TRAIL-transduced multipotent mesenchymal stromal cells ... - Nature

Cancer Gene Therapy (2011) 18, 229–239 r

2011 Nature America, Inc. All rights reserved 0929-1903/11

www.nature.com/cgt

ORIGINAL ARTICLE

TRAIL-transduced multipotent mesenchymal stromal cells (TRAIL-MSC) overcome TRAIL resistance in selected CRC cell lines in vitro and in vivo LP Mueller1, J Luetzkendorf1, M Widder, K Nerger, H Caysa and T Mueller Department of Internal Medicine IV, Martin-Luther-University Halle-Wittenberg, Halle (Saale), Germany Tumor-integrating multipotent mesenchymal stromal cells (MSC) expressing transgenes with anti-tumor activity may serve as vehicles for tumor therapy. Tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) represents such a factor; however, TRAIL-resistant tumor cells exist. Based on our previous work, here we investigated whether MSC with lentiviral TRAIL expression (TRAIL-MSC) inhibit the growth of TRAIL-resistant colorectal carcinoma (CRC) cells. Our data show that TRAIL-MSC induce apoptosis in selected TRAIL-resistant CRC cell lines and effectively inhibit the growth of TRAIL-resistant HCT8 cells. This sensitization to TRAIL-induced apoptosis required the presence of MSC-expressed TRAIL. However, for the first time we show that selected CRC cells are resistant to TRAIL-MSC. In the cell line HT29, this resistance could be overcome by concomitant subapoptotic genotoxic damage in vitro. However, such sensitization was not achieved in vivo as treatment of mixed HT29/TRAILMSC xenografts with 5-FU rather resulted in enhanced growth. Taken together, our data prove that TRAIL-MSC overcome TRAIL resistance in selected CRC cells through direct intercellular interaction and may, therefore, represent a clinical tool to overcome TRAIL resistance. However, such potential clinical use requires further preclinical studies as our data also prove that TRAIL-MSCresistant CRC cells exist. Our data add to the notion that TRAIL resistance of CRC cells is conferred by different mechanisms. Cancer Gene Therapy (2011) 18, 229–239; doi:10.1038/cgt.2010.68; published online 29 October 2010 Keywords: MSC; TRAIL; resistance; colorectal carcinoma

Introduction

Investigations in various animal models show that multipotent mesenchymal stromal cells (MSC) integrate in malignant tissue and may, therefore, serve as vehicles for a specific tumor therapy.1 A possible option for this approach comprises the ectopic expression of a factor with anti-tumor activity in MSC and the subsequent local or systemic application of these transduced MSC.2 Tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) represents such a factor as it induces apoptosis preferentially in tumor cells through binding of transmembraneous death receptors (DR) DR4 and DR5.3 Signaling is modulated by differential expression of DR and decoy receptors DcR1 and DcR2 as well as by differential intracellular signaling.4 Recombinant TRAIL as well as agonistic anti-DR antibodies are currently studied in clinical trials. TRAIL could, therefore, provide

Correspondence: Dr LP Mueller, Universitaetsklinik und Poliklinik fuer Innere Medizin IV, Martin-Luther-Universitaet HalleWittenberg, Ernst-Grube-Strasse 40, Halle D-06120, Germany. E-mail: [email protected] 1 These two authors contributed equally to this work. Received 21 March 2010; revised 14 June 2010; accepted 16 August 2010; published online 29 October 2010

a suitable transgene for MSC-based therapy of solid tumors. Advanced colorectal carcinoma (CRC) represents a solid tumor for which new treatment modalities are needed as it comprises a leading cause of tumor-related mortality in the western hemisphere.5 Depending on the cell type, TRAIL induces apoptosis independently of p53.4 Therefore, TRAIL promises to be also effective in CRC despite the high rate of mutated p53 in CRC cells.6 In our previous study, we had shown that MSC with lentiviral TRAIL expression can effectively inhibit the growth of CRC cells in vitro and in vivo.7 However, tumor cells show a differential sensitivity to TRAIL.4 For CRC cell lines like HCT8 and HT29, a relative resistance to soluble TRAIL (sTRAIL)8 as well as to agonistic anti-DR antibodies9 has been described. Various mechanisms have been suggested for the TRAIL resistance of tumor cells.4 In selected CRC cells, the differential sensitivity to TRAIL was associated with a differential expression of DR4 and DR5.10 In vivo, resistance to TRAIL appears to be transmitted by a complex interaction involving inflammatory signaling.11 Sensitization to TRAIL-induced apoptosis has been shown for various substances and appears mainly achieved by upregulation of DR4 and DR5 or involvement of the mitochondrial proapoptotic pathway.4 In TRAIL-resistant CRC cells, a sensitization to TRAIL

Reversal of TRAIL resistance by TRAIL-MSC LP Mueller et al

230

has been reported for oxaliplatin,12 for agents uncoupling mitochondrial functions8 as well as for histone deacetylase inhibitiors.9 In our previous work, we had observed that MSC with lentiviral TRAIL expression (TRAIL-MSC) inhibit the growth of cells of the CRC cell line HCT8 through apoptosis induction.7 As HCT8 cells are resistant to apoptosis induction by sTRAIL, we hypothesized that TRAIL-MSC can induce apoptosis and inhibit xenograft growth of TRAIL-resistant CRC cells in general. The data presented here prove that TRAIL-MSC can induce apoptosis in TRAIL-resistant CRC cells as well as effectively inhibit xenograft growth of TRAIL-resistant CRC cells. For the first time, we show that selected TRAIL-resistant CRC cells are also resistant to TRAILMSC. While concomitant subapoptotic genotoxic damage sensitized TRAIL-MSC-resistant CRC cells to TRAIL-MSC in vitro, no such sensitization could be achieved in vivo. In contrast, TRAIL-MSC seemed to enhance in vivo growth of TRAIL-MSC-resistant CRC cells despite concomitant genotoxic damage. These data caution for the clinical use of TRAIL-MSC.

