with 7-AAD viability dye (Immunotech, Marseille, France). Stained cells were analysed using an EPICS ALTRA flow cytometer (Beckman Coulter, Fullerton, CA, USA) equipped with the Expo 32 program.

Analysis of gene expression

Total RNA was isolated from 2 × 106 cells by RNeasy mini kit (Qiagen) and treated with RNase-free DNase (Qiagen). RNA was reverse transcribed with RevertAid H minus First Strand cDNA Synthesis Kit (Fermentas, Hanover, MD, USA). Two hundred nanograms of cDNA was subject to standard PCR with 35 cycles and gel resolved on 4% MetaPhor Agarose (Lonza, Rockland, ME, USA) for qualitative analysis.

Real-time PCR was performed in 1× PCR Master Mix (Fermentas), 2.5 mM MgCl2, 0.16 µM respective primers (Table 1), 1× Sybr GREEN (Molecular Probes) and 500 ng of template cDNA on RotorGene 2000 (Corbett Research, Sydney, Australia) and analysed by RotorGene Software, version 4.6. Gene expression was compared using delta cycle threshold (ΔCt = CtGOICtGAPDH) values for respective cells examined, where GAPDH expression was taken as endogenous reference gene (GOI = gene of interest). Change in gene expression was calculated according to the formula: fold increase = (reaction efficiency*2)ΔΔCt, where ΔΔCt = ([CtGOI(Fib)CtGAPDH(Fib)]− [CtGOI(AT−MSC)CtGAPDH(AT−MSC)]). Analysis was performed twice in triplicates and data are expressed as the means ± SE.

Table 1. Primer sequences
 Forward primer (5′ to 3′)Reverse primer (5′ to 3′)Product

Experiments in vivo

Six- to 8-week-old athymic nude mice (Balb/c-nu/nu) were used in accordance with institutional guidelines under the approved protocols. The following cell suspensions were injected in coinjection studies: 1.5 × 106A375 + 1.5 × 105 CDy-AT-MSC, 1.5 × 106A375 + 1.5 × 105 AT-MSC, 1.5 × 106A375 + 3 × 105 CDy-AT-MSC, 1.5 × 106A375 + 3 × 105 AT-MSC, and 1.5 × 106 A375 cells (in 100 µl of PBS s.c. into the flank). Animals were treated with 500 mg/kg/day of 5FC diluted in PBS i.p. in two rounds of 5 consecutive days with a break of 2 days. For the tumourigenicity test, suspension of expanded late passage 106 AT-MSC in 100 µl of PBS per animal (n = 3) was injected s.c.

In an independent study, tumours were induced with 1.5 × 106 A375 cells s.c. Animals (n = 6 per group) were divided into five groups according to treatment: CDy-AT-MSC i.v. or i.p./5FC, CDy-Fib i.v./5FC, AT-MSC i.v./5FC or 5FU controls. Mice received three cell doses spaced 7 days apart on days 3, 10 and 17. Suspension of 106 CDy-AT-MSC, or 106 CDy-Fib in 200 µl of PBS per each animal and dose was injected i.v. into the lateral tail vein or intraperitoneally. Four animal groups were treated with daily dose of 500 mg/kg 5FC i.p. for 4 consecutive days starting on days 5, 12 and 19 in three rounds. Control group received 20 mg/kg/day of 5FU i.p. in three consecutive 5-day rounds with a 2-day break starting on day 5.

Tumours were measured by caliper and volume was calculated according to formula: volume = length × width2/2. Animals were sacrificed at the point when the tumours exceeded 1 cm in diameter in animals within the control group. The results were evaluated as the mean of volume or weight ± SE.

For the CDy-AT-MSC targeting analysis, tumours were induced with 1.5 × 106 A375 cells s.c. and 106 CDy-AT-MSC were administered either intravenously, i.p. or s.c. into the other flank 3 days later. DNA was isolated from excised tumours on day 10 (7 days post-injection). Five hundred nanograms of tumour DNA was subjected to PCR amplification for 35 cycles with CDy primers to amplify specific product. Template integrity was checked with primers specific for human repetitive Alu sequences.


