A genetic system for direct selection of gene-positive clones during recombinational cloning in yeast

Abstract

Transformation-associated recombination (TAR) is a cloning technique that allows specific chromosomal regions or genes to be isolated directly from genomic DNA without prior construction of a genomic library. This technique involves homologous recombination during spheroplast transformation between genomic DNA and a TAR vector that has 5′ and 3′ gene targeting sequences (hooks). Typically, TAR cloning produces positive YAC recombinants at a frequency of ∼0.5%; the positive clones are identified by PCR or colony hybridization. This paper describes a novel TAR cloning procedure that selects positive clones by positive and negative genetic selection. This system utilizes a TAR vector with two targeting hooks, HIS3 as a positive selectable marker, URA3 as a negative selectable marker and a gene-specific sequence called a loop sequence. The loop sequence lies distal to a targeting hook sequence in the chromosomal target, but proximal to the targeting hook and URA3 in the TAR vector. When this vector recombines with chromosomal DNA at the gene-specific targeting hook, the recombinant YAC product carries two copies of the loop sequence, therefore, the URA3 negative selectable marker becomes mitotically unstable and is lost at high frequency by direct repeat recombination involving the loop sequence. Positive clones are identified by selecting against URA3. This method produces positive YAC recombinants at a frequency of ∼40%. This novel TAR cloning method provides a powerful tool for structural and functional analysis of complex genomes.

INTRODUCTION

Yeast artificial chromosome (YAC) and bacterial artificial chromosome (BAC) cloning systems have greatly facilitated the analysis and understanding of complex genomes (1,2). These techniques make it possible to isolate large DNA fragments, thereby greatly simplifying the physical mapping of chromosomes and genomes. However, the process of isolating a gene or specific chromosomal region of interest is labor intensive, requiring characterization of thousands of YAC or BAC clones and time consuming subcloning procedures. In addition, different regions of the same gene are often on different YACs or BACs, requiring multiple cloning steps to reassemble a copy of the gene. To clone DNA from the genome of a particular individual, a library must be constructed specifically for that purpose and standard YAC or BAC cloning strategies are not suitable for genomic regions in which rearrangements have occurred.

Recently, a recombinational cloning strategy was developed that allows genes and chromosomal regions to be isolated from a complex genome without prior construction of a genomic DNA library (3,4). This technique is carried out in yeast cells, which have a high level of homologous recombination. Development of the technique was stimulated by a pioneer work of David Botstein and colleagues (5), who showed that a double-strand DNA break is efficiently repaired when it is co-transformed into yeast with a linear DNA fragment that includes DNA sequence that is both 5′ and 3′ to the double-strand DNA break. The in vivo homologous recombination pathway that joins together two different DNA fragments sharing homology is now routinely used for construction of recombinant plasmids (6–8).

Ketner and colleagues (3) and Larionov et al. (4) proposed that recombinational cloning could be used to isolate specific chromosomal regions from total genomic DNA. Ketner et al. demonstrated that the 30 kb linear adenovirus genome could be selectively cloned by homologous recombination, even in the presence of a large excess of mouse DNA. They suggested that their recombinational cloning system would be sensitive enough to isolate a gene from mammalian genomic DNA that was present in more than 10 copies per cell. However, this cloning system has a high vector background (clones with no insert DNA) because the vector contains a yeast autonomous replicating sequence (ARS) which allows the vector to cyclize and replicate with high efficiency in yeast. In contrast, transformation-associated recombination (TAR) cloning uses vectors without an ARS element (4) that do not replicate in yeast unless an ARS, or a functional equivalent of an ARS, is acquired by recombination with genomic DNA. ARS sequences are frequently and randomly distributed throughout all eukaryotic genomes (i.e. one ARS per 20–40 kb, on average; 9). Thus, most mammalian chromosomal regions can be isolated by TAR cloning in yeast using an ARS-less vector.

