Pseudomonas aeruginosa plasmids as suicide vectors in Escherichia coli: resolution of genomic cointegrates through short regions of homology.

Pseudomonas aeruginosa plasmids which cannot replicate in Escherichia coli have been used to introduce specific modifications into the E. coli chromosome by homologous recombination ('gene targeting'). The E. coli gene (gpt) encoding guanine-xanthine phosphoribosyltransferase (Gpt) was used for initial targeting studies owing to the availability of a powerful positive selection for loss of the Gpt+ phenotype (6-thioguanine resistance or 6TGR or Gpt-). P. aeruginosa plasmids containing selectable markers flanked by gpt sequences were introduced as supercoiled DNA into an E. coli strain which contained a normal gpt locus. Primary cointegration of such plasmids into the E. coli genome results in a gene duplication event which maintains Gpt function; a secondary recombinational event which resolves the cointegrate either reverses the primary event or results in replacement of the original gpt copy with the modified version. A 316-bp region of homology was sufficient for cointegrate formation, and resolution of the cointegrates through a shorter (92 bp) homologous flank was selectable through loss of Gpt function. The frequency of cointegrate resolution under these conditions was significantly above the spontaneous gpt mutational loss rate.     

Complementation of sensitivity to alkylating agents in Escherichia coli and Chinese hamster ovary cells by expression of a cloned bacterial DNA repair gene.

Dual expression vectors derived from pSV2gpt and encoding all or part of the Escherichia coli ada+ gene have been constructed. Following transformation into an E. coli ada strain or transfection and stable integration into the genome of Chinese hamster ovary (CHO) cells, plasmid vectors containing the whole ada+ gene conferred resistance to both killing and mutagenesis by N-methyl-N'-nitro-N-nitrosoguanidine (MNNG). Thus, the bacterial DNA repair gene was functionally expressed in the mammalian cells. Plasmids containing an N-terminal fragment of the ada+ gene which encoded only one of the two methyltransferase activities of the Ada protein did not significantly protect E. coli or CHO cells against MNNG. These results are consistent with the central role of the intact ada+ gene in controlling the adaptive response to alkylating agents in E. coli. However, the data further suggest that some alkylation lesions in DNA, such as O6-methylguanine, may exert partly different biological effects in E. coli and mammalian cells.     

Cell-cycle-dependent repair of damage in alpha and bulk DNA of monkey cells

Excision repair of bulky chemical adducts in alpha DNA of confluent cultures of African green monkey cells has previously been shown to be deficient relative to that in the overall genome. We have compared the removal of adducts produced by treatment with aflatoxin B1 (AFB1) and N-acetoxy-2-acetylamino-fluorene (NA-AAF) from alpha DNA sequences in synchronized and exponentially growing cultures of monkey cells. Proficient removal of AFB1 adducts in alpha DNA was observed in exponentially growing cultures. However, as the cultures approached confluence, adduct removal from alpha DNA became deficient. Cells synchronized by subculturing confluent cultures exhibited proficient removal of adducts from both alpha and bulk DNA when treated in early G1 or late S/G2 while those cells treated in early S phase did not remove adducts from either alpha or bulk DNA. We conclude that the accessibility of chemical adducts to repair in alpha chromatin is influenced by the growth state and the cell cycle stage.     

