Introduction

Bacteria with flagella, such as Escherichia coli, have a competitive advantage for moving toward favorable conditions and for avoiding detrimental environments (Fenchel, 2002). The flagellum is a complex organelle that also has an important role in adhesion, biofilm formation and virulence (Pratt and Kolter, 1998; Wood et al., 2006). However, the operation and synthesis of flagella requires significant energy; for example, synthesis involves more than 50 genes at a cost of 2% of the biosynthetic energy in E. coli (Soutourina and Bertin, 2003). Hence, coordinating growth and flagella synthesis is necessary. As a result, flagella synthesis is a highly ordered cascade, and on top of the hierarchy, is the master regulator FlhDC required for the expression of all other genes of the flagella regulon (Liu and Matsumura, 1994). The operon that encodes FlhDC, flhDC, is one of the most highly regulated loci in the genome and the target for regulation by many environmental factors (Soutourina and Bertin, 2003). The flhDC promoter contains many transcription factor-binding sites, including those for global regulatory proteins, such as H-NS and the catabolite gene activator protein complex (Bertin et al., 1994; Soutourina et al., 1999).

Activating genes by insertion of motile DNA elements has been shown to provide benefits to the host under stress by facilitating adaptations to severe environments (Hall, 1998; Petersen et al., 2002; Zhang and Saier, 2009a). As an example, the well-studied bgl operon encoding the gene products for the fermentation of β-glucoside depends on the transposition of insertion elements IS1 or IS5 to become active (Schnetz and Rak, 1992; Hall, 1998). Also, the glycerol utilization operon (glpFK) can be activated by IS5 when it is inserted upstream (Zhang and Saier, 2009a). For insertion of IS5 into both the bgl and glpFK operons, the environment influences the mutation rate. In addition, in biofilms, some cells increase their mutation rate during stress by increasing competence during chronic infections (Ehrlich et al., 2010). Hence, these examples demonstrate a Lamarckian-type evolution (that is, environment-driven mutation) that differs from that of Darwin, which holds that mutation rates should not depend on the environment (Koonin and Wolf, 2009).

In recent years, insertion elements IS1 and IS5 have been identified in some E. coli strains that activate the promoter of flhDC and increase motility (Barker et al., 2004; Gauger et al., 2007); however, these mutations were rejected as cases of directed mutation (Barker et al., 2004). Here, we explored the regulation of flhDC by IS5 hopping under different environmental conditions and found that the environment influences the IS5 insertion upstream of flhDC. Hence, we provide the first example of Lamarckian-type evolution for an active locus.

Materials and methods

Bacterial strains, plasmids and growth conditions

The bacterial strains and plasmids used in this study are listed in Table 1. All experiments were conducted in Luria–Bertani (LB) medium (Sambrook et al., 1989) or M9C medium (Rodriguez and Tait, 1983) at 37 °C, except where indicated. Kanamycin (50 μg ml−1) was used for culturing strains with single-gene knockouts from the Keio Collection (Baba et al., 2006), chloramphenicol (30 μg ml−1) was used for maintaining pCA24N-based plasmids, and ampicillin (100 μg ml−1) was used to maintain plasmid pKD46.

Table 1 Escherichia col i bacterial strains and plasmids used in this study

Swimming motility assay

Cell motility was examined as described previously on low-salt, soft-agar plates (1% tryptone, 0.25% NaCl and 0.3% agar), where the wild-type BW25113 is motile (González Barrios et al., 2006). High-salt, soft-agar plates were also used, in which wild-type BW25113 is non-motile due to the high salt concentration (Wang et al., 2009).