Characterization of therapeutic cells CDy-AT-MSC

The transgene of interest was introduced into early passage AT-MSC by means of retrovirus transduction. Replication deficient bicistronic retrovirus vector with the yeast CD fused to uridine phosphoribosyltransferase gene (CDy) (Figure 1A) was transfected into mouse packaging cell line GP + E-86 cells. Viability of transfected cells in the presence of 100 µg/ml of 5FC decreased to 20% within 4 days of cultivation (Figure 1B), confirming the biological functionality of the transgene. To prepare mesenchymal stem cells expressing cytosine deaminase (CDy-AT-MSC), semi-confluent cultures of AT-MSC were transduced with high titre retrovirus-containing media. Transduction efficiency of each AT-MSC isolate was high, close to 90 ± 5%, as estimated from almost complete G418 resistance post-transduction in CDy-AT-MSC cultures. Expression of the transgene was confirmed to be maintained for at least 18 passages post-transduction (Figure 1C). CDy-AT-MSC cells were found to be stably transduced as expected, and the ΔΔCt method did not show significant difference in the CDy expression levels by quantitative RT-PCR.

Figure 1.

Stable transduction of human mesenchymal stem cells derived from adipose tissue (AT-MSC) by retroviral vector bearing yeast fusion cytosine deaminase gene. (A) Schematic diagram of plasmid pST2 constructed as replication deficient retrovirus with fusion gene yeast cytosine deaminase and phosphoribosyltransferase Fcy : Fur and harboring neomycin resistance. (B) Murine packaging cell line transfected with plasmid pST2 exhibited high sensitivity to otherwise nontoxic 5FC indicative of expression of CDy transgene and its biological function. IC50 (GP + E-86/pST2) = 5 ng/ml of 5FC. Data are expressed as means ± SE. (C) AT-MSC at passage 2 were retrovirally transduced, cultured in G418 selection media for 10 days and propagated in standard media. Total RNA was isolated from CDy-AT-MSC cells at passages 3, 5 and 18 and subjected to RT-PCR to detect amplification of CDy specific sequence. GAPDH was used as RNA integrity control and endogenous reference gene. CDy expression was confirmed in all CDy-AT-MSC samples analysed but not in untransduced AT-MSC control (data not shown)

Retrovirus transduction, G418 selection and subsequent expansion rendered CDy expressing AT-MSC sensitive to prodrug 5FC (Figure 2A). Moreover, cultivation of CDy-AT-MSC directly in the presence of active cytotoxic drug 5FU has shown the increase of the suicide effect on the cells at high 5FU concentrations (Figure 2B). These data demonstrate the functionality of the phosphoribosyltransferase domain of the CDy transgene. Significant CDy-AT-MSC proliferation inhibition mediated by 5FC and 5FU sensitivity is in contrast to our previous finding showing that AT-MSC possess intrinsic mechanism for increased chemoresistance to CDy produced cytotoxic 5FU and its metabolites 14. When addressed directly, resistance of CDy-AT-MSC to suicide effect of CDy transgene was found to be inversely correlated with the increasing passage number corresponding to the decreasing 5FU resistance in respective AT-MSC cells (Figure S1).

Figure 2.

