Nov 6, 1985 - dihydrofolate reductase mRNA species that were shorter than those of the ...... Splicing of messenger RNA precursors is inhibited by antisera.
MOLECULAR AND CELLULAR BIOLOGY, June 1986, p. 1926-1935
Vol. 6, No. 6
0270-7306/86/061926-10$02.00/0 Copyright © 1986, American Society for Microbiology
Spontaneous Splicing Mutations at the Dihydrofolate Reductase Locus in Chinese Hamster Ovary Cells PAMELA J. MITCHELL,t GAIL URLAUB, AND LAWRENCE CHASIN* Department of Biological Sciences, Columbia University, New York, New York 10027 Received 6 November 1985/Accepted 17 February 1986
We isolated and characterized three spontaneous mutants of Chinese hamster ovary cells that were deficient in dihydrofolate reductase activity. All three mutants contained no detectable enzyme activity and produced dihydrofolate reductase mRNA species that were shorter than those of the wild type by about 120 bases. Six exons are normally represented in this mRNA; exon 5 was missing in all three mutant mRNAs. Nuclease Si analysis of the three mutants indicated that during the processing of the mutant RNA, exon 4 was spliced to exon 6. The three mutant genes were cloned, and the regions around exons 4 and 5 were sequenced. In one mutant, the GT dinucleotide at the 5' end of intron 5 had changed to CT. In a second mutant, the first base in exon 5 had changed from G to T. In a revertant of this mutant, this base was further mutated to A, a return to a purine. Approximately 25% of the mRNA molecules in the revertant were spliced correctly to produce an enzyme with one presumed amino acid change. In the third mutant, the AG at the 3' end of intron 4 had changed to AA. A mutation that partially reversed the mutant phenotype had changed the dinucleotide at the 5' end of intron 4 from GT to AT. The splicing pattern in this revertant was consistent with the use of cryptic donor and acceptor splice sites close to the original sites to produce an mRNA with three base changes and a protein with two amino acid changes. These mutations argue against a scanning model for the selection of splice site pairs and suggest that only a single splice site need be inactivated to bring about efficient exon skipping (a regulatory mechanism for some genes). The fact that all three mutants analyzed exhibited exon 5 splicing mutations indicates that these splice sites are hot spots for spontaneous mutation.
Much recent progress has been made toward understanding the mechanism of pre-mRNA splicing in eucaryotic cells, especially since the development of cell-free systems for this process. Analysis of in vitro RNA splicing has led to the discovery of lariat structure splicing intermediates (35, 41) and has implicated Ul and U2 small nuclear ribonucleoproteins (snRNPs) as participants in the splicing process (3, 19). The extension of these studies is likely to define the chemical steps involved in the formation of spliced joints. A central problem that remains is how only correctly ordered splices are made in transcripts containing multiple introns, as is the case for the transcripts of most mammalian genes. The evidence on this point is confusing: wild-type transcripts require and exhibit great fidelity in making only the right connections, yet novel splicing can be readily produced experimentally. The alteration of a single base of a splice site can completely prevent splicing at that joint, but new cryptic splice junctions often appear as substitutes (51). It is possible to demonstrate efficient splicing between the exons of two different genes when these genes have been experimentally fused within an intron (12). Finally, splicing can be produced in molecules that would normally remain unspliced by the insertion of synthetic oligomers representing consensus 5' and 3' splice sequences (38). In contrast, not only are cryptic sites ignored in wild-type transcripts, but it is also probable that only the correctly ordered exon-to-exon connections are made, since myriad combinations of misspliced transcripts do not accumulate. A scanning mechanism (23) represents a simple solution for choosing the correct splice partners, but there is evidence against this model (22, 53). Moreover, in certain cases specific trans-acting factors probably mediate