Two general TAR cloning schemes were developed and characterized (10,11). If DNA sequence information is available from the 3′- and 5′-flanking regions of the gene of interest, the gene can be isolated using a vector with two short unique sequences that flank the gene. These ‘targeting hooks’ are cloned into the vector in such a way that the linear form of the vector releases the gene targeting sequences. The hooks can be as small as 60 bp (12). This approach has limitations, because DNA sequence information is often limited in the 3′- or 5′-flanking regions of the desired gene. Another limitation is that regions that lack an ARS-like sequence cannot be isolated by this method. A modified and improved version of TAR cloning, called radial TAR cloning, has also been developed. Radial TAR cloning also uses a vector with two targeting hooks; however, one hook is a unique sequence from the chromosomal region of interest and the other hook is a repeated sequence that occurs frequently and randomly in the genomic DNA (i.e. Alu repeats in human DNA or B1 repeats in mouse DNA). In the radial cloning method, a set of nested overlapping fragments is isolated that extends from the gene-specific targeting hook to different upstream or downstream Alu (B1) positions of the gene of interest. This approach increases the likelihood that a clone will be formed and isolated that includes an ARS-like sequence.

The amount of DNA damage (i.e. double-strand DNA breaks, etc.) in the genomic DNA used in a TAR cloning experiment will determine the size of inserts in the YAC clones. TAR cloning requires physical manipulation of the DNA, which causes some DNA shearing; thus the upper size limit of YACs obtained by TAR cloning is 250 kb. Circular YACs of ≤250 kb can easily be retrofitted into BACs and transferred into Escherichia coli for further characterization. TAR cloning has been used with success to isolate several single copy genes and specific chromosomal regions from human and mouse DNA (10–16; N.Kouprina, unpublished data).

The TAR cloning methods described above allow a gene to be isolated directly from total genomic DNA; however, these methods have a relatively high background. End joining and non-homologous recombination between the vector and genomic DNA generate YAC clones that propagate in yeast even though they do not carry the gene of interest. Typically, TAR cloning produces a set of clones in which the desired gene occurs at a frequency of ∼0.5%. Clones carrying the gene of interest are usually identified by PCR or colony hybridization.

This report describes a modified TAR cloning system with much higher specificity and lower background that utilizes positive and negative genetic selection for clones with the gene of interest. Because the entire gene isolation procedure can be accomplished in ∼10 days, this TAR cloning method is expected to be a powerful tool for structural and functional analysis of complex genomes.

MATERIALS AND METHODS

Yeast strains, transformation and selection of gene-positive clones

The highly transformable Saccharomyces cerevisiae strain VL6-48N (MATα, his3-Δ200, trp1-Δ1, ura3-Δ1, lys2, ade2-101, met14, cir^o), which has deletions of HIS3 and URA3, was used for transformations. This strain was generated from VL6-48 (11) by substitution of the ura3-52 gene by a KanMX cassette (17). Spheroplasts were generated as described previously (18). Agarose plugs (100 µl) containing ∼5 µg high molecular weight DNA were prepared with DNA from MRC-5 normal human fibroblasts (ATCC) or from liver cells of the Tg.AC mouse (12). Linearized TAR cloning vector (1 µg) was added to the DNA-containing plugs, treated with agarase and mixed with yeast spheroplasts. Transformants were selected on synthetic complete medium plates lacking histidine and uracil. To identify gene-positive clones, His⁺Ura⁺ primary transformants were replica plated on synthetic His^– plates containing 5-fluoroorotate (5-FO) to select clones with the unstable URA3 marker (19).

To estimate the density of an ARS sequence(s) in mouse and human genomic DNA fragments, DNAs were isolated from randomly selected clones of large size insert BAC libraries constructed with the pTARBAC1 vector containing a yeast selectable marker (HIS3) and centromere (CEN6) (20). The size of inserts in the libraries varied from 130 to 200 kb.