Materials and methods

MSC, cell lines and cell culture Cultivation of MSC isolated from human bone marrow was performed as described according to a protocol approved by the institutional Ethics Board.7 MSC growth medium was composed of low-glucose Dulbecco’s modified Eagle’s medium (PAA, Pasching, Austria) with 15% fetal calf serum (Invitrogen, Grand Island, NY) and 1% penicillin/streptomycin (PAA). The human CRC cell lines Colo205, Colo320DM, DLD1, HCT8, HCT15, HCT116, HT29, SW48 and SW480 as well as the derived respective clones with lentiviral DsRed expression were cultivated in RPMI-1640 (PAA) with 10% fetal calf serum and 1% penicillin/streptomycin. Human embryonic kidney 293T cells were cultivated in Dulbecco’s modified Eagle’s medium with 15% fetal calf serum and 1% penicillin/streptomycin. Screening for TRAIL sensitivity of CRC cell lines and MSC Sensitivity of CRC cell lines to sTRAIL was estimated by a sulforhodamine-B assay as described previously.7 Briefly, CRC cell lines were seeded at 5000 cells per well and MSC were seeded at 2000 cells per well in three 96-well plates with 8 wells per cell line on each plate, respectively. After 48 h, one plate was fixed with 10% trichloroacetic acid (control-0 h), one plate treated for 24 h with fresh growth medium (control-24 h) and one plate treated with 100 ng ml–1 soluble KillerTRAIL (Axxora, San Diego, CA) for 24 h (sTRAIL-24 h). After 24 h, the two treated plates were fixed with 10% trichloroacetic acid. Fixed plates were stained with 0.4% sulforhodamine-B (Sigma-Aldrich, St Louis, MO) as described.13 Optical density (570 nm) was determined in a mircoplate reader and the mean of the 8 wells for

Cancer Gene Therapy

each cell line at the respective condition was calculated. Values from control-0 h were set 100% and compared with control-24 h and with sTRAIL-24 h. To determine a dose-dependent response to sTRAIL, CRC cell lines were seeded at 5000 cells per well into 96-well plates, one plate per cell line. After 48 h of plating, media was removed and cells were treated for 24 h with sTRAIL at various concentrations (0.3–3000 ng ml–1) with one concentration per column of 8 wells. One column without added sTRAIL served as an untreated control. After 24 h, plates were fixed with 10% trichloroacetic acid and stained with 0.4% sulforhodamine-B. The mean value of the optical density of each column, that is of each sTRAIL dosage, was calculated and set in relation to mean values of the respective untreated controls, which were set 100%. The concentration that inhibited cell growth by 50% (inhibitory concentration (IC) 50) was determined for each treatment schedule from semilogarithmic dose–response plots.

Vector production and transduction of cells For lentiviral vector production, the packaging plasmids pMDLg/pRRE and pRSV-Rev14 and the envelope plasmid pVSVG were used. The transfer vector plasmids pFUGW (encoding GFP), pFUDW (encoding DsRed) and pFUTW (encoding human full-length TRAIL) were described previously.7 The plasmids pMDLg/pRRE, pRSV-Rev, pVSVG and pFUGW were kindly provided by Professor Dr T Braun, Max-Planck-Institute for Heart and Lung Research, Bad Nauheim, Germany. Vector particles were produced by transient transfection of human embryonic kidney 293T cells with 10 mg transfer vector plasmid and 5 mg of pMDLg/pRRE, pRSV-Rev and pVSVG, respectively, by calcium phosphate DNA precipitation as described previously.7 For transduction, MSC or CRC cells were grown to 50% confluence and fed with fresh growth medium containing 8 mg ml–1 polybrene (Sigma-Aldrich). Transduction of MSC was performed at passage 1 or 2. Titers used for transduction were about 5 105 viral particles per ml. Medium was replaced after 24 h and the transgene expression was analyzed after additional 24 h. Untransduced wild-type MSC (WT-MSC) or GFP-transduced MSC (GFP-MSC) served as controls. Coculture experiments For screening of TRAIL-MSC-mediated apoptosis induction in CRC cells, TRAIL-MSC (10 000 cells cm–2, respectively) were mixed and plated in 12-well plates with cells of one of the following CRC cell lines at the respective concentration: DLD1 (50 000 cells cm–2), Colo205, HCT8 and HT29 (each 60 000 cell cm–2), Colo320DM, HCT15, HCT116, SW48 and SW480 (each 100 000 cells cm–2). After 24 h cocultivation, lysates of whole coculture (that is all adherent and detached cells) were prepared and used for western blot analysis. Cocultures of CRC cells with WT-MSC served as controls. A TRAIL-neutralizing antibody (mouse anti-TRAIL clone 2E5; Axxora) was used at a concentration of 10 mg ml–1 to evaluate the specificity of TRAIL activity.


To evaluate an effect of WT-MSC on the sTRAIL sensitivity of HCT8 cells, WT-MSC (10 000 cells cm–2) were mixed with HCT8 cells (60 000 cells cm–2) and plated in three 96-well plates. Treatment with sTRAIL and analysis were performed in a sulforhodamine-B assay as described above. Unmixed cultures of HCT8 and MSC served as the respective negative controls. A coculture of TRAIL-MSC (10 000 cells cm–2) with HCT8 cells (60 000 cells cm–2) served as a positive control. After 24 h, cells were harvested by trypsinization, total cell number was counted and compared in relation to starting cell numbers. In order to analyze apoptosis induction in TRAILMSC-resistant CRC cells by concomitant exposure to TRAIL-MSC and subapoptotic genotoxic damage in vitro, TRAIL-MSC or WT-MSC (10 000 cells cm–2, respectively) were mixed with HT29 cells (60 000 cells cm–2) and plated in 5 wells of a 12-well plate. While one well was left untreated, 5-FU was added to two wells and oxaliplatin added to the remaining two wells at a final concentration of 3 mM, respectively. A medium change with repeated addition of 3 mM 5-FU or oxaliplatin to the treated samples was performed after 24 and 48 h. After 72 h, lysates of whole coculture (that is all adherent and detached cells) from the first treated well as well as lysates of detached cells from the second treated well were prepared and used for western blot analysis. To investigate the effect of TRAIL-MSC on the growth of HT29 cells, TRAIL-MSC or WT-MSC (10 000 cells cm–2, respectively) were mixed with DsRed-HT29 cells (60 000 cells cm–2) and plated in 6 wells each of a 12-well plate. A total of 3 mM 5-FU was added to three wells of the WT-MSC/DsRed-HT29-cocultures and TRAILMSC/DsRed-HT29-cocultures, respectively. After 24 h, detached cells were harvested and the cell number was determined. Adherent cells from one well were harvested, counted and the percentage of DsRed-positive cells was analyzed by flow cytometry. Fresh growth medium was added to the remaining wells including a repeated addition of 3 mM 5-FU to the treated samples. This procedure was repeated with the remaining wells 48 and 72 h after initial plating. Flow cytometric analyses were performed on an FACSCalibur using CellQuest software (all BD Biosciences, Franklin Lakes, NJ).

Flow cytometry for TRAIL-receptor expression The following antibodies were used: mouse antihuman TRAIL-R1 (DR4; clone HS101), mouse antihuman TRAIL-R2 (DR5; clone HS201), mouse anti-human TRAIL-R3 (DcR1; clone HS301), mouse anti-human TRAIL-R4 (DcR2; clone HS402; all Axxora), goat antimouse Ig FITC and mouse y1 FITC (clone X40; both BD Biosciences). Cells were detached, washed and incubated with primary antibody at 10 mg ml–1 for 30 min at 41C. After washing with saline, cells were incubated with secondary antibody or isotype control for 15 min at 41C. Analysis was performed on an FACSCalibur using CellQuest software. Geometric means of fluorescence intensities were determined and antibody-specific values were set in relation to values of the respective isotype controls, which were set 1.

