Lentiviral Transfer of an Inducible Transgene Expressing a Soluble

Home , GAS1

ORIGINAL ARTICLE Lentiviral transfer of an inducible transgene expressing a soluble form of Gas1 causes glioma cell arrest, apoptosis and inhibits tumor growth ALo´pez-Ornelas1, T Mejıá-Castillo2, P Vergara2 and J Segovia2 1Departamento de Farmacologıá, Centro de Investigacioń y de Estudios Avanzados del IPN, Av. Instituto Politećnico Nacional, Me´xico, DF, Me´xico and 2Departamento de Fisiologıá, Biofı´sica y Neurociencias, Centro de Investigacioń y de Estudios Avanzados del IPN, Av. Instituto Politećnico Nacional, Me´xico, DF, Me´xico

Gliomas are the most frequent primary tumors of the central nervous system, and their clinical prognosis remains very poor. Because of the characteristics of gliomas, gene therapy appears as a potentially relevant strategy for their treatment. However, the inability of viral vectors to transfer the therapeutic genes to a significantly high number of tumor cells, due to their limited diffusion and distribution, remains a critical obstacle for their application treating gliomas. We have demonstrated that the overexpression of growth arrest specific1 (Gas1) induces cell arrest and apoptosis and eliminates glioma cells in vitro and when implanted in mice. To improve the therapeutic range of Gas1, we generated lentiviral vectors coding for a soluble form of Gas1. Here, we show that cells infected with this virus produce the mutant protein, that acting both in autocrine and paracrine manners, causes death of infected and neighbor cells, thus importantly enhancing the effect of Gas1. Furthermore, the administration of this vector, or cells expressing it, inhibit the growth of tumors inoculated in mice. We present a gene therapy strategy that increases the effect of the therapeutic molecule by eliminating not just the infected cells that express Gas1, but neighbor non-infected cells. Cancer Gene Therapy (2011) 18, 87–99; doi:10.1038/cgt.2010.54; published online 1 October 2010 Keywords: glioma; lentivirus; Gas1; apoptosis; cell arrest

Introduction distribution within the brain parenchyma is very limited, thus they can only infect cells near to the site where the 3–5 Gliomas have an overall incidence rate of B15 per vector is administered. This constraint is particularly 100 000 person-years, and their clinical prognosis remains serious in the case of gliomas, where cells migrate and very poor, particularly for glioblastoma multiforme, the infiltrate into the surrounding tissue. The migrating most malignant form, with a median survival of roughly ability of glioma cells is one of the characteristics that 1 year after diagnosis.1 Although current treatments reduce the effectiveness of surgical resections; thus a gene include surgical resection, chemotherapy and radiotherapy approach that will affect more tumor cells than therapy, no significant improvement in survival has those directly expressing the therapeutic molecule will occurred in years. These facts underlie the need for new enhance its effect by eradicating the cells that produce the therapeutic approaches for these tumors, and gene molecule as well as neighbor cells, by both autocrine and therapy is a promising one. However, the efficacy of gene paracrine mechanisms. therapy has been hindered by several technical aspects, Growth arrest specific1 (Gas1) is a protein anchored to and one of the most critical is the incapacity of the vectors the external cell membrane by a glycosyl-phosphatidyli- 6 to transfer transgenes to a sufficiently large number of nositol (GPI) molecule that is capable of inducing cell cells to significantly reduce the bulk of the tumor.2 Viral arrest and apoptosis of different tumor and neural cells, 7–10 vectors are considered the most efficient systems to including gliomas. We also showed that Gas1 exhibits transfer therapeutic transgenes, but their diffusion and significant structural homology with Glial-cell-line-derived neurotrophic factor (GDNF) receptors (GFRas) and demonstrated that Gas1 inhibits GDNF-mediated Correspondence: Dr J Segovia, Departamento de Fisiologıá, intracellular signaling cascades, therefore, shutting down Biofı´sica y Neurociencias, Centro de Investigacioń y de Estudios 10–12 Avanzados del IPN, Av. Instituto Politećnico Nacional # 2508, cellular survival pathways. As the effect of Gas1 Me´xico 07300, DF, Me´xico. interfering with the function of the GFRa–GDNF-Ret E-mail: [email protected] complex is due to its structural similarity to GFRas, and Received 26 December 2009; revised 7 May 2010; accepted 14 June it is known that GFRa1 can be released from cells and in 2010; published online 1 October 2010 this soluble form binds and interacts with GDNF, at the Soluble Gas1 inhibits glioma growth ALo´pez-Ornelas et al 88 level of the first cysteine-rich domain of the GFRa1,13,14 we hypothesized that a secreted, soluble Gas1 protein, could exert its inhibitory effect both in the cells that produce soluble Gas1 and in neighbor cells, thus extending its therapeutic capacity to treat tumors. More- over, there is a report indicating that soluble Gas1, produced in COS7 cells, is functional.15 Taken all this information together, we decided to generate Tetracycline-inducible lentiviral vectors capable of expressing a mutant, secretable, form of Gas1 and to test the capacity of this protein causing growth arrest and apoptosis of glioma cells. The lentiviral vector encodes a Gas1 protein truncated at arginine 315, lacking the GPI consensus sequence, thus, this protein will not be anchored to the cell membrane, and will be secreted from the cell.14,16 Lentiviral vectors effectively infected C6 glioma cells, which produced a protein capable of Figure 1 Schematic representation of the different forms of inducing cell death in in vitro assays. Conditioned media Gas1. Full-length protein, including Phe 345 (Gas1); the truncated, taken from infected cultures caused cell death in soluble form (tGas1) including Arg 315, and the fusion proteins for independent cultures of glioma cells. Moreover, we each one of the forms, containing the V5 epitope (Gas1-V5 and showed that Gas1 was responsible for the effect, because tGas1-V5). when Gas1 was immunoprecipitated from conditioned medium, the medium without Gas1 did not prevent cell proliferation. Finally, we directly administered both the Gas1-producing lentivirus, and infected Gas1-producing gas1 (50CACCGCGATGGTGGCCGC30); the sequences cells to gliomas implanted in nude mice, and observed corresponding to the 30end of each of the fragments are significant inhibition of tumor growth. Our results show the following: truncated forms (50-ATACCCGGACCCG that the lentiviral-mediated transfer of a secretable form CGTCC-30 and 50-CCCGGACCCGCGTCCATT-30), and of Gas1 effectively induces cell death by autocrine and full length (50-GACGAACCCGGCGAGAAA-30 and paracrine mechanisms, and consequently importantly 50-GACGAACCCGGCGAGAAAATT-30). To amplify, enhances the potential capacity of Gas1 as a potential by PCR, the different fragments, the temperature and adjuvant for the treatment of gliomas. cycles were the following: 94 1C (1 cycle, 5 min), 94 1C, 63 1C (35 cycles) and 72 1C for denaturalizing, alignment and extension (1 min each step), respectively, and a final extension step 72 1C (1 cycle, 10 min). The vectors were Materials and methods sequenced to confirm the absence of unwanted mutations. Construction of plasmids expressing full-length and Subsequently, the four input vectors were subcloned, truncated forms of Gas1 according to the manufacturer’s instructions, by homo- To generate full-length and truncated forms of Gas1 we logous recombination, into the lentiviral vector pLenti/TO/ amplified, by PCR, the different cDNA fragments from V5-DEST (Invitrogen). In this vector, the expression of the the pGfa2-Gas1 vector.7 Four fragments were generated: gas1 cDNAs is driven by the CMV promoter; however, the two starting from the site of the initiation of translation lentiviral promoter also possesses a consensus repressor up to the codon for Arginine 315, thus obtaining the sequence, thus when the repressor element is co-infected in truncated form, lacking the GPI consensus signal, and the same cells, transgene transcription is inhibited, but it is two full-length fragments that included up to phenylala- activated in the presence of Tetracycline, thus allowing nine 345, the last amino acid of the Gas1 sequence. gene expression when the viral construct is used, and a However, one of each type of constructs (full length and Tetracycline-inducible system when both vectors are truncated) had a stop codon in the 30 oligonucleotide to present in the same cells. detain translation, and the other pair of constructs had no stop codons, thus when subcloned into the pENTRTOPO Transfection of Gas1-expressing plasmids expression vector (Invitrogen, Carlsbad, CA), the V5 To corroborate that the vectors expressed both a full- epitope tag would also be expressed as a fusion protein length Gas1 protein as well as a truncated, secretable, (Figure 1). The fragments are denominated gas1, for the Gas1 variety, they were transfected into rat glioma C6 full-length sequence; gas1-V5 for the fusion protein with cells, and their effect on cell proliferation was determined, the V5 epitope; tgas1, for the truncated form and tgas1- as a preliminary step before the construction of the V5 for the truncated form with the V5 epitope. The 50 lentivirus, as in the lentiviral pLenti/TO/V5-DEST PCR oligonucleotide was the same for all reactions, and plasmids, the expression of transgenes in eukaryotic cells contained a CACC sequence for directional subcloning is driven by the CMV promoter. into the pENTRTOPO vector, included a Kozak In all, 1 Â 106 C6 cells were seeded in 60 mm plates and sequence, and seven bases corresponding to the 50end of grown until they reached a confluence of 80–90%, as