Restricted repair of aflatoxin B1 induced damage in alpha DNA of monkey cells

We have investigated the processing of adducts formed by covalent binding of aflatoxin B1 (AFB1) to DNA in confluent cultures of African green monkey cells. Repair synthesis elicited by AFB1 adducts was deficient in alpha DNA sequences compared to that in bulk DNA, although the initial levels of modification were the same for these DNAs. The removal of the primary initial adduct, AFB1-N7-Guanine, was deficient in alpha DNA and the kinetics of its loss resembled those previously reported for removal from total DNA in xeroderma pigmentosum cells of complementation group A. Spontaneous loss of the AFB1 moiety or the concomitant loss of the guanine to yield an apurinic site account for these results. The formation of the more chemically stable secondary product, AFB1-triamino-Pyrimidine, occurred more rapidly and to a greater extent in alpha DNA than in bulk DNA, probably because of slower removal of the primary product. The excision repair patch size for AFB1 adducts in alpha DNA was only 10 nucleotides compared to 20 nucleotides for repair of AFB1 adducts in bulk DNA. Irradiation of cells with low doses of UV prior to or immediately after treatment with AFB1 increased the rate and extent of removal of AFB1 adducts from alpha DNA to the levels found in the bulk DNA, indicating that the formation of pyrimidine dimers or their repair may alter the chromatin structure of alpha DNA sufficiently to facilitate its repair.     

Characterization of gene-specific DNA repair by primary cultures of rat hepatocytes

At present, almost all the information on gene-specific DNA repair in mammals comes from studies with transformed cell lines and proliferating primary cells obtained from rodents and humans. In the present study, we measured the repair of specific DNA regions in primary cultures of nondividing rat hepatocytes (parenchymal cells). DNA damage was induced by irradiating the primary cultures of hepatocytes with ultraviolet (UV) light, and the presence of cyclobutane pyrimidine dimers (CPDs) was measured by using T4 endonuclease V in the following: a 21-kb BamHI fragment containing the albumin gene, a 14-kb BamHI fragment containing the H-ras gene, and the genome overall. The frequency of CPDs in the two BamHI fragments and the genome overall were similar and ranged from 0.5 to 1.3 CPDs per 10 kb for UV doses of 5–30 J/m². However, the removal of CPDs from the DNA fragment containing the albumin gene was significantly higher than from that of the genome overall and the DNA fragment containing the H-ras gene. Within 24 hr, approximately 67% of the CPDs was removed from the DNA fragment containing the albumin gene versus less than 40% for the genome overall and the DNA fragment containing the H-ras gene. The lower repair observed for the 14-kb fragment containing the H-ras gene is probably indicative of repair of the nontranscribed region of this fragment because the H-ras gene makes up only 2.4 kb of the 14-kb fragment. Primary cultures of hepatocytes removed CPDs from the transcribed strand of albumin fragment more efficiently than from the nontranscribed strand; however, no differences were observed in the repair of the two strands of the fragment containing the H-ras gene. These results demonstrate that primary cultures of nondividing rat hepatocytes show differential repair of UV-induced DNA damage that is comparable to what has been reported for transformed, proliferating mammalian cell lines. J. Cell. Physiol. 176:314–322, 1998. © 1998 Wiley-Liss, Inc.     

Cloning of the DNA repair genes mtcA, mtcB, uvsC, uvsD, uvsE and the leuB gene from Deinococcus radiodurans

A gene library from Deinococcus radiodurans has been constructed in the cosmid pJBFH. A 51.5-kb hybrid cosmid, pUE40, that transduced Escherichia coli HB101 from leucine dependence to independence was selected, and a 6.9-kb fragment which carried the leuB gene from D. radiodurans was subcloned into the EcoRI site of pAT153. The DNA repair genes mtcA, mtcB, uvsC, uvsD and uvsE, which code for two D. radiodurans UV endonucleases were identified by transforming appropriate repair-deficient mutants of D. radiodurans to repair proficiency with DNA derived from the gene library. Hybrid cosmid pUE50 (37.9 kb) containing an insert carrying both the mtcA and mtcB genes was selected and 5.6- and 2.7-kb DNA fragments carrying mtcA and mtcB, respectively, i.e., the genes that code for UV endonuclease alpha, were subcloned into the EcoRI site of pAT153. The three genes uvsC, uvsD and uvsE, that code for UV endonuclease beta, were all present in the 46.0-kb hybrid cosmid pUE60. The uvsE gene in a 12.2-kb fragment was subcloned into the HindIII site of pAT153 and the size of the insert reduced to 6.1 kb by deletion of a 6.7-kb fragment from the hybrid plasmid pUE62. None of the uvs genes introduced into the UV-sensitive E. coli CSR603 (uvrA-) was able to complement its repair defect. The mtcA, uvsC, uvsD and uvsE genes were found in the 52.5-kb hybrid cosmid pUE70. It is concluded that the DNA repair genes mtcA, mtcB, uvsC, uvsD and uvsE are located within an 83.0-kb fragment of the D. radiodurans genome.     

