Adaptive Reversion of a Frameshift Mutation in Escherichia coli by Simple Base Deletions in Homopolymeric Runs

Abstract

Spontaneous mutations are thought to occur primarily in growing cells. However, spontaneous mutations also arise in nutritionally deprived cells, and in some cases this process appears to be adaptive. Here it is reported that when a Lac⁻ strain of Escherichia coli is under selection for lactose use, the spectrum of Lac⁺ mutations that arises is different, and simpler, than that arising without selection. Mutations appearing during selection were mainly one-base deletions in runs of iterated bases. Similar mutations occurring in repetitive DNA elements are associated with a variety of human hereditary diseases and are increased in cells that cannot correct heteroduplex DNA.

The mechanisms by which spontaneous mutations arise in growing cells have been the subject of much research. Two causes of spontaneous mutation that are often cited are intrinsic polymerase errors and endogenous DNA lesions. Indeed, many of the mutations that arise spontaneously in bacteria and in their viruses are the same types of errors as those made by DNA polymerases replicating damaged and undamaged templates in vitro (1, 2). However, it is unclear if such mechanisms can account for the mutations that arise among cells that are apparently not growing or replicating their DNA (3). Many theories have been proposed to explain how mutations occur in nongrowing cells and why those mutations are adaptive (3–11). Comparisons between the types of mutations that arise during nonselected growth and those that arise under selective conditions may help to choose among these theories (12).

In previous publications (7, 8), we described the appearance of Lac⁺ revertants among populations of a strain of Escherichia coli, FC40, that cannot use lactose because of a frameshift mutation affecting the lacZ gene. Lac⁺ revertants occurring during exponential growth, which were detected 2 days after cells were plated with lactose as the only carbon source, arose at about 10⁻⁹ per cell per generation, a rate that is well within the normal range for mutations of this type. After day 2, Lac⁺ revertants continued to arise at a nearly constant rate of about 10⁻⁹ per cell per hour. The post-plating mutants quickly became the main class, and 90 to 95% of the Lac⁺ revertants that had appeared after a few days were apparently the result of mutations that occurred after plating. This class of mutants did not appear if the cells were starved in the absence of lactose or in the presence of lactose if there was another, unfulfilled growth requirement (7). In addition, mutation to Lac⁺ during lactose selection, but not during prior growth of the cultures, required some function or functions of the major recombination pathway, RecABC (7, 11). In the work reported here, we determined the sequence changes in Lac⁺ revertants that arose each day after FC40 was plated on minimal lactose plates.

The mutant lac allele, lacI33, carried by FC40 on an F′ episome, derives from an in-frame fusion of the lacI gene to the lacZ gene (13) but has a +1 frameshift mutation in lacI that is polar on lacZ (14). Because the lacI coding sequence is not essential, this allele can be reverted by any mutation that restores the reading frame but does not create a nonsense mutation. Such events include simple −1-bp deletions as well as more complex DNA rearrangements that restore the reading frame within the 130-bp target shown (Fig. 1).

Fig. 1.

The target for reverting mutations in the lacI33-lacZ allele. Sites at which −1-bp deletion mutations were found more than once are indicated in bold. Numbering is as in the Ecolac sequence in Gen-Bank up to base pair 1144, but the extra C at base pairs 1036 to 1038, which creates the lacI33 allele, is not numbered. The coding strand is shown. The lacI-lacZ gene fusion (13) eliminates the last five residues of LacI, all of the lac promoter and operator, and the first 23 residues (or 24 if the initiating methionine is included) of LacZ (base pairs 1145 to 1356 of the Ecolac sequence). The fusion protein is transcribed from the mutant lacI^q promoter. The frameshift mutation at base pairs 1036 to 1038 was induced by ICR191 (14). This mutation creates a stop codon at base pair 1145 (base pair 1358 of the Ecolac sequence), and a −1 frameshift mutation upstream of base pair 1016 would create a stop codon at base pair 1016. Thus, to revert lacI33, the reading frame must be restored between base pairs 1016 and 1145.

Newly arising, independent Lac⁺ mutants of FC40 were collected on days 2 to 5 after the cells were plated on minimal lactose plates (15). The DNA from these mutants was analyzed by amplification and sequencing (16), and the results are summarized in Tables 1 and 2 and in Fig. 1. Nearly all of the Lac⁺ mutants that arose after day 2 carried −1-bp deletions, and most of these occurred in runs of three to five bases. In contrast, about 50% of the Lac⁺ revertants that were isolated on day 2 carried mutations that resulted in deletions, duplications, and rearrangements. Thus, complex mutations are frequent in the absence of selection but are rare during selection. Considering only these two classes of intragenic revertants, there were nine complex mutations and 11 −1-bp deletions among the mutants isolated on day 2 but only one complex mutation and 30 −1-bp deletions among the mutants isolated after day 2 (χ² = 10.9, P = 0.001). Indeed, the one late-arising deletion mutant, isolated on day 3, was only weakly Lac⁺ (see Table 3), which suggests that this complex mutation may also have arisen during prior growth of the culture but took an extra day to appear as a Lac⁺ colony.

Table 1.

Summary of Lac⁺ revertants.

