Recovery and potential utility of YACs as circular YACs/BACs

Abstract

A method has been established to convert pYAC4-based linear yeast artificial chromosomes (YACs) into circular chromosomes that can also be propagated in Escherichia coli cells as bacterial artificial chromosomes (BACs). The circularization is based on use of a vector that contains a yeast dominant selectable marker (G418R), a BAC cassette and short targeting sequences adjacent to the edges of the insert in the pYAC4 vector. When it is introduced into yeast, the vector recombines with the YAC target sequences to form a circular molecule, retaining the insert but discarding most of the sequences of the YAC telomeric arms. YACs up to 670 kb can be efficiently circularized using this vector. Re-isolation of megabase-size YAC inserts as a set of overlapping circular YAC/BACs, based on the use of an Alu-containing targeting vector, is also described. We have shown that circular DNA molecules up to 250 kb can be efficiently and accurately transferred into E.coli cells by electroporation. Larger circular DNAs cannot be moved into bacterial cells, but can be purified away from linear yeast chromosomes. We propose that the described system for generation of circular YAC derivatives can facilitate sequencing as well as functional analysis of genomic regions.

INTRODUCTION

Current physical maps based on yeast artificial chromosomes cover most of the human genome as well as much of the genomes of a variety of organisms, for example, fugu fish, zebra fish, bovine, dog and Arabidopsis thaliana. The DNA inserts in YACs vary up to a megabase or more in length, sufficient to contain even very large genes in intact form for functional studies. However, the manipulation of YACs, even in successful instances (1–3), is cumbersome, and purification of YAC DNA is tedious and often incomplete.

In order to help mobilize the DNA inserts in YACs for a variety of studies, we have designed a vector that circularizes resident YACs in strain AB1380 (4) and permits their transformation into Escherichia coli cells as circular YAC/BACs. Even large circularized DNA molecules are less subject to shearing forces during lysis, facilitating handling and purification.

The method capitalizes on the capacity of yeast to repair a double-strand break in a plasmid with a transformed linear fragment containing DNA sequences that flank the break (5,6). This capacity was used to advantage to recover complete yeast genes from lambda clones using a linearized plasmid that contained phage lambda sequences flanking the cloned yeast DNA (7). In another application, Bradshaw et al. (8) used an F factor-based yeast-bacteria shuttle vector with target sequences that specifically rescued Hoxc and Hoxb genes from the YACs containing these genes.

We have aimed at a more general method to recover any YAC in a circularized pure form. A battery of available YAC libraries exists in the vector pYAC4. Utilizing the ability of the yeast to rescue double-strand breaks, we designed and constructed a targeting vector that circularizes pYAC4-based YACs to a form that is a circular artificial chromosome both in yeast (YAC) and bacteria (BAC). The circularizing vector, pNK-G418, is derived from pNKBAC39 (9), which contains an F′ origin of replication, yeast centromeric sequence (CEN), and the chloramphenicol resistance gene (CM^r) for selection in bacteria. The pNKBAC39 vector was modified to include sequences flanking the EcoRI cloning site in the pYAC4 vector as recombination targets, along with a KanMx4 module (10) for selection with G418 in yeast. We report here that this vector efficiently circularizes YACs with inserts of human DNA up to 670 kb. We also show evidence that circular molecules of up to 250 kb can maintain their integrity during circularization and subsequent transfer into bacteria as BACs, though problems associated with very large insert sizes have been encountered and are discussed. The methods are simple to use, and open up the possibility of the use of circular YACs for structural and functional studies.

MATERIALS AND METHODS

Yeast strains and media

All YACs described in this study were cloned in the Saccharomyces cerevisiae strain AB1380 (MATa ura3-52 trp1 lys2-1 ade2-1 can1-100 his5) (4). The yeast strain YNG96-2 [MATα leu2–Δ1 trp1–Δ2 his3–Δ1::ura3–X/GAL/RAD52/SUP11 ura3–Δ1 lys2–Δ1::kanMX imp⁺ (GAL2) cyh2^R CAN1^R kar1–Δ15 rad52–Δ2] (9) which has the HIS3 gene deleted, was used for YAC kar1-transfer experiments. Yeast cells were grown on complete medium (YPD or YPGal), YPD supplemented with 250 mg/l G418-sulfate (Life Technologies, Rockville, MD) or synthetic standard selective medium without either uracil (Ura^–) or tryptophan (Trp^–) or histidine (His^–) were prepared as described (11).