Curing IS5 from flhD+ Ω IS5 by a two-step method

Owing to the stable residence of IS5 after insertion into the flhDC regulatory region, curing IS5 via natural excision was not successful. Instead, we used a two-step method to replace the flhDC regulatory region of flhD+ Ω IS5 with that of the wild-type strain. To facilitate the screening for the right replacement, we first deleted flhD from flhD+ Ω IS5 using P1 transduction to transfer the ΔflhD KmR mutation from BW25113 flhD (Table 1) to make a non-motile strain, BW25113 flhD Ω IS5; this strain was verified by its complete lack of motility on low-salt, soft-agar plates. Then, the IS5 element was removed and the flhD+ gene was restored using the one-step inactivation method with pKD46 (Datsenko and Wanner, 2000) by replacing the flhD Ω IS5 region with PCR products amplified from the BW25113 wild-type genomic DNA covering the IS5 insertion site and the whole coding region of flhD+ using primer set PflhDC-4 (PflhDC-F4 and PflhDC-R4) (Table 2). Transformed cells were inoculated into low-salt, soft-agar plates, and cells with restored motility were selected and sequenced to verify the restoration of the wild-type flhD+ gene and its promoter region.

Table 2 Oligonucleotides used for PCR and qPCR in this study

Crystal violet biofilm assay

Biofilm formation was assayed in 96-well polystyrene plates using 0.1% crystal violet staining (Corning Costar, Cambridge, MA, USA) (Fletcher, 1977). Briefly, each well was inoculated at an initial turbidity at 600 nm of 0.05 and grown without shaking for 8 h in LB and M9C medium. Biofilm formation was normalized by the bacterial growth for each strain (turbidity at 620 nm). Two independent cultures were used for each strain.

Glass wool biofilms

Overnight cultures of the wild-type strain were inoculated into 250 ml LB with 10 g glass wool (Corning Glass Works, Corning, NY, USA) (Ren et al., 2004) to reach a final turbidity 0.05 (600 nm). The cultures were incubated for 8 h, 12 h, 15 h, 24 h and 39 h at 37 °C during which time the cells formed a biofilm on the surface of the glass wool. At each time point, planktonic cells in contact with biofilms were taken directly from the culture. Glass wool was taken from the culture and was rinsed twice with 100 ml 0.85% NaCl, and biofilm cells were collected after 2 min of sonication of the glass wool with 200 ml 0.85% NaCl in a bath sonicator (Fisher Scientific, Pittsburgh, PA, USA). A population of 108 cells (OD1) was inoculated onto a soft-agar plate for the motility assay.

Quantitative PCR (qPCR)

qPCR was performed using the StepOne Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). In all, 100–200 ng of genomic DNA was used for the qPCR reaction using the Power SYBR Green PCR Master Mix (Applied Biosystems). Primers were designed using Primer3Input Software (v. 0.4.0; Rozen and Skaletsky, 2000) and are listed in Table 2. The number of chromosomes that lack IS5 upstream of flhD+ was quantified using primer set PflhDC (PflhDC-F2 and PflhDC-R3), and the number of chromosomes with IS5 upstream of flhD+ was quantified using primer set PflhDC Ω IS5 (PflhDC-F2 and PflhDC-IS5-R). To quantify the total copy number of IS5 elements in the chromosome under different growth conditions, the chromosome number was quantified using primer sets purA and metG (Table 2), which amplify single-copy genes purA and metG, respectively, and the copy number of IS5 was quantified using primer sets IS5-1 and IS5-2 (Table 2), which amplify two different regions of the IS5 gene. The binding efficiencies of the six sets of primers were tested by varying concentrations of templates to generate a standard curve (Pfaffl, 2001), and the templates used were genomic DNA of the wild-type strain for primer sets PflhDC, purA, metG, IS5-1 and IS5-2, and genomic DNA of the flhD+ Ω IS5 strain for primer set PflhDC Ω IS5.

