Interaction between human brain tumour biopsies and fetal rat brain tissue in vitro

Human Patient-Derived Brain Tumor Models to Recapitulate Ependymoma Tumor Vasculature

Despite in vivo malignancy, ependymoma lacks cell culture models, thus limiting therapy development. Here, we used a tunable three-dimensional (3D) culture system to approximate the ependymoma microenvironment for recapitulating a patient's tumor in vitro. Our data showed that the inclusion of VEGF in serum-free, mixed neural and endothelial cell culture media supported the in vitro growth of all four ependymoma patient samples. The growth was driven by Nestin and Ki67 double-positive cells in a putative cancer stem cell niche, which was manifested as rosette-looking clusters in 2D and spheroids in 3D. The effects of extracellular matrix (ECM) such as collagen or Matrigel superseded that of the media conditions, with Matrigel resulting in the greater enrichment of Nestin-positive cells. When mixed with endothelial cells, the 3D co-culture models developed capillary networks resembling the in vivo ependymoma vasculature. The transcriptomic analysis of two patient cases demonstrated the separation of in vitro cultures by individual patients, with one patient's culture samples closely clustered with the primary tumor tissue. While VEGF was found to be necessary for preserving the transcriptomic features of in vitro cultures, the presence of endothelial cells shifted the gene's expression patterns, especially genes associated with ECM remodeling. The homeobox genes were mostly affected in the 3D in vitro models compared to the primary tumor tissue and between different 3D formats. These findings provide a basis for understanding the ependymoma microenvironment and enabling the further development of patient-derived in vitro ependymoma models for personalized medicine.

Pro-inflammatory cytokines in cystic glioblastoma: A quantitative study with a comparison with bacterial brain abscesses. With an MRI investigation of displacement and destruction of the brain tissue surrounding a glioblastoma

Cystic glioblastomas are aggressive primary brain tumors that may both destroy and displace the surrounding brain tissue as they grow. The mechanisms underlying these tumors’ destructive effect could include exposure of brain tissue to tumor-derived cytokines, but quantitative cytokine data are lacking. Here, we provide quantitative data on leukocyte markers and cytokines in the cyst fluid from 21 cystic glioblastomas, which we compare to values in 13 brain abscess pus samples. The concentration of macrophage/microglia markers sCD163 and MCP-1 was higher in glioblastoma cyst fluid than in brain abscess pus; lymphocyte marker sCD25 was similar in cyst fluid and pus, whereas neutrophil marker myeloperoxidase was higher in pus. Median cytokine levels in glioblastoma cyst fluid were high (pg/mL): TNF-α: 32, IL-6: 1064, IL-8: 23585, tissue factor: 28, the chemokine CXCL1: 639. These values were not significantly different from values in pus, pointing to a highly pro-inflammatory glioblastoma environment. In contrast, levels of IFN-γ, IL-1β, IL-2, IL-4, IL-10, IL-12, and IL-13 were higher in pus than in glioblastoma cyst fluid. Based on the quantitative data, we show for the first time that the concentrations of cytokines in glioblastoma cyst fluid correlate with blood leukocyte levels, suggesting an important interaction between glioblastomas and the circulation. Preoperative MRI of the cystic glioblastomas confirmed both destruction and displacement of brain tissue, but none of the cytokine levels correlated with degree of brain tissue displacement or peri-tumoral edema, as could be assessed by MRI. We conclude that cystic glioblastomas are highly pro-inflammatory environments that interact with the circulation and that they both displace and destroy brain tissue. These observations point to the need for neuroprotective strategies in glioblastoma therapy, which could include an anti-inflammatory approach.

Glioblastoma Organoids: Pre-Clinical Applications and Challenges in the Context of Immunotherapy

Malignant brain tumors remain uniformly fatal, even with the best-to-date treatment. For Glioblastoma (GBM), the most severe form of brain cancer in adults, the median overall survival is roughly over a year. New therapeutic options are urgently needed, yet recent clinical trials in the field have been largely disappointing. This is partially due to inappropriate preclinical model systems, which do not reflect the complexity of patient tumors. Furthermore, clinically relevant patient-derived models recapitulating the immune compartment are lacking, which represents a bottleneck for adequate immunotherapy testing. Emerging 3D organoid cultures offer innovative possibilities for cancer modeling. Here, we review available GBM organoid models amenable to a large variety of pre-clinical applications including functional bioassays such as proliferation and invasion, drug screening, and the generation of patient-derived orthotopic xenografts (PDOX) for validation of biological responses in vivo. We emphasize advantages and technical challenges in establishing immunocompetent ex vivo models based on co-cultures of GBM organoids and human immune cells. The latter can be isolated either from the tumor or from patient or donor blood as peripheral blood mononuclear cells (PBMCs). We also discuss the challenges to generate GBM PDOXs based on humanized mouse models to validate efficacy of immunotherapies in vivo. A detailed characterization of such models at the cellular and molecular level is needed to understand the potential and limitations for various immune activating strategies. Increasing the availability of immunocompetent GBM models will improve research on emerging immune therapeutic approaches against aggressive brain cancer.

Engineered Hydrogels for Brain Tumor Culture and Therapy

Brain tumors' severity ranges from benign to highly aggressive and invasive. Bioengineering tools can assist in understanding the pathophysiology of these tumors from outside the body and facilitate development of suitable antitumoral treatments. Here, we first describe the physiology and cellular composition of brain tumors. Then, we discuss the development of three-dimensional tissue models utilizing brain tumor cells. In particular, we highlight the role of hydrogels in providing a biomimetic support for the cells to grow into defined structures. Microscale technologies, such as electrospinning and bioprinting, and advanced cellular models aim to mimic the extracellular matrix and natural cellular localization in engineered tumor tissues. Lastly, we review current applications and prospects of hydrogels for therapeutic purposes, such as drug delivery and co-administration with other therapies. Through further development, hydrogels can serve as a reliable option for in vitro modeling and treatment of brain tumors for translational medicine.

Role of Microenvironment in Glioma Invasion: What We Learned from In Vitro Models

The invasion properties of glioblastoma hamper a radical surgery and are responsible for its recurrence. Understanding the invasion mechanisms is thus critical to devise new therapeutic strategies. Therefore, the creation of in vitro models that enable these mechanisms to be studied represents a crucial step. Since in vitro models represent an over-simplification of the in vivo system, in these years it has been attempted to increase the level of complexity of in vitro assays to create models that could better mimic the behaviour of the cells in vivo. These levels of complexity involved: 1. The dimension of the system, moving from two-dimensional to three-dimensional models; 2. The use of microfluidic systems; 3. The use of mixed cultures of tumour cells and cells of the tumour micro-environment in order to mimic the complex cross-talk between tumour cells and their micro-environment; 4. And the source of cells used in an attempt to move from commercial lines to patient-based models. In this review, we will summarize the evidence obtained exploring these different levels of complexity and highlighting advantages and limitations of each system used.

Embedded Spheroids as Models of the Cancer Microenvironment

To more accurately study the complex mechanisms behind cancer invasion, progression, and response to treatment, researchers require models that replicate both the multicellular nature and 3D stromal environment present in an in vivo tumor. Multicellular aggregates (i.e., spheroids) embedded in an extracellular matrix mimic are a prevalent model. Recently, quantitative metrics that fully utilize the capability of spheroids are described along with conventional experiments, such as invasion into a matrix, to provide additional details and insights into the underlying cancer biology. The review begins with a discussion of the salient features of the tumor microenvironment, introduces the early work on non-embedded spheroids as tumor models, and then concentrates on the successes achieved with the study of embedded spheroids. Examples of studies include cell movement, drug response, tumor cellular heterogeneity, stromal effects, and cancer progression. Additionally, new methodologies and those borrowed from other research fields (e.g., vascularization and tissue engineering) are highlighted that expand the capability of spheroids to aid future users in designing their cancer-related experiments. The convergence of spheroid research among the various fields catalyzes new applications and leads to a natural synergy. Finally, the review concludes with a reflection and future perspectives for cancer spheroid research.

