Glioblastoma Therapy with Cytotoxic Mesenchymal Stromal Cells Optimized by Bioluminescence Imaging of Tumor and Therapeutic Cell Response

Genetically modified adipose tissue derived mesenchymal stromal cells (hAMSCs) with tumor homing capacity have been proposed for localized therapy of chemo- and radiotherapy resistant glioblastomas. We demonstrate an effective procedure to optimize glioblastoma therapy based on the use of genetically modified hAMSCs and in vivo non invasive monitoring of tumor and therapeutic cells. Glioblastoma U87 cells expressing Photinus pyralis luciferase (Pluc) were implanted in combination with hAMSCs expressing a trifunctional Renilla reniformis luciferase-red fluorescent protein-thymidine kinase reporter in the brains of SCID mice that were subsequently treated with ganciclovir (GCV). The resulting optimized therapy was effective and monitoring of tumor cells by bioluminescence imaging (BLI) showed that after 49 days GCV treatment reduced significantly the hAMSC treated tumors; by a factor of 104 relative to controls. Using a Pluc reporter regulated by an endothelial specific promoter and in vivo BLI to image hAMSC differentiation we gained insight on the therapeutic mechanism. Implanted hAMSCs homed to tumor vessels, where they differentiated to endothelial cells. We propose that the tumor killing efficiency of genetically modified hAMSCs results from their association with the tumor vascular system and should be useful vehicles to deliver localized therapy to glioblastoma surgical borders following tumor resection.


Introduction
Glioblastoma multiforme (GBM), a brain tumor with high morbidity and mortality, accounts for the highest proportion of deaths from brain tumors [1]. The average survival of patients diagnosed with GBM is 14 months, and only 2% of the patients survive up to 3 years [2,3]. Current treatment protocols for GBM are ineffective due to its invasiveness, which precludes complete surgical removal and to persistence of radiation and chemotherapy resistant cells after therapy. A promising approach based on the use of cellular vehicles to deliver localized treatment targeting the primary tumor mass has been proposed [4]. Different types of stem cells have shown therapeutic potential against glioblastomas [5,6,7,8]. However, due to their tumor homing capacity and ease of genetic manipulation bone marrow mesenchymal stromal cells appear as particularly promising therapeutic vehicles [9]. Human mesenchymal stromal cells are abundant and easily extracted from adipose tissue and have properties similar to those of their bone marrow counterparts: ease of expansion from autologous tissue [10,11,12], demonstrated homing capacity towards tumors [13,14,15] and participation in the stroma of different types of tumors [16], ease of genetic manipulation [17], capacity to inhibit immune and inflammatory reactions [18] and genetic stability, not generating tumors when implanted during long periods of time [19]. Recently, a rat C6 glioblastoma tumor model was successfully treated with human adipose tissue mesenchymal stromal cells (hAMSCs) modified to express cytosine deaminase:uracil phosphoribosyltransferase [20].
Targeted cytotoxicity strategies, developed during the last 20 years for gene therapy, are based on the activation of a systemically delivered low toxicity prodrug to a cytotoxic agent in the vicinity of tumors, mediated by the action of a specific activating enzyme previously introduced in the therapy vehicle [21]. The delivery of cytotoxicity in the vicinity of a tumor has the advantage of avoiding systemic toxicity. The thymidine kinase encoded by herpes simplex virus (HSV-TK) was used in the first cell suicide gene therapy proof of principle and still is one of the most widely used systems in clinical and experimental applications [22,23]. This cytotoxic system has been applied to the treatment of gliomas with significant increases in survival [24]. HSV-TK is responsible for the conversion of deoxythimidine to deoxythyimidine monophosphate in infected cells. Due to its relaxed specificity, this enzyme can phosphorylate a variety of nucleotide analogs including ganciclovir (GCV), a 29-deoxiguanosine analog [25]. The incorporation of phosphorylated GCV in DNA results in DNA polymerase inhibition, chain termination and cell suicide [2,13,14,26]. Transfer of dying cell components through gap junctions and diffusion results in the killing of neighboring cells, with amplification of the therapeutic effect.
In the current work, we develop and demonstrate an efficient therapeutic strategy model against a Photynus pyralis luciferase (Pluc) and enhanced green protein (eGFP) expressing U87 glioblastoma model in SCID mice, using hAMSCs genetically modified to express a tri-functional reporter comprising Renilla reniformis luciferase (Rluc), red fluorescent protein (RFP) and a truncated version of HSV-TK (tTK). The latter enzyme has a 30 fold higher activity than wild type HSV-TK [26,27]. By taking advantage of bioluminescence imaging (BLI) to noninvasively monitor luciferase expressing tumor and therapeutic cells, we were able to fine tune and evaluate the therapeutic process in real time. In addition, the imaging platform allowed us to observe the behavior of therapeutic hAMSCs, offering insight into the therapeutic mechanism.

