Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Alternative Sigma Factor Over-Expression Enables Heterologous Expression of a Type II Polyketide Biosynthetic Pathway in Escherichia coli

Abstract

Background

Heterologous expression of bacterial biosynthetic gene clusters is currently an indispensable tool for characterizing biosynthetic pathways. Development of an effective, general heterologous expression system that can be applied to bioprospecting from metagenomic DNA will enable the discovery of a wealth of new natural products.

Methodology

We have developed a new Escherichia coli-based heterologous expression system for polyketide biosynthetic gene clusters. We have demonstrated the over-expression of the alternative sigma factor σ54 directly and positively regulates heterologous expression of the oxytetracycline biosynthetic gene cluster in E. coli. Bioinformatics analysis indicates that σ54 promoters are present in nearly 70% of polyketide and non-ribosomal peptide biosynthetic pathways.

Conclusions

We have demonstrated a new mechanism for heterologous expression of the oxytetracycline polyketide biosynthetic pathway, where high-level pleiotropic sigma factors from the heterologous host directly and positively regulate transcription of the non-native biosynthetic gene cluster. Our bioinformatics analysis is consistent with the hypothesis that heterologous expression mediated by the alternative sigma factor σ54 may be a viable method for the production of additional polyketide products.

Citation: Stevens DC, Conway KR, Pearce N, Villegas-Peñaranda LR, Garza AG, Boddy CN (2013) Alternative Sigma Factor Over-Expression Enables Heterologous Expression of a Type II Polyketide Biosynthetic Pathway in Escherichia coli. PLoS ONE 8(5): e64858. https://doi.org/10.1371/journal.pone.0064858

Editor: Vasu D. Appanna, Laurentian University, Canada

Received: March 29, 2011; Accepted: April 22, 2013; Published: May 28, 2013

Copyright: © 2013 Stevens et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This work was supported by NSERC, Ontario Ministry of Research and Innovation, and the University of Ottawa. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Bacterial polyketides possess an enormous range of chemical diversity and biological function. Many polyketides such as tetracycline [1], epothilone [2], and rapamycin [3] have been developed into key clinical pharmaceuticals in a broad range of therapeutic areas [4]. Sequencing of bacterial genomes, especially those of major polyketide producers such as Actinomycetes and δ-proteobacteria, have shown that there are many more polyketide biosynthetic pathways than polyketides isolated from standard cultivation techniques [5]–[7]. These genetically encoded polyketide natural products from cultivatable and uncultivatable bacteria represent one of the greatest remaining untapped reservoirs of new natural product diversity. Methods to effectively access this diversity will have a major impact on drug discovery [8].

To access this untapped diversity of polyketide products, a general method for heterologous expression of these pathways is needed. The selection of a heterologous host is contingent upon its ability to efficiently transcribe non-native pathways, translate the often GC rich transcripts, and possess all the starter and extender units necessary for polyketide production [9]. Hosts highly related to the native polyketide-producing organism often meet these requirements. Hosts such as Streptomyces coelicolor [10], Streptomyces lividans [11], and Myxococcus xanthus [12] have proved successful for heterologous production of a number of different polyketide products. Screening of multiple hosts related to the producing organism is often necessary to identify a successful heterologous host [13], [14]. In addition, various culture conditions also need to be screened to identify conditions that produce the desired compounds [15]. The lack of a general heterologous expression system has made heterologous expression incompatible with screening bacterial genomic DNA libraries for new polyketide products.

Escherichia coli has a number of advantages that make it an appealing host for a heterologous expression system. The ease of genetic manipulation and culturing as compared to other heterologous hosts makes E. coli far more useful for bioprospecting from genomic DNA libraries. Codon usage as well as starter and extender unit availability have proven not to be obstacles for heterologous production in E. coli [16]–[18]. The principle impediment to the use of E. coli as a heterologous host is its inability to effectively transcribe heterologous pathways. In all examples of heterologous expression of polyketides in E. coli, all of the biosynthetic genes have been placed under the control of the T7 promoter [16], [17]. The native promoters present in these heterologous pathways are not sufficient to ensure expression of all the genes in a given pathway under standard E. coli culture conditions.

