Optimum conditions for selective isolation of genes from complex genomes by transformation-associated recombination cloning

Abstract

Transformation-associated recombination (TAR) cloning in yeast is used to isolate a desired chromosomal region or gene from a complex genome without construction of a genomic library. The technique involves homologous recombination during yeast spheroplast transformation between genomic DNA and a TAR vector containing short 5′ and 3′ gene-specific targeting hooks. Efficient gene capture requires a high yield of transformants, and we demonstrate here that the transformant yield increases ∼10-fold when the genomic DNA is sheared to 100–200 kb before being presented to the spheroplasts. Here we determine the most effective concentration of genomic DNA, and also show that the targeted sequences recombine much more efficiently with the vector’s targeting hooks when they are located at the ends of the genomic DNA fragment. We demonstrate that the yield of gene-positive clones increases ∼20-fold after endonuclease digestion of genomic DNA, which caused double strand breaks near the targeted sequences. These findings have led to a greatly improved protocol.

INTRODUCTION

We recently developed a recombinational cloning strategy, named transformation-associated recombination (TAR) cloning, to isolate large genomic fragments (1,2). The method exploits a high level of recombination between homologous DNA sequences during transformation in the yeast Saccharomyces cerevisiae. For isolation, genomic DNA is transfected into yeast spheroplasts along with a TAR vector that contains targeting hooks homologous to the genomic DNA. Recombination between the vector DNA and the genomic DNA fragments results in establishment of yeast artificial chromosomes (YACs). Propagation of TAR-generated YACs in yeast cells depends on acquisition of genomic DNA with ARS-like sequences that can function as origin-of-replication sites in yeast. These short, degenerative, AT-rich sequences are distributed throughout all eukaryotic genomes, averaging one ARS per 20–40 kb (3,4). Thus, most eukaryotic chromosomal regions can be isolated by TAR cloning as YACs, with the insert size ranging from 50 to 250 kb.

When common repeats (such as SINEs and LINEs) are inserted into the vector as targeting hooks, TAR cloning produces genomic libraries in which each insert is flanked with a common repeat. A TAR approach has been applied successfully to the construction of genomic libraries for several organisms, including human (1,2) and mouse (5). The method has also been used to clone human DNA selectively from monochromosomal human–rodent hybrids and radiation hybrids containing chromosome fragments using a vector containing human-specific Alu repeats (1,2,6–8). It is worth noting that for some genomes, cloning in yeast may be the only way to construct large insert libraries. Dictyostelium discoideum DNA, for example, because of its high AT content and inverted repeats, is unclonable in bacterial artificial chromosome (BAC) vectors (9).

In addition, TAR cloning provides a unique opportunity to selectively isolate a chromosomal segment of interest or a gene from complex genomes using vectors with unique targeting hooks developed from 5′ and 3′ sequences flanking the segment. We have been using this method in our laboratory for the past few years and have isolated dozens of human and mouse single-copy genes and specific chromosomal fragments (10–18). Because TAR cloning produces multiple gene isolates, it allows the isolation of both parental alleles of the gene which can then be used for haplotype analysis (19). The technique also provides an opportunity to clone gene homologs and mutant genes from clinical material (20).

Now that the draft sequencing phase of the Human Genome Project (HGP) has been completed, TAR cloning can be applied to other aspects of the project, including closing the gaps and verifying contig assembly. It may enable the isolation and characterization of gap sequences that cannot be propagated as BACs in Escherichia coli cells while they are stable as YACs in yeast cells (21). Another potential application of the method to the HGP is for closing the biggest gaps in the human genome—the centromeric regions. These regions generally have been excluded as HGP targets because of the difficulty of cloning large blocks of repetitive DNAs (22). We recently used TAR cloning to isolate clones of centromeric DNA up to ∼300 kb in length with high selectivity (23).