Western blotting Western blot analyses were performed as described previously15 using the following primary antibodies: anti-TRAIL (0.5 mg ml1, mouse, clone HS501, Axxora), anti-caspase-3 (0.5 mg ml–1, mouse, MBL, Woburn, MA), anti-caspase-8 (0.5 mg ml–1, mouse, Invitrogen), antiPARP (1 mg ml–1, mouse, BD Biosciences) and anti-actin (0.025 mg ml–1, goat, Santa Cruz Biotechnology, Santa Cruz, CA). Immunocomplexes were visualized by enhanced chemiluminescence (Amersham Pharmacia Biotech, Little Chalfont, England) using horseradish peroxidase-conjugated anti-mouse IgG and anti-goat IgG (each 0.1 mg ml–1, Santa Cruz Biotechnology) and Roti-Lumin (Carl Roth, Karlsruhe, Germany). Animal studies Six- to eight-week-old athymic nude-fox n1 nu/nu mice (Harlan Winkelmann, Borchen, Germany) were used for the in vivo experiments according to institutional guidelines under approved protocols. Subcutaneous mixed xenografts were generated as follows: 3 106 DsRed-HT29 cells or DsRed-HCT8 cells were mixed with 7.5 105 TRAIL-MSC or control MSC (WT- or GFP-MSC), respectively. Unmixed DsRed-HT29 cells or DsRed-HCT8 cells served as controls. Cell suspensions were administered into the right (TRAIL-MSC) and the left (control MSC or unmixed CRC cells) flank of one animal on day 0. Animals with mixed HT29 xenografts received 5-FU intraperitoneally at doses of 30 mg kg–1 body weight or a respective volume of 0.9% NaCl on days 0–4. In a separate experiment, animals with mixed HT29 xenografts received intraperitoneally 5-FU or 0.9% NaCl at the same dose but on days 7–9 and 14–16. Grayscale images from both body sides of each animal were acquired over 25 days for determining tumor volume by measuring tumor dimensions. On day 25, animals were killed and dissected tumors were weighed. Every experiment was carried out at least in triplicates. None of the animals had to be killed prematurely because of tumor burden or impaired vital parameters. Image acquisition and determination of tumor volume A 2.2 CRi Maestro in vivo fluorescence-imaging system (CRi, Woburn) was used to acquire grayscale images. Measuring of tumor dimension was performed with GNU Image Manipulation Program 2.6.8 software (http://www.gimp.org/) using the measure tool. Tumor volume was calculated using the following formula: volume (pixel3) ¼ length (pixel)*(wide (pixel))2*pi/6. Tumor volume was converted in cm3 using the calibration factor of 0.1234837 mm per pixel, given for the CRi Maestro in vivo fluorescence-imaging system. Statistical analysis Statistical analysis was performed using SPSS 16.0 software (SPSS, Chicago, IL). Depending on the equality of variances as assessed by Levene testing, results from the t-test were used in case of equal variances and results from an updated t-test were used in case of non-equal variances. In any case, a P-value o0.05 was considered significant. Cancer Gene Therapy

231


232

Results

CRC cell lines show differential sensitivity for sTRAIL without general correlation to TRAIL-receptor expression Based on our hypothesis that TRAIL-MSC can induce apoptosis in CRC cells, which are resistant to apoptosis induction by sTRAIL, we first analyzed the sensitivity to sTRAIL in nine CRC cell lines. Cells were treated with sTRAIL for 24 h at a standard dose and growth was assessed by determining cell numbers at the start and at the end of treatment in comparison with untreated controls. Upon treatment, a reduction of the cell number below the starting values was observed in the cell lines Colo205, SW48, HCT116, HCT15 and DLD1. Thus, these cell lines were determined to be TRAIL sensitive. In contrast, for the cell lines SW480, HCT8, HT29 and Colo320DM, an increased cell number was observed at the end of treatment in comparison with starting values. These cell lines were, therefore, determined to be TRAIL resistant. In comparison with untreated controls, sTRAIL treatment resulted only in a slight reduction of proliferation in these TRAIL-resistant CRC cell lines (Figure 1a). In order to proof this concept of TRAIL resistance, the response of selected CRC cell lines to increasing doses of sTRAIL was tested by a sulforhodamine-B assay. The data showed a significant difference in the IC50 values of the sensitive cell lines DLD1 (45±7 ng ml–1) and HCT116 (40±11 ng ml–1) in comparison with each of the resistant

cell lines HCT8 (687±320 ng ml–1) and HT29 (490±215 ng ml–1), respectively (P ¼ 0.025 for HCT116 vs HCT8; P ¼ 0.022 for HCT116 vs HT29; P ¼ 0.026 for DLD1 vs HCT8; P ¼ 0.023 for DLD1 vs HT29). No differences for the IC50 values were observed between the sensitive cell lines HCT116 and DLD1 as well as between the resistant cell lines HCT8 and HT29. In resistant cell lines, no IC90 values were reached and MSC did not show any response to sTRAIL (Figure 1b). These data proof that the used dosage of 100 ng ml–1 sTRAIL is appropriate to discriminate sensitive and resistant CRC cell lines. Specifically, increased dosage up to 3 mg ml–1 did not abrogate this distinction, but only led to a similar increase of growth inhibition in all CRC cell lines (Figure 1b). Moreover, these data support the notion that MSC are completely resistant to sTRAIL qualifying them as ideal vehicles for ectopic TRAIL expression. As effects of sTRAIL are dependent on DR signaling and may be modified by differential expression of decoy receptors, we analyzed the expression of the respective receptors in the selected CRC cell lines. While expression of DR4 and DR5 was highest in TRAIL-sensitive DLD1 and lowest in the TRAIL-resistant HT29, no relevant difference for expression of both receptors was observed between TRAIL-sensitive HCT116 and TRAIL-resistant HCT8 (Figure 1c). Moreover, expression of decoy receptors was similar in all CRC cell lines, but lowest in TRAIL-resistant HT29. Thus, response to sTRAIL did not in general correlate with TRAIL-receptor expression.

