Synthetic Promoters Functional in Francisella novicida and Escherichia coli

Abstract

In this work, we describe the identification of synthetic, controllable promoters that function in the bacterial pathogen Francisella novicida, a model facultative intracellular pathogen. Synthetic DNA fragments consisting of the tetracycline operator (tetO) flanked by a random nucleotide sequence were inserted into a Francisella novicida shuttle plasmid upstream of a promoterless artificial operon containing the reporter genes cat and lacZ. Fragments able to promote transcription were selected for based on their ability to drive expression of the cat gene, conferring chloramphenicol resistance. Promoters of various strengths were found, many of which were repressed in the presence of the tetracycline repressor (TetR) and promoted transcription only in the presence of the TetR inducer anhydrotetracycline. A subset of both constitutive and inducible synthetic promoters were characterized to find their induction ratios and to identify their transcription start sites. In cases where tetO was located between or downstream of the −10 and −35 regions of the promoter, control by TetR was observed. If the tetO region was upstream of the −35 region by more than 9 bp, it did not confer TetR control. We found that three of three promoters isolated in F. novicida functioned at a comparable level in E. coli; however, none of the 10 promoters isolated in E. coli functioned at a significant level in F. novicida. Our results allowed us to isolate minimal F. novicida promoters of 47 and 48 bp in length.

INTRODUCTION

As synthetic biologists attempt to engineer the genomes of diverse species, there is a growing need for gene regulatory elements that function in species outside classic chassis organisms such as Escherichia coli and Saccharomyces cerevisiae. One approach to developing controllable promoters is to modify known, natural promoters so that they contain novel regulatory units. This has been done for promoters in a number of bacteria by inserting the sequence for a repressor protein binding site (operator) in close proximity to the core promoter region (1–5). In some cases, this has been successful; however, there are some drawbacks to this approach. In using natural promoters, one has to have knowledge of the exact limits of the promoter for precise engineering. Also, most natural promoters will be controlled by undefined multiple regulatory proteins, and this makes it difficult to predict how the promoter will function under different physiological conditions (6). Lastly, inclusion of a native promoter into a recombinant molecule could lead to a DNA construct integrating into the chromosome at the site of the promoter rather than at another, targeted site.

Francisella species are facultative intracellular bacterial pathogens that are found widely in nature (7). Many of the Francisella biotypes infect a wide variety of animals and humans and are extraordinarily infectious and virulent. Francisella novicida (alternatively called “F. tularensis subsp. novicida”) is normally noninfectious for humans but highly virulent in mice, and thus, it is often used as a research model for F. tularensis (8–10). These two species are closely related at the molecular level, and their nucleotide identity is about 98% (11, 12). All of the molecular tools developed in one species appear to function in the other.

Relatively little is known about the control of mRNA transcription and the nature of promoters in Francisella species. Analyses of genomic data from Francisella species have revealed that there are no complete two-component regulatory systems (13), there is only one alternative sigma factor, and there are two distinct alpha-subunits of RNA polymerase (14). The presence of two alpha-subunits is unusual and may be unique to Francisella (14). The two subunits appear to be expressed in about equal amounts, but it is not known if they associate as homo- or heterodimers.

Several studies provide evidence that promoters for antibiotic resistance cassettes that typically work in Escherichia coli and several other bacteria do not function in Francisella (15–17). For example, in one study, when investigators conducted transposon mutagenesis of F. novicida, they found that only insertions that had the antibiotic resistance gene oriented downstream of an F. novicida promoter resulted in antibiotic-resistant strains (18). The basic knowledge of Francisella gene regulation has allowed a few groups to develop systems to control Francisella protein production at either the transcription or translational level. Horzempa et al. showed that an endogenous promoter could be controlled by the addition of glucose (19). LoVullo et al. inserted the tet operator in the groEL promoter region and demonstrated transcriptional control by TetR (3). Finally, translation control was engineered into F. novicida and F. tularensis by using a riboswitch that was responsive to theophylline (20).

In this work, we describe the selection of constitutive and controllable promoters from a library of synthetic DNA molecules. We show that the strongest of these promoters have activity comparable to that of some of the strongest identified F. tularensis promoters. Synthetic promoters isolated in F. novicida functioned in E. coli with activity similar to that found in F. novicida; however, synthetic promoters isolated in E. coli did not promote transcription in F. novicida.

MATERIALS AND METHODS

Culture conditions and transformation of bacterial strains.

Unless otherwise indicated, E. coli strains were grown in modified LB broth (1% tryptone, 0.5% yeast extract, 0.5% NaCl) or on LB agar, and F. novicida strains were grown in tryptic soy broth (TSB) supplemented with 0.1% l-cystine (TSBC) or tryptic soy agar supplemented with 0.1% l-cystine (TSAC). Anhydrotetracycline (ATc) was used at 100 ng/ml, hygromycin B (Hyg) was used at 150 μg/ml, chloramphenicol (Cm) was used at 5 μg/ml for F. novicida and 25 μg/ml for E. coli, and 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-gal) was used at 20 μg/ml, as required. Transformation of F. novicida was done as described previously (21). Electroporation and chemical transformation of E. coli strains were done by using standard protocols (22).

DNA manipulations.