YAC/BAC retrofitting vectors

To retrofit linear YACs into circular YAC/BACs, three different vectors were utilized, pNKBAC39, pNKBACMX4 and pNK-G418. In addition to yeast selectable marker(s) (HIS3 or HIS3 and G418) and a yeast centromeric sequence (CEN6) all vectors contain an ∼6.0 kb BAC cassette [with the F′ origin of replication, and the chloramphenicol resistance gene (CM^r) for selection in bacteria].

The pNKBAC39 (pBelobac11-CEN6-HIS3), containing an inverted repeat of Alu sequences, is a transformation-associated-recombination (TAR) cloning vector developed for selective cloning of human DNA from human/rodent hybrid cell lines (9). The pNKBACMX4 vector was constructed by cloning a G418^R gene into the unique SacII site of pNKBAC39. This gene was PCR-amplified from the KanMX4 module (10) as a 1460 bp fragment using oligonucleotides 5 and 6 (see Table 1; the positions of these primers in the KanMX4 module are from –386 to –362 and from 1026 to 1050, respectively). The amplified fragment was gel purified, digested with SacII, and cloned into the unique site SacII of the pNKBAC39 vector. Alu targeting sequences in the vectors are at the termini after linearization of the vectors by SalI.

Table 1. Oligonucleotide sequences for circularizing vectors.

Oligo	Type	Sequence
1	5′ left target	GGATCCGGATCCCATGTTTGACAGCTTATCATCGATAAG
		(BamHI linker, underlined)
2	3′ left target	GAATTCCGTAGTGATTAATTAAAGTCTTGCGCC
		(PacI, bold and underlined)
3	3′ right target	TCTCGGTAGCCAAGTTGGTTTAAGG
4	5′ right target	GGATCCGGATCCCATAAATCGCCGTGACGATCAGCGGTC
		(BamHI linker, underlined)
5	5′ G418	CCGCGGCCGCGGGATATCAAGCTTGCCTCGTCCCCGC
		(SacII linker, underlined)
6	3′ G418	CCGCGGCCGCGGGTCGACACTGGATGGCGGCGTTAGT
		(SacII linker, underlined)
7	BAC3	GCGCGTTGGCCGATTCATTAATG
8	BAC4	GCCGCTCTAGAACTAGTGGATCC
9	LS2	TCTCGGTAGCCAAGTTGGTTTAAGG
10	Primer R	GCAAGTCTGGGAACTGAATGG

The circularizing vector, pNK-G418, was constructed from pNKBACMX4 by replacement of Alu targeting sequences with sequences specific for the pYAC4 vector. PCR primers were designed to amplify target fragments from the left arm (TRP1/ARS/CEN4 arm, using primers 1 and 2 in Table 1) and right arm (URA3 arm, using primers 3 and 4 in Table 1) from pYAC4 (GenBank accession no. U01086) between coordinates 5/559 and 504/1455, respectively. Sequences for the left arm specific primer were altered to create a unique PacI restriction site by changing a ‘T’ to an ‘A’ at base position 544 (12). The amplified sequences from the arms overlap by 57 bp. The fragments were mixed in equimolar ratio and primers 1 and 3 were used in a second PCR reaction to amplify the complete target sequence, 1451 bp in length, excluding the linker sequences, as a single fragment with the unique PacI site. It was then cloned into the BamHI site of pBluescript SK vector. The resulting clones with the insert were tested for the presence of the PacI site, and the presence of the target fragment sequence was verified by fluorescent-labeled dideoxy terminator sequencing. The target fragment was excised with BamHI endonuclease digestion and cloned into pNKBACMX digested with BamHI. BamHI treatment removes most Alu sequences from a targeting vector. (Only 52 bp of BLUR13 R1 Alu sequence remains in the vector after this treatment.) Linearization of the vector obtained by PacI leads to terminal exposure of the left and the right arm pYAC4 targeting sequences with sizes of 539 and 912 bp, respectively.

