Effect of excision repair by diploid human fibroblasts on the kinds and locations of mutations induced by (+/-)-7 beta,8 alpha-dihydroxy-9 alpha,10 alpha-epoxy-7,8,9,10- tetrahydrobenzo[a]pyrene in the coding region of the HPRT gene.

(+/-)-7 beta,8 alpha-Dihydroxy-9 alpha,10 alpha-epoxy-7,8,9,10- tetrahydrobenzo[a]pyrene (BPDE) is a direct-acting carcinogen that forms DNA adducts only with purines, predominantly (greater than 95%) with guanine. To investigate the effect of nucleotide excision repair on the kinds and locations (spectra) of mutations induced in diploid human fibroblasts by BPDE, we synchronized cells and exposed them to BPDE either at the beginning of S phase just when the target gene hypoxanthine (guanine) phosphoribosyltransferase (HPRT) is replicated or 12 hr prior to the beginning of S phase (early G1 phase). Clones resistant to 6-thioguanine were isolated, and the mRNA in lysates of 100-500 cells from each mutant clone was used to synthesize cDNA. HPRT cDNA was amplified 10(11)-fold by the polymerase chain reaction and then sequenced directly. The mutants derived from the two populations did not differ in the kinds of mutations; 19/20 of the base substitutions in cells taken from S phase and 19/19 of those from G1 phase involved G.C base pairs, predominantly G.C----T.A. However, they differed significantly in the distribution of the mutations in the coding region of the gene. In the cells from G1 phase, 29% of the mutations were clustered within a unique run of six guanine bases; in the S-phase cells, only 4% were located there. Assuming that the premutagenic BPDE-induced lesions involved purines, in the cells treated at the beginning of S phase, 24% of these lesions were located in the transcribed strand, whereas in the G1-treated cells, none were. This suggests that in the HPRT gene of diploid human cells excision repair of BPDE adducts occurs preferentially on the transcribed strand.     

Multiple point mutations in a shuttle vector propagated in human cells: evidence for an error-prone DNA polymerase activity.

Mutagenesis was studied at the DNA-sequence level in human fibroblast and lymphoid cells by use of a shuttle vector plasmid, pZ189, containing a suppressor tRNA marker gene. In a series of experiments, 62 plasmids were recovered that had two to six base substitutions in the 160-base-pair marker gene. Approximately 20-30% of the mutant plasmids that were recovered after passing ultraviolet-treated pZ189 through a repair-proficient human fibroblast line contained these multiple mutations. In contrast, passage of ultraviolet-treated pZ189 through an excision-repair-deficient (xeroderma pigmentosum) line yielded only 2% multiple base substitution mutants. Introducing a single-strand nick in otherwise unmodified pZ189 adjacent to the marker, followed by passage through the xeroderma pigmentosum cells, resulted in about 66% multiple base substitution mutants. The multiple mutations were found in a 160-base-pair region containing the marker gene but were rarely found in an adjacent 170-base-pair region. Passing ultraviolet-treated or nicked pZ189 through a repair-proficient human B-cell line also yielded multiple base substitution mutations in 20-33% of the mutant plasmids. An explanation for these multiple mutations is that they were generated by an error-prone polymerase while filling gaps. These mutations share many of the properties displayed by mutations in the immunoglobulin hypervariable regions.     

Site-specific rates of excision repair of benzo[a]pyrene diol epoxide adducts in the hypoxanthine phosphoribosyltransferase gene of human fibroblasts: correlation with mutation spectra.