Figure 1 CRC cell lines show differential sensitivity for sTRAIL in vitro without general correlation to TRAIL-receptor expression. (a) CRC cell lines were treated for 24 h with growth medium (control-24 h; white bars) or sTRAIL (sTRAIL-24 h; gray bars). Cell numbers were estimated by a sulforhodamine-B assay and compared with untreated controls (control-0 h; black bars). Data are shown as mean±s.d. of two independent experiments. (b) sTRAIL dose–response plots of selected CRC cell lines and MSC. Cells were treated for 24 h with growth medium or with the respective doses of sTRAIL. Cell numbers were estimated by a sulforhodamine-B assay and compared with the respective untreated control, which was set 100%. Data are shown as mean±s.d. of three independent experiments. (c) Expression of death receptors DR4 (TRAIL-R1) and DR5 (TRAIL-R2) as well as decoy receptors DcR1 (TRAIL-R3) and DcR2 (TRAIL-R4) in CRC cell lines and MSC was analyzed by flow cytometry. The geometric means of fluorescence intensities are shown in relation to the respective isotype controls, which were set 1. Depicted are the mean±s.d. of four independent experiments involving MSC preparations from four different donors. (d) CRC cell lines were cultivated in direct coculture with WT-MSC (W) and TRAIL-MSC (T) for 24 h. Whole cell lysates (that is adherent and detached cells) were harvested and analyzed by western blot for cleavage of PARP (ucl—uncleaved; cl—cleaved). Actin served as a loading control. Arrows indicate the respective molecular weight standards in kDa.

Cancer Gene Therapy


In accordance with their absolute resistance, MSC showed the lowest expression of DR4 and DR5. As sTRAIL-induced growth inhibition is partially conferred by apoptosis induction, we concluded that, while retarding proliferation, sTRAIL does not effectively induce apoptosis in TRAIL-resistant CRC cells. However, our data suggest that this differential TRAIL sensitivity of CRC cell lines cannot solely be explained by differential TRAIL-receptor expression. We, therefore, hypothesized that enhanced TRAIL signaling through direct cell contact with TRAIL-expressing cells may result in apoptosis induction upon direct cell contact not only in TRAIL-sensitive but also in TRAIL-resistant CRC cells. Owing to their inherent resistance to sTRAIL, MSC appeared as a suitable vehicle for such an approach.

TRAIL-MSC induce apoptosis in selected TRAIL-resistant CRC cell lines As shown previously, lentiviral expression of TRAIL in MSC yielded TRAIL-transduced MSC (TRAIL-MSC) with stable TRAIL expression as well as unaltered MSC-defining characteristics and did neither result in malignant transformation nor apoptosis induction of the transduced cells.7 To evaluate their respective response to TRAIL-MSC, cells of the nine CRC cell lines were directly cocultured for 24 h with either WT-MSC or TRAIL-MSC, respectively, and cleavage of PARP as a marker of apoptosis induction was analyzed by western blot in whole culture lysates. Relevant cleavage of PARP—characterized by a decrease of the uncleaved protein (116 kDa) and an increase of the cleaved fragment (85 kDa)—was observed in all TRAIL-sensitive CRC cell lines as well as in the TRAIL-resistant CRC cell lines SW480 and HCT8. No relevant cleavage of PARP—that is no increased signal for cleaved PARP as well as no decreased signal for uncleaved PARP—was observed in the TRAIL-resistant cell lines HT29 and Colo320DM (Figure 1d). Therefore, the cell lines HT29 and Colo320DM were determined

TRAIL-MSC-resistant in contrast to the TRAIL-MSCsensitive cell lines SW480 and HCT8. We regard the weak signal of cleaved PARP in all WT-MSC-containing cocultures to be the result of low level apoptosis induction upon in vitro cell culture. Our previous investigations had shown that TRAILMSC can effectively inhibit growth of TRAIL-MSCsensitive CRC cells in vitro.7 In addition, in vivo imaging had also suggested that TRAIL-MSC can inhibit xenograft growth of the TRAIL resistant, TRAIL-MSCsensitive CRC cell line HCT8. Here, we wanted to verify these data by conventional analysis of the effects of TRAIL-MSC on the growth of HCT8 cells in vivo.

TRAIL-MSC reduce xenograft growth of TRAIL-resistant, TRAIL-MSC-sensitive CRC cell line The effect of TRAIL-MSC on the in vivo growth of TRAIL-resistant HCT8 cells was examined in mixed xenografts using tumor weight as well as imaging-derived volume as markers of tumor growth. A significantly reduced tumor volume was seen in xenografts containing TRAIL-MSC compared with pure HCT8 xenografts (P ¼ 0.023) as well as compared with WT-MSC-containing xenografts (P ¼ 0.008). Similarly, tumor weight of xenografts containing TRAIL-MSC was significantly reduced compared with xenografts containing WT-MSC (P ¼ 0.014). The growth of pure HCT8 xenografts was similar as to that of WT-MSC-containing mixed xenografts (Figure 2). We concluded that TRAIL-MSC can inhibit growth of TRAIL-resistant, TRAIL-MSC-sensitive CRC cell lines effectively in vitro and in vivo. This observation pointed to an obvious discrepancy regarding the response to sTRAIL and to TRAIL-MSC in the cell line HCT8. In our previous study, we had showed that the induction of apoptosis by TRAIL-MSC in TRAIL-sensitive CRC cells as well as in TRAIL-resistant CRC cells was transferred by transmembraneous TRAIL in direct cell–cell contact between TRAIL-MSC and CRC cells. We hypothesized

Figure 2 TRAIL-MSC reduce xenograft growth of TRAIL-resistant CRC. DsRed-HCT8 cells were mixed with WT-MSC or TRAIL-MSC at a proportion of 4:1 (that is MSC representing 20% of the total cell number) and injected in nude mice subcutaneously (s.c.) into the left or right flank, respectively. DsRed-HCT8 xenografts without MSC served as controls. On day 25, tumor volumes on both sides of each animal were examined by measuring tumor dimensions in grayscale images, animals were killed, tumors were dissected and weighted. Tumor weight (left diagram) and calculated tumor volume (right diagram) on day 25 are presented as mean±s.d. No MSC n ¼ 5, WT-MSC and TRAIL-MSC each n ¼ 3. n Pp0.014 vs WT-MSC, wP ¼ 0.023 vs No MSC.

Cancer Gene Therapy

233


234

that in HCT8 cells, apoptosis induction by TRAIL-MSC is related to MSC-expressed transmembraneous TRAIL, but may be enhanced by an additional, non-TRAILrelated, MSC-related signal.

Apoptosis induction in HCT8 by cocultured MSC requires ectopic transmembraneous TRAIL expression in MSC without involving a sensitization for sTRAIL by native MSC In order to test this hypothesis, we first wanted to proof our previous data that apoptosis induction by TRAIL-MSC in cocultured HCT8 is indeed exerted by transmembraneous TRAIL ectopically expressed in MSC.