Photoprotein expressing hAMSC and U87 cells
To noninvasively monitor the behavior of tumor and therapeutic cells during therapy, hAMSCs were genetically modified by transduction with a lentiviral vectors for stable expression of a trifunctional chimeric reporter including Rluc, RFP and tTK activities, to generate Rluc-R-tTK-hAMSC or ''therapeutic vehicles''. Human glioblastoma U87 tumor cells were also genetically modified by transduction with a lentiviral vector for stable expression of a chimeric reporter with Pluc and eGFP activities, to generate ''Pluc-G-U87'' tumor cells. This strategy allowed independent imaging of therapeutic and tumor cells using non-cross-reacting luciferase substrates, coelenterazin and luciferin, respectively. In addition, the inclusion of two different fluorescent protein reporters, allowed selection of labeled cells by FACS, and detection of therapeutic and tumor cells by fluorescence confocal microscopy of tissue sections. Lentiviral transduction did not change doubling time or immunophenotype of Rluc-R-tTK-hAMSCs, as determined by flow cytometry analysis of hAMSC markers (Fig. S1).

Bystander killing of Pluc-G-U87 glioblastoma cells by Rluc-R-tTK-hAMSC in vitro
To demonstrate tTK gene was functional in the genetically modified therapeutic vehicles, replicate cultures of Pluc-G-U87, hAMSCs and Rluc-R-tTK-hAMSCs seeded in tissue culture plates were treated with ganciclovir (GCV) or PBS and cell survival assessed using the CellTiter 96 Aqueous One Solution Cell Proliferation Assay. Treatment with GCV had no effect on survival of Pluc-G-U87 alone (Fig. 1A) or on wt hAMSCs lacking the tTK gene (Fig. 1B). What is more, in the absence of GCV the growth pattern of hAMSCs was similar regardless of whether they expressed or not the tTK gene (Fig. S1B), indicating that the tTK gene had no negative effect on cell proliferation. However, as expected, GCV treatment did exert a strong cytotoxic effect, proportional to its concentration, on tTK expressing hAMSCs (Fig. S2) (*P,0,05).

