Primary gene-engineered neural stem/progenitor cells demonstrate tumor-selective migration and antitumor effects in glioma

Abstract

The prognosis of patients with glioblastoma multiforme (GBM) is generally poor after surgical tumor resection. With the aim of developing new adjuvant therapeutic strategies, we have investigated primary neural stem/progenitor cells (NSPC) in co-cultures with glioma cells, and in a model of gene therapy on aggressively growing malignant glioma. NSPC exhibited tropism towards medium conditioned by glioma cells, and in adherent low-cell density co-culture, were attracted to, and fused with, tumor cells. Similarly, within 24–48 hr of co-culture in suspension, NSPC-tumor hybrids were observed, representing 2–3% of the total cell population. NSPC were then coinjected into mouse brain with GBM cells, employing NSPC expressing cyclophosphamide (CPA)-activating enzyme cytochrome p450 2B6 (CYP2B6), which catalyzes CPA prodrug transformation into membrane diffusible DNA-alkylating metabolites. Upon CPA administration, NSPC containing CYP2B6 elicited substantial impairment of tumor growth. When implanted intracerebrally at a distant site from the tumor, gene-engineered NSPC specifically targeted GBM grafts, after traveling through brain parenchyma, and hindered tumor growth through local activation of CPA. Directed migration of primary NSPC corresponded closely with intracerebral and tumoral pattern of expression of vascular endothelial growth factor, which is a motility factor for NSPC. Overall, these findings indicate that therapeutic gene delivery mediated by primary NSPC is a potentially valid strategy for treatment of high-grade gliomas.

Glioblastoma multiforme (GBM) is one of the most deadly tumors of the central nervous system, associated with high risk of recurrence as result of the difficulty of complete removal by surgical excision.1, 2 The survival expectancy of patients prolonged by adjuvant therapies rarely reaches 2 years from diagnosis.1, 3 In the past years, new therapeutic strategies, based on delivery of antineoplastic molecules in the proximity of the tumor by neural stem/progenitor cells (NSPC) and other types of cells, have been proposed for high-grade gliomas.4, 5

The therapeutic function of NSPC is currently being appraised, in large part because established neural progenitor cell lines display tropism to tumor foci6; the pleiotropic factors vascular endothelial growth factor (VEGF), which acts as survival factor for adult NSPC,7 and hepatocyte growth factor (HGF), have also been recognized as motility factors for NSPC,8, 9 playing a role in guiding brain tumor-selective motility of NSPC.8, 10, 11 In addition, transplants of NSPC may conceivably contribute to relieve pathological conditions. To this respect, both primary NSPC, which form floating spheroids in culture called neurospheres, as embryonic neural progenitor cell lines, release factors while differentiating that directly inhibit glioma proliferation.5, 12, 13 When engineered for production and release of cytokines to boost the immune function, primary NSPC grafted intratumorally hamper glioma development.5, 14 Similarly, intratumoral injection of cytosine deaminase-producing neural progenitor cell lines ST14A and HB1.F3 and administration of 5-fluorocytosine prodrug, resulted in tumor size reductions of 50 and 75% against C6 rat glioma model and medulloblastoma, respectively.15, 16 In general, immortalized neural progenitor cell lines have been more extensively studied in animal models than primary cells for transport of therapeutic agents to sites of cancerous growth in the central nervous system.17, 18

Gene therapies involving activation of inert prodrugs are promising treatments for infiltrative gliomas, in part because formation of diffusible products results in enhanced cytotoxic action.19 Forclinical translation of cell-mediated gene therapies however, identifying suitable vector cells requires discerning how stem-like cells behave in presence of tumor. Here, primary NSPC characterized in co-cultures with GBM cell lines revealed tumor-tropism like that of neural progenitor cell lines and, when engineered for production of therapeutic molecules, tracked intracerebral aggressive glioma and produced antiproliferative effect.