Evaluation of Poly (Glycerol-Adipate) Nanoparticle Uptake in an In Vitro 3-D Brain Tumor Co-Culture Model

Despite the inherent problems associated with in vivo animal models of tumor growth and metastases, many of the current in vitro brain tumor models also do not accurately mimic tumor-host brain interactions. Therefore, there is a need to develop such co-culture models to study tumor biology and, importantly, the efficacy of drug delivery systems targeting the brain. So far, few investigations of this nature have been published. In this paper we describe the development of a new model system and its application to drug delivery assessment. For our new model, a co-culture of DAOY cell brain tumor aggregates and organo-typic brain slices was developed. Initially, the DAOY aggregates attached to cerebellum slices and invaded as a unit. Single cells in the periphery of the aggregate detached from the DAOY aggregates and gradually replaced normal brain cells. This invasive behavior of DAOY cells toward organotypic cerebellum slices shows a similar pattern to that seen in vivo. After validation of the co-culture model using transmission electron microscopy, nanoparticle (NP) uptake was then evaluated. Confocal micrographs illustrated that DAOY cells in this co-culture model took up most of the NPs, but few NPs were distributed into brain cells. This finding corresponded with results of NP uptake in DAOY and brain aggregates reported elsewhere.

Spherical cancer models in tumor biology

Three-dimensional (3D) in vitro models have been used in cancer research as an intermediate model between in vitro cancer cell line cultures and in vivo tumor. Spherical cancer models represent major 3D in vitro models that have been described over the past 4 decades. These models have gained popularity in cancer stem cell research using tumorospheres. Thus, it is crucial to define and clarify the different spherical cancer models thus far described. Here, we focus on in vitro multicellular spheres used in cancer research. All these spherelike structures are characterized by their well-rounded shape, the presence of cancer cells, and their capacity to be maintained as free-floating cultures. We propose a rational classification of the four most commonly used spherical cancer models in cancer research based on culture methods for obtaining them and on subsequent differences in sphere biology: the multicellular tumor spheroid model, first described in the early 70s and obtained by culture of cancer cell lines under nonadherent conditions; tumorospheres, a model of cancer stem cell expansion established in a serum-free medium supplemented with growth factors; tissue-derived tumor spheres and organotypic multicellular spheroids, obtained by tumor tissue mechanical dissociation and cutting. In addition, we describe their applications to and interest in cancer research; in particular, we describe their contribution to chemoresistance, radioresistance, tumorigenicity, and invasion and migration studies. Although these models share a common 3D conformation, each displays its own intrinsic properties. Therefore, the most relevant spherical cancer model must be carefully selected, as a function of the study aim and cancer type.

Laboratory models for central nervous system tumor stem cell research

Central nervous system (CNS) tumors are complex organ systems comprising of a neoplastic component with associated vasculature, inflammatory cells, and reactive cellular and extracellular components. Research has identified a subset of cells in CNS tumors that portray defining properties of neural stem cells, namely, that of self-renewal and multi-potency. Growing evidence suggests that these tumor stem cells (TSC) play an important role in the maintenance and growth of the tumor. Furthermore, these cells have also been shown to be refractory to conventional therapy and may be crucial for tumor recurrence and metastasis. Current investigations are focusing on isolating these TSC from CNS tumors to investigate their unique biological processes. This understanding will help identify and develop more effective and comprehensive treatment strategies. This chapter provides an overview of some of the most commonly used laboratory models for CNSTSC research.

Toward 3D biomimetic models to understand the behavior of glioblastoma multiforme cells

Glioblastoma multiforme (GBM) tumors are one of the most deadly forms of human cancer and despite improved treatments, median survival time for the majority of patients is a dismal 12-15 months. A hallmark of these aggressive tumors is their unique ability to diffusively infiltrate normal brain tissue. To understand this behavior and successfully target the mechanisms underlying tumor progression, it is crucial to develop robust experimental ex vivo disease models. This review discusses current two-dimensional (2D) experimental models, as well as animal-based models used to examine GBM cell migration, including their advantages and disadvantages. Recent attempts to develop three-dimensional (3D) tissue engineering-inspired models and their utility in unraveling the role of microenvironment on tumor cell behaviors are also highlighted. Further, the use of 3D models to bridge the gap between 2D and animal models is explored. Finally, the broad utility of such models in the context of brain cancer research is examined.

αB-crystallin is elevated in highly infiltrative apoptosis-resistant glioblastoma cells

We have previously established two distinct glioma phenotypes by serial xenotransplantation of human glioblastoma (GBM) biopsies in nude rats. These tumors undergo a gradual transition from a highly invasive nonangiogenic to a less-invasive angiogenic phenotype. In a protein screen to identify molecular markers associated with the infiltrative phenotype, we identified α-basic-crystallin (αBc), a small heat-shock protein with cytoprotective properties. Its increased expression in the infiltrative phenotype was validated by immunohistochemistry and Western blots, confirming its identity to be tumor-derived and not from the host. Stereotactic human GBM biopsies taken from MRI-defined areas verified stronger αBc expression in the infiltrative edge compared to the tumor core. Cell migration assays and immunofluorescence staining showed αBc to be expressed by migrating cells in vitro. To determine αBc function, we altered its expression levels. αBc siRNA depletion caused a loss of migrating tumor cells from biopsy spheroids and delayed monolayer wound closure. In contrast, glioma cell migration in a Boyden chamber assay was unaffected by either αBc knockdown or overexpression, indicating that αBc is not functionally linked to the cell migration machinery. However, after siRNA αBc depletion, a significant sensitization of cells to various apoptotic inducers was observed (actinomycin, tumor necrosis factor α, and TNF-related apoptosis-inducing ligand [TRAIL]). In conclusion, αBc is overexpressed by highly migratory glioma cells where it plays a functional role in apoptosis resistance.

Three-dimensional Gaussian model to define brain metastasis limits on 11C-methionine PET

PURPOSE

Since 11C-methionine (MET) heavily accumulates in brain tumors, PET with MET (MET-PET) is proposed for the image-guided planning of their targeted therapy. Determination of bulk tumor limits is therefore a crucial component of MET-PET image analysis. We aimed at validating a Gaussian model of tumor delineation on MET-PET. We choose MET-PET and MRI data obtained in brain metastases to adjust the model. Indeed, MRI limits of these non-infiltrative hypermetabolic brain lesions are efficiently used for their curative treatment.

METHODS AND MATERIALS

We developed a three-dimensional (3D) Gaussian model that relates the tumor-limit-defining threshold to maximum and mean count values in the defined tumor volume and to mean count values in a reference region. To adjust the model to experimental data, we selected 25 brain metastases following these criteria: (i) no surgery or classical radiotherapy within 6 months, (ii) no previous radiosurgery, (iii) MET-PET and MRI acquired within a 48-h interval, (vi) necrosis representing less than 25% of tumor volume on MRI. We applied a progressive thresholding procedure on MET-PET so as to match tumor limits on contrast-enhanced co-registered MRI.

RESULTS

In 22 tumors, a match could be reached between tumor margins on MET-PET and MRI. The relation between mean, maximum and threshold values closely fits the 3D-Gaussian model function. We found a quadratic relation between the mean-to-threshold ratio and the maximum-to-cerebellum activity ratio.

CONCLUSIONS

A 3D-Gaussian model may describe the limits of MET uptake distribution within brain metastases, providing a simple method for metabolic tumor delineation.

Adeno-associated virus (AAV) serotypes 2, 4 and 5 display similar transduction profiles and penetrate solid tumor tissue in models of human glioma

BACKGROUND

Adeno-associated viral (AAV) vectors are potent delivery vehicles for gene transfer strategies directed at the central nervous system (CNS), muscle and liver. However, comparatively few studies have described AAV-mediated gene transfer to tumor tissues. We have previously demonstrated that while AAV2 and Adenoviral (Ad) 5 vectors have similar broad host ranges in tumor-derived cell lines, AAV2 was able to penetrate human glioblastoma biopsy spheroids and xenografts more efficiently than Ad 5 vectors. These results suggested that AAV vectors could be suitable for therapeutic gene delivery to solid tumor tissue. In the present work, the transduction efficacy of AAV serotypes 4 and 5 were compared to AAV2, both in vitro and in intracranial GBM xenografts derived from patient biopsies implanted into nude rats.

METHODS

AAV vector serotypes 2, 4, and 5 containing either the green fluorescent protein (GFP) or the bacterial beta-galactosidase (lacZ) reporter gene were added to five different human glioma cell lines, to multicellular spheroids generated from glioblastoma patient biopsies, and to spheroids xenografted intracranially in nude rats. Transduction efficiency was assessed by fluorescence imaging, histochemistry, immunohistochemistry and flow cytometry.

RESULTS

While all three AAV serotypes were able to transduce the glioma cell lines when added individually or when they were administered in concert, AAV2 transduced the glioma cells most effectively compared to AAV4 or AAV5. Upon infecting glioblastoma spheroids in vitro, all three AAV serotypes efficiently transduced cells located at the surface as well as within deeper layers of the spheroids. In addition, similarly to what was observed for AAV2 16, both AAV4 and AAV5 were able to transduce human glioblastoma xenografts implanted intracranially.

