Natural DNA uptake by Escherichia coli.

Escherichia coli has homologues of the competence genes other species use for DNA uptake and processing, but natural competence and transformation have never been detected. Although we previously showed that these genes are induced by the competence regulator Sxy as in other gamma-proteobacteria, no conditions are known that naturally induce sxy expression. We have now tested whether the competence gene homologues encode a functional DNA uptake machinery and whether DNA uptake leads to recombination, by investigating the effects of plasmid-borne sxy expression on natural competence in a wide variety of E. coli strains. High- and low-level sxy expression alone did not induce transformation in any of the strains tested, despite varying the transforming DNA, its concentration, and the incubation conditions used. Direct measurements of uptake of radiolabelled DNA were below the limit of detection, however transformants were readily detected when recombination functions were provided by the lambda Red recombinase. This is the first demonstration that E. coli sxy expression can induce natural DNA uptake and that E. coli's competence genes do encode a functional uptake machinery. However, the amount of transformation cells undergo is limited both by low levels of DNA uptake and by inefficient DNA processing/recombination.

Introduction

Many bacteria can actively take up DNA from their environment, a genetically programmed ability called natural competence [1]. When this DNA recombines and changes the cell's genotype, the cell is said to be transformed. Natural competence has never been directly demonstrated in Escherichia coli, and most transformation instead relies on artificial permeabilisation to bring DNA into cells [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12]. Though this DNA can recombine with the host chromosome using the cell's recombination machinery if RecBCD is disabled, higher levels of recombination can be attained by controlled expression of phage recombination proteins (recombineering) [2], [6], [13], [14], [15]. Several studies have shown modest ‘natural’ uptake of plasmid DNA by E. coli under specific conditions, but this appears to bypass the classical DNA uptake machinery (the type IV pilus, TFP) so its relationship to natural competence is unclear [16], [17], [18], [19], [20], [21], [22], [23], [24], [25]. Palchevskiy and Finkel showed that E. coli can grow in medium containing DNA as the sole source of carbon and energy and that this growth depends on competence gene homologues, suggesting that E. coli can take up DNA, but they could not detect chromosomal transformation [26], [27].

Despite not being naturally transformable, E. coli and other Enterobacteriaceae have homologues of all but one of the competence genes (pilF2) common to their nearest transformable relatives, bacteria in the families Pasteurellaceae and Vibrionaceae [28], [29]. In the best-studied species, Haemophilus influenzae (Pasteurellaceae) and Vibrio cholerae (Vibrionaceae), expression of competence genes is controlled by two positive regulators, CRP and Sxy (also known as TfoX) [30], [31], [32], [33], [34], [35], [36], [37], [38]. The CRP regulator (cAMP receptor protein, also called catabolite activator protein (CAP)) responds to the availability of carbon and energy sources signaled by cAMP; it activates transcription by bending DNA to enhance contacts with RNA polymerase [39], [40], [41]. Much less is known about Sxy's function; it is required for competence-specific transcription of a subset of CRP-dependent promoters (CRP-S promoters) and is thought to make direct contacts with CRP [29].

In H. influenzae and V. cholerae, Sxy is absolutely required for competence development, and the combination of sxy induction and CRP activation are sufficient to induce high levels of competence and transformation [30], [33], [34], [37], [38], [42]. We have previously shown that expression of E. coli sxy induces all of its competence-gene homologues, but we were unable to detect natural transformation [29]. This raises the question of whether E. coli's CRP-S regulon is truly a competence regulon (i.e. one that encodes machinery used for DNA uptake). In this work, we investigate the ability of E. coli Sxy to induce natural competence in a wide variety of strains, using assays of transformation and DNA uptake. We show that although sxy expression alone is not sufficient for natural competence, transformation can occur when recombination functions are provided artificially.

Results

sxy expression in different E. coli strains does not induce transformation

We previously showed that expression of plasmid-borne sxy in strain BW25113 sxy::kan strongly induces all of E. coli's competence gene homologues, but we were unable to detect natural transformation with marked chromosomal DNA [29]. Because extensive strain-to-strain differences in competence and transformability are the norm in other species [43], [44], we investigated the effects of sxy expression on transformation in a wide range of E. coli strains.