bFGF: basic fibroblast growth factor; BSA: bovine serum albumin; CPA: cyclophosphamide; CYP2B6: cytochrome p450 2B6; DAPI: 4′,6-diamidino-2-phenylindole; DsRed: Discosoma species red; EGF: epidermal growth factor; FBS: fetal bovine serum; FITC: fluorescein isothiocyanate; GBM: glioblastoma; GFAP: glial fibrillary acidic protein; GFP: green fluorescent protein; H&E: hematoxylin and eosin; HGF: hepatocyte growth factor; NB: neurobasal medium; NSPC: neural stem/progenitor cells; T-PBS: phosphate buffered saline with 0.5% Tween-20; TRITC: tetramethyl rhodamine iso-thiocyanate; VEGF: vascular endothelial growth factor; VEGFR1: vascular endothelial growth factor receptor-1; VEGFR2: vascular endothelial growth factor receptor-2

Material and Methods

NSPC preparation

NSPC were obtained from C57BL/6 newborn mice (Harlan laboratories, Indianapolis, IN). Two-day-old pups were euthanized, and brains extracted. After excising the middle brain, the forebrain was minced with a scalpel, and disgregated in neurobasal medium (NB); medium supplemented with B27, 50 U/mL penicillin, 50 μg/mL streptomycin, 2 mM L-glutamine, 20 ng/mL epidermal growth factor (EGF) and 10 ng/mL basic fibroblast growth factor (bFGF) [all from Invitrogen (Carlsbad, CA), except EGF and bFGF from R&D systems (Minneapolis, MN)]. The mixture was filtered through 70 μm-cell strainer, and incubated at 37°C, 5% CO₂, in cell culture flasks. Neurospheres were decanted and transferred to fresh medium after 4 days. Following previously reported methods,20, 21 retroviral particles containing either green fluorescent protein (GFP)- or cytochrome p450 2B6 (CYP2B6)+GFP-pSF91 retroviral vector were used to transduce NSPC. Fluorescent cells were separated from non-transduced cells by fluorescence activated cell sorting in a BD FACSVantage cytometer and grown in culture.

Adherent and suspension co-cultures of NSPC with GBM cells

Either GBM cell lines U87MG or GL261, or Discosoma species red fluorescent clones (U87MG-DsRed and GL261-DsRed), derived from parental cells by a method described elsewhere,21 were cultured in NB for use in co-culture experiments with either NSPC or NSPC-GFP. Adherent co-cultures, kept in NB in absence of EGF and bFGF, were established by sequential centrifugation of GBM cells (160g; 500 cells/cm²) and NSPC (580g; 200 cells/cm²) (Thermo Fisher Scientific Forma 1L GPR centrifuge, with swinging bucket rotor for plates), onto plates previously coated with 10 μg/mL retronectin (Takara Bio, Otsu, Shiga, Japan) in phosphate buffered saline (PBS), and then with 2% bovine serum albumin (BSA). Suspension co-cultures in NB without EGF and bFGF consisted of NSPC and GBM cells seeded together in noncoated flasks.

NSPC chemotaxis to medium conditioned by U87MG cells

NSPC were assessed in 8-μm-pore FluoroBlock insert system (BD Biosciences, San Jose, CA) for GBM cell line-elicited tropism. 1.5 × 10⁴ NSPC were added per insert, and U87MG-conditioned NB without cells added to the bottom chambers. Plates were placed in 37°C incubator for 1 hr and NSPC counted on the underside of filters.

Co-immunostaining of NSPC and tumor markers in co-cultures

Immunolabeling of co-cultures was carried out on formalin-fixed cells by overnight 4°C incubation with primary antibody followed by 45 min incubation (when necessary) with secondary antibody at room temperature, using 1% BSA in 0.5% Tween-20-PBS (T-PBS) as nonspecific blocker, and T-PBS for washing. For nestin/DsRed double staining, co-cultures of NSPC and U87MG-DsRed were labeled first with rodent-specific nestin antibody (1:200, Chemicon, Temecula, CA) and peroxidase Vectastain ABC kit development (Vector Labs, Burlingame, CA), and then with DsRed antibody (1:300, BD Living Colors, Mountain View, CA) revealed by tetramethyl rhodamine iso-thiocyanate (TRITC)-secondary antibody (1:300, Jackson Immunoresearch, West Grove, PA). For GFP/CD44 double labeling on suspension co-cultures of NSPC-GFP and U87MG, cells were trypsinized, plated on retronectin and labeled with phycoerythrin-CD44 antibody (1:500, BD Biosciences, San Jose, CA) and then with fluorescein isothiocyanate (FITC)-GFP antibody (1:500, GeneTex, San Antonio, TX). Cells were examined in an Axiovert 200 M Zeiss fluorescence microscope.

