Mouse mutagenesis-systematic studies of mammalian gene function

doi:10.1093/hmg/7.10.1627

This Article

	Abstract
	FREE Full Text (PDF)
	Alert me when this article is cited
	Alert me if a correction is posted

Services

	Email this article to a friend
	Similar articles in this journal
	Similar articles in ISI Web of Science
	Similar articles in PubMed
	Alert me to new issues of the journal
	Add to My Personal Archive
	Download to citation manager
	Search for citing articles in: ISI Web of Science (67)
	Request Permissions

Google Scholar

	Articles by Brown, S. D.
	Articles by Nolan, P. M.
	Search for Related Content

PubMed

	PubMed Citation
	Articles by Brown, S. D.
	Articles by Nolan, P. M.

Social Bookmarking

	What's this?

Human Molecular Genetics Pages 1627-1633

Mouse mutagenesis-systematic studies of mammalian gene function
Introduction
Systematic Approaches To The Generation Of Mutations In The Mouse Genome
Genotype-Driven Approaches
Phenotype-Driven Approaches
Genetic approaches to the generation of new phenotypes in the mouse
Phenotypic screens
From Clock To Gene-From Phenotype Screens To Mutation
Conclusion
Acknowledgements
References

Mouse mutagenesis-systematic studies of mammalian gene function

Steve D. M. Brown* and Patrick M. Nolan

MRC Mammalian Genetics Unit and Mouse Genome Centre, Harwell, Oxon OX11 ORD, UK

Received April 30, 1998

The mouse will play a pivotal role in mammalian gene function studies as we enter the post-genomics era. The challenge is to develop systematic, genome-wide mutagenesis approaches to the study of gene function. The current mouse mutant resource has been an important source of human genetic disease models. However, despite an apparently large catalogue of mouse mutations, we have access to mutations at only a small fraction of the likely total number of mammalian genes-there is a phenotype gap that needs to be filled by the establishment of new mutagenesis programmes. Two routes, genotype- and phenotype-driven, can be used for the recovery of novel mouse mutations. For the former, gene trap embryonic stem cell libraries appear set to deliver a large number of mutations around the mouse genome. The advantage of genotype-driven approaches is the ease of identification of the mutated locus; the disadvantage that a priori assumptions have to be made concerning the function and likely phenotype of the mutated gene. In contrast, phenotype-driven mutagenesis emphasizes the recovery of novel phenotypes. One phenotype-driven approach that will play an important role in expanding the mouse mutant resource employs the mutagen N-ethyl-N-nitrosourea (ENU). The phenotype-driven route makes no assumptions about the underlying genes involved, and ENU mutagenesis programmes can be expected to play a significant role in uncovering novel pathways and genes; the disadvantage is that the identification of the mutant gene is still not trivial. Together, the complementary routes of genotype- and phenotype-driven mutagenesis will provide a much enlarged catalogue of mouse mutations and phenotypes for future gene function studies.

INTRODUCTION

Sometime during the first decade of the next millennium, the complete sequence of the human genome will be available, marking the beginning of the end of the genome project and heralding a new era of gene function studies in mammals. At the same time, there is a considerable expectation that the sequence of the mouse genome will also be determined. While there remain potentially significant obstacles to the identification and annotation of all of the gene sequences, attention is turning towards the post-genome challenge. This challenge is the interpretation of gene function.

There are many routes to building the profile of a particular gene and its function. These include database searches for comparisons of similarity with other sequences either from the same or other organisms (1,2); searching for potentially homologous structural motifs (3); and profiling the expression of the gene during development and differentiation (4,5) and how it responds to various challenges. This information undoubtedly will be acquired in the process of annotating the sequence map of the mouse and human genomes. Nevertheless, there is one route that can be expected to play an increasingly significant role in interpreting the genome sequence map and in determining and examining gene function in mammals. This is mutagenesis. The use of mutagenesis presages a return to classical genetic approaches of relating mutation to phenotype (2). The power of this approach is certainly evident in the lower eukaryotes. The completion of the genome sequence of Saccharomyces cerevisiae has led to an explosion in methodologies both to generate mutations systematically at every gene and to explore their phenotype (6). The challenge for mammalian systems in studying gene function is to devise routes that harness the power of systematic genome-wide mutagenesis to the growing genomic resources.

The mouse is the organism of choice if we are realistically to undertake genome-wide mutagenesis programmes. It may seem that we already have a large fund of mouse mutations for the examination of gene function, but the mouse mutant resource is relatively small. Although there are >1000 mouse mutations in the current mouse locus catalogue (see http://www.informatics.jax.org/ ), this represents only a small fraction of the likely total number of mammalian genes. We clearly do not have access to all phenotypes-there is a `phenotype gap' (7) that needs to be closed by the generation and archiving of new mouse mutations.