Cancer Gene Therapy Soluble Gas1 inhibits glioma growth ALo´pez-Ornelas et al 89 previously described.7,9 After 24 h, cells were transfected Construction of lentiviral vectors using Lipofectamine 2000 (Invitrogen) with each of the Viral producer cells four lentiviral construction plasmids. As a control, one The 293FT cells (Invitrogen) were used for the packaging plate was treated only with Lipofectamine 2000 and of lentivirus. Cells were grown in DMEM with high another one was only exposed to the media changes glucose (Invitrogen), with 10% fetal bovine serum specific to the transfection protocol. All transfections (Invitrogen), MEM non-essential amino acids 0.1 mM were carried out under the same conditions: 8 mgof (Invitrogen), 6 mML-glutamine (Sigma-Aldrich, St Louis, plasmid and 24 ml of Lipofectamine 2000. For some MO), 1 mM sodium pyruvate (Invitrogen) and 1% experiments, 48 h after transfection, the culture medium Penicillin/Streptomycin (PAA, Ontario, Canada). of transfected cells was collected and filtered with 0.22 mm filters (Millipore, Bedford, MA), for further experiments. Viable transfected cells were counted using Trypan blue; Construction of lentiviral vectors proteins and RNA were isolated using Tripure (Roche, Briefly, DNA complexes were prepared using Lipofecta- Indianapolis, IN) to perform western blot analysis and mine 2000, and the virus was obtained as described by the RT–PCR, respectively, as previously described, to deter- manufacturer (ViraPower T-Rex Lentiviral Expression mine the expression of the transgenes and the different System, Invitrogen). Two complexes were formed for the forms of Gas1.10 The medium collected from each of the subsequent construction of lentivirus; one virus with the cultures was transferred as conditioned medium to sequence for the truncated, secretable gas1 and the other independent cultures of proliferating C6 cells. For this virus expressing secretable gas1 together with the V5 assay, 300 000 cells were seeded in 35 mm plates and epitope as a fusion protein, and finally the virus with the grown until a confluence of 80%. Forty-eight hours after repressor element. As a control, a lacZ-expressing treatment, cells were counted using the Trypan blue lentivirus was also generated (the lentivirus with lacZ technique. As controls for the expression of the gas1 gene and the repressor element were constructed following the manufacturer’s protocol). The viral titers were obtained and protein, SH-SY5Y human neuroblastoma cells, that À1 express Gas1 24 h after serum withdrawal, were used as based on the sensitivity of the cells to Zeocin (250 mgml ) previously described.12 (Invitrogen) for the lentivirus coding for both truncated and truncated-V5 forms of Gas1, and with Blasticidin (5 mgmlÀ1) (Invitrogen) for the virus with the repressor element, according to the following equation: CFU RT–PCR To determine the expression of the different gas1 per ml ¼ p Â 20 Â 10X; where CFU is colony forming fragments, total RNA was isolated from transfected, units, p is the number of colonies per dilution of viral stock, infected cells and tumors, transcribed into DNA using 20 is a correction factor to express CFP per ml (1000 mlper reverse transcriptase M-MLV (Invitrogen). The PCR inoculated volume) and 10X is the lowest dilution used. reaction was performed under the same conditions For in vitro experiments, C6 cells were infected with previously described in the construction of plasmids and lentivirus with tgas1 and tgas-V5 (multiplicity of infection (MOI) 1–10) in the presence of Polybrene (Sigma-Aldrich), using as primers the oligonucleotides originally used À1 to amplify the fragments from the pGfa2-Gas1 plasmid. 6 mgml for 24 h, medium was removed, cells were washed As controls non-treated, Lipofectamine 2000 treated three times with phosphate-buffered saline and incubated without DNA and non-infected cells were used. Expres- in complete medium. Forty-eight hours after infection, cell sion of b-actin and lacZ was determined as previously viability was evaluated with the Trypan blue technique, described.10,12 then culture medium was collected and applied as conditioned medium to proliferating cultures of C6 cells, as previously described. Western blot analysis The isolation of proteins from transfected, infected cells, Stably infected C6 cells non-treated cells and tumors was performed using a C6 cells were infected with the lentivirus containing the commercial kit (Tripure) (Roche). Protein quantification repressor element of expression (MOI 10) and cells were À1 was performed using the bicinchoninic acid method selected with Blasticidin (Invitrogen) (5 mgml ). Subse- (Pierce, Rockford, IL). An anti-Gas1 polyclonal primary quently, these cells were infected with the lentivirus antibody (Proscience, Poway, CA) was used at a dilution expressing the truncated, secretable, form of Gas1 (MOI À1 of 1:200 and incubated for 12 h at 4 1C, as previously 10) and selected with Zeocin (Invitrogen) (250 mgml ); described.10,12 The secondary antibody (Jackson Immu- the selection with Blasticidin was also maintained. noResearch, West Grove, PA) (goat anti-rabbit) was used at a dilution of 1:4000 and incubated for 1 h at room Southern blot analysis temperature. As a positive control, we used an anti-b- Genomic DNA was obtained from both stably infected actin antibody,17 as previously described.9,10 Proteins C6 cells and from control non-infected cells, digested and were revealed using a Western Lightning Plus-ECL Kit ran in a 0.6% agarose gels. DNA was transferred onto (PerkinElmer, Waltham, MA). Images were digitally nylon membranes (Amersham; Piscataway, NJ) and captured with a BioDoc-It Imaging System (UVP, hybridized with a dioxigenin-labeled riboprobe trans- Upland, CA). cribed with T7 from a PCR template generated from the