Hoechst analysis on co-cultures of NSPC and tumor cells

Flow cytometric analysis of Hoechst DNA labeling was performed on suspension co-cultures of NSPC-GFP and U87MG-DsRed cells seeded together 2 days before analysis at equal cell numbers (10⁵ cells/mL). Cells were trypsinized, incubated for 90min at 37°C in medium containing 20 μg/mL Hoechst (Calbiochem, Gibbstown, NJ), fixed with formalin, washed with PBS-0.1% BSA and analyzed in a BD FACSVantage cytometer.

Cell implants in mouse brain

All animal procedures were performed under protocols approved by the University of South Alabama Animal Care and Use Committee. NSPC-GFP was injected intracerebrally into the right hemisphere of ketamine/xylazine-anesthesized 6-week-old female nude mice (Harlan laboratories, Indianapolis, IN). After making an incision in the scalp, a burr hole of 0.3–0.4 mm across was drilled in the skull at 2 mm from sagittal suture, 2 mm caudally from bregma. Using a stereotactically guided Hamilton syringe, NSPC resuspended in NB were injected (10⁵ NSPC/animal) at 3 mm depth from skull surface, 5 μL maximum volume per mouse. Mice were euthanized 2 or 10 days after implantation, whole brains harvested, fixed with formalin, bisected through the lateral plane of injection and paraffin-embedded for serial 5–10 μm thick coronal sections.

To study tumor tropism of NSPC, nude mice pre-implanted in the right hemisphere with 5 × 10⁴ GBM cells/animal at 2 mm from sagittal suture, 2 mm caudally from bregma and 3 mm depth, were inoculated 2 days later in the same cerebral hemisphere with 5 × 10⁵ NSPC-GFP/animal at 4 mm from sagittal suture; bregma and depth coordinates as for tumor. Mice were euthanized within 1–8 days post-NSPC injection and brains harvested for processing as described earlier.

Immunostaining of NSPC implants

Immunostaining on tissue sections, after de-paraffinization and boiling in 0.01 M citric acid pH 6 for antigen retrieval, was carried out by overnight 4°C incubation with primary antibody in 1% BSA-T-PBS and, if indicated, 3 hr incubation at room temperature with corresponding secondary antibody. Expression of nestin and neural differentiation markers was studied on NSPC-GFP implants by double immunofluorescence for GFP and corresponding marker. Sections were incubated with antibody (1:200) against either murine nestin, neuron-specific βIII-tubulin (R&D systems, MN), glial fibrillary acidic protein (GFAP) (Santa Cruz, CA), or oligodendrocyte progenitor marker O4 (R&D systems, MN), then with TRITC-secondary antibody (1:300), and finally with FITC-GFP antibody (1:300). Tissue was counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (1:1,000, Sigma, St. Louis, MO) and cover-slipped using fluorescent mounting media (Calbiochem, Gibbstown, NJ).

Implantation of NSPC-CYP2B6 for treatment of glioma with cyclophosphamide

Nude mice were inoculated into the right hemisphere with a suspension of U87MG cells and either NSPC-CYP2B6 or NSPC-GFP as control (5 × 10⁴ tumor cells plus 2.5 × 10⁵ NSPC in NB/animal) at 2 mm caudally from bregma, 2 mm from sagittal suture and 3 mm depth. Two days later, the scalp incision was opened to introduce subcutaneously in the back of each mouse an osmotic pump (Alzet, Durect Corporation, Cupertino, CA) containing sterile saline solution of cyclophosphamide (CPA). With a cannula penetrating 2 mm from the skull surface into the brain, CPA was infused intracerebrally through the burr hole at 75 μg/day until mice were killed; then, brains were harvested for processing as described earlier. Six coronal sections anterior, and 6 posterior, to the bisection, cut at 0.1 mm intervals, were stained with hematoxylin and eosin (H&E) to assess tumor volumes.