This article reviews progress towards systematic approaches for the generation genome-wide of new mouse mutations and the characterization of their phenotypes. It is important to note that ultimately we need to increase both the breadth and the depth of the mouse mutant resource. For the latter, the power of multiple alleles at a particular locus in terms of dissecting gene function has been amply demonstrated (for example, see ref. 8). Generation of allelic series may also come about through targeted studies of gene function rather than systematic approaches, but the development of systematic approaches promises to provide the material for molecular and quantitative geneticists for the study of gene function for decades to come.

Table 1. Genotype- and phenotype-driven approaches for systematic mutagenesis of the mouse genome

Approach Technical routes Advantages Disadvantages

Genotype-driven Gene trap libraries Easier identification of underlying locus Unpredictable phenotypes

Phenotype-driven ENU chemical mutagenesis No prior assumptions made about underlying genes or pathways Identification of underlyinggene is not trivial

SYSTEMATIC APPROACHES TO THE GENERATION OF MUTATIONS IN THE MOUSE GENOME

There are two potential approaches to the systematic generation of mutations across the mouse genome. One approach is genotype-driven, the other phenotype-driven (see Table 1). Genotype-driven approaches are sequence-driven and can involve targeted mutations engineered through homologous recombination in embryonic stem (ES) cells. However, this approach is not easily scaleable to the recovery of large numbers of mutations on a genome-wide basis. A more systematic approach can be taken using a gene trap strategy. Here an appropriate selectable construct is introduced into ES cells and large numbers of lines carrying disrupted transcription units can be derived in order to build an ES cell mutation bank (9). Mutant mice can be derived from the ES cell mutation bank for phenotyping. For the gene trap approach, recovering relevant coding sequence for the disrupted transcription unit is relatively straightforward. However, one disadvantage of the genotype-driven approach is that a prioriassumptions have to be made about the likely role and importance of any gene trap sequence that is the subject of further study.

For phenotype-driven approaches, the focus is on using random mutagenesis systems to derive mice with novel phenotypes from which subsequently the relevant genes and pathways are identified. Thus, no assumptions are made about the underlying genes involved, and phenotype-driven approaches can be expected to be a powerful route to the identification of novel pathways and their genes. Phenotype-driven approaches, however, require the application of appropriate screens to identify phenotypes of interest. Nevertheless, the identification of leptin (10), novel genes involved in genetic deafness (11) and the characterization of a first mammalian Clock gene involved in a circadian pathway (12,13) are all examples where new genes and pathways have been identified using a mouse mutant phenotype as the critical start-point.

Transgenic insertions are one potential route to generating new mutations genome-wide (14). Although the identification of the disrupted transcription unit can be straightforward, this approach is not readily amenable to scale-up in the search for new phenotypes. Instead, for systematic phenotype-driven mutagenesis, most mouse geneticists have turned their attention to chemical mutagenesis using N-ethyl-N-nitrosourea (ENU) (15). Saturation screens for a variety of developmental phenotypes using ENU mutagenesis have been reported recently in the zebrafish, Danio rerio (16,17), providing a paradigm for this approach in the mouse. Alternatively, the chemotherapeutic agent chlorambucil has also been employed in mouse mutagenic screens (18). Like ENU, chlorambucil is highly mutagenic in mouse male germ cells, but generates predominantly chromosomal rearrangements, including deletions and translocations. For the zebrafish and the mouse, however, the disadvantage of chemical mutagenesis as a phenotype-driven approach for systematic studies of gene function is that, having identified a novel phenotype, it still requires considerable effort to identify the underlying gene.

Finally, it is possible to adapt chemical mutagenesis programmes to a genotype-driven approach. DNAs from cohorts of mutant mice could be screened for changes in known gene sequences, and subsequently, having identified a mutation or series of mutations in a particular gene, phenotypes could be examined.

GENOTYPE-DRIVEN APPROACHES

Recently, two approaches have been reported that utilize a gene trap approach for the construction of a large bank of mutant ES cell lines. The first utilizes a promoter trap retrovirus shuttle vector to disrupt genes in murine ES cells (19). The vector carries a promoterless neomycin resistance gene, as well as a plasmid origin of replication and ampicillin resistance gene for the recovery of flanking cellular DNA from the promoter trap events. Flanking sequences recovered, designated promoter-proximal sequence tags or PSTs, were screened against the non-redundant GenBank database. Approximately 10% of PSTs showed a significant match, while another 5% matched anonymous cDNAs in dbEST. This approach, while clearly able to deliver a large number of mutations genome-wide, is limited to genes expressed in ES cells.

