Carcinogenesis Advance Access originally published online on September 8, 2005
Carcinogenesis 2005 26(12):2031-2045; doi:10.1093/carcin/bgi223
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
REVIEW |
Tumor suppressor genetics
Fred Hutchinson Cancer Research Center, C1-015, 1100 Fairview Avenue N, PO Box 19024, Seattle, WA 90109, USA
* To whom correspondence should be addressed. Email: cjkemp{at}fhcrc.org
 |
Abstract |
|---|
Top
Abstract
Defining tumor suppressor genesThe observation that mutations in tumor suppressor genes can have haploinsufficient, as well as gain of function and dominant negative, phenotypes has caused a reevaluation of the ‘two-hit’ model of tumor suppressor inactivation. Here we examine the history of haploinsufficiency and tumor suppressors in order to understand the origin of the ‘two-hit’ dogma. The two-hit model of tumor suppressor gene inactivation was derived from mathematical modeling of cancer incidence. Subsequent interpretations implied that tumor suppressors were recessive, requiring mutations in both alleles. This model has provided a useful conceptual framework for three decades of research on the genetics and biology of tumor suppressor genes. Recently it has become clear that mutations in tumor suppressor genes are not always completely recessive. Haploinsufficiency occurs when one allele is insufficient to confer the full functionality produced from two wild-type alleles. Haploinsufficiency, however, is not an absolute property. It can be partial or complete and can vary depending on tissue type, other epistatic interactions, and environmental factors. In addition to simple quantitative differences (one allele versus two alleles), gene mutations can have qualitative differences, creating gain of function or dominant negative effects that can be difficult to distinguish from dosage-dependence. Like mutations in many other genes, tumor suppressor gene mutations can be haploinsufficient, dominant negative or gain of function in addition to recessive. Thus, under certain circumstances, one hit may be sufficient for inactivation. In addition, the phenotypic penetrance of these mutations can vary depending on the nature of the mutation itself, the genetic background, the tissue type, environmental factors and other variables. Incorporating these new findings into existing models of the clonal evolution will be a challenge for the future.
Abbreviations: FAP, familial adenomatous polyposis; VNTR, variable number of tandem repeats
|
Defining tumor suppressor genes |
|---|
The categorization of ‘cancer genes’ into dominant-acting oncogenes and recessive tumor suppressors is rooted in historical context. The first ‘cancer genes’ identified were primarily derived from cancer-causing viruses and were found to transform cells in a dominant fashion. These genes became known as oncogenes due to their ability to drive oncogenesis. Evidence for so-called ‘antioncogenes’ or tumor suppressors was less forthcoming. Early experiments with somatic cell hybridization had suggested that tumor suppressor genes existed and were recessive, that is tumor suppressor genes must be completely inactivated for malignancy to occur. In 1969, a series of somatic cell hybrid fusion experiments demonstrated that A9 cells suppressed the tumorigenicity of malignant cells (1), indicating that a factor responsible for the suppression of malignancy likely existed in the A9 cells and that this factor was lost in malignant cells. Subsequent work by Al Knudson among others both predicted the existence of human tumor suppressors and led to the ‘two-hit’ model of tumorigenesis (2,3).
In a seminal analysis comparing unilateral and bilateral retinoblastoma patients, Knudson first delineated his hypothesis that the dominantly inherited form of retinoblastoma and the nonhereditary form are mechanistically linked (2). In the hereditary form, he postulated that one mutation (the first ‘hit’) is inherited while a second mutation occurs in somatic cells significantly accelerating the onset of retinoblastoma and often leading to bilateral forms of the disease. In the nonhereditary form, two mutations must occur somatically prior to retinoblastoma inititiation. Following the cloning of the retinoblastoma gene, RB, in 1986, Knudson's elegant and unifying prediction appeared to be born out (4). Loss of heterozygosity (LOH) and mutation analysis of RB revealed biallelic mutations in retinoblastoma as well as other tumor types and demonstrated that both hereditary and nonhereditary retinoblastomas sustain mutation or loss of both alleles of RB. This precedent has developed into one of the most dominant paradigms in cancer research leading to the tenet that tumor suppressor genes are recessive at the cellular level, requiring complete loss of function in order to reveal a phenotype. Conversely, germline mutations of tumor suppressor genes function dominantly at the organismic level, predisposing the carrier to early onset of disease by supplying one of the required two hits at birth.
In the years that followed, it became clear that RB was only the first example of an ever-expanding class of cancer gene, the tumor suppressor. In 1969, Li and Fraumeni described a dominantly inherited familial syndrome of cancer including rhabdomyosarcoma, osteosarcoma and breast cancer in female carriers (5). By analogy with RB, Li-Fraumeni syndrome was later ascribed to mutation in the TP53 gene and p53 was reinterpreted as a tumor suppressor rather than the oncogene that it was thought to be when originally identified in its mutant form (6,7). Biallelic mutations in TP53 were frequently (though not always) found in tumors, solidifying the belief that tumor suppressor genes require two hits for inactivation (Figure 1). The conventional wisdom that tumor suppressor genes are recessive, then, is based on historical context, not a priori logic.
