**3. Rodent models of gliomas**

Animal models are often used when researching a disease and trying to understand how a particular pathological process occurs, as well as a means of studying the efficacy of potential new therapies. This review will provide examples of commonly used and new experimental animal models for gliomas, which make up a large portion of primary brain tumors. The majority of models involve intracerebral implantation of rodent (rat or mouse) or human glioma cells into synergetic rats or mice, or immunocompromised rodents (e.g. nude or athymic rats or mice). There are also a limited number of transgenic mouse models for gliomas. One approach to better simulate a human tumor is to obtain human glioma neurospheres from patients during tumor resection, and then culture the cells prior to intracerebral implantation into immunocompromised rodents. Another recent approach is to implant non-replicating viruses that can stimulate neuronal stem cells to turn into glioma cells which develop into diffuse tumors similar to those found in high-grade or malignant gliomas called, glioblastoma multiforme (or GBM).

### **3.1 Intracerebral cell implantation models**

252 Advances in the Biology, Imaging and Therapies for Glioblastoma

information regarding alterations in tissue structure associated with glioma tumors. An extension of the diffusion-weighted imaging technique is diffusion tensor imaging (DTI), which can provide information on white matter neuronal fiber tractography. MR spectroscopy can be used to assess alterations in tumor metabolites associated with glucose, bioenergetics, amino acid or lipid metabolism, for example. Many of these advanced MR techniques are used in a clinical setting. More recently, molecular MRI (mMRI) which incorporates a MRI contrast agent as a signaling molecule and an affinity component that targets specific tumor markers associated with tumor growth, angiogenesis, cell invasion, inflammation, or apoptosis, can be used to characterize *in vivo* molecular events associated

This review will focus on various glioma models that have been studied *in vivo*, with particular emphasis on MR image and/or spectroscopy evaluation of these models, and the use of MR image criteria (morphological, biophysical, molecular and metabolic) to evaluate therapeutic treatments. The aims are to: (1) provide an overview on current rodent glioma models being studied; (2) provide an overview regarding currently used MR methods, including advanced MR techniques (e.g. MRA, DWI, PWI, MRS and mMRI), relevant to glioma research; and (3) summarize studies that have used MR methods to evaluate

Gliomas represent 40% of all primary central nervous system (CNS) tumors diagnosed. Among them, glioblastomas (GBM) are the most malignant, with a very poor survival time of about 15 months for most patients diagnosed with this grade IV brain tumor (CBTRUS 2011). High grade gliomas are the most common primary brain tumors in adults, and their malignant nature ranks them as the fourth largest cause of cancer death (Niclou *et al.*, 2010). There are four tumor grades for gliomas: Grade I which is a non-malignant, fairly circumscribed astrocytoma that is rare and appears in young adults; Grade II which are a more common diffusely infiltrating astrocytoma; Grade III which is an anaplastic astrocytoma; and Grade IV which is a glioblastoma multiforme (Niclou *et al.*, 2010). Grades II-IV gliomas generally are found in the adult population, and often recur following current treatment options (including surgical resection, radiotherapy, and chemotherapy) (Niclou *et* 

Grading and identification criteria that can be used to provide information regarding tumor behavior are cell proliferation (cellularity and mitotic activity), nuclear atypia, neovascularization and the presence of necrosis and/or apoptotic regions (Gudinaviciene *et al*., 2004). Grade II gliomas (also referred to as diffuse astrocytomas) and grade III gliomas (also referred to as anaplastic astrocytomas) only differ based on their mitotic activity, and this difference accounts for a substantial decrease in the 5-year survival for patients, from 47% to 29% (from grade II to III, respectively) (CBTRUS 2011). Grade IV gliomas (GBM) are often characterized by the presence of large necrotic areas (Gudinaviciene *et al*., 2004) and

Animal models are often used when researching a disease and trying to understand how a particular pathological process occurs, as well as a means of studying the efficacy of

therapeutic response in pre-clinical models for gliomas.

generally have a 3% 5-year survival (CBTRUS 2011).