When populations of repair-proficient diploid human fibroblasts were treated with (+/-)-7 beta, 8 alpha-dihydroxy-9 alpha, 10 alpha-epoxy-7,8,9,10-tetrahydrobenzo[a]pyrene (BPDE) during early S phase, just as the hypoxanthine phosphoribosyltransferase gene (HPRT) was being replicated, 5% of the induced base substitutions were found at nt 212, and 5% of the substitutions were found at nt 229 in exon 3. However, when the population was treated in early G1 phase to allow at least 12 hr for repair before the onset of S phase, 21% of the substitutions were found at nt 212, and 10% were found at nt 229. No such cell-cycle-dependent difference in distribution of base substitutions occurred in excision-repair-deficient cells. To test whether the increase in the relative frequency of mutations resulted from inefficient repair at these sites, we adapted ligation-mediated PCR to measure the rates of removal of BPDE adducts from individual sites in exon 3 of the HPRT gene. Cells were treated with 0.5 microM BPDE in early G1 phase and harvested immediately or after 10, 20, and 30 hr for repair. the nontranscribed strand of exon 3 was analyzed for the original distribution of adducts and those remaining after repair, using Escherichia coli UvrABC excinuclease to excise the adducts and annealing a 5' biotinylated gene-specific primer to the DNA and extending it with Sequenase 2.0 to generate a blunt end at the site of each cut. A linker was ligated to the blunt end, and the desired fragments were isolated from the rest of the genomic DNA by using magnetic beads, amplified by PCR, and analyzed on a sequencing gel. The distribution of fragments of particular lengths indicated the relative number of BPDE adducts initially formed or remaining at specific sites. The rates of repair at individual sites varied widely along exon 3 of the HPRT gene and were very slow at nt 212 and 229, strongly supporting the hypothesis that inefficient DNA repair plays an important role in the formation of mutation hotspots.     

Replication of acetylaminofluorene-adducted plasmids in human cells: spectrum of base substitutions and evidence of excision repair.

In rats fed the liver carcinogen 2-acetylaminofluorene (AAF), the two most abundant types of DNA adduct are N-(deoxyguanosin-8-yl)-2-acetylaminofluorene and its deacetylated derivative. When plasmids carrying AAF adducts replicate in bacteria, the predominant mutations are frameshifts, whereas with deacetylated (AF) adducts, they are mainly base substitutions, just as we found when plasmids carrying AF adducts replicated in human cells. We have investigated the frequency and spectrum of mutations induced when a shuttle vector carrying AAF adducts (85% bound to the C8 position of guanine, 15% to the N2 position) replicated in human cells. The frequency induced per initial AAF adduct was higher than with AF adducts, but the kinds of mutations were similar--i.e., 85% base substitutions, principally G.C----T.A transversions. There was good correlation between the "hot spots" for mutations and hot spots for AAF adduct formation, suggesting that mutational hot spots reflect preferential binding of the carcinogen to DNA. 32P-postlabeling analysis of the adducts before and after the DNA was transfected into the human cells showed that there was no deacetylation of AAF adducts and that 85% of both types of adducts were removed within 3.5 hr, most probably by excision repair.     

Restricted ultraviolet mutational spectrum in a shuttle vector propagated in xeroderma pigmentosum cells.

A shuttle vector plasmid, pZ189, carrying a bacterial suppressor tRNA marker gene, was treated with ultraviolet radiation and propagated in cultured skin cells from a patient with the skin-cancer-prone, DNA repair-deficient disease xeroderma pigmentosum and in repair-proficient cells. After replication in the human cells, progeny plasmids were purified. Plasmid survival and mutations inactivating the marker gene were scored by transforming an indicator strain of Escherichia coli carrying a suppressible amber mutation in the beta-galactosidase gene. Plasmid survival in the xeroderma pigmentosum cells was less than that of pZ189 harvested from repair-proficient human cells. The point-mutation frequency in the 150-base-pair tRNA marker gene increased up to 100-fold with ultraviolet dose. Sequence analysis of 150 mutant plasmids revealed that mutations were infrequent at potential thymine-thymine dimer sites. Ninety-three percent of the mutant plasmids from the xeroderma pigmentosum cells showed G X C----A X T transitions, compared to 73% in the normal cells (P less than 0.002). There were significantly fewer transversions (P less than 0.002) (especially G X C----T X A) and multiple base substitutions (P less than 0.00001) than when pZ189 was passaged in repair-proficient cells. The subset of mutational changes that are common to ultraviolet-treated plasmids propagated in both repair-proficient and xeroderma pigmentosum skin cells may be associated with the development of ultraviolet-induced skin cancer in humans.     

Xeroderma pigmentosum variant cells are less likely than normal cells to incorporate dAMP opposite photoproducts during replication of UV-irradiated plasmids.

