Human genomic sequences that inhibit splicing.

Mammalian genes are characterized by relatively small exons surrounded by variable lengths of intronic sequence. Sequences similar to the splice signals that define the 5' and 3' boundaries of these exons are also present in abundance throughout the surrounding introns. What causes the real sites to be distinguished from the multitude of pseudosites in pre-mRNA is unclear. Much progress has been made in defining additional sequence elements that enhance the use of particular sites. Less work has been done on sequences that repress the use of particular splice sites. To find additional examples of sequences that inhibit splicing, we searched human genomic DNA libraries for sequences that would inhibit the inclusion of a constitutively spliced exon. Genetic selection experiments suggested that such sequences were common, and we subsequently tested randomly chosen restriction fragments of about 100 bp. When inserted into the central exon of a three-exon minigene, about one in three inhibited inclusion, revealing a high frequency of inhibitory elements in human DNA. In contrast, only 1 in 27 Escherichia coli DNA fragments was inhibitory. Several previously identified silencing elements derived from alternatively spliced exons functioned weakly in this constitutively spliced exon. In contrast, a high-affinity site for U2AF65 strongly inhibited exon inclusion. Together, our results suggest that splicing occurs in a background of repression and, since many of our inhibitors contain splice like signals, we suggest that repression of some pseudosites may occur through an inhibitory arrangement of these sites.

Multiple splicing defects in an intronic false exon.

Splice site consensus sequences alone are insufficient to dictate the recognition of real constitutive splice sites within the typically large transcripts of higher eukaryotes, and large numbers of pseudoexons flanked by pseudosplice sites with good matches to the consensus sequences can be easily designated. In an attempt to identify elements that prevent pseudoexon splicing, we have systematically altered known splicing signals, as well as immediately adjacent flanking sequences, of an arbitrarily chosen pseudoexon from intron 1 of the human hprt gene. The substitution of a 5' splice site that perfectly matches the 5' consensus combined with mutation to match the CAG/G sequence of the 3' consensus failed to get this model pseudoexon included as the central exon in a dhfr minigene context. Provision of a real 3' splice site and a consensus 5' splice site and removal of an upstream inhibitory sequence were necessary and sufficient to confer splicing on the pseudoexon. This activated context also supported the splicing of a second pseudoexon sequence containing no apparent enhancer. Thus, both the 5' splice site sequence and the polypyrimidine tract of the pseudoexon are defective despite their good agreement with the consensus. On the other hand, the pseudoexon body did not exert a negative influence on splicing. The introduction into the pseudoexon of a sequence selected for binding to ASF/SF2 or its replacement with beta-globin exon 2 only partially reversed the effect of the upstream negative element and the defective polypyrimidine tract. These results support the idea that exon-bridging enhancers are not a prerequisite for constitutive exon definition and suggest that intrinsically defective splice sites and negative elements play important roles in distinguishing the real splicing signal from the vast number of false splicing signals.

Effect of differential polyadenylation and cell growth phase on dihydrofolate reductase mRNA stability.

We have constructed tetracycline-responsive dhfr minigenes and transferred them to a Chinese hamster ovary cell DHFR-deficient deletion mutant to obtained cells in which dhfr transcription can be repressed by tetracycline (tet-off). DHFR mRNA half-life measured after the repression of transcription by tetracycline in these transfectants is about 1.5 h, which is significantly shorter than previously reported. In addition, we observed that DHFR mRNA is less stable in serum-starved cells than in exponentially growing cells. Given that the dhfr gene contains multiple polyadenylation sites, we analyzed the role of polyadenylation site usage on the stability of the resultant mRNA molecules. We found that DHFR mRNA is more stable when a strong polyadenylation site is used. Finally, we have observed that the relative lengths of the poly(A) tails for the different DHFR mRNA species correlated with their relative stability in growing versus resting cells.

Control of 3' splice site choice in vivo by ASF/SF2 and hnRNP A1.

