|Home | About | Journals | Submit | Contact Us | Français|
Excision of transposable genetic elements from host DNA occurs at low frequencies and is usually imprecise. A common insertion sequence element in Escherichia coli, IS5, has been shown to provide various benefits to its host by inserting into specific sites. Precise excision of this element had not previously been demonstrated. Using a unique system, the fucose (fuc) regulon, in which IS5 insertion and excision result in two distinct selectable phenotypes, we have demonstrated that IS5 can precisely excise from its insertion site, restoring the wild-type phenotype. In addition to precise excision, several “suppressor” insertion, deletion, and point mutations restore the wild-type Fuc+ phenotype to various degrees without IS5 excision. The possible bases for these observations are discussed.
Insertion sequences (IS) are discrete DNA segments that are usually ~1 kb in length, flanked by inverted repeats. They can “hop” from one part of the chromosome to another or from a plasmid to the chromosome. These “jumping genes” can transpose by at least two distinct mechanisms, one involving replication (1, 12, 30) the other being replication independent (12, 16). More than 500 insertion sequences have been identified to date, and they are commonly distributed in the genomes of both prokaryotic and eukaryotic organisms (11). Once having hopped into a gene or operon, insertion sequences can inactivate gene function and have polar effects (23). Hopping into appropriate sites in regulatory regions, however, can activate or silence gene expression (20, 22, 26). The distributions and frequent movement of IS elements are believed to be primary causes of extensive DNA rearrangements, including chromosomal inversions and deletions (18, 21, 25).
IS5 is a 1,195-bp IS element that is commonly present in strains of Escherichia coli and its relatives, with copy numbers varying between 0 and 23 per genome (5, 29). This element is unique in that hopping often provides the host with various benefits such as the opportunity for genetic changes that relieve a stress condition. Under starvation conditions, IS5 has been shown to activate the normally cryptic β-glucoside (bgl) (22, 26) and fucose/propanediol fucAO catabolic operons in E. coli (9, 15). Then the cells gain the ability to metabolize β-glucosides and propanediol, respectively. IS5 insertion upstream of the master flagellar switch operon (flhDC) has been shown to enhance expression of the operon, thereby elevating cell motility (2). This element has also been found to provide E. coli with resistance to nitroheterocyclic and nitroaromatic compounds by inactivation of the nfsB operon that encodes enzymes converting these compounds to toxic metabolites (19, 31).
Our recent studies revealed that E. coli cells lacking the cyclic AMP receptor protein (Crp) can mutate to utilize glycerol (Glp+) at rates greater than wild type (33). All Glp+ mutants proved to contain IS5 upstream of the glpFK regulatory region, and the presence of IS5 was both necessary and sufficient for activation of expression of the glpFK operon, whose gene products are required for aerobic growth on glycerol (33). Mechanistic details by which IS5 activates glpFK operon expression were elucidated (34).
From the standpoint of adaptive evolution, it would be expected that IS elements might be beneficially excised when the stress conditions are relieved, particularly if the presence of the IS element creates another stress. Some transposable elements have been shown to be excised both precisely and imprecisely, with precise excision normally occurring at much lower rates than imprecise excision (4, 6, 17). In one study, IS5 was found to be excised imprecisely, causing deletion of adjacent regions (28). No information has been available regarding precise excision of IS5.
The fuc regulon for l-fucose (Fuc) uptake and metabolism consists of two divergent operons, fucPIK and fucAO (Fig. (Fig.1).1). fucPIK encodes a permease (FucP) for fucose uptake, an isomerase (FucI) that converts fucose to fuculose, and a kinase (FucK) that phosphorylates fuculose to fuculose-1-P. In the fucAO operon, fucA encodes an aldolase that converts fuculose-1-P to lactaldehyde and dihydroxyacetone-P. Under aerobic conditions, lactaldehyde is further metabolized by a series of enzymes to pyruvate that subsequently enters the tricarboxylic acid (TCA) cycle. Under anaerobic conditions, lactaldehyde is reduced to l-1,2-propanediol (PPD) by FucO; PPD then diffuses out of the cell (7). Expression of both operons relies on FucR, the activator of the fuc regulon, which, once bound to fuculose-1-P, stimulates transcription (9). Expression of these two operons is also dependent on Crp (9).
