Efficient Integration of Short Interspersed Element-flanked Foreign DNA via Homologous Recombination*

We investigated whether mouse short interspersed elements (SINEs) could influence the recombination frequency of foreign DNA. Vectors harboring a reporter gene in combinations of SINEs B1 and/or B2 or a portion of long interspersed element-1 were prepared and tested in vitro by a colony assay using HC11 murine mammary epithelial cells and in vivo by microinjection into fetilized mouse eggs. In transfected HC11 cells, the number of colonies surviving G418 selection increased by 3.5-fold compared with control when the reporter was flanked by fused B1-B2 sequences. Similar results were obtained from microinjection study; in fetuses 11.5 days post coitum, transgene positives in control and SINE-flanked vectors were 16 and 53%, respectively. Individual B1- and B2-harboring vectors showed equivalent activities with each other, as determined by the colony assay (2.8-fold versus 3.2-fold compared with control). We determined the contribution of homologous recombination to the SINE-mediated increase in integration frequency through a polymerase chain reaction-based strategy; in more than half of embryos transgenes underwent homologous recombinations involving B1 sequences. These results demonstrate that the SINE sequences can increase the integration rate of foreign DNA and that such an increase is most likely due to the enhancement of homologous recombination.

Mammalian genomes contain both unique and repetitive DNA sequences. In addition to the highly abundant satellite DNA, which occurs in tandem arrays mostly clustered in heterochromatic regions, the genome also contains repeated sequences that are interspersed among single-copy sequences of euchromatic regions, such as the short interspersed repeated DNA elements (SINEs) 1 and long interspersed repeated DNA elements. Many of these repetitive DNA family members are present in extremely high copy numbers, more than 10 5 copies/ haploid genome (reviewed in Refs. 1 and 2).
Both B1 (130 bp) and B2 (190 bp) sequences are murine SINEs with unknown functions (3). From 130,000 to 180,000 copies of the B1 repeat and from 80,000 to 120,000 copies of the B2 repeat are present in the mouse genome scattered on all chromosomes (4). Kramerov et al. (5) reported that members of the B1 family were nearly identical with one another. Individual members of the B2 family also show a high degree of homology, displaying only 3-5% deviations from the B2 consensus sequence (3). The presence of so many short homologous sequences throughout the mammalian and higher eucaryotic genomes has implications in genetic recombinations (1,6). There have been several reports about the participation of SINE equivalents in homologous recombinations; most of these data came from analyses of disease-related genes, such as the globin gene and low density lipoprotein receptor genes. Various DNA clones obtained from subjects with thalassemia, hereditary persistence of fetal hemoglobin, or familial hypercholesterolemia were shown to undergo deletions or insertions by Alu-Alu homologous recombinations or illegitimate recombinations between Alu family members and unrelated sequences and thus gave a clue of nonrandom rearrangements (7). Also, the duplications of the entire growth hormone gene (6), the lysozyme gene, (8), and a part of exons of the low density lipoprotein receptor gene (9) in evolution were proposed to have occurred via Alu-Alu recombinations. These results suggest that the number of recombinations involving Alu-like members is much higher than expected (1) and that these sequences, in particular, serve as hot spots for recombination events (6,10).
The introduction of exogenous DNA into mammalian cells has become an increasingly important procedure. An efficient integration of delivered DNA into the host genome is the most critical step in the whole procedure of many biological applications, especially in areas such as transgenesis and gene therapy. However, little information is available about the actual molecular events involved in the integration process. The integration phenomenon appears to be predominantly random and nonhomologous (11,12); thus, as far as there is no additional machinery or modulatory elements to accelerate the process, the integration of a foreign sequence into the genome might occur nearly by chance. Such a passive integration pattern may also limit the integration frequency. Therefore, the development of strategies that could lead to an efficient, nonrandom integration of foreign DNA could have a far reaching impact on areas where efficient gene transfer is a crucial step.