K-12 strains

The sxy plasmid used in our gene expression study (pEcsxy) [29] was transferred to the K-12 derivatives BW25113, C600 and W3110. IPTG was added to induce sxy expression in log phase, and cells were given marked chromosomal DNA before plating on selective and non-selective media. No transformants were ever recovered, despite varying (i) the duration of sxy induction/incubation with DNA (10, 30, 60, 90, 120, 180 minutes and overnight), (ii) the DNA concentration (1–20 ug/ml), (iii) the DNA type (chromosomal DNA or PCR product) and/or (iv) the DNA marker (purK::tet, crp::kan, fliC::kan, NalR point mutation). All combinations of assay conditions tested are listed in Table S2.

Sequencing confirmed that the plasmid-encoded sxy gene was free of mutations, and qRT-PCR confirmed that sxy had been strongly induced upon IPTG addition (Figure 1). The lack of transformation also could not be attributed to failure of Sxy and CRP to induce the CRP-S genes, as we confirmed transcription of ppdD and yrfD by qRT-PCR (Figure 1, black bars) and production of PpdD pilin by western blotting (Figure 2). Boosting CRP activity by addition of cAMP also had no effect.

Figure 1

Gene expression upon induction of plasmid-encoded sxy genes.

qRT-PCR quantification of changes in the expression of sxy, ppdD and yrfD upon sxy induction in strain BW25113 after 60 mins expression time. Each bar represents the average of two independent biological replicates for each gene, with error bars indicating standard deviation from the mean. Expression levels shown were normalised for each mRNA sample using 23S rRNA levels.

Figure 2

Expression of PpdD pilin protein upon induction of plasmid-encoded sxy genes.

Western blotting with PpdD antiserum was performed on whole-cell extracts from broth-grown BW25113 60 mins after IPTG induction of sxy expression. Lane 1: pEcppdD (positive control, His-tagged ppdD); Lane 2: pEcsxy; Lane 3: pEcsxy low. The positions of native processed (15 kDa) and His-tagged (17 kDa) PpdD proteins are indicated with arrows.

We also tested the effect of sxy expression in strain ZK126, which Palchevskiy and Finkel showed could use DNA for growth [26], [27]. As in other K-12 strains, sxy expression could not mediate natural transformation of ZK126 under any of the conditions described above (Table S2).

Other strains

This screen was further extended to natural populations of E. coli by investigating the ECOR collection, which is representative of this species' genetic diversity [45], [46]. None of the 72 ECOR strains produced PpdD pilin in overnight or log-phase cultures (data not shown), showing that CRP-S expression (and therefore likely sxy expression) is normally low in these strains. Attempts to transform pools of eight strains in overnight culture were also unsuccessful (Table S2). We transferred the sxy plasmid to 15 strains chosen to represent diversity in host species, pathogenicity, geographical distribution, subspecific grouping and MSLT type [45], [46]. Again, Sxy expression was never able to mediate chromosomal transformation.

Moderate sxy expression does not induce transformation

Since we had previously observed toxicity with long-term sxy expression from the high copy number vector pEcsxy, we reduced expression levels by recloning E. coli sxy in the low copy number vector pSU20 to generate pEcsxy low. qRT-PCR confirmed sxy expression from this plasmid, and showed induction of the CRP-S genes ppdD and yrfD (Figure 1, grey bars). This expression was lower than from pEcsxy, but still significant. Processed PpdD pilin could be detected by western blotting albeit at reduced levels (Figure 2), suggesting that the competence machinery could still be assembled. All of the transformation conditions described above were retested with pEcsxy low, but again sxy induction never led to chromosomal transformation.

