Escherichia coli Flagellar Genes as Target Sites for Integration and Expression of Genetic Circuits

E. coli is a model platform for engineering microbes, so genetic circuit design and analysis will be greatly facilitated by simple and effective approaches to introduce genetic constructs into the E. coli chromosome at well-characterised loci. We combined the Red recombinase system of bacteriophage λ and Isothermal Gibson Assembly for rapid integration of novel DNA constructs into the E. coli chromosome. We identified the flagellar region as a promising region for integration and expression of genetic circuits. We characterised integration and expression at four candidate loci, fliD, fliS, fliT, and fliY, of the E. coli flagellar region 3a. The integration efficiency and expression from the four integrations varied considerably. Integration into fliD and fliS significantly decreased motility, while integration into fliT and fliY had only a minor effect on the motility. None of the integrations had negative effects on the growth of the bacteria. Overall, we found that fliT was the most suitable integration site.


Introduction
The Gram-negative rod-shaped bacterium Escherichia coli K-12 is one of the most common microbes used for bioproduct manufacturing, metabolic engineering and as a chassis for Synthetic Biology devices [1][2][3][4][5]. Introduction of novel genes/ genetic circuits into E. coli on plasmids has many disadvantages, such as increased metabolic burden, variable copy numbers and the necessity for constant antibiotic selection pressure to maintain plasmid-borne genes [6,7]. Placing constructs in the bacterial chromosome can mitigate against these problems; however, chromosomal integrations can be highly variable in terms of integration efficiency and expression [8]. Moreover, there is limited information on the background and characterisation of integration sites. Identifying and characterising target loci for chromosomal expression is critical for reliable and rapid testing of genetic circuits. Since the expression of a gene is affected by its position in the chromosome [9,10], we looked for loci containing non-essential genes with high expression under common laboratory conditions. We selected the flagellar regions as putative integration sites based on RNA-polymerase profiling (Figure 1 and Figure S1) and the observation that they are located in highly expressed genomic regions that are enriched for interactions with DNA-binding proteins (highly expressed extended protein occupancy domains, or heEPODs) [9]. To test the suitability of chromosomal loci for expression of genetic constructs, we chose to integrate a simple genetic construct consisting of the gene encoding the thermosensitive l repressor cI857 and a constitu-tively active upstream promoter and ribosome binding site (RBS) elements from the Registry of Standard Biological Parts ( Figure  S2). The thermosensitive l repressor carries a mutation that leads to denaturation of the repressor protein upon a temperature shift from 30uC to 42uC. At 30uC the repressor protein negatively regulates gene expression from the bacteriophage l pR and pL promoters, but raising the temperature to 42 uC relieves repression allowing transcription from pR and pL [11]. We integrated this small genetic device into four promising sub-loci within flagellar region 3a. We characterised the sub-loci for efficiency of integration, level of transcription and expression. Finally, we propose fliT as standard site for the integration and characterisation of genetic devices.

Results and Discussion
Red recombinase and Gibson Assembly-based chromosomal integration strategy using pSB1K3(FRTK) In order to simplify and standardise our integration method, we modified the well-characterised high copy number plasmid pSB1K3 by introducing the kanamycin-resistance gene flanked by FRT sites from the original pKD13 plasmid [12] (Materials and Methods and Figure S2). The resulting plasmid pSB1K3(FRTK) can easily accept virtually any genetic circuit for integration in the E. coli genome. We tested the new vector by synthesizing a 913 bp insert (Repr-ts-1) consisting of the thermosensitive l repressor cI857, strong constitutive promoter (J23101), RBS (B0032) and double terminator (B0015) from the Registry of Standard Biological Parts. Isothermal Gibson Assembly was used to assemble Repr-ts-1 into pSB1K3(FRTK) to generate pSB1K3(FRTKr) ( Figure S2). Repr-ts-1 was modified with flanking sequences derived from four genes in the flagellar region (see below) to produce a series of constructs ready for integration into the E. coli genome.
Due to the low frequency of homologous recombination in E. coli, tools such as transposons, phages or more preferably phagederived elements are used for engineering this bacterium [12][13][14][15]. We employed the pKM208 plasmid-borne Red recombinase system of the bacteriophage l for integration of our DNA construct into the chromosome ( Figure 2). The Red recombinase system includes three proteins, namely Gam, Bet and Exo. Gam inhibits the host RecBCD exonuclease V so that Bet and Exo can gain access to DNA ends to promote homologous recombination [12]. Red recombinase system in pKM208 is controlled by the lacZ promoter and regulated by the addition of Isopropyl b-D-1thiogalactopyranoside (IPTG) to the growth media [16]. Our chromosomal integration protocol is simpler and more flexible than two other systems employing Red recombinase that have been developed recently. The first uses the yeast mitochondrial homing endonuclease I-SceI and requires a two step process for integration [15]. The second is based on knock-in/knock-out (KIKO) vectors and is restricted to integration at a selected few loci [14]. Other recent methods, such as 'the one-step clonetegration' [17] are capable of performing integration at similar if not faster rate but are limited in flexibility due to specific sequence requirements not present in subset of E. coli strains.