To convert E. coli into a general heterologous expression system, a mechanism for ensuring transcription of all the biosynthetic genes in a foreign pathway is required. We hypothesized that a general transcriptional regulator of polyketide biosynthesis should be present in bacteria because horizontal transfer of biosynthetic pathways is a key mechanism for their proliferation among species [19]. Strong evidence for substantial horizontal transfer of pathways can be seen in the large number of diverse Streptomyces strains which produce the tetracycline family of antibiotics, the presence of yersiniabactin-like gene clusters in diverse Actinomycetes [6], and the presence of highly related pederin, onnamide A and psymberin gene clusters in unidentified bacterial strains from the beetle Paederus fuscipes and the sponges Theonella swinhoei and Psammocinia aff. bulbosa respectively [20]. Because polyketide production is positively regulated in most bacteria by stress [21], we further hypothesized that alternative sigma factors downstream of the stringent response should positively regulate transcription of biosynthetic genes.

Herein we show that this new mechanism for heterologous expression of the oxytetracycline polyketide biosynthetic pathway, where a high-level pleiotropic sigma factor from the heterologous host is used to positively regulate transcription of the non-native biosynthetic gene cluster, is highly effective. We demonstrate that one of the six alternative sigma factors in E. coli can selectively, directly, and positively regulate heterologous expression from the Streptomyces rimosus oxytetracycline biosynthetic gene cluster [1] in E. coli. This is the first successful heterologous production of an aromatic polyketide in E. coli [22], demonstrating the utility of this method.

Results

Standard E. coli culture conditions do not enable heterologous expression

To test our hypothesis that an alternative sigma factor downstream of the stringent response could positively regulate transcription of polyketide biosynthetic pathways in E. coli, we investigated the ability of E. coli to heterologously express the oxytetracycline biosynthetic pathway from S. rimosus [1]. Oxytetracycline is produced by a 32-kb type II polyketide synthase gene cluster consisting of 21 genes. Thirteen of these genes encode proteins required for the biosynthesis of oxytetracycline [23] and thus must be transcribed for successful heterologous production of oxytetracycline. The small intergenic spaces between open reading frames and the changes in direction of transcription in the oxytetracycline gene cluster suggest that the entire pathway is expressed as, at minimum, five putative operons with at least one biosynthetic gene on each transcript (Figure 1a).

thumbnail
Download:
Figure 1. SYBR-green based qPCR analysis shows that transcription limits heterologous production of oxytetracycline in E. coli.

(a) The 32 kb oxytetracycline biosynthetic gene cluster is shown. Five putative operons, oxyABCDE, oxyIHGF, oxyJKLMNO, oxyRQP, and oxyST are predicted for this gene cluster (b) qPCR analysis shows that over-expression of the alternative sigma factors σ54, σS and FecI enable detectable levels of the oxyB transcript to be produced. Over-expression of no sigma factor, σE, σF and σH do not lead to detectable levels of the oxyB transcript. (c) qPCR analysis shows that over-expression of the alternative sigma factor σ54 lead to detectable levels of transcripts for all five putative operons in the oxytetracycline biosynthetic pathway. In the absence of σ54 over-expression, the oxyB transcript cannot be detected. See also Table S1. No oxyB, oxyF, oxyK, oxyP, oxyT transcripts were observed in the null strains which lacked the oxytetracycline gene cluster. (Figures S1, S2).

https://doi.org/10.1371/journal.pone.0064858.g001

To determine if E. coli could produce oxytetracycline under standard culture conditions, E. coli BAP1 [16] was transformed with the S. rimosus oxytetracycline pathway [12] and cultured in rich liquid media. BAP1 was used because it possesses a chromosomal insertion of the phosphopantetheinyl transferase from Bacillus subtillis, sfp, which ensures that polyketide synthase acyl carrier proteins (ACPs) are post-translationally modified [16]. LC-MS/MS analysis of the organic extracts from the culture media did not show any detectable oxytetracycline in the culture broth (see Figure S3). LC-MS/MS analysis was performed using multiple reaction monitoring (MRM) mode with a lower limit of detection of 100 µg/L. These results indicate that E. coli, under standard culture conditions, cannot heterologously produce oxytetracycline from the native S. rimosus gene cluster.

