TARGETED MUTATION OF GENES BY HOMOLOGOUS RECOMBINATION IN EMBRYONIC STEM CELLS

TOP ABSTRACT TARGETED MUTATION OF GENES... CONDITIONAL GENE-TARGETING... CHARACTERIZATION OF TISSUE-... NEW APPROACHES FOR... STRAIN BACKGROUND... CHEMICALLY INDUCED POINT... REFERENCES

Gene targeting in murine embryonic stem (ES) cells is a powerful technique that has been used to generate specific mutations in mice for investigations of gene function (8). Since the initial reports of successful germline transmission of gene-targeted ES cells during the mid-1980s, thousands of gene-targeted mutations or "knockout" lines have been reported. Studies of these mice have been very informative, elucidating gene function or identifying genetic pathways through mutations in single or, in some cases, multiple genes. Ultimately, gene-targeted mutations will be created in the majority of known genes of the mouse. This effort will be accelerated greatly by ongoing public and private efforts to sequence entire genomes of several different mouse strains as well as by continued analysis of the sequence to identify all expressed genes. The availability of the entire mouse sequence has facilitated the construction of gene-targeting vectors as has the development of thermostable polymerases with very high transcriptional fidelity. This has made PCR-based amplifications of large genomic DNA segments practical; it is no longer necessary to clone and map large portions of a targeted locus to construct a targeting vector.

By far, the most common gene-targeting strategy to date has been the introduction of null mutations that result in complete absence of the protein product in all cell types. Although insertion or replacement strategies can generate point mutations, small deletions, or other minor modifications of genes that do not lead to complete loss of function, these have not been extensively used in studies of gastrointestinal physiology. Null mutations can be informative; however, they do not always allow the full characterization of a gene's role in all cell types. For example, many genes are expressed in both embryonic stages and in the adult; null mutations in these genes can lead to embryonic lethality, preventing the analysis of the role of the protein in juvenile stages or in adult mice. Decreased viability, immunosuppression, sterility, or reduced breeding efficiency can also occur and may make it difficult to identify all functions of a particular gene in a homozygous null mutant mouse. Finally, targeted null mutations may promote compensatory functions by related genes that obscure the normal function of the targeted gene.

In some cases, it may be useful to develop a mutation that creates a specific change in the gene, such as a point mutation or small deletion, or a mutation in a promoter sequence that alters the expression of the gene but does not ablate it. As more single-nucleotide polymorphisms are identified in genes relevant to human intestinal disease phenotypes, it will be necessary to develop mice with similar alterations. Genes in the mouse could be further "humanized" through replacement of specific regions or domains of the murine counterpart with the exact human sequence. It is becoming increasingly apparent from genome analysis that a significant amount of the overall diversity of our genes comes through alternative splicing; therefore, simple deletion strategies could be used to remove specific domains of a protein to study the functions of these differentially spliced forms.

	CONDITIONAL GENE-TARGETING STRATEGIES FOR TEMPORAL OR TISSUE-SPECIFIC REGULATION OF MUTATIONS

TOP ABSTRACT TARGETED MUTATION OF GENES... CONDITIONAL GENE-TARGETING... CHARACTERIZATION OF TISSUE-... NEW APPROACHES FOR... STRAIN BACKGROUND... CHEMICALLY INDUCED POINT... REFERENCES

A number of conditional gene-targeting systems has now been developed that can be used to overcome some of the problems incurred by more traditional strategies (reviewed in Refs. 13 and 16). These methods generally involve the introduction of a mutation only in a defined cell type(s) and/or at a defined time period in the life of a mouse. Most of these strategies rely on site-specific recombinase proteins such as P1 bacteriophage protein, Cre, or, more recently, the yeast-derived Flip (FLP) recombinase. Both Cre- and FLP-mediated recombinations are specifically targeted to short (34 bp) DNA recognition sequences termed loxP or FRT sites, respectively, and both result in the excision or deletion of intervening DNA. loxP or FLP sites can be introduced into a gene of interest through homologous recombination in ES cells, using similar methods for generating standard knockout mice. Their integration, as well as the presence of the selectable marker gene, must not disrupt the coding region, expression, or splicing of the gene until the recombinase protein is expressed. This can be tested by developing lines of mice homozygous for the loxP-flanked ("floxed") or FRT-flanked ("flipped") allele and analyzing them for alterations in expression or function of the gene. The Cre/loxP system appears to be more efficient in mammalian cells than the FLP/FRT system and has been used more extensively for genome engineering.