The constitutive splicing factor ASF/SF2 has been shown to affect the choice between alternative splice sites by favoring the proximal as opposed to the distal choice. HnRNP A1 antagonizes ASF/SF2 by promoting the distal choice for competing 5' splice sites. We have tested the in vivo effects of these proteins on alternative 3' splice site choices. Cotransfection of a dihydrofolate reductase-calcitonin chimeric construct togetherwith a plasmid specifying the SR protein ASF/SF2 into cells of several mammalian lines increased use of a proximal 3' splice site, resulting in the inclusion of a terminal calcitonin exon. This stimulation of 3' proximal splicing was antagonized by cotransfection with an hnRNP A1 plasmid. This effect of hnRNP A1 in promoting distal splicing was also seen in an hnRNP A1-deficient MEL cell line. A similar effect of hnRNP A1 was demonstrated with mutant hamster adenine phosphoribosyltransferase (aprt) transcripts that are normally constitutively spliced, suggesting that hnRNP A1 may be a general inhibitor of proximal splicing. Intron size also influenced splice site choice in mutant aprt transcripts, with larger introns favoring proximal splicing. These results support the idea that the ratios of particular but general splicing factors and hnRNPs play a role in alternative splicing.

Retinoblastoma protein associates with SP1 and activates the hamster dihydrofolate reductase promoter.

The dihydrofolate reductase (dhfr) promoter is powerfully activated by the transcription factor Sp1. It has been suggested that Sp1 is a potential target for transcriptional regulation by the cell cycle regulator retinoblastoma protein (Rb), and so we have explored this possibility using the hamster dhfr gene as a model. By the use of DNA probes from the hamster dhfr gene promoter, containing the most proximal GC box (minimal promoter), and nuclear extracts from cultured hamster cells (CHO K1), we show that polyclonal and monoclonal antibodies against Rb supershift the binding of Sp1. Nuclear extract immunoprecipitation with anti-Rb followed by Western analysis using anti-Sp1 also shows that Rb is complexed to Sp1. Complementary Immunoprecipitation/WB analysis shows both forms of Rb protein in the anti-Sp1 immunoprecipitates. Moreover, nuclear extract immunodepletion of Rb abolishes Sp1 gel-shift. The interaction between Rb and Sp1 is maintained in all the phases of the cell cycle. Transient overexpression of Rb in dhfr negative cells co-transfected with a dhfr minigene driven by its minimal promoter increases DHFR activity and potentiates transcription when overexpressing Sp1. Both effects are severely reduced when the co-transfections are performed with a homologous dhfr minigene containing a single point mutation in the GC box. Thus, the activation by Rb of the dhfr gene may be exerted through Sp1. Stable transfectants of pCMVRb in K1 cells show an increase in both mRNA and DHFR activity. It is concluded that Sp1 is physically associated with Rb, and that this association increases Sp1-mediated transcription of the hamster dhfr gene.

Cointegration of DNA molecules introduced into mammalian cells by electroporation.

Electroporation was used to introduce a mixture of two plasmid-cloned genes into Chinese hamster ovary (CHO) cells, and the location of the two genes was subsequently determined by fluorescence in situ hybridization (FISH). The 25 kb Chinese hamster gene for dihydrofolate reductase (dhfr) in the form of a cosmid-derived 40 kb BglI fragment and the SV40 promoter-driven E. coli gene for guanine phosphoribosyltransferase (gpt) were co-electroporated and gpt + transfectants selected. Clones that had also integrated a single copy of the dhfr gene were studied by 2-color fluorescence in situ hybridization (FISH) to localize the integration site(s) of the exogenous DNA in metaphase chromosomes. All 9 clones examined showed co-localization of the two transgenes. The chromosomal site of integration was different in each clone. Co-integration was confirmed by co-amplification experiments. We conclude that, even when provided at low concentrations, separate soluble DNA molecules become linked upon gene transfer by electroporation, either by intracellular ligation prior to integration, or by co-integration at a common site in a given recipient cell.

Cell-growth regulation of the hamster dihydrofolate reductase gene promoter by transcription factor Sp1.

The dihydrofolate reductase (DHFR) gene (dhfr) promoter contains cis-acting elements for the transcription factors Sp1 and E2F. Given the ability of Sp1 to activate the dhfr promoter, we have evaluated the contribution of Sp1 to the cell-growth regulation of the dhfr gene. Using gel-mobility assays performed with DNA probes from the minimal promoter of the hamster dhfr gene and nuclear extracts from cultured hamster cells (CHO K1) we show that the binding of Sp1 to the dhfr promoter is cell-growth-phase regulated. Accordingly, dhfr transcription and mRNA levels in K1 cells increase upon serum stimulation. Cytological detection of Sp1 by immunofluorescence reveals a decrease of this protein in the process leading to the G0 state, and an increase upon serum stimulation of quiescent cells. These results were confirmed by western blot analysis. It is concluded that Sp1 progressively binds to the hamster dhfr promoter after stimulation of cell proliferation, which can account for the transcriptional regulation of the dhfr gene during the cell cycle. The role of Sp1 in the specific control of dhfr during the cell cycle was confirmed in vivo using cell lines derived from dhfr-negative cells transfected with dhfr plasmids carrying either the wild-type or mutated Sp1-binding or E2F-binding sites in the dhfr minimal promoter.

