| Nucleic Acids Research | Pages |
Analysis of a YAC with human telomeres and oriP from Epstein-Barr virus in yeast and 293 cells
Introduction
Materials And Methods
Plasmid construction
Yeast culture, transformation and YAC purification
Mammalian cell culture and transfection
DNA preparation and Southern analysis
Results
Construction of retrofitting constructs
Retrofitting constructs seed de novo telomeres efficiently in human cells
Telomere cassette plasmids can retrofit a YAC carrying the human CFTR gene
Two human telomeres cause instability and complex rearrangements of the telomere cassette in yeast
YAC yCFTRpTF2pTF5oriP forms circular, not linear molecules in 293-EBNA cells
Discussion
Acknowledgements
References
Analysis of a YAC with human telomeres and oriP from Epstein-Barr virus in yeast and 293 cells
Received May 5, 1999; Revised and Accepted July 23, 1999
ABSTRACT One approach to the construction and propagation of a mammalian artificial chromosome is to build it up in Saccharomyces cerevisiae, using a yeast artificial chromosome (YAC) base. We have demonstrated that circular YACs carrying the Epstein-Barr virus origin of plasmid replication (oriP)are maintained as stable, episomal elements in human cells. We wished to determine whether this technology could be extended, to generate linear extrachromosomal elements. Here, we describe the generation of retrofitting constructs, which permit the addition of human telomeres and the oriP domain to YACs. The constructs contain 0.8 kb of human telomere sequence separated by a unique NotI site from 0.7 kb of Tetrahymena telomere sequence. These constructs seed telomere formation with ~40-60% efficiency in human 293-EBNA and HT1080 cells whether or not the Tetrahymena sequence is removed by NotI digestion. A detailed analysis demonstrates that YACs carrying the human telomere cassettes on both arms show instability of the telomere sequences in S.cerevisiae at a frequency of ~50%. Introduction of correctly retrofitted, linear oriPYACs into human 293-EBNA cells by lipofection resulted in the generation of circular extrachromosomal elements varying in size from 8 to 300 kb. However, no apparently linear YACs could be detected, suggesting that extrachromosomal maintenance of DNA with the oriP/EBNA-1 system is not compatible with linear molecules capped by telomeres.
INTRODUCTION
Mammalian artificial chromosomes (MACs) are an attractive option for stable gene maintenance in mammalian cells and could be useful in genetic modification of organisms or in gene therapy. MACs would allow the introduction of large fragments of DNA consisting of intact genes with endogenous surrounding DNA, thus providing full and stable gene expression while not affecting the host genome (2-4). Yeast artificial chromosomes (YACs) are both a model for studying the DNA requirements for MACs and also a basis for their construction. YACs contain all the DNA sequences necessary for yeast chromosome function, replicate autonomously and segregate at cell division. As the biggest cloning vectors now available, YACs have allowed the cloning of megabase fragments of mammalian DNA including gene clusters with associated locus control regions (5). Transfer of YACs into mammalian cells in tissue culture or into the germline of mice is now well established (6). Furthermore, highly efficient homologous recombination in Saccharomyces cerevisiae allows easy manipulation of YAC DNA in vivo.
One route to MAC construction is to modify a YAC by addition of the functional elements of a mammalian chromosome and then to introduce this DNA into mammalian cells. Such a YAC-based MAC vector could be maintained extrachromosomally in both yeast and mammalian cells and could be shuttled between the two hosts. Such an approach has recently been shown to be partially successful. When a YAC carrying 100 kb of alphoid DNA and capped by human telomere sequences was introduced into human HT1080 cells, the YAC DNA formed a minichromosome in 8 out of 24 cell lines analysed (7). The minichromosomes (estimated to be 1-5 Mb) were much larger than the input DNA and it was impossible to conclude whether they were linear or circular. Similarly, a 1 Mb YAC containing [alpha] satellite DNA was modified by homologous recombination to include human telomere repeats and formed minichromosomes in 7 out of 24 cell lines when introduced into HT1080 cells (8). The minichromosomes (found to be 3.5-12.9 Mb in size) were again much bigger than the YAC, showing that the DNA must have undergone substantial rearrangement. Both groups used lipofection for introduction of the YACs into mammalian cells, showing that this method is suitable for the transfection of quite big DNA molecules.
Viral elements can be used as an alternative to the centromere to provide stable maintenance and segregation of large DNA in mammalian cells. These elements appear to be stable and, unlike alphoid DNA, do not undergo rearrangement in yeast cells. We have previously demonstrated that circular YACs up to 660 kb in size carrying the Epstein-Barr virus latent origin of replication, oriP, were maintained as stable, episomal elements in 293-EBNA cells which express the viral transactivator protein EBNA-1 (1). Such oriPYACs persist as unrearranged episomal elements for >120 generations, in the presence or absence of selection, and appear to segregate passively by association with host chromosomes. We wished to determine whether it would be possible to extend this technology to allow the stable, episomal maintenance of linear oriPYACs in human cells.