TAR cloning has become a routine and valuable procedure in our laboratory. Our continued investigation of conditions that maximize the efficiency and selectivity of gene capture, and their description here, will make the procedure available to the rest of the scientific community. Specifically, we determined the smallest homologous region required for gene isolation (4), and we developed a system allowing selection of gene-positive clones genetically (24). To generalize the technique, we developed a novel TAR cloning strategy that can isolate genomic regions regardless of the presence of ARS-like sequences (V.N.Noskov, N.Kouprina, S.H.Leem, I.Ouspenski, J.C.Barrett and V.Larionov, manuscript submitted); that approach allows isolation of genomic regions not only from eukaryotes, but also from prokaryotes.

Still, not all TAR cloning conditions are optimized, and that makes it difficult for other laboratories to adopt the technique for routine use. In the present study, we investigated various aspects of the technique and described what works best for us. We found the quantity and size of genomic DNA that yields the greatest number of transformants, and we demonstrated that double strand breaks (DSBs) adjacent to the targeted sequence in genomic DNA increase the yield of gene-positive clones dramatically.

MATERIALS AND METHODS

Yeast strains and transformation

The highly transformable S.cerevisiae strain VL6-48N (MATα, his3-Δ1, trp1-Δ1, ura3-Δ1, lys2, ade2-101, met14 cir^o), which has HIS3 and URA3 deletions, was used for the transformation experiments (20). Spheroplasts were prepared as described previously (24). Cell density of the culture of yeast used to prepare spheroplasts is critical, and ∼2 × 10⁷ cells/ml is optimal in our laboratory. We count cells with a hemacytometer rather than estimate their number by optical density because the OD₆₆₀ of a culture varies from ∼1.2 to 4.5, depending on the type of spectrophotometer. Yeast transformants were selected on synthetic complete medium plates lacking histidine (16). When an ARS-containing TAR vector was used for cloning, transformants were selected on synthetic his^– plates containing 1 mg/ml of 5-fluoro-orotate (5-FOA) to select against clones with a recircularized vector. 5-FOA was not included in the top agar, however, because of its toxic effect on yeast spheroplast regeneration.

TAR cloning vectors

Two centromere-based yeast TAR vectors with common mouse and human repeats, pVC-B1-B2 (B1-HIS3-CEN6-B2) (5) and pNKBAC39 (Alu-HIS3-CEN6-Alu) (7), were used for cloning genomic fragments. For isolation of human HPRT, the TAR vector pHPRT-AatII, containing two unique hooks, was constructed. The 163 bp 5′ and 194 bp 3′ targeting sequences were designed from the genomic sequence of HPRT near AatII restriction sites (positions 1274–1437 and 40 066–40 259 in the HPRT genomic sequence; accession number M26434, NCBI). The 5′ hook is 45 bp downstream from the proximal AatII site in genomic DNA; the 3′ hook includes a distal AatII recognition site. The targeting sequences were PCR-amplified from HPRT YAC DNA (12) with the following two pairs of primers: 5HP-F and 5HP-R, 5′-ATGCGGGCCCggtttacgg ccgccatgaagc-3′ plus 5′-ATGCATGCGTCGACgtttgcaggctc actaggtag-3′; and 3RT-F and 3RT-R, 5′-ATGCATGCGTC GACgtaggatatgcccttgactat-3′ plus 5′-ATGCGGATCCgacgtc tgtactagactacag-3′ (lower case letters denote sequences homologous to HPRT). PCR products were cloned into a polylinker of the basic TAR cloning vector pVC604 (16) as ApaI–SalI and SalI–BamHI fragments. Before being used for transformation, each TAR cloning vector was linearized with SalI to release its targeting hooks.