|
The two-hit model, however, is difficult to reconcile with another major paradigm of cancer research, Nowell's hypothesis of clonal evolution (8). If loss of one tumor suppressor allele truly has no phenotype, then both alleles would need to be lost within a single cell before any selective advantage could occur. However, the spontaneous mutation rate is very low in cells, and accumulation of multiple mutations is required to transform normal cells to a fully malignant tumor. Given the rate of cancer, Loeb has argued that cancer cells must have a mutator phenotype to supply the requisite number of mutations within an individual's lifetime (9). If the loss of one tumor suppressor allele had no selective advantage it would be even more difficult to account for the disparity between known mutation rates and human cancer incidence. As discussed next, tumor suppressor genes are not a unique class of genes with respect to their genetics and, like many other genes involved in metabolism and development, can be haploinsufficient, showing dramatic phenotypes with loss of only a single allele.
|
Origins and definition of haploinsufficiency |
|---|
In diploid organisms, with the exception of genes on the X and Y chromosomes and imprinted genes, two functional copies (alleles) of all genes are present. Nevertheless, for many genes, a single functional allele is sufficient to maintain normal operations. Some genes, and certain functions of other genes, however, are very sensitive to gene dosage levels. Haploinsufficiency represents that special circumstance in which one working allele of a gene is insufficient to accomplish the normal activity of that gene product in the cell. Drosophila geneticists described the concept of haploinsufficiency as early as the turn of the last century. ‘Minute’ mutations were first identified as dominant alleles that are lethal when homozygous. These minute mutations in many instances turned out to be chromosomal deficiencies. In the case of the haplo-IV minute in which the entire fourth chromosome is missing, it was found that a haploid complement of chromosome IV genes resulted in tardy development, reduced fertility and heavy mortality (10–12). In Genetics, a seminal journal of the early genetics field, the term ‘haplo-insufficiency’ was first applied by Curt Stern in his study of dosage effects on the cubitus interruptus allele of Drosophila (13) and was not used again until 1956 in describing one of Bridges' original deficiencies, M33a (14). Recently, however, there has been increasing interest in haploinsufficiency and the terms ‘haplo-insufficiency’ and ‘haploinsufficiency’ have appeared in more than 102 journal articles published in the journal Genetics alone during the last decade while at least 984 references in PubMed list ‘haploinsufficiency’ as a keyword (http://www.ncbi.nlm.nih.gov, August 11, 2005).
|
Haploinsufficient tumor suppressor genes |
|---|
Despite the long history of haploinsufficiency in genetics, translation of this concept to tumor suppressor genes has been slow. The delay is likely due to the lack of experimental evidence as well as a perceived conflict between haploinsufficiency (one allele is not enough) and the original definition of a tumor suppressor gene (two hits required). In traditional tumor suppressor genetics, inherited loss of one tumor suppressor allele leads to accelerated tumorigenesis due to the need to inactivate only one remaining allele. Haploinsufficient tumor suppressor genes also lead to accelerated tumorigenesis, however, without the requirement for inherited mutation of one allele (Figure 1).
In 1998, Fero et al. (15) reported that the cyclin-dependent kinase inhibitor p27kip1 is haploinsufficient for tumor suppression and Venkatachalam et al. (16) showed that p53 could suppress tumor development in a gene dosage-dependent manner. Similarly, Tang et al. (17) reported that Tgfß is haploinsufficient for tumor suppression, although its mechanism of haploinsufficiency is fundamentally different as Tgfß is a secreted protein and thus, functions in a non-cell-autonomous fashion.
In 1996, three labs independently created p27 knockout mice and reported similar phenotypes (18–20). p27–/– mice have an increased growth rate and adults are 20 to 30% larger than wild-type littermates. The increased size of the p27–/– animals was due to increased cellularity as opposed to increased cell size, pointing to a key role of p27 in controlling growth in all tissue compartments in vivo and providing the first clue to its gene dosage sensitivity. Interestingly, tissues from the p27+/– mice expressed roughly 50% the normal level of p27 protein and these mice showed an intermediate growth rate. The intermediate phenotype of the heterozygous animals indicated that control of proliferation and even adult animal size is extremely sensitive to the level of p27 protein. In addition, p27-deficient mice exhibited hyperplasia of the pituitary intermediate lobe and nearly 100% of p27–/– mice (on a 129/Sv genetic background) eventually succumbed as a result of benign pituitary adenomas (18). p27–/– mice have not demonstrated increased susceptibility to spontaneous tumor development in other tissues.
There is, however, an abundance of evidence from human cancers indicating that reduced p27 protein is associated with more aggressive tumors and reduced patient survival (reviewed in ref. 21). As cancer is a multistep process requiring several genetic events, a lack of spontaneous tumor predisposition in p27-deficient mice was insufficient evidence to rule out a tumor-suppressing role for p27. Other genetic events might be required to elicit latent tumor suppressing effects by p27. Indeed when p27-deficient mice were challenged with either the point mutagen ENU, or the broad spectrum mutagenic agent ionizing radiation, they showed tumor predisposition in multiple epithelial tissues (15). In a cohort challenged with a single dose of 1 Gy radiation, the median tumor-free survival was reduced from more than 70 weeks in wild-type control mice to 40 weeks in p27–/– mice due to increased tumor multiplicity in diverse sites including small and large intestine, lung, ovary, uterus and adrenal gland. p27+/– mice showed an intermediate susceptibility, both in tumor-free survival and tumor multiplicity. Genetic and biochemical analysis of tumors from the p27+/– mice revealed that the wild-type p27 allele was not mutated and protein expression was not silenced in p27+/– tumors. Similarly, when p27-deficient mice were challenged with the carcinogen 1,2-dimethylhydrazine (DMH), an alkylating agent that induces adenomas and adenocarcinomas specifically in the colon, overall tumor-free survival was significantly reduced in p27–/– mice and to an intermediate extent in p27+/– mice relative to wild-type littermate controls (22). The incidence of colorectal adenocarcinoma as well as the ratio of adenocarcinomas to adenomas and the histological aggressiveness of tumor behavior were all significantly increased in p27-deficient mice. While p27+/– mice exhibited a colorectal tumor latency and histologically aggressive tumor behavior intermediate between p27–/– and p27+/+ mice, the remaining wild-type allele was retained and continued to be expressed in DMH-induced colorectal tumors from p27+/– mice.