It has been shown that a cloned array of telomeric DNA (TTAGGG)n can fragment a mammalian chromosome and seed the formation of functional telomeres de novo in mammalian cells (9-12). When human telomere arrays are introduced into yeast they can seed yeast telomere formation with fairly high efficiency, but it has been shown that long mammalian telomeres are generally reduced in length from ~10-25 to <1.2 kb before being capped by a yeast telomere array (13,14). However, yeast telomere arrays (as found on YACs) are unable to form telomeres with any appreciable frequency when introduced into mammalian cells (15). This suggests that it is necessary to introduce human telomere arrays onto a YAC if it is to form a linear MAC in mammalian cells. It is possible to introduce human telomeric DNA onto the ends of a YAC by homologous recombination in yeast such that the human telomere seeds yeast telomere formation. However, the human array that remains is generally quite small and the resulting human/yeast telomere sequence has been found to be quite inefficient at telomere formation when the YAC is introduced into mammalian cells (16).
Here we describe three retrofitting vectors which allow the yeast telomeres at the ends of YACs to be replaced with a telomere cassette consisting of 800 bp of human telomere sequence separated by a NotI site from 700 bp of Tetrahymena telomere sequence. When introduced into the yeast host, the retrofitting vectors should recombine to replace the YAC vector telomere regions with the human/Tetrahymena cassette. The advantage of such a cassette is that the Tetrahymena sequence at the end will seed telomere formation in the yeast, thus shielding the human telomere fragment from abbreviation during propagation in yeast. In this paper we show that the retrofitting vectors do seed telomere formation efficiently in mammalian cells and that they can be used to introduce mammalian telomeric DNA onto the ends of the YAC arms. In addition, one of the vectors contains the EBV oriP domain, thus allowing construction of linear molecules carrying the oriP region and human telomeres, a combination which may allow the maintenance of linear extrachromosomal molecules in mammalian cells. However, transfection of linear YACs retrofitted with the telomere cassette and oriP gave rise to circular rather than linear molecules in human cells.
MATERIALS AND METHODS
Plasmid construction
To make the plasmid pTF2 (Fig. 1A), the yeast selectable marker LYS2 was isolated as a 5 kb EcoRI-HindIII fragment of pCH37, which was blunt-ended and cloned into the SmaI site of plasmid pSXneo-0.8-T2AG3 (from T. de Lange; 10). pCH37 is YCp50 with the 5 kb LYS2 fragment from Yp333 (17) cloned into the EcoRI and HindIII sites. The resulting plasmid, pSXneo-0.8-T2AG3LYS2, was digested with ClaI and a phosphorylated SalI linker was introduced. A 700 bp (T2G4)n-containing fragment was isolated from plasmid pCH5 by NotI + SalI digestion and cloned into the NotI and SalI sites of pSXneo-0.8-T2AG3LYS2 to generate plasmid pTF2. pCH5 contains a 700 bp Sau3AI fragment containing the Tetrahymena telomere from pYAC4 (5) cloned into BamHI-cut pBluescript (Stratagene).
Figure 1. Retrofitting plasmids pTF2, pTF5oriP and pTF6. (A) The left arm vector pTF2 carries the yeast selectable marker LYS2, the mammalian selectable marker neor and the human/Tetrahymena telomere cassette (Hum tel and Tetr tel). Homology to the left arm of pYAC4 is provided by an ~190 bp fragment of the ampr gene (hom 1). The plasmid was digested with SalI and BglI prior to retrofitting the YAC. (B) The right arm retrofitting vector pTF5oriP carries the yeast marker HIS5, the mammalian selectable marker hygror, the EBV oriP domain and the human/Tetrahymena telomere cassette. Homology to the right arm of the YAC is provided by an ~120 bp fragment (hom 2). The plasmid was linearised with SalI prior to retrofitting a YAC. (C) pTF6 is the same as pTF5oriP, but lacks the oriP domain.
To construct pTF5oriP (Fig. 1B), a deleted version of pTF2 was produced by digestion with KpnI and circularisation. This plasmid was cut with KpnI, blunt-ended and ligated with a blunt-ended 2.1 kb SalI fragment from plasmid pCH45 containingthe yeast selectable marker HIS5, generating plasmid pTF3. pCH45 is YCp50 containing a PCR fragment of HIS5 amplified from AB972 spanning positions -615 to 1451 of the HIS5 gene as published (18), cloned into the SalI site. The NruI-SalI fragment of pTF3 containing the telomere cassette and the HIS5 gene was cloned into the HindIII and SalI sites of the plasmid pHEBo (carrying oriP and the hygror gene) (from B. Sugden; 19) to generate pTF5oriP.
The fragment of pTF5oriP, containing the telomere cassette and the HIS5 gene, was obtained by partial digestion with ClaIand complete digestion with SalI. It was cloned into the ClaI and SalI sites of plasmid pHyg (from B. Sugden; 19), generating plasmid pTF6 (Fig. 1C).
Yeast culture, transformation and YAC purification
yCFTR is the YAC 37AB12 with the intact CFTR gene (20) which has been retrofitted with pRV1 (21). Yeast containing yCFTR were grown in selective medium lacking lysine and tryptophan. Yeast containing the YAC yCFTRpTF5oriP were grown in selective medium lacking histidine. Yeast containing YAC yCFTRpTF2pTF5oriP were grown in selective medium lacking histidine and lysine. All yeast media were supplemented with 40 mg/l adenine.
