Extracellular vesicle-mediated suicide mRNA/protein delivery inhibits glioblastoma tumor growth in vivo

Abstract

Extracellular vesicles (EVs) are considered as important mediators of intercellular communication, which carry a diverse repertoire of genetic information between cells. This feature of EVs can be used and improved to advance their therapeutic potential. We have previously shown that genetically engineered EVs carrying the suicide gene mRNA and protein—cytosine deaminase (CD) fused to uracil phosphoribosyltransferase (UPRT)—inhibited schwannoma tumor growth in vivo. To further examine whether this approach can be applied to other cancer types, we established a subcutaneous xenograft glioblastoma tumor model in mice, as glioblastoma represents the most common primary brain tumor, which is highly aggressive compared with the original schwannoma tumor model. U87-MG glioblastoma cells were implanted into the flanks of nude SCID mice, and the animals were intratumorally injected with the EVs isolated from the cells expressing EGFP or CD-UPRT. After the intraperitoneal administration of the prodrug 5-fluorocytosine, the tumor growth was assessed by regular caliper measurements. Our data revealed that the treatment with the CD-UPRT-enriched EVs significantly reduced the tumor growth in mice. Taken together, our findings suggest that EVs uploaded with therapeutic CD-UPRT mRNA/protein may be a useful tool for glioblastoma treatment.

Introduction

Extracellular vesicles (EVs) can be conceived as one of the intercellular communication and delivery systems of eukaryotic cells. They are generated and released by budding off from the cell membrane, and different stimuli affect the rate of EV release.^{1, 2, 3} Increasing evidence has demonstrated that EVs carry and transfer ‘biological cargo’ between cells, including mRNAs, noncoding RNA species, proteins and DNA.^{4, 5, 6} Recent studies have illuminated the therapeutic potential of EVs in the context of different diseases such as cancer,⁷ inflammation⁸ or cardiovascular diseases.⁹

Glioblastoma represents the most common and aggressive brain tumor in humans. Despite the aggressive multimodal treatment options, the prognosis is generally poor, and glioblastoma patients have poor life quality.¹⁰ Complete surgical resection is usually not an option as glioblastoma cells invade the neighboring cells in the tumor microenvironment.¹¹ Moreover, glioblastoma cells rapidly develop resistance to chemotherapy, which limits the efficacy of the existing treatment strategies. Therefore, there is an urgent need for development of novel and rational therapy strategies for glioblastoma.

Here we used an in-frame fusion of cytosine deaminase (CD) and uracil phosphoribosyltransferase (UPRT) as the suicide therapeutic molecule that converts 5-fluorocytosine (5-FC) to 5-fluorouracil (5-FU), causing defective DNA replication and apoptosis.^{12, 13} Our approach combines the administration of EVs uploaded with CD-UPRT suicide gene products into the tumor, together with the prodrug 5-fluorocytosine. The recipient cells containing the CD-UPRT fusion protein converts the prodrug 5-FC into 5-FU, leading to cell death. A successful implementation of this protocol in this study resulted in ~70% reduction in tumor growth in a xenograft mice model.

Materials and methods

Plasmids

The plasmids pCD-UPRT-EGFP and the control plasmid p-EGFP were constructed earlier.¹⁴

Cells

HEK293T, U87-MG, U251-MG and E98 cells were cultured in Dulbecco’s modified Eagle's medium (Sigma-Aldrich, St. Louis, MO, USA) containing 10% fetal bovine serum (Sigma-Aldrich), 2% l-glutamine (Sigma-Aldrich) and 0.2% normocin (Invivogen, San Diego, CA, USA). All cells were incubated at 37 °C in a 5% CO₂ atmosphere. Cells were determined to be mycoplasma negative by testing with LookOut Mycoplasma PCR Detection Kit (Sigma-Aldrich).