In contrast, a second group has developed an approach for gene trapping that should enable the targeting of genes independently of their expression status in ES cell lines. The bank of mutant ES cell lines recovered by these methods is known as Omnibank (20). The Omnibank approach is based on a gene trap vector with two functional units. The first consists of a PGK promoter linked to a puromycin N-acetyltransferase gene (conferring puromycin resistance) followed by a splice donor site (PGKpuroSD). Puromycin resistance is achieved when PGKpuroSD integrates within a transcription unit, splicing to exons downstream of the PGKpuroSD splice donor site followed by addition of a polyadenylation sequence from the endogenous gene. It is important to note that transcription of the endogenous gene in ES cells is not required for trapping, and a sequence tag from the trapped gene can be recovered by 3[prime] RACE. The second unit in the vector lying upstream of the PGKpuroSD cassette is the mutagenic cassette (SAIRES[beta]geobpA) consisting of a splice acceptor sequence fused to a selectable reporter gene ([beta]geo) followed by a polyadenylation sequence. Using these cassettes, the recovery of 3000 sequence tags (so-called Omnibank sequence tags: OSTs) has been reported (20). Many of these OSTs show matches to GenBank or SwissProt-18% match known genes and 10% match human and rodent expressed sequence tags (ESTs). A number of the unknown RACE products were chosen for further sequencing and, of these extended sequences, 27% subsequently showed hits to GenBank, giving confidence that the vector system is indeed trapping transcribed sequences. There was also evidence that the mutagenesis cassette (SAIRES[beta]geobpA) was functioning appropriately.

The OST database is being expanded rapidly. Searching this sequence tag database will allow investigators to identify novel transcription units for which a mutant ES cell line exists. Subsequently, mice can be generated from the mutant ES cell line and their phenotype explored. Clearly, as indicated above, some prior assumptions will have to be made about which gene trap mutants to investigate. Moreover, for those mutations that are analysed further, there will be prior expectations about the phenotype that will result. In this regard, depending on the screening protocols employed, subtle phenotypes may be missed. The development of systematic and comprehensive screening protocols which are so important for phenotype-driven screens (see below) will also be vital for the exploitation of gene trap banks.

PHENOTYPE-DRIVEN APPROACHES

Genetic approaches to the generation of new phenotypes in the mouse

ENU, the principle mutagen of choice for large-scale mouse mutagenesis programmes, delivers specific locus mutation rates in the region of 1 in 1000 gametes or less, reflecting the rate at which recessive mutations at loci genome-wide are likely to be generated (21,22). Dominant mutations are recovered at lower frequencies than recessive mutations. In one study, the ratio of loss-of-function mutations to gain-of-function was 4:1 (23). ENU is administered intraperitoneally to male inbred strain mice where it acts on the spermatogonial stem cells inducing a variable period of sterility; spermatogenesis comes to a halt for 10-15 weeks following injection while surviving mutagenized stem cell spermatogonia repopulate the testis. Subsequently, treated male mice are bred to recover mutant gametes.

Molecular characterization of a number of ENU mutations indicates that ENU induces point mutations, with the majority of changes occurring at A-T base pairs (for examples, see refs 8,11,12). It is important to stress that ENU point mutagenesis will allow the recovery of both recessive loss-of-function mutations as well as dominant mutations, including dominants arising from haploinsufficiency, dominant-negative and dominant gain-of-function mutations. Given that the 10th edition of McKusick's Mendelian Inheritance in Man (24) lists 2470 autosomal dominants and 647 autosomal recessives, the ability to undertake efficient screens for dominant phenotypes further underlines the value of ENU chemical mutagenesis and its utility for the development of mouse models of human genetic disease. Indeed, dominant screens represent one of the most straightforward and efficient routes to the systematic recovery of new phenotypes amongst a variety of protocols including recessive genome-wide and recessive targeted screens.