Western blot analysis showed expression of TRAIL in TRAIL-transduced MSC at the typical size of the transmembraneous protein (Figure 3a). When TRAILMSC were cocultured with HCT8, occurrence of detached cells was observed (Figure 3b), which we had previously shown to comprise apoptotic HCT8 cells.7 However, no detached cells were present in cocultures of HCT8 and TRAIL-MSC in the presence of an anti-TRAIL antibody (Figure 3b) as well as when HCT8 cells were incubated with cell-free supernatant of TRAIL-MSC (data not shown). Second, we wanted to test whether TRAIL-resistant HCT8 cells were sensitized to sTRAIL by directly cocultured native MSC. Cells of the TRAIL-resistant

Figure 3 Apoptosis induction in HCT8 by cocultured MSC requires ectopic transmembraneous TRAIL expression in MSC without involving a sensitization for sTRAIL by native MSC. (a) Whole protein lysates from WT-MSC and TRAIL-MSC at passage 2 were analyzed by western blot for expression of the transmembraneous form of the TRAIL protein (34 kDa). Actin (42 kDa) served as control for equal protein loading. Arrows indicate the respective molecular weight standards in kDa. (b) HCT8 cells were cocultivated with WT-MSC and TRAIL-MSC, respectively, for 24 h. For the experiment depicted in the far right panel, 10 mg ml–1 TRAIL-neutralizing antibody was supplemented (light microscopy; bars ¼ 200 mm). (c) HCT8 cells were cocultured with native MSC (HCT8 þ WT-MSC) in a 6:1 proportion. As controls served cultures of pure HCT8 (HCT8) or pure MSC (WT-MSC). Total cell number at the start of treatment was set at 100% (control-0 h). Samples of each culture were either treated with sTRAIL 100 ng ml–1 for 24 h (sTRAIL-24 h) or left untreated (control-24 h). Total cell number at the end of treatment was determined by a sulforhodamine-B assay and is shown in relation to values at the start of treatment. As a comparison, HCT8 cells were cocultured with TRAIL-MSC (HCT8 þ TRAIL-MSC) in a proportion of 6:1. Again, total cell numbers at the start of coculture (0 h) were set 100%. After 24 h, total cell number was determined in relation to starting values. Data are representative of two independent experiments, respectively.

Cancer Gene Therapy


HCT8 cell line were directly cocultured with native MSC and cocultures were treated with sTRAIL. Untreated cocultures as well as cultures of the respective cell types alone served as controls. Proliferation was assessed by determining cell numbers at the start and at the end of sTRAIL treatment. In comparison with the untreated control, no reduction in growth was observed upon sTRAIL treatment of cocultures of HCT8 with MSC (Figure 3c). Thus, concomitant presence of MSC and sTRAIL was not able to achieve the growth reduction as seen in cocultures of HCT8 with TRAIL-MSC (Figure 3c). As seen before, sTRAIL had no impact on the growth of HCT8 and MSC alone. In addition, growth of HCT8 cells was not altered by coculture with MSC. These data showed that apoptosis induction in TRAILresistant HCT8 cells by cocultured TRAIL-MSC is exerted by transmembraneous TRAIL. Direct contact with native MSC does not alter the response to sTRAIL in the TRAIL-resistant CRC cell line HCT8. Based on these data, we concluded that TRAIL-MSC can overcome TRAIL resistance only in selected CRC cell lines and that HT29 and Colo320DM represent TRAIL-MSC-resistant CRC cell lines. Sensitization to sTRAIL by subapoptotic genotoxic damage has been reported. Therefore, we next wanted to analyze whether relevant chemotherapeutic drugs at subapoptotic doses can sensitize the cell line HT29 to apoptosis induction by TRAIL-MSC.

Subapoptotic genotoxic damage sensitizes TRAIL-MSCresistant CRC cells to TRAIL-MSC-induced apoptosis in vitro To evaluate a putative sensitization, 5-FU or oxaliplatin were added at subapoptotic concentrations to direct cocultures of TRAIL-MSC with TRAIL-MSC-resistant HT29 cells. To ensure subapoptotic damage, the used concentrations of 5-FU and oxaliplatin were adjusted to the dose–response curves of the analyzed cells (Supplementary Figure). For 5-FU, the used dose of 3 mM represented a dose well below the IC at 96-h treatment of 90% for MSC (IC50 ¼ 5.0±2.2 mM, IC90 ¼ 765±150 mM) as well as for HT29 (IC50 ¼ 0.9±0.1 mM, IC90 ¼ 65±18 mM). Similarly, the used dose of 3 mM oxaliplatin represented a dose well below an IC of 90% for MSC (IC50 ¼ 2.4±1.6 mM, IC90 ¼ 94.5±10.2 mM) as well as for HT29 (IC50 ¼ 0.3± 0.04 mM, IC90 ¼ 15.3±4.6 mM). Nonetheless, these doses were equivalent to about 100% (5-FU) and 50% (oxaliplatin) of the respective in vivo serum levels.16,17 Cleavage of PARP, caspase-8 and caspase-3 were analyzed as markers of apoptosis induction in whole culture lysates and in detached cells if present. Cocultures with WT-MSC as well as the respective untreated cocultures served as controls. No detached cells were observed in all cocultures with WT-MSC as well as in the untreated coculture with TRAIL-MSC. Accordingly, no relevant cleavage of PARP and caspases was detectable in cocultures with WT-MSC, either treated or untreated, as well as in the untreated coculture with TRAIL-MSC (Figure 4).

235

Figure 4 TRAIL-MSC-resistant CRC cells are sensitized to TRAILMSC-induced apoptosis by subapoptotic genotoxic damage in vitro. HT29 cells were cultivated in direct coculture with WT-MSC (WT) and TRAIL-MSC (TRAIL) for 72 h. Cocultures were supplemented with 3 mM 5-FU or 3 mM oxaliplatin as stated. Whole cell lysates—that is adherent and detached cells (w)—as well as—if present— detached cells from a separate sample (d) were harvested and analyzed by western blot for cleavage of PARP and caspases-8 and caspase-3. Actin served as a loading control. Arrows indicate the respective molecular weight standards in kDa. The size of uncleaved PARP (116 kDa), cleaved PARP (85 kDa), procaspase-8 (55 kDa), cleaved fragments of caspase-8 (43, 41 and 18 kDa), procaspase-3 (32 kDa) and cleaved fragments of caspase-3 (20 and 17 kDa) are indicated. Data are representative of three independent experiments.

This confirmed first that the chosen concentrations of 5-FU and oxaliplatin represented subapoptotic doses, and second that TRAIL-MSC alone were not able to induce apoptosis in HT29 cells. However, treatment of the TRAIL-MSC-containing coculture with 5-FU or oxaliplatin resulted in the occurrence of detached cells. In these detached cells, extrinsic apoptosis induction was evident as marked by the cleaved fragments of caspase-8, caspase-3 and PARP, resulting in near elimination of the respective uncleaved protein. In the adherent cells of TRAIL-MSCcontaining coculture, cleavage of all three proteins was evident upon treatment with 5-FU, but only cleavage of PARP was observed upon oxaliplatin treatment (Figure 4). These data suggest that subapoptotic damage sensitizes TRAIL-MSC-resistant CRC cells to TRAIL-MSC-induced apoptosis. We hypothesized that apoptosis induction by concomitant presence of TRAIL-MSC and subapoptotic damage can effectively inhibit the growth of the TRAIL-MSC-resistant cell line HT29 in vitro.