CONCLUSIONS

In addition to the widely used AAV2 serotype, AAV4 and AAV5 serotypes may also be used to transduce biologically diverse glioma cell lines. They also penetrate and transduce solid human tumor tissue derived from patient biopsies. Therefore, the data presented here provide a proof of principle for developing AAV4 and AAV5 as treatment vehicles for human malignant gliomas.

Src Regulates Actin Dynamics and Invasion of Malignant Glial Cells in Three Dimensions

Malignant glioma is the major brain tumor in adults and has a poor prognosis. The failure to control invasive cell subpopulations may be the key reason for local glioma recurrence after radical tumor resection and may contribute substantially to the failure of the other treatment modalities such as radiation therapy and chemotherapy. As a model for this invasion, we have implanted spheroids from a human glioma cell line (U251) in three-dimensional collagen type I matrices, which these cells readily invade. We first observed that the Src family kinase-specific pharmacologic inhibitors PP2 and SU6656 significantly inhibited the invasion of the cells in this assay. We confirmed this result by showing that expression of two inhibitors of Src family function, dominant-negative-Src and CSK, also suppressed glioma cell invasion. To characterize this effect at the level of the cytoskeleton, we used fluorescent time-lapse microscopy on U251 cells stably expressing a YFP-actin construct and observed a rapid change in actin dynamics following addition of PP2 in both two-dimensional and three-dimensional cultures. In monolayer cultures, PP2 caused the disappearance of peripheral membrane ruffles within minutes. In three-dimensional cultures, PP2 induced the loss of actin bursting at the leading tip of the invadopodium. The inhibition of Src family activity is thus a potential therapeutic approach to treat highly invasive malignant glioma.

Evidence for a secreted chemorepellent that directs glioma cell invasion

Secreted chemotropic cues guide the migration of neuronal and glial cell precursors during neural development. It is not known if chemotropism contributes to directing the invasion of brain tissue by glioma cells. A model system has been developed that allows quantification of invasive behavior using gliomas spheroids embedded in collagen gels. Here we provide evidence that glioma spheroids secrete a chemorepellent factor(s) that directs cells away from the spheroid and into the collagen matrix. The relationship between total invasion, cell number, and implantation distance suggests that glioma cells respond to a gradient of the chemorepellent cue(s) that is well established at 48 h. C6 astrocytoma cells normally invade the collagen at an angle perpendicular to the spheroid edge. In contrast, an adjacent spheroid causes cells to turn away from their normal trajectory and slow their rate of invasion. Astrocytoma cells are repelled by an adjacent glioma spheroid but rapidly infiltrate astrocyte aggregates, indicating that astrocytes do not express the repellent cue. Uniform concentrations of repellent factor(s) in spheroid conditioned medium overwhelm endogenous gradients and render glioma cells less able to exhibit this chemotropic response. Concentration gradients of spheroid conditioned medium in cell migration assays also demonstrate the chemorepellent cue(s)'s tropic effect. Our findings indicate that glioma spheroids produce a secreted diffusible cue(s) that promotes glioma cell invasion. Identification of this factor(s) may advance current therapies that aim to limit tumor cell invasion.

Modulation of glioma cell invasion and motility by adenoviral gene transfer of PAI-1

A number of studies have emphasized the role of PAI-1 as an important regulator of tumor cell invasion and metastasis. The hallmark of primary tumors of the central nervous system and glioblastomas in particular is the diffuse invasion into the normal brain tissue. Since PAI-1 is expressed in such tumors, we studied the effect of adenoviral-mediated transfer of the PAI-1 gene in regulating the in vitro invasiveness of D54Mg glioma cells into Matrigel, and into fetal rat brain aggregates. Treatment of D54Mg cells with 50 MOI (multiplicity of infection) of the replication defective vector AdCMVPAI-1 increased PAI-1 expression 23-fold compared to control vectors, and the invasion through Matrigel was reduced by 67%. The motility of the cells was reduced by 58% compared to controls (indicating that inhibition of motility was the principal effect of PAI-1 in these cells). The ability of D54Mg tumor spheroids to invade fetal rat brain aggregates was not reduced by the PAI-1 gene transfer. The results show that overexpression of PAI-1 can inhibit glioma cell motility and invasion through extracellular matrix (ECM) components, like laminin and collagen, but does not inhibit tumor cell invasion in a three-dimensional invasion assay, simulating normal brain tissue having a different ECM and interstitial composition. The different results obtained in the two invasion assays reflect the complex biological effects of the uPA/PAI-1 system, and questions a simplistic view of PAI- I as an inhibitor of brain tumor invasion.

Model systems in neurooncology

Cell- and molecular biological techniques have had a major impact on experimental neurooncology in recent years; yet we are lacking suitable model systems. Monolayer cell cultures are rapid, reproducible and reliable systems, however, their validity is of major concern. Three dimensional culture systems, especially derived from primary biopsies, match better with the in vivo situation albeit being more tricky to handle. Animal models for glioma have to be orthotopic in order to draw any conclusions; most cell lines implanted into rodents still do not show the typical invasive phenotype. In addition, immunological phenomena have to be taken into account as well as changes of the biological features once cells have undergone the process of any transfection.

Invasion of human glioma biopsy specimens in cultures of rodent brain slices: a quantitative analysis

OBJECT

The reliable assessment of the invasiveness of gliomas in vitro has proved elusive, because most invasion assays inadequately model in vivo invasion in its complexity. Recently, organotypical brain cultures were successfully used in short-term invasion studies on glioma cell lines. In this paper the authors report that the invasiveness of human glioma biopsy specimens directly implanted into rodent brain slices by using the intraslice implantation system (ISIS) can be quantified with precision. The model was first validated by the demonstration that, in long-term studies, established glioma cells survive in the ISIS and follow pathways of invasion similar to those in vivo.

METHODS

Brain slices (400 microm thick) from newborn mice were maintained on millicell membranes for 15 days. Cells from two human and one rodent glioblastoma multiforme (GBM) cell lines injected into the ISIS were detected by immunohistochemistry or after transfection with green fluorescent protein-containing vectors. Preferential migration along blood vessels was identified using confocal and fluorescent microscopy. Freshly isolated (< or = 24 hours after removal) 1,1'-dioctadecyl-3,3,3',3'-tetramethylindocarbocyanine perchlorate-prelabeled human glioma biopsy specimens were successfully implanted in 19 (83%) of 23 cases, including 12 GBMs and seven lower grade gliomas (LGGs). Morphometric quantification of distance and density of tumor cell invasion showed that the GBMs were two to four times more invasive than the LGGs. Heterogeneity of invasion was also observed among GBMs and LGGs. Directly implanted glioma fragments were more invasive than spheroids derived from the same biopsy specimen.

CONCLUSIONS

The ISIS combines a high success rate, technical simplicity, and detailed quantitative measurements and may, therefore, be used to study the invasiveness of biopsy specimens of gliomas of different grades.

Adeno-associated viral vectors penetrate human solid tumor tissue in vivo more effectively than adenoviral vectors

The transduction efficiencies of adeno-associated viral vectors (AAV, serotype 2) and adenovirus vectors (ADV, serotype 5) were examined in three different models of cancer. First, we used flow cytometry to quantitate AAV-GFP or ADV-GFP transduction in 13 cell lines derived from malignant tissue (6 gliomas, 6 mammary cancers, and 1 leukemia). These experiments showed variable transduction efficiency (0%-81%) between the cell lines, with ADV being more effective compared to AAV in 9 of 13 cell lines. Second, spheroids prepared from human glioblastomas were infected with ADV or AAV expressing GFP or lacZ cassettes, and after 2 weeks, uniform reporter gene expression was observed on the spheroid. Whereas AAV produced consistent transduction throughout the spheroids, ADV infection was mainly limited to the outer cell layers of the spheroids, suggesting that AAV were more efficient at penetrating solid tumor tissue. Third, human biopsies from glioblastoma multiforme patients were xenografted into nude rats and grown for 4 weeks followed by viral vector injection. Combined use of high-resolution magnetic resonance imaging (MRI) and histologic analysis allowed the identification of transduced cells and their spatial distribution within the tumors. AAV-mediated transgene expression was observed in cell clusters through the entire tumor, while ADV-mediated transduction was restricted to cells at the tumor periphery. Thus, while AAV and ADV vectors may infect tumor-derived cell lines to a similar degree, AAV penetrated glioblastoma spheroids and xenografts more efficiently compared to ADV vectors. These results suggest that AAV may be suitable for therapeutic gene delivery to malignant tumors.