PCR was performed by using iProof high-fidelity DNA polymerase (Bio-Rad) for preparative PCR or with Taq DNA polymerase (NEB) for diagnostic PCR. Purification of DNA fragments was performed by using a NucleoSpin Gel and PCR Cleanup kit (Macherey-Nagel).

Strain and plasmid construction.

Bacterial strains and plasmids are described in Table 1. E. coli DH10B (Invitrogen) was used as the E. coli host for all cloning experiments. Reporter plasmid pMP829-cat/lacZ was created by ligating the chloramphenicol acetyltransferase (CAT) gene (cat) (PCR product using pBC SK+ as the template; Stratagene) and the E. coli β-galactosidase (lacZ) gene (PCR product using BioBrick part BBa_I732017 [http://parts.igem.org/] as the template) into pMP829 (23). To create a plasmid expressing Vgr, the lacZ gene of pMP829-cat/lacZ was removed by digesting the plasmid with PstI and XhoI, and a PCR product of the vgrG gene was inserted; the resulting plasmid was designated pMP829-cat/vgrG. VgrG is a 17.5-kDa F. novicida virulence factor that is part of the type VI secretion system encoded by the Francisella pathogenicity island (FPI) (24).

TABLE 1.

Strains and plasmids used in the study

Strain, plasmid, or oligonucleotide	Genotype, description, or sequence	Source or reference
Strains
F. novicida MFN245	hsdRI hsdRII res drg	39
F. novicida MFN45 tetR+	MFN45 attTn7::PrpsL-tetR+ res–aphA-1–res (Kmr)	This work
F. novicida ΔvgrG	MFN45 ΔvgrG	8
F. novicida ΔvgrG tetR+	MFN45 ΔvgrG tetR+ (Kmr)	This work
E. coli DH10B	F− mcrA Δ(mrr-hsdRMS-mcrBC) ϕ80lacZΔM15 ΔlacX74 recA1 endA1 araD139 Δ(ara leu)7697 galU galK λ− rpsL nupG	Invitrogen
E. coli MGZ1	E. coli MG1655 F− λ− ilvG rfb-50 rph-1; chromosomally integrated Z1 cassette expresses LacI and TetR	40
Plasmids
pMP720	Helper plasmid for mini-Tn7 integration; Hygr	26
pMP749	Mini-Tn7 Francisella integration vector; Apr Kmr	26
pMP749-tetR	pMP749 with tetR expressed from Pbla	This work
pMP823	Template for PCR of Pbla	23
pMP829	E. coli-Francisella shuttle vector; Hygr	23
pMP829-cat/lacZ	pMP829 with promoterless cat and lacZ	This work
pMP829-cat/vgrG	pMP829 with promoterless cat and vgrG	This work
pMP829-Px-cat/lacZ	Series of plasmids recovered from E. coli or F. novicida screen for functional promoters or control promoter x	This work
pMP829-Px-cat/vgrG	Series of plasmids with select promoters (x) driving vgrG expression	This work
Oligonucleotides
BamHI-N48-tetO	CACCTGACGTCTAAGAAGGATCC-Nx48-TCCCTATCAGTGATAGAGAa
BamHI-N30-tetOrc	ATTACCGCCTTTGAGTGAGCGGATCC-Nx30-TCTCTATCACTGATAGGGAa
PE-cat-FAM	(FAM)-CATTGGGATATATCAACGGTGGTATATCCA

^aWhere N is a random nucleotide at 30% G+C content.

An F. novicida strain expressing TetR was created by inserting the tetR gene at the unique Tn7 att site in the F. novicida chromosome. First, the tetR gene from Tn10 was joined to the 0.5-kb upstream promoter region of the β-lactamase gene found in plasmid pMP823 (23) by fusion PCR (25). This fusion product (P_bla-tetR) was ligated into the mini-Tn7 integration vector pMP749 (26) to make plasmid pMP749-tetR. A section of the plasmid consisting of tetR and the aphA-1 gene conferring kanamycin resistance (Km^r) and flanked by Tn7L and Tn7R sites was integrated into the F. novicida chromosome at the Tn7 att site by methods described previously (26), to create the F. novicida tetR⁺ strain. In order to introduce tetR into a ΔvgrG background, chromosomal DNA from the F. novicida tetR⁺ strain was used to transform the F. novicida ΔvgrG strain to kanamycin resistance, indicating that the aphA-tetR⁺ cassette was integrated into the F. novicida ΔvgrG chromosome. The ΔvgrG and aphA-tetR⁺ genotypes and phenotypes were verified, and the resulting strain was designated the F. novicida ΔvgrG tetR⁺ strain.

Synthetic tetO-containing DNA libraries.

