This essay has been epoduced fom the Geat Expeiments section of http://www.egito.com Gene Tageting: Alteing the Genome in Mice Maio Capecchi Backgound Gene tageting povides the means to ceate stains of mice with specific mutations in vitually any gene. This methodology pemits evaluation of the functions of genes in an intact mammal and systematic genetic dissection of the most complex biological pocesses, such as development, leaning and immunity. The pimay poblem in developing gene tageting was how to geneate a specific mutation in a mammalian cell without alteing any othe site in the genome. This became possible when we discoveed that cultued mammalian cells ae capable of homologous ecombination, allowing a mutant gene injected into a cell to eplace the chomosomal copy. Paallel advances in the isolation of embyonic stem cells made it possible to apply this technique to cells with sufficient developmental potential to allow mutations to be intoduced into the mouse gemline and thei effects ultimately obseved in whole animals. Ou contibutions to gene tageting poceeded in thee phases, the fist beginning in 1977. At that time I was attempting to impove the efficiency with which new genes could be intoduced into mammalian cells. Wigle and Axel had just epoted the fist successful tansfomation of mammalian cells with an exogenous gene, having succeeded in tansfeing a hepes simplex vius thymidine kinase gene (HSV-tk) into tk cultued cells by calcium phosphate copecipitation (Wigle et al., 1977). Although an impotant contibution to somatic cell genetics, this pocedue was not efficient. Appoximately one in one million cells exposed to the calcium phosphate-dna copecipitate acquied the exogenous gene in a functional fom. Using the same expeimental paadigm, I asked whethe I could intoduce tk genes into tk mouse fiboblasts by injecting DNA diectly into thei nuclei (Capecchi, 1980). This pocedue poved extemely
efficient. One thid of the cells that eceived the DNA stably integated the functional tk gene at andom positions within thei genomes and passed it on to thei daughte cells. This high efficiency of DNA tansfe by micoinjection made it pactical fo investigatos to geneate tansgenic mice containing andom insetions of exogenous DNA in thei genomes. This was accomplished by injecting the desied DNA into nuclei of one-cell zygotes and allowing these embyos to come to tem in foste mothes (see Geat Expeiments : Tansgenic Mice: Expession of Foeign Genes in Animals). Efficient functional tansfe of the tk gene into cells equied that it be linked to othe shot vial DNA sequences. When consideing how to impove the efficiency of tansfomation of mammalian cells, I had thought it plausible that vial genomes might contain bits of DNA that enhanced thei ability to establish themselves within a mammalian genome. I succeeded in finding such a sequence in the genome of SV40, a simian DNA vius. When coupled to the HSV-tk gene, the SV40 sequence inceased the efficiency of confeing a tk + phenotype to tk ecipient cells by a facto of moe than 100 (Capecchi, 1980). The enhancement did not appea to be the esult of HSV-tk plasmid eplication within the host cell befoe integation, and the SV40 DNA sequences wee integated into the host genome along with the HSV-tk gene. I concluded that the efficiency-enhancing sequence fom SV40 was eithe inceasing the fequency with which the exogenous DNA was integated into the host genome, o inceasing the pobability that the tk gene, once integated, was being expessed. The SV40 sequence poved to be what would late come to be called an enhance, and these expeiments contibuted to thei definition (Levinson et al., 1982). Ou ealy expeience with them would pove invaluable in developing a eliable tageting appoach. Although the ability to tansfom cells efficiently was itself extemely useful, the obsevation that I found most fascinating fom these ealy DNA micoinjection expeiments was that when multiple copies of the tk plasmid wee injected into cells, they wee always integated into the genome as head to tail concatemes. This stuctue esulted even though integation could appaently take place thoughout the host chomosomes. I ealized that concatemes could be geneated eithe by eplication (e.g., a olling cicle mechanism) o by homologous ecombination among the injected plasmids, as shown in Figue 1. We wee able to pove that they wee fomed by homologous ecombination (Folge et al., 1982). This was a statling discovey at the time, because it had always been assumed that homologous ecombination was esticted to