EV isolation

EV isolation was carried out as described previously.¹⁵ Shortly, the transfected HEK293T cells were cultured in EV-free media, and collected and centrifuged for 10 min at 300 g at 4 °C. The supernatant was filtered by a 0.8 μm filter, and then centrifuged for 20 min at 2000 g. The final supernatant was subjected to ultracentrifugation at 110 000 g for 90 min. Pellets were resuspended with 50 μl ice-cold phosphate-buffered saline and stored at −20 °C for further use.

Chemicals

5-FC and propidium iodide (PI) were purchased from Sigma-Aldrich.

Cell viability assay

U87-MG cells (1 × 10⁴ per well) were plated in 24-well plates and allowed to attach overnight at 37 °C in a humidified atmosphere containing 5% CO₂. Cells were then preincubated either with the isolated CD-UPRT EVs or with the control EGFP EVs (multiplicity of infection (MOI): 10) for 24 h. Subsequently, 5-FC (250 μg ml⁻¹) was added into the culture for another 24 h. Then, EVs and 5-FC were replaced and cells were incubated for additional 24 h. Cells were then harvested, stained with Trypan blue and counted by using a hemocytometer. Three independent experiments were performed and statistical significance was calculated by Student’s t-test.

TUNEL assay

To evaluate the apoptotic features caused by the uptake of CD-UPRT EVs in a tumor model in mice, terminal deoxynucleotidyl transferase (TDT)-mediated deoxy-UTP nick-end labeling (TUNEL) assay was performed according to the manufacturer’s instructions (Roche Applied Science, Penzberg, Germany). The paraffin-embedded tissue slides were pre-treated with permeabilization buffer (containing 0.1% sodium citrate and 0.1% Triton X-100), and labeled with fluorescein to detect incorporated nucleotide polymer. Nuclei were counterstained with DAPI (4',6-diamidino-2-phenylindole), and slides were analyzed by a fluorescence microscope. TUNEL-positive apoptotic cells were quantified by counting at least 500 nuclei in each random section obtained from the tumor specimens. Three independent counting were performed, and the values are expressed as mean±s.d. Statistical significance was calculated by Student’s t-test.

Note that EV-associated green fluorescence signal was mostly lost in the tissue sections, most likely due to paraffin embedding.

Cell-cycle analysis

U87-MG cells (4 × 10⁵ cells per well) were seeded in 6-well plates, and incubated at 37 °C overnight in a humidified atmosphere containing 5% CO₂. Cells were then preincubated with the isolated CD-UPRT EVs or the control EGFP EVs (MOI: 20) for 24 h, followed by treatment with 250 μg ml⁻¹ of 5-FC. EVs and 5-FC were replaced every 24 h. Cells were harvested at 72 h of incubation with EVs, and fixed in 1 ml of 85% ice-cold ethanol and stored at −20 °C. After washing once with phosphate-buffered saline, cells were labeled with PI solution (50 μg ml⁻¹) containing RNAse A, and subjected to cell-cycle analysis by a BD LSRFortessa flow cytometer (BD Biosciences, San Jose, CA, USA). The experiment was performed in triplicates and statistical significance was calculated by Student’s t-test.

Caspase-3/7 activity assay

U87-MG cells (5 × 10³ cells per well) were seeded in a 96-well plate and treated as described in the cell-cycle analysis. Seventy-two hours after incubation with EVs, caspase-3/7 activity was measured by Apo-ONE Homogeneous Caspase 3/7 Assay Kit (Promega, Madison, WI, USA) according to the manufacturer’s instructions. Three independent experiments were performed, and statistical significance was calculated by Student’s t-test.