YAC transfer to new yeast host strains

YAC transfer from AB1380 to a new host strain, YNG96-2 (13), for YAC circularization, using the HIS3 gene in a vector as a selectable marker, employed a modified kar1 mating protocol as previously published (14). The modification includes two steps of selection against diploid cells, using both the recessive cycloheximide cyh2^R and the canavanine resistant can1^R markers; the double selection greatly reduces the level of false-positive clones (9). Because YNG96-2 contains the RAD52 gene under the control of GAL1/GAL10 promoter, YAC modification was carried out with cells growing on galactose-containing medium when RAD52 is functional. Clones with retrofitted YACs were maintained on glucose-containing medium to reduce risk of YAC rearrangements by recombination.

Circularization of YACs

AB1380 strains carrying YACs were transformed with a linearized vector using a modified lithium acetate protocol (15). Briefly, the yeast cells were grown overnight in AHC medium (16) instead of YPD. The following day, cells were diluted into 50 ml of AHC to give an OD₆₀₀ of 0.5. The cells were grown to an OD₆₀₀ of 1.0. For transformation 1 µg linearized vector was added to ∼5 × 10^–8 lithium acetate-treated cells at room temperature. After a heat shock at 42° C for 20 min, cells were incubated at room temperature for 30 min. The cells were washed with water and incubated for 3–4 h at 30°C in 400 µl of YPD (non-selective medium). The cells were washed once more with water, plated onto YPD agar plates supplemented with 250 mg/l G418, and incubated at 30°C for 3–4 days. Transformants were replica plated onto fresh YPD/G418 plates and incubated at 30°C, or until primary isolates could be picked (2–3 days).

Primary isolates from the above plates were picked and streaked on AHC agar plates supplemented with tryptophan and uracil to give individual colonies. Single cell isolates were replica plated on a set of plates with otherwise complete medium lacking either tryptophan or uracil. Colonies with the phenotype predicted for YAC circularization (i.e. Ura^– Trp^– for pNK-G418 and pNKBACMX4) were further analyzed for sequence-tagged site (STS) content, size and fingerprint pattern. Ends of retrofitted YAC/BAC clones were sequenced using either Primer R and LS2 or BAC3 and BAC4 primers (Table 1, primers 7–10).

YACs transferred into the kar1-deficient strain, YNG96-2, were circularized using the HIS3 selectable marker in the vectors. Transformation of the strain and purification and analysis of primary transformants was done as described for transformation using G418 selection.

STSs used to verify the integrity of circularized YACs and following their transfer into E.coli included for (a) sWXD1229, sWXD633, sWXD178, sWXD636, sWXD634, sWXD635, sWXD733, sWXD1289, sWXD504 and sWXD754 for YAC yWXD1851, (b) sWXD207, sWXD513, sWXD682 and sWXD1520 for YAC1321, and (c) sWXD2342, sWXD1334, sWXD1916, sWXD1319, sWXD2457, sWXD27, sWXD385, sWXD415, sWXD791 and sWXD1151 for YAC yWXD931. The assay conditions, the GenBank accession numbers and the primer pair sequences can be obtained from http://www.ibc.wustl.edu/cgmdata/public/stsseq.html

Pulsed-field gel electrophoresis (PFGE)

Total yeast DNA was prepared in agarose plugs as described (17) and then subjected to PFGE in a Bio-Rad PFGE apparatus for 20 h at 14°C with switching beginning at 20/30 s and ramping to 40/60 s. The DNA was transferred to nylon membranes and Southern blot hybridizations were carried out with [³²P]dCTP-labeled total human DNA or DNA fragment coding for chloramphenicol acetyl transferase gene (Amersham random priming kit, Piscataway, NJ) (18). Hybridized membranes were washed, exposed to Kodak XOmat AR film and developed.