Tracking the invasiveness of human astrocytoma cells by using green fluorescent protein in an organotypical brain slice model

OBJECT

Although it is known that malignant astrocytomas infiltrate diffusely into regions of normal brain, it is frequently difficult to identify unequivocally the solitary, invading astrocytoma cell in histopathological preparations or experimental astrocytoma models. The authors describe an experimental system that facilitates the tracking of astrocytoma cells by using nonneoplastic cerebral tissue as the substrate for invasion.

METHODS

Cerebral tissue was cut into 1-mm-thick slices and cultured in the upper chamber of a Transwell culture dish on top of a polyester membrane (0.4-mm pore size) that was bathed in medium supplied by the lower chamber. Two astrocytoma cell lines, U-87 MG (U87) and U343 MG-A (U343), were selected because of their differing basal cell motilities in monolayer cultures. The astrocytoma cells were stably transfected with vectors that expressed green fluorescent protein (GFP), either alone or as a fusion protein with the receptor for hyaluronic acid-mediated motility (RHAMM) in either sense or antisense orientations. Stably transfected clones that had high levels of GFP expression were selected using the direct visualization provided by fluorescence microscopy and fluorescence-activated cell-sorter analysis. The GFP-expressing astrocytoma cell clones were implanted into the center of the brain slice and the degree of astrocytoma invasion into brain tissue was measured at different time points by using the optical sectioning provided by the confocal laser microscope. The authors observed that GFP-expressing astrocytoma cells could be readily tracked and followed in this model system. Individual astrocytoma cells that exhibited green fluorescence could be readily identified following their migration through the brain slices. The GFP-labeled U87 astrocytoma cells migrated farther into the brain slice than the U343 astrocytoma cells. The RHAMM-transfected GFP-labeled astrocytoma cells also infiltrated farther than the GFP-labeled astrocytoma cells themselves. The expression of antisense RHAMM virtually abrogated the invasion of the brain slices by both astrocytoma cell lines.

CONCLUSIONS

The authors believe that this organotypical culture system may be of considerable utility in studying the process of astrocytoma invasion, not only because it provides a better representation of the extracellular matrix molecules normally encountered by invading astrocytoma cells, but also because the GFP tag enables tracking of highly migratory and invasive astrocytoma cells under direct vision.

Spheroids: relation between tumour and endothelial cells

It has recently been established that the microenvironment plays a major role in many physiological and pathological events. Indeed cell-cell and cell-extracellular matrix contacts are necessary for much cellular function such as differentiation, proliferation, cell death, apoptosis and angiogenesis. For growth, proliferating tumour cells need to be fed by nutrients and oxygen brought by new vessels. In this context, scientists seek a new model that allows for the investigation of both angiogenesis and the influence of the microenvironment on this phenomenon. The purpose of this paper is to review the literature on the relation between tumour and endothelial cells grown as spheroids, a technique that allows us to study in three-dimensions the influence of cell contact on this growth. For the purpose of clarification, this review has recategorised the different studies on spheroids into three classes: (1) spheroids grown in vitro and then reimplanted in animals to follow endothelial cell infiltration; (2) spheroids grown in vitro and then cultured on endothelial cell monolayers; (3) tumours grown in vitro such as organotypic culture. This review attempts to demonstrate that spheroid cell cultures are useful for studying the relation between tumour and endothelial cells and to analyse physiological phenomena such as wound healing, extravasation and intravasation.

Retinoids inhibit human glioma cell proliferation and migration in primary cell cultures but not in established cell lines

OBJECTIVE

Retinoids are known to exhibit a broad spectrum of biological activities, and they participate in the onset of differentiation and the inhibition of growth in a wide variety of cancer cells. Some of these vitamin A derivatives are already in clinical use. However, data on retinoid actions in glial tumors are rather sparse. Therefore, we studied the effects of the natural retinoic acid (RA) forms all-trans-RA, 9-cis-RA, and 13-cis-RA on glioma cell lines and primary cultures from patients with glioblastomas multiforme.

METHODS

Six human glioma cell lines, one rat glioma cell line, and 20 primary cultures established from biopsies from patients with glioblastomas multiforme were investigated. Tumor cell proliferation was assessed using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide and cell-counting assays. Random migration out of tumor spheroids was quantified using a video-morphometry system. Invasion was investigated using a confrontational coculture test system. Retinoid receptor (RA receptor [RAR]alpha, -beta, and -gamma and retinoid X receptor [RXR]alpha, -beta, and -gamma) expression status was determined using reverse transcription-polymerase chain reaction studies.

RESULTS

Treatment of five human glioma cell lines with the different retinoids at concentrations up to 10(-5) mol/L produced no reduction of proliferation, using various incubation times. For one human glioma cell line (U343MG-A) and one rat glioma cell line (C6), which were previously reported to be sensitive to retinoids, we could confirm strong inhibitory effects on proliferation and clear changes in morphological features after retinoid treatment. Application of the different retinoids to low-passage primary cultures of human glioblastomas resulted in marked inhibition of proliferation (30-95%) for all tested samples. Using three-dimensional spheroid cultures, we detected retinoid-induced decreases in cell migration (24-65%). Invasion was not affected by these vitamin A derivatives. In an analysis of the expression patterns for retinoid receptors (RARs and RXRs), all primary culture samples yielded positive results for RAR gamma and RXR alpha and negative results for RAR alpha, RAR beta, and RXR gamma, whereas the results of RXR beta expression were heterogeneous among different patients. The cell lines, irrespective of their RA sensitivities, did not exhibit any major differences in receptor expression.

CONCLUSION

Retinoids strongly inhibit proliferation and migration in primary cultures of human glioblastomas multiforme. Our data support a clinical trial of retinoids for the treatment of human malignant gliomas. We observed that most established cell lines were not sensitive to RA. This difference between long-term cell lines and primary cultures cannot be explained by different retinoid receptor expression patterns.

Transforming growth factor-beta1 stimulates contraction of human glioblastoma cell-mediated collagen lattice through enhanced alpha2 integrin expression

Rapid invasiveness is a feature of the highly malignant glioblastoma tumor and is closely related to patient prognosis. The interaction between extracellular matrix (ECM) and cell surface receptors such as integrin heterodimers play a key role in the process of tumor invasion. We investigated the effects of transforming growth factor-beta1 (TGF-beta1), which is a mitogenic factor for glial cells, on integrin expression in T98G human glioblastoma cells using an in vitro model 3-dimensional collagen lattice. Exogenously applied TGF-beta1 dose-dependently enhanced collagen lattice contraction. Among the inhibitory antibodies tested against alpha integrin subunits, the anti-alpha2 antibody, P1-E6, alone prevented the enhanced contractile response by TGF-beta1, whereas any alpha integrin antibody (including P1-E6) had little effect on lattice contraction when cultured without TGF-beta1. RT-PCR analysis revealed that TGF-beta1 strongly increased alpha2 integrin transcript level. Furthermore, pretreatment with antisense phosphorothioate oligodeoxynucleotides against human alpha2 integrin using hemagglutinating virus of Japan (HVJ) liposome-mediated transfer prevented the effects of TGF-beta1 and also reduced the lattice contraction even in the absence of TGF-beta1. This data indicates that increased expression of alpha2 integrin is responsive to enhanced collagen lattice contraction by TGF-beta1. We suggest that TGF-beta1 exerts its effects on the invasive property of glioblastoma cells via upregulation of the alpha2 integrin subunit expression.

Dissecting glioma invasion: interrelation of adhesion, migration and intercellular contacts determine the invasive phenotype

The invasive cellular behavior of malignant gliomas is determined by receptor mediated cell-substratum contacts and cell-cell interaction as well as cellular locomotion. This study attempts to break down the complex phenomena of the invasive process into their components of attachment to neighboring cells, aggregate formation, adhesion to matrix substratum, migration and invasion into three-dimensional cellular aggregates separately analyzed in different in vitro assay systems. Using a panel of 13 glioma cell lines, adhesion to non-specifically or merosin coated surfaces was correlated to monolayer cell migration and dissemination of tumor cells from aggregates plated on these substrates. The formation kinetics of aggregates were determined and compared to the ability of these cells to rapidly attach and form mechanically stable cell-cell contacts. The motility rates in the different assay systems as well as cell-cell attachment was correlated to invasion of re-aggregated tumor cells into fetal rat brain. A tight positive correlation was found for substrate adhesion and monolayer migration. In contrast, cell-substratum contacts had little influence on dissemination of cells out of three-dimensional aggregates and no association between monolayer migration and migration of cells out of aggregates was detected. The ability of glioma cells to rapidly form aggregates was associated with enhanced migration out of aggregates. The capacity to invade fetal rat brain aggregates was correlated with the capacity to form stable intercellular adhesion as measured in a cell-cell adhesion assay. Invasion in this system was not found to be associated with migration in monolayer or with migration out of tumor aggregates. This study highlights that current in vitro assays for invasion only represent isolated aspects of the multi-cascade process which is involved in tumor cell invasion.

