Reading-frame Restoration by Transcriptional Slippage at Long Stretches of Adenine Residues in Mammalian Cells

We previously characterized a mutant apoB allele (the apoB86 allele) that produces both a truncated apoB (apoB86) and a full-length apoB100. The mutant allele contained a deletion of a single cytosine in exon 26, creating a stretch of eight consecutive adenines in the −1 reading frame. The altered reading-frame allele was restored, with ∼10% efficiency, by the transcriptional insertion of an extra adenine into the stretch of eight consecutive adenines, thereby accounting for the synthesis of the full-length apoB100. Here, we demonstrate that this reading-frame restoration does not occur when the long stretch of adenines is interrupted by a cytosine. To assess whether reading-frame restoration is unique to a single site in the apoB gene, the same mutation (eight consecutive adenines in the −1 reading frame) was inserted into another site within the apoB gene. Reading-frame restoration occurred at the second site and was abrogated when the stretch of adenines was interrupted by another base. Of note, a computerized analysis of human cDNA sequences revealed that long stretches of adenines in protein-coding sequences occur at a lower than predicted frequency, suggesting that evolution has selected against these sequences. We previously characterized a mutant apoB allele (the apoB86 allele) that produces both a truncated apoB (apoB86) and a full-length apoB100. The mutant allele contained a deletion of a single cytosine in exon 26, creating a stretch of eight consecutive adenines in the −1 reading frame. The altered reading-frame allele was restored, with ∼10% efficiency, by the transcriptional insertion of an extra adenine into the stretch of eight consecutive adenines, thereby accounting for the synthesis of the full-length apoB100. Here, we demonstrate that this reading-frame restoration does not occur when the long stretch of adenines is interrupted by a cytosine. To assess whether reading-frame restoration is unique to a single site in the apoB gene, the same mutation (eight consecutive adenines in the −1 reading frame) was inserted into another site within the apoB gene. Reading-frame restoration occurred at the second site and was abrogated when the stretch of adenines was interrupted by another base. Of note, a computerized analysis of human cDNA sequences revealed that long stretches of adenines in protein-coding sequences occur at a lower than predicted frequency, suggesting that evolution has selected against these sequences. In 1992, we reported the existence of a mutant apoB allele causing familial hypobetalipoproteinemia, the apoB86 allele (1Linton M.F. Pierotti V. Young S.G. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 11431-11435Crossref PubMed Scopus (42) Google Scholar). This allele, which resulted in the synthesis of both a truncated apoB (apoB86) and a full-length apoB100, contained a deletion of a single cytosine residue in exon 26 of the apoB gene. This frameshift mutation was predicted to yield a stretch of 20 novel amino acids, followed by a premature stop codon. That this mutation was responsible for the production of apoB86 was proved by immunochemical studies utilizing monoclonal antibodies and an anti-peptide antibody directed against the 20 novel amino acids at the carboxyl terminus of apoB86 (1Linton M.F. Pierotti V. Young S.G. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 11431-11435Crossref PubMed Scopus (42) Google Scholar). The production of apoB100 from this allele involved a novel mechanism: reading-frame restoration by transcriptional insertion of an extra nucleotide at the site of the 1-base pair (bp) 1The abbreviation used is:bpbase pair(s) deletion. The deletion of the single cytosine residue in the apoB86 allele created a stretch of eight consecutive adenine residues. Minigene expression studies in cultured hepatoma cells demonstrated that the single cytosine deletion, along with the stretch of eight consecutive adenines, was faithfully present in the genomic DNA, but that ∼10% of apoB86 cDNA clones actually contained nine consecutive adenines (1Linton M.F. Pierotti V. Young S.G. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 11431-11435Crossref PubMed Scopus (42) Google Scholar). Thus, the production of apoB100 by the apoB86 allele was due to the transcriptional insertion of an extra adenine, which restored the proper reading frame to the mutant mRNA. These studies provided the first in vivo evidence of “transcriptional slippage” (or “stuttering”) along a long stretch of consecutive adenines in mammalian cells. base pair(s) Slippage of RNA polymerase during transcription was first proposed by Chamberlin and Berg in 1962 (2Chamberlin M. Berg P. Proc. Natl. Acad. Sci. U. S. A. 1962; 48: 81-93Crossref PubMed Scopus (354) Google Scholar). In recent years, Wagner et al. (3Wagner L.A. Weiss R.B. Driscoll R. Dunn D.S. Gesteland R.F. Nucleic Acids Res. 1990; 18: 3529-3535Crossref PubMed Scopus (114) Google Scholar) described