Generation of tumor spheroids

For spheroid generation, cell suspensions with a density of 3x10⁴ cells per ml were prepared from the glioblastoma cell lines U87-MG, U251-MG and E98, containing 0.3% methylcellulose (Sigma-Aldrich). A total of 100 μl of cell suspensions was transferred into a 96-well round bottom plate (CELLSTAR 650185; Greiner Bio-one, Kremsmunster, Austria), and centrifuged for 5 min at 800 r.p.m. To allow formation of spheroids, cells were incubated in DMEM containing 5% fetal bovine serum for 3 days. Finally, 50 μl of fresh DMEM containing 10% fetal bovine serum was added on the third day. The spheroids were preincubated for 24 h with isolated CD-UPRT EVs or control EVs (MOI: 100). Subsequently, spheroids were treated with 250 μg ml⁻¹ of 5-FC. After 48 h, both EVs and 5-FC were replaced, and incubated for another 24 h. The images of spheroids were taken by an Axiovert microscope (Carl Zeiss, Jena, Germany). Circumferences (C) of at least 10 spheroids were measured by Axiovison Rel.4.8 (Carl Zeiss, Jena, Germany). Volumes of the spheroids were calculated by the formula , where radii (r) is . Statistical significance was calculated from at least three independent experiments by using Student’s t-test.

MTT assay

The MTT (3-(4,5-dimethythiazol-2-yl)-2,5-diphenyl tetrazolium bromide) assay was performed according to the manufacturer’s instructions with some slight modifications. An aliquot of 20 μl of CellTiter Blue Reagent (G8080; Promega) was added per 100 μl of media, and the spheroids were incubated for 4–6 h. Then, 100 μl of the suspension was transferred into a 96-well plate. The fluorescence was recorded at 560/590 nm.¹⁶ At least three independent experiments were performed, and statistical significance was calculated by Student’s t-test.

Establishing subcutaneous xenograft glioblastoma model in mice

A total of 2x10⁶ U87-MG cells were implanted into the flanks of nude SCID mice (n=11 per group, female, 10 weeks old). Ten days after implantation, the animals were randomly divided into two groups (control and treatment). The EGFP EVs (control group) and CD-UPRT-EVs (treatment group) were injected into mice intratumorally (MOI: 100). At 48 h after EV injection, prodrug 5-FC (250 μg ml⁻¹) was injected into mice intraperitoneally. EV samples were isolated freshly each week to minimize the possible effects related to EV stability. This treatment cycle was continued two times per week during a period of 4 weeks. Tumor growth was monitored by regular caliper measurements. Tumor volume was calculated by using the modified ellipsoid formula ‘0.5x(L × W2)’, where L is the length and W the width of the tumor. Data represent mean±s.e.m. of all animals within the groups (P=0.0021; unpaired t-test). This experiment was performed two times. This study had been approved by the Medical University of Vienna Ethic Commission under the protocol number: BMWF 66.009/0227-II/3b/2011.

Results

Considering the capacity and ability of EVs to transfer biological molecules between the cells, we aimed to use the engineered EVs to inhibit tumor growth in a mouse model of glioblastomas (Figure 1). We first introduced an inactive, suicide therapeutic molecule (mRNA and/or protein) into the EVs, which does not naturally exist in the cell. In the next step, we administered a systemic prodrug that is activated in response to the activity of the therapeutic molecule. We used an in-frame fusion of CD and UPRT as the suicide therapeutic molecule. This combination was shown to be effective in treatment of different cancers.^{17, 18, 19, 20} 5-FC is converted into 5-FU by the activity of CD, and successively 5-FU is converted into 5-fluoro-deoxyuridine monophosphate upon the activity of UPRT. Irreversible binding of 5-fluoro-deoxyuridine monophosphate to thymidylate synthase results in depletion of deoxythymidine triphosphate. Eventually, DNA replication is interrupted, and cells undergo apoptosis^{12, 13} (Figure 2).

Figure 1

Model of specific anticancer treatment via genetically engineered extracellular vesicles. An expression vector containing the desired gene is transfected into recipient cells. Plasmid DNA is transcribed and translated into protein. Protein and mRNA are transferred to extracellular vesicles (EVs), which can be isolated from the medium via ultracentrifugation. Harvested EVs can readily be used for transferring genetic material to recipient cancer cells for anticancer treatment. Tumor growth can be monitored by caliper or in a clinical setting via imaging modalities (e.g. magnetic resonance imaging).