These results differed from those reported previously in other murine models of tumor suppressor knockouts. For example, tumors from Rb+/– (23,24) and Apc+/– (25) mice all show frequent mutation or loss of the remaining wild-type allele, consistent with Knudson's two-hit model (2). In these cases the tumor suppressor appears recessive at the cellular level: complete loss of both alleles provides a much greater selective advantage than loss of a single allele. Even in these cases, however, the possibility that the loss of one allele confers a selective advantage perhaps early in tumor development cannot be excluded. The genetic inactivation of murine Rb and Apc genes mimics that seen in human tumors where biallelic mutations in these tumor suppressor genes are observed. In contrast, biallelic mutations in p27 are rarely seen in either human or murine tumors. The murine data unequivocally show a strong selective advantage to tumor development with loss of a single p27 allele. The data indicate that inhibition of growth (18) and tumor development (15) is highly sensitive to the gene dosage of p27. Thus, halving the normal amount of p27 is sufficient to result in unchecked growth, suggesting that there may not be a threshold for its activity but rather a dose–response continuum.
In another well-known example, Trp53 is inactivated via traditional ‘two-hit’ kinetics under certain circumstances and in certain tissues, but in other cases shows clear evidence for haploinsufficiency. Early evidence for haploinsufficient behavior of Trp53 at the cellular level was observed by Bouffler et al. (26). In this study, p53+/– mice showed a significantly higher number of spontaneous chromosome aberrations as compared to p53+/+ mice, with the incidence being intermediate between that of the p53+/+ and p53–/– mice. Similarly, Clarke et al. (27) reported that apoptosis is partially impaired in p53+/– mice. Venkatachalam et al. (16) later confirmed the haploinsufficiency of Trp53 by analysis of tumors from p53+/– mice. In a study of 217 p53+/– mice, roughly half of all tumors retained the wild-type p53 allele. In contrast to p53+/– tumors that lose the remaining wild-type allele, those tumors with wild-type allele retention expressed a functional p53 protein that preserved the ability to induce apoptosis following irradiation, to induce p21 and Mdm2 expression, and to repress PCNA expression.
Since 1998, a burgeoning number of tumor suppressor genes have shown evidence of haploinsufficiency (Table I). One of the recent examples with the best understood mechanism of haploinsufficiency is Nkx3.1. Nkx3.1 encodes a homeobox protein that is expressed specifically in the luminal epithelium of the prostate. Nkx3.1+/– mice develop prostatic hyperplasia without loss of the remaining wild-type allele (28). Microarray analysis identified a number of Nkx3.1 target genes that represented a range of responses to Nkx3.1 gene dosage (29). Some genes, e.g. probasin were relatively insensitive to gene dosage with 70 to 80% of normal expression in Nkx3.1 hemizygotes whereas expression of other genes such as intelectin was lost in Nkx3.1 hemizygotes. In between the two extremes were a number of genes whose expression in Nkx3.1 hemizygotes ranged from 16 to 55% of wild-type expression. The authors argued for a stochastic model of gene expression in which Nkx3.1 gene dosage affects the probability of target gene expression in any given cell. In the case of Nkx3.1, evidence of haploinsufficiency depended largely on the specific phenotype studied (i.e. probasin expression versus intelectin expression), indicating the difficulty inherent in identifying a multifunctional tumor suppressor as haploinsufficient or not.
|
In most cases, demonstration of the haploinsufficiency of a tumor suppressor gene has only been possible in the more rigorously controlled genetics of mouse models. One exception is the CBFA2/AML1 tumor suppressor. Song et al. (30) were able to demonstrate heterozygous missense mutations of CBFA2 segregate with familial platelet disorder with predisposition to acute myelogenous leukemia in four different pedigrees. In the leukemic cells from these patients, no somatic mutations in the coding sequence of the wild-type allele were identified and no evidence for deletion of the wild-type allele was observed. Further, the CBFA2 protein was expressed in leukemic cells and 100% of metaphases from the leukemic bone marrow contained the karyotypic marker for the chromosome carrying CBFA2. Apart from such exceptions, evidence for the existence of haploinsufficent tumor suppressors in humans has also been suggested by cell culture experiments. In chromosome transfer studies with breast cancer cell lines it was determined that a locus on the short arm of chromosome 8, a site of frequent LOH in human breast cancers, also behaves consistently with a haploinsufficient mechanism of tumor suppression (31). The researchers transferred chromosome 8 into breast cancer cell lines in which only one allele was present for all 8p microsatellite markers analyzed. In all cases, the presence of two full copies of chromosome 8p was incompatible with cell growth and in at least one case the donor chromosome 8 was retained while the recipient chromosome 8 was lost, as opposed to the expectation for ‘two-hit’ tumor suppression. This is similar to a report by Islam et al. (32) in which introduction of a derivative chromosome 9 was accompanied by loss of the recipient chromosome 10 and hints at a potential method, albeit labor intensive, for testing the haploinsufficiency of human tumor suppressors in cases where appropriate data on inherited mutations is not available.