Co-expression of sxy with H. influenzae pilF2 does not induce transformation

Although several genes of the H. influenzae CRP-S regulon are absent from the E. coli genome, only pilF2 is required for DNA uptake and transformation [28], [29], [47]. This gene is also present in V. cholerae and all transformable Pasteurellaceae. Since Pasteurellaceae are the closest relatives of Enterobacteriaceae, we cloned the H. influenzae pilF2 gene downstream of sxy in pEcsxy and pEcsxy low and tested whether co-induction of sxy and pilF2 permitted transformation in strain ZK126 under the conditions described above. Despite detectable plasmid-borne expression of both sxy and pilF2 from both plasmids (Figure 3), no transformants were detected (Table S2).

Figure 3

Induction of plasmid-encoded sxy and pilF2 genes in strain ZK126.

PCR reactions on cDNA (lanes 1 to 4) confirm that both sxy (515 bp; top) and pilF2 (265 bp; bottom) are strongly expressed when IPTG is added. As a control for DNA contamination of RNA preparations, PCR reactions were also performed on the RNA samples before reverse transcription. As expected, these reactions generated no product (lanes 4 to 8). Lanes 1 and 5: pEcsxy low; Lanes 2 and 6: pEcsxy low with IPTG; Lanes 3 and 7: pEcsxy; Lanes 4 and 8: pEcsxy with IPTG.

E. coli sxy expression does induce DNA uptake

The inability of diverse E. coli strains to be transformed could be due to a lack of DNA uptake across the outer membrane and/or to a defect in DNA processing. This was tested by direct measurement of uptake of radiolabelled chromosomal DNA fragments after sxy induction in BW25113 cells. For both pEcsxy or pEcsxy low, only 0.02% of the DNA added remained cell-associated after extensive washing (133 cpm per 109 induced cells). This baseline level was higher than background (17 cpm in tubes with no cells), but was not significantly higher than in uninduced cells (107 cpm per 109 cells). When 200 bp DNA fragments were used as donor DNA, levels of cell-associated cpm were not significantly above background (data not shown).

Since DNA uptake at our detection limit would give substantial numbers of transformants in known naturally transformable bacteria, we developed a more sensitive indirect test of DNA uptake. We measured sxy-inducible transformation when recombination functions were provided by the lambda Red recombinase system, which efficiently recombines mutations flanked by short homologous sequences into the chromosome [14]. These experiments used SW102 cells, which carry a chromosomally-encoded Red recombinase [48]. The results of these assays are shown in Figure 4.

Figure 4

Results of indirect DNA uptake assays.

SW102 cells with or without sxy plasmids were transformed with 1 µg fliC::kan or fliC::spec DNA. Cells without plasmids were electroporated. In other cells, sxy was induced with IPTG. Lambda Red expression was induced by heat shock. none = no plasmid, sxy = high-copy sxy plasmid, sxy low = low-copy sxy plasmid, sxy,pilF2 = high-copy sxy-pilF2 plasmid, sxy,pilF2 low = low-copy sxy-pilF2 plasmid, vector = high-copy no insert plasmid, vectorlow = low-copy no insert plasmid.

In control experiments, SW102 cells were first heat shocked to induce the recombinase and then electroporated with the test DNA fragment, a KanR cassette flanked by 50 bp segments of chromosomal homology to the fliC gene (fliC::kan). The transformation efficiency was 284 cfu/µg DNA±160 (Figure 4). To test whether sxy expression could induce DNA uptake, we repeated the experiment in SW102 cells carrying pEcsxy or pEcsxy low, replacing electroporation with IPTG-induction of sxy (Figure 4). Both plasmids gave small but reproducible numbers of KanR transformants: 6±3 transformants/ug DNA for pEcsxy and 23±5 transformants/ug DNA for pEcsxy low. Transformant genotypes were confirmed by PCR. Negative control cells carrying an empty plasmid (no sxy), or cells with no heat shock (no recombinase) never gave any transformants.

To confirm that the sxy-dependent transformation we observed uses the Sxy-regulated type IV pilus machinery for DNA uptake, we repeated the assay in a strain lacking hofQ, which is predicted to encode the outer membrane secretin through which DNA enters competent cells [1], [20], [26], [27], [29], [49], [50], [51]. Because the hofQ deletion contains a KanR cassette, the donor DNA was changed to fliC::spec. While SW102 hofQ::kan cells electroporated with fliC::spec DNA gave 215±60 transformants/ug DNA, cells expressing sxy gave none (Figure 4), confirming that the DNA uptake we observed used the competence machinery.