E. coli cells cannot utilize propanediol for growth, since neither of the two fuc operons can be activated by this carbon source. However, when IS5 hops into the intergenic region between the two operons (Fig. (Fig.1),1), the cells gain the ability to utilize propanediol (8, 9). The presence of IS5 causes constitutive expression of the fucAO operon but prevents expression of the fucPIK operon. This latter fact results because the insertion causes the displacement, away from the control region, of a Crp binding site required for the expression of this operon. Utilization of l-fucose is thereby prevented, but the constitutive expression of the fucO gene, encoding the oxidoreductase that normally reduces lactadelhyde, enables the cells to grow on propanediol in an oxidative pathway (20).
We have used this system to explore the possibility of precise excision of IS5 from its insertion site. The rationale was that precise (but not imprecise) loss of IS5 should render cells able to grow on fucose. We not only demonstrated precise excision but also documented the occurrence of “suppressor” mutations that restore the wild-type Fuc+ phenotype without loss of the inserted IS5 element. These results and their potential significance are presented and discussed here.
As described above, wild-type E. coli cells can utilize l-fucose but not PPD as the sole carbon source for growth. We first isolated PPD+ mutants by incubation of E. coli K-12 strain BW25113 cells on minimal medium 9 (M9) agar plates with 1% PPD as the sole carbon source. The plates were incubated at 30°C and examined daily for the appearance of colonies. After ~8 days, colonies began to appear and continued arising thereafter. As a control, PPD+ cells could form visible colonies on the same plates within 2 days.
More than 50 such colonies were isolated independently (i.e., from different plates and appearing on different days) and purified. Their intergenic regions between fucPIK and fucAO were then amplified using a pair of primers as described in Table Table1.1. Gel electrophoresis showed that all PPD+ mutants contained an insert in this intergenic region. Sequencing analysis showed that the insert was IS5, that it was always located in the same position, and that it was always oriented with its 3′ end proximally upstream of the fucAO promoter.
The four-base IS5 recognition site, CTAG, was directly repeated immediately upstream of IS5 (Fig. (Fig.1)1) as expected. Growth on PPD proved to be independent of FucR since deletion of fucR did not affect the growth rate (data not shown). These results are consistent with those reported by Chen et al. (9) and Cocks et al. (10). In addition, we found that growth of these mutants on PPD was still Crp dependent and that they had lost the ability to utilize fucose.
To rule out the possibility that other mutations in addition to IS5 insertion had occurred, we first sequenced the entire fucAO operon of one PPD+ mutant and found that the coding region of this operon was unchanged. We then transferred the IS5:fucAO mutation from this PPD+ strain to a new wild-type (PPD−) strain by P1 transduction. Colonies (transductants) arose on PPD minimal plates within 2 days. (Note that spontaneous PPD+ mutations lead to formation of visible colonies only after 7 to 8 days.)
Twenty transductants were isolated, purified and examined for the presence of IS5 and growth on PPD. All PPD+ transductants contained the IS5 element in the same position and orientation as that in the donor PPD+ strain, and they grew on PPD equally well. We repeated these experiments using two other independently isolated PPD+ mutants, and the same results were achieved. Thus, the IS5 insert appears to be both necessary and sufficient for the observed growth of mutant cells on PPD.
Since insertion of IS5 in the fucPIK-fucAO intergenic region is the sole mutation that causes the PPD+ and Fuc− phenotypes, the wild-type phenotype (PPD− and Fuc+) should be restored if IS5 is lost precisely from this region. We therefore applied fresh cells (~5 × 107 cells/plate) of a PPD+ strain onto M9 agar plates with 0.5% l-fucose as the sole carbon source. The plates were incubated at 30°C and examined for the appearance of colonies daily for 12 days. Colonies (i.e., Fuc+ revertants) appearing during incubation were purified on plates containing M9 plus fucose, and their fucPIK-fucAO intergenic regions were PCR amplified. PCR products from some of the colonies are shown in Fig. Fig.2.2. Among 61 Fuc+ revertants analyzed, four revertants were found to have lost the insert between fucPIK and fucAO (Table (Table2).2). Sequence analyses showed that IS5 had precisely excised from all four of these revertants, yielding the wild-type nucleotide sequence. The remaining 57 “pseudorevertants” still carried the same-sized insert as discussed below.