gem cells, whee its pupose was to shuffle the paental genetic taits to ensue thei boad dissemination among the offsping. Finding evidence of homologous ecombination in mouse fiboblasts implied that somatic cells also possessed the necessay machiney. We suspected that the homologous ecombination machiney in somatic cells was vey efficient, because when we injected moe than one hunded tk plasmid molecules pe cell they wee all incopoated into a single, odeed head-to-tail concateme. It was immediately clea that if we could haness this machiney to cay out homologous ecombination between a newly intoduced DNA molecule of ou choice and the same DNA sequence in a ecipient cell's chomosome, we would have the ability to specifically mutate o modify vitually any cellula gene. The second phase of ou quest fo gene tageting equied that we become familia with the substate pefeences and eaction poducts of the cellula ecombination machiney. We did this by studying ecombination between cointoduced DNA substates. These expeiments showed that the endogenous ecombination machiney could mediate a wide spectum of eactions. Although a distinct bias towads nonecipocal eactions was obseved, both ecipocal exchanges and nonecipocal ecombination events wee appaent among the poducts of ecombination (Folge, Thomas, and Capecchi, 1985) (fo a discussion of ecombination see Genes 2000 Molecula Biology : 15.3: Beakage and eunion involves heteoduplex DNA. Eithe type of eaction would suit ou pupose. We also found that the ability to cay out homologous ecombination depends on a cell's position in the cell cycle, showing a peak of activity in ealy S-phase (Wong and Capecchi, 1987), and that linea DNA molecules appeaed to be bette substates fo homologous ecombination than cicula o supecoiled molecules (Thomas, Folge, and Capecchi, 1986). Togethe, these studies demonstated that the cellula enzymatic machiney fo ecombination not only existed, but that when optimized it was sufficiently efficient that it might indeed be exploited to mediate homologous ecombination between exogenous DNA sequences and chomosomal sequences. The thid phase of ou effot was to devise a way to extend homologous ecombination to sites in the genome itself. To make things as simple as possible in the initial stages, we began with studies involving homologous ecombination with specific taget sites that we had ceated within the chomosomes. It was clea pio to the initiation of these expeiments that the fequency of tageting to a single copy gene in mammalian cells was likely to be low and that insetion of the tageting vecto at
sites othe than the taget locus would be fa moe common (simply because thee wee so many moe of them). To pemit detection of ae homologous ecombination events, we designed lines of ecipient cells with specific taget loci that would povide a selection potocol to eliminate cells not containing the desied ecombination poducts, as outlined in Figue 2. The fist step of this scheme equied geneation of cell lines containing andom insetions of a defective neomycin esistance (neo ) gene containing eithe a deletion o a point mutation. The second step involved intoducing into those cell lines a tageting vecto caying a neo gene with an inactivating mutation diffeent fom that of the taget neo gene. Homologous ecombination between the tageting vecto and the cognate sequence in the ecipient cell genome would geneate a functional neo gene fom the two defective pats. Cells containing this ecombination poduct would be esistant to the dug G418, which kills cells that lack a functional neo gene. We geneated ecipient cell lines containing single copies of the defective neo gene, lines containing multiple copies of the gene in head-to-tail concatemes and, by inhibiting concateme fomation, lines with multiple defective neo taget genes, each located on a sepaate chomosome. The diffeent cell lines allowed us to evaluate how the numbe and location of tagets within the ecipient cell's genome influenced the tageting fequency. Acquisition of neomycin esistance by cells injected with a neo tageting vecto occued at a fequency highe than we anticipated (Thomas, Folge, and Capecchi, 1986). One in a thousand injected cells yielded pogeny with functional neo genes. Southen tansfe and sequence analysis showed that the neomycin esistant phenotype was indeed due to coection of the defective chomosomal neo gene by homologous ecombination with the incoming DNA, as we had intended. Coection of a point mutation by gene tageting occued at a fequency that was five odes of magnitude highe than the spontaneous evesion fequency. Coection of a deletion mutation was effectively at an infinitely highe fequency elative to its spontaneous evesion fequency, since this mutation was neve obseved to evet. The fequency with which the defective chomosomal neo genes wee coected was shown to be independent of the numbe of neo tageting vecto molecules intoduced into the ecipient cells and also independent of the numbe of taget neo genes pesent within the ecipient cells. Thus, intoduction of one molecule pe cell (based upon an assumed Poisson distibution of input DNA) yielded the same successful tageting fequency