Inhibition of phospholipase C-gamma1 activation blocks glioma cell motility and invasion of fetal rat brain aggregates

OBJECTIVE

Phospholipase C (PLC)-gamma is a cytosolic enzyme activated by several growth factor (GF) receptors (epidermal GF receptor [EGFR], platelet-derived GF receptor, and insulin-like GF 1 receptor), and its activation is associated with increased cell motility (but not cell proliferation) in nonglioma cell lines. Because up-regulated activation of EGFR has been consistently linked to poor patient survival in patients with glioblastoma multiforme (GBM) and because inhibition of EGFR activation by tyrosine kinase inhibitors prevents glioma infiltration in vitro, we hypothesized that inhibition of PLC-gamma activation would inhibit glioma cell invasiveness.

METHODS

Our experimental model assesses tumor spheroid invasion of fetal rat brain spheroids by confocal microscopy. We treated U87 GBM spheroids, and those derived from a single patient, with the PLC inhibitor U73122. We also transfected rat C6 glioma cells with the PLCz complementary deoxyribonucleic acid coding for a dominant negative PLC-gamma1 src-homology-2/src-homology-3 peptide fragment, which blocks binding and activation of PLC-gamma1 by GF receptors. Two clones (C6F and C6E) were grown into spheroids and were tested for invasiveness in the spheroid model and for responsiveness to GFs in a standard in vitro motility assay.

RESULTS

The infiltration rate of the patient GBM cell line overexpressing wild-type EGFR was reduced by 2 micromol/L U73122 from a slope (percent invasion/h) of 0.74+/-0.08 (with the inactive congener U73343) to 0.04+/-0.053 (P = 8 x 10(-7) by two-tailed t test, 92% reduction); the integral rate, another measure of invasion, was reduced from 49.7+/-13 percent-hours per hour to 13.6+/-12 (P = 0.002, 72% reduction). The U87 spheroid invasion rate was reduced by 0.5 micromol/L U73122 from 46.7+/-8.5 percent-hours per hour to 11.2+/-4.6 (P = 3 x 10(-5)); the slope decreased from 1.7+/-0.41 percent per hour to 0.35+/-0.14 (P = 0.0001). The C6F and C6E clones demonstrated attachment to and "surrounding" of the fetal rat brain aggregate but no true invasion by confocal or light microscopy. PLCz blocked the motility response to epidermal GF, platelet-derived GF, and insulin-like GF. There was a significant decrease in PLC-gamma1-associated tyrosine phosphorylation.

CONCLUSION

These results support a key role for PLC-gamma activation as a common postreceptor pathway for GF-induced tumor infiltration and further identify PLC-gamma1 as a possible target for anti-invasive therapy for GBMs.

Growth of precultured human glioma specimens in nude rat brain

OBJECT

The aim of this study was to develop an improved animal model for brain tumor study. The need for better and more relevant brain tumor models is generally acknowledged. Glioma tissue can be cultured directly from the biopsy specimen as tumor spheroids. Using such precultured tissue, a new in vivo model for studying human gliomas was established.

METHODS

Precultured small tumor spheroids (< 300 microm) prepared from cell lines or tumor biopsy fragments were injected into the brains of immunodeficient rats by using a 5-microl Hamilton syringe that had a piston in the needle. Tumors could be established by injecting a single spheroid derived from the U-87MG cell line, whereas inoculation of 10 spheroids resulted in a tumor take comparable to that attained with injection of 10(6) single cells. Biopsy specimens obtained from six patients who underwent surgery for glioblastoma multiforme were cultured as organotypic spheroids for 11 to 18 days before inoculation into the rats. The animals were killed 3 months after spheroid implantation. Microscopic examination revealed tumor growth in 87.5 to 100% of the animals inoculated with tumor spheroids from all but one of the tumor biopsy specimens. Extensive invasion and cell migration along the nerve tracts of the corpus callosum was found in tumors that originated from four of the six biopsy specimens.

CONCLUSIONS

This approach, in which spheroids from precultured biopsy specimens are injected into the brains of immunodeficient animals, provides new means for experimental studies of human malignant brain tumors in a clinically relevant animal model.

Dynamic determination of human glioma invasion in vitro

OBJECT

The goal of this study was to evaluate whether there is any relationship between survival of patients with brain tumor and tumor proliferation or tumor invasion in vitro.

METHODS

Samples of freshly resected brain tumors from 14 patients with glioblastoma multiforme (GBM) were directly grown as three-dimensional multicellular spheroids. The tumor spheroids were cocultured with fetal rat brain cell aggregates (BCAs), used to represent an organotypical normal brain tissue model. Before the coculture, the tumor spheroids and the BCAs were stained with two different carbocyanine dyes, 1,1'-dioctadecyl-3,3,3',3'-tetramethylindocarbocyanine perchlorate (DiI) and 3,3'-dioctadecycloxacarbocyanine perchlorate (DiO), respectively. During the coculture, confocal laser scanning microscopy allowed a sequential analysis of tumor cell invasion by visualizing dynamic aspects of the invasive process. Single cocultures were examined at three different time points (24, 48, and 96 hours). During the observation period there was a change in the structural morphology of the cocultures, with a progressive decrease in BCA volume. Furthermore, the scanning confocal micrographs revealed a bidirectional movement of tumor cells and normal cells into brain and tumor tissue, respectively. It is also shown that there is a considerable variation in the rate of BCA destruction in cocultures of glioma spheroids generated directly from biopsy specimens. This variation is seen both between spheroids generated from the same biopsy as well as between spheroids that are grown from different biopsy specimens. Cell proliferation measured by Ki-67 immunohistochemical analysis of biopsy samples obtained in the same patients revealed a correlation between tumor cell proliferation and tissue destruction of the BCAs, as determined by a reduction in BCA volume (p = 0.0338). No correlation was found when survival was related to the same parameters (p > 0.05).

CONCLUSIONS

The present work provides a model for quick and efficient assessment of dynamic interactions between tumor and normal brain tissue shortly after surgery.

Connexin transfection induces invasive properties in HeLa cells

Gap-junctional coupling of tumor cells facilitates their invasion into normal tissue as was further investigated with connexin-transfected HeLa cells by an in vitro assay. Wild-type HeLa cells are coupling deficient and did not show invasive properties when multicell spheroids were confronted with precultured embryonic chicken heart fragments. After transfection with cDNA of Cx31, Cx40, or Cx43, these cells were homotypically coupled; Cx40- and Cx43-transfected HeLa cells also were heterotypically coupled with embryonic chicken heart cells in monolayer coculture. Transfected clones revealed invasive properties in the chicken heart in vitro assay. However, the pattern of invasion differed between these transfectants: After 4 days Cx43-transfected HeLa cells were found in the central part of heart spheroids; Cx40- and Cx31-transfected HeLa cells, however, did not invade the central core but were detected in the outer part of heart spheroids. We conclude that connexin expression supports invasion of tumor cells into normal tissue, but heterotypic gap-junctional coupling is not required for this process.

Expression of annexin II in glioma cell lines and in brain tumor biopsies

Annexin II is a calcium and phospholipid binding protein and a substrate for protein-tyrosine kinases. Increased levels of annexin II are observed in various cancer cells and tissues, and the molecule has been proposed as a marker of malignancy in vivo. Annexin II was expressed in four glioma cell lines (D-54MG, D-37MG, U251MG and GaMG), as determined by Western blot analyses, immunofluorescence staining and flow cytometric measurements. In addition, annexin II expression was also found in cryostat sections obtained from 15 consecutive brain tumor biopsies: Ten were histologically classified as glioblastomas, one as an astrocytoma, two as meningiomas and two as brain metastases. Cultured spheroids from the glioma cell lines and from three of the glioblastoma biopsies showed lower levels of annexin II, than found in the monolayers of the cell lines and in the freshly cut biopsies. The annexin II expression of the cell lines were not found to be related to their proliferative, migratory or invasive properties. These findings indicate that although annexin II may serve as a marker of malignancy in vivo, its expression can be reduced in vitro, and appear unrelated to malignant features of glioma cell lines.