Results

IS5 hops upstream of flhD+ only when cells swim on motility plates

While performing motility assays using a wild-type E. coli K12 BW25113 strain (which has low motility), we noticed that a temporal change occurs in the swimming pattern. When the wild-type cells were plated onto soft-agar plates (0.3% agar) after 8–12 h of incubation, a halo of 11–14 mm in diameter formed (Figure 1, Halo 1). Prolonged incubation (after 24 h) led to some cells swimming faster and forming a second halo with less cell density (Figure 1, Halo 2). The two halos were separated by a zone with very low cell density. Cells from Halo 2 were collected and sequenced in the regulatory region of flhDC; the sequencing results (GenBank accession JF342359) showed that an IS5 insertion element (1195 bp, accession J01735) (Kröger and Hobom, 1982) was upstream of flhD+ in these cells, but not for cells inside the original halo. IS5 was inserted in the opposite orientation into a 4-bp target site (5′-TTAA-3′) 96–99 bp upstream of the transcription start site of the flhDC operon, the same position described previously for the MG1655 fnr strain (Barker et al., 2004). These cells were named flhD+ Ω IS5. No insertion element was present upstream of flhD+ for the culture used to inoculate the soft-agar plate. Hence, on soft-agar plates, IS5 hops into the promoter region of flhDC.

Figure 1
figure 1

Swimming halo formed after 12 h (left) and 24 h (right) at 37 °C on low-salt, soft-agar plates. Halo 1 refers to the original halo formed by wild-type swimming cells, and Halo 2 refers to the outside halo formed by flhD+ Ω IS5 swimming cells.

To quantify the proportion of cells that have acquired IS5 during migration on motility plates, qPCR was conducted using a forward primer that binds upstream of the IS5 insertion site and to one of two reverse primers that bind either to the downstream chromosome region without IS5 or bind inside IS5. The copy numbers obtained from the two separate qPCR reactions were used to estimate the percentage of cells that acquired IS5 upstream of flhD+. Genomic DNA was isolated using cells collected from different regions of the swimming zone, and was used as templates for qPCR. Starting with a BW25113 wild-type colony (lacks IS5 upstream of flhD+), in the first stage of migration (less than 12 h) on soft-agar plates (Figure 1, Halo 1), the proportion of cells with IS5 upstream of flhD+ was very low (10−7), which is similar to that of the culture used for inoculation. During the development of the second halo, 9±4% cells acquired IS5 after 24 h, and 21±1% cells acquired IS5 after 36 h. Therefore, within 24 h, a substantial fraction of cells within the second halo had IS5 upstream of flhD+.

The frequency of IS5 hopping was also tested on plates with the same nutrient content as soft-agar plates, but with higher amounts of agar (1.5%), which serves to immobilize the E. coli cells, and on regular LB-agar plates with more nutrients (1.5% agar), which also does not support motility (Copeland and Weibel, 2009). In both cases, after 12 h, 24 h, 48 h and 72 h, the frequency of IS5 hopping upstream of flhD+ was unchanged and remained very low (10−7, estimated by qPCR). Thus, there was a 104-fold induction of IS5 hopping upstream of flhD+ on soft-agar plates compared with IS5 hopping on hard-agar plates. In contrast, when cultured in LB liquid medium or in minimal medium, the frequency of IS5 hopping upstream of flhD+ for planktonic cells remained low (10−7) and did not change for 11 days (200 generations in LB and 100 generations in minimal medium). This occurrence of a motile sub-population of cells was also reported for MG1655 fnr from an inoculum of poorly motile MG1655 fnr on soft-agar plates, and PCR screening confirmed that IS5 inserted into the flhDC regulatory region in the same manner (Barker et al., 2004). The sequenced K12 strain, MG1655, is a relative of BW25113 (Baba et al., 2006), and is highly motile because of a similar IS1 insertion sequence upstream of flhD+ (Blattner et al., 1997; Gauger et al., 2007).