Cancer Gene Therapy Soluble Gas1 inhibits glioma growth ALo´pez-Ornelas et al 90 tGas1 lentiviral vector (Epicentre, AmpliScribe T7, T3 apoptosis was determined using an in situ cell death and SP6 High Yield Transcription Kits, Madison, WI), detection kit (TUNEL), based on the TdT-mediated for 18 h at 60 1C. Membranes were incubated with an dUTP nick ending labeling, following the manufacturer’s antibody against dioxigenin (Roche) 1:5000, and signals instructions (Roche). Twenty fields from three indepen- revealed using a Western Lightning Plus-ECL Kit dent experiments were counted. (PerkinElmer). Images were digitally captured with a BioDoc-It Imaging System (UVP). In vivo transduction with lentivirus-expressing secretable Gas1 Immunocytochemistry Experiments were performed according to current 6 In all, 1 Â 10 stably infected C6 cells producing tGas1 Mexican legislation NOM-062-ZOO-1999 (SAGARPA) were grown until they reached 80–90% confluence in and in agreement with the Guide for the Care and Use of 60 mm plates and were subsequently divided into three Laboratory Animals of the National Institutes of Health groups; one was treated with Tetracycline (Invitrogen) (NIH) and internal (Cinvestav) guidelines. In all, 1 Â 106 1 (2 mgmlÀ ), the second group did not receive Tetracycline C6 cells were subcutaneously inoculated into the dorsal and the last group consisted of control non-transduced C6 flank of 6–8-week-old nude (nu/nu) mice, as previously cells. After 48 h in culture, cells were fixed with 4% described.8,9 When the tumors reached 4–5 mm (large p-formaldehyde prepared in phosphate-buffered saline axis), animals were divided into three groups; one received (pH ¼ 7.4) and incubated with a polyclonal antibody three doses of 5 Â 105 viral particles (v.p.) expressing 10 against Gas1 (Proscience) (1:200). The secondary anti- tgas1, another received the lacZ-expressing virus (5 Â 105 body was a biotinylated anti-rabbit IgG (Vector Labora- v.p.) and the last, control group, received 100 mlof tories, Burlingame, CA) (1:300). Proteins were revealed medium. Each dose was injected intratumorally every with fluorescein streptavidin (Vector Laboratories) third day, the tumor diameter was measured every third (1:200) and cells were counterstained with 4,6-diamino- day for 10 days, and the volume was calculated according 2 2-phenyldole (Vector Laboratories) to expose the nuclei. to the following equation: V ¼ (Dmax)(Dmin) /(2); where An Olympus BX-51 microscope and the Image-Pro Plus Dmax and Dmin represent the large and small axis, program Version 4.5 (MediaCybernetics, Bethesda, MD) respectively. After 10 days of starting the treatment, mice were used. SH-SY5Y were cultured and assayed as were euthanized and tumors were extracted. Tumor RNA 12 previously described. and protein were isolated for RT–PCR and western blot assays, respectively, as previously described. For the Immunoprecipitation second group of experiments, 5 Â 105 C6 cells were mixed Cells were seeded as previously described for immunocy- with 5 Â 105 C6 cells stably expressing the tgas1 lentivirus tochemistry, culture media were collected and incubated together with the repressor and were injected into the with the antibody against Gas1 (Proscience)9,12 and then flanks of nu/nu mice, as previously described. When immunoprecipitated with Tachizorb (Roche) following tumors reached 4–5 mm large axis, mice were divided into the manufacturer’s instructions; as a control, medium was two groups. One group received three injections of incubated with an unrelated polyclonal antibody, in Tetracycline (2 mg kgÀ1 in 100 ml of saline solution) every this case, against glial fibrillary acidic protein (Dako, third day, and the other group received the same volume Carpinteria, CA). After immunoprecipitation, western of vehicle. Tumor volume was determined as previously blot assays to detect Gas1 were performed as previously mentioned, and 10 days starting the treatment mice were described, including the control medium. The medium euthanized. from tgas1-infected cells with or without Tetracycline and from non-infected control cells subjected to the immunoprecipitation protocol were applied to cultures of C6 cells, and after 48 h, cellular viability was evaluated by the Results Trypan blue assay. Construction and expression of gas1 transgenes We obtained four different cDNA fragments expressing BrdU incorporation full-length and truncated gas1 sequences, amplifying by Cells were plated as previously described for the PCR the fragments from the pGfa2-Gas1 plasmid.7 immunocytochemistry assay, and incubated in the pre- Figure 1 shows the schematic representation of the Gas1 sence of 50 mM of BrdU (Roche) for 6 h at 37 1C. Twenty- proteins: two proteins contain the full-length Gas1 amino- four hours after transduction, cells were analyzed for acid sequence, Gas1, and the other is a fusion protein of BrdU incorporation. A BrdU staining kit (Invitrogen, Gas1 together with the V5 epitope, Gas1-V5. The other Camarillo, CA) was used according to the manufacturer’s two proteins contain the truncated, secretable, form of instructions. A total of 200–300 cells per coverslip were Gas1, one as the truncated Gas1 protein, tGas1, and the examined to determine the percentage of cells that other one, tGas1-V5, is the truncated Gas1 protein with incorporated BrdU. the V5 epitope. Each DNA fragment was directionally subcloned into Apoptosis assay the pLenti/TO/V5-DEST expression vector and trans- To determine apoptotic death, 48 h after infection with fected into C6 glioma cells. Figure 2a shows the expres- the tgas1 lentivirus, the number of cells undergoing sion of each of the constructs as demonstrated by RT–PCR.