Oligonucleotides BamHI-N48-tetO and BamHI-N30-tetOrc (Table 1) were added to a final concentration of 2 μM in 1× NEBuffer 2 (NEB) with 250 μM each deoxynucleoside triphosphate (dNTP). The mixture was brought to a boil and then allowed to cool slowly to facilitate the annealing together of the two oligonucleotides at their complementary tetO regions, which overlap each other by the full 19 nt of tetO. Klenow fragment (3′→5′ exo⁻; NEB) was added once the mixture cooled to 37°C, and the resulting reaction mixture was allowed to incubate for 1 h. This resulted in the extension of the partially overlapping oligonucleotides, each using the other as the template, resulting in a library of double-stranded DNA (dsDNA) fragments with an AT-rich random sequence flanking tetO (48 random base pairs to one side and 30 to the other) and a BamHI restriction site immediately following the random sequence to either side. The fragments were designed to include a short stretch of nonrandom DNA sequence at either end, which could be used as PCR primer binding sites, but no such PCR was performed as part of the experiments described here, and these nonrandom ends were removed as a consequence of the BamHI digestion step. The reaction mixture was heated to 75°C for 20 min to inactivate the polymerase before digestion with BamHI and ligation into the BamHI site upstream of the cat gene in pMP829-cat/lacZ (Fig. 1). The ligation products were dialyzed against distilled water (dH₂O) by floating the mixture on a 0.025-μm VSWP membrane filter (Millipore) for 2 h to reduce the salt concentration. Fifteen microliters of this product was used to transform 40 μl E. coli DH10B by electroporation. After recovery in 1 ml SOC (2% tryptone, 0.5% yeast extract, 10 mM NaCl, 2.5 mM KCl, 10 mM MgSO₄, 10 mM MgCl₂, and 20 mM glucose) for 1 h, the cells were spun down, resuspended in 200 μl SOC, and plated onto LB agar containing 200 μg/ml Hyg. After incubation at 37°C for 8 h, the thin lawn of bacterial growth was collected, and plasmid DNA was isolated. This plasmid preparation was used to transform the F. novicida tetR⁺ strain and E. coli MGZ1 by chemical transformation. Transformants were recovered for 1 h in medium containing ATc and then plated onto solid medium containing Hyg, Cm, and ATc. Plates used for E. coli also contained X-gal; however, since F. novicida is sensitive to a cleavage product of X-gal (27), this indicator was not added to plates used for F. novicida growth. The resulting clones were picked into TSB freezing medium (18) with 0.1% cysteine in 96-well plates containing Hyg. Clones were grown overnight and then spotted onto solid medium with Hyg, containing or lacking ATc (E. coli plates also contained X-gal), and then grown overnight at 37°C. E. coli plates were subsequently moved to 4°C for 18 h to allow greater color development. To assess β-galactosidase expression in F. novicida, colonies were overlaid with filter paper that had been soaked in X-gal (1 part 20 mg/ml X-gal in dimethyl sulfoxide [DMSO] and 3 parts dH₂O), and color was allowed to develop at 30°C for 8 h.

FIG 1.

Schematic of the approach for identifying inducible and constitutive Francisella promoters from semirandom DNA sequences. Oligonucleotides were hybridized at a complementary tetO sequence and made double stranded. These dsDNA fragments were ligated into a Francisella-E. coli shuttle vector upstream of cat and lacZ reporter genes and selected for the ability to drive cat expression.

Chemiluminescent LacZ assay.

β-Galactosidase levels were determined by using the luminescence generated by the cleavage of Galacton-Plus (Galacto-Light Plus system; Applied Biosystems). Cultures were grown to mid-exponential phase in 96-well plates in TSBC with Hyg for F. novicida and in EZ Rich defined medium (EZDM; Teknova) supplemented with 2% glucose and Hyg for E. coli MGZ1. F. novicida is naturally lacZ deficient. E. coli MGZ1 has the wild-type lac operon, but its activity was suppressed to minimal levels by the use of defined medium with the addition of glucose. Cultures were induced with ATc 2 h before harvesting, where appropriate. The A₆₀₀ of each culture was measured immediately before lysis. E. coli cultures were lysed directly by adding 20 μl of culture to 70 μl of lysis solution (100 mM potassium phosphate [pH 7.8], 0.2% Triton X-100, 500 μg/ml polymyxin B sulfate). F. novicida cells were pelleted by centrifugation for 20 min at 4,000 × g, and supernatant was removed before addition of 70 μl of lysis solution to each well. Twenty microliters of lysate was added to 70 μl of reaction buffer in a white, clear-bottom, 96-well plate (Griener Bio-One), followed by a 30-min incubation at 30°C. One hundred microliters of Accelerator-II (Applied Biosystems) was added to each well immediately before measuring luminescence for 1.0 s per well on a Molecular Devices SpectraMax M5 plate reader. A strain harboring pMP829-cat/lacZ was used as a blank, and luminosity values were normalized to cell culture density.

Western immunoblotting.

Cultures were grown to mid-exponential phase, and ATc was added 2 h before harvesting of cells, where appropriate. One milliliter of culture was pelleted by centrifugation and resuspended in 25 μl cold dH₂O containing protease inhibitors (cOmplete protease inhibitor cocktail, EDTA-free; Roche) before addition of 30 μl of 2× SDS loading buffer. Cultures were normalized based on cell density, separated by SDS-PAGE on a 12% gel (10 μl lysate loaded per lane), transferred onto nitrocellulose, and blocked in Odyssey blocking buffer (Li-Cor Biosciences). Primary antibodies were diluted in blocking buffer with 0.05% Tween 20 and used at the following dilutions: rabbit anti-TetR at 1:1,000 (ab14075; Abcam), rabbit anti-CAT at 1:1,000 (C9336; Sigma-Aldrich), and rabbit anti-VgrG at 1:5,000 (21). Primary antibody was detected by using IRDye800-conjugated goat anti-rabbit antibody (Rockland Immunochemicals) in Odyssey blocking buffer with 0.05% Tween 20 and 0.01% SDS (1:15,000) and visualized on an Odyssey scanner (Li-Cor Biosciences).

