Background

A large-scale, genome-wide knockout mouse project was first proposed in 2003 with the aim of developing knockout mouse resources for laboratory investigators. Curated databases at the time only described global knockout mice in about one-tenth of 25,000 estimated mouse genes [197]. Many of these resources were unavailable for public use. In light of this a worldwide, systematic, centrally-curated initiative to mass-produce knockout mouse lines could provide investigators with a vital resource. These mice could be used to generate previously-unavailable experimental models and, in doing so, advance biomedical research.

The 2003 working group (whose proceedings are summarized in a 2004 publication [197]) promulgated the following guidelines for the initiation of such a project: (1) a variety of methods (e.g., both gene trapping and gene targeting [198, 199]) should be employed in order to mass-produce null alleles, (2) null alleles should include reporter sequences in order to expedite the process of cell and tissue phenotyping for end-users, and (3) at a minimum, the project should produce, bank, and make available mutant ES cells to researchers across the country and around the world [197]. The authors recommended against the use of cis-elements (such as loxP or FRT sites) which could be used to generate conditional knockouts, due to the nascent state of the technology at the time.

Austin et al. [197] further described several tiers of phenotyping which could be centrally provided by the project. Once the absence of a particular gene in a particular mouse line has been firmly established, Tier 1 phenotyping would be employed to elucidate certain aspects of the mutant mouse’s physiology (for instance, blood chemistry and hematological profiles). Microarray transcriptome analysis would identify a subset of mice for deeper, system-specific examination. At the time, it was estimated that a majority of mouse genes could be knocked out

within five years, at a total cost of about $50 million (about $10 million per year). Production of 500 new mouse lines from the mutant ES lines was estimated to cost $12.5-15 million annually. Tissue of these 500 lines was expected to cost up to another $5 million (for tissue phenotyping), with another $2.5 million for Tier 1 phenotyping and another $18,000 for each mouse line selected for transcriptome analysis.

Critics of such large-scale projects raised concerns about the scope and scale of the project, citing insufficient demand for many of the commercially available knockout lines which were available at the time [200]. Further, given that many phenotypes are driven by multiple genes and redundant signaling pathways, single-gene knockouts may not alter physiology in a clinically meaningful way.

In spite of this, not one but four genone-wide knockout mouse projects have arisen in the years since the 2003 working group [199]: one in Europe (EUCOMM), one in Canada (NorCOMM), and two in the United States (the NIH-funded Knockout Mouse Project [196] and the state-funded Texas Institute for Genomic Medicine). The International Knockout Mouse Consortium (IKMC) was established in 2006 in order to limit redundancy and encourage data sharing between these four groups and any others willing to undertake the task [196, 199, 201].

4.5.1.1 Allele Types and Allele Conversion

Austin et al. [197] describe several deliverables for the knockout mouse project, including gene trapping constructs, frozen zygotes, frozen embryos, live mice, and phenotypic data. Critical to all is the development of an allele which can be easily modified via recombination using Cre-lox or FLP-FRT recombination.

homozygote is a complete knockout (due to the presence of two SV40 polyadenylation sequences which prevent transcription of the entire gene). A tm1a mouse is the “first” step in a breeding strategy to derive a floxed allele (hence the name, “knockout first”). Recombination of the tm1a allele using Cre or FLP-FRT technology yields further alleles. The tm1a allele and its derivatives are diagrammed in Figure 9.

Figure 9. Allele maps and recombination strategy for the tm1a allele. Figure made by author, adapted from [203].

The tm1b lacZ-tagged allele is generated when the tm1a allele undergoes Cre recombination [201]. Ideally, this process removes all genetic material between the most 3’ and 5’ loxP sites in the tm1a sequence, including the neo sequence, both poly-adenylating sequences, one or more targeted exons, and one of the FRT sites. Mice which are homologous for tm1b should also be complete knockouts. (In practice, Cre recombination can remove genetic material between any two of the three loxP sites in the allele – not just the most 3’ and 5’ sites. This means that two additional alleles can be generated in addition to the desired tm1b allele: one which lacks the neo sequence only, and the other lacks targeted exons only. In these cases, the allele is designed so that the recombination process creates a frameshift mutation which phenotypically resembles a knockout [201].)

The tm1c allele is generated when the tm1a allele undergoes FLP/FRT recombination, which excises all genetic material between the two FRT sites, including both polyadenylation sequences [201]. This recombination reverts the cell to a wild-type phenotype, and it should generate a normal protein. In particular, it should be noted that this recombination leaves two

loxP sites on either side of the target sequence which can be targeted with Cre recombination;

thus, a mouse homozygous for tm1c is a traditional “floxed mouse.”

Finally, Cre recombination of the tm1c allele yields the tm1d allele, which should not express normal protein. Like the tm1a and tm1b alleles, and the tm1d allele represents a “knockout” genotype and phenotype.

The tm1a allele for the CD47 gene was originally developed by the Knockout Mouse Project [203]. It should be noted that the target sequence of the CD47 tm1a allele includes exons 2 and 3 [203], whereas the commercially available CD47 null mouse was generated via the

removal of exon 2 only [188, 190]. The physiological implications of this difference are not completely known, however cell-specific analysis suggest the lack of exons 2 and 3 phenocopies gene expression patterns found in cells from the commercial CD47 null mice.