To design a vector that can mediate a highly efficient integration of foreign DNA, we searched for sequences that have a potential of increasing the recombination frequency between the input DNA and the host genomic loci. Because the SINE sequences such as B1 and B2 and the long interspersed repeated DNA element sequences such as L1 exist in high copy numbers dispersed throughout the mouse genome, we examined the possibility that these interspersed repetitive sequences could efficiently guide an exogenous DNA into genomic * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
loci. It was demonstrated that the mouse SINE repetitive sequences, such as B1 and B2, could significantly increase the integration frequency of flanking sequences in cultured cells and fetuses and that the elevated integration rate was mostly due to an increase in homologous recombination between the exogenous and endogenous SINE repetitive sequences.
To make p(B1) 2 ␤geo, p(B2) 2 ␤geo and p(B1 d ) 2 ␤geo vectors, pSP73 was digested with BglII and filled with Klenow enzyme. The linearized vector was added with d(T) to their 3Ј ends using Taq polymerase to prepare a kind of T-vector (15), into which a B1, B2, or deleted B1 (B1 d ) monomer was inserted (pSB1, pSB2, or pSB1 d ). The B1 d sequence was amplified like the B1 sequence but using another B1 3Ј primer, 5Ј-AG-CCCTAGCTGTCCTGG-3Ј. The pSB1, pSB2, and pSB1 d vectors were digested with PvuII, added with d(T) in the same mannner as described above, and inserted once again with B1, B2, or B1 d to make p(B1) 2 , p(B2) 2 , and p(B1 d ) 2 vectors. The orientations of the repetitive sequences were confirmed by restriction digestions with BamHI, BglII, and XhoI. The 4.7-kb PGK␤geo fragment was inserted into the HindIII/SalI sites of these vectors to generate p(B1) 2 ␤geo, p(B2) 2 ␤geo, and p(B1 d ) 2 ␤geo vectors.
Cell Culture, Transfection, and Southern Blot Analysis of Selected Colonies-An HC11 cell, a murine mammary epithelial cell line, was cultured as described previously (16). Transfection was performed by the calcium phosphate precipitation method. 10 g of test plasmid was cotransfected with 1 g of a Rous sarcoma virus-driven luciferase control plasmid. 48 h after the transfection cells were collected, a portion of them was assayed for luciferase activity to measure transfection efficiency, whereas the rest was replated onto a 100-mm diameter dish (Nunc) in various densities in a medium containing G418 (0.2 mg/ml, Life Technologies, Inc.) to obtain stable transfectants. The G418-resistant, Giemsa-stained colonies were counted after 10 -14 days of selection and normalized for transfection efficiency. For the luciferase assay a 100 g of whole protein extract was added to 350 l of assay buffer (25 mM glycylglycine, pH 7.8, 10 mM MgSO 4 , 2 mM ATP), and the luciferase activity was measured using a luminometer (Berthold model Lumat, Germany) that injects 100 l of a 0.05 mM luciferin (sodium salt, Sigma) solution into the reaction vessel. For Southern blot analysis G418resistant colonies were selected individually using a cloning cylinder (Sigma), expanded into large populations, and harvested to prepare genomic DNAs. 10 g of genomic DNA was digested with HindIII, which cuts the vectors only once, size-fractionated, and transferred to a nylone membrane. The resulting blot was hybridized with a radiolabeled 1.9-kb SacI fragment of p␤geo vector spanning a part of the LacZ and Neo R gene, washed, and exposed to an x-ray film according to the standard procedure.
