A Mouse Model of Chordoma

By Brian Harfe, PhD

An Introduction to Chordoma

Chordoma is a very rare type of primary, malignant bone tumor that affects 1 in 1,000,000 people. The treatment for this disease has traditionally been surgery. Recently, radiation therapy has been reported to be successful for the local control of the disease (Sciubba et al., 2008). However, these treatments are not optimal, as the median survival rate is still only about 7 years (McMaster et al., 2001).

Why Create a Mouse Model?

Research, and in particular the development of new drugs, is very expensive. It is extremely difficult for researchers to convince drug companies and funding agencies to devote the millions of dollars required to bring a novel drug therapy to the clinic for such a rare disease. As a result, chordoma patients have usually been enrolled in clinical trials geared towards treating other types of cancers. While clinical trials designed for other cancers may have benefits for chordoma patients, it is possible that there are chordoma-specific drug treatments that remain undiscovered.

As outlined by Dr. Rubin in a recent ESUN Op Ed article, development of a mouse model for a given disease has the potential to greatly speed up the development of new therapies. Some reasons why we use mice as the model system of choice in my laboratory and why this model system is ideal for screening new drugs are:

Mice are relatively inexpensive to maintain (feed and house). The key word here is "relative." At the University of Florida we are charged $304 a year per mouse cage (this number varies at different universities, with the University of Florida being close to "average" for a major research university). In each cage up to five mice can be housed. My laboratory currently has about 300 mouse cages, which leads to the not-so-inexpensive price tag of about $91,000/year in animal costs. However, compared to using large primates, mice are plentiful and cheap.
Using mouse embryonic stem (ES) cells allows genes to be altered - we can make "designer mice." This is the major reason we use mice in our laboratory. The ability to selectively introduce any genetic change into the mouse genome has allowed scientists to uncover the function of many genes and to successfully model some human diseases. The 2007 Nobel Prize in Medicine was awarded for the discovery of the power of mouse ES cells.
Mice reach reproductive age quickly and often have large litters, allowing us to create animals containing different combinations of genetic mutations. This means that research with mice can proceed faster than it often can with other research animals.

Starting the Model

Recent data has suggested that in humans, chordomas form from "benign notochordal cell tumors" (Yamaguchi et al., 2004; Yamaguchi et al., 2005). These benign tumors are thought to form from "notochordal remnants" that are left over from the embryonic notochord. The notochord is a rod of cells running along the entire vertebral column of the developing embryo (including humans). It has long been known to make proteins required for spinal column formation (Cleaver and Krieg, 2001; Hunter et al., 2003), but what happens to the notochord cells during later life has been debated for >50 years (Walmsley, 1953).

In my laboratory we recently performed a "fate map" of notochord cells using the mouse model system. In fate-mapping experiments, specific populations of cells are marked in the developing embryo. In our case, we marked cells within the notochord. The power of fate mapping is that once cells are marked during embryogenesis, these cells stay marked and can be identified during any stage of life. Our fate mapping of the notochord revealed that notochord cells form the middle part of the intervertebral disks, called the nucleus pulposus (Fig. 1).

Expressing the Cre Protein

The fate-mapping experiments were done by expressing a protein called CRE in the notochord using the Shhgfpcre or ShhcreERT2 alleles we created previously (Harfe et al., 2004). These alleles were crossed to mice containing either the R26R (Soriano, 1999) or eYFP (Srinivas et al., 2001) cre-inducible reporter alleles. Expression of Cre protein in a cell carrying a reporter allele instigates a recombination even that removes a translational stop cassette that is normally present upstream of the reporter allele. Removal of this piece of DNA results in activation of the reporter. Once the reporter is activated it cannot be turned off. The cell the reporter was activated in and all descendants of this cell will express the reporter. Using this system we were able to label notochord cells (an embryonic structure) and determine in two year-old mice that the notochord produced the nucleus pulposus of the intervertebral disk. The two Cre lines we created have been deposited at the Jackson Laboratory and can be obtained for a nominal fee from this mouse repository.

The nucleus pulposus has been proposed to contain both "notochordal" and "chondrocyte-like" cells (Hunter et al., 2003). Our data suggest that both of these cell types arise from the notochord since in our fate mapping experiments the entire nucleus pulposus is labeled at all stages of mouse development.

We were excited to discover that notochord cells produce the nucleus pulposus since this structure is frequently damaged in people that have recurrent back pain. In our laboratory we are currently investigating what genes are required to form the notochord using mouse models. In addition, we are culturing notochord cells to see if they can aid in healing damaged intervertebral disks.