For distant intracerebral delivery of NSPC-CYP2B6, nude mice were injected into the right hemisphere at 2 mm from sagittal suture, 2 mm caudally from bregma and 3 mm from skull surface, with GL261 (2.5 × 10⁴ cells/animal), and the next day inoculated with NSPC-GFP, NSPC-CYP2B6 (5 × 10⁵ cells/animal) or sham injection, at 4 mm from sagittal suture in the right hemisphere; bregma and depth coordinates as for tumor. One week after tumor injection, CPA delivery pumps were implanted as described earlier. Nineteen days after tumor implant, mice were euthanized to assess tumors on H&E coronal sections.

Results

NSPC undergo spontaneous fusion with GBM cells

In adherent low-cell density co-culture, NSPC displayed processes which, at early stages, made contact with scattered tumor cells at a distance of up to 300 μm (Fig. 1a). Within 4 days from plating, numerous individual NSPC and GBM cells had approached each other (Fig. 1b). Although both types of cells were motile, changing their positions rapidly, the fact that NSPC processes, which were 2–5 times the length of tumor processes, oriented gradually (Fig. 1a), suggested that GBM cells elicited attraction. A number of cells in contact formed single cells with double fluorescence (Fig. 1c). To exclude overlapping of cells or exchange of soluble marker proteins through pores as cause of color merging, U87MG-DsRed human cells co-cultured with NSPC for 7 days were analyzed for presence of rodent form of cytoskeletal protein nestin. Using a mouse-specific antibody unable to recognize human nestin, GBM cells containing patches of mouse nestin were found (Fig. 1d), indicating that these cells incorporated whole NSPC cytoplasms through membrane fusion; the occurrence of mononucleated cells, in addition to binucleated cells with fused cytoplasms, further suggested genuine cell–cell fusion.

Figure 1. NSPC fusion with GBM cells. (a) Adherent low-cell density co-culture of NSPC-GFP (green) with U87MG-DsRed cells (red) showing processes extending from NSPC towards tumor cells after 3 days of co-culture. (b) NSPC and glioma cells after 4 days of co-culture. (c) Left: A U87MG cell (red) with membrane apposition to NSPC (green) was followed-up; right: 17 hr after, a single recognizable cell has double (red and green) fluorescence. (d) DsRed/murine nestin double immunolabeling on U87MG-DsRed (human)/NSPC (mouse) adherent co-culture showing double label in a mononucleated hybrid cell; right: tumor cells lacking nestin reaction. (e) GFP/CD44 double immunofluorescence on NSPC-GFP (green)/U87MG (CD44 positive; red) suspension co-culture showing 1 cell (arrowheads) with uniform color coalescence. (f) Membrane fusion in NSPC-GFP/U87MG suspension co-culture cells double positive on GFP/CD44 immunofluorescence (insets). Nuclei stained with DAPI (blue). (g) NSPC-GFP/GL261-DsRed hybrid cell giving rise to 2 double-fluorescent daughter cells; pictures show successive time-points between day 5 and 6 from transference of cells from suspension co-culture to 10% FBS-medium. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

In suspension co-cultures of NSPC-GFP with U87MG, double-positive cells for GFP and U87MG marker CD44 were ascertained by double immunofluorescence (Fig. 1e), as well as cells undergoing membrane fusion (Fig. 1f). The viability of hybrids was assessed on NSPC-GFP/GL261-DsRed suspension co-culture; after 3 days of co-culture, cells were trypsinized and transferred to medium containing fetal bovine serum (FBS); division of hybrid cells was detected after 5 days in FBS (Fig. 1g). The rate of fusion was estimated in suspension co-cultures from the number of GFP/DsRed cells with high DNA content, as determined by flow cytometric Hoechst analysis, using a fixative to partially quench DsRed fluorescence. Red/green cells had relative increase of DNA content, as represented by higher Hoechst binding (Fig. 2). A 2–3% rate of fusion was assessed, measured as percentage of red/green cells with increased DNA content over total fluorescent cells.