Cancer Gene Therapy Soluble Gas1 inhibits glioma growth ALo´pez-Ornelas et al 91

Figure 2 Expression and effect of the different forms of Gas1 on cell growth. (a) C6 cells were transfected with the four different vectors, and the expression of the transgenes determined by RT–PCR, the different forms of gas1 mRNA were only observed in transfected cells, no expression was detected in control non-treated cells, or in cells subjected only to the transfection protocol (Lipofectamine), in the absence of gas1 vectors; b-actin expression was used as a positive control. (b) Protein expression was determined by western blot analysis, all transfected cells expressed Gas1 proteins of the predicted sizes; as a positive control we used serum-starved SH-SY5Y cells, and as negative controls non- treated C6 cells and C6 cells subjected only to the transfection protocol. (c) C6 cells were transfected with vectors expressing the different forms of Gas1, and the number of viable cells was determined 48 h after transfection, by Trypan blue exclusion for each treatment, controls were non-transfected cells and C6 cells subjected only to the transfection protocol, n ¼ 5. (d) To determine the effect of the conditioned medium of C6 cells transfected with the different forms of Gas1, various volumes of medium from each culture were taken and used to replace the medium of control non-transfected cells (from 25% of the volume to complete substitution of medium, 100%), n ¼ 3. (e) To determine the effect of the conditioned medium of C6 cells transfected with the different vectors, all the medium of control cells was substituted (100%) and BrdU incorporation determined in C6 cells, n ¼ 5. Data are means±s.e.m. Analysis of variance (ANOVA) followed by Tukey’s test. *Po0.05, **Po0.01, ***Po0.001, with respect to control.

As can be observed, the sizes of the products obtained cells. C6 cells were seeded and transfected with the from cells transfected with the complete gas1 sequence different constructs. Figure 2c shows that all forms of (gas1 and gas1-V5) are larger than the truncated, Gas1 reduced the number of viable cells with respect to secretable forms (tgas1 and tgas1-V5); no expression of control. To determine whether the capacity reducing the the transgene was observed in control non-transfected number of viable cells could be transferred to independent cells, or cells subjected to the transfection protocol cell cultures, we took conditioned media from cultures of (Lipofectamine) without cDNA (Figure 2a, lower panels). transfected cells and substituted the media of non- To corroborate the appropriate expression of the transfected cells with the conditioned media. A dose– transgenic proteins, we performed western blot analysis response curve was constructed substituting different of transfected cells. Figure 2b shows the expression of amounts of the medium of the independent cultures Gas1 in transfected cells, here we can also observe that the (from 25% of the volume to complete substitution of the different forms of Gas1 are of the expected sizes, that the medium, 100%) with conditioned medium. Figure 2d Gas1 forms are larger than the tGas1 constructs shows that there are no significant differences in the (B3 kDa), and the V5-fusion proteins are larger that number of viable cells, when comparing the medium from the Gas1 and tGas1 forms without the V5 epitope control cells, cells subjected to Lipofectamine, or cells (B1.4 kDa). To determine the molecular size of the expressing full-length Gas1, whereas media from cells transgenic proteins, we used commercial size markers, but expressing the truncated forms of Gas1 significantly also compared them with the endogenous Gas1 protein reduce the number of viable cells. The effects were seen obtained from SH-SY5Y cells grown in the absence of from the lowest level of substitution of medium (25%), serum.12 The next step was to determine the capacity of and in the tGas1, the number of viable C6 cells decreased the transgenic human Gas1 reducing the number of viable when the volume of conditioned media increased more

Cancer Gene Therapy Soluble Gas1 inhibits glioma growth ALo´pez-Ornelas et al 92 than with any of the other treatments (Figure 2d). and of the conditioned media obtained from infected cells, Interestingly, the form of truncated Gas1 without the as determined by the incorporation of BrdU (Figure 3d). V5 epitope is more effective that the fusion protein with Although both forms of truncated Gas1 inhibit cell the V5 epitope (Figure 2d). This result indicates that the proliferation, tGas1 was more capable than tGas1-V5 cell growth inhibitory effect is caused by soluble Gas1. We inhibiting proliferation of C6 cells. We constructed also determined the capacity of the conditioned medium response curves using different MOI for each virus and of C6-transfected cells to arrest the cell cycle, of control assays, from 1 to 10 MOI, obtaining the more pronounced non-transfected C6 cells, and observed that the media effects with a MOI of 10 (Figures 3c and d). Based on the from all transfected cultures had an effect, but that the previous information, we decided to use the tGas1 virus at truncated forms of Gas1 were more effective inducing this a MOI of 10 for the following experiments. outcome than the full-length forms, and that tGas1 We have previously shown that Gas1 induces apoptosis achieves a better result when compared with tGas1-V5 and cell arrest of glioma cells,9,10 and decided to (Figure 2e), similarly to what had been observed determine whether the tGas1 form acts in the same regarding their ability to decrease the number of viable manner. We observed that tGas1 induces a large increase cells (Figure 2d). of apoptotic cells, as determined by the TUNEL assay, when compared with control cells, and that the medium Expression and effect of tgas1-expressing lentivirus of infected cells can also induce apoptosis of a large After determining the effect of both the full-length and number of cells in independent cultures (Figure 4). tGas1 truncated forms of Gas1 reducing the number of viable from infected cells. These results are in agreement with transfected C6 cells and also from the conditioned media our previous data,7–10 and further support the concept of obtained from those cells, we concluded that the the apoptotic effect induced by the secretable form of truncated forms of Gas1 were effective reducing both Gas1 (tGas1) that acts on the cells that express the the viability and the proliferation of cells of independent transgene, as well as in a soluble form in the medium of cultures, thus suggesting that the soluble forms of Gas1 transfected and infected cells expressing this truncated were responsible for this effect. To increase the efficiency Gas1 protein. of the transfer of the transgenes, as well as to obtain a To confirm the capacity of the tgas1 lentivirus- system to regulate their expression, we constructed expressing tGas1 in a regulated manner, we generated a lentiviral vectors, in which the expression of the trans- stable cell line infected with the tGas1 lentivirus, and with genes can be induced by the administration of Tetracy- the virus containing the repressor element of transcrip- cline. It is worthwhile to remark that the lentivirus, in the tion. Cells were selected in the presence of Blasticidin and absence of the repressor element of transcription, will Zeocin, and maintained in the presence of the antibiotics. express the transgene under the control of the CMV We determined the expression of tgas1 in infected cells promoter, and that it is when co-expressed with the both in the presence and absence of Tetracycline repressor that the transgene will only be transcribed in (2 mgmlÀ1 of medium), and observed that tGas1 was only the presence of Tetracycline. In order to determine the expressed in infected cells in the presence of Tetracycline, number of vector copy number per cell, we performed as determined by RT–PCR (Figure 5a). We also verified Southern blot analysis of stably infected C6 cells and the expression of the tGas1 protein exclusively in infected found four copies of the tgas1 vector (Supplementary cells receiving Tetracycline, as ascertained by both Figure 1). western blot analysis and immunocytochemistry, as C6 cells were infected with two lentivirus-expressing positive controls for expression and protein size, we used tGas1 and tGas1-V5, respectively. The expression of the the SH-SY5Ycells, which express Gas1 after serum transgenes was confirmed by RT–PCR (Figure 3a) and withdrawal (Figures 5b and c). that of the transgenic protein by western blot analysis, Definitive demonstration of the presence of soluble where we can observe that the transgenic proteins are Gas1, as well as its capacity inhibiting cell growth and smaller than the endogenous full-length human Gas1 inducing cell death is a necessary prerequisite to protein obtained from serum-starved SH-SY5Y cells substantiate a gene therapy approach based on the (Figure 3b). We also determined the effect of the concept of a Gas1 protein that can induce both autocrine expression of the different forms of truncated Gas1 on and paracrine effects. To achieve these objectives, tGas1 the number of viable cells, in cells infected with the viral was immunoprecipitated from the medium of stably vectors, and that induced using media from cultures of infected cells, in the presence and in the absence of infected cells. We can observe that the number of viable Tetracycline. Figure 6a shows that tGas1 was only cells is significantly lower in cells infected with both tGas1 immunoprecipitated from the media of cultures of and tGas1-V5 lentivirus, compared with control, and that infected cells in the presence of Tetracycline, and not media from infected cultures also reduces the number of from medium of control non-infected cells, or from viable cells with respect to control, that is medium from infected cells grown in the absence of the drug. As positive non-infected cells (Figure 3c). We can also perceive that controls we used endogenous Gas1 obtained from serum- the effect of the tGas1 form is greater than that of the starved SH-SY5Y cells, as previously described, tGas1 tGas1 form expressing the V5 epitope (Figure 3c). We isolated from transfected cells, and as negative control, we assayed the capacity of the soluble forms of Gas1 to showed that the use of a non-related antibody does not inhibit the proliferation of cells, infected with the virus, precipitate tGas1 (Figure 6a). Furthermore, we showed