The fucPIK-fucAO intergenic regulatory regions from 20 of the pseudorevertants were sequenced, and all were shown to be unaltered with respect to the IS5 element (same location and same orientation as in the original PPD+ mutant). Podolny et al. (20) reported that a point mutation in crp—K53N in the unprocessed gene product or K52N in the N-terminal methionine-less gene product—could restore the expression of fucPIK in a PPD+ genetic background. We sequenced the crp gene in these 57 Fuc+ pseudorevertants (Table (Table2).2). Twenty-five of them contained a point mutation in the same codon of the crp gene, giving rise to a change in amino acid 53 (K53) in Crp. Thirty-two of the pseudorevertants still possessed the wild-type crp gene. In addition to the K53N mutation observed by Podolny et al. (20), we found three other mutations at this position: K53T, K53Q, and K53I (Table (Table2).2). No mutation in the crp gene was found in the 4 true revertants that lost IS5. fucR and its surrounding regions were unchanged in both types of Fuc+ pseudorevertants (see below).
The 32 Fuc+ pseudorevertants lacking a point mutation in the crp gene were further analyzed by sequencing the fucPIK promoter region between the 3′ end of IS5 and the +102 position relative to the fucPIK transcriptional start site. All of them contained a mutation in this region. Based on the mutations, these 32 Fuc+ pseudorevertants could be divided into two groups.
The first group of pseudorevertants each contained a mutation in the small region from −10 to −35. In this group, 5 pseudorevertants contained a one-base (T) insertion between −24 and −25, changing the spacing between the −10 and −35 hexamers from 17 bp to 18 bp. An 18-bp spacing has been reported to be optimal for some operons (14, 24). It should also be noted that the increased expression level, as indicated by the increased growth rate, was minimal. Seven mutants contained a base change (G to T) at −12, yielding a better −10 element (from GAAAAT to TAAAAT).
The second group of pseudorevertants each contained an insertion or a deletion in the region between +1 and the start codon (ATG) of the fucP gene. In this group, 10 mutants contained a 1-bp (A) insertion between +11 and +12; six mutants contained a 7-bp (AGTTCAT) insertion between +10 and +11, which is actually a tandem 7-bp direct repeat of the adjacent sequence; and four mutants contained a 13-bp (TGAGTTCATTTCA) deletion between +1 and +15.
The 1-bp insertion between +11 and +12 results in the creation of a potential −10 hexamer (CATTAT) that is positioned 15 bp downstream of an excellent potential −35 hexamer (TTGAAA). A comparable potential −10 hexamer (CATAGT) is positioned 15 bp downstream of the same potential −35 hexamer, created by the 7-bp insertion. Finally, the 13-bp deletion creates a potentially strong promoter by moving another putative −10 element (AATATT) so that it is exactly 17 bp away from the same potential −35 hexamer. Thus, the same potential −35 hexamer (located from positions −14 to −9) may function in conjunction with three different −10-like elements in the three types of mutants, resulting in the creation of a new promoter in each case (Fig. (Fig.3A3A).
Using the BCM search launcher (NNPP; http://searchlauncher.bcm.tmc.edu/seq-search/gene-search.html), slightly different new promoters with excellent scores (0.99 to 1.00) at positions −123 through −74 relative to the fucP start codon (13-bp deletion), −145 to −96 (7-bp insertion), and −139 to −90 (1-bp insertion) were predicted. Using the SoftBerry BPROM program (http://linux1.softberry.com/berry.phtml?topic=bprom&group=programs&subgroup=gfindb), the same promoters were identified in the 13-bp deletion and the 7-bp insertion strains but not in the 1-bp insertion strain. These novel potential promoters might therefore provide additional explanations for the activation observed for fucPIK operon expression.
Using the CONTRAfold program (http://contra.stanford.edu/contrafold), a prominent presumptive secondary structural element (stable hairpin) was found between the transcriptional start site of the fucPIK operon and the start codon of the fucP gene (Fig. (Fig.3B).3B). In principle, this structure might inhibit transcription elongation (for example, by functioning as a transcriptional attenuator), or it might inhibit translation of the fucPIK mRNA. Regardless, an insertion or deletion within the predicted base-paired region would be expected to increase expression of the operon. Thus, this secondary structural element might exert an inhibitory effect on operon expression, and each mutation isolated could destabilize it, giving rise to operon activation. The other potential reason for the activation of the fucPIK operon by these deletions or insertions between +1 and the fucP start codon could be that these mutations have created new promoter or promoters that are FucR independent (see above).
One IS5-excised true revertant, one pseudorevertant with an altered Crp protein (K53N), one pseudorevertant (with a G-to-T substitution at −12) from group 1, and one pseudorevertant (with a 13-bp deletion between +1 and +15) from group 2 were selected for growth in liquid M9 medium plus 0.5% xylose or 0.5% l-fucose. When cultured in xylose medium, all strains tested grew equally well (data not shown), indicating that the IS5 excision event, point mutations in the crp gene, and the insertion/deletion mutations in the fucPIK control region did not affect bacterial growth on this carbon source.