Preparation of circular YACs in agarose plugs for electroporation into E.coli

The original method for preparation of agarose plugs containing genomic DNA (17) can be routinely used for electrophoretic analyses, but this method is inefficient when retrofitted YACs are to be transferred into E.coli. Circular YAC/BACs >100 kb prepared by standard methods exhibit very low transformation efficiencies during electroporation. We have found that the transformation efficiency can be increased by more complete spheroplasting and shortening the incubation of agarose plugs in NDS (0.5 M EDTA, 0.01 M Tris–HCl pH 7.5, 1% N-Lauroyl Sarcosine pH 9.5, 5 mg/ml proteinase K) solution at 50°C. The modified protocol includes (i) short (30 min) yeast spheroplasting in EDTA mix solution (0.05 M EDTA, 0.01 M Tris–HCl pH 7.5) at 37°C versus spheroplasting in agarose plugs overnight using a 2-fold higher concentration of ICN Zymolase 20T (i.e. 30 U/ml), followed by embedding the spheroplasts in molten agarose, and (ii) short (1 h versus overnight) incubation with proteinase K in NDS solution at 50°C. Plugs prepared by this method are washed 3–4 times with EDTA mix (20 min each time at room temperature) before storage at 5°C. They are then dialyzed against water at room temperature before they are melted for electrophoration.

Yeast/bacterial DNA preparations for fingerprinting

High molecular weight yeast DNA was prepared as follows. Yeast cells were grown in 50 ml of media (AHC or YPD-G418 supplemented with 250 mg/l) to saturation (2–3 days at 30°C with shaking). The cells were centrifuged at 1100 g for 10 min, washed once with 50 mM EDTA, and the pellet resuspended in 2 ml SCE pH 7.0 (1 M sorbitol, 100 mM trisodium citrate, 50 mM EDTA) containing 2 mg/ml of Yeast Lytic Enzyme (84 700 U/g from ICN) and 10 mM β-mercaptoethanol, and incubated at 37°C for 30 min. The spheroplasts were centrifuged at 20°C at the same speed as above, and the supernatant was discarded. One milliliter of pre-warmed lysis buffer (100 mM Tris–HCl pH 8.6, 1.5 M NaCl, 50 mM EDTA, 8% dodecyltrimethylammonium bromide) was added to the spheroplasts. After 5 min at 65°C, 1 vol of chloroform was added and the preparation was spun at 16 000 g. The aqueous phase was transferred to a fresh tube and 1 ml of H₂O and 10 µg of RNase A were added to the aqueous phase, and incubated at 68°C for 20/30 min. A solution of 5% cetyltrimethylammonium bromide in 0.1 M NaCl was added to the preparation to a final concentration of 0.4%, mixed and centrifuged at 16 000 g in an Eppendorf centrifuge. The pellet was resuspended in 150 µl of 1.2 M NaCl and precipitated with ethanol. The DNAs were resuspended in 300 µl of 1× TE (10 mM Tris–HCl, 1 mM EDTA, pH 8.0) overnight at 4°C.

BAC DNAs were extracted from bacteria using Qiagen columns according to the manufacturer’s protocol for low copy plasmid isolation.

Transfer of retrofitted YACs into E.coli cells

The conditions for transferring YAC/BACs into DH10B electrocompetent cells were described in previous publications (8,9). Briefly, an agarose plug was dialyzed against water, melted at 68°C for 15 min, cooled to 45°C for 10 min, treated with 1.5 U of agarase for 1 h at 45°C, and chilled on ice for 10 min. The plug was then diluted 1:1 with water. One microliter of the mixture was electroporated into 20 µl of E.coli DH10B competent cells (Life Technologies) using a Bio-Rad Gene Pulser with settings of 2.5 kV, 200 Ω and 25 µF. Colonies were selected on LB plates containing chloramphenicol at a concentration of 12.5 µg/ml.

RESULTS

A battery of different genomic YAC libraries has been constructed using the cloning vector pYAC4. Most of these libraries are propagated in host strain AB1380. Unfortunately, this strain has no additional selectable markers that can be efficiently used for YAC retrofitting. (Three available markers, his5, ade2 and lys2, are point mutations exhibiting a high rate of reversion.) Therefore efficient modification of YACs cloned in AB1380 can be done either by transferring the YAC into a new genetic background, or by using a set of new dominant selectable markers developed for serial gene disruption experiments (11,19). In this report we describe construction of new YAC circularization vectors and demonstrate their usefulness for retrofitting YACs into BACs directly in AB1380, or after transfer of YACs into a new host strain with convenient genetic markers, employing kar1-induced YAC transfer (13,14).