Effects of nonsense mutations on nuclear and cytoplasmic adenine phosphoribosyltransferase RNA.

We have analyzed Chinese hamster ovary (CHO) cell mutants bearing nonsense codons in four of the five exons of the adenine phosphoribosyltransferase (aprt) gene and have found a pattern of mRNA reduction similar to that seen in systems studied previously: a decrease in steady-state mRNA levels of 5- to 10-fold for mutations in exons 1, 2, and 4 but little effect for mutations in the 3'-most exon (exon 5). Nuclear aprt mRNA levels showed a similar decrease. Nonsense-containing aprt mRNA decayed at the same rate as wild-type mRNA in these cell lines after inhibition of transcription with actinomycin D. Nonsense-containing aprt mRNA is associated with polysomes, ruling out a model in which stable residual mRNA escapes degradation by avoiding translation initiation. A tetracycline-responsive form of the aprt gene was used to compare the stability of nonsense-containing and wild-type aprt mRNAs without globally inhibiting transcription. In contrast to measurements made in the presence of actinomycin D, after inhibition of aprt transcription with tetracycline, a nonsense-mediated destabilization of aprt mRNA was indeed demonstrable. The increased rate of decay of cytoplasmic aprt mRNA seen here could account for the nonsense-mediated reduction in steady-state levels of aprt mRNA. However, the low levels of nonsense-bearing aprt mRNA in the nucleus suggest a sensibility of mRNA to translation or translatability before it exits that compartment. Quantitation of the steady-state levels of transcripts containing introns revealed no accumulation of partially spliced aprt RNA and hence no indication of nonsense-mediated aberrancies in splicing. Our results are consistent with a model in which translation facilitates the export of mRNA through a nuclear pore. However, the mechanism of this intriguing nucleocytoplasmic communication remains to be determined.

Large exon size does not limit splicing in vivo.

Exon sizes in vertebrate genes are, with a few exceptions, limited to less than 300 bases. It has been proposed that this limitation may derive from the exon definition model of splice site recognition. In this model, a downstream donor site enhances splicing at the upstream acceptor site of the same exon. This enhancement may require contact between factors bound to each end of the exon; an exon size limitation would promote such contact. To test the idea that proximity was required for exon definition, we inserted random DNA fragments from Escherichia coli into a central exon in a three-exon dihydrofolate reductase minigene and tested whether the expanded exons were efficiently spliced. DNA from a plasmid library of expanded minigenes was used to transfect a CHO cell deletion mutant lacking the dhfr locus. PCR analysis of DNA isolated from the pooled stable cotransfectant populations displayed a range of DNA insert sizes from 50 to 1,500 nucleotides. A parallel analysis of the RNA from this population by reverse transcription followed by PCR showed a similar size distribution. Central exons as large as 1,400 bases could be spliced into mRNA. We also tested individual plasmid clones containing exon inserts of defined sizes. The largest exon included in mRNA was 1,200 bases in length, well above the 300-base limit implied by the survey of naturally occurring exons. We conclude that a limitation in exon size is not part of the exon definition mechanism.

A mutational hot spot induced by N-hydroxy-aminofluorene in dihydrofolate reductase mutants of Chinese hamster ovary cells.

Eighteen mutants deficient in dihydrofolate reductase (DHFR) activity were induced with 0.5 microM N-hydroxy-aminofluorene in four separate experiments. This carcinogen dose killed approximately 80% of the treated cells and resulted in a mutational frequency approximately 3 x 10(-6). The nature of the induced changes in each of the mutants was determined by direct sequencing following polymerase chain reaction amplification, or in one instance, by Southern blot analysis. Nearly all (15/17) of the mutations were single base changes. Consistent with the binding specificity of this chemical, all mutations were targeted to guanine bases. The predominant change was G:C-->T:A transversion which was evident in 11/15 mutants. A single dG-AF mutational hotspot was noted at a site in the DHFR coding sequence of exon 4; one-third of the induced point mutations arose at this position. These results are compared with our previous analyses of mutants induced with the related aromatic amine, N-2-acetoxy-2-acetyl-aminofluorene.