Figure 2. Hoechst analysis of DNA content on NSPC-GFP/U87MG-DsRed suspension co-culture. Analysis by flow cytometry pointing out a population of cells (P3) emitting both red and green fluorescence. Median and mean values for Hoechst labeling of DNA are roughly 2-fold greater for P3 than for red only (Q1) or green only (Q2) cells. Experiment controls of Q1 and Q2 analyzed separately are shown. The fraction of cells with undetectable fluorescence was due to use of a quenching reagent. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

NSPC differentiation

In mouse brains transplanted solely with NSPC, the timing of entry of NSPC into advanced stages of differentiation was studied by double immunofluorescence for GFP and neural markers. Two days after transplantation, nearly 100% of NSPC were positive for nestin (Fig. 3), a marker of early neural progenitors, and lacked markers of neural lineage, indicating that NSPC maintained for some days a phenotype identical to that in vitro. Ten days after implantation, NSPC had engrafted, lacked nestin and showed GFAP expression indicative of astrocytic differentiation. In contrast to NSPC cultures under differentiating conditions, i.e. removal of EGF and bFGF and addition of 3% FBS, in which 30% of cells were positive for βIII-tubulin expression and 5% for O₄ (not shown), those markers were not detected in implanted cells.

Figure 3. Astrocytic differentiation of NSPC in brain. Double immunofluorescence on NSPC-GFP implants for GFP and either nestin (up) or GFAP (down) after 2 days (left) or 10 days (right) from injection into mouse thalamus. Nestin expression at day 2, indicative of undifferentiated NSPC, disappears after 10 days in tissue. Astrocyte marker GFAP, absent at day 2, is present in many implanted cells at day 10. Nuclei stained with DAPI (blue). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Correspondence between NSPC migration and VEGF expression in brain

NSPC were injected intracerebrally 2 mm apart, in the same hemisphere, from GL261 implants, that produced big tumors, or U87MG implants, which developed smaller tumor foci, often branched and extended ventrally. Within 8 days from injection, most NSPC migrated and targeted the tumor mass, to an extent comparable for both experimental tumors (Fig. 4). We then investigated tumor production of VEGF and HGF, 2 known chemotactic factors for NSPC, using immunohistochemistry. Slight staining with HGF antibody was seen in U87MG, but not GL261, and both tumors were strongly reactive with VEGF antibody (Fig. 5).

Figure 4. Directed migration of NSPC to glioma. (a) Immunofluorescences on mouse brain after inoculation with U87MG and with NSPC-GFP at 2 mm from the tumor, showing NSPC (green; arrowheads) at various times from injection, and migrating to the tumor (T). Middle panel: H&E-stained coronal view indicating injection paths for tumor (T.I.) and NSPC (N.I.). For orientating, the positioning of the pictures in general view is outlined. (b) Mouse brain immunofluorescence after injection with GL261 cells and with NSPC-GFP at 2 mm from the tumor, showing NSPC at day 5 from injection (left), and at day 8 (right), after targeting the tumor. Sections probed with FITC-GFP antibody and DAPI (blue). (c) Quantification of NSPC density in the tumors; bars represent mean ± standard deviation from 4 tumors. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Figure 5. VEGF and HGF immunostaining in gliomas. (a) Synthesis of VEGF in brain tumors of U87MG and GL261 cells. (b) HGF-positive immunoreaction in U87MG, and lack of staining in GL261 glioma. Tissue sections were probed with anti-Hu-Ms VEGF or HGF antibodies (1:200, Santa Cruz, CA), developed with peroxidase and counterstained with methyl green. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

To further explore NSPC migration in vivo, in a separate set of experiments we injected, 7 days after tumor injection, NSPC at the tumor site or at the corresponding tumor coordinates in the contralateral hemisphere. When injected contralaterally, NSPC migrated towards the glioma, reaching the interhemispheric fissure within 2 weeks, with longest trajectories along a vascular track running under the hippocampal area (Figs. 6a and 6b), and a layer of hippocampal neurons (Figs. 6a and 6c). VEGF staining of vessels disclosed strong positivity along the path of NSPC movement (Fig. 6d) as compared to vessels of small size, corresponding with strong staining for endothelial marker (Fig. 6e). Expression of VEGF was also evident in neurons, most prominently in the layers of packed neuronal bodies of hippocampus (Fig. 6f), where NSPC moved through. When injected intratumorally, NSPC often lined vessels of tumor vasculature; in C57BL/6 mice, NSPC also moved oriented to rims of inflammatory response, and occasionally long-range migration was observed along angiogenic vessels (Fig. 7).