Four-dimensional analysis of human brain tumor spheroid invasion into fetal rat brain aggregates using confocal scanning laser microscopy

The advent of confocal microscopy and fluorescence probes has made possible the routine visualization of the complex three-dimensional structures of thick fixed or live specimens. Four-dimensional (4-D) imaging of biological specimens (three-dimensional image reconstruction of the same living sample at different time points), remains a seldom-used application of confocal microscopy. In the present study we used 4-D imaging techniques to quantitate the invasion of human brain tumor spheroids into fetal rat brain aggregates (FRBAs), using the vital fluorescence membrane dyes, 3, 3'-Dioctadecyloxacarbocyanine perchlorate (DiO) and 1,1'-dioctadecyl-3,3,3',3'-tetramethylindocarbocyanine perchlorate (DiI) as visualization probes. We found invasion patterns similar to the in vivo behavior of these tumors in the brain. Glioblastoma spheroids showed diffuse and circumscribed infiltration accompanied by cystic degeneration or necrosis of FRBAs. Spheroids from cerebral metastasis, however, showed a sharp delimitation of the invasive margin, and did not penetrate the FRBA beyond a depth of 55 microns. Measured rates of glioblastoma invasion varied with the tumor specimens examined. The slopes of the mid-portions of plots of % infiltration vs. time (hours) for four glioblastoma cell lines were 1.7 +/- 0.21 (SD), 0.67 +/- 0.11, 1.4 +/- 0.22 and 1.3 +/- 0.18. We conclude that confocal microscopy with vital fluorescence probes is a practical method that allows for close monitoring and quantitation of the process of invasion in live tissue preparations, and may be used for assessing the in vitro effects of various tumor treatments.

Non-small-cell lung carcinoma cells invade human bronchial mucosa in vitro

To study invasion of lung cancer in vitro a novel three-dimensional coculture assay consisting of living human tissues has been developed. Multicellular spheroids initiated from a new large-cell lung carcinoma cell line (GaL23), found to be invasive in immunodeficient mice, were confronted with precultured bronchial fragments derived from mucosal biopsies obtained during routine fiberoptic bronchoscopy. The bronchial fragments consist of a stromal core with scattered fibroblasts covered by a continuous surface epithelium resting on a basal lamina. During the first 2 wk of confrontation, a gradual retraction of the bronchial epithelium with subsequent adhesion of the tumor cells to the underlying basal lamina occurred. The following week, a limited invasion of tumor cells into the bronchial stroma was seen. To facilitate the entrance of tumor cells through the mucosal surface, the surface epithelium was removed prior to coculture by ethylenediaminetetraacetic acid (EDTA) buffer treatment. Upon confrontation, GaL23 cells then rapidly attached to and migrated on the exposed basal lamina and an increasing number of tumor cells was seen in the stroma during the first week of culture. This model offers opportunities for studying mechanisms of lung cancer adhesion, migration, and invasion using human bronchial mucosa as the natural target tissue.

Stimulation of extracellular matrix components in the normal brain by invading glioma cells

Malignant gliomas are characterized by an extensive invasion of tumor cells into the normal brain parenchyma. A substantial amount of data indicates that cell movement in general is regulated by specific interactions between extracellular matrix components and specific cell-surface receptors. In the present work, multicellular spheroids from 4 human glioma cell lines (U-373Mg, A-172Mg, U-251Mg and HF-66) were confronted with normal rat brain cell aggregates in vitro, which resulted in a progressive invasion of tumor cells into the brain aggregates. The co-cultures were then sectioned and immuno-stained for specific extracellular matrix components (laminin, fibronectin and collagen type IV) and for specific cell-surface receptors which bind to these components (integrins beta1, beta4, alpha3, alpha6). In addition, flow-cytometric measurements and Northern blot analyses showed expression of several different integrins within the cell lines. The alpha3 subunit was expressed strongly in all cell lines. Whereas the beta1 subunit was expressed weakly in exponentially growing monolayer cultures, it showed a pronounced expression in multicellular spheroids, indicating that the integrin expression may vary depending on the micro-environment within a tumor. Furthermore, normal brain tissue was able to produce laminin when confronted with the glioma cells, which also was observed for fibronectin and collagen type IV. The relevance of our observations to the in vivo situation was investigated further by immuno-staining 5 human glioma biopsy samples for laminin. In some areas of the tumors, specific deposits of laminin were observed. In conclusion, we have shown that normal brain tissue has the ability to produce extracellular matrix components, such as laminin, collagen type IV and fibronectin, when confronted with invading glioma cells. Our results show that the glioma cells express specific integrins which can interact with these extracellular matrix components. Such interactions may facilitate tumor cell migration and invasion.

Mitogens as motogens

Numerous in vivo methodologies have documented the invasive behavior of glioma cells through normal brain parenchyma. Glioma cell locomotion has also been assessed with a number of in vitro assays including the Boyden chamber and other chemotaxis assays, colloidal gold cell tracking, analysis of migration of cells tumor cells from spheroids, confrontation cultures of glioma cells with aggregates of non-neoplastic tissue, time-lapse video microscopy, electron microscopic examination of the cytomorphologic correlates of cell motility, the radial dish assay, and quantitative enzyme immunoassay of proteins associated with invasion (e.g. laminin). Several of these techniques have been specifically modified to assess the effects of cytokines on glioma cell motility in vitro. Cytokines studied utilizing these methods include: epidermal growth factor (EGF), basic fibroblast growth factor (bFGF), the bb dimer of platelet-derived growth factor (PDGFbb), nerve growth factor (NGF), interleukin 2 (IL-2), transforming growth factors alpha and beta 1 (TGF alpha and TGFstraat1), and tumor necrosis factor alpha (TNF alpha). This review summarizes the investigational methods used to evaluate random and directional glioma cell motility and invasion in vivo and in vitro. The roles of specific mitogens as motogens, as evaluated with these methods are then presented.

Mitogens as motogens

Numerous in vivo methodologies have documented the invasive behavior of glioma cells through normal brain parenchyma. Glioma cell locomotion has also been assessed with a number of in vitro assays including the Boyden chamber and other chemotaxis assays, colloidal gold cell tracking, analysis of migration of cells tumor cells from spheroids, confrontation cultures of glioma cells with aggregates of non-neoplastic tissue, time-lapse video microscopy, electron microscopic examination of the cytomorphologic correlates of cell motility, the radial dish assay, and quantitative enzyme immunoassay of proteins associated with invasion (e.g. laminin). Several of these techniques have been specifically modified to assess the effects of cytokines on glioma cell motility in vitro. Cytokines studied utilizing these methods include: epidermal growth factor (EGF), basic fibroblast growth factor (bFGF), the bb dimer of platelet-derived growth factor (PDGFbb), nerve growth factor (NGF), interleukin 2 (IL-2), transforming growth factors alpha and beta 1 (TGF alpha and TGFstraat1), and tumor necrosis factor alpha (TNF alpha). This review summarizes the investigational methods used to evaluate random and directional glioma cell motility and invasion in vivo and in vitro. The roles of specific mitogens as motogens, as evaluated with these methods are then presented.

Localization of CD44 at the invasive margin of glioblastomas by immunoelectron microscopy

Glioblastoma multiforme is a highly invasive primary brain tumor, which is known to strongly express the CD44 cell adhesion receptor. A number of experimental studies suggest that the interaction of this receptor with extracellular matrix (ECM) proteins such as hyaluronic acid may in part mediate human glioma cell adhesion and invasion of brain tissue. Although the expression of CD44 and its spliced variants in brain tumors have been extensively studied, there have been no reports localizing its expression to the invasive margin of the tumor. The authors used immunoelectron microscopy to investigate the expression patterns of CD44 in an in vitro organotypic invasion assay. Tumor spheroids initiated from the U373 MG human glioblastoma line were confronted with fetal rat brain aggregates in a spheroid coculture system. The CD44 expression appeared at the interface between glioblastoma tumor spheroids and brain tissue, as well as in the spheroid itself. CD44 immunoreactivity was not detectable in mature 21-day fetal brain aggregates. The findings provide direct evidence that CD44 is expressed at the confrontational invasive border between glioblastomas and brain tissue, further supporting its role in glioma cell-ECM recognition and attachment.