For some haploinsufficient genes, expression profiling has proved a useful tool for dissecting the biological pathways affected by these genes (33). Recently, an attempt to identify early molecular changes associated with dominantly inherited predisposition to renal tumors via expression profiling (34) found that heterozygosity for either the von Hippel-Lindau tumor suppressor or the tuberous sclerosis complex genes (including the haploinsufficient renal tumor suppressor gene TSC2) significantly altered the expression profile of phenotypically normal renal epithelial cells in a gene-specific manner. Examination of transcriptional profiles for cells bearing a single mutation in a tumor suppressor gene may yield insight into the haploinsufficient phenotype(s) of other tumor suppressor genes, provided we develop appropriate guidelines for distinguishing dominant negative mutations from haploinsufficiency.
In humans and mice, however, the known haploinsufficient tumor suppressors likely represent an underestimate of the true number due to the inherent difficulty in reliably demonstrating an absence of mutations in both alleles as well as a lack of published experiments thoroughly examining the phenotype of hemizygotes. In addition to haploinsufficiency, tumor suppressor gene mutations can also lead to dominant negative and gain of function effects as described next.
|
Recessive versus dominant alleles |
|---|
Diploid organisms have two alleles for each autosomal gene. In simple Mendelian genetics, a dominant allele confers a phenotype in either the heterozygous or the homozygous state whereas a recessive allele confers a phenotype only in the homozygous state. Some of the earliest evidence for dominant and recessive alleles was observed in flowering pea plants. In Mendel's case, for example he found that round seed shape (R) was dominant and wrinkled seed shape (r) was recessive. The history of science is fraught with myth, however, and it is now thought that Mendel may have ‘rounded the rough edges’ of his data in order to simplify his conclusions. Thus, even from the beginning, the distinction between a dominant and a recessive allele was fuzzy. Words and concepts can help to clarify biological processes, but they can also limit our understanding of those same processes by imposing boundaries that do not exist in nature. The traditional view of genetics is that most recessive alleles are non-functional or have reduced functionality. However, the nature of dominant alleles is less obvious.
|
True nulls versus hypomorphic recessive alleles |
|---|
Researchers can begin to characterize the function of a gene by the creation of an allelic series, that is a series of unique mutations within the gene of interest. In an allelic series, a true null allele has the strongest phenotype and is genetically indistinguishable from a deficiency of the gene (that is a complete absence of the chromosomal region encoding the gene). By contrast, a hypomorphic recessive allele has reduced, but not completely absent, functionality. Functionality for a hypomorphic allele may be reduced, or in the case of genes that are responsible for multiple functions, one function may be affected while the others are preserved. Thus, even for tumor suppressors that appear to follow Knudsen's two-hit model, functionality at a given gene dosage will vary depending on whether the inactivating mutation is hypomorphic or a true null allele (Figure 2).
|
The distinction between a true null allele and a hypomorphic allele can sometimes be difficult to make. For example, the Apcmin mutation, which results in murine intestinal neoplasia via a classic two-hit mechanism, is often thought to be a loss of function allele. The Apcmin allele produces a truncated protein of roughly 850 amino acids. The Apc1638T mutation, in which the targeted mutation results in a stable but prematurely truncated protein containing 788 amino acids more than the Apcmin truncated protein, is homozygously viable (unlike either Apcmin or Apc1638N) and has no observable tumor phenotype. However, the germline Apc1638N, mutation in which almost no Apc1638N protein product is produced, results in reduced tumor multiplicity and increased tumor latency as compared to the Apcmin allele. Thus, Apc1638N, a mutation in which almost no protein is produced has a less severe phenotype than Apcmin, in which a stable albeit truncated protein is produced. So, is the Apcmin allele a gain of function allele? If so, then why is the remaining wild-type allele still selectively lost in most tumors from Apcmin/+ mice? One possible explanation is that Apc is gene dosage-dependent. Thus impairment of APC function might provide gastrointestinal epithelial cells with a selective advantage, with loss of the remaining wild-type allele providing a further selective advantage. Regardless, it is clear that a more flexible paradigm involving both quantitative as well as qualitative changes in function is required to fully appreciate the spectrum of tumor suppressor gene action.