Detection sensitivity of Pluc-G-U87 and Rluc-R-tTK-hAMSC in vivo
Skull bone and intervening brain tissue in live animals reduce the efficiency with which light photons reach the video camera detector used for BLI. To establish a correlation between light measurements and cell number, predetermined numbers of Pluc-G-U87 and Rluc-R-tTK-hAMSC were implanted in the brain of live mice at specific coordinates (0,6 mm posterior, 2 mm lateral and 2,75 mm depth respect Bregma) later used for experiments, and immediately after implantation imaged by BLI after administration of the corresponding luciferase substrate, luciferin or coelenterazine. The number of recorded photon events (PHCs) was extracted from recorded images and plotted, after background subtraction, versus the number of inoculated cells. As shown in Fig. 2, detection sensitivity (slope of the correlation curve) was similar for both cell types and there was a good linear correlation between the amount of light produced and the number of cells implanted, 34,2 PHCs/cell, R 2 = 0,994 and 20,64 PHCs/cell, R 2 = 0,974 for Pluc-G-U87 and Rluc-R-tTK-hAMSC, respectively.
Optimal Rluc-R-tTK-hAMSC dose for in vivo bystander effect Mixed cell populations comprising Pluc-G-U87 cells (4610 4 ) and different proportions of Rluc-R-tTK-hAMSCs, (1:1, 1:2, 1:4) respectively, were injected in the brain of SCID mice at the previously indicated coordinates (n = 15) to establish the best proportion of therapeutic to tumor cells. Two control groups (n = 10), were also included, one of them consisting of mice injected with 4610 4 Pluc-G-U87 cells only, that was treated with GCV; and the other consisting of mice injected with the same number of Pluc-G-U87 cells in a 1:1 proportion with Rluc-R-tTK-hAMSCs, that was treated with PBS. Imaging of Rluc activity was used to monitor implantation of Rluc-R-tTK-hAMSCs. Beginning six days after the cell implantation (T = 0), and during a 35 day period, the inoculated mice were treated daily by i.p. administration of GCV or PBS, as indicated. Light emission by Pluc luciferase was monitored by BLI from the beginning of GCV treatment and every week thereafter. Image sequences of tumors were used to monitor tumor growth during the experiment. Analysis of photons captured in all images and statistical evaluation showed ( Fig. 3A and B) that mixed cell tumors (1:1) treated with PBS grew at a similar rate than control tumors treated with GCV but without therapeutic cells (p.0,05). However, treatment with GCV of tumors bearing therapeutic cells inhibited tumor growth relative to controls. Although a trend in tumor growth inhibition appears from the start of the GCV treatment, a clear inhibition of tumor growth was apparent by day 21 and was evident by day 35 of GCV treatment in all the therapeutic cell proportions tested. The therapeutic effect directly correlated with the proportion of therapeutic cells used, being 1:4 the most effective cell dose, and in this group of mice, by the end of the experiment (T = 35) the amount of light produced by the tumors was 0,14% of the light produced by those in the control group (Pluc-G-U87 only) (p,0,05). Analysis comparing tumor size at the start (T = 0) and end (T = 35) of treatments shows a significant size reduction (p,0,05) for all the tumors treated with therapeutic cells plus GCV, but not for controls where either GCV or therapeutic cells were omitted. In the 1:4+GCV group, only 3% of light produced by tumor cells at T = 0 persisted at the inoculation sites following GCV treatment. Fig. 3C shows representative images from single animals from the groups (n = 5) treated with GCV and with or without therapeutic cells (1:4+GCV).
In vivo Rluc-R-tTK-hAMSC mediated bystander glioblastoma therapy A group of 24, 6 week old SCID mice was stereotactically inoculated at the previously indicated coordinates with either a mixture of 4610 4 Pluc-G-U87 cells and 1.6610 5 Rluc-R-tTK-hAMSCs (n = 16) or with 4610 4 Pluc-G-U87 tumor cells only (n = 8). Six days post implantation (T = 0), the mice were imaged by BLI to monitor Pluc activity from the Pluc-G-U87 tumors and Rluc activity from Rluc-R-tTK-hAMSC cells, and a treatment regime was initiated as described in the diagram of Fig. 4A. In brief, one half of the mice implanted with Pluc-G-U87 and Rluc-R-tTK-hAMSC cells received GCV, while the other half received the same volume of PBS. Mice implanted with only Pluc-G-U87 cells also received GCV. Treatment was administered during two 3 week periods, separated by a one week rest period. At day 49 all the control mice were sacrificed, as required by the animal welfare protocol.
Bioluminescence imaging (BLI) was used to monitor tumor development during the experiment. As indicated in Fig. 4B, showing representative images from single animals in each of the experimental groups, treatment of mixed cell tumors with PBS, or of control tumors (lacking therapeutic cells) with GCV had no inhibitory effect on tumor growth, and light production by Pluc-G-U87 cells progressively increased up to the end of the experiment. However, treatment of mixed cell tumors with GCV resulted in the progressive reduction of light produced by Pluc-G-U87 cells up to the end of the experiment, when light from Pluc-G-U87 cells reached background level ( Fig S3). Quantitative evaluation of therapeutic effect is shown by plots of total light events recorded in images from the tumor areas vs. time (Fig. 4D). This analysis shows that the amount of light produced by tumors containing therapeutic cells that were treated with GCV, by the end of the experiment was 10 4 times lower than that produced by control tumors treated with GCV or by tumors with therapeutic cells treated with PBS (p,0,05). Moreover, by the end of the experiment, light produced by tumors treated with therapeutic cells plus GCV was only 0,12% the amount of light produced by the same tumors at T = 0 (6 days after cell implantation) (n = 8).  Cell killing detected by BLI could also be verified at the macroscopic level by visual comparison of brain sections from control mice and those treated with therapeutic cells plus GCV (Fig. 4C), and while sections from control mice brains showed a clear tumor mass, those from mice inoculated with tumor plus therapeutic cells and treated with GCV had a normal macroscopic appearance and showed no signs of tumors.
Corroboration of BLI analysis was also provided by fluorescence confocal microscope observation of mouse brain sections that showed (Fig. 5) the presence of red fluorescent Rluc-R-tTK-hAMSCs evenly distributed throughout the tumor implantation site mixed with green fluorescent tumor cells, at T = 0 before starting GCV treatment (Fig. 5A). In well developed tumors (T = 49) from control group 1:4+PBS, red fluorescent Rluc-R-tTK-hAMSCs could also be detected in clusters within the green fluorescent mass of the well developed tumor (Fig. 5B). However, brains from mice in the 1:4+GCV group had few detectable tumor cells and no detectable Rluc-R-tTK-hAMSCs, in accordance with the BLI analysis (Fig. 5C).