Effective growth inhibition of TRAIL-MSC-resistant CRC cells in vitro by concomitant presence of TRAIL-MSC and subapoptotic genotoxic damage This analysis was performed in direct coculture of TRAIL-MSC-resistant DsRed-HT29 cells with TRAILMSC concomitantly treated with subapoptotic 5-FU doses. Cocultures with WT-MSC as well as the respective

Cancer Gene Therapy


236

untreated cocultures served as controls. In comparison with baseline values, total cell number as well as the number of detached cells was counted and the fraction of DsRed-positive cells—representing HT29 cells—was determined by flow cytometry. Concomitant presence of 5-FU in the TRAIL-MSCcontaining cocultures resulted in a reduction of HT29 cells in the adherent cell population in comparison with the baseline. In contrast, untreated cocultures with TRAIL-MSC as well as treated and untreated cocultures with WT-MSC showed a continuous increase in the number of HT29 cells. This differential response was accompanied by a clearly increased number of detached cells in the treated TRAIL-MSC-containing coculture, while only few detached cells occurred in the respective controls (Figure 5). In comparison with the untreated WT-MSC-containing coculture, 5-FU treatment of the WT-MSC-containing coculture resulted in reduced growth of HT29 cells, but in no relevant occurrence of detached cells (Figure 5). This observation confirmed that the chosen 5-FU concentration represented a growthretarding, but subapoptotic dose.

These data show that the simultaneous presence of TRAIL-MSC and subapoptotic damage by 5-FU can effectively inhibit the growth of TRAIL-MSC-resistant CRC cells in vitro. Based on this observation, we hypothesized that a combination of TRAIL-MSC and 5-FU can inhibit the xenograft growth of TRAIL-MSCresistant HT29 cells.

Systemic low dose 5-FU and tumorintegrated TRAIL-MSC do not inhibit growth of TRAIL-MSCresistant CRC cells in vivo The proposed effect of combined TRAIL-MSC and subapoptotic damage on the growth of TRAIL-MSCresistant HT29 cells in vivo was studied in nude mice using a mixed TRAIL-MSC/HT29 xenograft. 5-FU was systemically applied at a dose and schedule reflecting clinically relevant schedules and serum levels. Xenograft growth was assessed by tumor volume and tumor weight. In mice receiving 5-FU on days 0–4, growth of pure HT29 xenografts was reduced compared with shamtreated mice (Figure 6). This confirmed that the chosen dosage and schedule of 5-FU treatment represented a

Figure 5 Concomitant presence of TRAIL-MSC and subapoptotic genotoxic damage effectively inhibits the growth of TRAIL-MSC-resistant CRC cells in vitro. DsRed-HT29 cells were directly cocultured with WT- or TRAIL-MSC, respectively. Directly at the start of cocultures, 5-FU was added at 3 mM. Untreated cocultures served as controls. At the indicated time points, the number of detached cells (gray bars) as well as the proportion of DsRed-positive HT29 cells (white bars) within the total number of adherent cells (black bars) was determined. For comparison, cell numbers at start of coculture (0 h) were set as 100%. Data are representative of two independent experiments.

Figure 6 DsRed-HT29 cells were mixed with GFP-MSC or TRAIL-MSC at a proportion of 4:1 (that is MSC representing 20% of the total cell number) and injected in nude mice subcutaneously (s.c.) into the left or right flank, respectively. DsRed-HT29 xenografts without MSC served as controls. Animals received 5-FU intraperitoneally (i.p.) at 30 mg kg–1 body weight or a respective volume of 0.9% NaCl on days 0–4. On day 25, tumor volumes on both sides of each animal were examined by measuring tumor dimensions in grayscale images, animals were killed, tumors were dissected and weighted. Tumor weight (left diagram) and calculated tumor volume (right diagram) on day 25 are presented as mean±s.d.; n ¼ 3.

Cancer Gene Therapy


growth-retarding dose similar to the in vitro studies (Figure 5). In contrast to our hypothesis, growth of TRAIL-MSC-containing xenografts was not reduced by 5-FU treatment. Rather, the presence of GFP-MSC as well as TRAIL-MSC seemed to revert the growth retardation seen with 5-FU in pure HT29 xenografts (Figure 6). Similar results were obtained when 5-FU was applied at a later time point from days 7 to 9 and 14 to 16 (data not shown). These data show that the sensitization of HT29 cells for TRAIL-MSC by concomitant genotoxic damage as seen in vitro is not generally reproducible in vivo. This suggests that yet unknown factors have an additional influence on the interaction of TRAIL-MSC and TRAIL-MSC-resistant CRC cells in vivo.

Discussion

Our study yielded the following novel results: first, TRAIL-MSC can induce apoptosis in TRAIL-resistant CRC cells, as in the HCT8 cell line. Thereby, they overcome the TRAIL resistance leading to an effective growth inhibition of these cells in vitro and in vivo. In the cell line HCT8, this effect cannot be ascribed to a sensitization mediated by native MSC, but requires the presence of MSC-expressed TRAIL. Second, this effect, however, is limited to selected TRAIL-resistant CRC cell lines and TRAIL-MSC-resistant CRC cell lines like HT29 exist. Third, subapoptotic damage can sensitize TRAILMSC-resistant HT29 cells to TRAIL-MSC in vitro. However, such sensitization could not be achieved for HT29 cells in an in vivo xenograft model. Rather, TRAILMSC seemed to support growth of HT29 xenografts. Taken together, our data support the assumption that differential mechanisms confer TRAIL resistance in different CRC cell lines. Moreover, these findings caution for the clinical use of TRAIL-MSC. The differential response of CRC cell lines to sTRAIL is a well-established fact and is supported by the data of our initial screening. In particular, resistance of the CRC cell lines HCT8 and HT29 to sTRAIL has been described previously.8 The sTRAIL concentration used in our screening is in accordance with that used in other studies. It is sufficient to allow a distinction of TRAILresistant and TRAIL-sensitive cells as a further increase of the sTRAIL dose did not abrogate this distinction. This complies with the observation that an increased sTRAIL concentration—particularly in HT29—does not result in a further increase of proapoptotic effects.8 Regarding the cause of TRAIL resistance, several mechanisms like differential DR expression10 and differential downstream apoptosis induction8 have been described. The data from our study show that HCT8 and HT29 differ in their response to TRAIL-MSC-mediated TRAIL exposure. The data presented herein and our previous work show that apoptosis-inducing activity of TRAIL-MSC is exerted by transmembraneous TRAIL and not by a secreted protein. In contrast to a treatment