Conversion of linear YACs into circular BAC/YACs using a dominant selectable marker

YAC retrofitting vector pNK-G418 containing the KanMX4 module and short targeting sequences adjacent to the edges of the insert in the pYAC4 vector was investigated for its ability to circularize YACs propagated in AB1380. Three YAC clones, yWXD1321 (150 kb), yWXD1851 (350 kb) and yWXD931 (670 kb), were selected for these experiments, based on their sizes and gene contents. The first two YACs map to the EDA region in Xq13.3 (20); the third contains a segment of Xq26 that includes the HPRT gene (21), raising the possibility of using HPRT-based selection to move the retrofitted clone into mammalian cells for functional studies. As shown schematically in Figure 1, YAC circularization is accompanied by loss of most sequences of telomeric arms, including CEN4, TRP1, URA3 and the yeast origin of replication (ARS1) sequence. CEN4 sequence is replaced with CEN6 sequence in the retrofitting vector, so that the requisite centromere sequences are still present. However, the loss of the ARS sequence makes the circular YAC dependent on yeast ARS-like sequences present in the human DNA insert. Because such sequences are present on average every 30 kb in mammalian DNAs (22), DNA inserts >100 kb are in general capable of the initiation of replication in yeast cells.

Figure 1.

Schematic representation of circularization of YACs using the pNK-G418 vector. The targeting (‘hook’) sequences homologous to the YAC arms are shown by hatched and checkered boxes in the vector, and their corresponding positions in the TRP and URA3 arms respectively, are indicated. Recombination between the vector and a YAC results in loss of both arms and generation of a circular DNA molecule. HIS3 and KanMX are yeast selectable markers; CEN6 is a yeast centromeric sequence. The HIS3 marker and the CEN6 sequence permit selection in yeast. It is worth noting that the circularization vector does not contain yeast origin of replication. Therefore a stable propagation of a circular YAC depends on the presence of a yeast ARS-like sequences in the insert. The gray boxes on the target YAC show these sequences.

In a typical transformation experiment, the yield of primary G418^r colonies for each YAC clone following the transformation with linearized pNK-G418 plasmid DNA was ~10 colonies per 1 µg of linearized plasmid and 5 × 10⁸ yeast cells; i.e. comparable for YACs of all sizes. For instance, in a typical experiment, nine colonies were obtained for yWXD1851, 11 for yWXD1321 and 30 for yWXD931.

Most of the primary transformants contained a mixture of cells with linear and circular YAC molecules, as a result of either targeting one sister YAC after DNA replication or interaction between a vector and a YAC after the first cell division (see below). Cells with circular YAC DNA molecules can be easily purified from the primary transformant colonies by streaking cells on non-selective medium to obtain single cell colony isolates. Typically, only ~10% single cell colony isolates exhibit the phenotype predicted for YAC circularization with accompanying loss of both arms (i.e. Ura^– Trp^–). The frequency of G418R transformants with this phenotype was approximately the same for all three YACs analyzed, suggesting that efficiency of YAC circularization is not dependent on YAC size (at least in the size range analyzed).

Several lines of evidence support the implication of the genetic data that YAC ends have been lost during circularization by an incoming (transforming) vector. The first evidence comes from karyotypic analyses of the transformants. Chromosome-sized DNAs were subjected to pulsed-field electrophoresis, and the DNA was transferred to a nylon membrane and hybridized with human DNA as a probe.

Figure 2 shows CHEF analysis of a sampling of G418R transformants obtained for one of the YACs analyzed, yWXD1851, 350 kb. Control lane 11 contains genomic DNA from strain AB1380 containing no YAC; as expected, it shows no hybridization. Control lane 10 contains DNA from the parental, linear YAC strain, and shows the expected characteristic band at 350 kb. For comparison, lanes 1–9 contain DNA isolated from G418R colonies that include two which only require uracil (lanes 5 and 7) and one which requires neither uracil nor tryptophan (lane 9), as well as six colonies (lanes 1–4, 6 and 8) that are Ura^– Trp^– (i.e. the phenotype expected for the circular YAC DNA).

Figure 2.