Post-engraftment survival of therapeutic hAMSCs
The fate of Rluc-R-tTK-hAMSCs was monitored by BLI at the implantation day (T = 26) and 6 days later (T = 0), at the beginning of GCV treatment (Fig. 4E). The number of detected PHCs at T = 0 corresponded with that produced by 1.5610 4 Rluc-R-tTK-hAMSCs in standard plots (Fig. 2B), near to a 10 fold decrease relative to T = 26. However, due to their rapid proliferation the amount of light produced by Pluc-G-U87 tumor cells in the 1:4+GCV and 1:4+PBS groups, at T = 0 corresponded to approximately 2610 5 cells (Fig. 4E). Therefore, during the 6 days previous to GCV treatment, the ratio of PLuc-G-U87: Rluc-R-tTK-hAMSCs changed by 52 fold, from 1:4 to 13:1. Thus, the later is the actual ratio at which therapeutic effect is exerted.

Fate of tumor implanted Rluc-R-tTK-hAMSCs therapeutic cells
In an attempt to understand the mechanism by which the Rluc-R-tTK-hAMSCs+GCV treatment achieves the observed high degree of therapeutic effect, the differentiation behavior of therapeutic cells was analyzed. To do this, we isolated by FACS Rluc-R-tTK-hAMSCs and labeled them with a second Pluc-eGFP reporter under transcriptional control by the PECAM/CD31 promoter, as a reporter of differentiation to the endothelial lineage. With this strategy, changes in PECAM-regulated luciferase expression could be quantitatively related to those of the constitutively expressed Rluc, now an internal standard in the same cell that allows the elimination of potential artifacts related to changes in hAMSC number. The doubly labeled hAMSCs were mixed with un-labeled U87 tumor cells and implanted at the standard brain location. BLI analysis of Pluc and Rluc expression at days 0 and 7 post-implantation showed (Fig. 6A) that while the bulk of implanted hAMSCs disappears from the tumor during the 7 day period, cells expressing PECAM-promoter regulated Pluc remained associated to the tumor. Moreover, the ratio of PECAM/CD31 promoter-regulated Pluc activity to constitutively-expressed Rluc increased considerably by 92 fold, indicating that a subpopulation of the implanted hAMSCs had differentiated to the endothelial lineage and remained at the tumor site (Fig. 6B). In addition, analysis by laser confocal microscopy of tumor sections showed (Fig. 6C) that implanted cells expressing the RFP and eGFP regulated by the PECAM/CD31 promoter were closely associated to tumor microvascular structures and had endothelial morphology. The endothelial lineage of therapeutic cells association to vascular structures in the tumor was further verified by perfusing some of the mice with a fluorescein-conjugated high MW dextran not diffusible past the vascular endothelium. Confocal microscope analysis of sections of the perfused tumors showed the presence of RFP expressing therapeutic cells that positively stained with an anti-human PECAM/CD31 antibody and were closely associated to FITC-dextran labeled vascular structures (Fig. 6D).
In this experiment all the implanted therapeutic cells express RFP but only a fraction of them is induced to express the endothelial lineage reporters. This double label strategy offered the possibility of determining if the decrease in RLuc expressing therapeutic cells previously observed in tumors could be due to silencing of the CMV promoter that regulates RLuc, RFP and tTK expression. To do this, tumor sections were also analyzed to determine the proportion of eGFP expressing therapeutic cells that did not express RFP. Our assessment showed that most of eGFP expressing cells also expressed RFP and only 10% or less of the eGFP expressing cells did not express RFP, ruling out CMV promoter silencing as the cause for the decrease in production RLuc. photons within tumors.

Implantation of Rluc-R-tTK-hAMSC directly on established tumors also mediates bystander glioblastoma therapy
To verify that the Rluc-R-tTK-hAMSC mediated therapeutic approach is also effective under more clinical like conditions, 6 week old SCID mice were first stereotactically inoculated at the previously indicated coordinates only with 4610 4 Pluc-G-U87 cells. Six days post implantation (T = 0) one halve of the mice randomly chosen (n = 4) received an intracranial intratumoral injection of 1.6610 5 Rluc-R-tTK-hAMSCs. On day three mice were imaged by BLI to monitor Pluc-G-U87 tumor development, following which GCV treatment was initiated consistent of a daily dose of .GCV during a four week periods followed by 3 day rest periods. Weekly BLI was used to monitor tumor development during the experiment. On days 33 and 68, when the drop in Renilla luciferase emission indicated the disappearance of implanted Rluc-R-tTK-hAMSCs, additional intratumoral inoculations of 1.6610 5 Rluc-R-tTK-hAMSCs were performed.
As shown in Fig. 7A, the administration of Rluc-R-tTK-hAMSCs in combination with GCV results in the inhibition of tumor growth. Moreover, the tendency of treated tumors to regain growth when implanted therapeutic cells disappear is effectively conteracted by repeated inoculation of additional Rluc-R-tTK-hAMSC and GCV treatment. By day 53 right before all the control mice had died, light produced by tumors treated with therapeutic cells plus GCV was only 0,12% of that produced by the control tumors. Kaplan-Meier survival graph (Fig. 7B) showed that the treatment protocol resulted in significant (p,0.05) increase in animal survival. Median survival was 88,5 d (treated animals) and 51 d (control animals).

