New mouse models of cancer: Single-cell knockouts

Cancer is a clonal disease, initiating from a single cell that has accumulated sufficient genetic damage to cause uncontrolled growth. The malignant daughter cells within a clone interact with each other and their normal neighboring cells, the so-called microenvironment, influencing tumor progression (1). The time required to clonally expand a single malignant cell to a clinically detectable tumor can take many years. During this time, other genetic changes occur, leading to the evolution of the tumor phenotype. One fundamental mechanism that triggers cancer formation is the mutation of tumor suppressor genes followed by loss of heterozygosity (2). Tumor suppressor gene mutations in the mouse have been used extensively to model human cancer (3). However, current mouse models, including germ-line mutations and tissue-specific knockouts, have not been able to reproduce the clonal nature of human cancer (Fig. 1 Left and Center). In this issue of PNAS, Muzumdar et al. (4) and Wang et al. (5) exploit genetic technologies modeled after systems in Drosophila to induce a second hit in a tumor suppressor gene in very few somatic cells in mice (Fig. 1 Right). The exciting findings show that the resulting few homozygous mutant cells surrounded by otherwise normal cells behave differently than in mice in which every cell of the body or a particular organ is mutant. The ability to generate clones of mutant cells and follow their behaviors (e.g., cell cycle status or tumor formation) in the mouse offers new opportunities to determine the cellular mechanisms of tumor formation and more accurately model human cancer.

Fig. 1.

Mouse tumor suppressor gene mutation models. (Left) Homozygosity for a germ-line mutation. All cells of the animal (box) are homozygous mutant (red). (Center) Tissue-specific knockout. Tissues (circles) can be induced to become homozygous mutant (red circle). (Right) Mosaicism induced by mitotic recombination. Only very few cells are homozygous mutant (red dots).

In Drosophila, clones of homozygous mutant and wild-type cells can be created during development in heterozygous flies by mitotic recombination, creating genetic mosaics for in vivo analysis of gene function (6). Fly geneticists now use 34-bp sequences called Flpase recombination target (FRT) sites, located on homologous chromosome arms, and the DNA recombinase called Flpase, which specifically recognizes FRT sites, to induce interchromosomal recombination. For mutant analysis, a mutation of interest is recombined onto the FRT-containing chromosome telomeric to the FRT site. After Flpase-induced recombination, cell division yields wild-type and homozygous mutant daughter cells that can proliferate to generate genetically distinct cellular clones. In addition, the different genotypes of cells within each clone (homozygous mutant or wild type) and the neighboring heterozygous cells can be labeled with visual markers for cellular resolution (6, 7).

Recently, mitotic recombination systems have been developed for the mouse that result in <0.001–1% of cells undergoing recombination (8). The mosaic analysis with double markers (MADM) system developed by Luo and coworkers (8) is similar to Drosophila mitotic recombination systems, using interchromosomal recombination caused by DNA recombinases, unambiguously labeling wild-type, heterozygous, and homozygous mutant cells. Instead of using the Flp/FRT system, MADM uses the related Cre/loxP recombination system (9). There are two versions of MADM. GR is designed to yield homozygous mutant cells labeled by GFP, wild-type cells labeled by a red fluorescent protein (DsRed2), and heterozygous cells labeled yellow by a combination of GFP and DsRed2. Conversely, RG can yield homozygous mutant cells labeled red, wild-type cells labeled green, and heterozygous cells labeled yellow. Thus, the MADM system provides single-cell resolution for genetic mosaic analysis in the mouse. Until now, mitotic recombination systems had not been used in mammals for mutation analysis. The papers discussed here (4, 5) provide the first mutation analysis using mitotic recombination in the mouse. Both papers focus on the in vivo analysis of tumor suppressor genes.

