Gene Therapy (1997) 4, 700–709 1997 Stockton Press All rights reserved 0969-7128/97 $12.00 Efficient gene targeting in mouse embryonic stem cells NS Templeton1, DD Roberts2 and B Safer3 1ABL-Basic Research Program, NCI-FCRDC, Frederick, MD; 2Laboratory of Pathology, National Cancer Institute, National Institutes of Health, Bethesda, MD; and 3Molecular Hematology Branch, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD, USA We developed methods to improve the efficiency of gene cells. Due to the high frequency of targeting, corrected cells correction in mouse embryonic stem cells using homolo- could be isolated by screening colonies obtained after gous recombination of a replacement vector. The absolute growth without selection. Alternatively, colony formation frequency of homologous recombination in mouse embry- and the absolute frequency could be increased by co-plat- onic stem (ES) cells, defined as the frequency of homolo- ing the electroporated cells with nonelectroporated ES cells gous recombination per electroporated cell, is approxi- before the addition of selective medium. These parental mately 10−5 to 10−6 by current procedures. Our method for cells were nonirradiated but were killed in the selective gene targeting in mouse ES cells produces an absolute medium. Plating density and efficiency of colony formation frequency of 10−1. The protocol uses micro-electroporation are therefore critical factors for obtaining a high absolute chambers and a modified electroporation procedure that frequency of homologous recombination. Because this fre- does not cause significant cell death. Plating and growth of quency is extremely high, these methods can be used to the electroporated cells at an optimum density to maintain perform gene targeting without the use of selectable viability significantly increased the recovery of targeted markers. Keywords: gene replacement; stem cells; gene therapy; Lesch–Nyhan syndrome Introduction limited interest for gene therapy. The major concern in applying this technology has been the extremely low Recently, much progress has been attained in the efficient efficiency of this process, especially in pluripotent stem delivery of therapeutic genes using viral or plasmid vec- cells.1 Current methods used for gene targeting in mouse tors in vivo and ex vivo. These therapies show the greatest embryonic stem (ES) cells, for example, produce an abso- promise for treating diseases caused by recessive lute frequency of homologous recombination of approxi- mutations and diseases, such as cancer, that may respond mately 10−6 to 10−5.1,2 The low absolute frequency pre- to transient expression of therapeutic gene products. cludes the use of homologous recombination for gene These approaches cannot be used to treat diseases therapy in vivo, and necessitates the use of selectable mar- resulting from dominant mutations, are limited to deliv- kers in gene targeting ex vivo to facilitate the recovery of ery of relatively small cDNAs, and cannot achieve stable targeted cells. Our goal is to improve the current gene long-term gene expression without integration of the targeting technology to enable the use of homologous vector. Use of integrating vectors, however, presents recombination for gene therapy. additional risks of disrupting essential genes or inacti- Studies of the mechanism of homologous recombi- vation of tumor suppressor genes, leading to neoplasia. nation have identified several steps that could control the Integrating vectors often fail to support long-term gene absolute frequency, such as the absence of required expression. Furthermore, these introduced genes may recombinases or the presence of inhibitors of recombi- lack the complex transcriptional regulation that is essen- nation.1 Alternatively, homologous recombination may tial for the proper function of many genes. be relatively efficient, but the absolute frequency meas- The use of homologous recombination could overcome ured is limited by the efficiency of DNA delivery to the many of these limitations. The use of this approach to nucleus or the ability to clone and propagate cells that correct mutations in an endogenous genomic locus leads have undergone homologous recombination. Recently, to a functional gene in its natural context with complete there have been several reports of highly efficient homo- regulation of expression. Mutations can be corrected in logous recombination in mammalian cells,3–8 suggesting genes of any size, and dominant mutations can be elimin- that the lower frequencies commonly observed in ES cells ated. Furthermore, the corrected gene will be expressed are either cell specific or result from the inability to indefinitely if appropriate cells can be targeted. To date, recover homologous recombinants. Some of the new however, homologous recombination has attracted only technologies use small DNA fragments3 or DNA–RNA hybrid oligonucleotides4 to correct substitutions or three base pair deletions in genomic DNA. Because several Correspondence: NS Templeton, NCI-FCRDC, ABL-Basic Research