Order of intron removal during splicing of endogenous adenine phosphoribosyltransferase and dihydrofolate reductase pre-mRNA.

Using a strategy based on reverse transcription and the polymerase chain reaction, we have determined the order of splicing of the four introns of the endogenous adenine phosphoribosyltransferase (aprt) gene in Chinese hamster ovary cells. The method involves a pairwise comparison of molecules that retain one intron and have either retained or spliced another intron(s). A highly preferred order of removal was found: intron 3 > 2 > 4 = 1. This order did not represent a linear progression from one end of the transcript to the other, nor did it correlate with the conformity of the splice site sequences to the consensus sequences or to the calculated energy of duplex formation with U1 small nuclear RNA. By using actinomycin D to inhibit RNA synthesis, the in vivo rate of the first step in splicing was estimated for all four introns; a half-life of 6 min was found for introns 2, 3, and 4. Intron 1 was spliced more slowly, with a 12-min half-life. A substantial amount of RNA that retained intron 1 as the sole intron was exported to the cytoplasm. In the course of these experiments, we also determined that intron 3, but not intron 4, is spliced before 3'-end formation is complete, probably on nascent transcripts. This result is consistent with the idea that polyadenylation is required for splicing of the 3'-most intron. We applied a similar strategy to determine the last intron to be spliced in a very large transcript, that of the endogenous dihydrofolate reductase (dhfr) gene in Chinese hamster ovary cells (25 kb). Here again, intron 1 was the last intron to be spliced.

Splicing mutants and their second-site suppressors at the dihydrofolate reductase locus in Chinese hamster ovary cells.

Point mutants induced with a variety of mutagens at the dihydrofolate reductase (dhfr) locus in Chinese hamster ovary (CHO) cells were screened for aberrantly spliced dhfr mRNA by RNase protection and/or reverse transcriptase coupled with cDNA amplification by the polymerase chain reaction (PCR). Of 115 mutants screened, 28 were found to be affected in splicing. All exhibited less than 1% correct splicing, probably because the selection procedure was stringent. All 26 unique mutations were located within the consensus splice sequences; changes were found at 9 of 10 possible sites in this 25-kb six-exon gene. Mutations at the sites flanking the first and last exons resulted in the efficient recruitment of a cryptic site within each exon. In contrast, mutations bordering internal exons caused predominantly exon skipping. In many cases, multiple exons were skipped, suggesting the clustering of adjacent exons prior to actual splicing. Six mutations fell outside the well-conserved GU and AG dinucleotides. All but one were donor site single-base substitutions that decreased the agreement with the consensus and resulted in little or no correct splicing. Starting with five of these donor site mutants, we isolated 31 DHFR+ revertants. Most revertants carried a single-base substitution at a site other than that of the original mutation, and most had only partially regained the ability to splice correctly. The second-site suppression occurred through a variety of mechanisms: (i) a second change within the consensus sequence that produced a better agreement with the consensus; (ii) a change close to but beyond the consensus boundaries, as far as 8 bases upstream in the exon or 28 bases downstream in the intron; (iii) mutations in an apparent pseudo 5' site in the intron, 84 and 88 bases downstream of a donor site; and (iv) mutations that improved the upstream acceptor site of the affected exon. Taken together, these second-site suppressor mutations extend the definition of a splice site beyond the consensus sequence.

Direct selection for mutations affecting specific splice sites in a hamster dihydrofolate reductase minigene.