The second key component for conditional mutations involves the regulated expression of the recombinase (e.g., Cre), such that it is only active in specific cell types or during a specific period in the life of the mouse. This can be accomplished by developing transgenic mouse lines in which Cre expression is controlled by previously defined cell-specific promoters or conditional promoters such as tetracycline or hormone-inducible systems. Transgenic lines containing the Cre recombinase under control of a desired regulatory element are then bred with mice containing the floxed gene. Alternatively, modified viruses that express Cre can also be used to induce recombination in tissues following infection. This approach was previously used to circumvent the embryonic lethality associated with complete loss of function of the APC gene and conditionally mutate the adematous polyposis coli (APC) gene in the colon for studies of tumor development (18).

One of the most widely used methods for generating conditional knockouts via Cre or other recombinases is the tetracycline inducible system, which takes advantage of modified regulatory elements that normally control expression of tetracycline resistance genes in Escherichia coli (reviewed in Ref. 16). In E. coli, the tetracycline resistance-mediating genes are constitutively repressed by the tetracycline repressor protein (tetR). In the absence of tetracycline, this protein binds to the tetracycline operator (tetO) in the promoter and prevents transcription. In the presence of tetracycline, tetR repressor activity is blocked and transcription of the tet operon proceeds. Adaptation of this system for use in transgenic mice required several modifications. First, the tetR protein was converted into a transcriptional activator by fusing it to the transcriptional transactivation domain of the herpes simplex virus protein VP16 (5). This chimeric protein is termed the tetracycline transcriptional activator (tTA). The second modification involved the cloning of the tetO sequences upstream of a minimal promoter obtained from the human cytomegalovirus (CMV) gene (5). The tTA binds to the tetO sites in the absence of added tetracycline and can activate transcription of a target gene placed downstream of the tetO-CMV promoter. If tetracycline is added, it binds to tTA and renders it incapable of binding to tetO, thus removing the activation signal. This system is limited for conditional gene mutation studies, because it requires the continuous presence of tetracycline throughout development to block recombinase (e.g., Cre) activity. Furthermore, it has been difficult to prevent low-level "leakiness" of gene transcription, even at high doses of tetracycline.

An important modification of this system for use in conditional gene targeting has been the development of a mutant protein with reversed transcriptional properties compared with tTA (6, 9). This mutated protein is termed the reverse tetracycline transactivator (rtTA). Its binding to the tetO DNA element activates transcription of tetO-CMV in the presence of tetracycline derivatives such as doxycycline (tetracycline binds with relatively low affinity). In the absence of antibiotic, rtTA does not bind to DNA. A distinct advantage of the rtTA system is that the inducer does not need to be added until transcription is required. The rtTA system has now been used to activate Cre expression and mutate genes in a cell- or temporal-specific manner (10, 13). This is generally accomplished by generating two different transgenic lines, one that expresses rtTA under the control of a constitutive or tissue-specific promoter and a second that expresses Cre under control of the tetO-CMV or other variants. The transgenes from each of these lines are then bred to a mouse containing the loxP-flanked gene (or region) of interest. Cre is then expressed only after the addition of doxycycline and only in cells in which rtTA is also produced. Expression of Cre in the tissue of interest results in loxP-targeted recombination and deletion of the loxP-flanked gene.

	CHARACTERIZATION OF TISSUE-SPECIFIC PROMOTERS FOR GENETIC MANIPULATION OF THE INTESTINAL MUCOSA

TOP ABSTRACT TARGETED MUTATION OF GENES... CONDITIONAL GENE-TARGETING... CHARACTERIZATION OF TISSUE-... NEW APPROACHES FOR... STRAIN BACKGROUND... CHEMICALLY INDUCED POINT... REFERENCES