Manipulation and Culture of Mouse Embryos-The procedures for the manipulation of embryos were previously described (17). 4 -5-week-old BCF1 mice were superovulated by intraperitoneal injections (48 h apart) of serum gonadotrophin from a pregnant mare (Folligon, Intervet) and human chorionic gonadotrophin (Chorulon, Intervet) and mated with 8-week-old males of the same strain. Microinjection was performed 24 h after human chorionic gonadotrophin injection. The plasmid DNAs used in the microinjection were linearized by digestion with ScaI and purified through Elutip-d (Schleicher & Schuell). Microinjected eggs were allowed to develop in vitro into blastocyst-stage embryos in M16 media (Sigma) or transferred at 2-cell stage to pseudopregnant ICR foster mice to obtain 11.5-dpc fetuses. ␤-Galactosidase staining of embryos was carried out as described (18). Briefly, the embryos were rinsed with phosphate-buffered saline, fixed in 2% paraformaldehyde and 0.02% glutaraldehyde in phosphate-buffered saline for 30 min (for blastocysts) or 1 h (for fetuses). Staining was done for 3 h (blastocysts) or overnight (fetuses) at 30°C in a solution containing 0.1% X-gal, 2 mM MgCl 2 , 5 mM EGTA, 0.2% Nonidet P-40, 5 mM K 3 Fe(CN) 6 , and 5 mM K 4 Fe(CN) 6 ⅐6H 2 O. Stained samples were stored at 4°C in phosphate-buffered saline.
PCR-based Detection of Transgenes and Homologous Recombination Events-To detect transgenes in blastocysts the embryos were individually transferred to a 0.5-ml PCR tube containing embryo lysis buffer (19) and incubated at 50°C for 30 min, boiled at 95°C for 15 min, and then subjected to PCR. To analyze fetuses they were digested with proteinase K in an extraction buffer (10 mM Tris-Cl, pH 8.0, 0.1 M EDTA, pH 8.0, 0.5% SDS, and 20 mg/ml pancreatic RNase) and genomic DNAs were prepared as described previously (20). The PCR primers used were 5Ј-GCTTGGGTGGAGAGGCTATTCG-3Ј and 5Ј-GTAAAGC-ACGAGGAAGCGGTCAGCC-3Ј and were designed to amplify a 692-bp region of the Neo R sequence in the transgene. PCR was performed for 30 cycles (for fetuses) or 40 cycles (for embryos) at 95°C for 1 min, 55°C for 1 min, and 72°C for 1 min and for one cycle of final extension at 72°C for 10 min. For the detection of homologous recombinants in blastocyst embryos, fully expanded blastocyst embryos were treated as mentioned above and subjected to PCR analysis using a primer set, P137 (5Ј-CC-ACCAAAGAACGGAGCC-3Ј) and PB1d-3 (5Ј-AGACAGGGTTTCTCTG-TGTA-3Ј). PCR was carried out for 30 cycles at 95°C for 1 min, 58°C for 1 min, and 72°C for 0.5 min and for one cycle of final extension at 72°C for 10 min. The PCR product was size-fractionated by electrophoresis on a 1.75% agarose gel, transferred to a nylone membrane, and subjected to Southern blot analysis. The probe used was 500 bp of the PGK promoter region prepared from p␤geo by HindIII/BamHI digestion.

Construction of Repetitive Sequence-containing Plasmids and Characterization of Their
Integration-mediating Activity-B1 and B2 sequences were amplified from CBA genomic DNA by PCR and fused with each other in the same orientation (named B1/2). The 3Ј A-rich region that normally exists in endogenous copies was not included in B1/2. We also amplified about 710 bp of L1 sequence encompassing part of the 5Јuntranslated region and ORF1, a segment of L1 reported to be present at 4 ϫ 10 4 copies/genome (21), and confirmed its identity and orientation by digestion with PstI. Using these elements, we constructed a series of vectors having various combinations of the repetitive sequences (Fig. 1A). The rationale for designing a fused construct was such that if the integration rate of foreign DNA is largely determined by the frequency of collision between exogenous and endogenous DNA elements sharing high sequence homology, the fused heterodimer might be more beneficial in increasing the integration rate than either of the monomers. All the test vectors contained a PGK␤geo reporter gene. We analyzed the integration frequency of these constructs by a colony assay, which measured the number of G418-resistant colonies (22,23).