particular splices, because several mammalian genes use differential splicing to specify alternative gene products (for examples, see references 7, 16, 17, 31, and 40). Evidence for the role of specific bases in the splicing process has come from site-specific mutagenesis of cloned sequences and from the analysis of in vivo mutations. The human ,o-thalassemias have been particularly revealing, because a surprisingly high proportion of these mutations are due to splicing defects in the ,B-globin gene (reviewed in references 13 and 33). The most common 3-globin splicing defects are due to single-base substitutions that prevent splicing at a particular site; splicing then occurs at formerly silent cryptic splice sites. The deletion of splice sites has been observed in other cellular genes carrying splicing mutations. The albumin gene in mutant analbuminemic rats has suffered a 7-base-pair (bp) deletion involving part of a 5' splice consensus sequence; high-molecular-weight albumin transcripts accumulate in the nucleus of liver cells, but no mature albumin mRNA is found in the cytoplasm. Several somatic mutations that affect splicing of immunoglobulin genes have been shown to result from deletions spanning an entire splice site. In these cases a 5' donor site has been deleted, and the exon upstream of the deletion is skipped (5, 11, 43). Similar exon skipping has been suggested to explain a short human leukocyte antigen (HLA) mRNA in a mutant lymphoblastoid cell line (21) and short adenosine deaminase and argininosuccinate synthetase mRNAs present in individuals with deficiencies of these enzymes (1, 44). The isolation and analysis of in vivo splicing mutants is often more difficult than the site-directed mutagenesis of a cloned gene. However, such mutants provide some potential advantages. First, they may reveal the involvement of sequences that would be missed in an in vitro approach because of experimental bias. Second, in vivo splicing defects are exhibited within the context of the natural gene in
* Corresponding author. t Present address: Department of Biochemistry, University of California, Berkeley, CA 94720. 1926
VOL. 6, 1986
DIHYDROFOLATE REDUCTASE SPLICING MUTATIONS
its indigenous chromosomal location and are not subject to the possible influence of contiguous foreign cloning sequences. Finally, if splicing mutations are isolated in a selectable gene in a cultured cell line, there is the possibility of gaining additional genetic information through the isolation of revertants and suppressed mutants. For these reasons, we included a screen for splicing mutations in the course of characterizing dihydrofolate reductase (DHFR)deficient mutants of Chinese hamster ovary (CHO) cells. The Chinese hamster gene for DHFR spans 26 kilobases (kb) and contains six exons and five introns; the latter range in size from 0.3 to 9.4 kb (10). All of the introns occur in protein-coding regions of the gene and are positioned identically in the mouse (15) and human (52) dhfr genes. We have previously described a method for isolating mutants deficient in DHFR based on their resistance to killing by tritiated deoxyuridine (47, 49). Such mutants are readily induced in a hemizygous CHO cell line which contains only one copy of the dhfr gene (48). DHFR-deficient mutants are viable if provided with the end products of one-carbon metabolism: glycine, a source of purines and thymidine. Twenty DHFRdeficient mutants isolated after treatment with mutagens and three spontaneous mutants were screened for the presence of dhfr mRNAs of unusual size. In this report, we describe three mutants that produced short dhfr mRNAs lacking exon 5. These mutants each contained a different single-base substitution in the splice sites bordering exon 5. Revertants of two of these mutants contained secondary single-base changes. Remarkably, these three splicing defects were found in the three spontaneous mutants. MATERIALS AND METHODS Cell culture. The parental CHO cell line used for the isolation of the mutants described here was clone UA21; this cell line, as well as the similar clone UA41, are hemizygous for the dhfr gene and have been described previously (48). Clone DUll is a UV-induced deletion mutant derived from UA21 that contains no dhfr genes. RNA from this cell line was used as a negative control in Northern blots and nuclease S1 protection experiments; its isolation will be described elsewhere. Cells were routinely cultured as described previously (47). Mutant isolation. DHFR-deficient mutants were isolated by a modification of the previously described tritiated deoxyuridine suicide method. For the isolation of independent spontaneous mutants, 72 cultures were grown up from inocula of 5 to 10 cells in 2-cm wells of multiwell tissue culture dishes. When the cultures approached confluency, approximately 3 x 105 cells from each well were