PFGE of G418R clones derived from YAC yWXD1851 by circularization. After PFGE (Materials and Methods), DNA from nine independent transformants (lanes 1–9), from the parental YAC strain (lane 10), and from the AB1380 yeast strain containing no YAC (lane 11) were blotted onto a nylon membrane and hybridized with ³²P-labeled total human DNA to visualize YACs. The circular YACs that have the correct phenotype, requiring both tryptophan and uracil (lanes 1–4, 6 and 8) show only hybridization to labeled human DNA at the well and are inferred to be circular (see text); strains with other phenotypes (see text) show YACs at the position of a dimer (lanes 5 and 7) or the original YAC (lanes 9 and 10).

The Ura⁺ Trp⁺ colony in lane 9 contains a YAC indistinguishable from the original. Such transformants presumably result from integration of the KanMX4 module into a YAC or yeast chromosome. The Ura^– Trp⁺ colonies 5 and 7 show a predominant band that moves with a mobility about twice that of the original YAC (i.e. ∼700 kb). We have not studied these further; but because the URA3 and TRP1 genes are on opposite arms of the original YAC, and one marker is gone in these derivatives, the simplest interpretation is that they are generated from dicentric dimers of the original YAC. Such structures can arise as a result of recombination between two sister YACs initiated by inverted repeats (14).

In contrast, the clones with the phenotype expected for circularization show no species at the position of the linear YAC. Instead, all DNA-positive material was retained in the wells. Similar hybridization profiles were obtained during circularization of all YACs tested. This behavior is anticipated for open circular YACs, as has been shown by Wang and Lai (23). Circular DNAs larger than 100 kb never migrate into a gel under the conditions of PFGE.

More conclusive evidence for the circularization by retrofitting of the YACs comes from experiments in which the resultant YAC/BACs were transformed into and recovered from bacteria (see below) and the sequence at the ends of the insert using the primers LS2 and Primer R (Table 1). For example, in a BAC preparation of yWXD1851, sequence from the borders of the plasmid fragment, from Primer R to the PacI site, showed complete homology to the human X-linked anhidrotic ectodermal dysplasia protein gene, including exon 2 and flanking repeat regions (GenBank accession no. AF003528). The sequence obtained using the LS2 primer also showed direct contiguity to human DNA. The end-sequences were also identical to those in the linear YAC before circularization (A.Srivastava, personal communication).

Fidelity of circularized YACs

For circularization to be useful, the integrity of the human DNA insert must be maintained. Both STS content and restriction fragment patterns were used to assess the sequence content of the circular YACs. Several isolates from each of the three test YACs were checked. For derivatives of yWXD1321 and yWXD1851, four and 10 STSs, respectively, were tested (the STSs, which were all positive in the parental YACs, are further identified in Materials and Methods). All 4/4 and 10/10 were found to be present in the corresponding circularized YACs. For yWXD931, the results with the two best-characterized derivatives were somewhat different. One isolate, yWXD931-8, was positive for all 10 of 10 STSs present in the original YAC. In contrast, isolate yWXD931-2 contained only 7 of the 10 STSs found in the original YAC. Based on the physical map of the region (21), the missing STSs were contiguous, suggestive of deletion of part of the YAC insert.

The results of restriction enzyme digest fingerprints were consistent with the assays of STS content. High molecular weight DNA prepared in agarose plugs was digested with either EcoRI or TaqI, the digests electrophoresed and transferred to nylon filters, and Southern blot fingerprint patterns compared to the digestion pattern of the parental linear YACs with labeled total human DNA as a probe. Figure 3 shows sample results for circularized yWXD1851 and its parent for the 6 bp cutter EcoRI. The fingerprint patterns are identical, consistent with no gross rearrangements or deletions during the process. Comparable results were found for yWXD1321 and yWXD931-8; but yWXD931-2 was missing at least two large DNA fragments observed in both the parental YAC and in yWXD931-8, consistent with the loss of STS content that had been observed. Again, it is difficult to use the largest YAC, but at least one circular clone was obtained with a largely or completely intact human DNA insert.

Figure 3.

Restriction digest fingerprints of parental YACs and circular derivatives of yWXD1851 and yWXD1321. Southern blot hybridization with ³²P-labeled total human genomic DNA as a probe for electrophoretically separated fragments of YACs. Lanes 1–3, DNA from independent G418R transformants of yWXD1851 (lane 1–2) and from the original YAC strain (lane 3), all digested with EcoRI. Lanes 4–6, the same for two transformants and the original yWXD1321.