as injection of hundeds of molecules pe cell. These esults suggested that the tageting fequency might be limited by the availability of the machiney needed to mediate homologous ecombination athe than by the numbe of incoming vecto molecules o the numbe of tagets within the ecipient cell lines. We wee encouaged by the obsevation that the tageting fequencies in independent cell lines with taget sites at vaious genomic locations wee identical, suggesting that a lage faction of the genome should be accessible to gene tageting. Having established that gene tageting could be achieved in cultued mammalian cells and having detemined some of the paametes that influenced its fequency, we wee now eady to extend gene tageting to endogenous loci and to the whole mouse. The Expeiment At this point, the constaints of what we wee tying to accomplish made us face the poblem of what cells to use in ode to continue ou wok. We had developed the tageting pocedue in cells chosen lagely fo puposes of convenience. They, howeve, lacked the developmental plasticity that would be equied to povide a way of conveting a tageted cell into a mouse beaing the mutation. The obvious appoach was to pefom gene-tageting diectly in one-cell mouse embyos. This would both poduce a mutant mouse and guaantee intoduction of the tageted mutation into the mouse gemline so that the mutation could be popagated. Unfotunately, this appoach was impactical because of the low fequency of tageted homologous ecombination compaed to andom integation of the tageting vecto. Instead, it appeaed that ou only choice was to do gene tageting with some kind of cultued cells so that we could select the elatively ae tageted ecombinants. This appoach would equie that we use cells with the geatest possible developmental potential, since we would have to geneate a mutant mouse by intoducing tageted cells into an ealy embyo, whee they would have to be able to contibute daughtes to all of the tissues of the developing mouse, including the gemline. Even with mutant cells in all of its tissues, howeve, most of the cells in the mouse would lack the mutation, so it would have to be bed in ode to finally poduce mice with the mutation in evey cell. It initially appeaed that the only cells capable of meeting ou equiements fo developmental plasticity wee cultued multipotent embyonal cacinoma (EC) cells, a cell line deived fom tumos fomed by ealy embyos. In the late 1970s, EC
cells appeaed to offe the only potential avenue between the cultue dish and the whole animal. When these cells ae intoduced into a mouse peimplantation embyo, and the embyo allowed to implant into the uteus of a foste mothe, they contibute to the fomation of most somatic tissues. Howeve, afte attending seminas and confeences on EC cell biology ove a numbe of yeas, I became painfully awae of the fustations associated with attempts to obtain EC cell contibutions to the fomation of the mouse gemline. Finally, at a Godon Confeence in the summe of 1984, I head a discussion about ES cells (then called EK cells), cultued multipotent embyonic stem cells capable of populating the mouse gemline. These cells wee developed in Matin Evans's laboatoy, and diffeed fom EC cells in that they wee obtained fom a nomal ealy embyo athe than fom a mouse tumo. In the winte of 1985, I aanged to spend a week in Evans's laboatoy to lean how to cultue ES cells and geneate chimeic mice fom them. It was just befoe Chistmas, a mavelous time to be in Cambidge, England. In the beginning of 1986, all of ou effot switched to pefoming gene tageting in ES cells. We also decided to use electopoation as the means of intoducing the tageting vectos into ES cells. Although micoinjection was odes of magnitude moe efficient than electopoation as means fo geneating cell lines with tageted mutations, injection had to be done one cell at a time. With electopoation, 10 8 cells could be manipulated in a single expeiment, and if successful, the methodology of gene tageting would be moe eadily tansfeable to othe investigatos. The fist gene that we disupted in ES cells was that fo hypoxanthine phosphoibosyl tansfease (hpt), which we chose because we could select diectly fo cells containing a disuption of it. Futhe, because the hpt gene is located on the X chomosome and the host ES cell line was deived fom a male mouse, only a single locus had to be disupted to yield hpt cell lines. The stategy that we employed is shown in Figue 3. We placed a neo gene within an inton in the middle of cloned hptgenomic sequences, then intoduced the vecto beaing the disupted hpt gene into ES cells by electopoation, and finally selected fo cells esistant to both G418 and 6-thioguanine (6- TG), a dug toxic to cells with a functional hpt gene. All ecipient cell lines selected in this way had lost hpt function as a esult of tageted disuption of the hpt locus (Thomas and Capecchi, 1987). Because we foesaw that the neo gene could potentially be used as a positive selectable gene fo the disuption of many genes, it
