MOLECULAR AND CELLULAR BIOLOGY, Feb. 1992, p. 767-772 Vol. 12, No. 2 0270-7306/92/020767-06$02.00/0 Copyright © 1992, American Society for Microbiology Linked Spontaneous CG->TA Mutations at CpG Sites in the Gene for Protein Kinase Regulatory Subunit ROBERT A. STEINBERG* AND KAREN B. GORMAN Department ofBiochemistry and Molecular Biology, University of Oklahoma Health Sciences Center, Oklahoma City, Oklahoma 73190 Received 11 March 1991/Accepted 19 November 1991 CG-*TA transitions at CpG sequences account for many human point mutations and are thought to result from hydrolytic deamination of 5-methylcytosine residues in these sites. The gene for regulatory subunit of murine cyclic AMP-dependent protein kinase has two closely linked CpG sites, one of which is a strong hotspot for spontaneous CG-+TA mutations leading to cyclic AMP resistance in S49 mouse lymphoma cells. About 5% of mutants with a spontaneous mutation at this CpG site had also acquired a second CG-oTA mutation at the nearby CpG site. The two mutations were always at first positions of the Arg codons in which they occurred, and they were always together in a single regulatory subunit allele. Their linked appearance could be attributed to neither the selection conditions nor the preexistence of one mutation in the target cells. The high frequency of these double mutants suggests that their lesions result not from hydrolytic deamination but rather from an endogenous enzymatic mechanism. Spontaneous point mutations are thought to arise from tion with one or two closely linked additional mutations (10). errors in replicative or repair synthesis of DNA, the chem- An additional mutation found in two such isolates was a ical effects of background ionizing radiation, and/or the CG-*TA transition in the codon for Arg-332 causing its chemical instability of normal or modified DNA bases (6, replacement with Cys. These two mutations at Arg codons 18). Their low frequency at any one allele has suggested that caused a diagnostic two-charge-unit acidic shift in the mu- the flux of DNA-damaging events is low and distributed tant R subunits that could be detected easily by two- throughout the genome. Transition mutations at CpG dinu- dimensional gel electrophoresis (16). Here we report the cleotide sequences account for about 25 to 40% of known presence of these two Arg codon lesions in a number of point mutations leading to human genetic disorders or cancer spontaneous cAMP-resistant isolates and provide evidence and are thought to arise by spontaneous hydrolytic deami- that both lesions arose spontaneously and simultaneously. nation of 5-methylcytosine (5-Me-C) residues, which occur The high frequency with which these lesions are found to be most frequently at CpG sites (3, 7, 13). Although it is linked in spontaneous mutants suggests that they result from impossible to know a priori whether these human mutations a high flux of mutagenic activity localized to a very small were spontaneous or induced by exposure to mutagens, genomic target region. It is clearly inconsistent with their support for the deamination pathway comes from a recent production by spontaneous hydrolytic deamination. report showing that CpG hotspots for mutations in the low-density lipoprotein receptor and the p53 tumor suppres- sor genes are indeed methylated in vivo (19). MATERIALS AND METHODS Mutants of the cyclic AMP (cAMP)-sensitive S49 mouse Cell culture and mutant isolation. Wild-type and mutant lymphoma cell line selected for resistance to cAMP analogs isolates of S49 subline U36 were grown in suspension culture have provided abundant material for study of missense in Dulbecco's modified Eagle's medium containing glucose mutations in a mammalian gene. The most common mutants (3 g/liter), sodium bicarbonate (2.24 g/liter), and 10o heat- have lesions in the regulatory (R) or cAMP-binding subunit inactivated horse serum as described previously (16). For of cAMP-dependent protein kinase that increase apparent cloning, medium was solidified by the addition of 0.3% constants (Kas) of the enzyme for cAMP-dependent activa- SeaPlaque agarose (FMC); where appropriate, the cAMP tion (16, 23). In a recent study of sequence changes under- analog N6,02-dibutyryl-cAMP (Bt2cAMP) or 8-(4-chlo- lying such lesions in S49 sublines hemizygous for expression rophenylthio)-cAMP (CPT-cAMP) was included at a final of mutant R subunits with altered charge, we found 8 distinct concentration of 0.5 or 0.05 mM, respectively (16, 23). single-base-change mutations clustered, for the most part, in For mutant isolations, populations of 0.6 x 106 to 2.0 x regions identified with the two cAMP-binding sites of R 106 cells were grown from single-cell-derived colonies of subunit, sites A and B (22); subsequently, we have identified subline U36 (see Table 2, footnote a) and the entire popula- 14 additional point mutations associated with Ka phenotypes tions were plated in the presence of cAMP analogs. Overall in hemizygous or heterozygous