During the course of performing the fate mapping experiments we noticed that not all notochord cells ended up in the nucleus pulposus. It appeared that even in adult mice, a small number of notochord cells remained within each vertebra (see Fig. 1). Initially, we ignored these cells since we were focusing our efforts on disk formation. Eventually, we realized that we really should investigate these "notochordal remnants" since they were appearing in every animal we analyzed. A search of the literature revealed that notochordal remnants had been described in humans (Hunter et al., 2003; Yamaguchi et al., 2004; Yamaguchi et al., 2005) and were thought to form chordomas but had never been previously observed in mice.

In our fate-mapping experiments we activated a reporter gene (a "marker" gene) in the mouse notochord, which allowed us to see notochordal remnants. In humans, notochordal remnants are believed to undergo a very rare and still unidentified event leading to chordoma.

Creating the Model

Using a grant co-funded by the Liddy Shriver Sarcoma Initiative and the Chordoma Foundation, our laboratory activated a gene called Sonic Hedgehog (Shh), which is known to cause cancer in other tissues in the notochord in an attempt to create a mouse model of chordoma. Our hypothesis is that by forcing notochord cells, and as a result notochordal remnants, to express a cancer-causing gene, we will create mice that develop chordomas. Since no other type of cancer has been reported to form from notochordal remnants, it is unlikely that forced expression of cancer-causing genes will generate other types of sarcomas.

The gene Sonic Hedgehog was named after the Sega video game character Sonic the Hedgehog. In 1993 the race was on to find the homologue of the fly gene called Hedgehog (when flies lack this gene they had abnormal bristles, hence the name of the mutant). When the mouse gene was found it was realized that there were three different hedgehog genes in vertebrates. They were given the names Indian hedgehog, Desert Hedgehog (both hedgehogs found in the wild) and Sonic Hedgehog.

Currently, mice in which Shh was activated in the notochord are being "aged" in our mouse colony. Chordoma is a very slow growing type of tumor in humans. To increase the chance of generating chordomas we are letting our mice grow until old age (~1.5 years). The mice are currently 6 months old.

We know that the Shh signaling pathway has been activated in these mice since they have defects in their limbs, a known location of Shh expression (see Figure 2).

It is a bit frustrating to have to wait 1.5 years to determine if these mice have chordomas, so treated animals were shipped to Dr. Vijaya Ramesh at Massachusetts General Hospital for a non-invasive examination. Dr. Ramesh performed a CAT scan on the mice to determine if they have tumors. The advantage of this technique is that it can be used to directly determine if tumors are present in the mice without having to sacrifice the animals. At 5-months of age, no tumors were detected.

The Future

The presence of limb defects in our mice suggests that the experiment has worked (i.e. the Shh signaling pathway has been activated in notochordal remnants). Currently, these mice do not contain chordomas that can be detected by a CAT scan. Since chordomas grow very slowly it is possible that very small tumors are present that cannot yet be detected. The mice will be reexamined in 3 months when they are nine-months old. It is possible that the gene we chose, Shh, when overexpressed will not cause chordomas. Recently a number of additional genes have been shown to be altered in chordomas. We are currently taking the same approach described above, using two additional genes, to attempt to generate a mouse model for chordomas.

By Brian Harfe, PhD
Assistant Professor
Molecular Genetics and Microbiology
University of Florida

References

Cleaver, O. and Krieg, P. A. (2001). Notochord patterning of the endoderm. Dev Biol 234, 1-12.

Harfe, B. D., Scherz, P. J., Nissim, S., Tian, H., McMahon, A. P. and Tabin, C. J. (2004). Evidence for an expansion-based temporal shh gradient in specifying vertebrate digit identities. Cell 118, 517-28.

Hunter, C. J., Matyas, J. R. and Duncan, N. A. (2003). The notochordal cell in the nucleus pulposus: a review in the context of tissue engineering. Tissue Eng 9, 667-77.

McMaster, M. L., Goldstein, A. M., Bromley, C. M., Ishibe, N. and Parry, D. M. (2001). Chordoma: incidence and survival patterns in the United States, 1973-1995. Cancer Causes Control 12, 1-11.

Sciubba, D. M., Chi, J. H., Rhines, L. D. and Gokaslan, Z. L. (2008). Chordoma of the spinal column. Neurosurg Clin N Am 19, 5-15.

Soriano, P. (1999). Generalized lacZ expression with the ROSA26 Cre reporter strain. Nat Genet 21, 70-1.

Srinivas, S., Watanabe, T., Lin, C. S., William, C. M., Tanabe, Y., Jessell, T. M. and Costantini, F. (2001). Cre reporter strains produced by targeted insertion of EYFP and ECFP into the ROSA26 locus. BMC Dev Biol 1, 4.