Since transformation following sxy-mediated natural DNA uptake was lower than transformation following electroporation, we also tested whether transformation efficiency could be improved by co-expression of H. influenzae pilF2, again using plasmids pEcsxy-pilF2 and pEcsxy-pilF2 low. As shown in Figure 4, transformation efficiency was not improved.

These experiments confirm that E. coli can indeed take up small amounts of DNA using the DNA uptake machinery, and that this uptake is competence-specific and Sxy-inducible. Collectively, our experiments suggest that natural transformation is undetectable because artificial induction of sxy causes very little DNA to be taken up.

Discussion

E. coli can be transformed in the laboratory, but this requires artificial entry of DNA into cells. However E. coli should be naturally transformable because it has a set of inducible competence gene homologues in a functional regulon [29]. Here we show that sxy expression can lead to natural DNA uptake using the competence machinery. Nevertheless, the amount of DNA uptake is very low, and transformation only occurs when recombination functions are provided artificially, suggesting that recombination of incoming DNA itself is also inefficient. Recombination in E. coli may be limited by expression of the RecBCD exonuclease, though studies in H. influenzae have shown that transformation is not increased in recBCD mutants [6], [52], [53]. The combination of low uptake and limited recombination could explain why researchers have until now been unable to demonstrate natural competence in E. coli. A further hurdle is that none of the extensive conditions we and others have tested induce expression of sxy or the CRP-S genes it regulates [26], [29], [49], [50].

All competent species studied show extensive strain-to-strain variation in their ability to take up DNA and be transformed [43], [44]. This has caused some species initially classed as non-competent on the basis of one non-transformable isolate to be later re-classified when broader screening identified transformable strains. However none of the E. coli strains we tested could undergo chromosomal transformation upon sxy expression, so this is not a strain-specific problem.

Since E. coli's CRP-S genes do encode a functional DNA uptake machinery, natural conditions must exist under which higher levels of natural competence occur. Laboratory culture is a very unnatural condition, as is the IPTG induction of sxy expression, so it may be that other factors are required which are missing when Sxy is artificially induced. It is therefore important to find more natural conditions that induce sxy expression, which requires a better understanding of how it is regulated.

Insights into the regulation of sxy expression can be gained from studies in H. influenzae and V. cholerae, where sxy is controlled by both transcriptional and translational regulation [33], [36], [37], [54]. Transcriptional control is mediated by CRP. Like its homologues in H. influenzae and V. cholerae, E. coli sxy has a strong CRP binding site upstream of the transcriptional start site, strongly suggesting that CRP also induces sxy transcription in this species. Translational control in H. influenzae and V. cholerae occurs through the formation of a stem loop structure in sxy mRNA that limits the amount of translation [36], [37], [54]. This stem loop is thought to respond to competence-inducing signals, modulating the amount of Sxy protein cells make and thus controlling competence. We have mapped the transcriptional start site of E. coli sxy mRNA and found that transcription initiates at a G residue situated 115 bp upstream of the ATG start. Mfold simulations with E. coli sxy mRNA predict a stem loop structure very similar to those found in H. influenzae and V. cholerae. Moreover, Yamamoto et al. identified strong inverted repeats in the promoter region of V. cholerae sxy, and showed that these were essential for transcriptional and translational control of sxy expression [36], [37]. Such regions of strong base pairing are also found in the region upstream of H. influenzae and E. coli sxys, suggesting that similar mechanisms of sxy regulation exist in E. coli and perhaps all gamma-proteobacteria.

The signals that control sxy expression and induce competence in H. influenzae and V. cholerae are known (purine nucleotide depletion and chitin availability, respectively). Finding the signals that control sxy expression in E. coli is crucial to finding out how this species becomes competent in its natural environment. This signal is likely to be something that E. coli encounters in vivo, so it is conceivable that different classes of E. coli might respond to slightly different cues. Clearly these conditions are not being met in the laboratory setting.