Spheroplast transformation of yeast with plasmid DNA was carried out as described previously (22). Aliquots of 100 ng of plasmid DNA and 1 µg of salmon sperm carrier DNA were used in each transformation reaction. Transformants were plated on sorbitol plates lacking histidine (transformation with pTF5oriP) or lacking histidine and lysine (transformation with pTF2).
For transfection into mammalian cells, YAC DNA was isolated from a 1% low melting point agarose pulsed-field gel. The edges of the gel were cut off and stained with ethidium bromide to locate the YAC band. The gel was reassembled and the slice of gel containing the YAC was cut out and equilibrated in 20 ml of 1× agarase buffer (10 mM Bis Tris-HCl, 1 mM EDTA, 1 mM polyamines pH 8.0) at 4°C overnight. The agarose slice was cut into smaller slices, ~0.5 g/Eppendorf tube. The tubes were spun briefly, incubated at 68°C for 10 min and then at 40°C for 5 min. Ten units of [beta]-agarase (New England Biolabs) were added to each tube and mixed one to three times with a cut-off blue tip. Tubes were incubated at 40°C for 2 h, then at room temperature for 5 min. The integrity of the purified YAC was verified by PFGE.
Mammalian cell culture and transfection
293-EBNA is a human embryonic kidney cell line constitutively expressing the EBNA-1 protein and was obtained from Invitrogen. HT1080 is a human fibrosarcoma cell line obtained from Dr Howard Cooke (Edinburgh). Cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% foetal calf serum. For 293-EBNA, the medium was supplemented with 250 µg/ml G418 to maintain selection for the EBNA-1 gene.
Plasmid DNA and YAC DNA were introduced into cells by lipofection using lipofectAMINETM reagent (Gibco BRL). The number of cells used in each transfection was 2 × 105 (for HT1080) or 1 × 106 (for 293-EBNA) per well of a 6-well dish. For each transfection 100 ng of DNA and 6-12 µg of lipofectamine were combined in serum-free DMEM or OPTI-MEM (Gibco BRL). Liposomes and DNA were left in contact with the cells for 16-18 h. Cells were then allowed to recover in complete growth medium for 6 h before being split 24 h after lipofection; one well into three (HT1080) or four (293-EBNA) 90 mm dishes. After growth without selection for a further 1 (293-EBNA) or 2 (HT1080) days, selective medium was applied: 500 µg/ml G418 for transformation of HT1080 cells with plasmid pTF2; 200 µg/ml hygromycin B for transformation of 293-EBNA cells with plasmid pTF6; 300 µg/ml G418 + 200 µg/ml hygromycin B for transformation of 293-EBNA cells with YAC DNA. After 2 weeks colonies were picked and expanded for analysis.
DNA preparation and Southern analysis
High molecular weight yeast DNA was prepared in agarose blocks as described previously (23).
Genomic DNA from mammalian cells was isolated in agarose plugs. Cells were washed once with phosphate-buffered saline (PBS) and resuspended at 2 × 107 cells/ml in PBS. An equal volume of prewarmed 2% Seaplaque agarose in PBS was added to the cell suspension. The plugs were treated with LDS solution (1% dodecyl lithium sulfate, 100 mM EDTA, 10 mM Tris, pH 8.0) at 37°C with agitation for 1 h and then with fresh LDS overnight under the same conditions. The plugs were then washed in NDS (0.2% lauryl sarcosine, 100 mM EDTA, 2 mM Tris, pH 9.0) at 50°C overnight.
Restriction enzyme digests of DNA in plugs were carried out overnight using 40-60 U of enzyme in 200 µl of restriction enzyme buffer supplemented with 100 µg/ml BSA. Prior to digestion, plugs were washed twice for 30 min in TE buffer and twice for 1 h in 1× restriction enzyme buffer.
DNA was [gamma]-irradiated in a Gammacell 1000 Elite. The optimal irradiation dose was determined experimentally. A dose of 250 Gy was needed to linearise molecules of 100-200 kb, 1000 Gy was sufficient for molecules of 50 kb, while molecules of 10 kb were linearised only by doses of 2000 Gy.
DNA was transferred onto Hybond N+ nylon membrane (Amersham) by alkaline blotting as described by the manufacturer. Prehybridisation and hybridisation were carried out at 65°C in modified Church buffer (24) (16.8 g/l NaH2PO4·H2O, 21.4 g/l Na2HPO4, 7% SDS, 100 µg/ml salmon sperm DNA). Filters were washed at 65°C twice for 15 min in 2× SSC, 0.1% SDS and once for 15 min in 0.1× SSC, 0.1% SDS.
The neor probe is the 1.1 kb XhoI-SalI fragment of plasmid pMC1neopolyA (Stratagene). The Tel probe is the 0.8 kb SstI-NotI fragment of pTF2, which contains a human telomere array (the probe hybridises only to human telomeres, but not to Tetrahymena telomeres). The HIS5 probe is the 2.4 kb ClaI-SstI fragment of pTF6. The LYS2 probe is a 3 kb KpnI fragment of pTF2. The oriP probe is the 0.9 kb SmaI fragment from the plasmid pTF5oriP. DNA fragments were gel purified using the QIAEX II Gel Extraction Kit (Qiagen). The probes were labelled using the Megaprime DNA labelling system (Amersham).