To understand the consequences of deleting a tumor suppressor, Wang et al. (5) examined loss of the classic tumor suppressor, p53 (10), by using a mitotic recombination system they developed, placing FRT/loxP cassettes at the D11Mit71 locus located at 1.1 cM (i.e., very close to the centromere) on chromosome 11. p53, like most genes on chromosome 11, is located distal to D11Mit71. A comparison of mice heterozygous for p53 with mice heterozygous for a p53-null mutation on the FRT-marked chromosome 11 and FLP expression (called p53-FLP mice) showed important similarities and differences in tumor development. First, the survival curves of the two kinds of mice were very similar. These data suggest that the second hit that occurs spontaneously in a p53^+/− background and p53 loss induced by mitotic recombination occur with similar efficiency, emphasizing the value of using mitotic recombination to generate sporadic hits. Several important differences among the kinds of tumors that developed in these mice were also noted. p53-FLP mice had an increase in the number of carcinomas as compared with p53^+/− mice. Additionally, some rare tumor types were observed that include pancreatic duct adenocarcinoma and epithelial mesotheliomas of the pleural cavity. p53-FLP mice also developed dysplasia and carcinomas in situ in the skin, a phenotype not observed in p53^+/− mice. There may be several reasons for these important differences. One possibility is that the frequency with which a cell undergoes the second hit differs among cell types in spontaneous and induced systems in the mouse. The ability to generate the second hit randomly in p53-FLP mice exposes additional cell types to the probability of becoming a tumor. Thus, we no longer depend on unknown mechanisms that may differ among cells to generate the second hit. This system also should not discriminate among targeting stem cells or more fully differentiated cells and may lead to a deeper understanding of the nature of the cell of origin of human cancers. Second, and just as important, a clone of cells lacking p53 develops among a normal cell context, as occurs in human cancers. The microenvironment of a tumor cell is crucial for angiogenesis and metastatic potential. Recent studies in the mouse indicate the importance of a normal microenvironment on tumor evolution. In neurofibromatosis, for example, Nf1 loss in Schwann cells required an Nf1 heterozygous environment for tumor development (11). In prostate cancer, inhibition of retinoblastoma function in the prostate epithelium cooperated with p53 alterations in the stroma for tumor development (12). Thus, this new mitotic recombination system provides a better model of the single-cell events that lead to human cancers.

Muzumdar et al. (4), using the MADM system, examine mice with mosaic loss of the cell cycle inhibitor p27kip1 and further show the value of this sophisticated system for cell autonomous studies. The ability to mark p27kip1 wild-type and null cells was invaluable in this study. Clonal populations of cells lacking p27kip1 expanded to a much greater extent than their wild-type siblings (so-called “twin spots”) in a wild-type environment, as compared with mice lacking p27kip1 in all cells. Specifically, in the development of the central nervous system, a 6-fold increase in clonal expansion of p27kip1-null cells was observed. Expansion was due to a decreased ability to exit the cell cycle and thus an increase in the number of cells lacking p27kip1. The authors (4) suggest multiple reasons for the clonal expansion of p27kip1-null cells. The first reason is the cell-autonomous nature of proliferation that evades signals regulating organ size. Organ size thus depends on total cell number and not on the kind of individual cells that populate that mass. Another possibility is that this novel system allows acute deletion at later developmental time points. Last, the neighboring cells are normal.

The mosaic analysis with double markers system is not limited to tumor suppressor genes.

All of these potential mechanisms relate to tumor cell growth as well. Indeed, Muzumdar et al. (4) examined the effects of p27kip1 loss in the pituitary gland, because p27kip1-null mice develop pituitary adenomas. The authors (4) did not find clonal expansion of p27kip1-null cells in the pituitary gland, suggesting that a non-cell-autonomous mechanism is responsible for pituitary adenomas.

Our future understanding of the mechanisms of human tumor development looks bright, because this system allows analysis of a clonal population of cells lacking any known tumor suppressor in an otherwise normal environment. However, the limitation of these two novel systems at the moment is that only chromosomes 6 and 11 are marked. The mark on chromosome 6 is in the Rosa26 locus at 48.7 cM, limiting its utility to genes distal to this position. In addition, few tumor suppressor genes have been mapped to mouse chromosome 6. On the positive side, mosaic analysis using the MADM system is not limited to tumor suppressor genes. In contrast, essentially all of chromosome 11 can be screened by mosaic analysis, because the mark is located at 1.1 cM. In addition, chromosome 11 is dense in tumor suppressor genes, including p53, Brca1, Nf1, and others. Unfortunately, in contrast to chromosome 6, the system for chromosome 11 does not result in marked cells. Expansion of these mitotic recombination systems to other mouse chromosomes is necessary. Another advantage of these systems is that they will also allow coupling to tissue-specific Cre- and FLP-expressing mice that will target a specific organ. This capability is critical when the deletion of a tumor suppressor gene in the whole mouse results in one dominating tumor type that kills the mouse before the onset of other tumor phenotypes, as is the case for p53-null mice (13, 14). These marked chromosomes for mitotic recombination will also facilitate the identification of novel tumor suppressors using genetic recessive screens in the mouse.

Mosaicism in tissue-specific knockouts has been considered a technical flaw by mouse geneticists. However, the two papers discussed here (4, 5) clearly demonstrate that single-cell knockouts can provide great insight for understanding the cellular mechanisms that lead to tumor development.