CDy transgene expression renders G418-selected culture expanded CDy-AT-MSC sensitive to 5FC. (A) CDy-AT-MSC (passage 18) were cultured in the presence of 150 µg/ml of 5FC for the indicated time and the proliferation was measured by standard MTS assay. Cell proliferation in medium without 5FC was taken as 100%. (B) Sensitivity to 5FU was compared between parental AT-MSC and transduced cells CDy-AT-MSC on day 5. Biological activity of phosphoribosyltransferase domain increased the sensitivity CDy-AT-MSC cells to 5FU, although only at high concentrations. Data are expressed as means ± SE

To exclude the possibility, that AT-MSC culture expansion in vitro could have been achieved at the expense of some AT-MSC characteristics, we examined their immunophenotype and differentiation capability. AT-MSC (passage 15) were consistently positive for surface markers CD29, CD44, CD90 and CD105 and negative for CD14, CD34 and CD45. AT-MSC at passage 15 were capable of differentiation into osteocytes and adipocytes in vitro (Figure S2). Cells did not exhibit any morphological changes and were found not to be tumourigenic upon subcutaneous injection in nude mice (data not shown). The expression analysis of enzymes involved in 5FU metabolism and resistance 27, 28 revealed that both low passage and high passage AT-MSC express thymidylate synthase (TS), dihydropyrimidine dehydrogenase (DPD), thymidylate phosphorylase (TP), orotate phosphoribosyltransferase (OPRT) (Figure 3A). However, expression of thymidylate phosphorylase TP (platelet-derived endothelial cell growth factor; PD-ECGF) was not found in human fibroblasts. Moreover, AT-MSC express significantly more dihydropyrimidine dehydrogenase DPD (Figure 3B), that might have also contributed to 5FU catabolism to produce inactive metabolite dihydrofluorouracil. The level of expression of orotate phosphoribosyltransferase and thymidylate synthase was not significantly different in these cells. Overexpression of multidrug resistance-associated protein MRP1 (ABCC1) was reported in 5FU resistant cancer cells 29, 30. ATP-binding cassette members ABC A3 (ATP-binding cassette, subfamily A, member 3), ABC G2 (Bcrp) and ABC B5 expression was also shown to be correlated with increased chemoresistance 31, 32. All these genes were found to be expressed in low passage AT-MSC (Figure 3C). We found that the AT-MSC culture expansion was accompanied by loss of ABC B5 expression in AT-MSC beyond passage 20. ABC B5 is not expressed in fibroblasts (Figure 3C), tumour cells HT-29 and MDA-MB-231 (data not shown).

Figure 3.

Expression of 5FU-related enzymes and multidrug resistance genes in AT-MSC. Isolated DNase-treated total RNA from AT-MSC passage 4, AT-MSC passage 20 and human fibroblasts was reverse transcribed and PCR amplified. GAPDH served as RNA integrity control and internal standard. (A) Low and high passage AT-MSC express TP, DPD, OPRT and TS. TP expression could not be detected in fibroblasts. (B) Expression level of 5FU-related enzymes and selected multidrug resistance genes was compared by real-time quantitative PCR using Sybr GREEN dye. GAPDH served as endogenous reference gene. Relative expression was calculated from ΔΔCt values by comparison with the respective gene expression in human fibroblasts. AT-MSC express a significantly higher level of DPD in comparison to fibroblasts. Data are expressed as means ± SE. (C) AT-MSC at low passage express ABC B5, whereas high passage AT-MSC and fibroblasts do not

CDy-AT-MSC/5FC mediated cytotoxicity, bystander effect and apoptosis induction in vitro

The crucial point for the effective AT-MSC-directed enzyme prodrug therapy would be the capability of therapeutic cells to exert a bystander cytotoxic effect on target cancer cells. To test the CDy-AT-MSC efficiency, these G418-selected and expanded cells (at passages 15 up to 20) were directly cocultured with target tumour cells in the presence of 5FC. Proliferation inhibition by almost 65% could be observed even if cells were mixed in a ratio 1 : 50 CDy-AT-MSC/5FC to MDA-MB-361 cells in direct cocultures after 5 days, which represents 2% of transduced therapeutic cells within the coculture (Figure 4A). This extent of proliferation inhibition could have been achieved in MDA-MB-231 breast carcinoma cells when more than 5% of therapeutic cells were used. Most efficient inhibition of 91% was seen in A375 + CDy-AT-MSC/5FC 5-day-cocultures and was achieved with only 2% of therapeutic cells (Figure 4B). To confirm the bystander killing effect, fluorescently labelled tumour cells were mixed with CDy-AT-MSC in the presence of prodrug 5FC. Killing was statistically significant and dose-dependent with 5FC concentrations higher than 10 µg/ml (Figure 4C). Cocultures without 5FC or cocultures of untransduced AT-MSC with tumour cells exhibited no change in relative fluorescence (data not shown).