Figure 2

Proposed mechanism of cell killing brought by the action of cytosine deaminase and uracil phosphoribosyltransferase (CD-UPRT) on the prodrug 5-fluorocytosine (5-FC). The prodrug 5-FC is converted into 5-fluorouracil (5-FU) by the action of CD. 5-FU is immediately converted into 5-fluoro-deoxyuridine monophosphate (5-dUMP) by the activity of UPRT. 5-dUMP is an irreversible inhibitor of thymidine synthetase and causes deoxythymidine triphosphate (dTTP) depletion within the cell. Eventually, DNA replication is inhibited, and cells undergo apoptosis.

EVs carrying CD-UPRT, combined with 5-FC administration, inhibit glioblastoma tumor growth in vitro

HEK293T cells were transfected with an expression vector harboring the CD-UPRT fusion cassette, and the EVs released into culturing media were isolated by ultracentrifugation 3 days after the transfection. The isolated EVs were characterized by the nanoparticle tracking analysis on a NanoSight NS500 Instrument (Malvern Instruments, Malvern, UK) (Supplementary Figure 1a–d). In addition, we characterized EVs for the CD63 expression by using the CD63 ExoELISA Kit (System Biosciences, Palo Alto, CA, USA) (Supplementary Figure 2). To test the functionality of CD-UPRT mRNA/protein-carrying EVs, we treated U87-MG cells with EVs (7x10¹⁰ EV particle per well) for 24 h, followed by the treatment with the prodrug 5-FC. After 48 h, cell viability was assessed by the MTT assay. As shown in Figure 3a, we observed that the treatment of U87-MG cells with the EVs carrying CD-UPRT protein/mRNA, when combined with the exposure to 5-FC, inhibits glioblastoma cell viability by ~30%, compared with the control cells treated with the EGFP carrying EVs and 5-FC. EV uptake by U87-MG cells was confirmed by live-cell imaging (Supplementary Figure 3). To investigate the mechanism of defective cell proliferation of the U87-MG cells treated with the EVs carrying the suicide therapeutic molecule and 5-FC, we analyzed the cell-cycle distribution of the cells, and observed that the sub-G1 population of the cells treated with the EVs uploaded with the CD-UPRT system was slightly increased after the treatment with prodrug 5-FU (Figure 3b). In parallel, we found that cellular caspase-3/7 activity was relatively higher, to some extent, in the cells containing CD-UPRT/5-FC system, indicating apoptotic cell death triggered by the suicide therapy tool (Figure 3c). Taken together, these results provide evidence for the functionality of the therapeutic molecule transfer via engineered EVs in vitro.

Figure 3

Functional characterization of extracellular vesicle (EVs) in vitro. (a) U87-MG cells were pre-treated with cytosine deaminase and uracil phosphoribosyltransferase (CD-UPRT)-loaded or EGFP-loaded EVs (multiplicity of infection (MOI): 10) for 24 h, and 5-fluorocytosine (5-FC) was added for another 48 h. Cells were counted with a hemocytometer on a conventional light microscope. (b) U87-MG cells were prepared as in (a), and fixed in ice-cold 85% ethanol. After staining with a propidium iodide (PI) solution, cells were subjected to cell-cycle analyses by flow cytometry. (c) Cellular caspase-3/7 activity was measured by Apo-ONE Homogenous Caspase 3/7 Assay Kit (Promega) (Student's t-test; *P<0.05, **P<0.01).

