Although an antecedent to results from concerted mouse-based mutagenesis work, the ability to use cell lines in genome-wide approaches is impacted by issues of ploidy, genetic variation, and barriers to mapping studies using somatic cells. Instead of an approach such as that described above, which is oriented to recessive mutations, more recent work has used a screening strategy based on dominant mutations in Jurkat cells 2. For this study, a novel retrovector was generated that permits tetracycline regulation of the expression of inserted cDNAs (Figure 1). The vector is replication-deficient because the promoters of the retroviral long terminal repeats have been mutated, and it carries a GFP cDNA to allow selection of transduced cells. After insertion of mutated cDNA libraries into this retrovector downstream of a tetracycline-inducible promoter and transduction of the Jurkat T-cell line, individual cells were screened by fluorescence-activated cell sorting to identify those exhibiting decreased expression of a marker, CD69, known to be downstream from TCR ligation. Importantly, this study validated the ability to identify scores of genes for which mutations (generated while creating the cDNA library) attenuated TCR signaling to the CD69 gene 2. Furthermore, the incorporation of an on-off system controlling expression of the cDNA made a critical screening step possible by simple reversion testing - loss of the mutant phenotype when the expression of the cDNA insert was turned off. The completion of genome sequences and availability of tools sufficiently robust to handle the numbers of cDNAs screened (2,828 in the study in question 2, of which 1,323 were validated by reversion) link this approach to the post-genomic era.

Such a cell-based, dominant selection system offers a powerful tool for the identification of molecules that impact a signaling pathway, albeit at the cost of several drawbacks. Discovery will be limited to those molecules for which dominant negatives have an effect on the pathway, and there will be inherent pitfalls in the use of any cell line: for example, lipid-kinase signaling in most Jurkat T cells will be abnormal because of pre-existing loss-of-function mutations in the PTEN lipid phosphatase/tumor suppressor genes. Nonetheless, this or analogous retrovector systems can be used for the transduction of primary lymphocytes to perform further validation. Intriguingly, variations on this approach offer potentially great power in structure-function analyses with selected molecules. Single cDNAs could be subjected to high-density mutagenesis and, through the on-off capabilities of the approach, microarray analyses would allow measurement of the changes in global gene-expression profiles that result from graded interference with a given pathway.

The applicability of a finding to immune regulation ultimately depends on its validation in the immune response of animals. In principle, such retroviral technologies could also be applied to the transduction of hematopoietic cells (for example, lymphocytes or dendritic cells) derived from mouse lines in which loss-of-function mutations have been engineered by other approaches. This extension of the approach could allow high-throughput structure-function screening in primary cells independent of the need to achieve dominant-negative effects. In parallel, however, enhanced techniques for random mutagenesis, phenotypic screening, and gene identification have been important goals. To facilitate the process of cloning mutated genes after mutagenesis, another class of mutagen takes advantage of the fact that retrovirus insertion in the genome can be engineered so as to achieve loss-of-function mutations at high efficiencies (14; reviewed in 15). In particular, vectors based on oncoretroviruses are modified to encode a selectable trait whose expression depends on insertion into an active transcription unit. After bulk infection of pluripotent embryonic stem (ES) cells with high-titre retrovirions, individual clones can be micro-injected and transmitted to the germ line, so that progeny can be mated to homozygosity. As an example of the early success and potential of this strategy of retroviral mutagenesis, a novel link was identified between the mouse FUS gene, chromosomal stability, and B-lymphoid proliferation 16. This technology promises to provide further successes, which will be facilitated by the creation of banks of mutated ES cells for which a sequence tag identifying the mutated gene is publicly accessible. Two examples of ES-cell banks currently available are from a for-profit company, Lexicon Genetics 17, and a program funded by the National Heart Lung and Blood Institute and linked to BayGenomics 18.