Construction of TAR cloning vectors

A new TAR vector, pVC604-HP, containing the URA3 negative selectable marker, two targeting sequences (hooks) and a loop sequence, was constructed using the basic TAR cloning vector pVC604 (HIS3-CEN6) (18). Two hooks, 148 bp of the 3′ sequence of the human HPRT gene and 189 bp of the Alu BLUR13 sequence, were PCR amplified from the pVC-HP1 vector (HPRT-CEN6-HIS3-Alu) previously used for TAR cloning of the human HPRT gene (11). A 148 bp HPRT hook [positions 53695–53842 in the genomic sequence (accession no. M26434)] was cloned as a SalI–EcoRI fragment and a 189 bp Alu hook was cloned as a ApaI–XhoI fragment into the pVC604 polylinker. A 1001, 500 or 240 bp HPRT loop sequence flanking the gene-specific hook was PCR amplified from the HPRT YAC (11) and cloned in front of the unique hook as a BamHI–XbaI fragment. The loop sequence lies directly downstream of the hook sequence in the HPRT genomic sequence. This sequence [positions 52694–53694 in the genomic sequence (accession no. M26434)] is not entirely unique. A 379 bp region of the sequence corresponds to a LINE1 transposable element. The URA3 gene was PCR amplified from pRS306 as an ∼1.1 kb EcoRI–BamHI fragment and cloned between the unique hook and the loop sequence. A schematic representation of the pVC604-HP vector is shown in Figure 1.

Figure 1.

Schematic diagram of genetic selection of gene-positive YAC clones. The TAR vector carries a yeast centromere (CEN6), a yeast positive selectable marker (HIS3), two gene-specific targeting hooks (or one gene-specific hook and one common repeat hook) and a negative selectable marker (URA3). The TAR vector also contains a sequence called a loop sequence, which is distal to the gene-specific targeting hook sequence in the targeted chromosomal region and proximal to URA3 and the gene-specific targeting hook in the TAR vector. (In the diagram, only the end of the TAR vector carrying the gene-specific targeting hook is shown.) (A) Homologous recombination between the gene-specific targeting hook and a genomic fragment containing the gene of interest leads to duplication of the loop sequence in the YAC. The URA3 marker is flanked by a direct repeat of the loop sequence, which is mitotically unstable in yeast. Such clones can be easily detected by their ability to grow on media containing 5-FO (selects for Ura^– phenotype). (B) Non-homologous recombination between a hook and a genomic fragment (or non-homologous end joining) forms a YAC with one copy of the loop sequence. In these YACs the URA3 marker is stable and cells with these YACs do not grow on media containing 5-FO.

TAR vector pVC604-Tg, used for cloning of the mouse Tg.AC transgene, was constructed based on the previously described vector pVC604-B1/SV (12). A 160 bp transgene-specific hook and a 130 bp B1 repeat were re-cloned into the basic TAR cloning vector pVC604 as BamHI–XbaI and SacI–SacII fragments, respectively. Different sized loop sequences (211, 500 and 1000 bp of the v-Ha-ras gene sequence) lying distal to a targeting hook sequence in the transgene were PCR amplified from the transgene-containing YAC (16) and cloned in front of the hook as XhoI–EcoRI fragments. The URA3 gene was PCR amplified as a 1.1 kb EcoRI–BamHI fragment and cloned between the unique hook and the loop sequence. The HPRT TAR cloning vector was cut with SalI and transgene vectors were cut with NotI (these sites are located between the hooks) before transformation to yield linear molecules bounded by a gene-specific hook on one end and a common repeated DNA element (i.e. Alu or B1) as a hook on the other end.

PCR analysis

Two pairs of primers were used to characterize HPRT YACs by PCR. IN1R/IN1L amplify a 516 bp sequence of intron 1 and 46L/47R amplify a 575 bp sequence of exon 2 along with flanking introns (11). The results of this PCR reaction indicate which clones formed by recombination between the TAR vector and the 3′-region of the genomic HPRT. The HPRT YAC clones were further characterized using nine pairs of PCR primers that amplify HPRT exons 1–9 (11). A pair of primers, ZG-F and ZG-R, specific to a ζ-globin promoter region was used for PCR screening of transformants for the presence of the Tg.AC transgene sequence (12). These primers generate a 419 bp PCR product that is diagnostic for recombination between the TAR vector and genomic Tg.AC transgene sequences. Yeast genomic DNA was isolated from transformants and PCR amplified as described previously (11,16).

Characterization of YAC clones

Chromosomal size DNA from yeast transformants was separated by transverse alternating field electrophoresis (TAFE), blotted and hybridized with a gene-specific probe as described previously (11,16). The size of circular YACs was estimated by digesting agarose DNA plugs with NotI and TAFE gel analysis. Rescue of YAC ends for sequencing was done using standard protocols.