challenged in selective medium. The cells were exposed to a low concentration of [5-3H]deoxyuridine (1.5 x 10-8 M, 24 Ci/mmol; Amersham) in 150-mm dishes for 20 h, trypsinized, and stored frozen at -65°C for 14 days in medium containing 10% (vol/vol) dimethyl sulfoxide. The cultures were thawed and grown for 5 days, and then one-fifth of each was subjected to the double-pulse [3H]deoxyuridine selection previously described (49). Revertants to a DHFR+ phenotype were isolated following mutagenesis with ethyl methanesulfonate and selection in medium lacking glycine, hypoxanthine, and thymidine, as described previously (47). RNA analysis. Total RNA was extracted from cells grown in suspension culture, separated into polyadenylated [poly(A)+] and poly(A)- fractions, and analyzed by Northern blotting as previously described (29). S1 analysis was
1927
done with 2.3 ,ug of poly(A)+ RNA and 15- to 30-fold excess of uniformly labeled single-stranded probe. The procedures for the S1 analysis and the preparation of single-stranded probes from M13 recombinants have been described (29). The probe sequence representing exon 5 with adjacent parts of exons 4 and 6 was cloned by digesting the cDNAcontaining plasmid pA3-A35 (26) with SstI and BglII and cloning the appropriate 404-base fragment between the SstI and BamI sites of M13mpl8. Autoradiographic bands were quantitated by densitometry, with a correction factor for film nonlinearity (34). DNA analysis. The preparation of DNA and Southern blotting (44) analysis were carried out as described previously (48). Mutant and revertant dhfr genes were cloned as 41-kb BglI fragments with cosmid cloning vectors constructed for the specific cloning of this gene, as has been described for DS11 and DS31 (46). For DNA sequencing of the region around exon 5, each dhfr cosmid clone was digested with KpnI and MspI, and the resulting fragments were subcloned into M13mpl9 cut with KpnI and AccI. Plaques were transferred to nitrocellulose membranes and hybridized (2) with a probe containing exon 4 and exon 5 to identify recombinants carrying the 700-bp KpnI-MspI fragment (10) spanning exon 5. For sequencing the opposite strand, the wild-type M13 recombinant clone was cut with KpnI and ScaI, and the 350-bp fragment was cloned into M13mpl8 cut with KpnI and HincII. In a similar way, a 500-bp genomic region spanning the exon 4-intron 4 joint was cloned by cutting the cosmid DNA with HindlIl and SstI, cloning the fragments into the corresponding sites of M13mpl9, and screening the recombinant plaques with the same probe. DNA sequencing was carried out by the dideoxynucleotide method (42) with a 17-base universal M13 primer (New England Biolabs) and the Klenow fragment of DNA *polymerase (Bethesda Research Laboratories). For each of the six genes analyzed (three mutations, two revertants, and the wild type), two independent cosmid clones were isolated, sublcloned, and sequenced so as to be able to detect any base substitutions that might have arisen as cloning artifacts. Both strands of the exon 5 region were sequenced in the wild-type subclones. For the exon 4 region, only one strand was sequenced. In every case, the independent subclones showed perfect sequence agreement. The precise locations of exons 4 and 5 were assigned by comparing the genomic sequences with the published sequence for Chinese hamster lung cell dhfr cDNA (26). Our exonic sequences for CHO dhfr are in agreement with the cDNA sequence. RESULTS Isolation of spontaneous DHFR-deficient mutants. The starting cell line for the isolation of DHFR-deficient mutants was UA21, a CHO cell line that is hemizygous for the dhfr locus (48). This clone will be referred to as the wild type for comparison with the mutants derived from it. The independence of the mutants isolated was ensured by establishing separate cultures of UA21 from inocula of 5 to 10 cells; when these populations reached several hundred thousand cells, they were subjected to a tritiated deoxyuridine suicide technique that selects mutants deficient in DHFR activity (49). Of 72 cultures tested, only three yielded mutants deficient in DHFR activity. From these data a spontaneous mutation rate of 1.3 x 10' can be estimated, based on the average number of mutants per culture (9). This rate is comparable to estimates of spontaneous mutation rates for the inactivation of other single-allele loci in cultured mam-
1928
MITCHELL ET AL.
TABLE 1. Phenotypic characteristics of spontaneous dhfr mutants and revertants Growth requirementb DHFR activity' Clone None 1.00 UA21 GHT