Clearly, IS5 hopping upstream of flhD+ occurred at very different frequencies on plates with different agar concentrations. It is well known that for soft-agar plates (agar <0.4%) the pore size of the gel is larger than the bacterial cells, allowing them to penetrate into the polymer network and swim (Copeland and Weibel, 2009). While on hard-agar plates (agar >1.5%), cells grow into non-motile colonies and retain their planktonic morphology and phenotype. Cells isolated from the edge of the soft-agar plate have distinctively more flagella than cells from a hard-agar plate (Copeland and Weibel, 2009). Increased motility is thus an advantage on soft-agar plates because it allows cells to reach more nutrients. Hence, we conclude that soft-agar plates serve as a type of environmental cue, which promote cell motility by inducing IS5 hopping into the upstream region of flhD+.

IS5 does not hop upstream of flhD+ in flhD and flagella structure mutants

To determine whether IS5 hopping is dependent on the ability to swim, we plated cells with an flhD deletion, but which contain the upstream regulatory region and insertion site for IS5. As expected, the flhD cells did not swim or form a halo on mobility plates after 24 h, 48 h or 72 h incubation. All cells were collected from soft-agar plates and checked for the presence of IS5 by qPCR, and no cells had IS5 in the regulatory region of flhDC. Hence, motility is required for IS5 insertion into the regulatory region of flhDC.

To determine whether IS5 insertion relies on FlhD or relies on an active flagella component and rotor, the frequency of IS hopping was also tested for strains that lack essential components of the flagella apparatus. MotA is one of the two membrane proteins that form the stator of the flagellar motor (Dean et al., 1984), and FlgK is a hook-associated protein (Komeda et al., 1978). Critically, after 24 h and 36 h of incubation on low-salt, soft-agar plates, no IS5 was detected in strains with these two mutations. Hence, we conclude that IS5 hopping into the regulatory region of flhDC requires active motility.

This conclusion is corroborated by results with mutations in the global regulator H-NS (Soutourina et al., 1999). When we plated the hns mutant on soft-agar plates, as expected, the hns mutant was non-motile due to a reduced expression of flhDC and a subsequent lack of flagella biosynthesis (Bertin et al., 1994; Soutourina et al., 1999). However, after long incubation times (24 h) with the hns mutant, different from the non-motile flhD, flgK and motA mutants, a halo was formed that contained highly motile cells. These cells were found to have the IS5 insertion upstream of flhD+. Hence, non-structural mutations in the motility regulon permit IS5 hopping into the upstream region of flhD+ on agar plates that promote motility.

IS5 insertion upstream of flhD+ increases motility and biofilm formation while it decreases growth

The flhD+ Ω IS5 cells formed significantly more biofilm after 8 h than wild-type cells (5±1-fold more, Figure 2a) and have much greater motility (7±2-fold greater more, Figure 2b). All cells from four different spots of the outside halo (Figure 1, Halo 2) had the same IS5 insertion upstream of flhD+ and had the same motility. The specific growth rate of the cells with IS5 in flhD was slightly lower, and the cells had a lower yield (6±1% reduction) (Figure 2c). This result is consistent with an earlier study, which shows that MG1655 (IS1 inserted upstream of flhDC) grew more slowly than MG1655 that lacks IS1 upstream of flhDC due to the reduction in energy available for growth due to the enhanced motility (Gauger et al., 2007). For our work, when IS5 was cured from the upstream region of flhD+, growth, motility and biofilm formation were restored to the wild-type values (Figure 2), which demonstrates that the increase in motility, the increase in early biofilm formation, and the decrease in growth are due to the presence of IS5 in promoter of flhDC.

Figure 2
figure 2

(a) Early biofilm formation (8 h) at 37 °C in LB medium and M9C medium for the wild-type strain (white bar), strain flhD+ Ω IS5 that acquired IS5 upstream of the flhDC operon (gray bar), and strain flhD+ Δ IS5 in which IS5 was cured from strain flhD+ Ω IS5 (slash marks). Data are the average of 10 replicate wells from two independent cultures, and one standard deviation is shown. (b) Swimming motility of the three strains on low-salt, soft-agar plates after 12 h incubation at 37 °C. Three independent cultures were tested, and one representative image is shown. (c) Growth of the wild-type strain (open white circle), flhD+ Ω IS5 (filled gray circle) and flhD+ Δ IS5 (filled black triangle) in LB medium at 37 °C. Two independent cultures were used.