A Chinese hamster cell line containing an extra exon 2 (50 bp) inserted into a single intron of a dihydrofolate reductase (dhfr) minigene was constructed. The extra exon 2 was efficiently spliced into the RNA, resulting in an mRNA that is incapable of coding for the DHFR enzyme. Mutations that decreased splicing of this extra exon 2 caused it to be skipped and so produced normal dhfr mRNA. In contrast to the parental cell line, the splicing mutants display a DHFR-positive growth phenotype. Splicing mutants were isolated from this cell line after treatment with four different mutagens (racemic benzo[c]phenanthrene diol epoxide, ethyl methanesulfonate, ethyl nitrosourea, and UV irradiation). By polymerase chain reaction amplification and direct DNA sequencing, we determined the base changes in 66 mutants. Each of the mutagens generated highly specific base changes. All mutations were single-base substitutions and comprised 24 different changes distributed over 16 positions. Most of the mutations were within the consensus sequences at the exon 2 splice donor, acceptor, and branch sites. The RNA splicing patterns in the mutants were analyzed by quantitative reverse transcription-polymerase chain reaction. The recruitment of cryptic sites was rarely seen; simple exon skipping was the predominant mutant phenotype. The wide variety of mutations that produced exon skipping suggests that this phenotype is the typical consequence of splice site damage and supports the exon definition model of splice site selection. A few mutations were located outside the consensus sequences, in the exon or between the branch point and the polypyrimidine tract, identifying additional positions that play a role in splice site definition. That most of these 66 mutations fell within consensus sequences in this near-saturation mutagenesis suggests that splicing signals beyond the consensus may consist of robust RNA structures.

An alternatively spliced Pit-1 isoform altered in its ability to trans-activate.

Although alternative splicing has been shown to give rise to isoforms of a number of transcription factors, such isoforms have not previously been detected for the POU homeodomain protein Pit-1. Screening of a rat pituitary GH3 cell cDNA expression library yielded a clone, termed pCMVPit-1a, encoding a 35.8 kD protein (Pit-1a) containing a 26 amino acid insert in the Pit-1 trans-activation domain. The position of the insert, plus Southern blot analysis, implied that Pit-1a mRNA arises by alternative splicing of the Pit-1 gene transcript. Pit-1a mRNA was detected in GH3 rat pituitary tumor cells at levels about 1/7 that of Pit-1 mRNA. Pit-1a mRNA-specific sequences were also detected in rat and mouse pituitary, and in mouse thyrotropic tumor TtT cells. DNA mobility shift assays showed that Pit-1a binds specifically to Pit-1 binding sites in the proximal prolactin promoter, but produces DNA-protein complexes of markedly different mobilities than Pit-1. In stably transfected CHO cells which accumulated approximately equal levels of either of the two proteins, Pit-1 trans-activated a prolactin promoter-driven CAT construct, while Pit-1a yielded no detectable transactivation, implying a trans-activation ratio for Pit-1a/Pit-1 of less than 0.05. Thus, the insertion of 26 amino acids of similar composition into the activation domain of Pit-1 has at once affected both the mode of binding of this protein and its ability to function as a trans-activator.

Point mutational analysis of the hamster dihydrofolate reductase minimum promoter.

We have shown previously that 48 base pairs (bp) of 5'-flanking sequence are necessary for correct initiation at the major transcriptional start site of the Chinese hamster dihydrofolate reductase (dhfr) gene (Ciudad et al., 1988). As an upstream element, this sequence alone confers 25% of maximum promoter activity. The 5' half of this sequence is particularly well conserved among mammalian species; it contains one Sp1 binding site (GC box) and one CAA element. In the present work, we have analyzed the role of this region by extensive point mutational analysis. Twenty-three dhfr minigene constructs containing 1- or 2-base substitutions in this region of the promoter were tested by measuring their ability to transfect DHFR-deficient Chinese hamster ovary cells to a DHFR+ growth phenotype. Eight mutants, all in or near the GC box, exhibited reduced transfection efficiency. Promoter disfunction in these mutants was confirmed by RNase protection analysis of stable transfectants. Gel retardation experiments showed that mutants affected in the consensus sequence for Sp1 binding were deficient in binding a protein found in nuclear extracts of Chinese hamster ovary cells. Purified human transcription factor Sp1 was also unable to bind a promoter sequence bearing one of these single base substitutions, suggesting that Sp1 itself is involved in dhfr transcription in vivo. We conclude that most single base mutations in the GC box severely cripple or eliminate promoter function by inhibiting binding of transcription factors to this regulatory sequence and that Sp1 is likely to be involved in dhfr transcription in vivo. We also found that the well conserved CAA element is not absolutely necessary for transcription.

Use of a selectable reporter for the isolation of mammalian regulatory genes.