|
Dominant alleles of tumor suppressor genes |
|---|
A haploinsufficient tumor suppressor mutation is, by nature, difficult to distinguish from a dominant negative mutation. In both cases, the wild-type allele is retained, although the reason for retention is drastically different in each case. In the case of a dominant negative mutation, the wild-type allele does not need to be inactivated because the dominant negative mutation serves that function, often by binding the wild-type protein in non-functional complexes. In the case of a haploinsufficient mutation, the wild-type allele is retained because half the normal complement of wild-type protein is insufficient for functionality. These two mutation types can often be distinguished by factoring in other known traits of the gene. For example, in the case of p27, it is known that mice hemizygous for p27 show accelerated tumorigenesis without loss of the remaining wild-type allele indicating that one-half the normal complement of p27 is insufficient for tumor suppression (15) and that p27 is haploinsufficient, as opposed to inactivated by dominant negative mutation. In the case of the Wilms' tumor gene (WT1), however, it is known that the WT1 protein dimerizes through the N-terminal domain, indicating a potential mechanism for dominant negative action (35–38). Although milder cancer susceptibility syndromes associated with WT1 mutation follow traditional two-hit kinetics, a severe syndrome of genitourinary malformations associated with susceptibility to Wilms' tumor and due to missense mutations in WT1 nuclear localization signals, is thought to be caused by dominant negative mutations of WT1 (39).
|
Pleiotropy |
|---|
Many genes are pleiotropic, or in other words function in multiple cellular processes, and some of these functions may be more dosage-sensitive than others. For some phenotypes examined, deletion of a single allele may have little effect, while for other phenotypes deletion of a single allele of the same gene may be tantamount to a null phenotype. For example, p53 is a highly regulated gene with multiple inputs and outputs and p53 may be haploinsufficient for some functions, like inducing apoptosis (27) or maintaining genome integrity (26), but recessive for other functions like transcriptional regulation of certain genes. Quantitative phenotypic analyses of hemizygotes will improve understanding of the complex consequences conveyed by deletion of a single allele of a gene. RNAi can be useful in creating an allelic series of gene dosage as was demonstrated recently by Hemann et al. (40) for Trp53 and may be useful in further differentiating between minor changes in gene dosage. The spectrum of phenotypes between full complement of a functional tumor suppressor protein and complete deficiency for that tumor suppressor gene can be broad, especially for genes that are responsible for multiple functions within the cell. If one accepts that the definition of a gene as ‘recessive’ or ‘haploinsufficient’ is dependent on the cellular context of that mutation, then one can easily reconcile the perceived conflict between haploinsufficient tumor suppressor genes and the ‘two-hit’ model of tumor suppressor gene inactivation. The next sections discuss how the phenotypic penetrance of a mutation can be context-dependent.
|
History of modifier genes |
|---|
One of the first Drosophila mutants identified by Morgan was a fly with truncated wings. The inheritance of ‘truncate’ was difficult to reconcile with Mendelian genetics due to the inconstant penetrance of the truncated wing phenotype. As early as 1919, Muller and Altenburg recognized that the truncated wing phenotype in Drosophila could be phenotypically modified by extragenic loci (41).
In subsequent years, genetic screens for both enhancers and suppressors of a mutant phenotype have been used to identify additional members of a signaling network. It is well known that these modifier screens function best for identifying components of pathways sensitive to gene dosage. This strategy has been put to exquisite use in Drosophila. For example, a hypomorphic allele of the receptor tyrosine kinase controlling cell-fate choice in the Drosophila eye, sevenless (sev), provides just enough activity for most of the photoreceptor cells to form. Further reduction of signaling, however, leads to conversion of most of the photoreceptor cells to cone cells. This sensitized background was used to identify mutations in components of the receptor tyrosine kinase signaling pathway as dominant enhancers of the cell-fate phenotype in the eye (reviewed in ref. 42).
|
Modifiers of tumor suppressor gene mutations |
|---|
Numerous enhancers of tumor suppressor gene mutations have been identified. One of the first was the Mom1 or ‘Modifier of Min’ locus on chromosome 4 in mice (43). Apcmin mice in a Mom1 wild-type background, such as the AKR strain, are resistant to intestinal polyposis whereas Apcmin mice in a Mom1 mutant background, such as C57Bl6/J, develop hundreds of intestinal polyps throughout the gastrointestinal tract. Mutation in the Pla2g2a locus, which encodes a secretory phospholipase, was eventually shown to be responsible for 50% of the enhanced intestinal polyposis of Apcmin mice in the C57Bl6/J background (44). Among enhancers identified by a candidate gene approach, deficiency in Mlh1 was shown to accelerate the development of intestinal adenomas in Apcmin mice (45). In addition to its intrinsic tumor suppressor activity, p27 is one of the most promiscuous of these tumor suppressor enhancers. p27 functions as an enhancer of mutations in multiple genes important for human cancer, indicating that p27 may be a nodal component of diverse pathways for tumor suppression (46).
Suppressors of tumor suppressor gene mutations are known as well. Dnmt1 is the most prominent of the three known mammalian DNA methyl transferases. Intestinal tumor multiplicity in Apcmin mice heterozygous for a null allele of Dnmt1 (Dnmt1S/+) was reduced by 60% compared to Apcmin mice alone (47). Using a hypomorphic allele (Dnmt1N/+), Cormier et al. (48) demonstrated that the modifying effect of Dnmt1 on Apcmin is independent of both Trp53 and Mom1 status. When two distinct hypomorphic alleles of Dnmt1 were combined in an Apcmin background, intestinal polyposis was completely suppressed indicating that Dnmt1 is a very specific genetic suppressor of tumorigenesis in Apcmin mice (49). Similarly, deficiency for the human multidrug resistance gene, Mdr1, results in significantly fewer intestinal polyps in Apcmin mice (50). The existence of modifier loci have provided further evidence that the phenotypic expression of a primary mutation is highly dependent on multiple factors, including extragenic loci, tissue type and environmental stress.