Dominant and semi-dominant genome-wide screens. The first and most simple protocol (Fig. 1a) is to mate ENU-treated males to wild-type females and to score the F1 progeny for dominant and semi-dominant mutations. Mice carrying new phenotypes can be test bred to confirm heritability, and affected progeny recovered can subsequently be intercrossed to examine mutant homozygotes and semi-dominance. Given the simplicity of the protocol, this approach is being widely adopted for the recovery of new mouse mutations. It has been the route by which a number of overt visible and other anomalies have been recovered [e.g. biochemical anomalies (23) and cataract mutations (25)] and was the approach used to recover the dominant Clock mutation (12; see below). If we assume the mutation frequency for dominants to be one-quarter that of recessives, then an F1 screen of 20 000 animals is effectively a saturating screen and is within the capabilities of many large animal houses.

a, b

Figure 1. ENU mutagenesis approaches in the mouse. (a) Recovery of ENU-induced dominant viable mutations in a genome-wide screen. ENU-mutagenized males are mated to normal females and the F1 progeny are screened for dominant mutations. (b) Recovery of ENU-induced recessive viable mutations (*) targeted to a chromosome deletion region of the mouse genome. ENU-mutagenized males are mated directly to mice hemizygous for the appropriate deletion and viable progeny screened for mutations. This approach, while straightforward, would not distinguish new dominant mutations from recessive mutations recovered in the deletion region (see text). (c) Recovery of ENU-induced recessive lethal or recessive viable mutations targeted to a chromosome deletion region of the mouse genome using a two-generation scheme. Both the mutagenized chromosome and the deletion stock are genetically marked in the example shown (taken from ref. 29). The chromosome that is mutagenized (*) carries the albino mutation (c) while the deletion stock carries the mutant chinchilla allele (c^ch) at the albino locus. Three classes of progeny are recovered in terms of coat colour-albino, light chinchilla and wild-type. The albino class carries potential mutations along with the deletion chromosome. Absence of this class is indicative of a recessive lethal mutation. This class can also be tested for recessive viable mutations. The light chinchilla coat colour class (c/c^ch heterozygotes) are carriers and can be used to recover and propagate recessive lethal mutations. The advantage of a two-generation scheme is that dominant mutations can be identified in the first generation, thus allowing mice recovered in the albino class to be analysed for recessive viable mutant phenotypes (see text). The development of physical and transcript maps (ESTs a-k) across the chromosome deletion region provides panels of candidate genes that may underlie the gene of interest. Complementation analysis of the mutation (*) with a series of nested deletions from this chromosome region allows detailed localization of the mutation, further narrowing the pool of candidate genes.

Recessive genome-wide screens. In order to identify new recessive mutations, it is necessary to mate ENU-treated males to wild-type females, followed by breeding of the F1 progeny (G1) to wild-type mice to establish families of siblings (G2) sharing common mutations. G2 progeny can be backcrossed to G1, producing G3 progeny that can be assessed for recessive mutant phenotypes. Clearly, such a three-generation screen demands considerable animal house resources, and for this reason has only been attempted to date on a relatively small scale. Mouse mutants causing hyperphenylalaninaemia were recovered using just such a three-generation screen, including mutations in the gene for phenylalanine hydroxylase (26,27).

Targeted recessive mutation screens using chromosomal deletions. In this approach, recessive mutations can be recovered in a region-specific manner using stocks of mice carrying deletions for the relevant chromosome region (15). Viable mouse deletion stocks are available that cover ~15% of the mouse genome (28; see http://www.mgu.har.mrc.ac.uk ). In the simplest protocol (Fig. 1b), recessive viable mutations are recovered by mating ENU-treated males to the deletion stock and the F1 progeny analysed for novel phenotypes. This, however, will not distinguish the recovery of a dominant mutation, either within or outside the deletion region, from a targeted recessive mutation uncovered by the deletion, posing a particular problem if the frequency of dominant mutations recovered is high (see below). To recover targeted recessive lethal mutations, a two-generation screen must be employed whereby ENU-treated males are mated to wild-type females and the F1 progeny are mated to the deletion stock (15,29; Fig. 1c). The mutated chromosome usually carries a recessive visible marker that allows one to detect the presence of a recessive lethal mutant by the absence of the relevant visible phenotypic class in the second generation. This approach has been undertaken for the saturation mutagenesis of a 6-11 cM region encompassing the albino locus on mouse chromosome 7 (29). There are advantages of the two-generation protocol for the recovery of recessive viables in that one can screen the F1 generation for novel phenotypes and thus weed out dominant mutations before the second generation. Recessive viable mutations are detected by appropriate screens of the relevant visible phenotypic class in the second generation (Fig. 1c). Overall, the targeted approach to ENU mutagenesis has many advantages in that the recovered mutations are pre-localized and potential positional candidate genes can be assessed immediately. In addition, we can better delineate the potential candidates for any mutation within the deletion by focusing genomics efforts on the construction of transcript maps across the chromosome deletion region. In this way, mutagenesis and genomics can be combined, thus short-circuiting the process of gene identification.