Genetically engineered EVs target glioblastoma tumor cells in a three-dimensional culture model

Next, we investigated the therapeutic potential of the CD-UPRT carrying EVs in a three-dimensional culture system. First, we generated spheroids from three different glioblastoma cell lines (U87-MG, E98 and U251-MG) by using a methylcellulose approach. After confirmation of the spheroid formations by microscopic visualization at day 6, EVs derived from HEK293T cells expressing EGFP or CD-UPRT were added on the spheroid culture (MOI: 5x10³). After 24 h of preincubation with EVs, the spheroids were treated with 5-FC for another 48 h. Images of spheroids were taken, and the volumes of the spheroids were calculated. Our data showed that the volumes of the spheroids treated with the CD-UPRT EVs and 5-FC were significantly reduced in all glioblastoma cell lines used in the study, without any significant change in the number of spheroids among the groups (Figures 4a–c). Notably, the CD-UPRT EV treatment only or the 5-FC treatment only did not affect the growth of spheroids in any cell line tested in this study (Figures 4a–c). In addition, we determined the cell viability in the spheroid system by using MTT assay, and found that the cells treated with the CD-UPRT EVs and 5-FC showed reduced cell viability compared with the control cells treated only with the CD-UPRT EVs (Figures 4a–c). Considering that tumor spheroids provide a new intermediate model between in vitro and in vivo models in cancer research,²¹ our approach with genetically engineered EVs can be a useful tool for effective inhibition of tumor growth in a three-dimensional culture system.

Figure 4

Cytosine deaminase and uracil phosphoribosyltransferase/5-fluorocytosine (CD-UPRT/5-FC) system-targeted glioblastoma tumor cell spheroids. Spheroids were generated from glioblastoma cells U87-MG (a), E98 (b) and U251-MG (c), and pre-treated with extracellular vesicle (EVs) (multiplicity of infection (MOI): 5 × 10³) for 24 h. The prodrug 5-FC (250 μg ml⁻¹) was subsequently added into the culture for another 48 h. EVs and 5-FC were then replaced and incubated for additional 24 h. The spheroids were visualized, and photos were taken by an Axiovert microscope. Representative pictures of spheroids from each cell line were presented (a–c). Scale bar, 2 mm. The volumes were calculated after measurements of circumference and radii of spheroids by Carl Zeiss Zen Software (Jena, Germany). Cell viability was measured by MTT (3-(4,5-dimethythiazol-2-yl)-2,5-diphenyl tetrazolium bromide) assay, and the data were normalized to the values obtained from the CD-UPRT EV treatment only. Statistical significance between the cells treated with CD-UPRT EVs plus 5-FC and the cells treated with only CD-UPRT EVs was calculated by Student’s t-test (*P=0.03; **P<0.01).

Inhibition of glioblastoma growth in vivo via triggering apoptosis

To test our approach in an in vivo setting, we used a subcutaneous xenograft model of glioblastoma in mice.²² We implanted U87-MG cells into the flanks of nude SCID mice (n=11 per group). Ten days after implantation, we divided the animals randomly into two groups, and injected EGFP EVs (control group) and CD-UPRT-EVs (treatment group) intratumorally. At 48 h after EV injection, we injected the prodrug 5-FC intraperitoneally. We continued this treatment cycle two times a week, for 4 weeks, and monitored tumor growth by regular caliper measurements. We found ~70% reduction in tumor growth in the treatment group (Figures 5a and b). Further, we isolated RNA and prepared tissue sections from two randomly chosen tumors from each group, and analyzed the gene expression levels of PCNA as a proliferation marker. We found that the mRNA levels of PCNA were significantly reduced (~50%) in the treatment group compared with that in the control group (Figure 5c). In addition, the tissue sections prepared from the tumors established in mice were subjected to TUNEL staining for in situ cell death detection (Figure 5d). We observed that the tumor tissues treated with the CD-UPRT EVs showed ~3-fold higher apoptotic cell death compared with the tumor tissues treated with the EGFP EVs (Figures 5d and e). Thus, our findings suggest that EVs carrying the CD-UPRT system can effectively inhibit glioblastoma tumor growth in a subcutaneous xenograft mice model.