Pro- gram, PO Box B, Boyles Street, Building 535, Room 226A, Frederick, genetic disorders involve multiple mutations within a MD 21702–1201, USA large region of genomic DNA or involve larger deletions, Received 23 November 1996; accepted 24 March 1997 such as mutations in humans for hypoxanthine High frequency gene targeting NS Templeton et al 701 phosphoribosyltransferase (HPRT) and Factor VIII, we wanted to improve the efficiency of the traditional vec- tors that are used for gene replacement. The absolute frequency of homologous recombination in mouse ES cells, defined as the frequency of homolo- gous recombination per electroporated cell,1 is approxi- mately 10−5 to 10−6 by current procedures. In addition, the efficiency of colony formation for E14TG2a ES cells measured in nonselective medium has been estimated at 3%.2 This estimate is similar for E14TG2a cells grown on gelatinized tissue culture dishes or on irradiated feeder layers. On this basis, the measured absolute frequency of homologous recombination of 10−5 was estimated to reflect a true frequency of 10−3. The true absolute fre- quency was calculated by dividing the observed fre- quency by the efficiency of colony formation for these ES cells.2 The true frequency is therefore 100-fold higher than the observed absolute frequency of 10−5 that was achieved using an improved hprt gene targeting vector.2 Detection of homologous recombination, however, almost always requires growth in selective media. Because the 3% efficiency of colony formation was determined without selection, the possibility should be Figure 1 Partial HPRT genomic loci and targeting DNAs used for correc- considered that the efficiency of colony formation is tion. (a) The normal mouse 5′ HPRT locus.30 The mouse HPRT genomic ′ ′ further lowered by the procedure used for selection. region extending from the 5 end of exon 1 to the 3 end of exon 9, the last HPRT exon, is .33 kb.30 The diagonal lines represent the region that The hprt locus in mouse ES cells is often used to meas- is not drawn to scale. This region is estimated to be 10–15 kb in length. 2,9–13 ure the frequency of homologous recombination. Intron 1 and intron 2 are shown. (b) The deleted HPRT locus9,14 in − Targeting the hprt ES cells, E14TG2a,14 is particularly use- E14TG2a showing positions of the BamHI restriction enzyme sites. The ful for measuring this frequency. HPRT+ targeted cells deletion extends from at least 10 kb 5′ to the HPRT promoter into the can be selected by growth in hypoxanthine, aminopterin, second intron. The dashed lines with arrows indicate the deleted region. thymidine (HAT) medium. E14TG2a are male ES cells, (c) The pMP8-hisD construct. The unfilled rectangle represents the hisD gene expressed by the PGK promoter and is positioned in the 5′ ! 3′ and the HPRT gene is located on the X chromosome; direction with respect to pMP8. The dashed lines that are crossed from b therefore, only one allele needs to be corrected. These to c indicate the regions of homology between the E14TG2a hprt locus cells contain a genomic DNA deletion that extends at and pMP8-hisD. The 3′ BamHI restriction enzyme site of the E14TG2a least 10 kilobases (kb) 5′ to the HPRT promoter region hprt locus and approximately 700 bp 5′ to this site are not included in through to the second intron (Figure 1a, b). Several inves- the targeting DNAs. The BamHI restriction enzyme site used to linearize tigators have used DNA targeting vectors that can correct the DNA prior to electroporation of E14TG2a cells is shown. The hatched area (c, e) represents the pBluescript II SK (±) Phagemid sequences this locus.2,9–13 The absolute frequency of gene targeting −6 −5 (Stratagene). (d) The corrected HPRT locus after homologous recombi- at this locus has been in the range of 10 to 10 , nation with the pMP8-hisD targeting DNA. The filled square indicates depending on the donor DNA design. We have used this the location of the 0.2 kb RsaI mouse hprt genomic fragment. (e) The locus to reexamine the frequency of homologous recom- pMP8 targeting DNA previously described.2 bination in ES cells. Using efficient methods to deliver targeting DNA and to culture targeted cells, we can rou- tinely detect homologous recombination at an absolute encoding histidinol dehydrogenase. A PGK promoter, the frequency of 10−1. natural promoter for the phosphoglycerate kinase gene, was used for hisD gene expression. The second selectable marker cassette, PGK-hisD, was introduced in a location Results known not to disrupt production of normal hprt protein, and where a PGK-neo cassette had been placed in a pre- Analysis of homologous recombination in ES cells vious hprt targeting vector.2 Cells that have stably incor- Because of the low efficiency of colony formation of ES porated pMP8-hisD DNA can survive in medium that cells,2 we first sought to optimize recovery of ES cells lacks histidine and contains l-histidinol.15 Nonhomolo- undergoing homologous recombination without the iso- gous integration of the targeting DNA is operationally lation of colonies.