Xeroderma pigmentosum (XP) variant patients show the clinical characteristics of the disease, with increased frequencies of skin cancer, but their cells have a normal, or nearly normal, rate of nucleotide excision repair of UV-induced DNA damage and are only slightly more sensitive than normal cells to the cytotoxic effect of UV radiation. However, they are significantly more sensitive to its mutagenic effect. To examine the mechanisms responsible for this hypermutability, we transfected an XP variant cell line with a UV-irradiated (at 254 nm) shuttle vector carrying the supF gene as a target for mutations, allowed replication of the plasmid, determined the frequency and spectrum of mutations induced, and compared the results with those obtained previously when irradiated plasmids carrying the same target gene replicated in a normal cell line [Bredberg, A., Kraemer, K. H. & Seidman, M. M. (1986) Proc. Natl. Acad. Sci. USA 83, 8273-8277]. The frequency of mutants increased linearly with dose, but with a slope 5 times steeper than that seen with normal cells. Sequence analysis of the supF gene showed that 52 of 53 independent mutants generated in the XP variant cells contained base substitutions, with 62 of 64 of the substitutions involving a dipyrimidine. Twenty-eight percent of the mutations involved A.T base pairs, with the majority found at position 136, the middle of a run of three A.T base pairs. (In the normal cells, this value was only 11%.) If the rate of excision of lesions from supF in the two cell lines is equal, our data suggest that XP variant cells are less likely than normal cells to incorporate dAMP opposite bases involved in photo-products. If such incorporation also occurs during replication of chromosomal DNA, this could account for the hypermutability of XP variant cells with UV irradiation.     

Carcinogenic epoxides of benzo[a]pyrene and cyclopenta[cd]pyrene induce base substitutions via specific transversions.

We have determined the spectrum of base-pair substitution mutations induced in the lacI gene of a uvrB- strain of Escherichia coli by two polycyclic aromatic hydrocarbons--(+/-)7 alpha,8 beta-dihydroxy-9 beta,10 beta-epoxy-7,8,9,10 tetrahydrobenzo[a]pyrene (BPDE), and 3,4-epoxycylopenta[cd]pyrene (CPPE). Approximately 10% of all lacI mutations induced by either BPDE or CPPE are nonsense mutations, suggesting that base-pair substitutions are a large fraction of the mutational events induced by these agents in the uvrB- bacteria. Both carcinogens specifically induced the G . C leads to T . A and, to a lesser extent, the A . T leads to T . A transversions. One possible mechanism for transversion induction at G . C sites by BPDE might involve carcinogen binding to the exocyclic amino group of guanine in the template strand followed by a rotation of the modified base around its glycosylic bond from the anti to the syn conformation. This could allow specific pairing of modified bases with an imino tautomer of adenine.     

Mutations activating human c-Ha-ras1 protooncogene (HRAS1) induced by chemical carcinogens and depurination.

In vitro modification of plasmids containing the human c-Ha-ras1 protooncogene (HRAS1) with the ultimate carcinogens N-acetoxy-2-acetylaminofluorene and r-7, t-8-dihydroxy-t-9, 10-epoxy-7,8,9,10-tetrahydrobenzo[alpha]pyrene (anti-BPDE) generated a transforming oncogene when the modified DNA was transfected into NIH 3T3 cells. The protooncogene was also activated by heating the plasmid at 70 degrees C, pH 4, to generate apurinic/apyrimidinic sites in the DNA. DNA isolated from transformed foci was analyzed by hybridization with 20-mer oligonucleotides designed to detect single point mutations within two regions of the gene commonly found to be mutated in tumor DNA. Of 23 transformants studied, 7 contained a mutation in the region of the 12th codon, whereas the remaining 16 were mutated in the 61st codon. Of the codon-61 mutants, 6 were mutated at the first base position (C X G), 5 at the second (A X T), and 5 at the third (G X C). The point mutations induced by anti-BPDE were predominantly G X C----T X A and A X T----T X A base substitutions, whereas four N-acetoxy-2-acetylaminofluorene-induced mutations were all G X C----T X A, and a single depurination-induced activation that was analyzed contained an A X T----T X A transversion. Together, these methods provide a useful means of determining point mutations produced by DNA-damaging agents in mammalian cells.     

Base substitutions, frameshifts, and small deletions constitute ionizing radiation-induced point mutations in mammalian cells.

