Zebrafish (Danio rerio) are an emerging vertebrate model to study lymphocytic cancers. Like humans, D. rerio possess T-, B-, and NK-cells, a marrow for hematopoiesis, and a thymus where T-cell maturation and selection occurs. Virtually all zebrafish genes have human equivalents with conserved molecular functions, and vice versa. Just like people, D. rerio develop T cell cancers such as lymphoblastic lymphoma (T-LBL) and acute lymphoblastic leukemia (T-ALL), and the genes responsible for human and zebrafish T cell cancers are the same. Like human T-LBL and T-ALL, the D. rerio versions of these diseases frequently originate in the thymus or spread to the thymus, forming tumors. 

Since the molecular causes of T-LBL and T-ALL are not fully understood, zebrafish provide a system to investigate the genetic ‘explanations’ for these diseases. Likewise, because zebrafish cancers sometimes regress after treatment with the same medications used to treat human T-LBL and T-ALL, fish can be used to study the genetic basis of why some cancers respond to therapy but others do not. 

Finally, zebrafish with T-LBL and T-ALL can also be used to test new treatments. Therapeutic studies such as these are referred to as ‘pre-clinical testing.’ Besides determining whether a new medicine can effectively treat these cancers, fish can also reveal whether unexpected toxicities exist. Compared to in vitro studies, drug testing in living vertebrate animals has much better ability to predict which medicines may actually be feasible for use in human patients.  

The Frazer lab has several active projects based on these concepts, listed below:

I. Cancer-Prone Mutant Fish
II. Cancer Profiling
IV. Drug Testing

I.  Cancer-Prone Mutant Fish

We study 3 fish lines with unknown mutations that develop T cell cancer. These fish lines were created in an ENU phenotypic screen (Frazer et al, Leukemia, 2009) and dubbed hulk (hlk), shrek (srk), and oscar the grouch (otg). Studying these lines, their cancers, and how their cancers respond to different treatments is the backbone of our work.  We are mapping all 3 mutations to discover the genes responsible. Once identified, we will then analyze these genes to determine: (1) whether that gene is also mutated in human T-LBL and T-ALL, (2) what the normal function of that gene is, and (3) whether the molecular pathway each gene belongs to is perturbed in human T cell cancers. This work will involve combinations of human in vitro and zebrafish in vivo experiments.

A summary of each mutant line follows:

  • hlk is a dominant mutation, i.e., both hlk heterozygotes and hlk homozygotes develop T cell cancer. However, their frequencies are different: by 1 year, only ~6% of hlk heterozygotes develop cancer, but >35% of hlk homozygotes acquire T-LBL and/or T-ALL. We’ve created an extensive pedigree of hlk fish for genetic linkage studies, and collected DNA from over 350 fish that developed cancer. This DNA repository can be used to ‘positionally map’ the hlk mutation. In addition, we have performed exome sequencing from >80 hlk fish. In other words, we sequenced the promoter, 5’ and 3’ untranslated regions, introns, & exons for all known zebrafish genes. This identified a large collection of genetic differences, called single nucleotide polymorphism (SNPs), that distinguish hlk fish from wild-type zebrafish. Our SNP collection serves two purposes: (1) SNPs near the hlk mutation will show ‘linkage’ to hlk, meaning they will cluster near the hlk mutation (accelerating its identification). (2) It is likely that 1 SNP is in fact the hlk mutation, and careful scrutiny of candidate SNPs may be informative. For example, missense and nonsense SNPs, SNPs eliminating splice donor or acceptor sites, SNPs unique to hlk fish that are not present in control fish, and SNPs within known oncogenes and tumor suppressors all represent enticing leads. In addition to exome sequencing, we have also performed RNA microarray analyses of non-malignant hlk T cells as well as hlk cancers. In terms of their transcriptional profile, hlk T cells are very different from wild-type T cells. However, they also show key differences from hlk T-LBL and T-ALL samples. In essence, they represent a ‘pre-malignant’ scenario, with cells that are neither normal nor neoplastic. Understanding this pre-cancerous state makes hlk a fascinating opportunity to gain insight into how cancers develop, and will be even more informative once the hlk mutation has been identified. Lance Batchelor is working to determine the hlk mutation.