|
Mechanism of genetic modifier action |
|---|
Genetic modifiers serve to enhance or suppress the original mutation of interest via a number of different mechanisms. A genetic modifier can affect a parallel or redundant pathway or it can affect the localization of the mutant gene product of interest, thereby either enhancing or suppressing its activity. A genetic modifier sometimes takes the form of a stability factor affecting the gene product of interest. Finally, a genetic modifier can function as a downstream effector in a signaling cascade that amplifies the activity of the mutant gene product, as may be the case for p27.
|
Tissue-specific modifiers of tumor suppressor gene mutations |
|---|
Gene dosage-dependence of a tumor suppressor can vary not only with the genetic background of the organism, but also between tissues. In both DMBA-TPA-induced skin carcinogenesis (51) and Apc mutant mice (22), p27 tumor suppression shows a clear dosage-dependence, with p27 hemizygotes being intermediate to p27+/+ and p27–/– mice with respect to tumor-free survival. In other tissues, however, this is not the case. In the mammary gland, for example, p27–/– mice had longer tumor latency in MMTV-neu transgenics as compared to p27+/+ mice (52). p27+/– mice, however, showed reduced mammary tumor latency as compared to p27+/+ mice, as might be expected based on the established tumor suppressor function of p27. Counter intuitively, mammary tumors from p27–/– mice had a reduced mitotic index as compared to the mammary tumors in p27+/– mice. Similar results were seen in the prostate gland. In Nkx3.1-deficient, Pten+/– mice that were also hemizygous for p27, prostate tumor progression was enhanced whereas in Nkx3.1-deficient, Pten+/– mice that were p27–/–, prostate tumor progression was inhibited (53). This result was specific to prostate epithelium as other tumor types, especially lymphoma, were enhanced in Nkx3.1-deficient, Pten+/–, p27–/– mice. Thus, p27 haploinsufficiency is tissue-specific. Whether the unexpected effects of p27-deficiency seen in the mammary and prostate are due to extragenic tissue-specific modifiers or tissue-specific environmental exposures remains to be seen.
|
Modifiers of human tumor suppressor genes |
|---|
Evidence of tumor suppressor modifier loci has been less forthcoming in humans than in mouse models, likely owing to the controlled genetic background available in mouse experiments (reviewed in ref. 54). Nonetheless, there is strong evidence that such modifier loci exist in humans as well. Some of the strongest evidence comes from studying the variation in disease incidence and severity in familial adenomatous polyposis (FAP) families segregating a mutation in APC (55,56). Variations in the penetrance, number of polyps and the existence of extracolonic manifestations have been observed even within a given a family segregating a single mutation in APC (reviewed in ref. 57). The identification of Mom1 as a modifier of the Apcmin phenotype in mice precipitated efforts to determine whether the human homologue of Mom1 modifies the disease phenotype in human FAP patients. Early evidence indicated a weak linkage between the MOM1 locus at chromosome 1p35–p36 and modification of FAP phenotype (58), however, subsequent investigation failed to support the involvement of MOM1 (59,60). Whether MOM1 is capable of modifying APC in a restricted subset of families or is merely a mouse-specific modifier of Apc has yet to be determined. Modelling of segregation of disease severity in FAP families has indicated that a mixed model in which a single major modifier locus acts in concert with multiple minor loci is most likely to explain the intrafamilial variation in disease phenotype (55), although these modifier loci remain to be mapped. More recently, evidence from studies of human FAP patients has indicated the possible involvement of N-acetyl transferase polymorphisms in modifying disease phenotype in APC carriers (56).
Putative modifiers of the familial breast and ovarian cancer tumor suppressors BRCA1 and BRCA2 have also been identified. The earliest of these was the HRAS1 variable number of tandem repeats (VNTR) locus located 1 kb downstream of HRAS1 (61). Rare alleles of the HRAS1 VNTR increase the risk of ovarian cancer penetrance in BRCA1 mutation carriers by 2.11 times. The simplest explanation of this observation is that these rare VNTR alleles somehow affect the expression of HRAS1, a known proto-oncogene. The occurrence of the VNTR in a non-coding region of the HRAS1 locus does not allow a direct demonstration that HRAS1 is the tumor suppressor modifier in question, however, and the true identity of the BRCA1 modifier gene at this locus has not yet been demonstrated. More recently, polymorphisms of RAD51 (62–64), the androgen receptor (reviewed in ref. 65) and TP53 (66) have been implicated as modifiers of mutations in BRCA1 and/or BRCA2 and a putative BRCA1 modifier locus has been mapped to chromosome 5q (67). In addition to APC, BRCA1 and BRCA2, putative modifiers of tumor suppression by CDKN2A (68), VHL (69) and NF1 (70) have been identified.