The ARS-containing TAR vector pARS-Alu90 (Alu-pADH1-CEN6-ARSH4-HIS3-URA3-Alu) with two Alu targeting sequences was constructed using the Bluescript-based yeast–E.coli shuttle vector pRS313 (25). The cassette containing the URA3 gene driven by the Schizosaccharomyces pombe ADH1 promoter was derived from the pBL94 plasmid (26). In this vector, the TATA box and the URA3 initiation transcription site are separated with two 45 bp Alu targeting hooks. The XhoI–ClaI sites in the ADH1 promoter were used to clone the Alu targeting hooks, which were PCR-amplified from the 3′ end of the 290 bp Alu consensus sequence (27) with the alu45-F and alu45-R primer pairs, 5′-ATGCA TCGATgtcccagctacttgggaggc-3′ and 5′-ATGCATGCCT CGAGggagtgcagtggcgggatct-3′. The SmaI site that separates the Alu targeting hooks was used for linearization of the pARS-Alu90 vector before transformation. Insertion of the hooks does not prevent URA3 expression because the ADH1 promoter tolerates inclusion of up to a 125 bp sequence (26). A greater distance between the TATA box and the initiation transcription site inactivates URA3 expression. This configuration allows selection of TAR cloning events against vector recircularization.

Preparation of genomic chromosome-size DNA for TAR cloning experiments

Agarose plugs (60 µl) containing different amounts of genomic DNA were prepared, as previously described (16), from human fibroblast MRC-5 cells and mouse A9 cells (American Type Culture Collection). Prior to use, the DNA-containing plugs were melted and treated with agarase. High molecular weight human DNA was prepared in aqueous solutions as well, following the protocol developed for cloning of DNA in phage artificial chromosome (PAC) vectors (28). For some experiments, human genomic DNA was prepared with a Blood and Cell Culture DNA Maxi Kit (Qiagen). Genomic DNA size was reduced by careful pipetting of the DNA solution up and down several times in a 200 µl Pipetman tip. The size of the sheared DNA was determined by clamped homogeneous electrical field (CHEF) gel electrophoresis. Commercial female genomic DNA (catalog number G1521; Promega) and human genomic DNA (catalog number G3041; Promega) were used for cloning the human HPRT gene.

Analysis of transformants

Two pairs of primers were used to identify transformants containing the human HPRT sequence. 46L plus 47R amplified a 575 bp sequence of exon 2, along with the flanking introns (24). Yeast genomic DNA was isolated from the transformants and PCR amplified as described previously (16). To determine the size of the YAC inserts, the agarose plugs with yeast genomic DNA were melted at 70°C prior to CHEF analysis. This treatment linearized ∼20% of the YAC molecules in the 100–400 kb size range, and then Southern blot analysis of YAC DNA was performed with a diagnostic Alu probe (2).

RESULTS

The yield of transformants depends on the size and concentration of genomic DNA

In our original TAR cloning protocol, we prepared genomic DNA in agarose plugs to prevent it from shearing (16). Typically, ∼90% of genomic DNA consists of fragments >1 Mb. Because large molecules may penetrate into yeast spheroplasts less efficiently than small ones, we investigated how the size of genomic DNA affects transformant yield. For this purpose, we chose pNKBAC39, an Alu-containing TAR vector that has multiple potential targeting sites in the human genome. With 1 µg of vector, 2 µg of unsheared genomic DNA prepared in agarose plugs, and 5 × 10⁸ spheroplasts, the yield of transformants was approximately 400 colonies per plate. The agarose plugs were melted and the DNA was sheared to different extents by pipetting. Sheared DNA was transformed into yeast spheroplasts along with the linearized vector. The effect of DNA shearing on transformation frequency is summarized in Table 1. As can be seen, the yield of transformants increased dramatically and inversely with the size of the genomic DNA. With fragments between 50 and 150 kb, the transformant yield was approximately 16 times higher than with non-sheared DNA. Further decrease of genomic DNA size reduced the yield of transformants only slightly either due to the accumulation of small fragments that could not be recovered as YACs because they did not contain ARS-like elements (on average, there is one ARS per 20–40 kb) or due to the accumulation of fragments which do not have appropriately positioned Alu sequences. Thus, we concluded that the yield of transformants increases with a decrease in the size of genomic DNA fragments, possibly due to facilitating their penetration into yeast spheroplasts.

Table 1. Effect of shearing of genomic DNA on number of transformants obtained.