**3. Rodent models of gliomas** 

with gliomas.

**2. Human gliomas** 

*al.*, 2010).

Glioma cells (rat, mouse or human origin) are injected into the cerebral cortex of rats or mice (synergetic if cells are transplanted into the same species and strain that they were obtained from, or immune-compromised rats or mice if human cells are used) using a stereotaxic device for precise implantation into a brain region. As tumors grow over a period of 1-2 months, this model is considered a short-term model. Different cell lines varying in their degree of malignancy, such as rat C6, 9L/LacZ, F98 and RG2 cells, mouse GL261 cells and human U87 cells, provide a range of gliomas from moderately aggressive to GBM-like.

Many of these models have some characteristics associated with human gliomas, such as aggressive tumor growth, angiogenesis, and tumor necrosis (in a few models), however the diffuse nature of high-grade gliomas, glioblastoma multiforme (GBM), is not well represented. In many instances the intracerebrally-implanted rodent tumors have defined tumor boundaries, which do not represent the infiltrative nature of GBMs well. A comprehensive review that discusses the advantages and disadvantages of rat brain tumor models, most of them involving intracerebral implantation of rat glioma cells, is discussed in a paper by Barth and Kaur (2009).

The rat C6 cell line produces diffusively invasive astrocytomas (Barth, 1998; Barth and Kaur, 2009), which have been found to be similar to human glioma cells regarding the expression of genes mainly involved in tumor progression (Sibenaller *et al*., 2005). C6 gliomas were induced in an outbred Wistar rat strain repeatedly injected with methylnitrosourea (MNU), which makes it non-syngeneic in inbred strains, and increases its potential to evoke an alloimmune response (Barth and Kaur, 2009). As a result of some genetic similarities to human gliomas, the C6 model has been widely used as a GBM model for a number of years (Grobben *et al*., 2002; Barth and Kaur, 2009). The 9L/LacZ-derived tumors are aggressive and infiltrative, and are angiogenic (Plate *et al*., 1993), which are some of the characteristics associated with human GBM (Weizsaecker *et al*., 1981). Although the aggressive 9L/LacZ gliomas are highly invasive (Szatmori *et al*., 2006) and have extensive neovascularization (Plate *et al*., 1993), due to their pronounced immunogenicity (Barth, 1998) and the fact that they are classified as gliosarcomas (Sibenaller *et al*., 2005), makes these cells a poor choice for glioma studies. F98 gliomas are classified as anaplastic malignant tumors, which have an infiltrative pattern of growth, and also have attributes associated with human GBM (Barth, 1998; Barth and Kaur, 2009). The aggressive and invasive nature (Barth, 1998; Barth and Kaur, 2009) of RG2 tumors (Groothuis *et al*., 1983), as well as the highly tumorigenic human glioblastoma U87 MG cell line *(Martens et al*., 2006; Cheng *et al*., 1996), both mimics human

Assessment of Rodent Glioma Models Using Magnetic Resonance Imaging Techniques 255

Hupp, 2011; Muller et al., 2011). PTEN, which stands for phosphatase and tensin homolog, is a tumor suppressor gene also involved in the regulation of the cell cycle (Natsume et al., 2011; Alexiou and Voulgaris, 2010). Rb/p53 mice developed malignant tumors in approximately 9 months (Jacques *et al*., 2010). PTEN/Rb/p53 tumors had an appearance that was similar to the Rb/p53 tumors (Jacques *et al*., 2010). Deletion of Rb/p53 or Rb/p53/PTEN resulted in the formation of primitive neuroectodermal tumors (PNET), which alludes to the role of an initial Rb loss involved in driving the PNET phenotype (Jacques *et al*., 2010). It was found that targeted deletion of PTEN and p53 in subventricular zone (SVZ) stem cells resulted in glioma formation with a latency period of approximately 7-8 months (Jacques *et al*., 2010). The tumors from the recombination of PTEN/p53 were histologically infiltrative, diffuse, necrotic and had signs of micro-vascular proliferation,