Cancer Gene Therapy Soluble Gas1 inhibits glioma growth ALo´pez-Ornelas et al 93

Figure 3 Infection of C6 cells with lentivirus expressing the secretable forms of Gas1 (tGas1) reduces cell proliferation. (a) Expression of tgas1 and tgas1-V5 as determined by RT–PCR from C6 cells 48 h post-infection with tgas1 and tgas1-V5 lentivirus, respectively, non-infected C6 cells were used as control, b-actin is a positive control. (b) Expression of tGas1 and tGas1-V5 determined by western blot analysis of proteins extracted from infected C6 cells 48 h after infection, SH-SY5Y is from protein obtained from SH-SY5Y serum-starved cells, b-actin is a positive control. (c) Viability of C6 cells, as determined by Trypan blue exclusion, after infection with tgas1 and tgas1-V5 virus, and after taking conditioned medium from infected cells and substituting it, in control non-infected cells, using different MOI, controls are non-infected cells, or cells incubated in the presence of Polybrene, and media obtained from those cells, n ¼ 5(d). Cell proliferation determined by BrdU incorporation of C6 cells after infection with tgas1 and tgas1-V5 virus, or after taking conditioned medium from infected cells and substituting it, in control non- infected cells, using different MOI, controls are non-infected C6 cells, or medium from non-infected cells, n ¼ 3. Data are means±s.e.m. Analysis of variance (ANOVA) followed by Tukey’s test. *Po0.05, **Po0.01 ***Po0.001, with respect to control.

that medium from which tGas1 had been removed by mice, and when the tumors reached 4–5 mm of the longer immunoprecipitation had no effect on cell growth, with axis, we administered intratumorally, vehicle (control), respect to cells cultured with medium from control non- lacZ-expressing virus, or tgas1 virus. Treatments were infected cells, whereas medium from Tetracycline-treated applied every other day for a total of three injections, and tGas1-infected cells, induced a significant decrease in the we determined the growth of the tumor for 10 days after number of viable C6 when compared with both control initiating the treatment. Tumors receiving the tgas1 and medium in which tGas1 had been removed lentivirus showed a 74% reduction in growth, compared (Figure 6b). Taken together, our data demonstrate that with non-treated controls, and were significantly different tgas1-infected cells produce a truncated form of Gas1, from both control and lacZ-infected tumors (Figure 7a). under the control of an inducible promoter, which is Moreover, after treatment, mice were euthanized and released to the medium, in a soluble form that can inhibit expression of transgenes assessed in dissected tumors. the growth of glioma cells. Figure 7b shows the expression of the tgas1 and lacZ transgenes exclusively in tumors receiving the correspond- In vivo effects of tGas1 ing virus, and in Figure 7c, we observed that tGas1 was To ascertain the therapeutic potential of tGas1, we only expressed in tgas1-infected tumors. These results examined the effect of the gene transfer of tgas1 demonstrate that the lentivirus-mediated transfer of the determining the growth of C6 cells inoculated in athymic tgas1 transgene is very effective in inhibiting the growth of nude mice. C6 cells were injected into the flanks of nu/nu gliomas in an in vivo mouse model.

Cancer Gene Therapy Soluble Gas1 inhibits glioma growth ALo´pez-Ornelas et al 94

Figure 4 Secreted tGas1 induces apoptosis. (a) C6 cells were infected with the tgas1-expressing lentivirus and apoptosis determined using the TUNEL assay. Top panel shows apoptotic cells in a culture of C6 control cells; second panel shows apoptotic C6 cells 48 h after infection with the tgas1-expressing lentivirus; the third panel is a culture of C6 cells in which the medium was replaced with medium from control C6 cells and the lower panel shows C6 control cells, in which medium had been replaced with conditioned medium from infected C6 cells, photographs are from representative experiments, 4,6-diamino-2-phenyldole (DAPI) was used to reveal nuclei, calibration bar is 1 mm. (b) Table shows the percentage of TUNEL-positive cells, and data are obtained from three independent experiments for each of the previously described conditions; data were obtained from 20 fields of three independent experiments, and the percentage of apoptotic cells is presented.