|
Summary |
|---|
For the past 30 years, Knudsen's ‘two-hit’ model has provided a useful framework for interpreting the kinetics of tumor suppressor gene inactivation. However the growing list of tumor suppressor genes that exibit haploinsufficiency as well as dominant negative mutations indicates a more complex view. The Platonic view that genes function in a completely dominant or a recessive fashion is oversimplified. Past models of simple on/off switches for gene function do not account for the pleiotropy of phenotypes observed for a given gene. Much as rheostats have gradually replaced the ‘toggle’ light switches of old, models of tumor suppressor gene function must now encompass mutations that modulate activity without abolishing it entirely. In addition, the phenotypic expression of a primary mutation is highly dependent on multiple factors, including genetic background, tissue type, and environmental stress. Recessivity or haploinsufficiency are not absolutes but are context-dependent. Future research will need to focus more carefully on the phenotype of tumor suppressor gene mutation heterozygotes and their pleiotropic effects. A more complete understanding of tumorigenesis will only be achieved by a detailed analysis of tumor suppressor gene dosage effects on cellular phenotypes embedded within the complexity of the organism.
|
Notes |
|---|
The online version of this article has been published under an open access model. Users are entitled to use, reproduce, disseminate, or display the open access version of this article for non-commercial purposes provided that: the original authorship is properly and fully attributed; the Journal and Oxford University Press are attributed as the original place of publication with the correct citation details given; if an article is subsequently reproduced or disseminated not in its entirety but only in part or as a derivative work this must be clearly indicated. For commercial re-use, please contact journals.permissions@oxfordjournals.org
|
Acknowledgments |
|---|
Research in the author's lab was funded by grants from the American Cancer Society, the National Institutes of Health and the Life Possibilities Fund.
Conflict of Interest Statement: None declared.
|
References |
|---|
- Harris,H., Miller,O.J., Klein,G., Worst,P. and Tachibana,T. (1969) Suppression of malignancy by cell fusion. Nature, 223, 363–368.[CrossRef][Medline]
- Knudson,A.G. (1971) Mutation and cancer: statistical study of retinoblastoma. Proc. Natl Acad. Sci. USA, 68, 820–823.
[Abstract/Free Full Text] - Comings,D.E. (1973) A general theory of carcinogenesis. Proc. Natl Acad. Sci. USA, 70, 3324–3328.
[Abstract/Free Full Text] - Friend,S.H., Bernards,R., Rogelj,S., Weinberg,R.A., Rapaport,J.M. and Dryja,T.P. (1986) A human DNA segment with properties of the gene that predisposes to retinoblastoma and osteosarcoma. Nature, 323, 643–646.[CrossRef][Medline]
- Li,F.P. and Fraumeni,J.F.Jr (1969) Soft-tissue sarcomas, breast cancer, and other neoplasms. A familial syndrome? Ann. Intern. Med., 71, 747–752.[ISI][Medline]
- Finlay,C.A., Hinds,P.W. and Levine,A.J. (1989) The p53 proto-oncogene can act as a suppressor of transformation. Cell, 57, 1083–1093.[CrossRef][ISI][Medline]
- Malkin,D., Li,F.P., Strong,L.C. et al. (1990) Germ line p53 mutations in a familial syndrome of breast cancer, sarcomas and other neoplasms. Science, 250, 1233–1238.
[Abstract/Free Full Text] - Nowell,P.C. (1976) The clonal evolution of tumor cell populations. Science, 194, 23–28.
[Abstract/Free Full Text] - Loeb,L.A. (1991) Mutator phenotype may be required for multistage carcinogenesis. Cancer Res., 51, 3075–3079.
[Free Full Text] - Bridges,C.B. (1917) Deficiency. Genetics, 2, 445–465.
[Free Full Text] - Bridges,C.B. (1921) Genetical and cytological proof of nondisjunction of the fourth chromosome of Drosophila melanogaster. Proc. Natl Acad. Sci. USA, 7, 186–192.
[Free Full Text] - Li,J.-C. (1927) The effect of chromosome aberrations on development in Drosophila melanogaster. Genetics, 12, 1–58.
[Free Full Text] - Stern,C. (1943) Genic action as studied by means of the effects of different doses and combinations of alleles. Genetics, 28, 441–475.
[Abstract/Free Full Text] - Abramson,S., Herskowitz,I.H. and Muller,H.J. (1956) Identification of half-translocations produced by X-rays in detaching attached X-chromosomes of Drosophila melanogaster. Genetics, 41, 41–419.