with sTRAIL, such cell-based TRAIL treatment may result in a more sustained exposure to higher levels of TRAIL. We, therefore, propose that in HCT8, the prolonged exposure to TRAIL by TRAIL-MSC can overcome a reduced signaling via DR. This corresponds to the observed and well-known differential expression of TRAIL receptors in CRC cell lines. While receptor expression was not predictive for response to sTRAIL, TRAIL-MSC may compensate for the reduced DR4 and DR5 expression in HCT8, but fail to compensate for the even lower expression in HT29. This interpretation suggests a threshold of DR expression for TRAIL-MSC activity. Although we did not investigate other cell lines in detail, a similar mechanism may be assumed for the cell line SW480 as it also showed a resistance to sTRAIL, but sensitivity to TRAIL-MSC-induced apoptosis. Conclusively, HT29 and Colo320DM differ from HCT8 and SW480, as an exposure to TRAIL-MSC could not induce apoptosis in HT29 or Colo320DM. However, concomitant genotoxic damage sensitized HT29 to TRAIL-MSC in vitro in our model. This resembles the well-described sensitization of tumor cells to sTRAIL by various agents. Specifically, a sensitization to sTRAIL by oxaliplatin has been described for HT29.12 A possible mechanism for such sensitization comprises a modulation of TRAIL sensitivity through disruption of mitochondrial function.8 In accordance with these in vitro data, the combination of TRAIL with additional damage has been shown to effectively reduce tumor growth in vivo too.18 Given the lack of relevant apoptosis induction in HT29 cells upon treatment with 5FU, oxaliplatin or TRAIL-MSC alone, we propose a synergistic activity of subapoptotic genotoxic damage and TRAIL-MSC in this particular CRC cell line. Therefore, our finding that a combination of TRAILMSC with genotoxic damage not only failed to reduce growth of HT29 xenografts in two different settings, but rather seemed to support tumor growth, requires a specific explanation. As the growth of pure HT29 xenografts was reduced upon 5-FU treatment, an insufficient activity or intratumoral presence of 5-FU in our model can be precluded. As reported by us,7 MSC may support the growth of CRC xenografts. Thus, the anti-tumor effect of combined TRAIL-MSC and 5-FU may be overridden by an MSC-mediated growth support in our model. Interestingly, a support of in vivo tumor growth has been reported for TRAIL in TRAIL-resistant cells.19 This effect was associated with increased metastasis20 as well as with increased proliferation.21 Several mechanisms have been suggested mainly involving Bcl-2 family-dependent prevention of apoptosis and TRAIL-induced NFkappaB-dependent activation of proinflammatory genes.20 Specifically in HT29, TRAIL can result in activation of IGF-binding protein 3 resulting in NFkappaB activation.22 This, however, does not yet explain the differential response of HT29 to concomitant exposure to TRAIL-MSC and 5-FU in vitro and in vivo. An explanation may lay in the simultaneous induction of apoptosis and proliferation depending on TRAIL concentration, caspase activity and duration of

Cancer Gene Therapy

237


238

incubation. Such complex response has been described for HT29, related to a receptor-proximal apoptosis defect with over-expression of cFLIP.23 Intensity and duration of interaction with TRAIL-MSC as well as caspase activity are likely to differ between the in vitro and in vivo settings in our model. Putative relevant factors include the tumor stroma or residual immune response leading to activation of NFkappaB, thus inhibiting TRAIL-MSC/5FU-induced apoptosis24 in the nude-mice model. In addition, a selection of cells in the inherently inhomogenous tumor cell population seems possible depending on the differential mode of interaction of HT29 and TRAIL-MSC. The relevance of such cell-to-cell variability for differential apoptosis induction by TRAIL has been described.25 It remains speculative whether and which of these features are responsible for the observed differential response in our model. Obviously, mixed xenografts are an inferior model to show a putative clinical relevance as the clinical reality requires a systemic or local application of transduced MSC in the presence of an established tumor. In our own work, we had observed no reduction in tumor growth upon intravenous application of TRAIL-MSC in mice bearing a subcutaneous CRC xenograft.7 As we also observed a low frequency of tumor integration, but a pronounced pulmonary tropism of MSC, we conclude that the missing anti-tumor activity in our model was the result of an insufficient tumor integration of TRAILMSC. However, others have shown that MSC with ectopic gene expression can inhibit growth of pulmonary and subcutaneous xenografts after systemic transplantation.26,27 In accordance with preliminary data in our model, we hypothesize that under optimized conditions, tumor integration of TRAIL-MSC may be enhanced, resulting in a potent anti-tumor activity. Taken together, our data prove that TRAIL-MSC can inhibit the growth of selected TRAIL-resistant CRC cells in vitro and in vivo and may, therefore, represent a tool to overcome TRAIL resistance in tumor treatment. However, such potential clinical use of TRAIL-MSC requires further detailed preclinical studies as our data also prove that TRAIL-MSC-resistant tumor cells exist. Thereby, our data add to the notion that resistance of CRC cells to TRAIL in general—sTRAIL as well as TRAIL-MSC—is conferred by different mechanisms. These mechanisms may gain complexicity through additional, resistanceconferring signaling in vivo resulting in TRAIL-related growth support of selected tumor cells. This cautions for the clinical use of TRAIL-MSC; however, the identification of molecular markers may help to identify tumors, which will respond to such treatment.

Conflict of interest

Dr LP Mueller, J Luetzkendorf and Dr T Mueller have filed a patent on the lentiviral TRAIL-MSC described in this report. All other authors, M Widder, K Nerger and H Caysa, have no conflict of interest to declare.

Cancer Gene Therapy

Acknowledgements

We are grateful to the donors who consented the use of their bone marrow aspirates. We thank Professor Dr K Maeder, Institute of Pharmacy, Martin-LutherUniversity Halle-Wittenberg for the non-invasive in vivo multispectral fluorescence imaging. The study was supported in part by grants from the Federal State of Saxonia-Anhalt and the German Ministry of Education and Research through the Wilhelm-Roux-Program at the Medical Faculty of the Martin-Luther-University Halle-Wittenberg (NBL-3 program 1/09, 3/15, 6/08, 3/27) as well by a grant from the Federal State of SaxoniaAnhalt (FKZ: 3646A/0907). Author contributions: LPM, JL and TM designed the study; LPM, JL, MW, KN, HC and TM performed the research; LPM, JL, HC and TM controlled and analyzed data; LPM, JL and TM wrote the paper.