To further determine the capacity of tGas1 as a can deliver therapeutic genes into tumors, that very rarely potential therapeutic molecule, we inoculated a mixture metastize outside of the CNS. Viral vectors have been the of C6 cells with C6 cells stably infected with the tgas1 system most widely used to deliver transgenes in both lentivirus, into the flanks of nu/nu mice. When tumors preclinical and clinical trials, and though they appear safe, reached the sized previously described, we treated one clinical results have not yet generated a viable alternative group of mice with Tetracycline (2 mg kgÀ1) to induce the to current glioma treatments.18,19 The most common expression of the transgene, whereas the other group therapeutic transgene used for glioma gene therapy, both received the same volume of vehicle. Figure 7d illustrates experimentally and in clinical trials, has been the herpes that there is a significant reduction (60%) in the volume thymidine kinase (HSV-tk) gene.20–22 The product of the of the tumors receiving Tetracycline, compared with HSV-tk gene transforms the prodrug ganciclovir to a saline controls, thus showing that the Tetracycline- precursor that arrests DNA replication, inducing cell induced production of secretable Gas1 is effective death. Interestingly, the effects of this type of vectors are diminishing glioma cell proliferation (Figure 7d). amplified by the bystander effect, caused by the passage of the toxic metabolites to cells connected by gap junctions.23–25 This effect increases the therapeutic capacity of this approach; however, a limitation to this system is that the Discussion level of intercellular communication varies in different tumors, and may even be blocked in the later stages of Gliomas, particularly glioblastoma multiforme, have a carcinogenesis.24,26,27 We consider, as we will discuss in more very poor prognosis that has not improved for decades. detail presently, that the procedure we put forward in this Although current treatments include sophisticated surgi- paper shows improvements with respect to the transfer of cal approaches, radiotherapy and chemotherapy,1 the HSV-tk, because it affects neighboring cells that need not to ability of tumor cells to infiltrate into the brain be joined by gap junctions, thus increasing its range of parenchyma as well as their enhanced resistance to drugs therapeutic activity. and radiation, deter the efficacy of existing treatments. There are several factors that need to be considered for Gene therapy offers a promising approach for the treat- viral vectors to deliver the therapeutic transgene to ment of brain tumors, particularly of gliomas, because it sufficient tumor cells to induce a significant beneficial

Cancer Gene Therapy Soluble Gas1 inhibits glioma growth ALo´pez-Ornelas et al 95

Figure 5 Expression of tGas1 in stably infected cells is induced by Tetracycline. C6 cells were infected with the tgas1-expressing lentivirus, with the lentivirus expressing the repressor element, and selected in the presence of Blasticidin and Zeocin. (a) The expression of tgas1 mRNA as determined by RT–PCR, in the absence (tgas1), or in the presence of Tetracycline (tgas1/Tetra); control þ is the PCR product from the tgas1 vector, b-actin is a positive control. (b) Expression of tGas1 in stably infected cells, determined by western blot, lanes represent C6 cells without or with Tetracycline as described above; SH-SY5Y is wild-type Gas1 from SH-SY5Y serum-starved cells, b-actin is a positive control. (c) Expression of tGas1 as determined by immunocytochemistry; C6 cells are control, non-infected cells; C6 tGas1 are stably infected cells in the absence or in the presence of Tetracycline; lower two panels are SH-SY5Y cells incubated with complete medium or in the absence of serum (W/S). Cells were stained with an antibody against Gas1, counterstained with 4,6-diamino-2-phenyldole (DAPI) to reveal nuclei, and the images merged to show co-expression, calibration bar is 1 mm.

effect in patients. Physical barriers and immune responses tumor, and experimental approaches to modulate the inhibit the distribution of viral vectors. On the other matrix, and thus allowing a better distribution of the virus hand, the short half-life of some virus, like retrovirus, also have been proposed.31,32 affect their therapeutic capacity.28,29 Virus diffuse pas- Another strategy to obtain a more complete and sively once they are administered into the tissue, and their homogeneous distribution of the therapeutic agent is to inadequate distribution and propagation within the tumor use soluble molecules that can be released from cells diminish their therapeutic capacity.30 Viral vectors are expressing the transgene,33 and in this manner more freely very large compared with chemotherapeutic agents (over diffuse within the tumor, thus enhancing the beneficial 100-fold), causing their distribution within the tumor to effect of the treatment. Gas1 is a molecule that can induce be very limited, thus the penetration of these vectors from cell arrest and apoptosis of different tumor cells,34–36 the site of injection or from the vasculature is severely including glioma cell lines and human primary gliomas.7–10 restricted.4,19 The extracellular matrix has an important Bioinformatics analysis uncovered significant structural role limiting the distribution of viral vectors within the homology between Gas1 and GFRas.11 Based on this

Cancer Gene Therapy Soluble Gas1 inhibits glioma growth ALo´pez-Ornelas et al 96

Figure 6 tGas1 is present in the conditioned medium of stably infected cells and inhibits cell growth. (a) Immunoprecipitation of tGas1 from culture media. Medium was obtained from C6 cells; infected with tgas1 in the presence (tGas1/Tetra) or absence (tGas1) of Tetracycline, or control non-treated C6 cells; lanes represent, for each set of experiments: (1) only medium; (2) immunoprecipitated in the absence of Gas1 antibody; and (3) immunoprecipitated in the presence of the Gas1 antibody. Negative control (À) is immunoprecipitation in the presence of an unrelated polyclonal antibody (against glial fibrillary acidic protein); and positive control ( þ ) is western blot from an extract of C6 cells stably infected with tgas1 in the presence of Tetracycline, and SH-SY5Y is from an extract of serum-starved cells. (b) The culture medium of control C6 cells was changed with media from control non-infected, from medium of C6 stably infected cells in the presence of Tetracycline, from which tGas1 had been removed by immunoprecipitation, IPtGas1, and from infected cells from which tGas1 was not immunoprecipitated, tGas1. Analysis of variance (ANOVA) followed by Tukey’s test. *Po0.05, **Po0.01, n ¼ 3.

information, we showed that the effect of Gas1 arresting undergoing the same protocol, but that did not receive cell cycle and inducing apoptosis is caused by its ability Tetracycline. This experiment shows that tGas1 produced inhibiting the GDNF-mediated intracellular signaling in infected cells inhibits cell growth and induces apoptosis cascade, a mechanism consistent with the effect of GDNF of neighbor cells. inducing glioma proliferation.10,12,37 It has been reported There are reports indicating that soluble apoptogenic that the carboxy terminal domain of Gas1 is not necessary molecules, like TRAIL, applied into gliomas are effective for this molecule to induce growth arrest, and that a reducing tumor growth.40–42 Thus, a potentially relevant soluble form is active.15 These data allowed us to form of application is the concept of injecting tGas1- hypothesize that a secreted, soluble, form of Gas1 would producing cells into tumors. A possible source of the cells arrest cell cycle and induce apoptosis in cells expressing are cells from the same tumor, other somatic cells from the the transgene, but also in neighbor cells not expressing it. same patient, or from a different source, so the secreted To obtain a secretable form of Gas1, we generated protein could inhibit the growth of the tumor, and in the truncated forms of Gas1, lacking the GPI-link consensus case of applied tumor cells would also kill the genetically sequence, so these forms of Gas1 will be secreted from the modified cells and increase the safety of the system. cells.14,38,39 The autocrine and paracrine effects of Gas1 The capacity of regulating the production and release of will significantly increase its therapeutic capacity, by the therapeutic molecule will also facilitate control of the allowing the diffusion and distribution of the therapeutic treatment, both temporally, and if cells can be marked and molecule away from producer cells. To reach this traced in the patient, also by its localization in the brain. As objective we generated lentivirus expressing a truncated a first attempt to explore these potential therapeutic form of Gas1 (Arg 315), and determined that infected strategies, we injected a mixture of C6 cells, and stably cells release a soluble form of Gas1 that can induce cell infected C6 cells into nude mice, and determined the effect arrest and apoptosis when transferred to independent of the administration of Tetracycline in the growth of the cultures of C6 cells. Furthermore, the intratumoral tumors. As seen in Figure 7d, there is a significant reduction injection of the tgas1 lentivirus inhibited the growth of in the growth of the tumors that received Tetracycline C6 cells inoculated in nu/nu mice. Finally, we also compared with control mice. This experiment shows it is observed decreased tumor growth, when a mixture of feasible to regulate the expression of tGas1 from infected non-infected C6 cells and cells stably infected with the cells in an in vivo model, and inhibit tumor growth. tgas1 virus were inoculated together in mice and were In the present work, we have demonstrated that a administered with Tetracycline, when compared with mice truncated form of Gas1, produced in transfected or