IS5 insertion upstream of flhD+ is stable and heritable

The stability of IS5 upstream of flhD+ was tested in planktonic cells harboring IS5 upstream of flhD+ using serial dilutions every 24 h for 11 days in LB medium and in minimal medium (M9C). On the basis of qPCR, IS5 is not removed from the upstream of flhD+ after 1, 5 and 11 days. This result indicates that once IS5 inserted into flhDC, it is stable and is not removed (it may make copies and insert into other areas of the chromosome, but it leaves a copy of itself). Moreover, wild-type cells were also assayed using the same conditions, and no cells acquired IS5 in the flhDC regulatory region after 1, 5 and 11 days. These results provide another line of evidence that IS5 hopping upstream of flhD+ occurs only in a motility-inducing environment.

IS5 insertion upstream of flhD+ increases diversity in biofilm cells

Flagella are important for initial attachment in biofilms (Pratt and Kolter, 1998), for the mature structure (Wood et al., 2006), and for the dispersal of biofilms (Kaplan, 2010). Here, we tested whether IS5 is inserted upstream of flhD+ during biofilm development by screening biofilm cells for increases in motility. In the presence of glass wool, planktonic cells were collected from the culture, and biofilm cells were collected from the glass wool by sonicating. Motility results indicated that a small proportion of biofilm cells (as early as 8 h) migrated much faster than the rest of the non-motile cells, which appeared as a bright dot in the inoculum ring (Figure 3a, left side of the plate). For the planktonic cells in contact with biofilms, a small proportion of highly-motile cells emerged after 15 h (Figure 3a); however, these motile cells are probably derived from the biofilms as no highly motile cells arise in shaking flasks without biofilms, and cells frequently detach from biofilms (Kaplan, 2010). After 15 h, 24 h and 39 h of incubation, the proportion of highly motile cells increased for both biofilm and planktonic cells in contact with biofilms. A total of 500 cells from biofilms after 24 h were screened on high-salt, soft-agar plates, and 2.2% of the cells had increased motility (Figure 3b). PCR screening for the insertion element upstream of flhDC for these cells showed that IS5 was inserted upstream of flhDC. Similar results were obtained for cells grown statically in 96-well polystyrene plates; cells with higher motility were found after 12 h of incubation (Figure 3c). Thus, we conclude that IS5 hopping in flhDC occurs during biofilm formation and creates a sub-population of cells with increased motility but slower growth.

Figure 3
figure 3

Emergence of highly motile cells (flhD+ Ω IS5) during biofilm development. (a) Biofilm (left) and planktonic cells in contact with biofilms (right) were collected at different time points from cultures with glass wool and plated on high-salt, soft-agar plates and incubated for 12 h. Cells with high motility were identified at these stages by the formation of swimming halos. (b) Screening of biofilm cells with high motility after 24 h with high-salt, soft-agar plates where cells were non-motile, unless IS5 was inserted upstream of flhD+. A total of 500 cells were tested, and only 108 cells are shown. (c) Emergence of highly motile cells (flhD+ Ω IS5) in biofilms at different stages in 96-well polystyrene plates. Planktonic cells in contact with biofilms (left) and biofilm cells (right) were collected at different time points and plated on high-salt, soft-agar plates and incubated for 12 h. Cells with high motility were identified at these stages by the formation of swimming halos.

IS5 hops upstream of flhD+, but not into other operons on motility plates and in biofilms

IS5 is present in multiple (10–23) copies in the chromosome of E. coli K12 (Deonier, 1996), and in K12 strain W3110, which is closely related to BW25113 (Jensen, 1993), IS5 is present at 23 copies (Umeda and Ohtsubo, 1990). To test whether the transposition of IS5 upstream of flhD+ is specific or due to a general increase in IS5 transposition during swimming, the copy number of IS5 in the chromosome was quantified by qPCR and compared with the copy number of purA and metG, which are present as single copies. For flhD+ Ω IS5 cells collected from outside halos on soft-agar plates, the estimated copy number of IS5 was 23.6±0.7, and for BW25113 wild-type cells growing planktonically, there were 22.5±0.8 copies of IS5. Therefore, the highly motile cells have only one additional IS5 element.