DNA repair in an active gene: removal of pyrimidine dimers from the DHFR gene of CHO cells is much more efficient than in the genome overall

DNA repair was measured in the dihydrofolate reductase gene in Chinese hamster ovary cells, amplified for the gene, by quantitating pyrimidine dimers with a specific UV-endonuclease. More than two thirds of the dimers had been removed from a 14.1 kb restriction fragment of the gene by 26 hr after irradiation (20 J/m2), while little removal was detected in fragments upstream of the gene and only 15% were removed from the genome overall. This suggests that damage processing can vary according to function or activity of affected sequences, which has general implications for correlations of DNA repair with survival and mutagenesis. Perhaps preferential repair of vital sequences facilitates UV-resistance of these cells despite low overall repair levels.     

High spontaneous mutation frequency in shuttle vector sequences recovered from mammalian cellular DNA.

The recombinant shuttle vector pSV2gpt was introduced into V79 Chinese hamster cells, and stable transformants expressing the Escherichia coli gpt gene were selected. Two transformants carrying tandem duplications of the plasmid at a single site were identified and fused to simian COS-1 cells. Plasmid DNA recovered from the heterokaryons was used to transform a Gpt- derivative of E. coli HB101, and the relative frequency of plasmids carrying a mutation in the gpt gene was determined. The high frequency of Gpt- plasmids (ca. 1%) was similar to that observed when plasmid was recovered from COS-1 cells which had been transfected with pSV2gpt. Most of the mutant plasmids had rearrangements in the region containing the gpt gene.     

Escherichia coli gpt gene provides dominant selection for vaccinia virus open reading frame expression vectors.

Mycophenolic acid, an inhibitor of purine metabolism, was shown to block the replication of vaccinia virus in normal cell lines. This observation led to the development of a dominant one-step plaque selection system, based on expression of the Escherichia coli gpt gene, for the isolation of recombinant vaccinia viruses. Synthesis of xanthine-guanine phosphoribosyltransferase enabled only the recombinant viruses to form large plaques in a selective medium containing mycophenolic acid, xanthine, and hypoxanthine. To utilize the selection system efficiently, we constructed a series of plasmids that contain the E. coli gpt gene and allow insertion of foreign genes into multiple unique restriction endonuclease sites in all three reading frames between the translation initiation codon of a strong late promoter and synthetic translation termination sequences. The selection-expression cassette is flanked by vaccinia virus DNA that directs homologous recombination into the virus genome. The new vectors allow high-level expression of complete or partial open reading frames and rapid construction of recombinant viruses by facilitating the cloning steps and by simplifying their isolation. The system was tested by cloning the E. coli beta-galactosidase gene; in 24 h, this enzyme accounted for approximately 3.5% of the total infected-cell protein.     