Flagellar genes as target sites for chromosomal integration
The lack of well-characterized and reliable chromosomal integration sites is a major limitation for E.coli engineering. Sabri et al have recently reported the successful integration of foreign DNA into three E. coli chromosomal loci, namely the arsenite transporter arsB, the ribose transporter operon rbsA-rbsR, and the lactose catabolic enzyme-encoding lacZ [14]. Identification of other target sites for reliable chromosomal integration would significantly aid progress. We chose the E. coli K12 MG1655 flagellar gene region 3a ( Figure 1) as a potentially suitable integration site for several reasons. First, the genes involved in flagella biogenesis and regulation are well-conserved in a wide variety of E. coli strains [10,18]. Second, flagellar genes are considered to be non-essential for E. coli [13], thus we hypothesized that integration into these loci woud not be lethal for the cell. Third, the flagellar genes are well characterized and the interactions between them have been studied extensively [10,18]. Fourth, suitable chromosomal integration sites during exponential growth should be located in chromosomal regions that show constitutive and high expression. RNA-seq and RNAP binding data obtained by Kahramanoglou et al [10] show that the flagellar genes are among the top 0.5% most highly expressed during E. coli exponential growth ( Figure 1 and Figure S1). Fifth, the flagellar genes are found in heEPODs [9].
We identified fliD, fliS, fliT, and fliY as putative chromosomal integration sites for our study. fliD encodes the cap that promotes filament-assembly [19], fliS and fliT encode specific export chaperones [18,20], while fliY encodes cystine-binding periplasmic protein [21]. BLAST search revealed that all four selected flagellar genes are conserved among commonly-used E. coli strains, such as MG1655, W3110, DH10B and BL21-DE3. The location of the flagellar regions on the E. coli MG1655 chromosome and the location of the four target genes are shown in Figure 1 and Figure  S1. These four genes are located in flagellar region 3a, which shows the highest levels of RNA polymerase binding (of the flagellar regions) at mid-exponential growth phase ( Figure 1 and Figure S1). This indicates that this region of the chromosome is easily accessed by RNA polymerase, suggesting that DNA integrated into fliD, fliS, fliT, and fliY would be highly expressed. To confirm this, the relative expression of fliD, fliS, fliT, and fliY was measured by RT-PCR ( Figure 3A). The expression at all four target genes showed higher expression (2-5 fold) compared to the mean expression of the housekeeping genes arcA and rpoD [22,23]. Notably, the expression at fliS was 2-3 fold higher than at fliD, fliT, and fliY ( Figure 3A).