No. of pipettings	DNA size	No. of transformants per platea	Increase of transformation efficiency
0	>1.0 Mb	420	×1
3	350 kb (400–50 kb)b	1960	×4.7
6	300 kb (350–50 kb)	3360	×8.0
9	250 kb (300–50 kb)	5680	×13.5
12	200 kb (250–50 kb)	6420	×15.3
15	150 kb (200–50 kb)	6280	×15.0
18	100 kb (150–50 kb)	6820	×16.2
21	50 kb (100–50 kb)	6440	×15.3

^aAverage number of transformants. Ten independent transformation experiments were carried out.

^bPipetting leads to a linear decrease of size of genomic DNA and accumulation of ∼50 kb fragments. In the range of DNA size, the first number corresponds to the maximal size of DNA molecules after pipetting.

Unexpectedly, the sizes of the DNA inserts in the YACs obtained with untreated and moderately sheared DNA were approximately the same and vary from 50 to 200 kb (Fig. 1). That result suggests that when untreated plugs are used for transformation, the fraction of spontaneously sheared DNA in the plugs (3–5%) penetrates the spheroplasts preferentially.

Figure 1.

Physical characterization of YACs generated by an Alu-containing TAR cloning vector from human genomic DNA sheared to different size fragments. The pNKBAC39 vector containing Alu targeting hooks was used to generate mini-YAC libraries from high molecular weight genomic DNA (A), and DNA sheared to fragments of 50–150 kb (B), 50–200 kb (C) and 50–250 kb (D). Chromosome size DNA was isolated from yeast transformants, separated by CHEF gel electrophoresis, and blot-hybridized with an Alu probe for visualization of the linearized YAC DNA molecules. The molecular weight of each band was estimated by comparing it with the high molecular weight marker. Two hundred YACs were analyzed.

Genomic DNA prepared in aqueous solutions is roughly between 300 and 500 kb long, which is optimal for TAR cloning of fragments >200 kb. When we repeated our TAR cloning experiments with genomic DNA isolated from human fibroblasts using a method developed for constructing PAC libraries (28), the frequency of transformation was reproducibly five times higher when the DNA was prepared in aqueous solutions than when it was prepared in agarose plugs and sheared to the same size (data not shown). In a reconstruction experiment, we showed that agarose fragments inhibit spheroplast transformation.

In our previous TAR cloning experiments, the concentration of genomic DNA varied from 1 to 20 µg per µg of vector, and approximately 5 × 10⁸ spheroplasts. Because the concentration of genomic DNA is an important variable, we examined its effect on transformant yield. Different amounts of human (or mouse) genomic DNA were presented to yeast spheroplasts along with linearized Alu-containing (or B1-B2-containing) vectors. Figure 2 summarizes the results of several such experiments. As can be seen, the yield of transformants increased curvilinearly with the amount of genomic DNA up to 2.0 µg, and then decreased. This may be due to the inhibition of transformation by DNA clumps formed during the precipitation of high molecular weight DNA with polyethylene glycol. Thus, the optimal amount of genomic DNA in the transformation mix was ∼2 µg.

Figure 2.

Effect of DNA concentration on yield of transformants during TAR cloning. Different amounts of human or mouse genomic DNA prepared in aqueous solutions with size fragments roughly between 50 and 200 kb were presented to yeast spheroplasts together with 1.0 µg of an Alu-containing TAR vector (pNKBAC39 or pVC-B1-B2) containing B1 and B2 mouse-specific repeats. Open and black circles represent experiments with human and mouse DNA, respectively. The y-axis represents the number of colonies per experiment.

Thus, to achieve the highest yield of transformants during TAR cloning, we recommend that the genomic DNA be prepared in aqueous solution in a concentration of 2 µg per µg of vector and 5 × 10⁸ spheroplasts, and that its size be reduced, if necessary.