Discussion
In the current work we demonstrate a procedure to develop therapeutic strategies against a human U87 glioblastomas and an effective therapy model in SCID mice. The procedure is based on the use of hAMSCs genetically modified to express the herpes Simplex tTK as vehicles to deliver bystander toxicity to tumors. To analyze the behavior of tumor and therapeutic cells, we labeled the former with a CMV promoter regulated Pluc-eGFP chimerical reporter, and the latter with a different trifunctional chimerical reporter comprising Rluc-RFP and tTK activities [27]. This approach allowed us to monitor the location and number of tumor and therapeutic cells in live mice during treatment with GCV, and optimize cell and prodrug treatment.
We established that both, tTK activity in genetically modified hAMSCs, and administration of GCV were both required for bystander killing of tumor cells co-cultivated with hAMSCs.
In vivo BLI of predetermined numbers of hAMSC and U87 cells, stereotactically implanted at a specific site of the mouse brain, allowed us to correlate the number of photon events recorded by BLI with that of implanted cells, for their quantification during therapy.
An optimum therapeutic ratio of 4 hAMSCs to 1 U87 cell was empirically determined by implanting predetermined proportions of both cell types in mouse brains and administering GCV i.p. after a 6 day period, allowed for tumor establishment. Analysis of hAMSC behavior during the 6 day pre-treatment period indicated that a large proportion of the therapeutic cells disappeared, while tumor cells proliferated. This lead to a 50 fold, reduction in the proportion of therapeutic to tumor cells, that was 1:13 by the time the GCV treatment began. This result not only emphasizes the effectiveness of hAMSCs as therapeutic vehicles, but also suggests that procedures or scaffold materials for their protection would likely improve even more the therapeutic outcome. The disappearance of a large proportion of hAMSCs when injected in immunosupressed mice has been previously observed [28]. However, the reduction in hAMSC number should not be attributed to CMV promoter silencing. In a previous study by Vilalta et al. [19], hAMSCs expressing Pluc regulated by the constitutively active SV40 promoter also lost between 75-90% of cells within the first 10 days post inoculation in SCID mice. Moreover, since in our cell differentiation experiment all the therapeutic cells are initially red fluorescent and become also green fluorescent upon differentiation to the endothelial lineage, silencing of the CMV promoter should result in the appearance of green fluorescent cells that show no red fluorescence. Thus CMV silencing can be evaluated by determining the proportion of therapeutic cells expressing GFP (regulated by the inducible PECAM/CD31 promoter) that do not express RFP (regulated by the CMV promoter) in histological sections of tumors. While extensive silencing of the CMV promoter would be required to produce the 92 fold increase in PECAM/CD31-PLuc expression observed in the BLI experiments, our assessment found less than 10% of GFP positive cells that were not RFP positive. This result excludes CMV promoter silencing, and pointing to cell death or diffusion from the tumor sites as the likely causes for the drop in RLuc production following cell implantation.
The therapeutic effectiveness of hAMSCs was evaluated by comparing their capacity to inhibit U87 tumor growth by implanting the optimal 4:1 (hAMSC to U87) proportion in mice brains, and treating with GCV or PBS. Such experiments showed that treatment with GCV of tumors containing hAMSCs reduced the number of tumor cells by a factor of 10 4 , relative to tumors without hAMSCs, also treated with GCV, or tumors with hAMSCs treated with PBS.
For the same tumor, by the end of the experiment, treatment with hAMSCs plus GCV had reduced tumor cell number to 0,12% of that at T = 0. However, in tumors without hAMSCs but also treated with GCV tumor cell number increased by 319 fold in the same period.