To determine if transcription of the genes in the oxytetracycline gene cluster was limiting production of oxytetracycline in E. coli, we quantified mRNA levels using quantitative PCR (qPCR) for five genes in the pathway. The five genes oxyB, oxyF, oxyK, oxyP, and oxyT, one per putative operon, were selected. oxyB encodes the β-ketosynthase, oxyF and oxyT encode methyltransferases and oxyK encodes an aromatase all of which are required for oxytetracycline biosynthesis [23]. oxyP encodes a malonyl-ACP transferase, which may not be required for biosynthesis of oxytetracycline [23]. Transcripts for four of the genes, oxyF, oxyK, oxyP and oxyT could be easily detected and quantified (Figure 1c). Transcript levels for oxyB were indistinguishable from negative control samples that lacked the oxytetracycline gene cluster, indicating that oxyB was not transcribed (Figure 1b, Figure S1, Table S1). As OxyB is required for generation of the polyketide backbone of oxytetracycline, the lack of transcription of this gene is sufficient to account for the inability of E. coli to produce oxytetracycline under standard culture conditions. These results, in conjunction with published work [16], [17], confirm that transcription of heterologous polyketide pathways in E. coli limits the ability of the heterologous host to produce polyketide products when using the native promoters.

Over-expression of alternative sigma factors σ54, σS, and FecI positively regulate transcription of oxyB in E. coli

To test our hypothesis that an alternative sigma factor downstream of the stringent response could positively regulate transcription of polyketide biosynthetic pathways in E. coli, we investigated the impact of alternative sigma factor over-expression on transcription from the oxytetracycline gene cluster. In their native hosts, polyketide biosynthetic gene clusters are principally regulated by one or more pathway specific regulators, which can respond to a diversity of stimuli [21], [24]. The most general observed positive regulator of polyketide biosynthesis is nutrient limitation [21], which initiates the stringent response in bacteria. During stringent response the levels of the pleiotropic regulators phosphorylated guanosine nucleotides, ppGpp and pppGpp, increase, leading to an increase in alternative sigma factor-mediated transcription [24]. As increases in (p)ppGpp have been shown to correlate with polyketide production in diverse bacterial species [25], [26], we hypothesized that this positive regulation could be mediated by alternative sigma factors.

Sigma factors control the specificity of gene transcription by binding to RNA polymerase (RNAP) and recruiting it to promoter sequences upstream of the gene to be transcribed [27]. In addition to the household sigma factor σ70, E. coli has six alternative sigma factors, σ54 which is involved in nitrogen assimilation, σH which controls heat shock promoters, σS which controls stationary phase promoters, σF which controls flagellum related genes, σE controls response to extracytoplasmic stress, and FecI which is involved in iron transport [27]. Over-expression of sigma factors has been shown to increase expression of genes in their regulon in many bacteria, including E. coli [28]–[30]. If an alternative sigma factor is a positive regulator of polyketide biosynthesis, over-expression of that sigma factor will increase transcription from genes in the oxytetracycline gene cluster.

To determine if σ54,σH, σS, σF, σE, and FecI could positively regulate polyketide biosynthesis in E. coli, we quantified mRNA levels using quantitative PCR for oxyB. Six E. coli BAP1 strains possessing the oxytetracycline biosynthetic gene cluster and over-expressing either σ54,σH, σS, σF, σE, or FecI were investigated. No oxyB transcripts could be detected when σH, σF, and σE were over-expressed. Over-expression of σS and FecI led to detectable but low levels of oxyB transcription. When σ54 was over-expressed substantial levels of oxyB transcripts could be detected (Figure 1b, Figure S2, Table S1). Quantification of mRNA levels for oxyB, oxyF, oxyK, oxyP and oxyT in E. coli BAP1 possessing the oxytetracycline biosynthetic gene cluster and over-expressing E. coli σ54 showed that we were able to detect all five transcripts (Figure 1c). These data demonstrate that all of the putative transcripts required for oxytetracycline are present when σ54 is over-expressed.