  • srk is similar to hlk, as it also shows dominant inheritance. It has higher penetrance than hlk, with ~14% of srk heterozygotes and >40% homozygotes developing T cell cancer by 12 months. Fish from the srk line also have shorter latency (time to disease onset) than hlk. Due to these differences, collecting cohorts of srk fish with T-LBL/T-ALL for mapping will be much easier and faster. Phenotypically, srk and hlk fish are indistinguishable. Both lines develop thymic tumors that spread to the marrow and throughout the animal. Strategies used to identify the hlk mutation will be applied in identical fashion to srk. Likewise, exome sequencing and transcriptional profiling of srk T cells and T cell cancers will be performed.

  • otg is recessively inherited, an important difference from hlk and srk. In other words, only otg homozygotes develop T cell cancer. However, otg has the highest penetrance of our 3 mutant lines, with ~50% of otg homozygotes acquiring T-LBL/T-ALL by one year. Since all otg fish that acquire cancer are known homozygotes, this simplifies mapping. However, otg fish have another feature that is even more valuable: they exhibit an embryonic phenotype. This means that rather than waiting months for cancer to develop, homozygous otg fish can be identified less than 1 week post-fertilization. And their embryonic phenotype is not only convenient, but fascinating. Normally, if wild-type fish embryos are irradiated, this induces massive apoptosis in the central nervous system. In contrast, if otg homozygous embryos are irradiated no apoptosis occurs. This phenotype is identical to p53-mutant fish, but experiments have ruled-out the possibility of otg being a p53 mutation. Similar studies have excluded several other candidate genes from the intrinsic and extrinsic apoptosis pathways. Because regulation of apoptosis is integral to both normal development and oncogenesis, otg holds great promise as a model to study these phenomena. Barbara Squiban is researching otg.
Back to Top

II.  Cancer Profiling

We have investigated the genomes and transcriptomes of zebrafish T cell cancers using array comparative genomic hybridization (aCGH) and RNA microarrays, respectively. We examined normal T cells, pre-cancerous T cells, malignant T cells, and serially-transplanted T cell cancers (a method to select for highly-aggressive cancers, Rudner et al, Oncogene, 2011 ). In so doing, we have identified genes that are recurrently deleted or amplified, and other transcripts that are up- or down-regulated by cellular transformation or iterative transplantation. Depending on the experimental context that resulted in their discovery, these genes represent candidate oncogenes, tumor suppressors, progression factors, or anti-progression factors. These genes either promote or retard oncogenesis, or they enhance or suppress cancers’ aggressiveness once it’s already established. To examine these candidates further, we’re using a combination of in vitro (with human T cell lines) and in vivo (with zebrafish) approaches:

  • Human cell lines: Using cell lines from human T-LBL and T-ALL cases, we first determine genomic copy number & expression patterns for the human homologue of the candidate zebrafish gene. Then, to assess function, we use over-expression or siRNA vectors to modify that gene. After successful gene augmentation or knock-down, we test parameters like growth rate, cell cycle profile, apoptosis resistance, cytokine-independent growth, etc.

  • Transgenic zebrafish: Because transgenesis is straightforward in zebrafish, candidate genes with interesting in vitro properties can be over-expressed using ubiquitous (i.e., b-actin) or T cell-specific (i.e., lck) promoters in transgenic lines. These lines can then be bred to other lines, such as our hlk, srk, and otg mutants. We then determine whether over-expression of that gene is oncogenic, tumor suppressive, or if it alters cancer in vivo aggressiveness. We use several assays for this, including tracking of cancer incidence, latency, engraftment, and progression rates. To assess loss-of-function for tumor suppressor and anti-progression genes, either dominant-negative transgenics or TALEN-engineered knock-outs are utilized in these same assays. 
Back to Top