fliT supports high efficiency integration
We integrated our DNA construct Repr-ts-1 into each of four open reading frames of fliD, fliS, fliT, and fliY. Integration of Repr-ts-1 into the target loci and successful flipping out of the kanamycin resistance cassette was verified by diagnostic PCR and DNA sequencing ( Figure S3). Interestingly, the integration efficiencies for Repr-ts-1 differed significantly for the four integration sites, despite their being located within the same ChIP-seq data from cells at mid-exponential growth phase [10]). E. coli flagellar genes are among the top 0.5% most highly expressed genes during exponential growth. The finding that the flagellar genes are among the top 0.5% of highly expressed genes in E. coli was derived from the RNA-Pol binding data obtained by Kahramanoglou et al [10]. Figure was generated by uploading the RNA-Pol binding data to the UCSC microbial genome browser for E. coli K12 MG1655 (http://microbes.ucsc.edu/cgi-bin/ hgGateway?db=eschColi_K12). doi:10.1371/journal.pone.0111451.g001 flagellar region. The integration into fliT yielded three times the number of recombinants per mg DNA than integration into either fliD or fliY, and six times more recombinants than integration into fliS ( Figure 4). The integration efficiencies do not appear to be operon-specific since fliD, fliS, and fliT are all part of the same operon, while fliY is part of a separate operon ( Figure S1). The differences in the DNA integration efficiency might be attributed to the variable secondary structures in the sequences adjacent to target loci; further explanation would require investigation of the local 3D architecture of the E. coli chromosome. This suggests that the integration efficiency for any given locus must be determined empirically.
Integration of DNA into flagellar genes reduces motility but does not inhibit cell growth For the purpose of characterising genetic constructs, target loci for chromosomal integration should typically not negativelly impact growth rate. Essential genes should therefore be avoided [24][25][26][27]. Integration of the DNA fragment into all four flagellar genes yielded viable transformants whose growth was not inhibited relative to the wild type ( Figure S4). The growth of the four engineered strains was impaired neither at the restrictive (37uC), nor permissive (30uC) temperatures for the cI857 repressor ( Figure  S4). Interestingly, strains with the integrated construct grew slightly better than the wild type strain ( Figure S4 and Table S2). The engineered strains with Repr-ts-1 integrated in the flagellar genes had reduced motility when compared to the wild type ( Figure S5). fliD and fliS mutants were most affected, while integration into fliT and fliY had only a minor effect on the motility ( Figure S5). This could result in a lower metabolic burden to the cell due to the disruption of flagellar function. This experiment confirms that the flagellar region is suitable for the integration of DNA into the E. coli chromosome.  Figure 3B). Expression from Repr-ts-1 was assessed qualitatively via the repressive activity of cI857. We constructed plasmid pSB1A1(GFP) harboring a GFP-expressing gene under the control of the cI857-repressible pR promoter and electroporated it into the E. coli strain containing chromosomally integrated Repr-ts-1. At 30uC GFP expression was inhibited but a temperature shift to 42uC triggered GFP expression ( Figure 3C and D). The GFP expression was confirmed both qualitatively by visual observation ( Figure 3C) and quantitatively by measuring GFP fluorescence over time ( Figure 3D). These experiments confirmed that functional cI857 was expressed from the integrated constructs. Notwithstanding the differences in transcription levels of cI857 integrated at the four flagellar loci (Figure 3B), the level of derepression of GFP expression did not seem to be affected ( Figure 3D).

Conclusions
We set out to provide the Synthetic Biology community with well-characterised integration sites for expression of genetic circuits from the E. coli chromosome. The flagellar region appears to be a good target for the integration of genetic circuits into the E. coli chromosome. The integration of Repr-ts-1 into four genes of the flagellar region, namely fliD, fliS, fliT and fliY did not have any negative impact on growth. The mutants grew slightly better than the wild type strain, possibly as a result of the lower metabolic burden due to the disruption of the flagellar function. Notably, the four targeted genes of the flagellar region differed significantly in the efficiency of integration and level of transcription. Our results show that for the flagellar genes analysed, fliT is the best candidate site because of the high frequency of integration and high relative transcription. We also present a modified method for generating chromosomal integrations in E. coli. The method combines Isothermal Gibson Assembly of DNA fragments [28] with the Red  Figure also shows that the integration of the repressor in the four flagellar genes had similar effect on the GFP expression. The starting absorbance (OD600) of the investigated strains was 0.05. Wt (wild type), wt+GFP (wild type transformed with the plasmid with the GFP located downstream of cI857-regulated pR promoter), rep+GFP (strain with the chromosomally-integrated cI857 transformed with the plasmid harboring pR controlled GFP). doi:10.1371/journal.pone.0111451.g003 system of the bacteriophage l [12]. It is simpler and more flexible than the previously developed procedures of chromosomal integration employing Red recombinase, such as the I-SceI and KIKO vectors-based method. The transcriptional profile of flagellar genes [10] is better than that of phage attachment att sites utilized by the 'clonetegration method' [17]. Furthermore, whilst phage attachment sites [29] are often missing from some E. coli strains due to genotype changes, flagellar genes are wellconserved. BLAST search revealed that the four targeted flagellar genes are present in in all commonly-used E. coli strains, including MG1655, W3110, DH10B and BL21-DE3. Our identification and characterisation of flagellar integration sites will facilitate rational design of Synthetic Biology devices in E. coli.