Figure 4.

CDy-AT-MSC/5FC mediated cytotoxicity towards human tumour cells. Therapeutic efficacy of expanded CDy-AT-MSC in combination with 5FC was analysed by coculture experiments. Direct coculture experiments and the bystander effect were evaluated quantitatively by proliferation assay or fluorescence intensity. Indirect coculture was evaluated by flow cytometry. Data are expressed as means ± SE. (A) Tumour cells were plated with increasing amounts of CDy-AT-MSC and incubated with 150 µg/ml of 5FC for 5 days. Cell viability was measured by MTS assay. MDA-MB-361 cells cultured in the presence of CDy-AT-MSC together with 5FC show 64.6% proliferation inhibition, even when the ratio of CDy-AT-MSC : MDA-MB-361 tumour cells was as low as 1 : 50. Significant 26.9% inhibition of MDA-MB-231 proliferation by CDy-AT-MSC/5FC could be achieved at the ratio 1 : 50. In the absence of 5FC, there was no effect on cell proliferation because it is a nontoxic prodrug. (B) Time course of the proliferation inhibition in direct cocultures of CDy-AT-MSC and human melanoma A375 (mixed in a ratio 1 : 50, 1 : 10 and 1 : 5) in the presence of 5FC. Inhibition of 95.0% could be achieved in a mixture containing 17% CDy-AT-MSC after 5 days. (C) Fluorescently (CFDA-SE) labelled tumour cells were mixed with same amount of CDy-AT-MSC and incubated in increasing concentrations of 5FC. Three days later, fluorescence intensity corresponding to the number of viable cells was quantified. A significant decrease in relative fluorescence was observed at concentrations higher than 10 µg/ml of 5FC. (D) Indirect cocultures of CDy-AT-MSC or AT-MSC with HT-29 were evaluated by flow cytometry. HT-29 cells were seeded in six-well plates and same number of CDy-AT-MSC or AT-MSC in inserts. Cells were cocultured in media with or without 5FC for indicated time. Proportion of apoptotic cells was calculated from number of Annexin-V positive cells. The apoptotic population peaked at the 48-h time point. A significant increase in the proportion of apoptotic and secondary apoptotic (annexin V and 7-AAD positive cells) was observed exclusively in HT-29 cultures indirectly cocultured with CDy-AT-MSC/5FC

To mimic the situation in vivo, where the therapeutic cells might not have reached and/or been evenly distributed within the different parts of the tumour mass, we performed indirect coculture experiment to asses the cytotoxic effect of CDy-AT-MSC/5FC in the absence of cell–cell contacts. As soon as 48 h post-coculture start, 35% of target HT-29 cells were undergoing apoptosis, as indicated by the annexin V positive and 7-AAD negative staining in flow cytometric analysis (Figure 4D). One day later, there was a significant shift towards annexin V and 7-AAD double positive cells exhibiting secondary necrosis. Even in the absence of cell–cell contact, CDy-AT-MSC/5FC combination could induce apoptosis in target cancer cells. Taken together, these data validate the capability of transduced CDy-AT-MSC to exert a potent cytotoxic bystander effect on the tumour cells without the need for the transgene expression in tumour cells directly.