mutant cells (10). Among mutation rates were estimated from the distributions of spontaneous isolates, the most frequent mutation was a numbers of analog-resistant colonies per population by using CG-+TA transition in the site B region causing substitution published tables (14) after correction for control cloning ofTrp for Arg-334. Among mutants heterozygous for expres- efficiencies. Single isolates from each population containing sion of mutant R subunits, we found several mutagen- mutants were grown up under nonselective conditions and induced isolates that had this Trp-334 mutation in combina- analyzed for the presence or absence of site B mutations by DNA sequence analysis as described below. To subclone cells or to determine cloning efficiencies in * Corresponding author. the presence or absence ofcAMP analogs, about 300 to 1,000 767 768 STEINBERG AND GORMAN MOL. CELL. BIOL. cells were plated per 60-mm dish (varying with subline and selective conditions to yield 200 to 400 colonies per dish). a b c The diluted cell populations used for plating were counted AC GT A CG T AC G T with a Coulter electronic particle counter to determine the 4-_ input cell numbers. Wild-type cloning efficiencies on nonse- i A. q. - lective dishes were generally greater than 0.80. ....* 4-- Sequence analysis of mutant cDNAs. Poly(A)-containing *K RNAs were isolated from postnuclear supernatants of cell -4 extracts by oligo(dT)-cellulose chromatography and reverse 4- transcribed into cDNA as described elsewhere (9, 22). auw cAMP-binding site B regions of type I R subunit were then .p amplified by polymerase chain reaction (PCR), using PST 4 and 3PR primers as described previously (22). The amplified , 01 DNA fragments were purified by electrophoresis in gels of Wm - * ... .s NuSieve-GTG agarose and sequenced in both directions, Wm using PST and MLU primers (9, 22). To determine whether double mutants had the two muta- tions in the same or different alleles, a somewhat larger FIG. 1. Analysis of amplified cDNA sequences from R-subunit site B regions of spontaneous single (Trp-334) and double (Cys-332, interval was amplified by using BCL and 3PR primers, and a Trp-334) mutants. Site B regions were amplified from cDNA copies site B fragment cut from the amplified DNA with restriction of poly(A)-containing RNAs by using PCR, and the PCR products endonucleases EcoRI and MluI was subcloned into the M13 were purified and sequenced as described in Materials and Methods, bacteriophage vector M13um2l (IBI) as described previ- using the negative-strand MLU PCR primer. RNAs were from ously (22). Five to six recombinant phage plaques were control (a), singly mutant (b), or doubly mutant (c) cells. A, C, G, isolated from each preparation and sequenced with a Seque- and T designate dideoxynucleotides in termination reactions; arrow- nase II kit (U.S. Biochemical) as described elsewhere (22) heads indicate wild-type and mutant versions of mutated residues. but using the internal PST positive-strand primer. The wild-type sequence shown is TGGCAGCCCG*AGGACG* Two-dimensional gel analysis of mutant R subunits. Mutant ATTCATCAGCA, with asterisks indicating the mutated residues. cells were labeled with [35S]methionine, extracted, and purified by cAMP-affinity chromatography; the purified R subunits were then resolved by high-resolution two-dimen- tion for 2 h in ice, mixtures were diluted and assayed for sional gel electrophoresis as described previously (16). Pat- protein kinase activity as described previously (22); in all terns from well-characterized sublines (16) were used to cases, more than 85% of the C subunit was reconstituted into calibrate gel patterns for one- and two-charge-unit shifts in complexes that were inactive in the absence of cAMP. R-subunit positions. Statistical analyses. The significance of finding no double RESULTS mutants among Bt2cAMP-selected subclones of a Trp-334 mutant was analyzed by application of an algorithm for Identification of spontaneous Cys-332, Trp-334 double mu- computing exact significance levels in r x c contingency tants. Mutants were analyzed for the presence of mutations tables as described previously (16) but using Microsoft at Arg-332 and Arg-334 codons by amplifying the site B Fortran to compile the program and an AT&T PC6300 region from R-subunit cDNA and sequencing the resulting computer to run it. DNA products. Figure 1 shows representative sequencing Expression and characterization of mutant R subunits with gel patterns for site B regions from wild-type and mutant S49 a Cys-332, Gln-334, Leu-334, or Trp-334 mutation. The cell material; since the mutants tested were heterozygous for coding sequence for murine type I R subunit was introduced expression of mutant and wild-type alleles, the mutant from cDNA plasmids into a pET-8c expression vector (24) patterns showed evidence for both G and A at the mutated from which the EcoRI restriction site had been removed (by positions in these negative-strand sequences.