Mapping of transcription start sites by primer extension.

Cultures of the F. novicida tetR⁺ strain and E. coli MGZ1 harboring promoter plasmids were grown in TSBC with Hyg (for F. novicida) and EZDM supplemented with 2% glucose and Hyg (for E. coli). Cultures were induced with ATc 1 h before harvesting in mid-exponential phase. A total of 0.5 ml of culture was added to 1 ml RNAprotect Bacteria reagent (Qiagen), and RNA was isolated by using the RNeasy minikit (Qiagen). The RNA was quantified spectrophotometrically, and FAM (6-carboxyfluorescein)-labeled cDNA was produced in a reverse transcription reaction by using Moloney murine leukemia virus (M-MuLV) reverse transcriptase (NEB) with 5 μg of RNA as the template, according to the manufacturer's protocol, in a reaction mixture containing 20 U RiboLock RNase inhibitor (Thermo Scientific) and FAM-labeled primer PE-cat-FAM (Table 1). The resulting products were concentrated by ethanol precipitation and resuspended in 10 μl HiDi formamide (Life Technologies) and 0.3 μl GeneScan 500 ROX size standards (Life Technologies). The mixture was heated at 95°C for 5 min, cooled on ice for 1 min, and then subjected to electrophoresis on an AB3730 DNA analyzer (Applied Biosystems). Data were analyzed by using GeneMapper software (Applied Biosystems).

Intracellular growth assay.

J774A.1 mouse macrophage-like cells were used to seed 96-well plates at 5 × 10⁴ cells/well in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum and 2 mM l-glutamine and allowed to adhere overnight. F. novicida strains were added to the macrophages at a multiplicity of infection of 50 to 1 and incubated for 1 h. After 1 h (t = 0), wells were washed with phosphate-buffered saline (PBS) containing 10 μg/ml gentamicin (Gm) three times, before addition of fresh DMEM supplemented with 10% fetal bovine serum, 2 mM l-glutamine, and 2 μg/ml Gm, with or without ATc, as appropriate. Infected macrophages were lysed at different time points by washing three times with PBS before addition of 0.1% deoxycholic acid in PBS. Lysates were serially diluted in PBS with 0.1% gelatin and spread onto TSAC with Hyg to enumerate viable bacteria by plate counts.

Creation of minimal Francisella promoters.

Plasmids containing promoters P143, P146, and P165 were amplified by inverse PCR using 5′-phosphorylated primers that were extended away from each other on the circular template so that the entire plasmid was amplified, excluding approximately 56 nt consisting of the tetO region up to and including the upstream BamHI site. The deleted region of each promoter was replaced by a 26-bp stretch of a randomly generated DNA sequence (http://www.faculty.ucr.edu/∼mmaduro/random.htm) containing a unique PstI site, which allowed truncated promoters to be identified by restriction digestion. Each resulting PCR product was ligated to itself to reform the circular plasmid, each one now missing the upstream portion of its synthetic promoter. Ligation products were used to transform E. coli. The plasmid was isolated, and the modified promoters were sequence verified before F. novicida was transformed with these plasmids. Expression of LacZ activity in F. novicida was assayed side by side with LacZ activity produced by the corresponding full-length promoters.

Statistical analysis.

Statistical analysis was carried out by using the GraphPad Prism 5 software package (GraphPad Software, Inc.).

Nucleotide sequence accession numbers.

The sequences of characterized F. novicida tetO-containing promoter regions described in this work have been deposited with GenBank and have been assigned accession numbers KF279494 to KF279508. Sequences that are too short to be submitted to GenBank can be found in the text or supplemental material.

RESULTS

Selection of synthetic promoters in F. novicida.

We created a library of 97-bp-long (not including the flanking BamHI restriction sites) synthetic DNA fragments with a nearly central tetO region surrounded on either side by random nucleotides (Fig. 1). The randomized regions were designed to have 30% G+C content in order to be slightly below the average 32% G+C content of the F. novicida chromosome. Our reasoning was that promoter regions would have a lower G+C content than the protein-coding regions of the chromosome. These fragments were ligated into the BamHI site of an F. novicida-E. coli shuttle vector and allowed to insert in either orientation with respect to a selective marker, the chloramphenicol acetyltransferase gene (cat). The ligation mixture was electroporated into E. coli, and selection was made for hygromycin resistance. The transformed cells were pooled, and plasmid DNA was isolated from the entire library.

An F. novicida strain was constructed to constitutively express the tetracycline repressor protein, TetR, from a chromosomal location at the unique Tn7 att site (26). This F. novicida strain was chemically transformed with the library of random inserts, and the transformed cells were selected separately on either hygromycin or chloramphenicol agar plates. We found that about 0.5% of the hygromycin-resistant colonies were also chloramphenicol resistant. A chloramphenicol concentration of 5 μl/ml was used for selection, which is well above the MIC that we determined to be in the range of 1 to 1.5 μg/ml.

