Beyond the petri dish: The glioblastoma xenograph model

For many years, investigations of tumor formation, growth and response to treatment have relied on tumor cell lines established in the late 1960s and cultured in vitro. It is now known that prolonged tissue culturing can alter cellular genetic and morphological characteristics. As a result, such cell cultures do not accurately reflect key features of human tumors.

For example, established glioblastoma multiforme (GBM) cell lines cultured in a petri dish lack a growth factor receptor found in more than 40 percent of human primary GBM tumors. Prolonged cell culturing can also lead to hypermethylation of specific genes, a DNA modification present in twice the number of cultured versus human GBM tumors.

Mayo's glioblastoma xenograft model

Xenografting, a means of introducing foreign tissue into an organism, can be used to generate in vivo cell lines by implanting human tumor specimens directly in animal models. One of the first laboratories to do so was at Mayo Clinic in Rochester, Minn., under the direction of Jann N. Sarkaria, M.D., a radiation oncologist and head of the Animal Core of the Specialized Program of Research Excellence (SPORE) program, where investigators have developed more than 50 glioblastoma xenograft models since 2002.

Tissue specimens from human tumors are implanted into the mouse flank and, if growth occurs, can subsequently be implanted into mouse brains. Serial transplantation into further generations of mice continues the maintenance of in vivo xenograft cell lines.

The singular advantage of xenografting for basic and translational research is that xenografted cells generate tumors that maintain the important morphological, molecular, and histopathologic features of primary human tumors, including, for example, the invasive features of GBM. Different subtypes of GBM tumors can be observed to see how quickly they grow in mice, how invasive they are and what their individual pathologic features are. When phenotyping is concluded, the molecular mechanisms can be examined and manipulated.

Currently, Mayo has 52 cell lines that represent four of the five subtypes of GBMs. Of those cell lines, 35 have been implanted in mouse brains and later dissected out for gene-expression profiling, measurement of messenger RNA globally, and detection of chromosomal deletions or amplifications. The laboratory is about to begin whole-exome sequencing. It is exacting, painstaking and labor-intensive work.

Because of its scope relative to the number of cell lines and the degree of phenotyping and genetic characterization, Mayo's xenograft model is considered a premier, state-of-the-art model. Research grant applicants to the National Institutes of Health are often told they need a model that is similarly characterized, and Mayo has shared its cell lines with researchers around the world — 55 to date. The SPORE grant provides financial support for this type of distribution.

GBM xenograft modeling in action

As Dr. Sarkaria points out, the majority of clinical trials are launched with limited human or animal data. It is gradually becoming the standard to have animal data, but most preclinical drug trials use tumors that are cultured in vitro.

The xenograft model allows evaluation of therapeutic drugs and combined drug and irradiation regimens in living tissue, using tumor lines that more closely replicate human tumor therapeutic sensitivity. Another advantage is that identical tumor subtypes can be implanted in mouse brains and the mice randomly assigned to a treatment or a no-treatment group.

Dr. Sarkaria notes that in this way, "you have a very robust understanding of how the drug works in a particular tumor with a particular gene signature and molecular characteristics. Human patients, on the other hand, may develop medical conditions, such as a blood clot, that can cloud the outcome."

Variable response to EGFR therapies

Increased signaling of epidermal growth factor receptor (EGFR) inhibitors is thought to contribute to the malignant characteristics of certain tumors, including GBM. Among these tumors, however, only subgroups within a given type are responsive to EGFR inhibitor therapy such as erlotinib or gefitinib. Several laboratory approaches have been used to determine the molecular factors that explain the varying response, but they have had mixed results.

The xenograft approach successfully replicated the clinical finding that mutations in certain molecular markers were a factor in EGFR inhibitor sensitivity and resistance. The model also demonstrated that an additional mutation contributed to sensitivity in two specific GBM subtypes (Sarkaria et al. Mol Cancer Ther. 2007;6[3]:1167-74).

New mechanisms of acquired resistance to TMZ

Another validation of correlations between the animal model and human tumor mechanisms has come from an investigation into why gliomas develop chemoresistance to the commonly used drug temozolomide (TMZ). TMZ is known to induce cellular apoptosis. Although GBM is initially responsive, more than 90 percent of recurrent GBM tumors show no response to a second round of TMZ.

Using TMZ-resistant GBM xenograft cell lines from Mayo, a research team at the University of Alabama found that both primary and recurrent human GBM biopsies and primary and TMZ-resistant GBM xenograft lines exhibit a similar, although unexpected, remodeling adaptation to TMZ. This finding not only helps explain the nature of TMZ resistance but also will inform future drug development (Oliva et al. J Biol Chem. 2010;285[51];39759-67).

Do PARP Inhibitors Work?

PARP inhibitors are a class of drugs that inhibit production of an enzyme called poly (ADP-ribose) polymerase, or PARP. Several preclinical studies, using established cell lines, suggested that PARP inhibitors could enhance the efficacy of TMZ in both TMZ-sensitive and TMZ-resistant tumors and would also improve the effects of radiation therapy.

In anticipation of a large clinical trial to evaluate a PARP inhibitor, Dr. Sarkaria and his team tested it in a panel of glioblastoma xenografts. The results, unlike those predicted from established cell lines, showed that although the PARP inhibitor was effective in newly formed tumors, recurrent tumors remained resistant. The investigators also discovered that specific expression of a DNA repair protein, MGMT, makes tumors resistant to the sensitizing effects of PARP inhibitors.

A clinically significant finding was the fact that some PARP inhibitors were more effective in mouse flank tumors than tumors in the brain, a dissociation that could be demonstrated only by contrasting in vivo tumor sites using a xenograft model. The differences in sensitivity of brain versus flank tumors suggest a failure of these specific PARP inhibitors to penetrate the blood-brain barrier. This finding is in contrast to the commonly accepted paradigm that because portions of GBM tumors have an open blood-brain barrier, drugs will penetrate all of the tumor.

Mayo's results suggest that failure of penetration is an important issue in designing novel treatments such as PARP inhibition. Mimicking human GBM tumor growth, the xenograft model has much to offer in understanding the fundamental biology of GBM tumors, their resistance to standard and experimental treatments, and the development of novel therapies.