RESULTS
Construction of retrofitting constructs
Three retrofitting vectors were constructed to allow introduction of the telomere cassette, with or without oriP, onto YACs (Fig. 1). All three retrofitting vectors, pTF2, pTF5oriP and pTF6, contain the same 1500 bp telomeric cassette, consisting of 800 bp of human telomeric sequence separated by a unique NotI site from 700 bp of Tetrahymena telomeric DNA. Both telomere arrays are orientated in the same direction, with the G-rich strand running 5[prime]->3[prime] towards a unique SalI linearisation site at the 3[prime]-end of the Tetrahymena sequences. During growth in S.cerevisiae, the Tetrahymena sequences should act as a telomeric seed preventing abbreviation of the human telomeric array. pTF2 introduces the telomere cassette and neor gene onto the left arm of any YAC based on the vector pYAC4 and has LYS2 as the yeast selectable marker. pTF5oriP and pTF6 introduce the telomere cassette and hygror gene onto the right arm of any YAC based on the vector pYAC4 and have HIS5 as the yeast selectable marker. In addition to this, pTF5oriP introduces the oriP domain of Epstein-Barr virus onto the right arm of a YAC. Thus, the telomere cassette can be introduced onto both arms of a YAC along with the neor and hygror genes and the EBV oriP domain (optional).
Retrofitting constructs seed de novo telomeres efficiently in human cells
Prior to YAC modification with the vectors, we tested the ability of the cloned telomeric cassette to seed de novo telomere formation in human cells with or without the presence of Tetrahymena telomeric sequences distal to the human telomere.
Two outcomes are possible after stable introduction of telomeric DNA into mammalian cells. First, integration of the human telomeric DNA may lead to breakage of the chromosome at the site of integration and formation of a telomere. Secondly, the telomere fragment will integrate into a chromosome without seeding the formation of a new telomere (9-11). These two possibilities can be distinguished by looking at the length of telomere-containing fragments formed after integration. De novo telomere formation will result in a terminal restriction fragment which is heterogeneous in size (appearing as a smear on a Southern blot) while simple integration will result in a discrete band (9-11).
Plasmid pTF6 was utilised to test de novo telomere formation by the telomere cassette after introduction into 293-EBNA cells. The three DNA constructs shown in Figure 2A were each introduced into 293-EBNA cells by lipofection and DNA from 12 hygromycin-resistant clones was analysed for each construct by BglI digestion and hybridisation with the HIS5 probe. The results for pTF6-S, in which both telomeres are present and the human telomere fragment is shielded by the Tetrahymena telomere fragment, are shown in Figure 2B. No signal was detected in 2 out of 12 clones, a distinct smear was seen in 4 out of 10 clearly hybridising clones (indicated by an asterisk) and one or more discrete bands are seen in the other clones where single or multiple integrations had occurred without de novo telomere formation. For construct pTF6-SN, in which the human telomere fragment was exposed at the end of the molecule, 6 out of 10 strongly hybridising cell lines had smeared bands (Fig. 2C). The size range of smeared fragments was from ~9 to >15 kb. So the amount of added telomere array could be estimated to range from ~4 to >10 kb. In contrast, no smeared bands were observed in the case of pTF6-SS, in which both telomere fragments were removed (data not shown).
Figure 2. Testing de novo telomere formation in 293-EBNA cells. (A) The constructs used for testing were obtained from pTF6 by digestion with SalI (pTF6-S), SalI + NotI (pTF6-SN) and SalI + SstI (pTF6-SS). DNA was gel purified and introduced into 293-EBNA cells by lipofection. (B and C) Detection of de novo telomere formation in 293-EBNA cells by capped and nude human telomere fragments. High molecular weight DNA isolated from the clones of 293-EBNA cells that had been stably transfected with pTF6-S (B) and pTF6-SN (C) was digested with BglI, separated on 1% agarose gels, transferred to membranes and hybridised with 32P-labelled HIS5 probe. Clones in which telomere seeding activity was detected are labelled with *, clones in which no hybridisation signal was detected are marked with -.
De novo telomere formation was also tested in HT1080 cells using a similar strategy. Three pTF2-based fragments were produced by digestion with SalI (capped human telomere), SalI + NotI (nude human telomere) and SalI + SstI (without any telomeric DNA) (Fig. 1A) and these were introduced into HT1080 cells. DNA from 15-20 G418-resistant stable transformants were analysed for each construct by digestion with HpaI and hybridisation with the LYS2 probe. Smeared telomeric bands were observed in the case of cell lines obtained by transfection of DNA with either just the human telomere (44%) or the human/Tetrahymena telomere cassette (50%), while no telomere formation was detected for the construct without any telomeric DNA (data not shown). The final results for both cell lines are shown in Table 1.