RESULTS

TAR cloning strategy using negative and positive selection

A schematic diagram of a novel TAR cloning strategy that uses positive and negative selection is shown in Figure 1. Genomic DNA and linearized TAR cloning vector are combined with yeast spheroplasts. The TAR cloning vector contains a yeast centromere (CEN6), a positive selectable marker (HIS3), a negative selectable marker (URA3) and two targeting hooks. (Only one hook is shown in Figure 1; the other hook can be either a unique gene-specific sequence or a common repeat.) The TAR vector also carries an additional gene-specific DNA sequence, called a loop sequence, that is immediately adjacent but distal to the gene-specific targeting hook sequence in the chromosomal DNA. In the TAR vector this loop sequence is proximal to the gene-specific hook and separated from it by URA3, the negative selectable marker. This allows for negative selection against the URA3 gene, which is destabilized in clones with the desired gene, because it is flanked by direct repeats in the YAC clone.

Figure 1 shows the results of two different recombination events involving chromosomal DNA and the TAR cloning vector (for simplicity only one gene-specific hook is shown). In Figure 1A, the TAR vector recombines with the gene-specific targeting hook by homologous recombination; the YAC product then transiently carries (from proximal to distal) the loop sequence, URA3, the gene-specific targeting hook and a second copy of the loop sequence. Direct repeats are extremely unstable in yeast, so there is a high probability of loss of URA3 by spontaneous mitotic recombination involving the direct repeat copies of the loop sequence. After this loop-out recombination event, the YAC carries (from proximal to distal) one copy of the loop sequence followed by distal chromosomal DNA from the gene of interest (Fig. 1A, bottom). If the negative selectable marker is URA3, positive YAC clones carrying the desired gene can be selected by growth in the presence of 5-FO. Other negative selectable markers can be used in place of URA3.

Figure 1B shows the results of non-homologous recombination or non-homologous end joining between a chromosomal DNA fragment and the TAR cloning vector. In this case, the loop sequence is not duplicated, no direct repeat forms and the URA3 marker is mitotically stable (Fig. 1B).

Highly efficient cloning of a multicopy mouse transgene

The TAR cloning strategy described above was used to isolate the v-Ha-ras Tg.AC transgene cassette from mouse DNA. The Tg.AC transgenic mouse carries ∼40 copies of the transgene integrated into a unique site on chromosome 11 (16 and references therein). Each transgene includes the v-Ha-ras gene and a simian virus 40 (SV40) polyadenylation signal and is under control of a ζ-globin promoter. The transgene and flanking genomic sequences were recently isolated by radial TAR cloning using a vector carrying 160 bp of SV40 as a transgene-specific targeting hook and a common mouse repeat B1 as a second hook (12). Transgene-positive clones were obtained at a frequency of ∼2%, at least in part because of the high copy number of the target. In this experiment, the TAR cloning vector was modified to include a loop sequence, namely 211, 310, 500 and 1000 bp of the v-Ha-ras gene sequence that lies distal to the transgene-specific targeting hook in the transgene on chromosome 11. In the TAR cloning vectors the loop sequences and URA3 negative selectable marker were arranged in the configuration described above (Fig. 1).

Cloning experiments with the modified vectors containing different size loop sequences demonstrated that the yield of transgene-positive clones is also ∼2% (Table 1). For a vector with 1000 bp loop sequences six positive YAC clones were identified by screening 250 His⁺Ura⁺ transformants by PCR with primers for the ζ-globin promoter of the transgene. CHEF analysis showed that the YACs were circular and ranged in size from 50 to >200 kb (data not shown), which is typical of clones isolated by radial TAR cloning (12,16). Stability of the URA3 marker in the transgene-positive transformants was determined by replica plating. All six transformants exhibited papillae growth on 5-FO plates, indicating loss of the URA3 marker by ‘looping out’. This recombinational loss of URA3 occurs because the loop sequence is located in both the vector and the insert of the recombinant YAC, forming a direct repeat that flanks the URA3 gene (Fig. 1A). Stability of the URA3 marker was also characterized in the 250 His⁺Ura⁺ transformants used in the above analysis. Nine of 250 colonies exhibited papillae growth in the presence of 5-FO and six of these nine (66%) carried the v-Ha-ras transgene, as determined by PCR. Detection of transgene-positive clones was less efficient with smaller size loop sequences. The fraction of v-Ha-ras-positive YACs gradually decreased with a decrease in size of the loop sequence (Table 1).