was citical that neo expession be mediated by an enhance that would function egadless of its location within the ES cell genome. It was hee that ou ealie expeience with what came to be known as enhances poved of value. To encouage such stong neo expession, the enhance we used to dive it was a duplicated, mutant polyoma-vius enhance known to pemit obust vial eplication in mouse embyonal cacinoma cells (Thomas and Capecchi, 1987). Subsequently, this stategy of using a ubiquitously-expessed, positive selectable make has been used by many investigatos to disupt a wide spectum of genes in ES cells. Ou expeiments with the hpt gene showed that ES cells wee able to mediate homologous ecombination. They also clealy demonstated the need fo the ability to select tageted cells. Unde optimal conditions, the absolute tageting fequency of the hpt disuption was one cell caying a tageted mutation pe 10 5 cells that wee electopoated. Fo each successfully tageted cell, appoximately one thousand contained a andom insetion of the neo taget vecto into the genome. The selection potocols woked, howeve, and, just as impotantly, did not alte the pluipotent state of ES cells in cultue (as we demonstated in sepaate expeiments). I believe these disuption expeiments with the hpt locus played a pivotal ole in the development of the field by making othe investigatos ealize that specific modification of a mammalian genome in vivo was actually possible. Many people wee encouaged to exploe the use of tageted gene disuption as a means fo detemining the function of mammalian genes they wee studying. The high atio (~1-1000) of nonhomologous to homologous ecombination events that we had obseved in the hpt disuption expeiment pesented a potential poblem fo the geneal applicability of gene tageting. Because the disuption of most genes is not expected to poduce a phenotype, like hpt -, that is selectable at the cell level, an investigato seeking a specific gene disuption would theefoe eithe need to conduct a tedious DNA sceen though many colonies of cells to identify the ae colony containing the desied tageting event, o devise a selection scheme able to enich fo cells containing the tageting event. Late in 1986, I conceived of a vey geneal stategy to enich fo cells in which the homologous tageting event had occued. The stategy, known as positive-negative selection, uses two components (see Figue 4). One component is a positive selectable gene, neo, used as a make to select fo ecipient cells that have incopoated the tageting vecto anywhee in thei genome (i.e., at the taget site by homologous
ecombination o at othe sites by nonhomologous ecombination). The second component is a negative selectable gene, HSV-tk, which is located at the end of the lineaized tageting vecto and is used to select against cells containing andom insetions of the tageting vecto. Thus the "positive" selection eniches fo all ecipient cells that have incopoated the intoduced vecto, and the "negative" selection eliminates those that have incopoated it at nonhomologous sites. The net effect is to enich fo cells in which the desied homologous tageting event has occued. The scheme was based on the citical obsevation, in ealie expeiments, that insetion of exogenous DNA at nonhomologous sites included the ends of the tageting vecto. Convesely, homologous ecombination eliminates the ends of the tageting vecto, theeby discading the "negative" selectable gene. The stength of this enichment pocedue is that it is independent of the function of the gene that is being disupted and succeeds whethe o not the gene is expessed in the cultued ecipient ES cell. We tested this scheme and poved it to be wokable in an expeiment involving the disuption of the poto-oncogene int-2(mansou, Thomas, and Capecchi, 1988), and positive-negative selection has since become the most fequently used pocedue to enich fo cells containing tageting events. The Legacy The use of gene tageting to evaluate the functions of genes in the living mouse is now a outine pocedue and is used in hundeds of laboatoies all ove the wold. It is vey gatifying to be able to pick up almost any majo jounal in the biological sciences and find the desciption of yet anothe so-called knockout mouse. The in vivo functions of well ove 7,000 genes have been analyzed using the gene tageting technology. This numbe is paticulaly impessive consideing that no conceted pogam was involved to geneate this collection of mouse lines with tageted mutations. Rathe, it was accomplished as a cottage industy, by the collective wok of many individual laboatoies. The gene tageting potocol is now geneally pefomed as follows: Fist, the desied DNA sequence alteation is intoduced into a cloned copy of the gene of inteest by standad ecombinant DNA technology. Second, as shown in Figue 5, the alteation is then tansfeed, by means of homologous ecombination, to the cognate genomic locus in pluipotent, mouse embyo-deived stem (ES) cells and cultued ecipient ES cell lines caying the alteation ae selected. Thid, ES cells