Table 1. De novo telomere formation in mammalian cells
| Construct | Cell line | Number of cell lines tested | Cell lines with no signal | Frequency of de novo telomere formation |
| Human/Tetrahymena cassette | 293-EBNA | 12 | 2 | 40% (4/10) |
| HT1080 | 19 | 5 | 50% (7/14) | |
| Human telomere only | 293-EBNA | 12 | 2 | 60% (6/10) |
| HT1080 | 20 | 4 | 44% (7/16) |
Telomere cassette plasmids can retrofit a YAC carrying the human CFTR gene
In this paper we retrofitted the YAC yCFTR, which carries the intact cystic fibrosis transmembrane regulator gene (CFTR) (Materials and Methods). Replacement of the YAC arms was carried out in two steps, with careful analysis of the telomeres after the retrofitting events (Fig. 3). The retrofitting vectors pTF2 and pTF5oriP have the selectable markers LYS2 and HIS5, respectively, and so can be used in either order on a YAC cloned with pYAC4. However, the YAC retrofitted in this paper has previously been retrofitted with the plasmid pRV1 (21), which introduces a LYS2 gene onto the right arm, therefore pTF5oriP was used for the first round of retrofitting.
Figure 3. General scheme of retrofitting the YAC arms with pTF5oriP and pTF2. (A) The structure of the YAC vector arms prior to retrofitting is shown at the top. Fragments hom 1 and hom 2 are regions of homology used in the two steps of homologous recombination. The structure of the arms after retrofitting with the positions of the diagnostic restriction sites and probes (not to scale) are underneath. Symbol KpnI* indicates two closely situated KpnI sites. The fragments expected with various digests and probes are shown at the bottom. (B) Analysis of clones obtained after retrofitting the left arm of the YAC with pTF2 (second retrofitting).DNA from yCFTRpTF2pTF5oriP clones 2, 5, 7 and 9 (as indicated above each lane) was digested with HpaI, HpaI + NotI and HpaI + SstI as indicated.The filter was probed with the LYS2 probe. The positions of size markers are shown on the right, while the positions and expected sizes of the fragments are shown by the arrows on the left. The constant 4.3 kb band is due to hybridisation with the endogenous LYS2 gene and is present in a lane with an unretrofitted YAC (not shown). (C) Analysis of clones obtained after retrofitting the right arm of the YAC yCFTR with pTF5oriP (first retrofitting). DNA from four clones (1, 2, 10 and 11 as indicated above each lane) was digested with BglI, BglI + NotI and BglI + SstI and probed with the HIS5 probe (a larger fragment due to hybridisation with the endogenous gene is off the top of the region of gel shown). DNA from the same clones digested with NotI + SstI and probed with the Tel probe is shown below. The positions of size markers are shown on the right, while the positions and the expected sizes of the fragments are shown by the arrows on the left.
pTF5oriP plasmid DNA was linearised with SalI and introduced into the yeast strain containing yCFTR by spheroplast transformation. Twelve transformants with the desired LYS- HIS+ TRP+ phenotype were shown by PFGE and Southern blotting to contain the intact 320 kb YAC. The size of the telomeric region inthe 12 clones was checked by SstI digestion and hybridisation with a telomere probe (data not shown). Clones with the biggest and brightest telomere fragment (clones numbers 1, 2, 10 and 11) were chosen for more detailed analysis of the telomere region as shown in Figure 3C. All four clones had the right pattern of fragments when probed with the HIS5 probe: a 4.7 kb BglI fragment which is reduced to 4.0 kb on NotI digestion and to 3.2 kb on SstI digestion (Fig. 3A and C). All four clones also contained the intact 800 bp human telomere fragment released with NotI + SstI digestion (Fig. 3C). This indicates that all four YACs have the intact 0.8 kb human telomere fragment separated by a NotI site from the Tetrahymena telomere fragment.
For the second round of retrofitting, pTF2 was digested with BglI and SalI prior to introduction into the yeast spheroplasts. DNA from 24 transformants with the desired LYS+ HIS+ phenotype was separated by PFGE and hybridised with the neor probe, which is specific for pTF2. Four clones in which the 330 kb YAC hybridised strongly with the neor probe (2, 5, 7 and 9) were analysed in more detail as shown in Figure 3B, using the LYS2 probe. There is a constant 4.3 kb HpaI band from the yeast genomic LYS2 gene, which co-migrates with the band from the YAC. If the NotI site is present, the 4.3 kb HpaI fragment should be reduced to 3.6 kb with NotI digesion, as seen for clones 5, 7 and 9. However, smaller fragments are seen in clones 2 and 9 (Fig. 3B) and these fragments (though fainter) are also present in the HpaI digests without NotI digestion, indicating that in a proportion of the YACs the NotI site has been lost. In these clones the reduction in size of the total telomere cassette indicates that the yeast telomere array is now probably seeded directly off the human telomere array. All eight YACs contained a population of molecules which had lost the NotI site, but the proportion of such molecules differed significantly between the clones, with clone 7 having the fewest deleted telomeres.
In clone 7 the size of the HpaI + NotI and HpaI + SstI fragments were 3.6 and 2.8 kb as expected, indicating that the human telomere fragment between the NotI and SstI sites had retained its initial 800 bp size. However, the HpaI fragment was somewhat larger than the expected size of 4.3 kb (Fig. 3B). The additional sequence (~500 bp) has been acquired in the Tetrahymena telomere region (presumably by addition of yeast telomeric DNA) but this should not hinder the ability to seed de novo telomere formation in mammalian cells.