The relative role of point mutations and large genomic rearrangements in ionizing radiation-induced mutagenesis has been an issue of long-standing interest. Recent studies using Southern blotting analysis permit the partitioning of ionizing radiation-induced mutagenesis in mammalian cells into detectable deletions and major genomic rearrangements and into point mutations. The molecular nature of these point mutations has been left unresolved; they may include base substitutions as well as small deletions, insertions, and frame-shifts below the level of resolution of Southern blotting analysis. In this investigation, we have characterized a collection of ionizing radiation-induced point mutations at the endogenous adenine phosphoribosyltransferase (aprt) locus of Chinese hamster ovary cells at the DNA sequence level. Base substitutions represented approximately equal to 2/3 of the point mutations analyzed. Although the collection of mutants is relatively small, every possible type of base substitution event has been recovered. These mutations are well distributed throughout the coding sequence with only one multiple occurrence. Small deletions represented the remainder of characterized mutants; no insertions have been observed. Sequence-directed mechanisms mediated by direct repeats could account for some of the observed deletions, while others appear to be directly attributable to radiation-induced strand breakage.     

Mutagenic specificity of a potent carcinogen, benzo[c]phenanthrene (4R,3S)-dihydrodiol (2S,1R)-epoxide, which reacts with adenine and guanine in DNA.

Mutations were induced in the supF gene of the pS189 shuttle vector by treatment with optically active benzo[c]phenanthrene (4R,3S)-dihydrodiol (2S,1R)-epoxide in vitro and replication in human cells. The induced mutation frequency was 60-fold greater than the spontaneous rate, and most of the mutations analyzed were transversions (86%), which principally consisted of similar numbers of A.T----T.A and G.C----T.A changes. The unusual susceptibility of A.T pairs to mutation by this chemical agent is consistent with its chemical reactivity toward adenine and argues that the mutations are targeted to the adducts formed. The central base in the sequences 5'-AGA-3', 5'-AAC-3', and 5'-GAG-3' was particularly susceptible to mutation. Twelve "hotspots" in the supF gene accounted for most mutations seen. Some of these hotspots differed from those found by others for racemic benzo[a]pyrene dihydrodiol epoxide and, even when a hotspot was common, the mutagenic changes were not always the same. Although adenine insertion opposite a noninstructional lesion could account for most of the data, no single mutagenic mechanism could encompass all of it. The cellular machinery that converts chemical damage to mutations must determine the mutational result to a large extent, but the findings herein show that the chemical agent itself plays a large role in determining both the location and the nature of the mutations that arise.     

Stimulation of recombination between homologous sequences on plasmid DNA and chromosomal DNA in Escherichia coli by N-acetoxy-2-acetylaminofluorene.

A plasmid containing a wild-type lac operon and a tetracycline-resistance gene was covalently modified by N-acetoxy-2-acetylaminofluorene and used to transform two series of Lac- Escherichia coli cell types. Each set contained wild-type and repair-deficient mutants. One set of cells contained a lacY mutation and the other a deletion of the entire lac operon. Survival and mutagenesis of the plasmid were measured as a function of the N-acetoxy-2-acetylaminofluorene concentration. The results indicate that when no homologous sequences are present in the chromosomal DNA, mutations occur at a low frequency: at 10% survival the frequency was 1-2 X 10(-4) mutants per transformant. When homologous sequences, the lacY allele, are present in the chromosomal DNA, Lac- plasmids are found at a high frequency in a recA-dependent, lexA-independent fashion: at 10% survival the frequency was 5-10 X 10(-2) mutants per transformant. Southern blot analysis of the restriction enzyme profiles of the resulting plasmid and host-cell DNA sequences showed recombinational transfer of host sequences to the N-acetoxy-2-acetylamino-fluorene-treated plasmid had occurred. When the host chromosomes contained Lac+ homologous sequences no mutants were found, indicating that the results were not caused by error-prone recombination.     

Sodium azide mutagenesis: preferential generation of A.T-->G.C transitions in the barley Ant18 gene.