To visualize the relative transcriptional strength of and control by TetR, we examined the amount of β-galactosidase produced by the reporter gene lacZ, which was downstream of the cat gene (Fig. 1). Since F. novicida is sensitive to the cleaved products of X-gal, we designed experiments that exposed F. novicida to X-gal following the growth of colonies. We robotically picked approximately 9,000 Cm^r colonies and gridded them onto agar with or without the TetR inducer ATc. Once colonies were fully grown, the agar plates were overlaid with filter paper saturated with a solution of X-gal to visualize cells expressing β-galactosidase. Clones with a wide range of blue intensity were observed indicating a wide range of lacZ expression levels. Some clones produced blue colonies only in the presence of ATc, and others were blue under both conditions, while the remainder did not produce any obvious blue color under either condition.

After qualitatively assaying the β-galactosidase levels, 187 colonies were picked into liquid medium in 96-well plates, grown, and then gridded onto solid medium with and without ATc (see Fig. S1A and S1B in the supplemental material). These 187 clones were chosen from the original screen plate to represent promoters of various strengths with a preference for clones that produced intense blue staining on the ATc/X-gal plate. After repeated qualitative observations of β-galactosidase levels, 15 clones (10 TetR controlled and 5 constitutive) were quantitatively tested for levels of β-galactosidase expression by cleavage of the luminescent substrate Galacton-Plus. Both TetR-controlled and TetR-insensitive promoters were tested with and without the addition of the TetR inducer ATc (Fig. 2). Two recombinant clones were constructed to contain two strong F. tularensis LVS promoters, Pbfr and PZ12 (promoters for a bacterioferritin-encoding gene and a tRNA gene, respectively) (28). Although none of the synthetic promoters expressed β-galactosidase as strongly as the strongest known natural promoter in F. tularensis (Pbfr), all of the synthetic promoters were expressed as strongly as or stronger than almost all of the natural promoters found previously by Zaide et al. (28). For comparison, the PZ12 promoter (originally called “P12” but designated here PZ12 to distinguish from promoters identified in our work) was the fourth strongest natural promoter found by Zaide et al. (28) and about twice as strong as an average-strength promoter defined as “strong” by those researchers.

FIG 2.

β-Galactosidase expression in F. novicida driven by synthetic promoters. Clones were chosen from a qualitative assay (see Fig. S1 in the supplemental material) and quantitatively assayed for β-galactosidase activity with and without the addition of the TetR inducer ATc. Six independent replicates of cultures containing the various promoter-reporter plasmids were grown to mid-exponential phase and induced with ATc, or mock induced, for 3 h. Cell number was normalized by determining the A₆₀₀. β-Galactosidase activity is indicated in arbitrary luminosity units. The 10 promoters on the left side of the graph (P40 to P21) are inducible with ATc, and the next 5 promoters (P142 to P165) are unresponsive to ATc addition. Both sets of promoters are ordered from strongest to weakest. The strong, natural F. tularensis promoters Pbfr and PZ12 were identified previously by Zaide et al. (28) and are included for comparison. Error bars represent standard errors of the means.

The data presented in Fig. 2 also show that some synthetic promoters were inducible by the addition of ATc, whereas others were not. Those promoters that were inducible showed increases of reporter activity of >10-fold when the inducer was added compared to activity in cultures without the inducer. Curiously, the strains carrying the synthetic, constitutive promoters, and the natural F. tularensis promoters, showed a slight decrease in activity when ATc was added. This may be due to a low level of antitranscriptional activity of ATc.

Our cloning strategy (Fig. 1) allowed the synthetic BamHI fragments to insert in either orientation, as determined by the direction of tetO and by the length of the flanking random sequence. When we sequenced 184 DNA fragments that had promoter activity, we found that almost all of them were unique (169 of 184) (see Data Set S1 in the supplemental material) and that of 56 fragments oriented in the “forward” direction (tetO closer to the 3′ end of the DNA insert), all 56 yielded promoter activity that was controlled by TetR. This is understandable, as the 30 bp downstream of the tetO region would presumably not be long enough to represent a promoter without extending into the tetO region. Of the DNA fragments that were in the reverse orientation, 27 were inducible with ATc and 25 were constitutive. This suggests that the 48-bp region downstream of tetO (in the reverse orientation) is sufficient to constitute a promoter in F. novicida.

Our selection and screening assays relied on promoter activity to produce a chloramphenicol resistance phenotype or β-galactosidase activity. As a separate measure of the activity of the promoters, we wanted to directly observe chloramphenicol acetyltransferase (CAT) production by using Western immunoblotting. Figure 3 and Fig. S2 and S3 in the supplemental material show the activity of selected promoters in producing CAT. Promoters that exhibited inducibility with ATc in producing β-galactosidase (P20, P39, P40, P94, and P135) all showed TetR control of CAT expression in Western blot assays. P39 and P40 showed a small amount of CAT expression in the absence of inducer. The promoter P142, which was constitutive in the β-galactosidase assay, showed production of CAT with or without ATc addition; promoters P146 and P165 also produced CAT in the absence of ATc.

FIG 3.