Two human telomeres cause instability and complex rearrangements of the telomere cassette in yeast
To test how quickly rearrangements of the telomere cassette occurred during growth of the yeast, we analysed eight single colonies from clone 7 (see above) after streaking out onto an agar plate. The results for four clones (7-5, 7-6, 7-7 and 7-8) after hybridisation with the telomere probe which gives information on the structure of both arms are shown in Figure 4. The pTF5oriP arm (first retrofitting, bands indicated by solid arrows in Fig. 4) gives the expected bands in clones 7-5, 7-6 and 7-8. However, a large proportion of the clone 7-7 molecules have lost the NotI site giving a 4 kb BglI fragment and a 0.7 kb fragment with BglI + SstI digestion (Fig. 4A). The pTF5oriP human telomere fragments revealed with NotI + SstI digestion (indicated by solid arrows in Fig. 4B) are ~800 bp in clones 7-5 and 7-6 and slightly larger in 7-8.
Figure 4. Analysis of rearrangements of the telomere regions of the retrofitted YAC during growth. (A) DNA from four single colonies of clone 7 (7-5, 7-6, 7-7 and 7-8) of yCFTRpTF2pTF5oriP (see Fig. 3) was digested with BglI, BglI + NotI and BglI + SstI and probed with the Tel probe. (B) DNA from the same clones digested with NotI + SstI and NotI + KpnI probed with the Tel probe. Colonies 7-5, 7-6, 7-7 and 7-8 are indicated by 5, 6, 7 and 8 above each lane. Fragments derived from the pTF2 arm are shown by white arrows, fragments from the pTF5oriP arm are shown by black arrows. The expected sizes of the fragments are shown at the left of the arrows (see Fig. 3).
On the pTF2 arm (second retrofitting, open arrows in Fig. 4) the NotI site is still present in clones 7-5, 7-7 and 7-8, because on digestion with NotI + KpnI the 2 kb fragment distal to the SstI site (present in 7-5, 7-7 and 7-8, but 1.5 kb in 7-6) is reduced to between 700 and 900 bp (Fig. 4B). The human telomere fragment is ~850 bp in 7-5, ~700 bp in 7-6, between 700 and 800 bp in 7-7 and ~720 bp in clone 7-8 (open arrows in Fig. 4B).
The different sizes of the human telomere fragments in all four colonies indicates that more complex rearrangements had taken place. Clone 7-5 appears to be the least rearranged because it appears to have 850 bp of human telomeric array on the pTF2 arm and 750 bp on the pTF5oriP arm. This clone was chosen for introduction into 293-EBNA cells.
YAC yCFTRpTF2pTF5oriP forms circular, not linear molecules in 293-EBNA cells
The YAC yCFTRpTF2pTF5oriP, which carries human telomeres at both ends and oriP from EBV, was then introduced into human 293-EBNA cells (which express EBNA-1) in order to determine whether it is possible to generate linear molecules maintained by the oriP/EBNA-1 system in human cells. As we had found that the human/Tetrahymena telomere cassette is quite unstable during growth of the yeast (see above), it was necessary to carefully analyse the two telomere regions prior to introduction of the YAC into human cells. Four colonies of clone 7-5 with yCFTRpTF2pTF5oriP (shown in Fig. 4) were grown up and the telomere regions analysed as before in Figure 4A. Only two of the four colonies (clones 751 and 753) contained YACs with the NotI site on both arms (data not shown), again indicating how unstable the cassette is during culture. Both the human telomere arrays in clones 751 and 753 were >750 bp. Intact YAC DNA was purified from PFGE gels of high molecular weight DNA of clones 751 and 753 (the same DNA which had been analysed for telomere structure). The DNA was checked for intactness by PFGE and then introduced into 293-EBNA cells by lipofection. Colonies were selected for G418 and hygromycin B resistance and 16 colonies from each YAC were expanded for analysis.
In order to find out whether the YAC was being maintained extrachromosomally and to distinguish between linear and circular molecules, we carried out PFGE of high molecular weight DNA with or without prior treatment with [gamma]-irradiation. If the YAC has integrated it will not enter the PFGE gel without [gamma]-irradiation and will form a smear of DNA at the same size as the genomic fragments after irradiation. If the YAC is a linear extrachromosomal molecule, it will enter a PFGE gel with or without [gamma]-irradiation and will run with the same mobility in each case. If the YAC is a circular extrachromosomal molecule, it will enter the PFGE gel if it is less than ~300 kb in size (and supercoiled). After irradiation the circular molecules will be cut into linear molecules which will run with quite different mobility to the circular molecules. More irradiation will be needed to linearise smaller circular molecules. Finally, supercoiled circular molecules migrate at approximately the same rate in PFGE irrespective of the switching time, unlike linear molecules, where larger molecules are resolved with longer switching.
High molecular weight DNA from the 32 different clones, with and without different amounts of [gamma]-irradiation treatment, were analysed by PFGE using various switching times, followed by Southern blotting and hybridisation with an oriP probe specific for the YAC. The results for six clones are shown in Figure 5. Clones 751-13 and 751-14 appear to contain a mixture of circular molecules; the molecules of 50-100 kb are linearised and visible in Figure 5 but the additional smaller circles are not linearised with this amount of irradiation and run near the limit of resolution on this gel. Clone 751-15 has a circular molecule of ~190 kb which runs just below the limit of resolution on a PFGE gel without irradiation but is linearised with irradiation. Clone 753-9 has a supercoiled molecule of ~30 kb, which is too small to be linearised with 250 Gy of irradiation, as shown in Figure 5, but with 1000 Gy of irradiation the linear form is seen (data not shown). Clones 753-11 and 753-12 have circular molecules of ~80 kb which are completely linearised with irradiation. In clone 753-12 there is additional integrated DNA or extremely large circular DNA, which runs at the limit of mobility.