Bacterial strains, plasmids, and growth conditions
Plasmids and bacterial strains used in this study are listed in Table 1. Luria-Bertani broth (LB) was used for cultivating Escherichia coli strains. When required, LB media was supplemented with ampicillin (100 mg/ml) or kanamycin (50 mg/ml). All plate cultures were cultivated for approximately 24 hours at either 30uC, 37uC or 42uC. Liquid cultures of E. coli were grown in LB and incubated at 200 r.p.m. on a rotatory shaker at 30uC, 37uC or 42uC, depending on the requirements.

PCR amplification and DNA modification
Gel extraction and plasmid isolation were performed using standardized kits (Qiaquick Gel Extraction Kit and Qiaprep Spin Miniprep kit, respectively) from Qiagen according to the manufacturer's instructions. Oligonucleotide primers used in this study were synthesized by IDT. PCR amplifications were routinely performed in 50 ml volumes using Phusion DNA polymerase (Thermo Scientific) or Dream Taq master mix kit (Thermo Scientific) according to the supplier's instructions. Recombinant plasmids were confirmed by PCR amplification and sequencing.

Gibson Isothermal Assembly of DNA fragments
A modified Gibson Isothermal Assembly method [28,30] was used to clone target sequences into the plasmids. Briefly, 0.3 ml of the vector and 0.9 ml of the instert were added to 4 ml of the 1.336 Assembly Master Mix consisting of ISO Buffer, T5 exonuclease, Phusion polymerase, Taq ligase, and H 2 O to a final volume of 5.2 ml (http://www.srcf.ucam.org/,wac26/gibson/index.html). The assembly reaction was incubated at 50uC for 60 min and the whole volume was trasformed into competent E. coli.

Construction of pSB1K3(FRT)
Plasmid pSB1K3(FRTK) was constructed by cloning the kanamycin resistance marker flanked by directly repeated FRT sites (originally from pKD13) [12] into pSB1K3. This was performed by joining DNA fragments with 40 to 60 bp overlapping sequences employing Isothermal Gibson Assembly. The first of the assembled fragments consisted of the kanamycin resistance gene flanked by FRT sites, while the second comprised the pSB1K3 backbone.

Preparation and transformation of parental E. coli
Chemically competent E. coli have been prepared by a variation of the Hannah protocol using CCMB80 buffer [31], while electrocompetent E. coli were generated using the modified method of Miller and Nickoloff [32]. Briefly, parental E. coli strain K12 MG1655 was transformed with pKM208, recovered and selected on ampicillin plates at 30uC. 500 ml of LB with ampicillin was inoculated with overnight incubation (1:100 dilution) of parental E. coli K12 MG1655 with pKM208, grown at 30uC to OD 600 0.2 prior to addition of IPTG (1 mM), then grown to a final OD 600 of approximatelly 0.5. Cells were harvested by centrifugation, washed twice with 10% (v/v) glycerol and resuspended in a final volume of 100 ml of 10% glycerol per 100 ml of culture. 100 ml of the gel-purified PCR product   (approximatelly 5 mg of DNA) was electroporated into 100 ml of prepared electro-competent cells carrying pKM208. Electrotransformation was performed on a Bio-Rad micropulser with the settings recommended for E. coli by the manufacturer.