HC11 murine mammary epithelial cell line was transfected with these vectors and allowed to form stable transformants under the drug selection; for later calibration a Rous sarcoma virus-driven luciferase vector was included as an internal standard. After 10 -14 days of selection, the surviving colonies were stained with X-gal or Giemsa, and the number of colonies was counted. As shown in Fig. 1B, the number of colonies was reproducibly higher in repetitive sequence-containing vectors. Compared with the p␤geo control, the colony number increased by about 1.8-fold when the ␤geo sequence was flanked with a single copy of a fused B1/2 sequence and by about 3.5-fold when the ␤geo sequence was entrapped within a pair of B1/2. A significant level of increase in integration frequency was also observed when one or two copies of the L1 sequence was used as the source of the repetitive sequences. These results suggest that the repetitive sequences, especially the B1/2 sequence, positively influenced the integration rate of a flanking transgene. The B1 and B2 sequences used to prepare the B1/2 in this experiment were sequenced and compared with the previously reported ones (Fig. 2); the 130 bp of B1 sequence showed 90% homology with the consensus sequence reported by Maraia (24), and the 135 bp of B2 sequence shared 92% homology with the B2 sequence reported in GenBank TM (accession number M31441).
To estimate the copy number of transgenes genomic DNAs from each of independent G418-resistant clones transfected with the ScaI-linearized p␤geo or p(B1/2) 2 ␤geo vector were subjected to Southern blot analysis after digestion with HindIII, which cuts the vector only once. When hybridized with a 1.9-kb Neo R sequence as the probe (Fig. 1A), about 49% (18/37) of the p(B1/2) 2 ␤geo clones turned out to carry only one or two copies of transgenes, whereas about 28% (5/18) of the p␤geo clones showed one or two copies of transgenes.
Detection of Transgenes in 11.5-dpc Mouse Fetuses Microinjected with p␤geo and p(B1/2) 2 ␤geo Vectors-The results of colony assay represent the combined consequences of the integration and subsequent expression of the exogenous gene. However, there could be colonies that did not survive the drug selection because of inefficient expression of the Neo R gene, even though they had the transgenes integrated in the genome. To assess the effect of B1/2 on the integration efficiency alone and to clarify whether the SINE-mediated integration and expression is maintained through the later stage of mouse development, we evaluated the integration frequency mediated by B1/2 in an in vivo system by microinjecting the B1/2-containing recombinant DNA into the pronuclei of fertilized mouse one-cell embryos.
BCF1 mouse eggs were microinjected with linearized p(B1/ 2) 2 ␤geo or p␤geo DNA and transferred to pseudo-pregnant ICR mice. At 11.5 dpc the fetuses were isolated, stained with X-gal, and analyzed for the presence of transgenes by PCR and Southern blot analysis. The stage of 11.5 dpc was chosen because it is one of the most critical stages in the murine development that are associated with major changes in the developmental pattern and thus the gene expression pattern (25). As shown in Table I, 4 of 25 fetuses (16%) from p␤geo were identified as transgene-positive, while 17 of 32 (53.3%) from p(B1/2) 2 ␤geo turned out to carry the transgene. Among the 17 fetuses carrying the p(B1/2) 2␤ geo sequence, 8 stained positive for ␤-gal in isolated regions of the body, 2 showed an unrestricted expression pattern over the whole body, and the remaining 7 (Fig. 3) did not show any ␤-gal expression. In the case of the p␤geo control, two of the four fetuses expressed ␤-gal. Thus, the integration rates, not only in cultured HC11 cells but also in fetuses, were shown to be significantly higher in the SINEcontaining vector than in the control.
Contribution of Individual Components of the Fused B1/B2 Sequence to the Increase in Integration Frequency-Although both B1 and B2 sequences belong to the mouse SINE family, they apparently do not share any evolutionary relationships. Because they are not physically associated with each other in the mouse genome, the contribution of each element to the enhancement of flanking DNA integration could be significantly different if the process is dependent on the DNA sequence rather than on the copy number of the repetitive sequence in the genome.
To determine which of the two SINE members played a dominant role in increasing the integration rates of foreign DNA, two vectors containing either monomeric B1 or B2 sequence at both sides of the ␤geo reporter gene were constructed (Fig. 4A). In the newly prepared p(B1) 2 ␤geo and p(B2) 2 ␤geo vectors, the orientation of the B1 or B2 sequences relative to the reporter gene was reversed compared with that of p(B1/ 2) 2 ␤geo to examine whether the orientation of the repetitive sequences could affect the expression of the reporter gene and thus the integration frequency. Colony assay was performed according to the above-mentioned procedure. The result showed that each of the B1 and B2 monomer sequences can enhance the integration frequency to a similar degree (Fig. 4B).