Figure 5. Analysis of the yCFTRpTF2pTF5oriP DNA in the 293-EBNA cell lines. High molecular weight DNA from the cell lines, either without [gamma]-irradiation treatment (-) or irradiated with 250 Gy (+), was run on a pulsed-field gel with switching time of 30 s for 24 h. The gel was blotted and the filter hybridised with the oriP probe. The positions of the wells, the limit of resolution and the size markers are indicated at the right of the gel.
A strong hybridisation signal with the oriP probe was detected in 30 of the 32 clones analysed. In all 30 clones the YAC DNA was found to be extrachromosomal (only 753-12 has an additional, probably integrated, copy), indicating that oriP and EBNA-1 are functional. In every case the extrachromosomal DNA was circular as opposed to linear and the molecules ranged from 8 to 300 kb in size (Table 2).
Table 2. Size of episomal molecules found in the 293-EBNA cells transfected with yCFTRpTF2pTF5oriP DNA
| Size (kb) | Number of cell lines |
| 8-50 | 10 |
| 60-100 | 7 |
| 110-150 | 10 |
| 160-200 | 5 |
| 210-250 | 2 |
| 260-300 | 1 |
Further analysis was carried out to determine which functional elements were still present on the circular molecules. The hygror gene on the right arm of the YAC was used for selection. The oriP element is next to the hygror gene and is present in all the cell lines, including the line with an integration. The HIS5 probe hybridised to discrete BglI fragments ranging from 3 to 20 kb in most of the cell lines. The CFTR gene spans ~200 kb of the YAC and hybridisation with the CFTR probe indicated that about half of the circular elements do contain some of the CFTR gene, but it was lacking from the very small elements, as expected. LYS2 is located on the opposite arm to hygror and oriP and was found to be lacking in all the cell lines except the one with an apparent integration.
DISCUSSION
We describe three retrofitting vectors that introduce a human/Tetrahymena telomere cassette onto the arms of YACs made in the vector pYAC4. The left arm vector (pTF2) also introduces a neor gene while the right arm vectors (pTF5oriP and pTF6) introduce a hygror gene, allowing for selection of both arms in mammalian cells. pTF5oriP also introduces the oriP domain of Epstein-Barr virus, which, with EBNA-1, allows extrachromosomal maintenance of DNA in mammalian cells.
The telomere cassette used in each construct consists of 800 bp of human telomere array separated from 700 bp of Tetrahymena telomere array by a NotI site, which was selected because of its rare occurrence within genomic DNA (at an average 1 site/100 000 bp of human genomic DNA). Plasmid DNA carrying this telomere cassette was shown to form telomeres with 40-60% efficiency in both 293-EBNA and HT1080 cells in tissue culture. Telomere formationabsolutely required the presence of the human telomere fragment. However, there was no significant difference in efficiency of telomere formation whether or not the Tetrahymena telomere region was present, indicating that it is not advantageous to remove the Tetrahymena sequence prior to transfection into mammalian cells. This is very similar to previous results where the same human telomeric array of 800 bp seeded telomere formation in 60-70% of HeLa cell lines and an extra 0.9 kb of non-telomere DNA did not affect the efficiency of telomere formation (10).
The retrofitting vectors were also shown to efficiently recombine with a YAC to replace the yeast telomeres with the human/Tetrahymena telomere cassette. After the first round of retrofitting with pTF5oriP, four of four clones were found to contain the NotI site and the human telomere fragment was unrearranged at 800 bp. This indicates that the cassette was functioning as desired, leaving the full 800 bp of human telomere sequence unchanged while the yeast uses the Tetrahymena sequence to seed telomere formation. The human telomere sequence at the end of the YAC would then be expected to function with 40-60% efficiency in human cells, as the only difference from the original plasmid is that the Tetrahymena telomere sequence has probably largely been replaced by yeast telomere sequence (such a Tetrahymena/yeast telomere sequence seeds telomere formation with very low efficiency in mammalian cells; 15).
The second round of retrofitting with pTF2 led to YACs with both arms carrying human telomere sequences. However, the telomere arrays in YACs with both arms retrofitted appeared to be more unstable and prone to rearrangements than YACs with one retrofitted arm. When the right arm, which had initially been correctly retrofitted with pTF5oriP, was analysed after the second retrofitting it appeared to have undergone substantial telomeric rearrangements, including seeding of the yeast telomere directly off the human telomere sequence and changes in size of the human telomere array. Our data also show that the process of rearrangement of the dual-telomere cassette continues during growth of individual clones after retrofitting of both ends; rearranged clones comprise ~50% of the population, but can be avoided by careful screening prior to use of retrofitted clones. The presence of the unique NotI site separating the two telomere arrays allowed the presence and the size of the human telomere fragments to be monitored. This is important as the smaller human telomere arrays which remain after addition of yeast telomeres directly onto a human telomere have been observed to give a lower frequency (~20%) of telomere formation in mammalian cells (16). Ikeno et al. used a similar dual-telomere cassette containing 1.1 kb of mammalian TTAGGG telomere repeats separated by an I-SceI site from 0.3 kb of yeast telomere repeats. However, the final state of the telomere arrays in yeast and the efficiency of telomere formation in mammalian cells were not described.