Because the success of conditional gene targeting in the gastrointestinal tract is contingent on identification of gene-regulatory elements with the appropriate features of tissue specificity, implementation of this technology has required identification of genes that are expressed predominantly or uniquely by the gastrointestinal epithelium. The best studied genes to date are the family of fatty acid-binding protein (Fabp) genes that have been extensively and elegantly characterized (1, 20-22). The murine intestinal (Fabpi) and liver (Fabpl) fatty acid-binding protein genes are controlled by a relatively compact collection of activating and repressing promoter elements that can regulate gene expression along two axes in the gastrointestinal tissues: the gastrocolonic axis and the crypt-villous axis. The use of transgenic promoter-reporter approaches has permitted relatively detailed mapping of the principal transcriptional regulatory elements that control endogenous Fabp expression along these two axes. Inclusion of a "full-length" rat Fabpl promoter (region 4000 to +21 of the Fabpl gene) drives expression of reporter genes along the length of the small intestine in the adult mouse with onset at embryological day 16-17 (E16-17). Expression is greatest in the middle one-third of the small intestine and diminishes progressively toward the duodenum and ileum. There is limited expression in the cecum of the large intestine, with minimal or no expression in the stomach or colon. Similarly, the full-length promoter derives maximal expression toward the apex of the small intestinal villi, with lesser expression at the crypt-villous interface and no expression in Paneth cells at the crypt depths. The Fabpi promoter has an overlapping pattern of gastrocolonic expression but is normally silent in the gastric and distal colonic epithelium as well as in all other epithelium not associated with the gastrointestinal tract.

Sequential deletions of the full-length rat Fabpl 5'-untranslated region have permitted identification of discrete regulatory elements that control temporal and spatial patterns of expression (1). cis-Acting suppressors of transcription in the cecum and colonic epithelium were identified between nucleotides 4000 and 1600. Suppressors of gastric epithelial expression were defined outside of the 4000-bp promoter region. A suppressor of expression in the proximal tubular epithelium of the kidney was also found. Although the endogenous gene remained silent in the mouse and rat kidney throughout adulthood, transgenic mice that expressed a truncated Fabpl promoter element (132 to +21) demonstrated strong expression in the renal proximal tubular epithelium. A 35-nucleotide sequence positioned at nucleotides 167 to 133 in the Fabpl locus contains a heptad repeat element responsible for suppression of Fabpl gene expression in the renal proximal tubular epithelium as well as in hepatocytes and mucous-producing pit cells of the gastric epithelium and absorptive enterocytes in the proximal small intestine (19). Conversely, the same heptad repeat activates gene expression in the colonic epithelium, such that all proliferating and nonproliferating cells and colonic crypts distributed from the cecum to the rectum support transgene expression. This extensive mapping of the Fabp regulatory elements via transgenesis has made it feasible to direct exogenous gene expression to specific regions of the intestinal epithelium while silencing extraintestinal expression (e.g., liver and kidney).

By coupling these tissue-specific transcriptional regulatory elements with the systems for conditional gene expression or mutation described above, promising new opportunities for genetic manipulation of the gastrointestinal tract are emerging. A study that demonstrates the feasibility of this approach was recently reported (17). A variant of the Fabpl transcriptional regulatory region (Fabpl^{4x at 132}) was employed to direct transgenic expression of the Cre recombinase to the distal ileum, cecum, and colon. Mice carrying the Fabpl^{4x at 132}/Cre transgene were crossed with mice carrying a floxed hygromycin reporter transgene, and reporter gene deletion was monitored in multiple tissues during development. The Fabpl^{4x at 132}/Cre transgene was able to direct recombination of the floxed reporter in the correct distribution of the intestinal epithelium (distal small intestine and colon) with apparent high efficiency and stability as early as embryonic day 13.5.

A second experiment examined the potential for inducible Cre-mediated recombination limited to the intestinal epithelium (17). A transgene that included the reverse tetracycline-regulated transactivator (rtTA) under control of the same Fabpl transcriptional element (Fabpl^{4x at 132}) permitted the induction of Cre-mediated recombination of a floxed reporter gene at anytime during adulthood following administration of doxycycline in the drinking water. Although this study dramatically marks the first success of conditional gene targeting in the mouse intestine, it also highlights some of the technical difficulties that will require refinement before more generalized application of this complex genetic approach. First, efficiency of Cre-mediated recombination of the reporter within individual crypts was clearly <100%. This could reflect an intrinsic inefficiency of the Cre recombinase or heterogeneity in its levels of expression. Alternatively, this might reflect the special problem presented by an intestinal epithelial population that is undergoing rapid and perpetual renewal. Unlike cell populations that are nonrenewing and/or are long-lived, the rapid turnover of the intestinal epithelium makes stable conditional mutants somewhat daunting. If Cre is expressed under the control of transcriptional elements that only operate in differentiating intestinal epithelial cells, it is likely that the window of time available for recombination will be too narrow for efficient target gene silencing. Second, in the inducible Fabpl/rtTA system, unexplained regional differences in the induced expression of the Cre recombinase were evident, possibly due to animal-to-animal variation in rtTA expression, the dose of doxycycline received, or other unknown factors. Finally, given the intimate, reciprocal relationship between the gastrointestinal epithelium and enteric microflora, the use of antibiotic-responsive inducible systems, such as the doxycycline-rtTA system, could introduce unanticipated consequences by modifying the composition of commensal strains resident in the gut. Thus, although not without drawbacks, hormone-regulated inducible systems, such as the ecdysone- or tamoxifen-based systems, may gain favor for conditional gene regulation studies in the gastrointestinal tract (4, 7, 15, 23).