Moreover, the integration frequency of the B1/2 fusion sequence was only slightly higher than those with either of the monomeric sequences inserted in the reverse orientation. Thus, the three constructs carrying the repetitive modules seemed to be functionally equivalent in guiding the transgenes into the genome. Considering the results of Figs. 1B and 4B, it appears that the number of isolated repetitive sequence modules in a vector is a critical variable in increasing the chances for these sequences to collide with the endogenous homologs.
Identification of Homologous Recombination Events in (p(B1 d ) 2 ␤geo-injected Blastocyst Embryos by Hord-PCR-Because the only difference between the repetitive sequenceflanked vector and the control vector was the presence of the repetitive sequences, it is reasonable to infer that the repetitive sequence was responsible for the increase in integration rates of transgenes. One of the mechanisms underlying this increase could be the enormous chance of homologous recombination of the incoming B1 or B2 sequence of the vector with the corresponding, high copy numbered cellular homologs. To determine the nature of the integration events, we prepared a vector that can test the sequence-dependent (homologous) recombination events between the two homologs of B1 sequences. Basically, the construct was equal to p(B1) 2 ␤geo except for a deletion of a part (20 bp of B1 3Ј-terminal region) of the repetitive sequence (p(B1 d ) 2 ␤geo; Fig. 5A). Lehrman et al. (9) reported that the examined recombinational breakpoints of Alu sequence were mostly centered in the middle region, especially between the two RNA polymerase III promoters. Therefore, the p(B1 d ) 2 ␤geo construct was designed to retain the entire middle region of the B1 sequence, lacking the proximal part only, so that the inherent recombination activity would not be disturbed. The p(B1 d ) 2 ␤geo vector has been evaluated in vitro and in vivo for the integration capacity before it is applied for the analysis of B1-mediated integration events. In a colony assay, the p(B1 d ) 2 ␤geo was shown to be capable of integrating at a frequency about two times greater than that of the p␤geo control vector (Fig. 5B). Also, 52% (36/69) of the blastocyst embryos microinjected with linearized p(B1 d ) 2 ␤geo DNA were transgene-positive when analyzed by a combined PCR and Southern blot procedure (Fig. 5C). These results once again demonstrate the high integration-mediating capability of the B1 d sequence. The extra bands in the PCR-positive lanes (3, 5, 9, 18, 19, 27, and 29) in Fig. 5C appear to be artifacts associated with the positivity because it was also shown in the positive control.
With the integration-enhancing ability of B1 d proven, we next went on to analyzing the nature of the B1 d -mediated integration events in developing embryos. For this, microinjection was performed with p␤geo and p(B1 d ) 2 ␤geo vectors, and the eggs were cultured into the blastocyst embryos. Individual embryos were analyzed by a PCR process designed to detect only the homologous recombination event (named Hord-PCR; see Fig. 6). When homologous recombination occurs, the B1 d sequence near the reporter gene turns into an intact B1 sequence by acquiring from cellular B1 sequence the truncated 20-bp sequence with which the primer B1-3-3Ј can anneal. The other primer was designed to anneal with the 5Ј PGK promoter region, and thus, with this primer pair, PCR of the homologous recombination product should yield a 300-bp fragment consisting of a part of the PGK promoter and a part of the B1 sequence from homologous recombinants. The specific PCR product of the homologous recombinants was detected by Southern blot analysis using a 500-bp PGK promoter sequence as the probe. In Fig. 7, the extra bands associated with PCR positivity can be seen in lanes 12, 20, 21, 25, and 27 as well as in the positive control; only the results for 40 out of 75 samples are shown.