The molecular basis for the absence of anthocyanins and proanthocyanidins in four independent sodium azide-induced ant18 mutants of barley was examined by sequencing the gene encoding dihydroflavonol 4-reductase in these mutants. Sodium azide generated 21 base substitutions, which corresponds to 0.17% of the 12,704 nucleotides sequenced. Of the substitutions, 86% were nucleotide transitions, and 14% were transversions. A.T-->G.C base pair transitions were about 3 times more frequent than G.C-->A.T transitions. No deletions or mutation hot spots were found. The absence of dihydroflavonol 4-reductase activity in ant18-159, ant18-162, and ant18-164 plants is caused by missense mutations in the respective genes. By using microprojectile bombardment, a plasmid harboring the wild-type Ant18 gene was introduced into ant18-161 mutant cells and resulted in the development of anthocyanin pigmentation, which demonstrates that the mutation is corrected by expression of the introduced gene. On the other hand, a plasmid derivative with the two ant18-161-specific base transitions at the 5' splice site of intron 3 prevented complementation. It is concluded that the absence of detectable mRNA for dihydroflavonol 4-reductase in ant18-161 cells is due to the mutations in the pre-mRNA splice donor site.     

Construction of mutants of Moloney murine leukemia virus by suppressor-linker insertional mutagenesis: positions of viable insertion mutations.

A highly efficient method for the generation of insertion mutations is described. The procedure involves the use of a 220-base-pair (bp) EcoRI fragment bearing the SuIII+ suppressor tRNA gene as an insertional mutagen. The plasmid DNA to be mutagenized is linearized by a variety of means, and the suppressor fragment is ligated into the site of cleavage. Successful insertion mutants can be readily detected in Escherichia coli carrying lac- amber mutations on MacConkey lactose plates; virtually 100% of the red colonies contain insertions of the fragment. Subsequent removal of the SuIII+ gene and recyclization leaves a 12-bp insertion if the original cleavage was blunt-ended and a 9-bp insertion if the original cleavage generated 3-bp cohesive termini. This technique, as well as conventional linker mutagenesis with decamer and dodecamer linkers, was used to generate a large library of insertion mutations in cloned DNA copies of the genome of Moloney murine leukemia virus. A number of viable mutants were isolated bearing 9-, 10-, and 12-bp insertions in various domains of the genome. The map positions of the viable mutations suggest that the viral long terminal repeats and portions of the gag and env genes are quite insensitive to alteration. Although most of the mutations were stable for many passages, some of the mutants lost the inserted DNA; we presume that the insertion was somewhat deleterious in these mutants and that continued passage of the virus selected for overgrowth by a revertant.     

DNA base changes and alkylation following in vivo exposure of Escherichia coli to N-methyl-N-nitrosourea or N-ethyl-N-nitrosourea.

Dideoxy chain-termination DNA sequencing was used to determine the specific DNA base changes induced after in vivo exposure of Escherichia coli to N-methyl-N-nitrosourea (MNU) and N-ethyl-N-nitrosourea (ENU) using the xanthine guanine phosphoribosyltransferase (gpt) gene as the genetic target. The resultant mutation spectra were compared with the levels of O6-alkylguanine and O4-alkylthymidine in genomic DNA immediately after exposure. All (39/39) of the MNU-induced mutations were G X C----A X T transitions. In contrast, 24/33 point mutations isolated following ENU treatment were G X C----A X T transitions, 7/33 were A X T----G X C transitions, 1/33 was a G X C----C X G transversion, and 1/33 was an A X T----C X G transversion. Three large insertions, probably of spontaneous origin, were also isolated. O4-alkylthymidine/O6-alkylguanine ratios were 0.014 for MNU and 0.28 for ENU. These data suggest that the difference in the mutation spectrum of MNU versus ENU may be attributed, in part, to the different ratio of O6-alkylguanine versus O4-alkylthymidine produced in the DNA. Of the G X C----A X T transitions, 82% of the MNU- and 71% of the ENU-induced mutations occurred at the middle guanine of the sequence 5'-GG(A or T)-3'.     

Single adduct mutagenesis: strong effect of the position of a single acetylaminofluorene adduct within a mutation hot spot.

2-Acetylaminofluorene (AAF), a potent rat liver carcinogen that binds covalently to the C-8 position of guanine residues in DNA, is an effective frameshift mutagen. The mutations are distributed nonrandomly, in that most are located at a few specific DNA sequences (i.e., mutation hot spots). Among these hot spots, the Nar I sequence (GGCGCC) is especially susceptible to the induction of -2 frameshift mutations (GGCGCC----GGCC). Due to the nature of the Nar I sequence, G1G2CG3CC, three different molecular events, each involving the deletion of two contiguous base pairs (i.e., G2C, CG3, G3C), can give rise to the observed end point (GGCC). To compare the potential role of each of the three possible guanine-AAF adducts in the Nar I site to induce the -2 frameshift mutation, we constructed double-stranded plasmid molecules containing a single-AAF adduct bound to one of the three guanine positions. Using these plasmids, we found that only the adduct in the G3 position induces the -2 frameshift mutation. This strong effect of the position of the -AAF adduct within the Nar I site is discussed in relation to the possible involvement of an unusual DNA conformation in the mutagenic processing.     