Clearly, more extensive studies will be needed to define more tightly regulated, inducible systems to permit tissue- and temporal-specific targeting of intestinal genes. Nevertheless, refinements including higher-efficiency Cre recombinases and tighter ligand control of Cre expression to avoid background levels of recombinase activity are in development and should result in better genetic systems for inducible, tissue-specific gene regulation in the intestine. Furthermore, as more lines of mice that express recombinases in specific intestinal tissues become available, it will become easier for more investigators to apply them to their own systems. Likewise, the development of more lines of mice with genes flanked by integrated loxP or FLP sites will also aid this effort.

	NEW APPROACHES FOR IMPLEMETATION OF BAC-BASED TRANSGENIC STRATEGIES

TOP ABSTRACT TARGETED MUTATION OF GENES... CONDITIONAL GENE-TARGETING... CHARACTERIZATION OF TISSUE-... NEW APPROACHES FOR... STRAIN BACKGROUND... CHEMICALLY INDUCED POINT... REFERENCES

An important limitation of transgenic approaches for tissue-specific gene expression concerns insertion site-dependent effects of the local chromatin "neighborhood." The random nature with which transgenes insert into the host genome often preclude normal regulation of transgene expression. As a remedial approach to this shortcoming, there has been renewed interest in utilization of larger transgenic constructs that include greater flanking DNA to insulate the transgene from local chromatin effects. A significant impediment of this technology has been the difficulty of cloning and manipulating large segments of the genome in either bacterial (BACs) or yeast artificial chromosomes. The recent introduction of -prophage-based strategies for homologous recombination in BAC clones will significantly alleviate this technical barrier. Copeland et al. (2) and Yu et al. (25) have developed elegant new systems for -prophage-mediated "recombineering" as a tool for efficiently introducing gene disruptions or modifications into either BAC plasmids or the bacterial host genome. This technology should catalyze renewed interest in BAC-based transgenesis to better control transgene expression characteristics in vivo.

	STRAIN BACKGROUND EFFECTS-OBSTACLES AND OPPORTUNITIES

TOP ABSTRACT TARGETED MUTATION OF GENES... CONDITIONAL GENE-TARGETING... CHARACTERIZATION OF TISSUE-... NEW APPROACHES FOR... STRAIN BACKGROUND... CHEMICALLY INDUCED POINT... REFERENCES

It is now evident from many studies that the phenotype of a particular gene-targeted mutation can be dramatically altered by the genetic background (12). These strain-specific phenotypes have become evident after crossing certain mutations into non-129/Sv backgrounds. These observations are not surprising, because significant variations in the phenotypes of different spontaneous murine mutations have already been documented. Considerable effort has been expended to cross transgenes or targeted mutations onto uniform backgrounds. This is necessary to control the contribution of genetic background to the phenotypic expression of a transgenic line or targeted mutation. Unfortunately, the traditional method for backcrossing a mutation involves repeated rounds of breeding onto the strain of interest and requires several years to generate mice in which the majority of nonlinked loci have been converted to the backcross strain. This represents a significant obstacle in experiments involving multiple transgenes and/or mutations, as is required for conditional targeting strategies.