Progress in genome sequencing and annotation, and saturation with sufficient genetic markers for gene mapping, have dramatically enhanced our capacity to identify a gene mutated by chemical mutagens that create point mutations. These capabilities, coupled to recognition of the powerful advantages of ethylnitrosourea (ENU) as a mutagen 19, set the stage for several of the recent successes noted at the beginning of this article. ENU is a highly efficient mutagen, able to create multiple mutations in the germ line of each male breeder, which can be coupled with breeding strategies to perform recessive screens (Figure 2) 202122. In the case of mice with mutations in the Lps2 gene, this approach has revealed an essential component of signaling by some, but not all, Toll-like receptors (TLRs) operating as part of a pathway independent of the MyD88 relay protein 123. Mutagenesis and screening for defects in ex vivo production of TNFα by macrophages were initiated on an extensively used inbred genetic background, C57BL/6. Notably, this work involved approximately 2,650 G1 mice, heterozygous for any mutation, and around 4,600 G3 mice, bred for the generation of homozygotes, underscoring some of the cost and logistical issues in such screens. The authors of these studies 123 focused on a homozygote whose macrophage response to endotoxin (lipopolysaccharide) and double-stranded RNA was abrogated while activators of other TLRs elicited normal TNFα responses. Gene mapping and high-throughput sequencing identified a mutation in the gene encoding a previously known adapter protein, Trif/Ticam-1, that leads to a small carboxy-terminal substitution. Although this gene product had already been implicated in TLR signaling, the Lps2 mutant mouse lines establish several novel features of the both the protein (Trif/Ticam-1) and the MyD88-independent signaling pathway. In particular, these studies make clear the essential role of the Trif/Ticam-1 protein in signaling by several members of the TLR gene family, including some whose signaling also uses MyD88. The results also indicate that other TLRs use an Lps2-independent, MyD88-dependent means of signaling.

Another triumph of functional screening at a genome-wide level has emerged from a separate program, resulting from similar ENU mutagenesis and recessive screening in the same C57BL/6 background 34. This large-scale program focussed on the identification of mutations leading to immune dysregulation 22. Founder mice were identified by a quantitative screen that detected differences in the profile of surface immunoglobulin M (sIgM) on circulating B lymphocytes that also carried IgD. In addition to a two-fold increase in sIgM, closer examination also revealed a modest increase in another molecule linked to signaling, CD21. Gene mapping localized the point mutation to the Carma1/CARD11 gene. Rather than a loss of expression, as found in the prior work in Jurkat cells (see above, 13) and concomitant papers using gene targeting 2425, this ENU-induced point mutation led to normal levels of a full-length protein with altered function. Comparison of the consequences of this mutation to the outcome of gene-targeting experiments which eliminated Carma1 expression 24 is illuminating in terms of what each strategy can offer. The ENU mutant, which affects a coiled-coil region of the protein, led to a phenotype far more prominent in B lymphocytes than their T-cell counterparts, while the lack of Carma1 in the loss-of-expression studies led to a profound impairment of TCR/CD28-induced production of interleukin-2 (IL-2) and proliferation, along with B-cell signaling defects 132223. Conversely, the point-mutant Carma1 led to a partially penetrant hyper-IgE syndrome with other atopic manifestations in the mice, whereas this abnormality was not noted in the conventional knockouts. Thus, the random generation of a point mutant uncovered new features of the protein's function in vivo and some selectivity in the impact of these functional traits on particular lymphoid-cell types.

So, what are the global lessons from these advances, and what future refinements incorporating genome-wide approaches may be possible? The first lesson is that these approaches establish the ability of large, well-funded, and well-organized operations to generate new insights into immune regulation using functional-genomic screens. In addition, some of the findings will differ from those obtained using conventional knockouts. Subsequent to these initial 'test-of-principle' papers, the pace of discovery using this type of approach will accelerate. The second lesson is that although there undoubtedly is some selection bias, a strikingly high frequency of 'old friends' turn up in these screens. Thus far, most of the listed gene products have been previously identified and implicated in the relevant signaling pathways by conventional approaches. This suggests that intensive gene discovery by pregenomic techniques has been quite successful. Nonetheless, because a substantial fraction of predicted genes do not fall into known functional groups it is to be expected that the gene products uncovered will extend more into groups not previously linked to the functions being screened. Finally, the collected findings indicate that it is already timely to be generating systems-based strategies to exploit these advances. Approaches of this type will gain power as they are combined and linked to other forms of genomics 262728, proteomics and systems analysis. Thus, for example, the tetracycline-regulated dominant retrovector system can be combined with transductions of hematopoietic cells from mutant mice and to microarrray analyses of gene-expression profiles. Such approaches may allow the identification of selective roles of protein modules and quantitation of the effects on function of a wide range of sequence variants generated in vitro. Similarly, when coupled to screens of protein-protein interactions in cell lines, genome-scale approaches will rapidly increase our insights into structure-function relationships in immunity and disease.