Splicing mutations in the CHO DHFR gene preferentially induced by (+/-)-3 alpha,4 beta-dihydroxy-1 alpha,2 alpha-epoxy-1,2,3,4- tetrahydrobenzo[c]phenanthrene.

Cultured Chinese hamster ovary (CHO) cells were treated with the polycyclic aromatic hydrocarbon racemic 3 alpha,4 beta-dihydroxy-1 alpha,2 alpha-epoxy-1,2,3,4- tetrahydrobenzo[c]phenanthrene. Mutants deficient in dihydrofolate reductase activity were isolated. A carcinogen treatment at 0.1 microM yielded at 46% survival of the treated population and an induced frequency of mutation of 1.7 x 10(-4), 10(3)-fold greater than the spontaneous rate. By polymerase chain reaction amplification and direct DNA sequencing, we determined the base changes in 38 mutants. Base substitutions accounted for 78% (30/38) of the mutations. We obtained, in addition, four frameshift and four complex mutations. The preferred type of mutation was transversion (A.T----T.A and G.C----T.A) occurring in 69% of the analyzed mutants. A purine was on the 3' side of the putative adduct site in every mutant. Mutations were favored at sequences AGG, CAG, and AAG (the underlined base is the target). Surprisingly, 42% of the mutations created mRNA splicing defects (16/38), especially at splice acceptor sites for each of the five introns. Thus, this chemical carcinogen may recognize some aspect of DNA structure in regions corresponding to pre-mRNA splice sites.     

Spectrum of cisplatin-induced mutations in Escherichia coli.

Using a forward-mutation assay based on the inactivation of the tetracycline-resistance gene located on plasmid pBR322, we have determined the mutation spectrum induced in Escherichia coli by cisplatin [cis-diamminedichloroplatinum(II)], a widely used antitumor drug. Cisplatin is known to form mainly intrastrand diadducts at ApG and GpG sites. We found that cisplatin efficiently induces mutations in an SOS-dependent way (i.e., dependent upon UV irradiation of the host bacteria). More than 90% of the mutations are single-base-pair substitutions occurring at the potential sites of cisplatin adducts (ApG and GpG). Taking into account the relative proportions of ApG and GpG adducts, we found that the ApG adducts are at least 5 times more mutagenic than the GpG adducts. Moreover, a strong mutation specificity was seen at the 5' side of the ApG adducts (A X T----T X A transversions). The observation that most mutations occur at the 5' end of the adduct at both ApG and GpG sites is discussed in relation to recent structural data.     

Efficient recovery and sequencing of mutant genes from mammalian chromosomal DNA.

A retroviral shuttle vector was constructed by introducing the Escherichia coli xanthine (guanine) phosphoribosyltransferase gene (gpt) into the pZip-NeoSV(X)1 vector [Cepko, C. L., Roberts, B. E. & Mulligan, R. C. (1984) Cell 37, 1053-1062]. This vector was packaged into infectious virus which then was used to infect a hypoxanthine (guanine) phosphoribosyltransferase-deficient mouse cell line. Cell lines that expressed the gpt gene were isolated, and it was found that these cells contained a single integrated copy of the vector in a proviral form. Treatment of these cell lines with either ethyl methanesulfonate or BrdUrd produced a greater than 10-fold increase in the frequency of 6-thioguanine-resistant (Sgur) mutants. Intact gpt genes have been recovered from a number of Sgur cell lines after COS cell fusion and introduced into E. coli as part of a plasmid. The complete DNA sequences of three mutant genes have been determined. Two of the mutant genes have a single base substitution, whereas the third has a 34-base-pair deletion. This system should be valuable for analyzing mutagenic specificity and the molecular mechanisms of chemical mutagenesis in mammalian cells. A potentially important feature of the system relative to other shuttle-vector systems is that the mutations are induced in genes integrated into mammalian chromosomes rather than in genes existing as part of autonomously replicating plasmids.