Table 1. Isolation of the human HPRT gene and mouse Tg.AC transgene by a new TAR cloning system.

		Size of loop sequence (bp)	Yield of gene-positive clones detected by
			PCR	Genetic screen
Tg.AC transgene
	Experiment 1a	1000	6/250 (2.4%)b	6/9 (66.7%)
	Experiment 2	500	5/250 (2.0%)	3/7 (42.9%)
	Experiment 3	310	4/240 (1.7%)	3/11 (27.2%)
	Experiment 4	211	7/450 (1.6%)	2/17 (11.8%)
HPRT gene
	Experiment 1	1001	7/1210 (0.6%)	7/17 (41.2%)
	Experiment 2	500	2/500 (0.4%)	2/7 (28.6%)
	Experiment 3	240	2/500 (0.4%)	1/15 (6.7%)

^aEach experiment is a summary of two or three independent spheroplast transformations.

^bNumbers in parentheses indicate the percentage of gene-positive clones detected by PCR and genetic assay.

Between 5 and 100 Ura^– ‘pop-out’ events were observed on replica plating of gene-positive colonies (Fig. 2). In contrast, most negative transformants produce one or two Ura^– colonies when replica plated on 5-FO medium. These Ura^– cells resulted from rare mutations in URA3.

Figure 2.

Direct selection of gene-positive clones on 5-FO-containing medium. TAR cloning vector containing a 1000 bp loop sequence was used for isolation of the Tg.AC mouse transgene. Two transgene-positive and 48 transgene-negative transformants were replica plated on 5-FO complete medium lacking histidine. Colonies containing the transgene YACs exhibit papillae growth as a result of ‘pop-out’ of the URA3 marker. Between 5 and 100 Ura^– ‘pop-out’ events were observed on replica plating of gene-positive colonies. ‘Pop-out’ events are explained by generation of an unstable duplication of a loop sequence in the gene-positive YAC clones as predicted from the scheme (see Fig. 1A). Transgene-positive colonies were confirmed by colony hybridization.

This result indicates that this novel TAR cloning method efficiently identifies targeted recombinants carrying the gene of interest using negative selection against URA3. Thus, when used to isolate a multicopy gene, this TAR cloning strategy provides a highly efficient method to genetically select for positive clones.

Highly efficient cloning of a single copy human gene

TAR cloning with negative selection can also be used to isolate a single copy gene from the human genome. The human HPRT gene was recently isolated by radial TAR cloning (11). The TAR cloning vector used to isolate human HPRT carried a 381 bp HPRT-specific 3′ targeting hook and a 189 bp Alu hook. For this radial TAR cloning experiment, ∼0.6% of the clones were HPRT-positive (11,12; unpublished data).

A similar experiment was performed with a modified TAR cloning vector which included a shorter HPRT-specific hook (148 bp) and a 1001 bp HPRT gene fragment (the loop sequence) adjacent to the specific hook sequence in chromosomal DNA. As above, the modified vector also carried the negative selectable marker URA3 (Fig. 1). 1210 random His⁺Ura⁺ transformants from three experiments were selected and replica plated on medium with 5-FO to identify clones with a mitotically unstable URA3. Seventeen colonies exhibited papillae growth in the presence of 5-FO and 7 of these 17 (41%) also carried all the HPRT exon sequences based on a PCR assay (Table 1). The fraction of HPRT YAC clones gradually decreased with a decrease in size of the loop sequence (Table 1). CHEF analysis indicated that the HPRT-positive YACs were circular and ranged in size from 70 to 300 kb (data not shown). When the same 1210 transformants were analyzed by PCR for the presence of HPRT sequences, seven clones were found that included all exon sequences of HPRT. These results indicate that this novel TAR cloning system is highly efficient, highly selective and sufficiently sensitive to isolate a single copy gene from a large and complex mammalian genome.