Overlapping circular BACs from a megabase size linear YAC generated with an Alu-containing circularizing vector

A YAC insert can also be converted into a set of overlapping circular YAC/BACs, using Alu-containing targeting vectors pNKBAC39 or pNKBACMX4. In this case the generation of circular YAC derivatives results from recombination of the targeting sequences with multiple Alu repeats within the YAC insert. The KanMX4 cassette in pNKBACMX4 allows YAC retrofitting in AB1380, but the yield of the transformants selected for G418 is very low (see above). The frequency of transformation is much higher when auxotrophic markers (i.e. HIS3 or TRP1) are used for selection of transformants. But these markers cannot be used in AB1380, and a YAC of interest was therefore transferred into a new genetic background using kar1 transfer techniques (13,14).

We recently developed kar1 strains containing a conditional RAD52 gene. They permit the transfer of a YAC from any host into a recombination-deficient background (13). Propagation of YACs in a strain with regulated RAD52 expression increases YAC stability while retaining the possibility of clone modification by homologous recombination.

A 1.0 Mb clone, YAC77, from human chromosome 1 was transferred by kar1-fusion from AB1380 into YNG96-2, as described in Materials and Methods. YAC transfer was confirmed by PFGE and Southern blot hybridization. The YAC strain was then transformed to histidine prototrophy using the linearized pNKBACMX4 vector carrying two Alu targeting sequences. The yield of His⁺ transformants was approximately 500 times higher than the yield of G418R transformants. More than 98% of the transformants (59 of 60 analyzed) contained YAC77 sequences based on STS analysis (data not shown).

To detect the insert sizes of the resultant YAC/BACs, we carried out studies with YACs treated in several ways. From standard preparations, the circular YACs showed no mobility in PFGE (as in Fig. 2; data not shown). If instead, the agarose plugs were irradiated before electrophoresis with a low dose of gamma rays (~5 krad), double-stranded breaks were generated that linearized some of the YACs, which then migrated with a mobility characteristic of linear species (Fig. 4A). Southern blot hybridization with a probe of labeled total human DNA showed that the circular YACs derived from YAC77 ranged from ∼50 to 700 kb. This is similar to the size range of YACs (60 to 700 kb) that were recovered previously with the pNKBAC39 vector during TAR cloning of human DNA from hybrid cell lines containing whole human chromosomes (9,24).

Figure 4.

PFGE of circular YACs generated from YAC77 with pNKBACMX4 vector. (A) Agarose plugs containing DNA of 11 independent transformants with circular YAC derivatives were electrophoresed after irradiation with a low dose of gamma irradiation (see text) to linearize a fraction of the circular molecules so that they migrate into the gel. (B) Analysis of an additional set of 16 clones after digestion with NotI restriction enzyme. The YACs are visualized by Southern blot hybridization with ³²P-labeled total human genomic DNA.

Similar results were obtained by linearizing the YACs by NotI digestion with an additional set of clones. The resultant linear YACs show a similar range of sizes (Fig. 4B).

Transfer of circularized YAC/BACs into bacteria

The transfer of YAC/BACs from yeast to E.coli has been previously described (8,9), but we have found that efficiency of transfer was low when genomic DNA was prepared by a standard method (18). In this paper we describe a modification of this method that increases the efficiency of YAC/BAC transfer ~100-fold. The modification includes more complete spheroplasting and shortening of the incubation of agarose plugs in NDS solution, as described in Materials and Methods. In a typical experiment with a ∼100 kb YAC/BAC, 1 µl of melted plug prepared by the modified method yields 100 to 500 chloramphenicol resistant colonies.

In general, the efficiency of electroporation was the same for clones of 50 to 150 kb; but for circular clones ~200 kb, it was reduced 5-fold. In the size range of 250 to 300 kb, transformation of E.coli was ~20-fold less efficient; and we obtained only single transformants with YAC/BACs >300 kb.