When cultured in the fucose medium (Fig. (Fig.4),4), the IS5-excised true revertant grew at the same rate as wild-type cells, as expected. The pseudorevertant with the Crp(K53N) mutation also grew on fucose, but exhibited an ~4-h-longer lag phase than that of the wild-type cells (lag phase of ~8 h) before initiating growth. Its growth rate was also slower than that of wild-type cells (Fig. (Fig.4).4). The pseudorevertant with a G-to-T substitution grew on fucose slowly. The pseudorevertant with a 13-bp deletion between +1 and +15 grew on fucose even better than the wild type. The lag phase was ~4 h shorter, and the growth rate was higher than that of the wild type. As a negative control, the original PPD+ mutant did not grow detectably in 33 h in the same medium.
To determine if FucR is required for growth on fucose in wild-type and mutant strains, the fucR gene was deleted using the method of Datsenko and Wanner (13) in the wild type and some Fuc+ pseudorevertants. The fucR deletion mutant in the wild-type background lost the ability to grow on fucose as expected. However, after the fucR gene was deleted from all types of the Fuc+ pseudorevertants, growth on fucose was still observed. For Fuc+ pseudorevertants with a point mutation in Crp or a 1-bp substitution at −12, the loss of FucR decreased the growth rates by ~20% (data not shown). For Fuc+ revertants with a 13-bp deletion downstream of +1, the loss of FucR did not affect growth on fucose. These results show that expression of the fucPIK operon in these Fuc+ pseudorevertants is not or is only slightly dependent on FucR.
To determine if Crp is required for growth on fucose in wild-type and mutant strains, we used P1 transduction to replace the crp gene with a kanamycin resistance gene in a Fuc+ pseudorevertant with the Crp(K53T) mutation, a Fuc+ revertant with a 1-bp substitution at −12, and a Fuc+ revertant with a 13-bp deletion downstream of +1. These crp strains lost the ability to utilize fucose for growth, showing that Crp is essential for growth of all types of Fuc+ pseudorevertants on fucose.
It was expected that loss of Crp would abolish growth of Fuc+ pseudorevertants that resulted from point mutations in Crp since expression of the fucPIK operon is positively dependent on the modified Crp. However, in the absence of Crp, the lack of growth of the Fuc+ pseudorevertants that resulted from a 1-bp substitution at −12 or from the 13-bp deletion downstream of +1 in the control region may not be due to alterations of fucPIK expression, as expression of the other operon, fucAO, which is also required for growth on fucose, is dependent on Crp as well. Since the native Crp (plus FucR) cannot activate the fucPIK promoter when IS5 is present upstream of the fucPIK regulatory region (20), the fucPIK promoter in these Fuc+ pseudorevertants (with a 1-bp substitution at −12 or a 13-bp deletion downstream of +1) may still be active, regardless of Crp. If this is the case, the lack of growth of these mutants on fucose in the absence of Crp could be due to insufficient expression of the fucAO operon.
To determine if the point mutations in Crp are both necessary and sufficient for the restoration of growth of the PPD+ Fuc− cells on fucose, a native crp gene or a crp gene with the K53N point mutation was transduced into Fuc+ revertants with the modified crp gene deleted. Growth assays showed that transductants with the K53N Crp mutation, but not those with the native crp gene, recovered fucose utilization. Furthermore, the Fuc+ phenotype from all four point mutations in Crp could be transferred to a PPD+ Fuc− strain by P1 transduction. The transductants formed visible colonies on plates containing M9 plus fucose in 2 days. (Fuc+ colonies from wild-type cells appeared in ~8 days.) These results demonstrate that the point mutations in Crp are sufficient to restore growth of PPD+ cells on fucose.
Similar experiments were conducted using Fuc+ pseudorevertants with point mutations in the fucPIK promoter or insertions/deletions between +1 and the start codon for fucP. The Fuc+ phenotypes for each of these mutant types could be moved to another PPD+ Fuc− strain by P1 transduction. The transductants formed visible colonies on plates containing M9 plus fucose in 2 to 3 days. These results provide further evidence for the conclusion that these mutations are solely responsible for the positive growth phenotype.