References 1 Studeny M, Marini FC, Dembinski JL, Zompetta C, Cabreira-Hansen M, Bekele BN et al. Mesenchymal stem cells: potential precursors for tumor stroma and targeted-delivery vehicles for anticancer agents. J Natl Cancer Inst 2004; 96: 1593–1603. 2 Kumar S, Chanda D, Ponnazhagan S. Therapeutic potential of genetically modified mesenchymal stem cells. Gene Ther 2008; 15: 711–715. 3 Ashkenazi A. Targeting the extrinsic apoptosis pathway in cancer. Cytokine Growth Factor Rev 2008; 19: 325–331. 4 Wang S. The promise of cancer therapeutics targeting the TNF-related apoptosis-inducing ligand and TRAIL receptor pathway. Oncogene 2008; 27: 6207–6215. 5 Center MM, Jemal A, Smith RA, Ward E. Worldwide variations in colorectal cancer. CA Cancer J Clin 2009; 59: 366–378. 6 Walther A, Johnstone E, Swanton C, Midgley R, Tomlinson I, Kerr D. Genetic prognostic and predictive markers in colorectal cancer. Nat Rev Cancer 2009; 9: 489–499. 7 Luetzkendorf J, Mueller LP, Mueller T, Caysa H, Nerger K, Schmoll HJ. Growth-inhibition of colorectal carcinoma by lentiviral TRAIL-transgenic human mesenchymal stem cells requires their substantial intratumoral presence. J Cell Mol Med 2010; 14: 2292–2304. 8 Tillman DM, Izeradjene K, Szucs KS, Douglas L, Houghton JA. Rottlerin sensitizes colon carcinoma cells to tumor necrosis factor-related apoptosis-inducing ligandinduced apoptosis via uncoupling of the mitochondria independent of protein kinase C. Cancer Res 2003; 63: 5118–5125. 9 Nawrocki ST, Carew JS, Douglas L, Cleveland JL, Humphreys R, Houghton JA. Histone deacetylase inhibitors enhance lexatumumab-induced apoptosis via a p21Cip1-dependent decrease in survivin levels. Cancer Res 2007; 67: 6987–6994. 10 Drosopoulos KG, Roberts ML, Cermak L, Sasazuki T, Shirasawa S, Andera L et al. Transformation by oncogenic RAS sensitizes human colon cells to TRAIL-induced apoptosis by up-regulating death receptor 4 and death receptor 5 through a MEK-dependent pathway. J Biol Chem 2005; 280: 22856–22867.


11 Luo JL, Maeda S, Hsu LC, Yagita H, Karin M. Inhibition of NF-kappaB in cancer cells converts inflammation-induced tumor growth mediated by TNFalpha to TRAIL-mediated tumor regression. Cancer Cell 2004; 6: 297–305. 12 Toscano F, Fajoui ZE, Gay F, Lalaoui N, Parmentier B, Chayvialle JA et al. P53-mediated upregulation of DcR1 impairs oxaliplatin/TRAIL-induced synergistic anti-tumour potential in colon cancer cells. Oncogene 2008; 27: 4161–4171. 13 Skehan P, Storeng R, Scudiero D, Monks A, McMahon J, Vistica D et al. New colorimetric cytotoxicity assay for anticancer-drug screening. J Natl Cancer Inst 1990; 82: 1107–1112. 14 Dull T, Zufferey R, Kelly M, Mandel RJ, Nguyen M, Trono D et al. A third-generation lentivirus vector with a conditional packaging system. J Virol 1998; 72: 8463–8471. 15 Mueller LP, Luetzkendorf J, Mueller T, Reichelt K, Simon H, Schmoll HJ. Presence of mesenchymal stem cells in human bone marrow after exposure to chemotherapy: evidence of resistance to apoptosis induction. Stem Cells 2006; 24: 2753–2765. 16 Terret C, Erdociain E, Guimbaud R, Boisdron-Celle M, McLeod HL, Fety-Deporte R et al. Dose and time dependencies of 5-fluorouracil pharmacokinetics. Clin Pharmacol Ther 2000; 68: 270–279. 17 Levi F, Metzger G, Massari C, Milano G. Oxaliplatin: pharmacokinetics and chronopharmacological aspects. Clin Pharmacokinet 2000; 38: 1–21. 18 Zhu H, Zhang L, Huang X, Davis JJ, Jacob DA, Teraishi F et al. Overcoming acquired resistance to TRAIL by chemotherapeutic agents and calpain inhibitor I through distinct mechanisms. Mol Ther 2004; 9: 666–673. 19 Malhi H, Gores GJ. TRAIL resistance results in cancer progression: a TRAIL to perdition? Oncogene 2006; 25: 7333–7335.

20 Trauzold A, Siegmund D, Schniewind B, Sipos B, Egberts J, Zorenkov D et al. TRAIL promotes metastasis of human pancreatic ductal adenocarcinoma. Oncogene 2006; 25: 7434–7439. 21 Ehrhardt H, Fulda S, Schmid I, Hiscott J, Debatin KM, Jeremias I. TRAIL induced survival and proliferation in cancer cells resistant towards TRAIL-induced apoptosis mediated by NF-kappaB. Oncogene 2003; 22: 3842–3852. 22 Williams AC, Smartt H, AM HZ, Macfarlane M, Paraskeva C, Collard TJ. Insulin-like growth factor binding protein 3 (IGFBP-3) potentiates TRAIL-induced apoptosis of human colorectal carcinoma cells through inhibition of NF-kappaB. Cell Death Differ 2007; 14: 137–145. 23 Baader E, Toloczko A, Fuchs U, Schmid I, Beltinger C, Ehrhardt H et al. Tumor necrosis factor-related apoptosisinducing ligand-mediated proliferation of tumor cells with receptor-proximal apoptosis defects. Cancer Res 2005; 65: 7888–7895. 24 Plantivaux A, Szegezdi E, Samali A, Egan L. Is there a role for nuclear factor kappaB in tumor necrosis factor-related apoptosis-inducing ligand resistance? Ann N Y Acad Sci 2009; 1171: 38–49. 25 Spencer SL, Gaudet S, Albeck JG, Burke JM, Sorger PK. Non-genetic origins of cell-to-cell variability in TRAILinduced apoptosis. Nature 2009; 459: 428–432. 26 Studeny M, Marini FC, Champlin RE, Zompetta C, Fidler IJ, Andreeff M. Bone marrow-derived mesenchymal stem cells as vehicles for interferon-beta delivery into tumors. Cancer Res 2002; 62: 3603–3608. 27 Kucerova L, Altanerova V, Matuskova M, Tyciakova S, Altaner C. Adipose tissue-derived human mesenchymal stem cells mediated prodrug cancer gene therapy. Cancer Res 2007; 67: 6304–6313.

Supplementary Information accompanies the paper on Cancer Gene Therapy website (http://www.nature.com/cgt)

Cancer Gene Therapy

239