transcriptional slippage by Escherichia coli RNA polymerase during RNA elongation at runs of 10 or more adenines or thymines, resulting in the addition of untemplated thymine or adenine residues and restoration of the proper reading frame to out-of-frame lacZ constructs. Interestingly, RNA polymerase stuttering was not observed when similar experiments were performed in yeast (3Wagner L.A. Weiss R.B. Driscoll R. Dunn D.S. Gesteland R.F. Nucleic Acids Res. 1990; 18: 3529-3535Crossref PubMed Scopus (114) Google Scholar). This study was undertaken to define further the DNA sequence requirements for reading-frame restoration in the apoB86 allele. We also sought to determine whether the reading-frame restoration was somehow unique to the apoB86 allele or would occur at similar sequences elsewhere in the apoB gene. Finally, because transcriptional slippage could potentially introduce frameshift mutations, we hypothesized that evolution may have selected against the presence of long stretches of adenines in protein-coding sequences. To assess this possibility, we analyzed whether long stretches of adenines occur at a lower than predicted frequency in human proteins containing three consecutive lysine residues. Blood was collected from H. J. B. (4Steinberg D. Grundy S.M. Mok H.Y.I. Turner J.D. Weinstein D.B. Brown W.V. Albers J.J. J. Clin. Invest. 1979; 64: 292-301Crossref PubMed Scopus (42) Google Scholar, 5Young S.G. Peralta F.P. Dubois B.W. Curtiss L.K. Boyles J.K. Witztum J.L. J. Biol. Chem. 1987; 262: 16604-16611Abstract Full Text PDF PubMed Google Scholar, 6Young S.G. Bertics S.J. Curtiss L.K. Dubois B.W. Witztum J.L. J. Clin. Invest. 1987; 79: 1842-1851Crossref PubMed Scopus (49) Google Scholar, 7Young S.G. Bertics S.J. Curtiss L.K. Witztum J.L. J. Clin. Invest. 1987; 79: 1831-1841Crossref PubMed Scopus (65) Google Scholar), and plasma was used to prepare very low density lipoproteins, which were used as a source of apoB size standards for Western blot experiments. H. J. B. is a compound heterozygote for hypobetalipoproteinemia with two mutant apoB alleles: an apoB37 allele (7Young S.G. Bertics S.J. Curtiss L.K. Witztum J.L. J. Clin. Invest. 1987; 79: 1831-1841Crossref PubMed Scopus (65) Google Scholar, 8Young S.G. Northey S.T. McCarthy B.J. Science. 1988; 241: 591-593Crossref PubMed Scopus (54) Google Scholar) and an apoB86 allele. The apoB37 allele yields exclusively apoB37 (6Young S.G. Bertics S.J. Curtiss L.K. Dubois B.W. Witztum J.L. J. Clin. Invest. 1987; 79: 1842-1851Crossref PubMed Scopus (49) Google Scholar, 7Young S.G. Bertics S.J. Curtiss L.K. Witztum J.L. J. Clin. Invest. 1987; 79: 1831-1841Crossref PubMed Scopus (65) Google Scholar), whereas the apoB86 allele yields apoB100, apoB86, and apoB48 (1Linton M.F. Pierotti V. Young S.G. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 11431-11435Crossref PubMed Scopus (42) Google Scholar, 6Young S.G. Bertics S.J. Curtiss L.K. Dubois B.W. Witztum J.L. J. Clin. Invest. 1987; 79: 1842-1851Crossref PubMed Scopus (49) Google Scholar, 7Young S.G. Bertics S.J. Curtiss L.K. Witztum J.L. J. Clin. Invest. 1987; 79: 1831-1841Crossref PubMed Scopus (65) Google Scholar). ApoB48 is produced by the apoB86 allele as a result of apoB mRNA editing in the intestine (9Powell L.M. Wallis S.C. Pease R.J. Edwards Y.H. Knott T.J. Scott J. Cell. 1987; 50: 831-840Abstract Full Text PDF PubMed Scopus (708) Google Scholar). The apoB86 allele contains a single cytosine deletion (cDNA nucleotide 11840) in exon 26 of the apoB gene, which generates a stretch of eight consecutive adenines in the −1 reading frame (i.e. the mutation changes the sequence AAA AAC AAA to AAA AAA AA). An apoB fusion minigene vector (pB18/86) containing the 1-bp deletion found in the apoB86 allele and a wild-type vector lacking the mutation (pB18/100) have been described (1Linton M.F. Pierotti V. Young S.G. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 11431-11435Crossref PubMed Scopus (42) Google Scholar). For this study, we mutated the pB18/100 expression vector to delete one of the adenines immediately preceding cytosine 11840 instead of cytosine 11840 itself. This construct contained the sequence AAA ACA AA and was designated pB18/86:4AC3A (because it had four adenines, one cytosine, and then three more adenines) (Fig. 1). Thus, pB18/86:4AC3A contained the same −1 frameshift as the apoB86 allele, but lacked the long stretch of adenines. To create pB18/86:4AC3A, we used the mutagenesis technique of Deng and Nickoloff (10Deng W.P. Nickoloff J.A. Anal. Biochem. 1992; 200: 81-88Crossref PubMed Scopus (1078) Google Scholar) and a mutagenic primer (5′-GCCAGTTTGAAAACAAAGCAGAT-3′). In addition, we generated pB18/86cDNA (Fig. 1), which is identical to pB18/86, except that it was constructed entirely from cDNA clones. For this clone, a 2119-bp Bam HI-Hin dIII cDNA fragment was ligated to the pB18 cDNA vector, rather than to the 3567-bpBam HI-Hin dIII genomic fragment. The apoB86 mutation was introduced into the vector by site-directed mutagenesis. All mutations were confirmed by DNA sequencing.