All 61 Fuc+ revertants and pseudorevertants were tested for growth on plates containing M9 plus 1% PPD agar. As shown in Table Table2,2, those mutants with IS5 present were still capable of growth on PPD, while the four revertants with precisely excised IS5 lost the ability to utilize PPD, as expected. The same Fuc+ revertants previously examined for growth on fucose (Fig. (Fig.4)4) were tested for growth in liquid M9 plus 1% PPD. As shown in Fig. Fig.5,5, no growth was observed during a >40-h period of shaking for either the wild-type strain or the true Fuc+ revertants lacking IS5. Three other strains—the original PPD+ mutant, the Fuc+ pseudorevertant with the K53N Crp point mutation, the Fuc+ pseudorevertant with a mutation at −12, and the pseudorevertant with a 13-bp deletion downstream of +1—grew at the same rate, showing that the crp point and control region mutations have no effect on PPD utilization.
Using real-time PCR, we determined mRNA levels of fucP (the first gene in the fucPIK operon) in various types of Fuc+ pseudorevertants compared to the wild-type strain and the original PPD+ strain. Strains were cultured in LB with or without 0.5% l-fucose at 37°C. Real-time PCR was conducted as described previously (32). The results are summarized in Table Table3.3. Levels of expression of fucP were similar for the wild type and the IS5-excised true Fuc+ revertants. They exhibited low basal levels in the absence of fucose and an ~30-fold increase due to the presence of exogenous fucose. In the absence of fucose, the levels of fucP expression in the Fuc+ pseudorevertants with (i) the Crp(K53N) mutation, (ii) the 1-bp substitution at −12, and (iii) the 13-bp deletion downstream of +1 were 22-, 5-, and 39-fold higher than in the wild-type strain, respectively. This shows that the crp point mutations and the fucPIK control region mutations led to various levels of constitutive expression. In the presence of fucose, fucP expression increased a further ~1.5-fold for the pseudorevertant with the Crp(K53N) mutation and ~3-fold for the pseudorevertant with the 1-bp substitution at −12, suggesting that in these cells, activation of PfucPIK is still mildly dependent on FucR. In the presence of fucose, no further increase of fucP expression was observed in the Fuc+ pseudorevertant with the 13-bp deletion downstream of +1, suggesting that fucP expression is independent of FucR in this cell type. In the case of the original PPD+ strain, fucP was expressed at extremely low levels, regardless of the presence of fucose.
In summary, this paper reports four mechanisms by which PPD+ Fuc− cells harboring an IS5 in the control region can regain the capacity to utilize fucose. All deal with the expression levels of the two divergently expressed operons, fucPIK and fucAO. The first of the four mechanisms involves precise IS5 excision, the first case in which this phenomenon has been demonstrated. The second probably involves strengthening the promoter for fucPIK operon expression, in one case by improving the −10 element, in the other by increasing the spacing within the −10 and −35 region by 1 base pair. The third involves alterations of the downstream region between the +1 transcriptional start site and the fucP ATG start codon (Fig. (Fig.1).1). These mutations may activate fucPIK operon expression by destabilizing a secondary hairpin structure in the mRNA, which overlaps all of the downstream mutations, or by creating new promoter(s). The detailed mechanisms by which these mutations activate expression of the fucPIK operon will be the subject of another communication. The fourth mechanism involves changes in a single amino acyl residue in Crp, the K53 residue (K53N, K53T, K53Q, and K53I), which provided activation of gene expression by unmasking a cryptic activating region termed AR3 (3, 20, 27). These results provide information about the complex mechanisms of fucPIK operon activation and therefore a knowledge base for understanding novel pathways for selectable directed evolution.
In outline form, we have demonstrated that (i) IS5 can exactly excise; (ii) no other alteration except for IS5 integration is responsible for the PPD+ phenotype; (iii) Crp is required for growth of the PPD+ mutants on PPD; (iv) additional mutations are necessary and sufficient to allow growth of PPD+ Fuc− cells on fucose; (v) expression of the fucPIK operon is constitutive at various levels in all Fuc+ pseudorevertants and is largely independent of FucR; and (vi) the mutations in this region may enhance gene expression either by creating new promoters, by disrupting a potential inhibitory stem-loop structure in the RNA, or by another yet-to-be-defined mechanism.
Taken together, these findings provide an example of the amazing plasticity of a bacterial cell to genetically adapt to changing environments.
We thank Ann Hochschild for providing strain ECL56. Carl Welliver and Jeeni Criscenzo provided expert assistance in the preparation of the manuscript.
This work was supported by NIH grants GM 64368 and GM 077402.
Published ahead of print on 22 January 2010.