Moreover, upstream regions of bgl, ade, fuc and glpFK operons were also screened for the presence of insertion elements for flhD+ Ω IS5 cells collected from outside halos and for flhD+ Ω IS5 cells collected from 24 h biofilms using PCR screening around the insertion sites (four pairs of primers list in Table 2). Insertion of IS5 or IS1 elements into these operons that derepress gene expression have been reported under specific conditions (Hall, 1998; Petersen et al., 2002; Zhang et al., 2010). As expected, based on the qPCR results, no IS element was found in any of these operons that corroborates that IS5 insertions occurred only upstream of flhD+. In addition, we also quantified the copy number of IS5 for all BW25113 cells collected from soft-agar plates after 12 h or 24 h incubation to obtain a population average, and the increase in copy number of IS5 in the population ranged from 0.3 to 1.5 when compared with cells growing planktonically after 24 h in LB, which corroborates again that the IS5 insertions occurred only upstream of flhD+.

IS5 hopping disrupts H-NS binding to the flhDC operon

In vitro transcription experiments showed that H-NS represses flhDC transcription at the AT-rich region that contains the insertion site for IS5 (5′-TTAA-3′) in cells without IS5 upstream of flhDC (Soutourina et al., 1999). We hypothesized that once IS5 is inserted into this region, it may lead to a loss of repression of flhDC by H-NS due to the disruption of the AT-rich region. As expected, we found that H-NS significantly repressed motility when overproduced in the wild-type strain without IS5 upstream of flhD+, which agrees with earlier studies (Soutourina et al., 1999). However, this repression of motility by H-NS was abolished in cells that had IS5 upstream of flhD+, for example, in flhD+ Ω IS5 and hns flhD+ Ω IS5. Hence, after its insertion, IS5 prevents H-NS from repressing motility.

Discussion

Here, we show that IS5 hopping into the promoter region of flhDC occurs at a high frequency only under environmental conditions that promote high flagella activity. It is well known that most transposition is not random, and hot spots have been identified where transposition events occur more frequently. IS5 has been found to be present in multiple (10–23) copies in the chromosome of E. coli K12 (Deonier, 1996) and shows a preference for 5′-YTAR-3′ (Y=C or T) (R=A or G) target sequences (Mahillon and Chandler, 1998). Upstream of the flhDC operon, there is one hot spot (5′-TTAA-3′) for IS5 insertion (Barker et al., 2004). However, we provide evidence that this IS5 hopping event is not solely due to the presence of a hot spot for IS5 in the upstream of flhDC because prolonged incubation of a mutated strain without flhD, flgK or motA did not increase IS5 hopping even though the hot spot upstream of flhDC remains for these three mutants. Another line of evidence is that prolonged incubation leads to increased IS5 hopping on soft-agar plates (swimming is allowed), but not on hard-agar plates (swimming is inhibited).

Once inoculated on agar plates, cells replicate, causing an increase in the population and cell density, until nutrients and resources become limited. For cells on soft-agar plates, when the cell density becomes limited by nutrients and wastes, the highly motile cells have an advantage when they migrate further to obtain nutrients and areas with reduced waste concentrations. Hence, for our experiments, a starvation condition led to IS5 hopping, which allowed cells to obtain a better environment. Thus, we reasoned that the increased transposition occurs when increased motility leads to an increase in fitness, and we propose that transposition of IS5 upstream of flhD+ represents a quasi-Lamarckian phenomenon, because the induced genetic change is beneficial to the organism. Just as Koonin and Wolf (2009) found for the prokaryotic CRISPR-cas system, the lines of evidence for IS5 hopping in a quasi-Lamarckian way rather than a Darwinian way are as follows: (i) hopping occurs in a motility-driven environment (soft agar) that leads to the insertion of the motile element, (ii) the resulting modification directly affects motility, which is the same cue that caused the modification, and (iii) the modification is adaptive and is inherited by the progeny of the cell that encounters the mobile element.