The TAR vector recombines preferentially with the ends of the targeted genomic fragment

The sizes of DNA inserts in YAC libraries generated by Alu-containing vectors vary from 50 to 250 kb (7,8). Analysis of more than 400 clones revealed only two YACs with inserts <50 kb. One explanation for this is that the density of ARS-like sequences in human DNA (one per 20–40 kb) prevents the rescue of YACs with small inserts. Another explanation is that not all Alus in the targeted fragment are recognized equally by the targeting Alus in the vector. It is possible that the Alus nearest the ends of the targeted fragment are preferential targets for recombination. To elucidate the mechanism of recombination between vector and genomic DNA, we cloned human DNA with a TAR vector containing a yeast origin of replication site—a technique that renders the human DNA capable of replicating in yeast cells regardless of whether it contains its own ARS-like elements (V.N.Noskov et al., manuscript submitted). pARS-Alu, the vector constructed for the purpose, had the following structure: Alu-pADH1-CEN6-ARSH4-HIS3-URA3-Alu, where ‘Alu’ represents two 45 bp sequences from an Alu repeat. The vector was linearized between the targeting hooks and presented with the human genomic DNA into yeast spheroplasts. To eliminate the background due to vector re-circularization, transformants were selected for on synthetic selective medium containing 5-FOA (see Materials and Methods). Per 1 µg of vector and 2 µg of genomic DNA, the average yield of His⁺Ura^– transformants was 1000. Roughly 95% of the clones contained recombinant YACs with human DNA inserts. Among the 105 YAC clones characterized by CHEF gel electrophoresis, inserts varied from 50 to 200 kb, i.e. they did not differ in size from the inserts obtained with a vector lacking an ARS sequence (Table 2). Because Alus are present in genomic DNA roughly once per 4 kb, based on this result, we may suppose that Alu sequences at the ends of the targeted genomic DNA fragment recombined much more efficiently than internal Alu sequences with homologous hooks in the vector.

Table 2. Analysis of YACs obtained with ARS-containing and ARS-lacking TAR vectors.

Plasmid	Size of human DNA inserts (kb)	No. of YACs
pNKBAC39 (ARS–)a	<50	20 (11%)
	50–100	100 (59%)
	100–150	60 (19%)
	>150	20 (11%)
		Total 200
pARS-Alu90 (ARS+)	<50	17 (16%)
	50–100	60 (57%)
	100–150	18 (17%)
	>150	10 (10%)
		Total 105

^aBoth plasmids contain Alu-targeting sequences as hooks. pNKBAC39 does not contain a yeast origin of replication. pARS-Alu90 plasmid contains an origin of replication.

Specific double strand breaks at the targeted genomic region increases the yield of gene-positive clones

During single-copy gene isolation, TAR cloning usually produces positive YACs with the desired insert at a frequency of ∼0.5% (16). The gene-specific fragment is represented in the transformation mix by a population of overlapping DNA molecules due to the random shearing of genomic DNA during its isolation. Thus, the targeted sequences are positioned differently in relation to the ends of fragments. Because Alu-containing vector recombined much more efficiently with Alu targeted sequences located at the ends than internal Alu homologous sequences, we would expect an increase in gene capture efficiency if gene-specific targeted sequences are located at the end of the genomic fragment.

To investigate that possibility, we introduced DSBs into the human HPRT region by AatII digestion of genomic DNA, which produces a 39 kb HPRT fragment. The vector pHPRT-AatII, constructed for the experiment, contained two unique targeting sequences homologous to the HPRT genomic regions near the AatII sites (Fig. 3A). When the AatII-digested genomic DNA and the vector were transfected into yeast spheroplasts, the yield of transformants varied from 2 to 10 colonies per µg of vector and 2 µg of genomic DNA. Such a low yield of transformants is characteristic for TAR vectors with two unique hooks. Among 115 transformants analyzed, 19 (17%) contained the HPRT sequence. All gene-positive clones contained circular YACs with 39 kb DNA inserts (data not shown). In a control experiment with untreated genomic DNA, the yield of HPRT-positive clones was <1% (Fig. 3B). To exclude the possibility that the increased yield of gene-positive clones was simply the result of the decreased size of genomic DNA following from AatII digestion, we repeated the experiments with genomic DNA cut with MluI. MluI digestion produced an ∼98 kb HPRT-containing fragment with the targeted sequences located ∼11 and 48 kb from the 5′ and 3′ ends of the fragment, respectively (Fig. 3A). No increase of HPRT-positive clones was observed with MluI-treated DNA (Fig. 3B). Thus, the highly selective cloning of HPRT from the AatII-digested genomic DNA was the result of the enrichment of HPRT genomic fragments with targeted sequences at their ends.