Macroscopic and laser confocal microscope examination of brain slices at the end of experiments corroborated BLI imaging results. While brains from mice with untreated mixed-cell tumors had large tumor masses containing green and red fluorescent cells, no macroscopic tumors and few green fluorescent cells were found in the brains of mice implanted with tumors plus therapeutic cells and treated with GCV.
In experiments better modeling a clinical situation, we show that implantation of therapeutic hAMSCs on preestablished glioblastoma U87 tumors and treatment with GCV also results in tumor growth inhibition that lasts as long as there were surviving therapeutic hAMSCs. Moreover, repeated inoculations of therapeutic hAMSCs resulted in a progressive reduction of tumor size and a significant extension of mice survival, relative to untreated controls.
Insight on the therapeutic mechanism mediated by hAMSCs was gained using a double label strategy that allows monitoring of changes in gene expression on a ''per cell'' basis. This was achieved by labeling of the same therapeutic cell that already expressed Rluc-RFP reporter constitutively, with a different Pluc-eGFP reporter regulated by the inducible human PECAM promoter. By co-implanting these cells with unlabelled U87 tumors and monitoring by BLI we could observe that within the first 7 days post implantation, a large proportion of the hAMSCs disappeared from the tumor site, either by death or dispersion. However, a subpopulation of the cells that expressed PECAMpromoter regulated luciferase remained in close association with the tumor. Moreover, in the tumor associated cell population, the ratio of PECAM-regulated to CMV-regulated luciferase activity increased with time reaching by day 7 a value 92 fold higher than that at implantation time. This results indicated that tumor associated hAMSCs were actively differentiating to the endothelial lineage. Independent analysis by fluorescence laser confocal microscopy of tumor slices revealed RFP and eGFP expressing hAMSCs with endothelial morphology in close association with tumor vascular structures. This latter result was further corroborated by positive staining of microvessel associated red fluorescent hAMSCs with an anti human PECAM antibody.
Association of bone marrow derived MSCs with vascular structures in tumors has been reported previously [29,30]. However, the vascular associated cells were shown to express pericyte specific antigens while PECAM/CD31 is considered an endothelial specific marker. Both findings are not mutually exclusive, and in the absence of additional data, could be reconciled by considering that differentiation of mesenchymal stromal cells may depend on their tissue of origin as well as on the specific tumor environment in which they are implanted [31]. Our results suggest that GCV induced suicide of hAMSCs not only kills neighboring tumor cells by bystander effect but also by eliminating the associated vascular system that supplies oxygen and nutrients.
In conclusion, we demonstrate a general and versatile strategy to develop cell based tumor therapies and evaluate their effectiveness based on the use of chimeric bioluminescent and fluorescent reporters introduced in tumor and therapeutic cells, followed by analysis of their behavior by non invasive real time BLI and confocal microscopy.
With this strategy we were able to show that tTK expressing hAMSCs are very effective vehicles to deliver localized cytotoxic therapy to U87 glioblastomas and that hAMSCs within tumors differentiate to endothelial lineage cells that associate with vascular structures. Implantation of genetically modified hAMSCs may provide an effective procedure to eliminate residual tumor cells in surgical borders after tumor removal.
In future research we will use the above strategy to improve the preclinical significance of the therapy model. We will focus on evaluating procedures for therapeutic cell delivery, including the use of a biomaterial to protect and control the delivery of therapeutic cells implanted in tumors. To evaluate the relevance of the immune environment on the therapeutic effectiveness, we will use immune-competent mice implanted with singeneic glioma tumor model that will be treated with autologous therapeutic hAMSCs. Finally, we will seek to validate the generality of the therapeutic procedure using glioma tumors from human biopsies implanted in immunedefficient mice.