Materials and Methods

Bacterial growth conditions, strains, and plasmids

Details on the strains used in this study are given in Table 1. E. coli was grown on Luria Bertani (LB) broth or LB agar (1.2%) at 30°C or 37°C. Plasmids were introduced by CaCl2–heat shock transformation. When required, antibiotics were used at the following concentrations: kanamycin 15 µg/ml, chloramphenicol 20 µg/ml, nalidixic acid 20 µg/ml, ampicillin 100 µg/ml, tetracycline 10 µg/ml. Where required, isopropyl β-D-1-thiogalactopyranoside (IPTG) and cyclic adenosine monophosphate (cAMP) were used at 1 mM.

Table 1

Bacterial strains used in this study.

Strain name	Genotype	Source/Reference
C600	F- λ- tonA21 thi-1 thr-1 leuB6 lacY1 glnV44 rfbC1 fhuA1	[11], [59]
W3110	F- λ- rph-1 INV(rrnD, rrnE)	[60]
ZK126	W3110 ΔlacU169 tna-2	From Steven Finkel, USC Los Angeles, USA
SW102	F- mcrA Δ(mrr-hsdRMS-mcrBC) Φ80dlacZΔM15 ΔlacX74 deoR recA1 endA1 araD139 Δ(ara, leu)7649 rpsL nupG [λcI857 (cro-bioA) <> tet] ΔgalK	Recombineering strain [48]
NK6051	Hfr Hayes Δlac-proxIII thi-1 purK::tn10	From Stan Maloy, University of Illinois, USA
JW5702	BW25113 crp::kan	Keio collection [57]
JW1908	BW25113 fliC::kan	Keio collection [57]
JW3354	BW25113 hofQ::kan	Keio collection [57]
BW25113-NalR	BW25113 NalR	[29]
SW102 hofQ::kan	F- mcrA Δ(mrr-hsdRMS-mcrBC) Φ80dlacZΔM15 ΔlacX74 deoR recA1 endA1 araD139 Δ(ara, leu)7649 rpsL nupG [λcI857 (cro-bioA) <> tet] ΔgalK hofQ::kan	This work
ECOR-10	phylogroup A, subspecies group I, isolated from human in Sweden	ECOR collection [45], [46]
ECOR-15	phylogroup B1, subspecies group I, isolated from human in Sweden	ECOR collection [45], [46]
ECOR-22	phylogroup A, subspecies group I, isolated from animal in Bali	ECOR collection [45], [46]
ECOR-31	phylogroup A, subspecies group II, isolated from animal in USA	ECOR collection [45], [46]
ECOR-34	phylogroup B1, subspecies group II, isolated from animal in USA	ECOR collection [45], [46]
ECOR-35	phylogroup D, subspecies group II, isolated from human in USA	ECOR collection [45], [46]
ECOR-37	phylogroup E, subspecies group II, isolated from animal in USA	ECOR collection [45], [46]
ECOR-42	phylogroup E, subspecies group II, isolated from human in USA	ECOR collection [45], [46]
ECOR-47	phylogroup D, subspecies group II, isolated from animal in New Zealand	ECOR collection [45], [46]
ECOR-48	phylogroup D, subspecies group II, isolated from human in Sweden	ECOR collection [45], [46]
ECOR-55	phylogroup B2, subspecies group III, isolated from human in Sweden	ECOR collection [45], [46]
ECOR-65	phylogroup B2, subspecies group III, isolated from human in USA	ECOR collection [45], [46]
ECOR-67	phylogroup B1, subspecies group III, isolated from animal in Indonesia	ECOR collection [45], [46]
ECOR-72	phylogroup B1, subspecies group III, isolated from human in Sweden	ECOR collection [45], [46]

Strain SW102 hofQ::kan was constructed using recombineering to transfer the mutation from the Keio collection strain JW3354 to the chromosome of strain SW102 [48]. Plasmid pEcsxy contains the coding sequence of E. coli sxy, cloned downstream of the p-T5-lac promoter in the high-copy number vector pCA24N [55]. To create plasmid pEcsxy low, an 800 bp segment flanking and including the coding sequence of E. coli sxy (b0959) was PCR-amplified from the genome of strain BW25113 using primers with integrated ClaI and EcoRI restriction sites (all PCR primers are listed in Table S1). The PCR product was digested with ClaI and EcoRI, and cloned into the low-copy number vector pSU20 digested with the same enzymes, putting sxy under control of the lac promoter [56].