Cancer Gene Therapy Soluble Gas1 inhibits glioma growth ALo´pez-Ornelas et al 97

Figure 7 In vivo anti-tumoral action of tGas1. (a) Growth curves of C6 cells subcutaneously implanted in nude mice, and treated with the indicated lentivirus, LacZ, expressing lacZ, tGas1-expressing tgas1, and control receiving only medium. Data are means±s.e.m. Arrows show the application of the treatments. Analysis of variance (ANOVA) followed by Tukey was used. *Po0.05, ***Po0.001 compared with both control and LacZ, n ¼ 3 mice per group. (b) Expression of the transgenes as assessed by RT–PCR, RNA was obtained from control tumors, receiving medium, lacZ or tgas1 lentivirus; Control þ is DNA from the respective expression vectors, and Control À, is the PCR reaction without adding cDNA, b-actin is a positive control. (c) Expression of tGas1 from tumors receiving the different treatments, SH-SY5Y is from proteins obtained in serum-starved SH-SY5Y cells, b-actin is a positive control. (d) Tetracycline regulates the expression tGas in vivo. Tumors consisting of a mixture of 5 Â 105 C6 cells and 5 Â 105 stably infected C6 cells were subcutaneously implanted in nude mice, one group received the local injection of Tetracycline (2 mg kgÀ1) and the control group only received vehicle (saline solution). Results are means±s.e.m. Arrows show the different applications of Tetracycline. ANOVA followed by Tukey’s test. **Po0.01, ***Po0.001 compared with medium group, n ¼ 4–5 mice per group. infected cells, can be secreted into the medium, and induce blot assay, Dr G Domıńguez-Monzoń and N Zarco for cell arrest and apoptosis of other cells not expressing the useful comments and discussion of the data and R therapeutic transgene. Furthermore, the application of Sańchez for technical support. the lentiviral vector, or cells infected with this virus inhibit the growth of tumors inoculated in mice. We consider that the results shown here support the potential use of Gas1 as an adjuvant in the treatment of gliomas, and when we References consider the soluble nature of tGas1, expand its therapeutic capability. 1 Benı´tez JA, Domıńguez-Monzoń G, Segovia J. Conventional and gene therapy strategies for the treatment of brain tumors. Curr Med Chem 2008; 15: 729–742. Conflict of interest 2 Wang Y, Yuan F. Delivery of viral vectors to tumor cells: extracellular transport, systemic distribution, and strategies The authors declare no conflict of interest. for improvement. Ann Biomed Eng 2006; 34: 114–127. 3 Tayi VS, Bowen BD, Piret JM. Mathematical model of the rate-limiting steps for retrovirus-mediated gene Acknowledgements transfer into mammalian cells. Biotechnol Bioeng 2009; 105: 195–209. This work was partially supported by Conacyt grant 4 Mok W, Boucher Y, Jain RK. Matrix metalloproteinases-1 54756 (JS). We thank Dr L Gutie´rrez-Escolano and E and -8 improve the distribution and efficacy of an oncolytic Lo´pez for their invaluable assistance with the Southern virus. Cancer Res 2007; 67: 10664–10668.