- Fero,M.L., Randel,E., Gurley,K.E., Roberts,J.M. and Kemp,C.J. (1998) The murine gene p27Kip1 is haplo-insufficient for tumour suppression. Nature, 396, 177–180.[CrossRef][Medline]
- Venkatachalam,S., Shi,Y.P., Jones,S.N., Vogel,H., Bradley,A., Pinkel,D. and Donehower,L.A. (1998) Retention of wild-type p53 in tumors from p53 heterozygous mice: reduction of p53 dosage can promote cancer formation. EMBO J., 17, 4657–4667.[CrossRef][ISI][Medline]
- Tang,B., Bottinger,E.P., Jakowlew,S.B., Bagnall,K.M., Mariano,J., Anver,M.R., Letterio,J.J. and Wakefield,L.M. (1998) Transforming growth factor-beta1 is a new form of tumor suppressor with true haploid insufficiency. Nat. Med., 4, 802–807.[CrossRef][ISI][Medline]
- Fero,M.L., Rivkin,M., Tasch,M. et al. (1996) A syndrome of multiorgan hyperplasia with features of gigantism, tumorigenesis, and female sterility in p27-deficient mice. Cell, 85, 733–744.[CrossRef][ISI][Medline]
- Nakayama,K., Ishida,N., Shirane,M., Inomata,A., Inoue,T., Shishido,N., Horii,I. and Loh,D. (1996) Mice lacking p27 display increased body size, multiple organ hyperplasia, retinal dysplasia, and pituitary tumors. Cell, 85, 707–720.[CrossRef][ISI][Medline]
- Kiyokawa,H., Kineman,R.D., Manova-Todorova,K.O., Soares,V.C., Hoffman,E.S., Ono,M., Khanam,D., Hayday,A.C., Frohman,L.A. and Koff,A. (1996) Enhanced growth of mice lacking the cyclin-dependent kinase inhibitor function of p27(Kip1). Cell, 85, 721–732.[CrossRef][ISI][Medline]
- Slingerland,J. and Pagano,M. (2000) Regulation of the cdk inhibitor p27 and its deregulation in cancer. J. Cell Physiol., 183, 10–17.[CrossRef][ISI][Medline]
- Philipp-Staheli,J., Kim,K.H., Payne,S.R., Gurley,K.E., Liggitt,D., Longton,G. and Kemp,C.J. (2002) Pathway-specific tumor suppression. Reduction of p27 accelerates gastrointestinal tumorigenesis in Apc mutant mice, but not in Smad3 mutant mice. Cancer Cell, 1, 355–368.[CrossRef][ISI][Medline]
- Jacks,T., Fazeli,A., Schmitt,E.M., Bronson,R.T., Goodell,M.A. and Weinberg,R.A. (1992) Effects of an Rb mutation in the mouse. Nature, 359, 295–300.[CrossRef][Medline]
- Hu,N., Gutsmann,A., Herbert,D.C., Bradley,A., Lee,W.H. and Lee,E.Y. (1994) Heterozygous Rb-1 delta 20/+mice are predisposed to tumors of the pituitary gland with a nearly complete penetrance. Oncogene, 9, 1021–1027.[ISI][Medline]
- Luongo,C., Moser,A.R., Gledhill,S. and Dove,W.F. (1994) Loss of Apc+ in intestinal adenomas from Min mice. Cancer Res., 54, 5947–5952.
[Abstract/Free Full Text] - Bouffler,S.D., Kemp,C.J., Balmain,A. and Cox,R. (1995) Spontaneous and ionizing radiation-induced chromosomal abnormalities in p53-deficient mice. Cancer Res., 55, 3883–3889.
[Abstract/Free Full Text] - Clarke,A.R., Gledhill,S., Hooper,M.L., Bird,C.C. and Wyllie,A.H. (1994) p53 dependence of early apoptotic and proliferative responses within the mouse intestinal epithelium following gamma-irradiation. Oncogene, 9, 1767–1773.[ISI][Medline]
- Abdulkadir,S.A., Magee,J.A., Peters,T.J., Kaleem,Z., Naughton,C.K., Humphrey,P.A. and Milbrandt,J. (2002) Conditional loss of Nkx3.1 in adult mice induces prostatic intraepithelial neoplasia. Mol. Cell. Biol., 22, 1495–1503.
[Abstract/Free Full Text] - Magee,J.A., Abdulkadir,S.A. and Milbrandt,J. (2003) Haploinsufficiency at the Nkx3.1 locus. A paradigm for stochastic, dosage-sensitive gene regulation during tumor initiation. Cancer Cell, 3, 273–283.[CrossRef][ISI][Medline]
- Song,W.J., Sullivan,M.G., Legare,R.D. et al. (1999) Haploinsufficiency of CBFA2 causes familial thrombocytopenia with propensity to develop acute myelogenous leukaemia. Nat. Genet., 23, 166–175.[CrossRef][ISI][Medline]
- Wilson,P., Cuthbert,A., Marsh,A. et al. (2003) Transfer of chromosome 8 into two breast cancer cell lines: total exclusion of three regions indicates location of putative in vitro growth suppressor genes. Cancer Genet. Cytogenet., 143, 100–112.[CrossRef][ISI][Medline]
- Islam,M.Q. and Islam,K. (1999) Lack of tumor suppression but induced loss of copies of indigenous chromosome 10 in vitro following microcell-mediated transfer of a deleted human der(9)t(X;9) chromosome to Syrian hamster BHK-191-5C cells. Cytogenet. Cell Genet., 87, 11–18.[CrossRef][ISI][Medline]
- Moseley,A.E., Huddleson,J.P., Bohanan,C.S., James,P.F., Lorenz,J.N., Aronow,B.J. and Lingrel,J.B. (2005) Genetic profiling reveals global changes in multiple biological pathways in the hearts of Na, K-ATPase alpha 1 isoform haploinsufficient mice. Cell. Physiol. Biochem., 15, 145–158.[CrossRef][ISI][Medline]
- Stoyanova,R., Clapper,M.L., Bellacosa,A. et al. (2004) Altered gene expression in phenotypically normal renal cells from carriers of tumor suppressor gene mutations. Cancer Biol. and Ther., 3, 1313–1321.
- Haber,D.A., Timmers,H.T., Pelletier,J., Sharp,P.A. and Housman,D.E. (1992) A dominant mutation in the Wilms tumor gene WT1 cooperates with the viral oncogene E1A in transformation of primary kidney cells. Proc. Natl Acad. Sci. USA