To create plasmids pEcsxy-pilF2 and pEcsxy low-pilF2, the H. influenzae pilF2 gene (HI0366) was first PCR-amplified and cloned into pGEMT-Easy (Promega) to generate pHipilF2. The pilF2 insert was then released from pHipilF2 by NotI or EcoRI digestions, and cloned into the NotI site of pEcsxy to generate plasmid pEcsxy-pilF2, or the EcoRI site of pEcsxy low to generate plasmid pEcsxy low-pilF2. In both plasmids pilF2 is cloned in the forward orientation downstream of E. coli sxy and under control of the lac promoter.

Transformation assays

Overnight cultures of strains carrying sxy-expressing plasmids were diluted 1∶100 in LB-Cm and grown at 37°C. At OD600 0.3, sxy expression was induced with IPTG and 1–20 ug/ml donor DNA was added. The donor DNA was either E. coli chromosomal DNA or purified PCR product with one of the following genotypes: purK::tet, crp::kan, fliC::kan or BW25113-NalR (see Table 1). Where necessary, cAMP was added at the same time as IPTG. Cells were incubated with DNA for 10, 30, 60, 90, 120 or 180 minutes, or overnight, before being plated on selective and non-selective media. The full list of combinations tested is given in Table S2.

DNA uptake assays

BW25113 chromosomal DNA was digested with XbaI and EcoRI and end-labelled with Klenow using 33P-dATP. Cultures of strain BW25113 carrying pEcsxy or pEcsxylow were grown at 37°C and sxy expression was induced at OD600 0.3 for 30 minutes. Cells were then incubated with 3 µg/ml 33P-labelled DNA for 30 minutes at 37°C, pelleted and washed twice with LB, and finally resuspended in LB for scintillation counting.

Indirect DNA uptake assays

These assays used the recombineering strain SW102, which contains chromosomally encoded exo, bet and gam genes from phage λ, controlled by the inducible cI857 promoter [48]. In positive control assays, overnight cultures were diluted in 3 mL LB and grown at 30°C to OD600 0.6, when recombination functions were induced by transfer to 42°C for 15 mins. Cells were then pelleted, washed three times with ice-cold dH20, concentrated 100-fold and electroporated with 1 µg donor DNA. Electroporated cells were allowed to recover in 1 mL LB for 1 h at 30°C before plating.

Assays testing the effect of sxy on DNA uptake omitted the electroporation step. Instead, sxy expression was induced with IPTG at OD600 0.3 and cells were given 1 µg/ml DNA. When the OD600 reached 0.6, recombination functions were induced by heat shock as above. Cells were then returned to 30°C and allowed to grow for a further 1.5 h at 30°C before plating. The donor DNAs used in these experiments were ΔfliC PCR products amplified from either the chromosome of the Keio strain JW1908 (fliC::kan) (Table 1) [57] or plasmid pRSM2832 (fliC::spec) [58].

Gene and protein expression methods

Expression of plasmid-borne sxy and pilF2 genes was confirmed by RT-PCR (primer sequences are given in Table S1). For each sample, 1 µg RNA was reverse-transcribed with iScript™ (Biorad), and 1 µL cDNA was used in PCR. Quantitative PCR was performed as described in Cameron and Redfield [28], using primers specified in Sinha et al. [29] (listed in Table S1). Protein sample preparation and immunoblotting were performed as described in Sinha et al. [29].

Primers used in this study.

(PDF)

Click here for additional data file.

Conditions tested in transformation assays. None gave any transformants.

(PDF)

Click here for additional data file.