To compare the extent of proliferation inhibition by CDy-AT-MSC/5FC combination with other cancer cell types, IC50 values were calculated as a percentage of CDy-AT-MSC/5FC in direct coculture to achieve 50% proliferation inhibition in comparison to 5FC untreated controls in a standard coculture assay as described above. Inhibition of proliferation values varied from 0.01% to 15% and reflect the tumour cell sensitivity to CDy-AT-MSC/5FU based chemotherapy in vitro (Figure 5A). Furthermore, the proliferation inhibition value was compared for each cell line tested, if 5% CDy-AT-MSC/5FC were present (Figure 5B). This was the estimated value of achievable CDy-AT-MSC proportion in HT-29 induced tumours upon systemic administration from our in vivo data 14. It is obvious, from the data comparison, that even tumour cells originating from same tissue exhibit major differences in sensitivity to CDy-AT-MSC/5FC mediated killing. Taken together, these data highlight the capability of expanded CDy-AT-MSC to exert a potent bystander effect, even though they were sensitive to the CDy/5FC suicide effect, which may serve as intrinsic self-inactivation of the cellular delivery vehicle in vivo. These experiments validate the rationale for using adipose-tissue-derived mesenchymal stromal cells for cell-directed enzyme prodrug therapy of several tumour types found to be responsive to CDy-AT-MSC/5FC mediated chemotherapy in vitro.

Figure 5.

Inhibitory efficiency of CDy-AT-MSC/5FC treatment in human tumour cell lines in vitro. Therapeutic efficacy of expanded CDy-AT-MSC in combination with 5FC was analysed by direct coculture experiments evaluated by MTS proliferation assay. (A) Comparison of IC50 values for different human tumour cell lines. IC50 value was expressed as a percentage of CDy-AT-MSC/5FC in direct coculture to achieve 50% proliferation inhibition in comparison to 5FC untreated control. (B) Proliferation inhibition achieved by 5% CDy-AT-MSC/5FC in direct cocultures with tumour cells. Tumour cells were plated with 5% of CDy-AT-MSC and supplemented with 5FC. Cocultures were evaluated after 5 days. Cell viability was expressed as the mean ± SE percentage of 5FC untreated control (set to 100% by default)

A375 melanoma regression upon CDy-AT-MSC/5FC mediated bystander killing in vivo

Next we evaluated the tumour growth inhibition in vivo. Previous results suggested human melanoma A375 to be the candidate target tumour type to be tested in vivo. Therefore, mixtures of A375 + 10% CDy-AT-MSC and A375 + 10% AT-MSC were injected subcutaneously in nude mice. Tumour growth was significantly inhibited in A375 + 10% CDy-AT-MSC/5FC group of animals up to day 10 (Figure 6A). The time to the 100% tumour onset was two-fold higher (7 versus 14 days) in the therapeutic versus control group (Figure 6B). However, by the experimental endpoint on day 24, both tumour volume and tumour weight were smaller in therapeutic group, although this was not statistically significant and all animals developed tumours in both groups.

Figure 6.

Tumour growth was effectively inhibited by expanded CDy-AT-MSC/5FC in vivo. Mixtures of A375 tumour cells with 10% CDy-AT-MSC or AT-MSC were injected s.c. into flank of each nude mouse (n = 5–6 in each treatment group) to induce tumours. A375 alone induced tumours were used as control. All animals were treated with daily dose of 500 mg/kg of 5FC i.p. for 5 consecutive days starting day 3 post-implantation. A second round of 5FC treatment was started at day 10 post-implantation. (A) Co-injection of 1 : 10 AT-MSC + A375 resulted in the growth of tumours in all experimental animals showing some tendency to support growth of tumours in comparison to induction with A375 cell alone. The group coinjected with A375 + CDy-AT-MSC/5FC exhibited significant inhibition of tumour growth up to day 10. (Kruskal–Wallis test for multiple comparisons, P < 0.05) (B) All animals developed tumours, although the time to the 100% tumour onset was two-fold longer in the therapeutic compared to the control group

To test whether the higher proportion of CDy-AT-MSC cells coinjected would increase the therapeutic effect, we injected animals with A375 + 20% CDy-AT-MSC in the therapeutic group and A375 + 20% AT-MSC in the control group. Even though there were six out of nine animals having palpable tumours (less than 25 mm3) by day 3 of the experiment prior to 5FC treatment start, five of them regressed and resulted in very low tumour burden at the experimental endpoint (Figure 7). Moreover, 89% of the animals (eight out of nine) within the experimental group were tumour-free in contrast to control group, where all of the animals developed tumours.