Walmsley, R. (1953). The development and growth of the intervertebral disc. Edinburgh Med J 60, 341-64.

Yamaguchi, T., Suzuki, S., Ishiiwa, H. and Ueda, Y. (2004). Intraosseous benign notochordal cell tumours: overlooked precursors of classic chordomas? Histopathology 44, 597-602.

Yamaguchi, T., Watanabe-Ishiiwa, H., Suzuki, S., Igarashi, Y. and Ueda, Y. (2005). Incipient chordoma: a report of two cases of early-stage chordoma arising from benign notochordal cell tumors. Mod Pathol 18, 1005-10.

A Mouse Model of Chordoma: Experimental Plan

By Brian Harfe, PhD

Introduction

Why Create a Mouse Model?

Research, and in particular the development of new drugs, is very expensive. It is extremely difficult for researchers to convince drug companies and funding agencies to devote the millions of dollars required to bring a novel drug therapy to the clinic for such a rare disease. As a result, chordoma patients have usually been enrolled in clinical trials geared towards treating other types of cancers (e.g., see the Chordoma Foundation's Clinical Trial webpage). While clinical trials designed for other cancers may have benefits for chordoma patients, it is possible that there are chordoma-specific drug treatments that remain undiscovered.

As outlined by Dr. Rubin in a recent ESUN Op Ed article (Of mice and men – why we need to embrace mouse models of human sarcomas), development of a mouse model for a given disease has the potential to greatly speed up the development of new therapies. Below I list some of the reasons why we use mice as the model system of choice in my laboratory and why this model system is ideal for screening new drugs.

Mice are relatively inexpensive to maintain (feed and house). The key word here is "relatively." At the University of Florida, we are charged $304 a year per mouse cage (this number varies at different universities, with the University of Florida being close to "average" for a major research university). In each cage up to five mice can be housed. My laboratory currently has about 300 mouse cages, which leads to the not-so-inexpensive price tag of about $91,000/year in animal costs. However, compared to using large primates, mice are plentiful and cheap.
Using mouse embryonic stem (ES) cells allows genes to be altered. We can make "designer mice." This is the major reason we use mice in our laboratory. The ability to selectively introduce any genetic change into the mouse genome has allowed scientists to uncover the function of many genes and to successfully model some human diseases. The 2007 Nobel Prize in Medicine was awarded to Mario R. Capecchi, Sir Martin J. Evans, and Oliver Smithies for the discovery of the power of mouse ES cells.
Mice reach reproductive age quickly and often have large litters. This allows us to create animals containing different combinations of genetic mutations. This means that research with mice can proceed faster than it often can with other research animals.

Mouse Costs

As a vertebrate developmental biologist my biggest headache is worrying about mouse costs. A standard large grant request from NIH (an R01 grant) is for $250,000/year. In the current NIH budget environment, all budgets are cut 10-20% (or more), leaving researchers that are lucky enough to obtain a grant with perhaps $210,000/year. Once the cost of maintaining the research mice is subtracted and paying salaries, very little money is left for actually performing experiments. For a description of the cost issues affecting vertebrate researchers, please see "Animal research: Mighty mouse" by Jane Qiu (Nature 444, 814-816, 14 December 2006) and for a description of the cost issues affecting vertebrate researchers, please see "The Mouse House as a Recruiting Tool" (Science, Vol. 288, No. 5468, April 2000, pp. 254-255).

A vertebrate developmental biologist is a scientist who uses a vertebrate model system, for example mice, to investigate how different tissues form during the lifespan of an organism. Traditionally, a developmental biologist would observe how different cell types moved during growth of an organism and then try to infer the mechanisms responsible for forming a given organ. With the onset of molecular biology in the 1970’s scientists began to move from looking at cells to investigating the role individual genes played in forming different organs.

How Will We Create a Mouse Model for Chordoma?

Recent data has suggested that in humans, chordomas form from "benign notochordal cell tumors" (Yamaguchi et al, 2004; Yamaguchi et al, 2005). These benign tumors are thought to form from "notochordal remnants" that are left over from the embryonic notochord. The notochord is a rod of cells running along the entire vertebral column of the developing embryo (including humans). It has long been known to make proteins required for spinal column formation (Cleaver and Krieg, 2001; Hunter et al, 2003), but what happens to the notochord cells during later life has been debated for more than 50 years (Walmsley, 1953).