Immunoblot analysis of TetR control of cat gene expression. The production of CAT (indicated by arrows at right) is shown for strains expressing TetR with or without ATc addition and with the cat gene with no promoter or downstream of the inducible, synthetic promoters P20, P39, P40, P94, and P135; the constitutive synthetic promoters P142, P146, and P165; or the natural promoters PZ12 and Pbfr. Digital overexposure of the immunoblots (see Fig. S3 in the supplemental material) reveals nonspecific antibody-reactive protein bands that are present relatively evenly in all of the lanes. The normalized intensities of the CAT bands are listed in Table S1 in the supplemental material. MW, molecular weight.

Promoter control of the Francisella virulence factor VgrG.

The gene products of cat and lacZ are both foreign to F. novicida. In order to test the utility of the synthetic promoters in controlling native genes in F. novicida, we engineered plasmids with the strong P40 or the weak P18 inducible promoter. These plasmids were placed upstream of a two-cistron operon (cat-vgrG) so that they controlled expression of CAT and the virulence factor VgrG. The VgrG protein is part of the type VI secretion system encoded by the Francisella pathogenicity island (FPI) and is required for virulence (24). As shown in Fig. 4A, the P40 and P18 promoters showed the expected TetR-regulated vgrG expression. In strains with plasmids with no promoter upstream of the cat-vgrG operon, there was no detectable CAT or VgrG. When P40 or P18 was placed before cat-vgrG, it was controlled if TetR was expressed in the cell but was not controlled if no TetR was expressed (Fig. 4B). If TetR was expressed, the production of CAT and VgrG occurred only if ATc was added to the culture. A possible exception was the strain carrying the plasmid with P40 driving the cat-vgrG operon: a small amount of CAT production was seen in the absence of ATc. Similar TetR-regulated expression was seen with another FPI-encoded virulence factor, DotU (see Fig. S5 in the supplemental material).

FIG 4.

Immunoblot analysis of expression of the virulence factor VgrG by a strong promoter and a weak promoter. (A) The test plasmid used in these experiments has an artificial operon of the cat and vgrG genes. The production of CAT and VgrG is shown for F. novicida strains expressing or not expressing TetR; strains expressing TetR with or without ATc; strains with cat and vgrG downstream of no promoter; strains with the strong, inducible promoter P40; or strains with the weak, inducible promoter P18. The wild-type (WT) F. novicida strain carrying an empty control plasmid is shown at the left. Digital overexposure of the immunoblots (see Fig. S4 in the supplemental material) reveals nonspecific antibody-reactive protein bands that are present relatively evenly in all of the lanes. The normalized intensities of the CAT and VgrG bands are listed in Tables S2 and S3 in the supplemental material. (B) Immunoblot detection of TetR in F. novicida strains. Arrows point to the 23-kDa TetR band.

Because of the incomplete control of CAT expression by TetR in the plasmid containing the P40 promoter, we suspected that a small amount of VgrG might also be produced when vgrG is downstream of P40. A potentially more sensitive assay for the control of VgrG expression is to measure the intracellular growth of an F. novicida ΔvgrG mutant harboring a plasmid containing vgrG controlled by a tetO-bearing promoter. We found that a ΔvgrG tetR⁺ F. novicida strain carrying a plasmid with P40-vgrG regained the ability for intracellular growth upon addition of ATc (see Fig. S6 in the supplemental material). However, as we suspected, even in the absence of ATc, there was moderate growth of the ΔvgrG complemented strain, probably due to a low level of activity of the P40 promoter in the absence of the inducer.

To test if a weak, TetR-controlled promoter could tightly control VgrG expression yet express sufficient VgrG when induced, we placed the P18 promoter in front of the cat-vgrG plasmid-borne operon. The control of vgrG by P18 yielded the expected virulence phenotype, as measured by the ability of F. novicida to grow in the macrophage-like cell line J774 (Fig. 5). An F. novicida ΔvgrG strain lacking tetR and with vgrG downstream of P18 on plasmid pMP829 grew as well as the wild-type (tetR⁺) strain. Similarly, a tetR⁺ strain with the same plasmid grew like the wild type when ATc was added but grew like the mutant F. novicida ΔvgrG strain when ATc was absent (Fig. 5). When no promoter was placed in front of the plasmid-borne vgrG gene, there was no enhanced growth of the F. novicida ΔvgrG strain (see Fig. S7 in the supplemental material). Hence, a weak- to moderate-strength TetR-controlled promoter has sufficient dynamic range to properly regulate virulence factors in F. novicida.

FIG 5.

Intracellular growth of F. novicida strains having vgrG controlled by the TetR-responsive promoter P18. Induction of plasmid-encoded VgrG expression by the addition of ATc induces the ability for intracellular growth. The strain without a TetR repressor to control P18-vgrG also exhibits wild-type intracellular growth. In the absence of ATc, the strain with P18-driven vgrG grows the same as the ΔvgrG strain. Error bars represent standard errors of the means. Analysis of the differences among the growth patterns of different strains was done by a two-way analysis of variance [P = <0.0001 for the ΔvgrG tetR⁺(829::P18-vgrG) strain with ATc versus the ΔvgrG tetR⁺(829::P18-vgrG) strain; P = 0.1370 for the ΔvgrG tetR⁺(829::P18-vgrG) strain with ATc versus the WT tetR⁺ strain; P = 0.56 for the ΔvgrG tetR⁺(829::P18-vgrG) strain versus the ΔvgrG strain].