Selection of transformants and removal of antibiotic resistance gene
Transformants were selected on kanamycin plates at 37uC overnight. Kanamycin resistant colonies were incubated on LB plates without antibiotic at 42uC overnight to cure pKM208 plasmid and subsequently screened for kanamycin resistance and ampicillin sensitivity. Chromosomal integration was confirmed by diagnostic PCR with the flanking primers. Competent cells of the successful transformants were prepared by the method described above and transformed with the temperature sensitive plasmid pCP20 harboring FLP recombinase and selected on ampicillin plates at 30uC, overnight. Transformants were grown again on LB plates without antibiotic at 42uC overnight to cure pCP20 plasmid, tested for loss of kanamycin and ampicillin resistance and confirmed by PCR.

Qualitative verification of the integrated DNA construct
To confirm the integration of cI857 l repressor, constructed E. coli K12 MG1655 strain with integrated Repr-ts-1 was transformed with pSB1A1(GFP) which has GFP under the control of l promoter ( Figure 3C). Transformants were cultivated on plates at 30uC or 42uC.

Plate reader measurement of the GFP fluorescence and absorbance
The overnight cultures were diluted to OD 600 0.05 in a total volume of 200 ml and transferred into flat-bottomed 96 well plates. Greiner BioOne black plates and Sterilin Sero-Well clear plates were used for the measurement of the GFP fluorescence and absorbance, respectively. The plates were incubated in a Fluostar Omega fluorimeter (BMG Labtech) at 37uC and 30uC for the measurement of absorbance and at 30uC and 42uC for the measurement of the GFP fluorescence with an automatically repeated protocol. Absorbance measurement was performed with the following parameters (600 nm absorbance filter, cycle time 30 min, number of cycles 48, double orbital shaking at 500 rpm). GFP fluorescence measurement was performed with the following parameters (excitation filter 485-12, emission filter EM520, gain 1400, cycle time 30 min, number of cycles: 6 at 30uC followed by 21 at 42uC, double orbital shaking at 200 rpm).

RT-PCR expression analysis
Total RNA was isolated from strains grown into midexponential phase (OD 600 = 0.7) using Isolate II RNA Mini Kit (Bioline) and purified from genomic DNA contamination with the help of a TURBO DNA-free Kit (Applied Biosystems). cDNA was prepared from 1 mg of total RNA employing SuperScript III Reverse Transcriptase (Invitrogen). Quantification of cDNA targets was performed with a QuantiTect SYBR Green PCR Kit (Qiagen). A 7500 Fast Real-Time PCR System (Applied Biosystems) was used to measure the expression levels of the target DNA sequences in MicroAmp Fast Optical 96-Well Reaction Plates (Applied Biosystems). Primers for RT-PCR were designed employing Primer3 Software. The relative expression levels were quantified with the help of REST9 Software (Qiagen) employing Pfaffl method [33]. The experiments were carried out in triplicate and the means and standard errors were calculated.

Motility assay
Motility agar plates were prepared by pouring 100 ml of the motility agar (10 g tryptone, 5 g NaCl, 0.25% Bacto-Agar (Difco)) in the 13 cm plate and let to set overnight. Motility plates were pre-warmed at 37uC before the experiment. Overnight cultures of the tested bacterial strains were normalized to an absorbance (OD600) of 1.0 prior to spotting 2 ml of the cultures into the middle of the motility plates. Plates were grown for 4-6 hours at 37uC.

Sequence analyses and databases
DNA sequences were compared with the help of the National Center for Biotechnology Information (NCBI) website (http:// ncbi.nlm.nih.gov) using the BLASTN [34], TBLASTX algorithms and position-specific iterated BLAST (PSI-BLAST) [35]. The annotated E. coli K-12 genome was obtained from the E. coli K-12 project website (http://www.xbase.ac.uk/genome/escherichiacoli-str-k-12-substr-mg1655   Table S1. (TIF) Figure S5 Motility Assay. The engineered strains harboring integrated thermosensitive repressor in the flagellar genes had reduced motility when compared to the wild type (wt). Overnight cultures of the tested bacterial strains were normalized to an absorbance (OD600) of 1.0 prior to spotting 2 ml of the cultures into the middle of the motility plates. Picture was taken after 5 hours of incubation at 37uC. (TIF)