Figure 7.

Local 5FC conversion in tumours leads to tumour regression in vivo. Mixtures of A375 tumour cells with 20% CDy-AT-MSC or AT-MSC were injected s.c. into flank of each nude mouse (n = 9 in each treatment group). Animals were 5FC treated as above. At day 24, animals were sacrificed, tumours were excised and weighted. (A) Tumours started to significantly regress after the start of 5FC therapy in the experimental group treated with 20% CDy-AT-MSC. Standard deviations are omitted for clarity. (B) Tumour burden in A375 + 20% CDy-AT-MSC/5FC treated animals was significantly lower compared to other animal groups. Tumour weight was taken as a measure of tumour burden at the experimental endpoint (day 24) and expressed as mean ± SE weight. (Kruskal–Wallis test for multiple comparisons, P < 0.01) (C) Group coinjected with A375 + 20% CDy-AT-MSC/5FC exhibited strong inhibition of tumour growth beyond day 7. (Data compared by the Mann–Whitney U-test for single comparisons, *P < 0.05, **P < 0.01) (D) Eight out of nine animals in the therapeutic group were tumour-free at the experimental endpoint. Within the group, three animals did not develop a tumour at all, and six of them had palpable tumours at day 4. Five out of these six tumours responded to therapy by complete regression. All animals developed tumours in A375 + 20% AT-MSC group

To examine tumour homing capabilities of CDy-AT-MSC to pre-established A375 tumour xenografts, three delivery routes were tested (Figure 8A). CDy transgene as a marker of CDy-AT-MSC homing could be detected in tumours after intravenous administration only, and it was absent when therapeutic cells were injected i.p or s.c. Moreover, when the influence of this therapeutic paradigm on the tumour growth was examined, three doses of CDy-AT-MSC injected intravenously combined with 5FC prodrug application inhibited tumour growth by 90.6% in comparison to control AT-MSC/5FC group by day 23 (Figure 8B). A different delivery route (i.p.), other cell type (fibroblasts used as cellular vehicle) or treatment with cytotoxic drug 5FU directly have proven inefficient in tumour growth inhibition under these conditions.

Figure 8.

Systemically administered CDy-AT-MSC exert antitumour effect in the presence of 5FC in vivo. (A) Delivery of CDy transgene into subcutaneous tumour xenografts upon CDy-AT-MSC administration in vivo was evaluated by PCR to detect CDy specific product in isolated tumour DNA. Tumours were induced with 1.5 × 106 A375 cells s.c. and CDy-AT-MSC (106 cells) were administered either i.v., i.p. or s.c. into the other flank 3 days later. DNA was isolated from excised tumours on day 10 (7 days post-injection). The presence of CDy transgene within the tumour DNA confirms the capability of CDy-AT-MSC to deliver the transgene into subcutaneously induced tumours after intravenous administration. DNA isolated from cultured CDy-AT-MSC served as a control. DNA integrity was confirmed by detection of human Alu sequences. (B) Tumours were induced with A375 cells as above. Animals (n = 6 per group) were divided into five groups according to treatment: CDy-AT-MSC (106 cells) i.v. or i.p., CDy-Fib (106 cells) i.v., AT-MSC (106 cells) i.v. and the group that received 5FU only. Three cell doses were injected either into the lateral tail vein or intraperitoneally on days 3, 10 and 17 (#). Animals were treated with daily dose of 500 mg/kg of 5FC i.p. for 4 consecutive days starting on days 5, 12 and 19 in three rounds in the indicated groups (↑). The control group received 20 mg/kg/day 5FU i.p. in three consecutive 5-day rounds with a 2-day break starting on day 5. Therapeutic regimen CDy-AT-MSC i.v./5 FC showed an inhibitory effect on tumour growth in all experimental animals. None of the animals exhibited complete tumour regression. The results are expressed as mean tumour volume (SE omitted for clarity)

From these data, we conclude that expanded CDy-AT-MSC were able to home into melanoma A375 subcutaneous xenografts upon systemic administration and were efficient in combination with prodrug 5FC in tumour growth inhibition, although treatment did not result in complete tumour regression as observed in coinjection studies.