Implantation of C6 astrocytoma spheroid into collagen type I gels: invasive, proliferative, and enzymatic characterizations

A three-dimensional model has been developed in which C6 astrocytoma spheroids of defined sizes are embedded into collagen type I gels. The authors have monitored cell invasive behavior; obtained quantitative data on cell invasion, proliferation, and enzymatic activity; assessed cell-cell interactions by altering the spheroid size used; and studied cell-matrix interactions by modifying the matrix components. Their results show that C6 astrocytoma cells detach from the spheroid surface and invade the gel as single cells by means of a system that appears to be dependent on metalloprotease function. These invasive cells have a low proliferative index. Larger spheroids with central hypoxic microregions possess cells that invade the gel at faster rates; this could be correlated with the release of increased collagen type I degrading activity. Extracellular matrix proteins, such as laminin, fibronectin, and collagen type IV have no significant influence on invasive activity, whereas hyaluronic acid decreases and human central nervous system myelin increases invasion. New strategies directed at the treatment of malignant gliomas must take into account the subpopulation of malignant cells located long distances from the major tumor mass. The spheroid invasion model may provide specific insights into the behavior of these invasive cells.

Glial ontogeny and glial neoplasia: the search for closure

The last ten years have seen rapid progress in both our understanding of the normal progression and control of gliogenesis and in the laboratory techniques necessary to sustain and study most glial cell types, including progenitor cells of both type-1 astrocyte (T1A) and oligodendrocyte-type-2 astrocyte (T2A) lineage. These studies have direct relevance for the lineage analysis of human gliomas, optimizing in vitro glioma models, and suggesting potentially fertile new grounds for glioma biology research. We do not yet known whether malignant transformation occurs only in mature glia that then 'de-differentiate' into cells with glial precursor phenotypes and behavior characteristics, whether neoplastic transformation occurs in O-2A progenitor cells, or whether both mechanisms may occur in different patients. However, preliminary results suggest that astrocytomas can arise from two different glial lineages, that oligodendrogliomas and mixed oligo-astrocytomas arise exclusively from the O-2A lineage, and that medulloblastomas may also have a connection with the O-2A lineage. An ontogeny-based glioma classification system may lead to better prognostic patient data and better predict patient response to treatment than currently available classification systems. Available data from the study of developmental glial biology raises serious doubts about the fidelity and relevance of in vitro glioma models that rely on culture media supplemented with animal serum and suggest that relying on chemically-defined media conditioned by astrocytes may be the better research strategy. Findings from the study of normal gliogenesis also suggest that growth factors are likely to act as much more than simple mitogens in glioma biology. Potentially fertile areas of research for glioma biology include studying the cooperative effect of multiple growth factors, potential growth factor effects as survival factors, inhibitors of differentiation, and differentiation inducers, and studying potential positive humoral feedback loops between glioma cells and normal glial cells, as well as normal non-glial cells, within and surrounding each glioma.

Inhibition of in vivo tumorigenicity and invasiveness of a human glioblastoma cell line transfected with antisense uPAR vectors

Our previous studies showed that glioblastomas express increased urokinase-type plasminogen activator receptors (uPARs) in comparison to low-grade gliomas (Yamamoto et al., Cancer Res., 54, 5016-5020, 1994). To explore whether downregulation of uPAR inhibits tumor formation and invasiveness, a human glioblastoma cell line was transfected with a cDNA construct corresponding to 300 bp of the human uPAR's 5' end in an antisense orientation, resulting in a reduced number of uPA receptors. Co-culture studies with tumor spheroids and fetal rat brain aggregates showed that antisense SNB19-AS1 cells expressing reduced uPAR failed to invade fetal rat brain aggregates. Intracerebral injection of SNB19-AS1 stable transfectants failed to form tumors and were negative for uPAR expression in nude mice. Thus uPAR appears in this model to be essential for tumorigenicity and invasion of glioblastomas in vivo.

Inhibition of epidermal growth factor receptor-associated tyrosine kinase blocks glioblastoma invasion of the brain

OBJECTIVE

Glioblastoma multiforme is a malignant primary brain tumor associated with short patient survival despite aggressive treatment, in part because of its propensity to aggressively infiltrate into brain tissue. Glioblastoma multiforme is also unique because it is the only nonepithelial human tumor for which excessive activation of epidermal growth factor receptor (EGFR) has been consistently linked to tumor growth and patient survival, and EGFR activation promotes glioblastoma multiforme infiltration in vitro.

METHODS

Cocultures of human glioblastoma spheroids (derived from three separate patients) and fetal rat brain aggregates were examined for infiltration using confocal microscopy, in the presence of 0 to 100 mumol/L genistein, a tyrosine kinase (TK) inhibitor, and 3 mumol/L tyrphostin A25, a specific EGFR-TK inhibitor.

RESULTS

Infiltration (not attachment) was completely inhibited by genistein at 10 mumol/L, the IC20 for monolayer growth inhibition in two cell lines. Tyrphostin A25 at 3 mumol/L (the IC20 for monolayers) reduced invasion in a third cell line from 38.8 +/- 6.1% invasion-hour per hour (n = 5) to 2.9 +/- 1.2% invasion-hour per hour (n = 6) (P = 0.0002, two-tailed t test, 93% inhibition), and from 0.54 +/- 0.065% per hour (slope) to 0.028 +/- 0.018% per hour (P = 0.00001, 95% inhibition). Maximal percent invasion was reduced from 100 +/- 0 to 7.4 +/- 5.6% of the fetal rat brain aggregate. No change was detected in EGFR-associated tyrosine phosphorylation at those doses in monolayers by 32P immunolabeling, consistent with the known effects of low concentrations of TK inhibitors. An increase in expression of wild-type and truncated EGFR was demonstrated by Western blotting. Invasion was equally well inhibited by a monoclonal antibody to the high-affinity ligand binding domain of EGFR and not by antibody to an inactive domain.

CONCLUSION

Our observations support the role of EGFR activation as a determinant by which glioblastoma invades normal brain tissue, and we show that invasion can be effectively inhibited at much lower concentrations of TK inhibitors than are necessary for growth suppression.

Quantification of glioma cell invasion by confocal laser scanning microscopy in an in vitro co-culture system

Malignant invasion in co-cultures of spheroids from the glioma cell line GaMg into brain cell aggregates (BA) was determined by two different techniques: by confocal laser scanning microscopy (CLSM) and by conventional light microscopic observations of semi-thin sections obtained from co-cultures. The remaining BA volumes were detected by CLSM in vital dye-fluorescent-stained co-cultures. The same specimens were fixed and embedded in Epon, and cut for histologic and morphometric analyses. The results show that CLSM can be used for continuous determination of progressive glioma invasion. Compared to the light microscopic observations, the degree of invasion appeared slightly lower when analyzed by CLSM. We conclude that the CLSM provides the possibility for continuous studies on interaction between normal and malignant cells. Therefore it considerably improves existing methods for studying tumor cell invasion in vitro.

Quantitative NMR microscopy of multicellular tumor spheroids and confrontation cultures

In cancer research, tumor spheroids are a well established system to study tumor metabolism resembling the situation in vivo more closely cell monolayers. Spherical aggregates of malignant melanoma cells (MV3) and their invasion into rat brain aggregates have been investigated by quantitative NMR microscopy. Relaxation times (T1, T2) and diffusion parameter images were acquired with an in-plane resolution of 14 x 14 microns2. The authors were able to demonstrate that the morphology of the spheroids can be visualized on these NMR maps. The contrast was mainly manifested in relaxation maps, where average relaxation times T1 = 1.94 +/- 0.17 s and T2 = 42.8 +/- 6.3 ms were obtained for proliferating cells, and T1 = 2.49 +/- 0.31 s and T2 = 104.3 +/- 29.4 ms for the necrobiotic center. The mean diffusion coefficients were 0.59 +/- 0.12 micron2/ms and 0.85 +/- 0.14 micron2/ms, respectively. The authors could follow the dynamic process of tumor cell invasion in the investigated co-culture system. Knowledge about tumor cell migration and tumor cell invasion is essential for the understanding of cancer and its therapy. Quantitative NMR microscopy can study this dynamic process noninvasively and therefore may help to assess the influence of therapy on the micromilieu of these spheroids.

Organ culture of a glioblastoma from a patient with an unusually long survival

Multicellular tumor spheroids were directly initiated in vitro from the biopsy specimens of a patient who is alive and who has had no neurological changes in 7 years after the gross removal of a glioblastoma. The spheroids were studied alone and in confrontation with aggregates of fetal rat brain tissue. Both in the biopsy and in the tumor spheroids, a very high proportion of cells were proliferating, as flow cytometric deoxyribonucleic acid measurements showed that 40% of the cells in the biopsy specimens and in the tumor spheroids were in the S and G2M phases of the cell cycle. Despite this high proliferation rate, the volume of the spheroids decreased, indicating an even greater cell loss. Light and scanning electron microscopic studies also indicated cell death in the spheroids. This behavior may be related to the long-time survival.