Transcription start sites and position of tetO in F. novicida promoters.

In order to localize the promoter regions in each recombinant clone, we used primer extension of 10 mRNA species corresponding to controlled promoters to find the transcription start site and, thus, the putative location of the −10 and −35 regions of the promoters (Fig. 6A). We carried out the same experiment with five constitutive promoters. Of the 10 inducible promoters, the tetO region overlapped the putative −35 region in 5 promoters, overlapped the −10 region in 1 promoter, was downstream of the −10 region in 2 promoters, and was upstream of the −35 region in 2 promoters. In the two promoters with the tetO region upstream of the putative −35 region, the tetO region was within 2 or 5 bp of the −35 region. All of the constitutive promoters had the tetO region upstream of the putative −35 region (Fig. 6B; also see Fig. S8 in the supplemental material). In all five constitutive promoters for which the transcription start site was identified, the tetO region was at least 10 bp upstream of the −35 region and was in the reverse orientation.

FIG 6.

Organization of synthetic DNA fragments that function as promoters in F. novicida. (A) TetR-regulated promoters. Strong to weak promoters are shown from top to bottom. The transcription start, as determined by primer extension of mRNA, is indicated by large boldface letters. The −10 and −35 regions are indicated by letters in boldface italic type. The tetO regions are indicated by an arrow. Underneath each promoter sequence is a diagrammatic representation of the promoter organization showing the relationship of the tetO region to the −10/−35 regions and the transcriptional start site. The TetR-regulated promoters P39 and P21 had 5 and 2 bp, respectively, between the −35 region and tetO. (B) Sequence and organization of the unregulated promoter P146. All of the constitutive promoters that were examined showed the same organization (see Fig. S8 in the supplemental material). In the diagrams, the squares represent the −10/−35 regions, the circle represents the transcriptional start site, and the arrow represents the tetO region. The DNA sequences of 185 synthetic F. novicida promoters are provided in Data Set 1 in the supplemental material.

Synthetic F. novicida promoter activity in E. coli.

The accumulated, circumstantial evidence in the literature suggests that E. coli promoters function poorly in Francisella. However, this idea has never been directly tested, and it is not known if Francisella promoters function in E. coli. In order to investigate the cross-species functionality of promoters, we wanted to test E. coli promoters in F. novicida, and F. novicida promoters in E. coli. To aid in studying cross-species promoter activity, we isolated synthetic promoters in E. coli, using an approach similar to that used to isolate synthetic promoters in F. novicida (Fig. 1). Thousands of Cm^r colonies resulted when E. coli MGZ1 cells were transformed with the same library of random, tetO-containing dsDNA fragments ligated into pMP829-cat/lacZ when selected for on Cm plates in the presence of ATc. The promoterless parent plasmid was unable to produce a Cm^r phenotype in E. coli under these conditions. Eighty-eight of these Cm^r transformants were subjected to further analysis. Sequencing revealed that all 88 clones had received a synthetic fragment upstream of cat and that 67 of these consisted of unique sequence (see Data Set S2 in the supplemental material). The majority of these synthetic E. coli promoters displayed TetR repression and ATc induction, as determined by an X-gal spot assay (see Fig. S1C and S1D in the supplemental material). Ten of these ATc-inducible E. coli promoters had expression levels quantitated by a LacZ assay. Furthermore, E. coli MGZ1 was transformed with a selection of the synthetic promoters isolated from Francisella in the experiment described above to allow comparison to those promoters isolated in E. coli. We found that the approximate relative strengths of the strongest promoters selected in E. coli were the same as those of the stronger F. novicida promoters when expressed in E. coli (Fig. 7). Surprisingly, two controlled and one constitutive F. novicida-selected synthetic promoter induced expression of β-galactosidase in E. coli at levels equivalent to those induced by the selected E. coli promoters. The strongest known F. tularensis promoter, Pbfr, functioned in E. coli but exhibited a lower level of expression, relative to P40 and P20, than it did when tested in F. novicida. The bfr promoter was almost twice as strong as the strongest synthetic promoter (P40) in F. novicida (Fig. 2) but was less strong than P40 in E. coli (Fig. 7). All of the synthetic E. coli promoters functioned poorly in F. novicida (see Fig. S9 in the supplemental material), providing firm evidence for the widely held, but previously untested, consensus that E. coli promoters function poorly in Francisella species.

FIG 7.

β-Galactosidase expression in E. coli driven by synthetic promoters. Promoters with a “PE” prefix were selected in E. coli, and their functionality in driving the expression of β-galactosidase in E. coli MGZ1 expressing TetR is shown. The activities of the synthetic, inducible (P20 and P40); constitutive (P146); and natural (Pbfr) promoters selected or isolated in F. novicida or F. tularensis are also shown. Values on the y axis are arbitrary luminosity units. Error bars represent standard errors of the means.

Minimum size of F. novicida promoters.

Our data suggest that tetO confers promoter repression when positioned within 5 bp of the −35 region but does not induce repression when positioned more than 9 bp from this region. Taken together, this implies that a region from the transcriptional start to 10 bp upstream of the −35 region is sufficient to form a Francisella promoter. To test this notion, we deleted the tet operator and all of the synthetic DNA sequence upstream of tetO from three plasmids containing constitutive Francisella promoters (P143, P146, and P165). In place of the deleted sequence, we inserted a 26-bp randomly generated spacer DNA sequence upstream of the minimal promoter region to serve as an insulating sequence between the plasmid sequence and the remaining promoter sequence (see Fig. S11 in the supplemental material).