Figure 6. Correspondence between NSPC migration and expression of VEGF in tissue. (a) Immunofluorescence of mouse brain showing directional migration of NSPC (green; arrowheads) along trajectories of brain vessels (white arrow) and hippocampal neurons (block arrow). The position of NSPC injection (dashed arrow) was contralateral to the site of glioma inoculation. (b) Higher magnification of the vascular trajectory in (a) to show the narrow positioning of NSPC along the track. (c) High power view of area in (a) where NSPC trajectories split, showing cells moving along the granular layer (GL). (d) VEGF immunofluorescence disclosing more reactivity (red) in NSPC-attracting brain vasculature as compared to smaller vessels (arrowheads). Sections counterstained with DAPI (blue). (e) Immunohistochemistry for von Willebrand factor showing strong reaction with endothelium (E) along the NSPC pathway; antibody labeling (1:200, Chemicon, Temecula, CA) was developed with peroxidase. (f) VEGF immunohistochemistry on mouse brain staining neurons of dentate gyrus area that synthesize significant amount of VEGF. Sections counterstained with methyl green. VEGF sections probed with antibody and developed with either TRITC-secondary antibody (d) or peroxidase reaction (f). H, hippocampus. SP, stratum pyramidale. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Figure 7. NSPC migration along blood vessels. (a) Up: Immunofluorescence showing intratumorally-injected NSPC (green) migrating along a vascular track. Down: H&E staining of tumor (T), indicating the route of migration (arrowhead). (b) Immunofluorescence revealing the end point of NSPC perivascular migration in a ventral outgrowth located at 3 mm from the tumor. Nuclei stained with DAPI (blue). Right: Leukocytic infiltration in the outgrowth as revealed by H&E staining. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Anti-VEGFR1 antibody inhibits U87MG-induced chemotaxis of NSPC

To investigate whether VEGF expression in tissue was responsible for tropism, the effect on chemokinesis of selective blockage on NSPC of VEGF receptor-1 (VEGFR1) or -2 (VEGFR2) was assessed, using medium conditioned by U87MG, which elicited high migration response, as stimulus. VEGFR1 antibody, at a concentration in the range of the neutralization dose₅₀ recommended by manufacturer, decreased significantly the number of chemoattracted cells (p < 0.0001) (Table 1), resulting in 80% inhibition of migration, expressed as ratio of migratory NSPC, subtracted those of negative controls, versus IgG control.

Table 1. Effect of VEGFR neutralizing antibodies on U87MG-induced NSPC tropism¹

Inhibition of glioma growth by CPA activation mediated by NSPC-CYP2B6

As proof-of-concept of anti-glioma effect of CPA activation in vivo, U87MG cells and NSPC-CYP2B6 were grafted together in mouse brain; administration of CPA resulted in inhibition of tumor growth in all animals with orthotopic grafting of NSPC-CYP2B6, as compared to grafting of control NSPC-GFP. After CPA administration, nearly 75% of CYP2B6-treated tumors had no NSPC remaining, whereas the other 25% showed 3-fold diminished number of cells, as compared to number of NSPC-GFP persisting in controls. The average brain tumor volume was reduced by 8-fold in mice implanted with NSPC-CYP2B6 (Table 2), and, consistent with the antiproliferative effects of CPA-derivatives,19 the remaining tumor masses had significantly lower tumor cell proliferation rates. These results suggested effective prodrug activation in vivo driven by NSPC-CYP2B6.

Table 2. Effect on U87MG tumor growth of CPA-activating therapy mediated by NSPC-CYP2B6

In a clinical situation, gene-engineered cells should migrate towards tumor cells that have infiltrated the brain parenchyma. For this reason, distant injection of NSPC-CYP2B6 was tested for anti-glioma effect on GL261 murine tumors. Injection of NSPC-CYP2B6 at 2 mm distance from the tumor, combined with CPA administration near the tumor, reduced glioma growth in all animals tested. Brain tumors in control mice with either NSPC-GFP or sham injection (tumor alone), and both with or without CPA infusion, were similar in size; by contrast, in mice treated with extratumoral inoculation of NSPC-CYP2B6 and CPA delivery, brain tumor volumes ranged from 5 to 50% that of the average volume in controls (Table 3), indicating that primary NSPC elicited a significant anti-glioma effect under conditions requiring tumor tracking.