Tumour cell migration in the central nervous system

Intrinsic tumours of the central nervous system (CNS) are set apart from solid, non-neural primary neoplasms in that, although they seldom metastasize to distant organs, they are generally characterised by a diffuse local invasive pattern. Indeed, it is this important biological characteristic which precludes successful therapeutic intervention in the majority of brain and spinal cord neoplasms. While tumours metastasising to the brain are generally well-circumscribed lesions, sub-populations of neoplastic cells from intrinsic, neuroectodermal tumours may migrate several millimeters away from the brain/tumour interface, resulting in a poor demarcation of the neoplasm. These migratory cells give rise to recurrent tumours following surgical and adjuvant chemo- and radio-therapeutic intervention. The mechanisms which facilitate such migration of neoplastic neural cells into the contiguous normal nervous tissue are poorly documented. However, migration in this context is likely to be a complex multifaceted phenomenon involving cell/cell and cell/extracellular matrix (ECM) adhesion, locomotion, angiogenesis and enzymic degradation of the ECM. In particular, cell adhesion molecules, ganglioside, paracrine and autocrine growth and motility factors and matrix metalloproteinases (MMPs) and their inhibitors probably all play important and inter-dependent roles in the migration of neoplastic neural cells.

Interaction between human medulloblastomas and foetal rat brain aggregates in vitro

The invasiveness of three medulloblastoma permanent cell lines (D-283, D-341, and DAOY), a human medulloblastoma biopsy, and in addition, a human rhabdomyosarcoma permanent cell line (TE-671), which previously had been regarded as a human medulloblastoma, was studied in an organ co-culture assay. All the four cell lines and the biopsy were co-cultured with normal rat brain cell aggregates for up to six days in vitro. The medulloblastoma biopsy, the D-283 and the D-341 cells invaded the brain tissue by diffuse single cell infiltration. The medulloblastoma cell line (DAOY) showed an invasive pattern similar to that observed earlier for most glioblastoma cell lines. This was characterized by massive cell replacement and destruction of normal brain tissue. The rhabdomyosarcoma cell line (TE-671) presented a solid invasive pattern with a fairly well defined border between normal brain and tumour tissue. Thus, the organ co-culture assay system in vitro seems to mimic several aspects of the in situ invasive behaviour of medulloblastomas. It may, therefore, provide new perspectives for pretreatment investigations with chemotherapy and radiotherapy of these malignancies.

Role of plasminogen activator and of 92-KDa type IV collagenase in glioblastoma invasion using an in vitro matrigel model

The invasive nature of human gliomas represents a major factor in preventing their total resection. The exact nature of the underlying mechanisms of tumor cell invasion are still unclear. In this study, we have quantitatively assayed a glioblastoma cell line for its ability to migrate through a polycarbonate filter coated with matrigel which contains a complex of multiple basement membrane components. At 48 h the glioblastoma cell line (U251) showed a rate of invasiveness of 42% and also dependent on the concentration of matrigel. The U251 cell line produced a urokinase type plasminogen activator and a 92-KDa type IV collagenase. Both enzymes were inhibited by the addition of uPA and 92-KDa type IV collagenase antibodies. Those same antibodies reduced the invasion rate of U251 cells from 42% to 12 and 21%, respectively. Similarly, the addition of epsilon-aminocaproic acid (a plasmin inhibitor) or tissue inhibitor of metalloprotease (TIMP2, a collagenase inhibitor) reduced the invasiveness of U251 cells from 42% to 14% and 10%, respectively. Additionally, the other two glioblastoma cell lines (LG11, UWR1) and astrocytes showed a rate of invasiveness at 41%, 61% and 12%, respectively. Finally, the addition of hyaluronic acid to the matrigel, a constituent of brain extracellular matrix, enhanced the rate of invasion. These findings provide evidence for the role of serine proteases and metalloproteases in facilitating the invasion of extracellular matrix components by glioblastoma cell line and suggest a therapeutic role for protease inhibitors in attempting to minimize the invasive propensity of gliomas.

Comparative aspects of neoplastic invasion of the brain

This review of the invasion of the human brain by tumors is based upon a comparative approach. The taxonomic distribution of a structure of the CNS to be considered a brain is present in the following taxonomic groups: Mollusca (Cephalopoda, Pulmonata) Sipunculida, Echiurida, Annelida, Arthropoda and Vertebrata (fishes, amphibians, reptiles, birds and mammals). The comparative approach (inter-, and intraspecies specific) provides a more informative, indicative, understanding of invasion of the human brain; especially in view of the morphological and functional heterogeneity of structures playing a role in the neoplastic invasion to and from the brain. It can be distinguished among primary invasiveness (as cancer in situ) and local recurrence on the one hand and remote spreading, such as metastasis, on the other. A review of the recent literature arranged according to tumor types is provided and comparative conclusions elucidated, especially to underline the tumor specificity of the invasion of brain tumors, especially the second or intra-species specific approach is therapeutically important. This is set forth against a background of epidemiology and species specificity. Due to the lack of bony skull capsule an extensive knowledge of the brain of cephalopod molluscs with its functional and morphological similarity to the human eye, and also that of the pulmonate Helix pomatia, the vineyard snail, could well serve as excellent oncologic models. Some conclusions can be drawn dealing with the theoretical possibility of first appearance of the different and heterogeneous components of the complicated structure known as brain including its supporting accessory organs during phylogeny.

Development of an in vitro extracellular matrix assay for studies of brain tumor cell invasion

Invasion of brain by tumor cells is an inherent feature of the malignant phenotype. Assays to quantitate invasiveness should provide a powerful tool to investigate this phenomenon. We have developed a modified in vitro assay to measure tumor cell invasion, attachment, and chemotaxis using a barrier of the complex basement membrane Matrigel on gelatin-coated filters. Within 5 hours, 7.8% of U251MGp and 2.6% of SF126 human malignant glioma cells invaded the Matrigel and filter, compared with 0.8% of normal human leptomeningeal cells. The extent of invasion was directly proportional to incubation time and filter pore size and inversely proportional to the Matrigel concentration. Cells from exponentially growing U251MGp cultures invaded more readily (10.9%) than cells from plateau-phase cultures (2.3%); however, labeling studies with bromodeoxyuridine showed that quiescent cells and rapidly dividing cells were equally capable of invading. This suggests that the mechanisms underlying invasion by malignant glioma cells are distinct from those underlying proliferation and indicates the need for therapy aimed specifically at invasive behavior. In a practical application of this assay to test a potential anti-invasive strategy, monoclonal antibodies to the beta subunit of an integrin receptor mediating attachment to the extracellular matrix inhibited invasion by U251MGp cells in a dose-dependent manner. This assay should allow evaluation of the cellular and molecular basis of brain tumor progression and perhaps aid the development of rationally designed drugs that limit tumor invasion. It may also allow prediction of the clinical behavior of neoplasms in individual patients.

Effects of EGF, bFGF, NGF and PDGF(bb) on cell proliferative, migratory and invasive capacities of human brain-tumour biopsies in vitro

Spheroids initiated directly from human primary gliomas were used to investigate the effects of EGF, bFGF, NGF and PDGF(bb) on cell proliferation, migration and invasion into foetal rat brain tissue. EGF increased tumour spheroid volume in 10 of 13 glioblastomas studied, whereas 5 of 11 tumours responded to bFGF. NGF increased the spheroid volume in 2 of 5 tumours. In 8 tumours, PDGF(bb) had no effect on tumour spheroid volume. An increase in BUdR-labelling indices confirmed that cell proliferation was responsible for the volume increase observed in stimulated spheroids. EGF stimulated cell migration in 5 and bFGF in 3 of 8 tumours studied. NGF stimulated cell migration in 1 of 5 glioblastomas, whereas 1 of 3 glioblastomas responded to PDGF(bb). The effects of growth factors on the invasion of spheroids prepared from the glioblastoma biopsy specimens were also studied in vitro using foetal rat brain aggregates as target tissue. EGF stimulated invasion in 7 of 8 glioblastomas studied, whereas bFGF stimulated invasion in 2 of these tumours. NGF or PDGF(bb) did not increase the invasiveness of the glioblastoma tissue. Our results represent the net effect of the growth factors on a complex tumour-cell population. We conclude that exogenously administered growth factors, EGF in particular, increase the cell proliferation as well as migratory and invasive capacities of cultured primary brain tumour biopsies in vitro.