The integrity of circular YAC/BACs after transfer from yeast into E.coli was tested over the same size range. For a circular derivative of yWXD1321 (150 kb), all 21 CM^r E.coli transformants analyzed contained molecules indistinguishable from that propagated in yeast, based on size, STS content (data not shown) and fingerprint pattern. At the same time only 1/9 tested transformants of yWXD1851 (350 kb) was as faithful to the original (Fig. 5); and all of seven derivatives of yWXD931 (670 kb) lacked most of the content of the original clone. To estimate a maximum size of circular molecules that can be accurately transferred into E.coli cells, we analyzed a set of circular YAC/BAC derivatives in size range from 200 to 320 kb. We observed that clones >250 kb are deleted frequently during electrophoration. Figure 6 shows several examples of YAC/BAC transferring into bacterial cells. Clones of 200, 250 and 270 kb were unchanged in size; but three transformants of a YAC/BAC of 320 kb were all extensively and variably deleted.

Figure 5.

Restriction digest fingerprints of circular YAC/BACs in yeast and after transfer into bacteria. EcoRI-digested YAC/BAC 1851 DNAs from yeast (lane 2) and one of the isolates following transfer to bacteria (lane 1) were electrophoresed in agarose gels, and Southern blot hybridization was done with ³²P-labeled total human genomic DNA as a probe.

Figure 6.

Transfer from yeast to E.coli of YAC/BACs of different size. Circular YAC/BACs of 200 (lane 2), 270 (lane 3), 250 (lane 4) and 320 kb (lane 5) generated from YAC77 were transferred into E.coli. Three CM^r colonies were randomly selected for each YAC/BAC transfer, linearized by NotI digestion, and separated by PFGE. Staining with ethidium bromide shows the position of the linearized YACs. Lane 1, lambda DNA ladder.

DISCUSSION

Both the natural occurrence and artificial construction of circular chromosomes in yeast have precedents in the literature. In pioneering work, Strathern et al. (25) showed that an ∼200 kb circular form of a chromosome in yeast (in their case, chromosome III) is quite stable during mitotic cell divisions. Frequency of loss of this chromosome was no higher than once per 1000 cell divisions. Circular YACs with human and mouse DNA inserts were created either by in vivo ligation (26) or recombination in yeast (9,24,27). While there are no systematic studies, it is clear from published data that circular molecules up to ∼1.0 Mb can be constructed in yeast, and the mitotic stability of these circles is no different from linear YACs of the same size.

Circular YACs have two main advantages compared to their linear counterparts: (i) they can be separated from linear yeast chromosomes using standard alkaline lysis methods (28); and (ii) alternatively, circular YACs can be modified by homologous recombination into BACs to be transferred into E.coli cells to simplify DNA isolation for further physical and functional analyses. But despite these advantages, all existing genomic YAC libraries are built with linear telomere-containing vectors. In order to help mobilize the DNA inserts in YACs for a variety of studies, we have designed vectors that permit YAC circularization in different host strains and transformation into E.coli cells as circular YAC/BACs.

In the size range of 100 to 200 kb, where BACs and YACs overlap in insert capacity, they are generally self-consistent in sequence content (29). In the larger size range where only YACs provide coverage, up to 40% of the YACs are chimeric (30), requiring care in the selection of targets for circularization; but YACs are the only current vector system that can provide coverage in this size range, and because essentially every genomic region has been recovered in at least one YAC that is not chimeric, the technique should be widely applicable. In this paper we have shown that YACs up to ∼700 kb can be efficiently converted to circular molecules by in vivo recombination. We also demonstrated that circular DNA molecules with sizes up to ~250 kb can be efficiently and faithfully transferred from yeast cells into E.coli cells by electroporation. Larger size circular DNAs cannot be moved into bacterial cells intact, but still can be purified from yeast by alkaline lysis preparation or by making use of their differential mobility in circular and relinearized form (Figs 2 and 4).

Circular YAC/BACs may be useful in various genome studies, including DNA sequencing and isolation of entire genes for further functional studies. One potential use of this system is its employment for the generation of autonomously replicating artificial mammalian chromosomes without the need for telomeres. Thus, they might provide a route to the controlled provision of cloned single genes and regions and the study of chromosomal structure and dynamics, both in yeast and possibly, when fitted with selectable markers, in mammalian cells.