Fortunately, novel methods have now been developed to perform accelerated breeding strategies using microsatellite marker-assisted genomic scans (11, 24). This technology permits selection of backcrossed littermates with more limited "contamination" of the starting strain background at each backcross generation. Implemetation of such "speed-congenic" approaches can accelerate the introgression of targeted modifications into defined strain backgrounds as much as twofold by significantly decreasing the number of generations needed to create congenic mice. An alternative approach to rapid backcrossing is to generate mutations directly in non-129/Sv ES cell lines, thus obviating the need to perform additional backcrossing. ES cell lines have now been derived from several different strains, including C57BL/6, DBA/1, and BALB/c. The generation of mutations in these and other inbred ES cell lines will allow the rapid assessment of potential phenotypic differences resulting from the same mutation.

Although strain background-specific phenotypes can complicate interpretation of a gene's function, it can also be an opportunity for gene discovery and identification of modifier genes. A recent report by Farmer et al. (3) highlights the use of classic quantitative trait loci (QTL) analyses in mapping potential modifier genes. In this study, the known contribution of strain background on the severity of spontaneous colitis that develops in IL-10-deficient mice was used to advantage. Specifically, mice homozygous for a disrupted IL-10 gene show severe disease development when crossed onto the C3H genetic background, whereas disease severity is significantly reduced on the B6 background. Detailed histological scoring of colitic lesions in an F₂ population of IL-10-deficient mice generated by intercrosses of C3H.IL-10^/ and B6.IL-10^/ mice identified several distinct phenotypes. Subsequent mapping defined six major QTL, termed cytokine deficiency-induced colitis susceptibility modifiers. Notably, susceptibility loci were contributed both by the colitis-sensitive C3H and the colitis-resistant B6 backgrounds, either as additive, recessive, or epistatic traits. Several candidate genes that map to these loci encode proteins known to play important roles in epithelial integrity (e.g., epidermal growth factor gene), inflammation or immunomodulatory signaling (e.g., NF-B₁ gene, CTLA4, and transforming growth factor- signaling component, SMAD4). Although this study highlights the complexity of genetic interactions contributing to susceptibility to chronic intestinal inflammation, it also demonstrates the power of disease phenotypes imparted by targeted, single-gene mutations to elucidate complex trait phenomena. The refinement of mapping in this model using interval-specific congenic stocks should contribute to discovery of novel genetic loci that affect the interplay of gut microbes, intestinal epithelium, and the innate and adaptive immune systems, and will be a model for future studies aimed at defining the genetic contributions to inflammatory bowel disease.

	CHEMICALLY INDUCED POINT MUTATIONS IN THE MOUSE GENOME: N-ETHYL-N-NITROSOUREA MUTAGENESIS

TOP ABSTRACT TARGETED MUTATION OF GENES... CONDITIONAL GENE-TARGETING... CHARACTERIZATION OF TISSUE-... NEW APPROACHES FOR... STRAIN BACKGROUND... CHEMICALLY INDUCED POINT... REFERENCES

Random genome-wide mutagenesis screens with chemical agents such as N-ethyl-N-nitrosourea (ENU) may find increased use as a strategy for identifying genes involved in normal and pathological gastrointestinal physiology (14). Unlike the gene-targeting methods described above, in which a mutation is generated in a specific gene, ENU can be used to rapidly produce a large number of random mutations that can then be screened for a specific phenotype. Thus ENU mutagenesis screens are said to be "phenotype driven" and can be used to identify both dominant and recessive mutations. Another advantage of this technique is that it generates mainly point mutations, which do not generally result in the total elimination of a protein's function. Point mutations often inactivate or alter the function of specific protein domains, and studies of these variants may provide new information regarding the functions of a specific gene. Multiple centers around the world are now performing large-scale ENU mutagenesis screens, and many of the mutants discovered are being made available to the scientific community. However, these approaches are not limited to large groups or centers. Small-scale screens can be done in individual laboratories but require significant mouse colony space and resources. The future usefulness of this technique for investigators interested in gastroenterology will depend mainly on the development of methods to rapidly screen large numbers of mice for specific defects in the gut or liver.

In summary, advances in genetic engineering technologies coupled with the sequencing of the mouse and human genome are accelerating the discovery of genes and gene function. The development of new conditional gene-targeting techniques, characterization of gastrointestinal tract-specific gene regulatory elements, and advanced techniques to identify modifier genes make this an exciting time for studies of gastrointestinal physiology and disease. Although many of these emerging technologies remain quite sophisticated and will require further refinement, it is clear that their implementation will be required to elucidate the complex interplay among the microflora, epithelium, and immune cells of the gastrointestinal tract.