After introduction of the retrofitted YAC into 293 cells expressing the EBNA-1 protein, 30 of 32 cell lines were found to contain episomal circular molecules with sizes ranging from 8 to 300 kb. Selection was for the hygror gene on the right arm that is therefore present in all the cell lines. In most of the clones the HIS5 gene is located on a new discrete BglI fragment. These novel fragments represent the junctions formed during circularisation of the YACs and may contain residual telomere DNA (this could not be determined due to the large amount of endogenous telomere DNA in the cell lines). In the other cell lines that do not contain the HIS5 gene, the junction must have formed between the hygror and HIS5 genes, removing all the telomere sequence in the process. Many of the extrachromosomal molecules have lost the CFTR gene and all have lost the LYS2 gene on the left arm, suggesting that circularisation occurred within the YAC rather than at the very ends (this is consistent with the resulting molecules being smaller than the input YAC). The only cell line with an apparent integration contained the LYS2 gene, suggesting that the intact YAC had been transfected into the cells and that the left arm was lost during circularisation. OriP was present in all the molecules and the presence of extrachromosomal elements indicates that oriP and EBNA-1 are fully functional in this experiment.
All 30 cell lines contained circular rather than linear extrachromosomal molecules, even though they had human telomere cassettes at both ends that should function with 40-60% efficiency in the 293-EBNA cell line. A trivial reason for this could be that the oriPYACs have been degraded during purification, but the DNA was checked by PFGE prior to lipofection. Furthermore, although it is more technically challenging to introduce large purified YACs into mammalian cells by lipofection, lipid-mediated transfer has also been used by others to successfully introduce a 1 Mb YAC into HT1080 cells (8).
There are several alternative explanations for the lack of linear molecules in the cell lines. First, linear molecules may be unable to replicate completely. Replication of the circular form of EBV is initiated bidirectionally in the dyad symmetry region of oriP but one fork is blocked at the family of repeats region or oriP leaving the other to replicate most of the genome unidirectionally (25). On a linear molecule this would prohibit complete replication if there were no other replication origin between the family of repeats and the telomere. However, the plasmid pTF5oriP was constructed in such a manner that the dyad symmetry is situated closer to the end of the YAC, while the family of repeats is positioned towards the CFTR gene, so that replication of the end of the short YAC arm should be efficient and replication of the long arm would rely on replication initiating within the 300 kb of the CFTR genomic DNA. It has previously been shown that ~10 kb of mammalian DNA permits replication of EBV-based vectors lacking the dyad symmetry domain (26), thus 300 kb would be expected to contain sequences that could function as origins in human cells.
Secondly, it is possible that oriP/EBNA-1 function is not compatible with the linear conformation of the molecule. Ascenzioni et al. (27) previously analysed the replication and maintenance of small circular and linear molecules (with or without short telomere arrays) replicated from the SV40 origin in COS-7 cells. They observed that only circular molecules replicated, either DNA transfected as circular molecules or circularised versions of the linear molecules. This suggests that the SV40 origin has a requirement for circularly constrained DNA in vivo. Although EBV is packaged as a linear molecule within the viral capsid, circularisation of the genome is believed to occur shortly after infection. Thus linear conformations do not appear to represent a predominant cellular form of the virus.
Finally, chromatin assembly may differ on linear and circular templates. In vitro studies with HeLa cell extracts have indicated the presence of cellular factors that inhibit the replication of linear SV40 templates (28) and correct chromatin structure is essential for assembly of the replication complexes. It is possible that linear forms of the oriPYACs are not assembled into a permissive chromatin structure.
ACKNOWLEDGEMENTS
K.S. was a Wellcome Trust student and this work was also supported by the MRC.
REFERENCES
*To whom correspondence should be addressed. Tel: +44 207 594 3224; Fax: +44 207 594 3015; Email: t.tolmachova{at}ic.ac.uk Present address: Kaetrin Simpson, Cold Spring Harbor Laboratory, PO Box 100, 1 Bungtown Road, Cold Spring Harbor, NY 11724, USA
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Lindenbaum, E. Perkins, E. Csonka, E. Fleming, L. Garcia, A. Greene, L. Gung, G. Hadlaczky, E. Lee, J. Leung, et al. A mammalian artificial chromosome engineering system (ACE System) applicable to biopharmaceutical protein production, transgenesis and gene-based cell therapy Nucleic Acids Res., December 7, 2004; 32(21): e172 - e172. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Perea, A. Robertson, T. Tolmachova, J. Muddle, R. H. M. King, S. Ponsford, P. K. Thomas, and C. Huxley Induced myelination and demyelination in a conditional mouse model of Charcot-Marie-Tooth disease type 1A Hum. Mol. Genet., May 1, 2001; 10(10): 1007 - 1018. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Huertas, S. Howe, A. McGuigan, and C. Huxley Expression of the human CFTR gene from episomal oriP-EBNA1-YACs in mouse cells Hum. Mol. Genet., March 1, 2000; 9(4): 617 - 629. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||