containing the alteed genetic locus ae injected into mouse blastocysts, which ae in tun bought to tem in foste mothes, geneating chimeic mice that ae capable of tansmitting the modified genetic locus to thei offsping. This geneates mice heteozygous fo the mutation in evey cell. Finally, if the intent is to evaluate the consequences of any type of ecessive mutation, heteozygous siblings ae intebed to yield mice homozygous fo the mutation. Figue 6 outlines these steps, fom the injection of cultued ES cell lines containing the desied tageted gene modification to the geneation of gemline chimeas and thei offsping. Gene tageting in mice has thus fa been used pimaily to completely disupt specific genes in ode to ty to detemine thei function by ceating the seveest possible phenotype. Howeve, it can be used to manipulate the mouse genome in almost any desied manne. Fo example, it is possible to geneate an allelic seies of mutations in a specific gene to evaluate the effects of gain-of-function o patial loss-of-function mutations, as well as the effect of complete loss-of-function mutations. To pemit the evaluation of multiple potential oles of a gene, paticulaly if the loss-of-function allele compomises the embyo at ealy stages of development, the Ce-loxP and Flp/FRT site-specific ecombination systems, in concet with gene tageting, can be used to estict the effect of a mutation to specific cells, tissues, o tempoal peiods (Gu et al., 1994; Moon, Boulet, and Capecchi, 2000; Moon and Capecchi, 2000). Thee is no question that the mouse is a vey complex oganism, but it povides a model vey simila to ouselves, and the boad ange of genetic manipulations in it made possible though gene tageting pomises the potential fo decipheing even the most elaboate of its biological pocesses. The Autho Maio R. Capecchi is Distinguished Pofesso of Human Genetics and Investigato of the Howad Hughes Medical Institute at the Univesity of Utah School of Medicine. D. Capecchi is a membe of the National Academy of Sciences. His honos include the Bistol-Myes Squibb Awad fo Distinguished Achievement in Neuoscience Reseach, the Gaidne Foundation Intenational Awad fo Outstanding Achievements in the Field of Medical Science, the Geneal Motos Copoation's Alfed P. Sloan J. Pize fo Outstanding Basic Science Contibutions to Cance Reseach, the Bio-Analytica Pize fo Distinguished Contibutions to the Field of Molecula Biology (Gemany), the Kyoto
Pize fo Eminent Achievements in the Field of Life Sciences, the Fanklin Medal fo Advancing Ou Knowledge of the Physical Sciences, the Rosenblatt Pize fo Excellence in Academic Achievements, and the Baxte Awad fo Distinguished Reseach in the Biomedical Sciences. Contact Details Maio Capecchi Howad Hughes Medical Institute Univesity of Utah School of Medicine 15 Noth 2030 East Room 5100 Salt Lake City, UT 84112-5331 Phone: 801-581-7096 Fax: 801-585-3425 E-mail: maio.capecchi@genetics.utah.edu Last Revised on 9-1-2001 The citation fo this eview is http://www.egito.com/sevlet/lookup?expt=capecchi Capecchi, M. R. (1980). High efficiency tansfomation by diect micoinjection of DNA into cultued mammalian cells. Cell 22, 479-488. Abstact Folge, K. R., Thomas, K., and Capecchi, M. R. (1985). Nonecipocal exchanges of infomation between DNA duplexes coinjected into mammalian cell nuclei. Mol. Cell Biol. 5, 59-69. Abstact Folge, K. R., Wong, E. A., Wahl, G., and Capecchi, M. R. (1982). Pattens of integation of DNA micoinjected into cultued mammalian cells: evidence fo homologous ecombination between injected plasmid DNA molecules. Mol. Cell Biol. 2, 1372-1387. Abstact Gu, H., Math, J. D., Oban, P. C., Mossmann, H., and Rajewsky, K. (1994). Deletion of a DNA polymease beta gene segment in T cells using cell type-specific gene tageting. Science 265, 103-106. Abstact Levinson, B., Khouy, G., Woude, G. V., and Guss, P. (1982). Activation of SV40 genome by 72-base pai tandem epeats of Moloney sacoma vius. Natue 295, 568-572. Abstact Mansou, S. L., Thomas, K. R., and Capecchi, M. R. (1988). Disuption of the poto-oncogene int-2 in mouse embyo-deived stem cells: a geneal stategy fo tageting mutations to non-selectable genes. Natue 336, 348-352. Abstact Moon, A. M., Boulet, A. M., and Capecchi, M. R. (2000). Nomal limb development in conditional mutants of
Fgf4. Development 127, 989-996. Abstact Full Text Moon, A. M. and Capecchi, M. R. (2000). Fgf8 is equied fo outgowth and pattening of the limbs. Natue. Nat. Genet. 26, 455-459. Abstact Full Text Thomas, K. R., Folge, K. R., and Capecchi, M. R. (1986). High fequency tageting of genes to specific sites in the mammalian genome. Cell 44, 419-428. Abstact Thomas, K. R. and Capecchi, M. R. (1987). Site-diected mutagenesis by gene tageting in mouse embyo-deived stem cells. Cell 51, 503-512. Abstact Wigle, M., Silvestein, S., Lee, L., Pellice, A., Chen, Y. and Axel, R. (1977). Tansfe of puified Hepes Vius thymidine kinase gene to cultued mouse cells. Cell 11, 223-232. Abstact Wong, E. A. and Capecchi, M. R. (1987). Homologous ecombination between coinjected DNA sequences peaks in ealy to mid-s phase. Mol. Cell Biol. 7, 2294-2295. Abstact This essay is epoduced fom the Geat Expeiments section of http://www.egito.com