A general method is described for isolating the genes encoding differentiation-specific activators of transcription using genetic selection. Employing regulation of the prolactin encoding gene (PRL) as a model, we have shown that the hamster dihydrofolate reductase-encoding gene (dhfr) is an effective dominant selectable reporter in this methodology. The dhfr coding region was ligated to the rat PRL promoter, and the resultant construct was stably transfected into DHFR- Chinese hamster ovary (CHO) cells, where it had little or no activity. Transfection of these cells with plasmid DNA, containing the coding region of a pituitary-specific transcription factor (Pit-1/GHF-1) in a eukaryotic expression vector, resulted in transfectants in which activation of the chimeric construct, pPRLdhfr, had occurred, enabling these cells to be selected on the basis of their DHFR+ phenotype. Our results suggest that this strategy could be used to isolate unknown genes that regulate a variety of differentiated functions.

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.

DNA base changes and RNA levels in N-acetoxy-2-acetylaminofluorene-induced dihydrofolate reductase mutants of Chinese hamster ovary cells.

Formerly, we isolated a series of dihydrofolate reductase-deficient Chinese hamster ovary cell mutants that were induced by N-acetoxy-2-acetylaminofluorene. Deletions and complex gene rearrangements were detected in 28% of these mutants; 72% contained putative point mutations. In the present study, we have localized the putative point mutations in the 25,000 base dhfr gene by RNase heteroduplex mapping. Assignment of a position for each mutation was successful in 16 of 19 mutants studied. We cloned DNA fragments containing the mapped mutations from nine mutants into a bacteriophage lambda vector. In the case of 11 other mutants, DNA was amplified by the polymerase chain reaction procedure. Sequence analysis of cloned and amplified DNA confirmed the presence of point mutations. Most mutants (90%) carried base substitutions; the rest contained frameshift mutations. Of the point mutations, 75% were G.C to T.A transversions in either the dhfr coding sequence or at splice sites; transition G.C to A.T mutations were found in two mutants (10%). In one of these transition mutants, the base substitution occurred at the fifth base of the third intron. Of the frameshift mutations, one was a deletion of G.C pair and the other was an insertion of an A.T pair. Of the mapped mutants, 38% exhibited greatly reduced (approximately 10-fold) steady-state levels of dhfr mRNA. All eight sequenced mutants displaying this phenotype contained premature chain termination codons. Normal levels of dhfr mRNA were observed in five missense mutants and in five mutants carrying nonsense codons in the translated portion of exon VI. Taken together with the results of other mutagens at this locus, we conclude that the low dhfr mRNA phenotype is correlated with the presence of nonsense codons in exons II to V but not in the last exon of the dhfr gene.

Nonsense mutations in the dihydrofolate reductase gene affect RNA processing.

Steady-state dihydrofolate reductase (dhfr) mRNA levels were decreased as a result of nonsense mutations in the dhfr gene. Thirteen DHFR-deficient mutants were isolated after treatment of Chinese hamster ovary cells with UV irradiation. The positions of most point mutations were localized by RNA heteroduplex mapping, the mutated regions were isolated by cloning or by enzymatic amplification, and base changes were determined by DNA sequencing. Two of the mutants suffered large deletions that spanned the entire dhfr gene. The remaining 11 mutations consisted of nine single-base substitutions, one double-base substitution, and one single-base insertion. All of the single-base substitutions took place at the 3' position of a pyrimidine dinucleotide, supporting the idea that UV mutagenesis proceeds through the formation of pyrimidine dimers in mammalian cells. Of the 11 point mutations, 10 resulted in nonsense codons, either directly or by a frameshift, suggesting that the selection method favored a null phenotype. An examination of steady-state RNA levels in cells carrying these mutations and a comparison with similar data from other dhfr mutants (A. M. Carothers, R. W. Steigerwalt, G. Urlaub, L. A. Chasin, and D. Grunberger, J. Mol. Biol., in press) showed that translation termination mutations in any of the internal exons of the gene gave rise to a low-RNA phenotype, whereas missense mutations in these exons or terminations in exon 6 (the final exon) did not affect dhfr mRNA levels. Nuclear run-on experiments showed that transcription of the mutant genes was normal. The stability of mature dhfr mRNA also was not affected, since (i) decay rates were the same in wild-type and mutant cells after inhibition of RNA synthesis with actinomycin D and (ii) intronless minigene versions of cloned wild-type and nonsense mutant genes were expressed equally after stable transfection. We conclude that RNA processing has been affected by these nonsense mutations and present a model in which both splicing and nuclear transport of an RNA molecule are coupled to its translation. Curiously, the low-RNA mutant phenotype was not exhibited after transfer of the mutant genes, suggesting that the transcripts of transfected genes may be processed differently than are those of their endogenous counterparts.