The resulting plasmids with the minimal promoters expressed β-galactosidase in F. novicida at the same levels as the corresponding plasmids with the longer promoter regions. The minimal promoters drove expression at levels that were nearly identical to those of the parental promoters (Fig. 8), suggesting that all of the promoter activity originated from the promoters and was minimally influenced by DNA sequence upstream of the minimal promoter sequence: each promoter-upstream junction changed, but the level of reporter expression remained the same. When the three minimal promoters were used in a BLASTN query of the F. novicida chromosome, they showed maximum identities of 17-, 11-, and 16-bp stretches of identical nucleotides for promoters mP143, mP146, and mP165, respectively. Thus, we have identified three DNA sequences that can serve as constitutive, minimal promoters that have low sequence identity to F. novicida chromosomal DNA sequences.

FIG 8.

β-Galactosidase expression driven by minimal promoters in F. novicida. Open bars represent activity in strains having the original promoters that contained tetO. Filled bars represent activity expressed from strains carrying promoters from which the nucleotides following the upstream BamHI region through the tetO region were deleted. PZ-12 serves as a control for expression relative to a natural promoter and has not been modified. Error bars represent standard errors of the means.

DISCUSSION

Our understanding of gene control is advancing to such a degree that in many organisms, one is able to assemble elements of known strength, such as ribosome binding sites (RBSs) (29), promoters (30), and transcriptional stop signals (31), in order to precisely control the expression of one or multiple genes. In any one system, one needs multiple elements that provide a variety of expression levels in order to create gene clusters, with each gene product being expressed at the desired level. Typically, this type of synthetic biology approach is used for metabolically engineering a microbe to produce biofuels (32) or therapeutic small molecules (33). However, the science of genetically engineering pathogens and commensal bacteria to act as vaccines or as therapies is growing (34) and will need advanced genetic tools to construct sophisticated synthetic genomic regions.

For robust genomic engineering, it is best to have a broad type of any one element. Clearly, one needs elements of different strengths for different constructs. Also, the use of exogenous genetic elements, especially promoters, often results in different outputs when used in different genetic environments. For example, the amount of a gene product will be affected not only by promoter strength but also by RBS context (29), protein stability, and factors such as premature termination of either transcription (35) or translation. For several of the synthetic promoters described here, we tested their activity in driving the expression of multiple genes, thus providing evidence that the promoters can function in a variety of genetic environments.

Both controlled and constitutive synthetic promoters have been developed in E. coli, and usually, these have been based on defined, natural promoters or consensus sequences. The promoters described in this work were generated randomly. The random nature of these promoters is especially important in F. novicida, since it has a very high rate of recombination and readily takes up and integrates linear or circular DNA (36). The random nature of these synthetic promoters minimizes the chances that any recombinant construct will integrate into the F. novicida chromosome at an off-target site that has a long stretch of identity with the promoter used in the recombinant DNA.

The data presented here, as well as data from previous studies of Francisella promoters (28), suggest that the consensus sequence of the core Francisella promoter is very similar to that of the E. coli core promoter. Thus, the failure of Francisella promoters to function in E. coli is perplexing. As promoters can be recognized generally as areas of low G+C content (37), and the consensus −10 and −35 promoter elements are made up of A or T nucleotides in 10 of 12 sites, it seems that an organism with a chromosome with a low G+C content, such as Francisella, will have a genome highly enriched for promoter-like sequences and may need an extra identifying feature to reduce nonspecific transcription initiation. Alignment at the −10 box of the 17 Francisella promoters for which the transcription start sites have been identified revealed a “TGn” motif directly upstream of the −10 box in 8 of the 17 promoters (47%). Given the 30% G+C content of the promoters, this “extended −10” sequence would be expected in 5.2% of promoters if appearing by chance alone. This motif is also known to be conserved in E. coli promoters but in only a subset of promoters termed “extended −10 promoters,” which make up 19% of the total promoters (38) and were found in 3 of 11 (27%) of our characterized E. coli promoters. Although our sample size is too small to draw any conclusions, this analysis provides a testable hypothesis for future work.

This work focused on the effect of promoter sequences in producing different gene products. Recent work (29) has shown that changes in the genetic context of an RBS can have a dramatic effect on the amount of gene product that is made. However, in many of our experiments, we measured the amount of gene product produced from the second cistron in an operon; therefore, the immediate genetic context of this RBS was not changed with the various promoter sequences, minimizing any effect resulting from changes in RBS strength. Also, the anti-Shine-Dalgarno sequence in the F. novicida 16S rRNA (11) is identical to that in E. coli, and one would expect similar translation initiation rates in both species.

Francisella species have a poor repertoire of transcription control tools, and it appears that importing controlled promoters from other species is not a viable option. In this work, we demonstrated that regulated transcription control elements can be produced through a simple selection-and-screening process for a semirandomized DNA fragment. Using this method, we were able to create a series of tightly repressed, strongly induced promoters for F. novicida and E. coli and to produce small, defined promoters that can be used for genomic engineering.