Sequences that direct significant levels of frameshifting are frequent in coding regions of Escherichia coli

Abstract

It is generally believed that significant ribosomal frameshifting during translation does not occur without a functional purpose. The distribution of two frameshift-prone sequences, A_AAA_AAG and CCC_TGA, in coding regions of Escherichia coli has been analyzed. Although a moderate level of selection against the first sequence is evident, 68 genes contain A_AAA_AAG and 19 contain CCC_TGA. The majority of those tested in their genomic context showed >1% frameshifting. Comparative sequence analysis was employed to assess a potential biological role for frameshifting in decoding these genes. Two new candidates, in pheL and ydaY, for utilized frameshifting have been identified in addition to those previously known in dnaX and nine insertion sequence elements. For the majority of the shift-prone sequences no functional role can be attributed to them, and the frameshifting is likely erroneous. However, none of frameshift sequences is in the 306 most highly expressed genes. The unexpected conclusion is that moderate frameshifting during expression of at least some other genes is not sufficiently harmful for cells to trigger strong negative evolutionary pressure.

Fig. 1. Measurement of frameshifting efficiency on the A_AAA_AAG sequences. (A) Schematic representation of constructs used to assay frameshifting. (B) Pulse–chase analysis of the products expressed from cassettes with the A_AAA_AAG contexts from different genes. The areas from the gels corresponding to the termination and frameshifting products are shown. The GST lane shows the corresponding products from the parental vector in which the stop codon is located after GST; the GST–MBP lane shows products from the parental vector in which the GST and MBP genes are in-frame. The (–) lane contains labeled proteins from the uninduced control (Materials and methods). (C) Quantitation of the efficiency of frameshifting. Average frameshifting in three independent pulse–chase experiments was calculated for each construct and is represented by black bars. Error bars show standard deviations.

Fig. 2. Pulse–chase analysis of the products expressed from the construct with the entire sequence of the ycdB gene. The (–) lane contains labeled proteins from the uninduced control.

Fig. 3. Distribution of occurrences of slippery sequences in 1000 randomized genomes. (A) A_AAA_AAG. (B) CCC_TGA.

Fig. 4. Measurement of the frameshifting efficiency in cassettes with sequences from genes ending with CCC_UGA. (A) Schematic representation of expression constructs for analysis of frameshifting efficiency. (B) Pulse–chase analysis of the products expressed from vectors containing inserts of different genes ending with CCC_UGA. The areas from the gels corresponding to the termination and frameshifting products are shown. FS indicates frameshift product; TER indicates termination product. The GST lane shows the corresponding products from the parental vector in which the stop codon is located after GST; the GST–MBP lane shows products from the parental vector in which the GST and MBP genes are in-frame. The (–) lane contains labeled proteins from uninduced control. (C) Quantitation of the frameshifting efficiency. Average frameshifting efficiency of three independent pulse–chase experiments was calculated for each construct and is represented by black bars. Sequences in which frameshifting is <1% are omitted. Error bars show standard deviations.

Fig. 5. Mass spectrum of the GST–MBP fusion protein synthesized from a cassette containing the yjeF sequence. The major peak at 73 628.15 Da corresponds to the predicted mass of the fusion protein (73 629.91 Da). The satellite peak at 73 703.07 Da corresponds to the β-mercaptoethenol adduct of the fusion protein.

Fig. 6. Analysis of the frameshifting efficiency on different CCN-Stop combinations. (A) Pulse–chase experiments with expression vectors containing mutations in pheL. CCC denotes the wild-type pheL context. Other abbreviations indicate the mutation of either a Pro or a Stop codon. The areas from the gels corresponding to the termination and frameshifting products are shown. The GST lane shows the corresponding products from the parental vector in which the stop codon is located after GST; the G–M lane shows products from the parental vector in which the GST and MBP genes are in-frame. The (–) lane contains labeled proteins from uninduced cultures. (B) Quantitation of the pulse–chase results with mutated pheL constructs. Average frameshifting in three independent pulse–chase experiments was calculated for each construct and is represented by black bars. Error bars show standard deviations.

Fig. 7. Sequence of the ycdV gene. Annotated initiation codon, termination codon and termination codon in the –1 frame are in bold. The A_AAA_AAG sequence in the ‘0’ frame is underlined. Repeated sequences are differently highlighted.

Fig. 8. Comparative sequence analysis of cdd genes from B.subtilis (upper sequence) and B.firmus (lower sequence). FS site, position of frameshifting site in cdd gene from B.subtilis; SD, Shine–Dalgarno sequence facilitating initiation on initator AUG (marked as start) and known to stimulate frameshifting. stop, stop codon in B.firmus cdd.