Besides identifying recurrently-amplified and -deleted genes shared by zebrafish and human T cell cancers (Rudner et al, Oncogene, 2011), our aCGH studies also revealed an unexpected finding: recurrent amplification of a zebrafish retrovirus called ZFERV (Frazer et al, Advances in Hematology, 2012). Retroviruses are single-stranded RNA viruses that are reverse transcribed into DNA and then integrate into their hosts’ genome. If they integrate into the germline (i.e., genomic DNA in spermatogonia or oogonia), they become a permanent part of their host’s genome and can be passed from generation-to-generation. Once this occurs, the retrovirus is termed an ‘endogenous retrovirus’ (ERV). It is estimated that 2-10% of the human genome is made up of ancestral ERV integrants. ERV are typically inactivated by mutations and other silencing mechanisms, rendering them transcription, integration, and replication incompetent. In contrast, ZFERV has none of these features: its ORFs remain intact, and retroviral transcripts are abundant in zebrafish T cells, proving it is transcriptionally-competent and suggesting that retroviral proteins are likely made. ZFERV has integrated into multiple zebrafish genomic loci, showing it remains replication- & integration-competent too. Moreover, in zebrafish T cells and T-LBL/T-ALL cells, we have demonstrated that even more copies of ZFERV are present.

While we have shown that new ZFERV integrations occur, we have not shown where these new retroviral insertions occur. It is likely they differ from T cell-to-T cell and cancer-to-cancer. Identifying new integration sites is of paramount importance, because we suspect new ZFERV insertions contribute to oncogenesis by disabling tumor suppressor genes and/or integrating upstream of proto-oncogenes and augmenting their transcription. The compelling precedent for this hypothesis is that ZFERV’s mammalian relatives—the murine leukemia-related virus (MLV) class of retroviridae—are oncogenic by an insertional mutagenesis mechanism. To prove this hypothesis, we are identifying new ZFERV integration sites in T cell cancers. Besides answering whether new ZFERV insertions occur in oncogenic loci, since cancers are clonal populations, ZFERV integration patterns could be used to track clonal evolution in a particular malignancy. Nonetheless, if our hypothesis is correct, identifying ZFERV integrations sites is likely to reveal unrecognized oncogenes and tumor suppressor genes, which would make it an incredibly powerful tool.  

Lastly, we have shown that ZFERV transcription is higher in cancer-prone T cells (T cells from our hlk mutant as well as MYC transgenic fish) that it is in normal T cells. This suggests that retroviral silencing is impaired in these lines, but we do not understand why. Furthermore, both hlk- and MYC-driven cancers have even higher ZFERV transcription levels, independent of genomic copy number. Because high ZFERV transcription correlates with cancer-predisposition and outright malignancy, this implies that one or more ZFERV transcripts/proteins may have oncogenic properties independent of ZFERV’s ability to act as an insertional mutagen. 

Back to Top

IV.  Drug Testing

Zebrafish with T-LBL/T-ALL respond to the same therapies used in people with these diseases. We have used g-irradiation and the synthetic glucocorticoid dexamethasone (DXM) to induce remissions (disappearance of all detectable cancer) in fish. For DXM, we simply add the drug to water. Fish absorb it through their gills and trans-dermally. We call this system an ‘immersion assay.’ Since their cancers are GFP+, we quantitate responses using fluorescence microscopy.

Glucocorticoids are the key drugs used to treat lymphocytic cancers, including T-LBL and T-ALL. Of course, what this also means is that all patients who die of these diseases have cancers that have either innate or acquired steroid resistance, so it is a huge clinical problem. In our trials, ~90% of fish with MYC-driven cancer respond to DXM. However, only ~10% of hlk fish respond to DXM. This is a perfect scenario to study why some cancers respond to steroid treatment, but others are refractory. With 2 responsive populations (90% of MYC fish, 10% of hlk fish) and 2 non-responsive populations (10% of MYC fish, 90% of hlk fish), we can identify the genetic factors underlying steroid resistance. In addition, since many cancers go into remission after irradiation or DXM treatments but then go on to relapse, we can also query the factors that allow this to happen.

Besides studying established therapies, we also use zebrafish with T-LBL and T-ALL to test new medicines. If a new drug is water-soluble and absorbable by fish, we can test it in immersion assays. For new agents discovered using test tube in vitro assays, this is the first opportunity to see not only whether they work in a living animal, but whether they have toxic side effects. To date, we have tested two new tyrosine kinase inhibitors, two histone deacetylase inhibitors, an mTOR inhibitor, and a new medicine developed from a zebrafish-based drug screen (Ridges et al, Blood, 2012). We anticipate testing many more drugs by immersion assay, and for each, if some fish fail to respond or go on to relapse later, we can also examine what it was about those particular cancers that allowed them to do so. 

Back to Top