Cancer Gene Therapy Soluble Gas1 inhibits glioma growth ALo´pez-Ornelas et al 98 5 Mok W, Stylianopoulos T, Boucher Y, Jain RK. Mathema- 23 Asklund T, Appelskog IB, Ammerpohl O, Langmoen IA, tical modeling of herpes simplex virus distribution in solid Dilber MS, Aints A et al. Gap junction-mediated bystander tumors: implications for cancer gene therapy. Clin Cancer effect in primary cultures of human malignant gliomas with Res 2009; 15: 2352–2360. recombinant expression of the HSVtk gene. Exp Cell Res 6 Stebel M, Vatta P, Ruaro ME, Del Sal G, Parton RG, 2003; 284: 185–195. Schneider C. The growth suppressing gas1 product is a GPI- 24 Dilber MS, Abedi MR, Christensson B, Bjo¨rkstrand B, linked protein. FEBS Lett 2000; 481: 152–158. Kidder GM, Naus CC et al. Gap junctions promote the 7 Zamorano A, Lamas M, Vergara P, Naranjo JR, Segovia J. bystander effect of herpes simplex virus thymidine kinase in Transcriptionally mediated gene targeting of gas1 to glioma vivo. Cancer Res 1997; 15: 1523–1528. cells elicits growth arrest and apoptosis. J Neurosci Res 2003; 25 Touraine RL, Ishii-Morita H, Ramsey WJ, Blaese RM. The 71: 256–263. bystander effect in the HSVtk/ganciclovir system and its 8 Zamorano A, Mellstro¨m B, Vergara P, Naranjo JR, Segovia relationship to gap junctional communication. Gene Ther J. Glial-specific retrovirally mediated gas1 gene expression 1998; 5: 1705–1711. induces glioma cell apoptosis and inhibits tumor growth in 26 Ishii-Morita H, Agbaria R, Mullen CA, Hirano H, vivo. Neurobiol Dis 2004; 5: 483–491. Koeplin DA, Ram Z et al. Mechanism of ‘bystander 9 Benı´tez JA, Arregui L, Vergara P, Segovia J. Targeted- effect’ killing in the herpes simplex thymidine kinase simultaneous expression of Gas1 and p53 using a bicistronic gene therapy model of cancer treatment. Gene Ther 1997; adenoviral vector in gliomas. Cancer Gene Ther 2007; 14: 4: 244–251. 836–846. 27 Yamasaki H. Aberrant expression and function of gap 10 Domıńguez-Monzoń G, Benı´tez JA, Vergara P, Lorenzana junctions during carcinogenesis. Environ Health Perspect R, Segovia J. Gas1 inhibits cell proliferation and induces 1991; 93: 191–197. apoptosis of human primary gliomas in the absence of Shh. 28 Kyritsis AP, Sioka C, Rao JS. Viruses, gene therapy and Int J Dev Neurosci 2009; 27: 305–313. stem cells for the treatment of human glioma. Cancer Gene 11 Schueler-Furman O, Glick E, Segovia J, Linial M. Is GAS1 a Ther 2009; 16: 741–752. co-receptor for the GDNF family of ligands? Trends 29 Sack BK, Herzog RW. Evading the immune response upon Pharmacol Sci 2006; 27: 72–77. in vivo gene therapy with viral vectors. Curr Opin Mol Ther 12 Lo´pez-Ramı´rez MA, Domıńguez-Monzoń G, Vergara P, 2009; 11: 493–503. Segovia J. Gas1 reduces Ret tyrosine 1062 phosphorylation 30 Kim SM, Lim JY, Park SI, Jeong CH, Oh JH, Jeong M et al. and alters GDNF-mediated intracellular signaling. Int J Dev Gene therapy using TRAIL-secreting human umbilical cord Neurosci 2008; 26: 497–503. blood-derived mesenchymal stem cells against intracranial 13 Paratcha G, Ledda F, Baars L, Coulpier M, Besset V, glioma. Cancer Res 2008; 68: 9614–9623. Anders J et al. Released GFRalpha1 potentiates downstream 31 McKee TD, Grandi P, Mok W, Alexandrakis G, Insin N, signaling, neuronal survival, and differentiation via a novel Zimmer JP et al. Degradation of fibrillar collagen in a mechanism of recruitment of c-Ret to lipid rafts. Neuron human melanoma xenograft improves the efficacy of an 2001; 29: 171–184. oncolytic herpes simplex virus vector. Cancer Res 2006; 66: 14 Virtanen H, Yang J, Bespalov MM, Hiltunen JO, Leppa¨nen 2509–2513. VM, Kalkkinen N et al. The first cysteine-rich domain of the 32 Yun CO. Overcoming the extracellular matrix barrier to receptor GFRalpha1 stabilizes the binding of GDNF. improve intratumoral spread and therapeutic potential Biochem J 2005; 387: 817–824. of oncolytic virotherapy. Curr Opin Mol Ther 2008; 10: 15 Ruaro ME, Stebel M, Vatta P, Marzinotto S, Schneider C. 356–361. Analysis of the domain requirement in Gas1 growth 33 Jeong M, Kwon YS, Park SH, Kim CY, Jeun SS, Song KW suppressing activity. FEBS Lett 2000; 481: 159–163. et al. Possible novel therapy for malignant gliomas with 16 Beaufort N, Leduc D, Rousselle JC, Magdolen V, Luther T, secretable trimeric TRAIL. PLoS One 2009; 4: e4545. Namane A et al. Proteolytic regulation of the urokinase 34 Zhao L, Pan Y, Gang Y, Wang H, Jin H, Tie J et al. receptor/CD87 on monocytic cells by neutrophil elastase and Identification of GAS1 as an epirubicin resistance-related cathepsin G. J Immunol 2004; 172: 540–549. gene in human gastric cancer cells with a partially 17 Garcia-Tovar CG, Perez A, Luna J, Mena R, Osorio B, randomized small interfering RNA library. Biol Chem Aleman V et al. Biochemical and histochemical analysis of 71 2009; 284: 26273–26285. kDa dystrophin isoform (Dp71f) in rat brain. Acta 35 Gobeil S, Zhu X, Doillon CJ, Green MR. A genome-wide Histochem 2001; 103: 209–224. shRNA screen identifies GAS1 as a novel 18 Asadi-Moghaddam K, Chiocca EA. Gene- and viral-based melanoma metastasis suppressor gene. Genes Dev 2008; 22: therapies for brain tumors. Neurotherapeutics 2009; 6: 547–557. 2932–2940. 19 Chiocca EA, Broaddus WC, Gillies GT, Visted T, Lamfers 36 Kokkinakis DM, Brickner AG, Kirkwood JM, Liu X, ML. Neurosurgical delivery of chemotherapeutics, targeted Goldwasser JE, Kastrama A et al. Mitotic arrest, apoptosis, toxins, genetic and viral therapies in neuro-oncology. and sensitization to chemotherapy of melanomas by J Neurooncol 2004; 69: 101–117. methionine deprivation stress. Mol Cancer Res 2006; 4: 20 Pulkkanen KJ, Yla-Herttuala S. Gene therapy for malignant 575–589. glioma: current clinical status. Mol Ther 2005; 4: 585–598. 37 Wiesenhofer B, Weis C, Humpel C. Glial cell line- 21 Hutterer M, Gunsilius E, Stockhammer G. Molecular derived neurotrophic factor (GDNF) is a proliferation therapies for malignant gliomas. Wien Med Wochenschr factor for rat C6 glioma cells: evidence from antisense 2006; 156: 351–363. experiments. Antisense Nucleic Acid Drug Dev 2000; 10: 22 Maä¨ttaÄM, Samaranayake H, Pikkarainen J, Wirth T, 311–321. Yla¨-Herttuala S. Adenovirus mediated herpes simplex 38 Yang J, Tiong J, Kennard M, Jefferies WA. Deletion of the virus-thymidine kinase/ganciclovir gene therapy for GPI pre-anchor sequence in human p97—a general ap- resectable malignant glioma. Curr Gene Ther 2009; 9: proach for generating the soluble form of GPI-linked 356–367. proteins. Protein Expr Purif 2004; 34: 28–48.

Cancer Gene Therapy Soluble Gas1 inhibits glioma growth ALo´pez-Ornelas et al 99 39 Orlean P, Menon AK. Thematic review series: lipid 41 Hingtgen S, Ren X, Terwilliger E, Classon M, Weissleder R, posttranslational modifications. GPI anchoring of protein Shah K. Targeting multiple pathways in gliomas with stem in yeast and mammalian cells, or: how we learned to stop cell and viral delivered S-TRAIL and Temozolomide. Mol worrying and love glycophospholipids. Lipid Res 2008; 48: Cancer Ther 2008; 7: 3575–3585. 993–1011. 42 Menon LG, Kelly K, Yang HW, Kim SK, Black PM, Carroll 40 Kim MH, Billiar TR, Seol DW. The secretable form of RS. Human bone marrow-derived mesenchymal stromal trimeric TRAIL, a potent inducer of apoptosis. Biochem cells expressing S-TRAIL as a cellular delivery vehicle for Biophys Res Commun 2004; 321: 930–935. human glioma therapy. Stem Cells 2009; 27: 2320–2330.

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

Cancer Gene Therapy