Of the 75 blastocyst embryos injected with linearized  Note that the orientation of the repetitive sequence elements in p(B1) 2 ␤geo and p(B2) 2 ␤geo is opposite to that in p(B1/2) 2 ␤geo. B, the result of colony assays for the constructs outlined in A. HC11 cells were transfected with each vector, cultured for about 12 days in the presence of G418, and then stained with Giemsa. Values shown represent the means from two independent experiments. p(B1 d ) 2 ␤geo DNA, about 21% (15/75) was clearly shown to have undergone homologous recombination (Fig. 7). Because there were two identical B1 d sequences on the vector and the Hord-PCR was applied only to a single B1 d sequence, this value might represent only half of the actual recombination frequencies, assuming that both B1 d sequences have a similar capability of recombination. Thus, considering the frequency of the B1 d -mediated foreign DNA integration events as shown in Fig. 5C, it could be reasoned that more than half of the p(B1 d ) 2 ␤geo-injected transgene-positive embryos had the transgenes integrated via the homologous recombinations involving the B1 sequences.
In conclusion, these results demonstrated that repetitive sequences such as B1 and B2 could substantially improve the integration frequency of flanking foreign DNA and that such an increase occurred through an enhancement in the homologous recombination events between the incoming and resident repetitive sequences.

DISCUSSION
In this paper we presented evidence that the B1 and/or B2 repetitive sequences can mediate highly efficient integration of flanking genes into the genome of both differentiated HC11 cells and developing mouse embryos. In an accompanying paper the integration frequency of the SINE-flanked vector has been examined in mouse preimplantation embryos. 2 In that FIG. 6. The Hord-PCR strategy that can detect homologous recombinants. The primers P137 and PB1d can prime the amplification of a 300-bp DNA fragment only when the input DNA has undergone homologous recombination with an intact endogenous B1 sequence. P137 was derived from the 5Ј region of the PGK promoter and thus could prime the DNA synthesis regardless of the occurrence of homologous recombination. PB1d, derived from the 3Ј region of intact B1 sequence and deleted from the p(B1 d ) 2 ␤geo, cannot prime the DNA synthesis unless there has occurred a homologous recombination event between the B1 d sequence and the endogenous B1 sequence. report, the B1/B2-flanked vector was microinjected in a linear or covalently closed circular form, and it showed high ␤-gal expression rates (more than three times than the control) in either conformation. Further, we demonstrated in the present work that a significant proportion of the augmented integration rate is due to the homologous recombination between the endogenous and incoming repetitive sequences.
The ability of human Alu repeats to promote homologous recombination has been investigated in several cases (27)(28)(29)(30). In those reports, the effects of Alu sequences on the homologous recombination events were shown to be controversial; some reported to be efficient and others indistinguishable from entirely random recombination. The inconsistency might, in part, derive from the absence of an assay to detect the homologous recombination events easily and directly. The homologous recombination events were detected by cloning integration sites on the genome followed by clone mapping and sequencing (27) or, less directly, by examining the colonies surviving the G418 and gancyclovir selection (29). The former is a direct but laborintensive and time-consuming method and thus could not examine an enough number of recombination events, whereas the latter could not exclude the possibility of Alu-involved illegitimate recombinations in the doubly selected colonies. Our results demonstrated that rodent SINE sequences, such as B1 and B2, could consistently increase (about three times) the integration rates of SINE-flanked vector in various experimental systems, such as cultured HC11 cells, preimplantation embryos 2 and 11.5-dpc fetuses. Moreover, using the Hord-PCR method, the homologous recombination events could be easily and definitively verified at the molecular level.
It has been suggested that the sequence of Alu repeat diverged too much among the family members or the sequence length of Alu repeat (300 bp) is too short to promote speciesspecific recombination in mammalian cells (28,29). Waldman and Liskay (31) have shown that it is the length of uninterrupted homology that is important for efficient integration but not the overall homology. p(B1 d ) 2 ␤geo, the vector used for the detection of homologous recombination events in our experiment, carries only 110 bp of B1 sequence and yet showed a homologous integration frequency two or three times greater than the control. Moreover, because of the sequence divergence among the family members, the longest stretch showing perfect match with the B1 consensus in the B1 d sequence extends only 18 bp, which is less than the homology length of 25 bp known to be sufficient for recombination (32). Therefore, our result demonstrated that a SINE repeat that is as short as 110 bp in length and carries only an 18-bp stretch of complete sequence homology (Fig. 2) could mediate an efficient integration of flanking foreign DNA into the endogenous B1 loci.