Figure 5

Cytosine deaminase and uracil phosphoribosyltransferase (CD-UPRT)-loaded extracellular vesicles (EVs) inhibit glioblastoma (GBM) tumor growth in vivo. (a) A total of 2x10⁶ U87-MG cells were subcutaneously implanted into the flanks of nude SCID mice. At 10 days after implantation, animals were randomly divided into two groups (control and treatment). Animals were treated with EGFP-loaded EVs (control) or CD-UPRT-loaded EVs (treatment) two times per week. Prodrug 5-FC was injected intraperitoneally (250 μg ml⁻¹) after EV injections. This experiment was performed two times. (b) Comparison of tumor growth in groups by manual caliper measurements (**P=0.0021; unpaired t-test). (c) Real-time PCR analysis of PCNA (proliferating cell nuclear antigen) expression in tumor tissues: total RNA was isolated from two randomly selected tumors from each group as described in (a). mRNA levels of PCNA were detected by real-time PCR and the data were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH). The values are expressed as mean± s.d. from three independent experiments (Student’s t-test; **P<0.01). (d) TUNEL staining: Paraffin-embedded tissue slides were pre-treated with permeabilization buffer (containing 0.1% sodium citrate and 0.1% Triton X-100) and labeled with fluorescein to detect incorporated nucleotide polymer and analyzed by a fluorescence microscope. (e) TUNEL-positive green cells were normalized to the total number of the counted cells. Three independent counting were performed, and the values are expressed as mean±s.d.. Asterisks indicate statistical significance (Student’s t-test; **P^<0.01). EGFP, enhanced green fluorescent protein; ND, not detected; 5-FC, 5-fluorocytosine.

Discussion

We have previously described an application of suicidal mRNA/protein-containing EVs in schwannoma.¹⁴ In this study, we extended the therapeutic application of EVs carrying suicidal genes to glioblastoma. As glioblastoma is a devastating disease, with a limited response to current therapy regimens, implementation of CD-UPRT carrying EVs could be beneficial for patient outcome. Current gene therapy approaches aim specifically for monogenetic diseases rather than multigenetic diseases such as cancer. Use of viral vectors is a common strategy in the gene therapy field. However, there are certain drawbacks associated with the viral methods: (i) immune response to viral antigens, which limits the therapy efficacy, (ii) subsequent inflammation that leads to severe complications in patients. On the contrary, EVs can be collected from transfected non-tumorigenic human cells, or alternatively from the cells of the affected individual, which would diminish severe immune reactions against viral gene therapy. Additionally, we have recently proposed a mechanism for the mRNA uptake by EVs, demonstrating that mRNA molecules having a conserved ‘zip-code’ sequence in their 3′-untranslated regions are more likely to be loaded into EVs.¹⁵ On the basis of our findings, the 3′-untranslated region sequence of any given mRNA molecule can be engineered to possess a zip-code sequence, allowing improved delivery of the mRNA/protein of interest into target cells/tissues through EVs.

One limitation of our proposed protocol is the route of EV delivery. Our protocol involves the direct intratumoral injection of EVs; however, direct injection may not be possible for the treatment of tumors that are not readily accessible. Systemic delivery of EVs can be proposed to overcome this problem, but this approach needs to be addressed carefully, in respect to the parameters including stability and efficacy. A protocol published by El-Andaloussi et al.²³ has proven the possibility of systemic small interfering RNA-loaded exosome delivery. In this report, the authors propose that exosomes can be engineered so that specific ligand and small interfering RNAs can be effectively loaded into these engineered exosomes. Importantly, this study provides evidence for the small interfering RNA-loaded exosomes to cross the blood–brain barrier and exert the functions brought by small interfering RNAs. It should nevertheless be noted that exosomes may be subject to systemic clearance by the liver.²³ Taken together, our findings provide a novel area of application of suicide gene carrying EVs.