As determined by qPCR, the number of pre-existing, highly motile mutants in liquid culture was very low (less than 10−7) in a total of 107–108 wild-type cells that were used to inoculate the soft-agar plates. Clearly, selection of pre-existing mutants alone would not be expected to produce 104 more cells with high motility, which supports our conclusion that environment-induced mutations contribute more substantially to bacterial evolution under these conditions compared with selection of random pre-existing mutator strains. Moreover, general IS5 transposition was not induced on the soft-agar plates, as the increase in IS5 copy number was limited to basically one IS5 transposition per cell, under these conditions. However, the increase in the proportion of flhD+ Ω IS5 cells from 24 to 36 h on the soft-agar plates may be due to selective amplification of these highly motile cells.

IS5 has also been shown to behave as a transcriptional enhancer of the otherwise cryptic E. coli bgl operon to allow cells to utilize β-glucoside (Schnetz and Rak, 1992). IS5 is also capable of activating a metabolic operon glpFK to allow cells to utilize glycerol in the absence of the catabolite gene activator protein complex (Zhang and Saier, 2009b). In addition, the insertion of IS element in the cryptic ade promoter region results in relief of the H-NS-mediated silencing of ade operon, which allows the cells to utilize adenine as the sole source of purines (Petersen et al., 2002). However, in contrast to IS transposition into the ade, bgl and glpFK operons, expression of flhDC was not silenced before the transposition, and cells were motile in the absence of IS5 in its promoter (Figure 1, Halo 1); hence, the directed mutation here is fundamentally different in that the transcription of an active gene was increased.

In biofilms, we show that the transposition of IS5 upstream of flhD+ helps to activate flhDC expression and generates a proportion of cells with higher motility and slower growth. Consistent with an earlier study with Pseudomonas aeruginosa, where biofilm-grown cells exhibited more variation in swimming motility than did those from the inoculum (Boles et al., 2004; Boles and Singh, 2008), we also observed cells with increased motility and reduced growth arising in E. coli biofilms. Genetic diversity has been proposed as a beneficial feature of biofilms and has been reported in various species (Boles et al., 2004; Allegrucci and Sauer, 2007). The differences in swimming motility phenotypes found in P. aeruginosa biofilms were heritable and not produced by planktonic growth, and were dependent on RecA function (Boles et al., 2004). Here, we show that increased variation in motility and growth inside biofilms is caused by transposition of insertion elements, which results in a heritable change. Therefore, IS hopping is a novel mechanism for the production of genetic and phenotypic variations in biofilms.

We have shown previously that phage elements have an important role in generating diversity inside E. coli biofilms (Wang et al., 2009), and excision of e14, CP4-57, rac and CPS-53 prophages generate isogenetic strains with different phenotypes, including those of motility and growth (Wang et al., 2010). Induction of transposition in prokaryotes under cell-stress conditions is potentially important in creating diversity facilitating adaptation to stressful environments (Hall, 1998; Petersen et al., 2002; Zhang and Saier, 2009a), and IS5 hopping upstream of flhD+ is an adaptive mutation. Moreover, as IS5 is also found in E. coli bacteriophages (Kröger and Hobom, 1982), horizontal gene transfer might influence dissemination of IS5 elements to different strains of E. coli. Hence, horizontal gene transfer appears to be a form of quasi-Lamarckian inheritance (Koonin and Wolf, 2009) in prokaryotes, in which transferred genes confer selective advantages for growth in that environment. Here, we show clearly that cells can become highly motile to adapt to starvation conditions due to IS5 hopping and perhaps this is facilitated by horizontal gene transfer.