In my laboratory, we recently performed a "fate map" of notochord cells using the mouse model system. In fate-mapping experiments, specific populations of cells are marked in the developing embryo. In our case, we marked cells within the notochord. The power of fate mapping is that once cells are marked during embryogenesis, these cells stay marked and can be identified during any stage of life. Our fate mapping of the notochord revealed that notochord cells form the middle part of the intervertebral disks, called the nucleus pulposus (Fig. 1).

Benign Notochordal Cell Tumors are tumors that resemble chordomas but are not cancerous. The clinical definition is that these cells are intraosseous benign lesions of notochordal origin (Yamaguchi et al., 2004). It has been proposed that these benign lesions very rarely, and for unknown reasons, directly give rise to chordomas in humans (Yamaguchi et al., 2005).

Fate Mapping

The fate-mapping experiments were done by expressing a protein called CRE in the notochord using the Shhgfpcre or ShhcreERT2 alleles we created previously (Harfe et al., 2004). These alleles were crossed to mice containing either the R26R (Soriano, 1999) or eYFP (Srinivas et al., 2001) cre-inducible reporter alleles. Expression of Cre protein in a cell carrying a reporter allele instigates a recombination even that removes a translational stop cassette that is normally present upstream of the reporter allele. Removal of this piece of DNA results in activation of the reporter. Once the reporter is activated it cannot be turned off. The cell the reporter was activated in and all descendants of this cell will express the reporter. Using this system we were able to label notochord cells (an embryonic structure) and determine in two year-old mice that the notochord produced the nucleus pulposus of the intervertebral disk. The two Cre lines we created have been deposited at the Jackson Laboratory and can be obtained for a nominal fee from this mouse repository (the Shhgfpcre allele and the ShhcreERT2 allele).

During the course of performing the fate mapping experiments, we noticed that not all notochord cells ended up in the nucleus pulposus. It appeared that even in adult mice, a small number of notochord cells remained within each vertebra (see Fig. 1). Initially, we ignored these cells since we were focusing our efforts on disk formation. Eventually, we realized that we really should investigate these "notochordal remnants" since they were appearing in every animal we analyzed. A search of the literature revealed that notochordal remnants had been described in humans (Hunter et al, 2003; Yamaguchi et al, 2004; Yamaguchi et al, 2005) and were thought to form chordomas but had never been previously observed in mice.

In our fate-mapping experiments, we activated a reporter gene (a "marker" gene) in the mouse notochord, which allowed us to see notochordal remnants. In humans, notochordal remnants are believed to undergo a very rare and still unidentified event leading to chordoma. In our laboratory, we are currently activating genes known to cause cancer in other tissues in the notochord in an attempt to create a mouse model of chordoma. Our hypothesis is that by forcing notochord cells, and as a result notochordal remnants, to express cancer-causing genes, we will create mice that develop chordomas. Since no other type of cancer has been reported to form from notochordal remnants, it is unlikely that forced expression of cancer-causing genes will generate other types of sarcomas.

The Pitfalls and Drawbacks of Using Mice to Model Chordoma

Mice are not humans #1: While at face value we may share 99% of our genes (Waterston et al, 2002), it is the other 1% that makes mice mice and humans human. It is possible that drugs that are designed to treat chordomas in mice will not work as well (or at all) in people. This is true of all treatments that are developed using the mouse model system. But we still need to test treatments in animals before they can be applied to people.

Mice are not humans #2: Humans walk on two legs while mice use all four. It has been suggested that mice are not a good model system for studying intervertebral disk development because we use such different mechanisms to move around (Alini et al, 2008) and put different stresses on our vertebrae. Since chordomas have been proposed to form from cells that normally make the intervertebral disks (Yamaguchi et al, 2005), it is possible that chordomas will not occur in mice by the same mechanisms as in humans. While this may be true, I believe that the formation of chordomas in mice, even if they occur by a pathway not normally occurring in humans, would greatly aid the search for new drugs to treat this deadly disease. Generation of a mouse that contains chordomas that look like and express markers associated with human chordomas will be extremely useful for screening novel drugs, even if mice and humans produce chordomas through mutation of different sets of genes.

What’s Next?

Dr. Rubin in his Op Ed article envisions a research environment in which large numbers of drugs can be quickly and cheaply screened against a bank of sarcoma mouse models to identify the most promising drug for future investigation. Our goal is to successfully create a chordoma mouse model to be used in these types of studies.

By Brian Harfe, PhD
Assistant Professor
Molecular Genetics and Microbiology
University of Florida

References

Alini, M., Eisenstein, S. M., Ito, K., Little, C., Kettler, A. A., Masuda, K., Melrose, J., Ralphs, J., Stokes, I. and Wilke, H. J. (2008). Are animal models useful for studying human disc disorders/degeneration? Eur Spine J 17, 2-19.