Vector constructs
Three lentiviral vectors constructs were used to label cells. hrlmrfp-tk vector contains a chimeric trifunctional reporter comprising Rluc, monomeric red fluorescent protein (mRFP1) and a truncated version of the herpes simplex virus thymidine kinase coding sequences (sr39tk) regulated by the Cytomegalovirus promoter, was constructed [27]. pRRL-Luc-IRES-EGFP vector, containing the PLuc under control of the cytomegalovirus (CMV) promoter followed by an IRES to express an eGFP, was a kind donation from L. Alvarez-Vallina [32]. pLox:Pecam-Luc:eGFP vector that expresses chimeric Pluc and eGFP regulated by the endothelial specific promoter PECAM was constructed in our group. Lentiviral vector containing a fusion reporter comprising PLuc and eGFP, was obtained by polymerase chain reaction amplification and standard cloning procedures using the Pluc and eGFP genes from plasmid pGL4.10:PLuc (Promega Corporation, Madison, WI, USA) and pEGFP-N1 plasmid (Clontech Lab., Palo Alto, CA, USA). hPECAMp was kindly provided by Dr. Carmelo Bernabéu.

Lentiviral particle production
Production of viral particles was performed using human embryonic kidney cells 293T (ATCC, CRL-11268 TM ) grown in Dulbecco's Modified Eagle's Medium-high glucose (DMEM-hg) (Sigma, Steinheim, Germany), 10% heat-inactivated fetal bovine serum (FBS) (Sigma), 2 mM L-glutamine (Sigma), 50 units/ml penicillin/streptomycin (Sigma), and 2 mM HEPES. The day previous to transfection, 3610 6 cells were seeded on 10 cm 2 poly-D-lysine (Sigma) treated plates. 6 mg of each lentiviral transfer vector, were mixed with 2 mg of viral envelope plasmid (pMD-G-VSV-G) and 4 mg of packaging construct (pCMV DR8.2) in 250 ml of 150 mM NaCl and then mixed with 48 ml of 1 mg/ml polyethylene amine (Polyscience, Warrington, PA, US) in 250 ml of 150 mM NaCl, and incubated at room temperature (RT) for 20 min. This DNA solution was then added drop wise to the plate containing the 293T cells plus medium, swirled gently and incubated for 16 h at 37uC with 5% CO 2 . The following day, the transfection solution was removed, the cells were rinsed with PBS 1X and medium without FBS was added to the cells. Following a 48 h incubation the supernatant was collected, centrifuged at 2000 rpm to remove cell debris, and filtered through a 0,45 mm low protein binding filter (Corning, Bath, UK).
Cell lines and cell culture hAMSCs were isolated from adipose tissue derived from cosmetic subdermal liposuctions, with patient consent, as described previously [26]. Liposuction samples were obtained after written informed consent by anonymous donors from Hospital de la Santa Creu i Sant Pau, Barcelona, Spain. Work with human samples was approved by written consent by the Ethical Committee of Clinical Investigation of Hospital Santa Creu i Sant Pau, Barcelona, Spain; and Bioethical Subcommittee of Superior Council of Scientific Research. Briefly, lipoaspirate was suspended in 1X collagenase type I (Invitrogen, Carlsbad, CA) solution and incubated at 37uC and inactivated by addition of DMEM+10% FBS. hAMSC were isolated by plastic adherence technique. hAMSCs were grown in DMEM-hg with 20% FBS (Hyclone, Logan, UT), 2 mM L-glutamine (Sigma) and 50 units/ ml penicillin/strepotomycin (Sigma), expanded for until 70% of confluence and infected with either the hrl-mrfp-tk, or with both, first with the hrl-mrfp-tk and then with the pLox:Pecam-Luc:eGFP viral vector stocks, using 10 mg/ml polybrene (Sigma) and incubated for 24 to 48 hours (MOI = 20). Human glioma cells U87 (ATCC, HTB-14), were grown in nutrient mixture DMEM/ Hams F12 HAM containing 10% heat-inactivated FBS (Sigma), 2 mM L-glutamine (Sigma) and 50units/ml penicillin/strepotomycin (Sigma). U87 were infected, as described, with pRRL-Pluc-IRES-eGFP viral stock (MOI = 20) to obtain Pluc-G-U87, as described above. Expression of fluorescent proteins was used to select positively transduced cells.

Cell proliferation assays
Quadruplicates of 3610 3 Rluc-R-tTK-hAMSC or untransduced hAMSCs an 5610 3 Pluc-G-U87 cells were plated into 96well plates and incubated overnight at 37uC. Culture medium was replaced with medium containing 4 mg/ml GCV 24 h later (T = 0). Cell proliferation was evaluated spectrophotometrically at 490 nm, using the CellTiter 96 Aqueous One Solution Cell Proliferation Assay (Promega) at days 0, 5 and 9. Results were expressed as the percentage of proliferation, where the proliferation of cells in culture medium at day 0 was set to 100%. Mixed cell cultures containing 10 4 Pluc-G-U87 with several proportions of Rluc-R-tTK-hAMSC were seeded on 48-well plates and analyzed by BLI at days 0 (the onset of GCV treatment), 5 and 9, with changes of medium every 3 days. Medium was changed every 3 days.