Density of ARS-like sequences in human and mouse genomes

So far TAR cloning has been used to isolate chromosomal regions that included an ARS or an ARS-like sequence that acts as an origin of replication in yeast. Using random clones with small inserts of mammalian DNA, Stinchomb and co-authors (9) have shown that 20–40 kb carry on average one ARS-like sequence. ARS activity was detected as the ability of the clones to transform yeast cells with high efficiency. This estimate was based on the limited number of clones with genomic DNA inserts. Because the frequency of ARS-like sequences is a critical parameter for gene capture, we re-estimated the density of ARS sequences using a large number of DNA inserts from representative human and mouse BAC libraries. These BAC libraries were constructed with the pTARBAC1 vector containing a yeast selectable marker (HIS3) and centromere (CEN6) (20). The size of inserts in the libraries varied from 130 to 200 kb. BAC DNAs from 100 randomly selected clones from each library were purified from E.coli and transformed into yeast spheroplasts. Between 200 and 1000 transformants are typically obtained during standard spheroplast transformation with 10 ng BAC DNA if the BAC contains an ARS-like sequence (for example with a 150 kb HPRT YAC retrofitted to a BAC), while no transformants are obtained with DNA isolated from a BAC clone lacking an ARS sequence. All 200 BAC clones analyzed exhibited a high transformation efficiency comparable with that for a yeast ARS/CEN vector, indicating that most mammalian chromosomal fragments >130 kb can be cloned using the TAR cloning method. It is worth noting that ∼5% of BACs produced mitotically unstable transformants, suggesting that initiation of DNA replication is not efficient from these ARS-like sequences. Earlier we demonstrated that it does not prevent isolation of functional genes (14).

Molecular mechanism of non-targeted recombination during TAR cloning

Without negative selection TAR cloning produces recombinant YACs that in most cases carry random genomic fragments instead of the desired gene. These background clones may form by non-homologous end joining between the vector ends and chromosomal DNA or by homologous recombination between similar but not identical sequences in the vector and chromosomal DNA. To understand these mechanisms, background YAC clones were characterized from a radial cloning experiment that used a TAR vector with a 60 bp HPRT-specific targeting hook and a 189 bp hook from the 5′-end of Alu. This vector had a similar cloning efficiency as the TAR cloning vector containing a larger sized HPRT targeting hook (11,12) and made it easier to obtain DNA sequence from the insert of the background YACs. The terminal sequences of YAC inserts were rescued as plasmids in E.coli and sequenced using T3 or T7 primers. All of the YAC inserts had an entire Alu sequence at one end, as predicted from homologous recombination between the TAR vector and a chromosomal Alu sequence. The YAC sequences adjacent to the gene-specific targeting hook are summarized in Figure 3. The majority of the clones (38 of 44 YAC analyzed) had the entire hook sequence. The sequences of 25 YAC inserts were found in the draft human genome sequence. These sequences had no detectable homology to the HPRT-specific targeting hook in the targeted chromosomal region. This result strongly suggests that the end of the linear TAR vector was ligated to a random chromosomal fragment by an end-joining reaction. A minor fraction of the clones (6/44) contained a partial HPRT-specific targeting hook that was 6–50 bp long. End sequencing of these clones also showed no homology between the cloned genomic fragments and the HPRT-specific targeting hook. These clones could have formed by a combination of nuclease degradation and non-homologous end joining. In summary, these data indicate that non-homologous end joining is the main mechanism by which background clones are generated during TAR cloning in yeast.

Figure 3.

Molecular analysis of background clones. The ends of 44 randomly selected background YACs (lacking HPRT) obtained during HPRT cloning were rescued as plasmids in E.coli. Terminal sequences of the YAC inserts were determined. Thirty-eight clones (87%) have a non-rearranged 60 bp gene-specific hook sequence; these clones form by non-homologous end joining rather than by homologous recombination. Other clones have a partially deleted gene-specific targeting hook and could form by degradation of the end of the hook followed by non-homologous end joining or by homologous recombination.

DISCUSSION

TAR cloning is a method that facilitates isolation of large fragments of genomic DNA without prior construction of a random genomic DNA library. This method greatly simplifies the physical and functional analysis of mammalian and other complex genomes (10–16), for example by facilitating the study of polymorphic regions, separation of haplotypes and cloning of regions that are difficult to clone in BAC or YAC libraries. With the recent completion of the human genome sequence, it is expected that TAR cloning will become a useful tool for isolating and characterizing human genes and gene regulatory elements.