The results obtained in the present study further contribute to potential clinical utility of human adipose-tissue derived mesenchymal stromal cells in a modified cell-directed approach to enzyme mediated prodrug conversion for molecular chemotherapy of cancer. The safety of the CD combination with prodrug 5FC been shown in many clinical settings, although with the limited therapeutic efficiency. It was shown that yeast CD produced a 15-fold higher amount of 5FU compared to bacterial enzyme 33. Moreover, fusion of this yeast CD to phosphoribosyltransferase further significantly increased the bystander effect in vitro34. These enzyme modifications lead to optimized prodrug conversion to achieve maximum level of cytotoxic compound in vivo. Indeed, the cell survival in a mixture containing only 6.25% CDy-expressing embryonic endothelial progenitor cells was only 60% (instead of the expected 93.75%), demonstrating a strong bystander effect 2. In the present study, a mixture containing 2% CDy-AT-MSC exhibited A375 cell survival of 9%. On the other hand, similar experiments with bacterial CD expressing human neural stem cells have shown approximately 80% of growth inhibition in a mixture containing 33% of transduced cells 4. However, although these data cannot be directly compared, they reflect the enzyme optimization for maximum production of the cytotoxic agent. The outcome clearly depends on the target cell sensitivity, the intrinsic properties of the therapeutic cells and the interplay between these two in vitro.

By contrast to these studies employing embryonic or fetal cells, we decided to test adult human mesenchymal stromal cells as a cellular vehicle for transgene expression. These cells represent an easily obtainable abundant source of adult multipotent cells for autologous use. Transgene delivery into AT-MSC could be achieved by oncoretroviral and lentiviral vector, although with rather low efficiency (2.7% versus 13.1%, respectively). An efficiency higher than 90% was achieved by employing adenoviral or lentiviral vector with cytomegalovirus promoter in AT-MSC cells 22. Adenoviral transgene delivery for short-term transgene expression in bone marrow-derived MSC has also been shown to be feasible with high efficiency 8, 23. Transduction efficiency in the range 28–46% was reported for AT-MSC when modified MoMuLV retrovirus was used 35. We have employed the ping-pong strategy to increase viral titres and succeeded in highly efficient AT-MSC transduction by amphotropic retrovirus. Up to this point, there were no reports from other laboratories (including our) of any negative effect of transgene expression in MSC on the proliferation potential, morphology, differentiation capabilities or transformation properties 14, 22, 23, 35.

A combination of fusion yeast CD and adult adipose-tissue derived mesenchymal cells as a cellular vehicle resulted in pronounced inhibition of cell proliferation. Prolonged propagation of AT-MSC in vitro did not impair strong tumour cell targeted cytotoxicity, although there was a negative impact on AT-MSC chemoresistance. As reported previously, mesenchymal stromal cell cultures represent a heterogeneous cell population 36, 37. Several studies have shown that prolonged culture expansion in vitro was accompanied by phenotypic changes, signs of ageing and senescence 38–40. We hypothesize that the loss of chemoresistance in culture propagated AT-MSC could be attributed to the gradual decrease of the cell proportion with properties of uncommitted progenitors. This cell subpopulation described to be resistant to 5FU contains quiescent, uncommitted and undifferentiated mesenchymal cells 41. It is possible that in vitro culture conditions favour the expansion of less chemoresistant proliferative MSC. We consider that this might be beneficial in therapeutic approaches, where it is desirable for therapeutic cells to exert their function and to be inactivated by the self-suicide mechanism thereafter.