The expression of the Escherichia coli uvrA gene in human cells

Cells cultured from xeroderma pigmentosum (XP) patients are defective in excision repair of damaged DNA specifically at the incision step. In Escherichia coli this step is mediated by the UvrA, UvrB and UvrC gene products. Our goal is to express each of these genes in XP cells, singly or in combination, and to determine the most suitable conditions for generating faithful E. coli Uvr protein copies in functional concentrations and properly localized for the eventual repair of damaged chromosomal DNA or DNA which is introduced exogenously. The E. coli gpt gene in pSV2gpt is used as a selection marker for uvr gene transfection into XP cells. The uvr genes were cloned into composite pBR322, SV40 and gpt vectors in which each E. coli gene is flanked by individual SV40 regulatory elements. SV40-transformed XP-A cells were transfected with pSV2uvrASV2gpt, gpt+ colonies were selected, and cell lines established. Several lines were examined in detail. Cell lines 714 and 1511 contain uvrA together with flanking SV40 regulatory elements integrated intact in genomic DNA and express UvrA protein as well as a 95,000-dalton UvrA-related protein. The expression of uvrA was found to be 50-100-fold lower than the expression of gpt. Attempts were made to assay the mammalian UvrA protein for functionality, but endogenous activities interfered with assays for each of the UvrA protein's three activities. The peptide maps derived from partial proteolysis of the "mammalian" UvrA protein are identical to the E. coli UvrA protein. The sub-cellular location of UvrA protein in uvrA+ XP cells was investigated by fractionation of cell extracts in which an indirect immunofluorescence method revealed its location as being largely extra-nuclear. Two uvrA+ cell lines were examined for their UV-resistant phenotype and not unexpectedly were found not to be reverted to a state of repair proficiency.     

Introduction and recovery of a selectable bacterial gene from the genome of mammalian cells.

The simian virus 40 (SV40)-pBR322 recombinant, pSV2, carrying the origin of SV40 replication and the gpt gene of Escherichia coli, has been stably introduced into Chinese hamster ovary hprt- cells. All gpt-transformed cell lines were found to contain one or more insertions of pSV2 sequences exclusively associated with high-molecular-weight DNA. Additional analyses showed that at least one integrated copy in each cell line retained an intact gpt gene and flanking SV40 sequences required for expression of xanthine-guanine phosphoribosyltransferase. Most cell lines contained pSV2 sequences which had integrated with partial sequence duplication. Upon fusion with COS-1 cells, a simian cell line permissive for autonomous pSV2 replication, most gpt-transformed cell lines produced low-molecular-weight DNA molecules related to pSV2. The majority of these replicating DNAs were indistinguishable from the original transfecting plasmid in both size and restriction enzyme cleavage pattern. In addition, the recovered DNA molecules were able to confer ampicillin resistance to E. coli and to transform mouse L cells and Gpt- E. coli to a Gpt+ phenotype. These studies indicate that all of the genetic information carried by this SV40-plasmid recombinant can be introduced into and retrieved from the genome of mammalian cells.     

tert-Butoxyl radicals generate mainly 7,8-dihydro-8-oxoguanine in DNA

Like hydroxyl radicals, alkoxyl radicals have been implicated in the generation of cellular oxidative DNA damage under physiological conditions; however, their genotoxic potential has not yet been established. We have analyzed the DNA damage induced by a photochemical source of tert-butoxyl radicals, the water soluble peroxy ester [4-(tert-butyldioxycarbonyl)benzyl]triethylammonium chloride (BCBT), using various repair endonucleases as probes. The irradiation (UV(360)) of BCBT in the presence of bacteriophage PM2 DNA was found to generate a DNA damage profile that consisted mostly of base modifications sensitive to the repair endonuclease Fpg protein. Approximately 90% of the modifications were identified as 7,8-dihydro-8-oxoguanine (8-oxoGua) residues by HPLC/ECD analysis. Oxidative pyrimidine modifications (sensitive to endonuclease III), sites of base loss (AP sites) and single-strand breaks were only minor modifications. Experiments with various scavengers and quenchers indicated that the DNA damage by BCBT+UV(360) was caused by tert-butoxyl radicals as the ultimate reactive species. The mutagenicity associated with the induced damage was analyzed in the gpt gene of plasmid pSV2gpt, which was exposed to BCBT+UV(360) and subsequently transfected into Escherichia coli. The results were in agreement with the specific generation of 8-oxoGua. Nearly all point mutations (20 out of 21) were found to be GC-->TA transversions known to be characteristic for 8-oxoGua. In conclusion, alkoxyl radicals generated from BCBT+UV(360) induce 8-oxoGua in DNA with a higher selectivity than any other reactive oxygen species analyzed so far.     