Figure 3.

Effect of DSBs in the targeted genomic region on yield of gene-positive clones. (A) DSBs were introduced into the human HPRT genomic region by AatII or MluI digestion. AatII digestion produced a 39 kb HPRT fragment with targeted sequences at both ends of the fragment. MluI digestion produced a 98 kb HPRT-containing fragment with targeted sequences located 11 and 48 kb from the 5′ and 3′ ends of the fragment, respectively. The TAR vector pVC-HP39 contains two unique targeting sequences that are homologous to HPRT genomic regions adjacent to the AatII sites. (B) Yield of gene-positive clones when specific DSBs are introduced into the HPRT genomic region.

DISCUSSION

TAR cloning could become a routine procedure in any laboratory, with adherance to a few guidelines. Hooks should be unique sequences containing no repeated sequences. They can be as small as 60 bp (greater length does not increase the selectivity of gene isolation) (4), and should be free of yeast ARS-like sequences. Also important to TAR cloning is high efficiency in yeast transformation. We routinely prepare spheroplasts using yeast strain VL6-48, which exhibits a uniquely high transformation efficiency.

In the present study, we investigated additional factors that could affect TAR cloning efficiency. We showed that transformant yield increased curvilinearly with increasing amounts of genomic DNA in the transformation mixture, up to 2 µg. Further increases inhibited transformation efficiency, possibly due to formation of DNA clumps during precipitation of high molecular weight DNA with polyethylene glycol. Thus, under standard conditions of TAR cloning (1 µg of vector DNA and 5 × 10⁸ spheroplasts), the optimal concentration of genomic DNA is 2 µg.

Our results also demonstrated that the yield of transformants depended on the size of the genomic DNA in the transformation mix. The yield was approximately one order of magnitude greater with genomic DNA sheared to 100–250 kb than with Mb size genomic DNA. The smaller size range allows isolation of the majority of mammalian genes. Regardless of size, the yield of transformants was about five times higher when the DNA was prepared in aqueous solution than when it was prepared in agarose plugs. The reason is that agarose fragments in the melted plug inhibit yeast transformation. It is worth noting that aqueous solutions of human and mouse DNA between 50 and 150 kb are commercially available. We successfully used commercial human genomic DNA (Promega) for isolation of the 70 kb hTERT gene (18) and three ∼80 kb genomic regions from human chromosome 5 (21). Isolation of genes and chromosomal segments >250 kb requires the use of pre-selected genomic DNA to remove small molecules, which penetrate yeast cells preferentially. TAR isolation of large chromosomal regions may also be achieved by searching for new host strains that do not exhibit a strong dependence on the size of transforming DNA.

The study also sheds light on the mechanism of TAR cloning. Based on our results with ARS-containing vectors, targeted homologous sequences at the ends of the genomic DNA fragment recombined much more efficiently than internal sequences. This observation is critical to improving TAR cloning protocols. The yield of gene-positive clones increased approximately 20 times when specific DSBs were introduced into the genomic DNA close to the targeted sequences. The question remains of how far a DSB can be from the targeted sequence and still induce a recombination interaction with a vector. Until that is clarified, we recommend choosing targeting sequences as close as possible to the endonuclease recognition site(s) that will be used for digestion of genomic DNA before cloning. The HGP’s recently released draft sequences of genomes simplify the choice of specific targeting sequences. Alternatively, specific DSBs can be introduced into genomic DNA using RecA-assisted restriction endonuclease cleavage (29).