BLI determination of luciferase activity
In vitro cultures. For BLI of tissue culture plates, medium was removed from the wells, the wells were rinsed twice with PBS 16, and imaged immediately following addition of 100 ml/well of Pluc or Rluc substrate stock reagent (Caliper, Hopkinton, MA, US and Prolume, Pinetop, AZ, US). Imaging of PLuc and Rluc activities was performed in consecutive days.
For imaging, plates are placed in the detection chamber of a high-efficiency ORCA-2BT Imaging System (Hamamtsu Photonics, Hamamatsu City, Japan) provided with a C4742-98-LWG-MOD camera fitted with a 5126512 pixel charge couple device (CCD) cooled at 280uC. Images were acquired during 1 min using 161 array (binning 161), and in order to register the position of the light signal, an additional image was obtained using a white light from a lamp in the detection chamber. Light events were calculated using the Wasabi image analysis software and expressed as PHC after discounting the background of wells without cells. The net number of PHCs in the area of interest was calculated using the formula: PHCs = (total number of PHCs in the area of interest)2[(number of pixels in the area of interest)6(average background PHCs per pixel)].
Pseudo color images were generated using arbitrary color bars representing standard light intensity levels for Pluc (blue: lowest; red: highest) and for Rluc (black: lowest; blue: highest).
In vivo bioluminescent imaging. In vivo BLI of engrafted SCID mice was performed as described previously [26]. Briefly, mice were anesthetized i.p. and then injected i.p. with 150 ml of luciferin (Caliper) (16.7 mg/ml in saline) or through tail vein with 25 ml of benzyl coelenterazine (1 mg/ml in 50/50 propilenglycol/ ethanol) (Prolume) diluted in 125 ml of water. When BLI was performed after the surgery no anesthesia was required. Animals were placed in the detection chamber of the high efficiency ORCA-2BT Imaging System and images acquired from the dorsal direction during a 5 min period. A second image of the animal was obtained using a white-light source inside the detection chamber, to register the position of the luminescence signal. To increase detection sensitivity, the readout noise of the recorded signal was reduced by adding together the light events registered by arrays of 868 adjacent pixels that are read simultaneously (binning 868) of the camera CCD. Mice were imaged weekly during experiments. Quantification and analysis of photons recorded in images was done using the Wasabi image analysis software (Hamamatsu Photonics) as described above.
In order to establish the correlation between the number of transplanted cells and the amount of emitted light, standard curves were generated by imaging either 1610 4 , 2,5610 4 , 5610 4 and 1610 5 Pluc expressing cells or 1610 4 , 3610 4 , 5610 4 and 1.5610 5 Rluc expressing cells, stereotactically injected in the brain (0.6 mm posterior, 2 mm lateral and 2.75 mm depth respect bregma). Light measurements were expressed as PHCs following subtraction of background as previously described. Data were represented as the number of PHCs versus number of grafted cells.

Animal experiments
Adult 6-8 weeks old SCID mice were purchased from (Charles Rivers, Wilmington, MA, USA) and kept under pathogen-free conditions in laminar flow boxes. Animal maintenance and experiments were performed in accordance with established guidelines of Catalan Government and protocol num. 4565 approved by Direcció General del Medi Natural, Generalitat de Catalunya. Animals were anesthetized by i.p. injections of xylazine (Henry Schein, Melville, NY,USA) 3.3 mg/kg and ketamine (Merial, Duluth, GA, USA) 100 mg/kg. Subsequently mice were mounted in a stereotactic frame (Stoelting, Wood Dale, IL, U.S.A.) and their heads were secured using a nose clamp and two ear bars, and a skin flap was lifted to expose the skull surface and further anesthetized, by injection of fentanil (Kernpharma, Barcelona, Spain). For stereotactic cell implantation, a burr hole was drilled and the cell suspensions was injected at a 250 nl/min rate using a Hamilton syringe series 700 at the previously indicated coordinates. The injections were delivered at a rate of 0.25 ml/ min, and the needle was slowly withdrawn after an additional 5 minutes. The scalp was closed by suture and the animals were placed in individual recovery cages and supplied with buprenorfine (Buprex, Schering Plough SA, Madrid, Spain) in the drinking water. GCV sodium (Cymevene, Roche, Basel, Switzerland) was injected i.p. daily at a dose of 100 mg/kg [33,34].Control animals were inoculated with the same volume of PBS.

Histology
Brain from sacrificed mice were harvested, washed with physiological serum and fixed with paraformaldehyde during 48 hours. Brains were then washed with PBS 16, embedded in OCT, sliced in 20 mm sections and mounted on glass slides. Hoechst staining was performed for detection of cell nuclei. For macroscopic view analysis 20 mm brain sections were stained with Hematoxylin and Eosin (HE). Immunohistochemical detection of endothelial cells was performed on 10 mm thick section using a primary hPECAM-1 (4 mg/ml) (Abcam, Cambridge, UK). Secondary antibody conjugated with Cy5 (3 mg/ml) (Jackson ImmunoResearch, Suffolk, UK) was applied and sections counterstained with Bisbenzimide (Hoechst, Sigma).

Fluorescence angiography and Laser Confocal microscope observation of brain implanted hAMSCs
For tumor microvessel imaging, mice were anesthetized and injected trough the lateral tail vein with 200 ml (10 mg/ml) of a high molecular weight (2,000,000 MW) FITC-conjugated dextran Sigma (St Louis, MO, US). Ten minutes after the injection mice were sacrificed and brains were retrieved and fixed in formalin solution 10% (Sigma) during 24 hours. The fixed brains were sliced and analyzed for microvessel formation and presence of fluorescent hAMSCs using confocal laser scanning microscopy (Leica TS1 SP2).