Our data suggest a possible mechanism of initial low passage AT-MSC chemoresistance by the expression of DPD that contributed to 5FU catabolism 28. TP expression in AT-MSC in combination with DPD has been ascribed to 5FU resistance 29. TP is identical to PD-ECGF and exerts angiogenic activity 42. It has been shown that the expression of TP in cancer cells has a very different outcome to that in stromal cells of the tumour. The expression of TP in tumour stromal cells was a favourable prognostic marker in colorectal carcinoma. Studies reported enhanced anticancer effect of 5FU-derivative chemotherapy in patients with stromal TP expression 43. These data are further supported a report from Kato et al.44 showing that the overexpression of TP (PD-ECGF) could potentiate the bystander cytotoxic effect of 5FU prodrugs even in the absence of cell–cell contacts. The expression of TP by CDy-AT-MSC in situ might have enhanced the anticancer effect of 5FU-based chemotherapy, although the biological implications of stromal TP remain to be clarified.

Coinjection studies in vivo have confirmed the efficiency of CDy-AT-MSC/5FC therapy in a human melanoma A375 xenograft. To achieve tumour regression, a relatively high dose of 20% therapeutic cells had to be used. However, this extent of tumour-free survival could not be achieved in our previous study with the different tumour type (colon carcinoma HT-29 xenograft), even at 50% of coinjected therapeutic cells 14. The survival of CDy-AT-MSC/5FC can be estimated from in vitro studies up to several days. It correlates to the time-frame of tumour growth inhibition in the group coinjected with 10% therapeutic cells, where tumours started to grow significantly beyond day 10. By this time, the majority of the therapeutic cells might have been eradicated by the self-suicide effect. Nevertheless, initially coinjected 20% CDy-AT-MSC were sufficient to mediate tumour regression. We tend to speculate that the extensive local production of cytotoxic drug and/or concomitant disturbance of the tumour milieu due to the extensive cell death in situ resulted in substantial damage to the tumour initiating cells within the tumour mass. The amount of the systemically administered therapeutic cells to home within the tumour was estimated to less then 10% of the tumour mass 8, 9. Therefore it is rational to use multiple repeated injections of the therapeutic cells to achieve tumour regression in systemic treatment. Indeed, a single intravenous CDy-AT-MSC dose only marginally inhibited growth of subcutaneous melanoma (data not shown). Therefore, we used repeated intravenous injections of therapeutic cells to achieve inhibition of tumour growth. As it is obvious from the data shown in Figure 8, the other delivery route (i.e. intraperitoneal) was not suitable for CDy-AT-MSC tumour targeted homing accompanied by an absence of therapeutic effect.

Taken together, the results of the present study provide data for further exploitation of mesenchymal stromal cells as delivery vehicles for targeted cancer chemotherapy. AT-MSC biological properties in combination with suitable transgene/prodrug may exert a very efficient tumour cell cytotoxic effect that could be exploited for treatment of tumours in vivo. These cells possess many of the attributes necessary for the optimal cellular vehicle that may prove beneficial in clinical studies.

Supporting Information

Supporting information may be found in the online version of this article.


We thank M. Dubrovcakova, V. Frivalska for technical assistance; K. Hlubinova for help with the preparation of virus producing cells; L. Hunakova and J. Bodo for help with flow cytometry analysis; and J. Fabry MD, Posonium Clinic, Bratislava, and D. Guba MD, Institute of Medical Cosmetics, Bratislava, for providing us with material for AT-MSC isolation. We acknowledge the help of Associate Professor V. Witkovsky with statistical evaluation. The study was supported by VEGA grant nos. 2/6061/26 and 2/7060/27; financial support from FIDURA Capital Consult GmbH, Munich, SPP Foundation; and grant support awarded by the League against Cancer (V.A.). The authors indicate no potential conflict of interest.