The strong recombinational propensity of the SINE sequences has played a significant role in the evolution of the mammalian genome. In fact, members of the Alu repetitive DNA family have been implicated in many different recombination events and thus can serve as "hot spots" for recombinational events. However, within the life of an organism these recombinational processes may be occurring only rarely because of the high topological constraints exerted on the cellular genome (33). In contrast, the B1 and B2 repetitive sequences placed on free-floating transgenes are free from such constraints and thus may be easily brought into close proximity to genomic cognates by diffusion, resulting in an increased chance of recombination.
The expression frequency of a transgene, in addition to the integration rate, is considered to be one of the most important factors in transgenesis. The repetitive sequences B1 and B2 may help increase the expressivity of the transgene by guiding the integration of flanking genes into a transcriptionally favorable region in the host genome. In fact, as resolved by the chromosomal painting with the B1 or B2 sequence probe, both B1 and B2 sequences are known to be preferentially clustered in the R bands (34). Genes in R bands are known to replicate early in the S phase, and virtually all of the widely expressed housekeeping genes map to these chromosomal regions (35,36). Because the major proportion of the transgene integration mediated by the B1 and/or B2 sequences occurred through the homologous recombination at the corresponding genomic loci according to our results (Fig. 7), it is most likely that the transgenes are located at euchromatic regions and have a high probability of being expressed properly. Therefore, the mammalian germline transformation system involving the SINE sequences has dual benefits; an increased frequency of transgenesis and an improved expressivity of the transgene. In fact, as described in our another report, the p(B1/2) 2 ␤geo-injected blastocyst embryos showed much higher levels of ␤-gal expression compared with the p␤geo-injected control embryos. 2 It is generally accepted that the frequency of random integration into a genome is highly affected by the sequences of transgene ends (26,37). In measuring the integration efficiency of SINE-flanked vectors by the microinjection experiment we excluded such potential bias by linearizing both control and SINE-tagged vectors with the ScaI restriction enzyme, which cuts both plasmids once within the ampicillin-resistant gene leaving the repetitive sequences positioned internally in the vectors. Therefore, the free ends of both linearized vectors should be identical in the nucleotide sequence and the frequency of illegitimate recombination is expected to be similar in both vectors.
Finally, we are currently establishing an experimental pro- FIG. 7. PCR and Southern hybridization analyses of preimplantation embryos microinjected with linearized p␤geo or p(B1 d ) 2 ␤geo vectors. Blastocyst embryos microinjected with p␤geo or p(B1 d ) 2 ␤geo were subjected to PCR analysis. The PCR products were size-fractionated and analyzed by the Southern technique using the 300-bp probe described in the legend to Fig. 6. The hybridization probe was prepared as described in the legend to Fig. 6. The results for 40 out of 75 embryos microinjected with p(B1 d ) 2 ␤geo DNA are shown; those for remaining 35 are not shown. A number of positive signals with varying intensities, representing homologous recombination events, can be seen. The signal in the 22nd sample has a unique pattern, i.e. smaller size and no smearing, and therefore, it may not represent the normal Hord-PCR product. Ps, p(B1) 2 ␤geo DNA as the positive control; N1 and N2, no template DNA; N3 and N5, p(B1 d ) 2 ␤geo; C1-C10, blastocyst embryos injected with the p␤geo control vector. tocol to detect mouse embryos that have transgenes already integrated into their genome at the preimplantation stage by applying the Hord-PCR method to isolated blastomeres; when properly established, this method is expected to allow the preselection of the transgene-positive embryos that have a high propensity of expressing the transgene by virtue of its integration into the B1 or B2 loci. Furthermore, this method may find a highly valuable application in the production of transgenic domestic animals, such as goats, sheep, and cows, where the application of transgenic technology is limited mainly because of their long life cycles and high cost.