The use of antibodies to 5-bromo-2'-deoxyuridine for the isolation of DNA sequences containing excision-repair sites

We have developed an immunological method for isolation and identification of DNA sequences containing 5-bromo-2'-deoxyuridine (BrdUrd) incorporated during UV-induced excision-repair synthesis. DNA fragments containing BrdUrd incorporated during repair synthesis were incubated with goat anti-BrdUrd and rabbit anti-goat IgG, and the antibody-DNA complexes were separated from bulk DNA by nitrocellulose filter binding. With this method, 80% of DNA sequences containing BrdUrd-labeled excision-repair sites were recovered, contaminated with less than 1% of DNA fragments devoid of excision-repair sites. Recovery of DNA fragments containing repair sites was independent of size from 2 to 20 kilobases. We have used this method in conjunction with blot hybridization to demonstrate that repair synthesis occurs in human ribosomal gene sequences in cells treated with UV.     

Molecular cloning and biological characterization of a human gene, ERCC2, that corrects the nucleotide excision repair defect in CHO UV5 cells.

The UV-sensitive Chinese hamster ovary (CHO) cell line UV5, which is defective in the incision step of nucleotide excision repair, was used to identify and clone a complementing human gene, ERCC2, and to study the repair process. Genomic DNA from a human-hamster hybrid cell line was sheared and cotransferred with pSV2gpt plasmid DNA into UV5 cells to obtain five primary transformants. Transfer of sheared DNA from one primary transformant resulted in a secondary transformant expressing both gpt and ERCC2. The human repair gene was identified with a probe for Alu-family repetitive sequences. For most primary, secondary, and cosmid transformants, survival after UV exposure showed a return to wild-type levels of resistance. The levels of UV-induced mutation at the aprt locus for secondary and cosmid transformants varied from 50 to 130% of the wild-type level. Measurements of the initial rate of UV-induced strand incision by alkaline elution indicated that, whereas the UV5 rate was 3% of the wild-type level, rates of cosmid-transformed lines were similar to that of the wild type, and the secondary transformant rate was about 165% of the wild-type rate. Analysis of overlapping cosmids determined that ERCC2 is between 15.5 and 20 kilobases and identified a closely linked gpt gene. Cosmids were obtained with functional copies of both ERCC2 and gpt. ERCC2 corrects only the first of the five CHO complementation groups of incision-defective mutants.     

Expression of adenovirus E1a and E1b gene products and the Escherichia coli XGPRT gene in KB cells

The recombinant plasmid pSV2-gpt, which contains the Escherichia coli XGPRT gene under the control of a simian virus 40 early promoter, was modified to contain the type 2 adenovirus (Ad2) XhoI-C (0 to 15.5 map units) restriction endonuclease fragment. Plasmid (pLB206) DNA was introduced into human KB cells by Ca2+-mediated DNA transfection, and transformants were selected in medium containing xanthine, aminopterin, and mycophenolic acid, as a consequence of expression of the dominant, selectable XGPRT gene. A series of 13 gpt+ cell lines were isolated and tested for their ability to complement Ad5 deletion mutants in E1a (H5dl312) and E1b (H5dl315). Four classes of gpt+ KB cell lines were identified, including clones constitutively expressing both E1a and E1b, only E1a, or only E1b or not expressing either E1a or E1b. DNA and RNA filter transfer hybridization analysis substantiated the conclusions that those cell lines capable of complementing viral host range mutants contained the appropriate viral DNA sequences and cytoplasmic polyadenylated RNA species. DNA filter transfer hybridization studies also revealed that the transfected vector DNA was stably integrated into chromosomal DNA in the KB transformants and the number of integrated sites ranged from 1 to 3. The gpt+ KB cell line that only expressed E1b gene functions only contained viral E1b gene sequences; those cell lines that expressed neither E1a nor E1b gene function contained only small or no regions of Ad2 DNA. When weaned off the selective medium, transformed KB cell lines stably maintained their inserted DNA in the absence of selective pressure and could easily be adapted to growth in suspension culture.