Mutation	Day 2	Day 3	Day 4	Day 5	Days 3 to 5
Loss of bases*	5	1	0	0	1
Gain of bases†	4	0	0	0	0
−1-bp deletion	11	9	12	9	30
Extragenic	0	0	2	1	3
Total	20	10	14	10	34

^*Detected by a decrease in the size of the amplified product.

^†Detected by an increase in the size of the amplified product, except for one, a duplication of base pairs 1087 to 1094 (Fig. 1), that was detected by sequencing.

Table 2.

Number of occurrences of various frameshift mutations. Blank spaces indicate that no mutant was found.

Mutation*	Length of run	Day 2	Day 3	Day 4	Day 5	Days 3 to 5
−GC at 1020	3			1	1	2
−GC at 1036	4	4	7	8	5	20
−GCat 1039	1				1	1
−AT at 1056	5	2			2	2
−AT at 1067	1	1†
−GCat 1068	2		1			1
−GCat 1072	3	2		3		3
−AT at 1078	1	1
−GC at 1093	4	1
−GC at 1106	2		1			1

^*Only the first base pair in a run is given.

^†The culture contained a jackpot of 193 mutants per 10⁷ cells. A total of five mutants were amplified and three were sequenced, all of which were the same, which confirmed that the −AT frameshift was the mutation giving rise to the jackpot (25).

Table 3.

The β-galactosidase activity of various mutants. Lac⁺ mutants were grown to saturation in M9 0.1% minimal glycerol medium (7). The β-galactosidase activity was measured as in (26), and the results are given in Miller units. The numbers are for independently isolated mutants of each class. Blank spaces indicate either that no mutant was found or that none were assayed

Mutation*	Day 2	Day 3	Day 4	Day 5
Loss of bases	76; 180; 422	80
Gain of bases	73; 183; 272
−GCat 1020			269	283
−GCat 1036	223	240		265
−GCat 1039				285
−AT at 1056				182
−AT at 1067	225
−GCat 1068		165
−GCat 1072			127; 170; 241
−AT at 1078	216
−GCat 1093	135
−GCat 1106		165
Extragenic			53; 59	76

^*Only the first base pair in a run is given.

Because late-arising Lac⁺ colonies might have been the result of mutations that gave only a weak Lac⁺ phenotype, we measured the β-galactosidase activity of an assortment of mutants (Table 3). The mutant with the largest deletion had the greatest activity, 422 Miller units, which suggests that this deletion creates a strong promoter. Not surprisingly, the three extragenic mutants that were isolated had rather low β-galactosidase activity. Among the −1-bp deletion revertants, the levels of β-galactosidase activity were about the same—most ranged from about 200 to 300 Miller units regardless of the day on which they appeared.

These results limit the types of mechanisms that might be responsible for the Lac⁺ revertants of FC40 that arose during selection. Although we did not sequence regions outside the target region, we read approximately 200 bases of sequence for each mutant and found no sequence changes other than the ones that reverted the frameshift. In addition, the reverting mutations were found at seven different sites. The lack of silent mutations and the diversity of mutational sites argue against the theory that the late-arising mutations resulted from recombination with homeologous (similar but nonhomologous) sequences located elsewhere in the genome (10). A distinctive feature of the late-arising frameshift mutations is that 90% occurred in runs of three or more bases. This site specificity is typical of −1-bp frameshift mutations made by DNA polymerases in vitro (2) and can be explained by replication of a misaligned template (17). Thus, the late-arising revertants of FC40 might be the result of simple polymerase errors, which supports our previous conclusion, based on genetic evidence, that adaptive mutations require some form of DNA replication (8). In vivo, mutations at iterated sequences are greatly enhanced by the loss of mismatch repair functions (18), which raises the possibility, previously suggested (5), that in nutritionally deprived cells, error correction activity may be low, or that mispaired or misaligned DNA may be inaccessible to the error-correcting enzymes. Similar types of mutations, apparently due to slippage during replication, repair, or recombination of repetitive DNA elements, have recently been associated with a variety of human hereditary diseases (19).

The late-arising revertants of FC40 are a subset of the kinds of mutations that can revert lacI33, yet they are distinguished from the mutations that arise during nonselective growth by their requirement for RecABC (20). One straightforward explanation for this result is that − 1-bp deletions are produced by the same mechanism both before and during selection but that there is an extra requirement for some function or functions of RecABC in the latter case. For example, if the −1-bp deletions are the result of polymerase errors, RecABC may be required to initiate DNA synthesis, to preserve the products, or both. RecABC can initiate DNA synthesis by producing D-loops (21) and (perhaps by the same mechanism) by catalyzing unequal crossing-over between regions of homology (22). Both of these mechanisms would require that more than one copy of the lac region be present in a cell. Thus, the mutations might arise in a subpopulation that has duplicated the lac region (8, 23) or simply among cells (probably the majority) that have more than one copy of the episome. If DNA synthesis is initiated only infrequently or is restricted in extent, yet is error-prone, this would help to explain why the amount of DNA synthesis measured in non-dividing cells appears to be inadequate to account for the mutations that arise (3). If the products of this synthesis are transitory unless the cell achieves a useful mutation and begins to divide, this would explain why the only mutations that are recovered are adaptive (4, 7–9).