Table 3. Inhibition of GL261 tumor growth with tumor-targeted CYP2B6/CPA therapy mediated by NSPC

Discussion

In this study, spontaneous fusion between NSPC and glioma cells expressing different fluorescent markers was observed under various conditions of in vitro co-culture. The uniform color coalescence of double-fluorescent cells suggested fusion as mechanism of formation rather than apoptotic cell engulfment. Unlike nonreplicating fused cells that arise in cultures of NSPC,22 NSPC-GBM cell hybrids underwent cell divisions, suggesting a considerable hybrid viability due to the contribution of tumor cells. In suspension co-cultures of NSPC with glioma cells, 2–3% of NSPC fused with tumor cells, as indicated by formation of a cell population with both NSPC and tumor fluorescent labels and high DNA content; the finding of a rate of fusion comparable to that observed in adherent co-culture implied that fusion was consequence of contact between both types of cells, not methodological artifacts of the co-culture system. Although it remains to be clarified whether human NSPC would behave differently, tumor cells of both mouse and human origin and with or without retroviral vector rendered similar results. Evidence of tumor cell fusion in humans after receiving allogeneic transplants of hematopoietic stem cells has been presented, and tumor cell fusion documented as a mechanism that possibly underlies tumor progression.23 The rationale of the CYP2B6/CPA therapy used here for enzymatic prodrug oxidation into diffusible cytotoxic derivatives, which incidentally would imply direct exposure of putative sporadic hybrids to cytotoxic anticancer activity, was that intact NSPC expressing CPA-converting enzyme are entirely responsible for prodrug activation. An advantage of CPA activation over other prodrug activating therapies is that transfer of drug metabolites does not necessarily require cell–cell contact,24 and the killing effect on both cells containing CYP2B6 and tumor cells lacking CYP2B6 (by-stander effect) occurs regardless of their proliferative phase.19

Using NSPC as vehicle for cytochrome P450 therapy, the growth of intracerebral high-grade gliomas, a type of tumor previously refractory to some extent to CPA activation treatment mediated by cells,4 was significantly reduced by administration of CPA, in agreement with the cytotoxic effect in vitro previously reported for these cells on both neurospheres and GBM cells.21 NSPC selectively migrated through brain parenchyma in the direction of glioma, and both after local and distant delivery of NSPC carrying the therapeutic gene, all animals treated exhibited reduced intracerebral tumor masses, with tumor growth inhibition of nearly 90% when locally delivered. Transplanted NSPC maintained a stem-like phenotype for few days and, while keeping the ability for transgene expression, moved through the recipient tissue directed to sites of tumor growth. Both VEGF⁺/HGF⁺ and VEGF⁺/HGF⁻ gliomas attracted NSPC. Together with other molecules whose upregulation is associated to tumor growth,10, 25 VEGF and HGF act as motility factors for NSPC,8–10 the latter being identified as NSPC chemoattractant of highest potency.10, 26 In agreement with this, we found that NSPC chemokinesis induced by U87MG was blocked by selective inhibition of c-met receptor on NSPC (data not shown). However, neither intracerebral migration of NSPC nor their directional migration towards tumors could be related to synthesis of HGF in tissue. Under our experimental conditions, diffusion of VEGF from brain tumors could explain tumor-oriented migration of NSPC. Also, NSPC showed tendency to migrate along perivascular spaces of the brain parenchyma that release high amount of VEGF from endothelial cells. Therefore, the close correspondence between migration of NSPC, directed to either gliomas or rims of inflammatory response, with local patterns of expression of VEGF, supported the idea that VEGF plays a major role as motility factor for NSPC guidance in the brain.

In summary, intracerebral inoculation of primary NSPC for enzyme/prodrug therapy effectively inhibited the growth of aggressive high-grade gliomas, which might in part be due to efficient targeting of tumor cells by diffusible cytotoxic derivatives. Taken together, data presented here indicate that, as result of their tumor-selective migratory properties, primary NSPC have broad potential for cell-based therapeutics of glioma.

Acknowledgements

We thank B. Hanks for flow cytometry analysis and L. Owen for critical reading of the manuscript.