Two approaches will allow us to extend the range of available deletions covering the mouse genome. The first employs cre-lox to engineer deletions (30). A number of new deletions ranging from 90 kb to 3-4 cM have been generated on mouse chromosome 11 using this method. Alternatively, using selectable markers targeted into ES cells, it is possible by radiation mutagenesis to recover cell lines carrying deletions encompassing the integrated marker (31). A number of deletions on mouse chromosome 17 in the region of the t complex have been rescued. The potential to develop sets of nested deletions across any region will not only aid the recovery of recessive mutations but also help refine the detailed location of any mutation, further narrowing the pool of potential candidate genes (Fig. 1).

Phenotypic screens

As well as implementing mutagenesis and breeding protocols for the recovery of mutations, it is necessary to devise and employ screens that will identify mutant phenotypes of interest efficiently amongst the progeny of ENU-treated mice (32). Clearly, any cohort of mice can be the focus of one or two intensive tests for phenotypes of particular interest to the investigator. However, given the importance of generating a wide range of novel phenotypes, the challenge is to develop efficient, systematic and comprehensive phenotype screens that will permit comparison between laboratories, over time and between different strains of mice. Such screens will need to employ a battery of tests and must allow for a rapid assessment of large numbers of mutant mouse progeny for potential deficits. The corollary of this is that such screening protocols will be hierarchical. The primary screen will be systematic and broad-ranging but rapid, able to process potentially hundreds of mice a week. Subsequently, mice demonstrating potential deficits in particular areas will be the focus of more sophisticated secondary and tertiary screens to better define the mutant phenotype.

A systematic and hierarchical protocol for phenotype assessment called SHIRPA has recently been proposed (33), involving primary, secondary and tertiary screens. Analysis of data from any stage of the test provides a comprehensive profile of function and gives indications of deficits in muscle and lower motor neuron function, spinocerebellar function, sensory function, neuropsychiatric function and autonomic function. The primary screen involving a battery of ~30 rapid tests is semi-quantitative and based on the Irwin screen (34) developed for drug effects on wild-type animals. The testing is carried out in a simple testing arena and it is possible to add further tests to expand the range of phenotype observations (for a detailed protocol, see http://mgu.har.mrc.ac.uk/mutabase/ ). The secondary screen involves a comprehensive behavioural screening battery and pathological analysis. Tertiary screening follows using more sophisticated approaches to assess behaviour and phenotype including, for example, the Morris water maze and investigations employing electroencephalography (EEG), nerve conduction, electromyography and magnetic resonance imaging (MRI).

In a similar fashion, Crawley and Paylor (35) have proposed a battery of tests comprising a primary screen for neurological and neuropsychological deficits, along with a constellation of behavioural paradigms for investigation of behavioural phenotypes. Targeted mutagenesis studies have underlined the fact that mutations in single genes can affect components of complex mammalian behaviours such as sensorimotor gating, a constituent of many complex neuropsychiatric disorders (36). The use of systematic phenotypic screens to identify multiple components of complex behaviours promises to be an important element of the post-genome challenge.

SHIRPA has been employed to score mutant phenotypes from a large dominant genome-wide screen underway at MRC Harwell. To date, from 3000 mice analysed, some 30 new mutations with confirmed inheritance have been identified and a similar number of mutations are in inheritance testing (J. Peters, P. Nolan, J. Hunter and S. Brown, unpublished data). These results give us confidence that combining ENU mutagenesis approaches along with efficient, comprehensive phenotype screens will enable us to make a significant contribution to the mouse mutant resource and the closure of the phenotype gap.

FROM CLOCK TO GENE-FROM PHENOTYPE SCREENS TO MUTATION

The evolution of comprehensive, systematic screens will be allied to the continual development and application of novel phenotyping protocols, many of which ultimately will be incorporated as standard tests within systematic screening approaches. For example, the identification of the mouse Clock mutation and the cloning of the gene is a paradigm of ENU mutagenesis, from the application of a novel screen to the identification of new genes and pathways. The impetus for locomotor activity screens in mice was somewhat influenced by phenotypic screens in Drosophila and Neurospora where dominant and semi-dominant circadian period mutations at the Per (37), Tim (38) and Frq (39) loci were characterized successfully. The Clock mutant was originally identified in a dominant screen for circadian locomotor activity mutants in progeny of ENU-mutagenized mice (40). Phenotypically, Clock is a specific circadian rhythm mutation, with a lengthening of the circadian period of locomotor activity by 1 h in heterozygotes and by longer in homozygotes (see, for example, Fig. 2c). The culmination of this work was the isolation of a candidate gene, Clock, by positional cloning and the successful identification of a point mutation that results in a skipped exon in the mutant gene (12). Furthermore, the nature of the Clock mutation was confirmed by in vivo complementation using bacterial artificial chromosome (BAC) transgenesis (13).

Human Molecular Genetics