Another successful transgenic mouse model involves the deletion of the *TP53* (tumor protein 53) gene (*Trp53 null* background), and over-expressing human PDGF (plateletderived growth factor) under the control of the GFAP (glial fibrillary acidic protein) promoter, which developed tumors with human glioblastoma-like features and with the integrated development of PDGFRα+ tumor cells and PDGFRβ+ Nestin+ vasculature in 2-6 months (Hede *et al.*, 2009). The tumor suppressor gene *TP53* is either lost or commonly mutated in astrocytic brain tumors, and these *TP53* alterations are often combined with excessive growth factor signaling via the PDGF/PDGFRα complex (Hede *et al*., 2009). PDGF is one of many growth factors that regulate cell growth and division, and has been found to be widely associated with malignant gliomas (Calzolari and Malatesta, 2010; Shih and

GBM cancer-initiating cells have been found to mediate resistance to chemotherapy and radiation treatment, both used as follow-up therapies following surgical resection of the main tumor mass (Wei *et al*., 2010). Cells isolated from GBM that possess the capacity for self-renewal following radiation and chemotherapy, can form neurospheres when cultured *in vitro* (Wei *et al*., 2010). The glioma-associated cancer-initiating cells were found to express MHC-I (major histocompatibility I) but not MHC-II, CD-40 or CD80, which induces T-cell immune deficiency, and express the costimulatory inhibitory molecule, B7-H1, which plays a role in mediating immune resistance in gliomas and induces T-cell apoptosis (Wei *et al*., 2010). These neurospheres can be intracerebrally implanted into immune-compromised rodents to develop tumors *in vivo,* and therefore provide an experimental model that more closely resembles recurrent human GBM (radiation and chemotherapeutic resistant and induce immunosuppression) to evaluate new therapies. Another approach that takes into consideration the role of tumor-initiating stem cells, is to orthotopically implant tiny fragments of surgically-resected tumors, containing brain tumor stem cells within the glioblastoma tissue, into immunocompromised mice (xenograft model) brains with the use

Glial progenitor cells in the white matter and subventricular zone within the central nervous system were recently found to be the likely candidates for glioma-initiating cells (Assanah *et* 

which are all characteristics of human high-grade gliomas (Jacques *et al*., 2010).

Holland, 2006).

**4. Human glioma neurospheres** 

of a trocar system (Fei *et al*., 2010).

**4.1 Viral-induced glioma models** 

high-grade gliomas via inducing vascular alterations. U87 pcDNA3 and U87 IRE1 DN human glioma cells were selected as malignant glioma models that form highly versus poorly vascularized tumors, respectively (Drogat *et al*., 2007; Wehbe *et al*., 2010). GL261 cells give rise to quickly growing, and diffusively invasive intracranial tumors in C57BL/6 mice (Szatmori *et al*., 2006). RG2 and F98 glioma cell lines were both obtained from chemical induction as a result of administering ethylnitrosourea (ENU) to pregnant rats, where the progeny developed brain tumors that were isolated, and propagated and cloned in cell culture (Barth and Kaur, 2009). Human U87 cells are of high interest for angiogenesis studies (Cheng *et al*., 1996). The immunogenicity issue of the 9L/Lacz model can be resolved by using non-immunogenic models (e.g. RG2).

Xenograft models, induced by orthotopic (into native tumor sites) injection of primary tumor cells or tumor cell lines, represent the most frequently used *in vivo* cancer model systems for glioma research (Waerzeggers *et al*., 2010). Both cell culture and xenograft model systems lack the stepwise genetic alterations that are thought to occur during tumor progression, and often do not represent the genetic and cellular heterogeneity of primary tumors, as well as the complex tumor-stroma interaction (Waerzeggers *et al*., 2010). Genetically engineered mouse models (discussed below in the "Transgenic Mouse Models" section) better represent the causal genetic events and subsequent *in situ* molecular evolution, the tumor-stroma interactions, and consist of cellular subpopulations such as cancer stem cells (discussed further in the "Human Glioma Neurospheres" and "Viral-Induced Glioma Models" sections below), that occur in native tumors (Waerzeggers *et al*., 2010).