This study demonstrates that a new genetic approach can be used to directly select a gene-positive clone from primary yeast transformants using positive and negative selection. Enrichment for gene-positive clones with this novel procedure is 50-fold higher compared with the original TAR cloning method. Therefore, the method eliminates the need for labor-intensive screening of hundreds of YAC clones by PCR or colony hybridization.

The modified TAR cloning strategy uses a TAR vector with two new elements: the URA3 negative selectable marker, which can be selected against by growth on 5-FO, and a loop sequence which is located distal to the gene-specific targeting hook sequence in the chromosome fragment, but upstream and proximal to the URA3 gene in the TAR vector. When this TAR vector recombines with a chromosomal fragment containing the gene of interest by homologous recombination, the YAC product then transiently carries (from proximal to distal) the loop sequence, URA3, the gene-specific targeting hook and a second copy of the loop sequence. There is a high probability of loss of URA3 by spontaneous mitotic recombination between the two copies of the loop sequence, so growth on 5-FO can be used to identify the desired recombinants and clone the gene of interest. Positive–negative selection produces a library in which each second clone contains the gene of interest.

In a typical TAR cloning experiment with 10 µg genomic DNA and 2 µg vector DNA, several clones are isolated with the desired gene. Therefore, the technique can be used even when the amount of genomic DNA is a limiting factor (i.e. for clinical studies or to isolate a gene from an obligate parasite that cannot be cultivated outside its host).

A TAR vector must be designed in a specific manner in order for the TAR cloning experiment to successfully isolate the desired gene with high efficiency. The gene-specific targeting hook should be a 60–800 bp unique sequence from the gene of interest (12). A short targeting hook is preferred, because it is easier to identify a short unique sequence and short sequences are easier to synthesize. The loop sequence, distal to the targeting hook sequence in the chromosomal DNA, should be sufficiently long to recombine at a reasonable rate with the proximal loop sequence in the TAR cloning vector. With a loop sequence that was ∼200 bp long, only 8% of clones carrying the gene of interest were revealed by positive–negative selection. So a minimal recommended length for the loop sequence is 1000 bp, which in this study allowed selection of all positive clones. The loop sequence need not be entirely unique. The presence of repeated elements in the sequence does not prevent either cloning or selection of the desired gene. Because the loop sequence is ∼1.5 kb from the end of the linearized vector, it cannot target genomic fragments with the same efficiency as a targeting hook (unpublished data). The presence of a repeat within a loop sequence should also not prevent selection of positive clones because most of the repeated sequences in mammalian genomes are divergent. It is well documented that divergency decreases the level of homologous recombination in yeast (21). For example, HRPT YACs were identified in this work using a 1 kb loop sequence containing a 379 bp LINE1 repeat.

So far TAR cloning has been used to isolate chromosomal regions that include an ARS or an ARS-like sequence that acts as an origin of replication in yeast. ARS-like elements are short (∼50 bp) AT-rich sequences containing a non-conserved 17 bp core consensus (22). Random clones with small inserts of mammalian DNA carry on average one ARS-like sequence in 20–40 kb (9), as detected as the ability to transform yeast cells with high efficiency. The density of ARS sequences was also re-estimated using random clones from two BAC libraries with large inserts in a vector carrying a yeast selectable marker and a yeast centromere (20). These experiments were in agreement with the previous estimate and indicate that most mammalian chromosomal fragments >130 kb can be cloned using the TAR cloning method.

Some regions of mammalian genomic DNA, including heterochromatin blocks and centromeric and telomeric regions, may not contain ARS-like sequences. It may not be possible to use an ARS-containing TAR vector to clone such gene regions, because TAR vectors with ARS sequences cyclize efficiently and have a very high vector background (4,12). Recombinational cloning of these regions will require novel approaches that can select against or prevent vector circularization events. They may include either use of mutants with defects in non-homologous end joining (23) or development of new vectors allowing elimination of the vector background.

In summary, this work presents a novel TAR cloning system that uses positive and negative selection and targeted recombination. This system is a significant advance in recombinational cloning that provides a powerful tool for structural and functional analysis of complex genomes.