#### **3.2 Chemical-induced model**

Slow-growing, low- and high-grade, spontaneous gliomas can be generated with a chemically-induced model from the administration of ENU (Kish *et al*., 2001; Koestner, 1990). Transplacental ENU exposure of a pregnant female a day before gestation, results in the formation of low-grade oligodendrogliomas and mixed gliomas, with a tumor incidence approaching 100%, in rat pups at approximately 3-6 months of age (Koestner, 1990). In addition to oligodendrogliomas and mixed gliomas, unfortunately the ENU-induced model also results in the formation of meningiomas (Koestner *et al*., 1971), spinal cord tumors (Koestner *et al*., 1971) and other primitive neuroectodermal tumors (Vaquero *et al*., 1992), decreasing its potential as a reproducible model. In addition to the isolation of RG2 and F98 rat glioma cells from ENU induction, A15A5 neoplastic astrocytes have also been cloned (Davaki and Lantos, 1980).

#### **3.3 Transgenic mouse models**

As we are beginning to understand the genetic mutations associated with gliomas, it is possible to generate transgenic mouse models that have these genetic mutations. Recent findings suggest that brain tumors originate from neural stem or progenitor cells. Some examples of transgenic mutations include deletions of gene combinations, such as Rb/p53, Rb/p53/PTEN or PTEN/p53 (Jacques *et al.*, 2010). pRb is a retinoblastoma protein, which is a tumor suppressor protein that is dysfunctional in many cancers. Rb controls excessive cell growth by inhibiting cell cycle progression until the cell is ready to divide (Chinnam and Goodrich, 2011; Lohmann, 2010). p53 which is also known as protein 53 is a tumor suppressor protein responsible for regulating the cell cycle (Kim et al., 2011; Maclaine and Hupp, 2011; Muller et al., 2011). PTEN, which stands for phosphatase and tensin homolog, is a tumor suppressor gene also involved in the regulation of the cell cycle (Natsume et al., 2011; Alexiou and Voulgaris, 2010). Rb/p53 mice developed malignant tumors in approximately 9 months (Jacques *et al*., 2010). PTEN/Rb/p53 tumors had an appearance that was similar to the Rb/p53 tumors (Jacques *et al*., 2010). Deletion of Rb/p53 or Rb/p53/PTEN resulted in the formation of primitive neuroectodermal tumors (PNET), which alludes to the role of an initial Rb loss involved in driving the PNET phenotype (Jacques *et al*., 2010). It was found that targeted deletion of PTEN and p53 in subventricular zone (SVZ) stem cells resulted in glioma formation with a latency period of approximately 7-8 months (Jacques *et al*., 2010). The tumors from the recombination of PTEN/p53 were histologically infiltrative, diffuse, necrotic and had signs of micro-vascular proliferation, which are all characteristics of human high-grade gliomas (Jacques *et al*., 2010).

Another successful transgenic mouse model involves the deletion of the *TP53* (tumor protein 53) gene (*Trp53 null* background), and over-expressing human PDGF (plateletderived growth factor) under the control of the GFAP (glial fibrillary acidic protein) promoter, which developed tumors with human glioblastoma-like features and with the integrated development of PDGFRα+ tumor cells and PDGFRβ+ Nestin+ vasculature in 2-6 months (Hede *et al.*, 2009). The tumor suppressor gene *TP53* is either lost or commonly mutated in astrocytic brain tumors, and these *TP53* alterations are often combined with excessive growth factor signaling via the PDGF/PDGFRα complex (Hede *et al*., 2009). PDGF is one of many growth factors that regulate cell growth and division, and has been found to be widely associated with malignant gliomas (Calzolari and Malatesta, 2010; Shih and Holland, 2006).