Over-expression of σ54 enables oxytetracycline production in E. coli

If transcription limits heterologous production of polyketides in E. coli, then over-expression of σ54 in the presence of the oxytetracycline gene cluster is predicted to lead to the production of oxytetracycline. To test this hypothesis we over-expressed σ54 in E. coli BAP1 containing the oxytetracycline gene cluster and analyzed the organic extracts with LC-MS/MS to determine if oxytetracycline was detectable. LC-MS/MS analysis using the MS2 scan mode clearly identified the presence of oxytetracycline (Figure 2b). The LC retention time and MS2 spectrum of our heterologously produced oxytetracycline was indistinguishable from authentic oxytetracycline standards (Figure 2c). Using MRM mode and a standard curve generated from authentic oxytetracycline, the titer of heterologously expressed oxytetracycline was determined to be 2.0±0.1 mg per L of culture broth. These results demonstrate that when σ54 is over-expressed in E. coli containing the oxytetracycline gene cluster, all the biosynthetic genes are transcribed and translated, all the necessary proteins are functional in vivo, and the required starter and extender units are available, enabling oxytetracycline production.

thumbnail
Download:
Figure 2. Over-expression of σ54 enables E. coli to heterologously produce oxytetracycline.

(a) The enzymatic pathway responsible for the biosynthesis of oxytetracycline. (b). ESI-LC-MS/MS analysis of an oxytetracycline standard and the organic extracts from E. coli cultures containing the oxytetracycline gene cluster and over-expressing σ54, σS, or FecI. These traces show the ion extraction data from the Q3 scan of MS/MS experiments. Q1 was used to select the [MH]+ ion for oxytetracycline (m/z = 461). Ion extractions were performed for the oxytetracycline fragment m/z = 283 from the Q3 scan. Signals with a peak width of less than 0.1 s were regarded as noise and removed with the noise filter application. (c). The MS2 spectrum of the m/z = 461 peak for the oxytetracycline authentic standard and heterologously produced oxytetracycline. See also Figure S3.

https://doi.org/10.1371/journal.pone.0064858.g002

Because over-expression of σS and FecI also produced detectable oxyB transcripts, it is possible that these alternative sigma factors may also enable heterologous production of oxytetracycline. We therefore investigated the ability of E. coli BAP1 cultures over-expressing σS and FecI to heterologously produce oxytetracycline. LC-MS/MS analysis showed that no detectable oxytetracycline was present in these culture broths (Figure 2b). As expected from our transcription data, over-expression of σH, σF, and σE did not lead to heterologous production of detectable levels of oxytetracycline (Figure S3). These data demonstrate that only the alternative sigma factor, σ54, can positively regulate transcription of the oxytetracycline biosynthetic pathway, leading to heterologous production of oxytetracycline in E. coli.

Antibiotic activity of oxytetracycline does not limit heterologous expression

Oxytetracycline, like tetracycline, is a broad spectrum antibiotic that binds to the 30S subunit of the ribosome, inhibiting protein synthesis [31]. The minimum inhibitor concentration of E. coli strains to oxytetracycline can range from 0.5–16 µg/mL with a typical value of 8 µg/mL [32]. Because the oxytetracycline gene cluster used in this study lacked a resistance gene, we were concerned that the activity of oxytetracycline at our maximum titer of 2 ug/mL may limit heterologous expression. To address this concern, we generated a tetracycline resistance vector pLRVP09 based on pBR322 and transformed this into a BAP1 strain with the oxytetracycline gene cluster and a σ54 expression vector. Heterologous production of oxytetracycline from this strain was identical to strains lacking the pBR322 derived TetC resistance determinant. This data indicates that the antibiotic activity of oxytetracycline does not limit the heterologous production of oxytetracycline to 2 mg/L.

σ54 over-expression does not lead to substantial accumulation of the oxytetracycline biosynthetic proteins

While most classes of bacterial natural products have been heterologously expressed in E. coli, type II polyketide biosynthetic pathways have not been [22]. Even when key type II biosynthetic genes, such as those encoding the ketosynthase (KS) or chain length factor (CLF), are placed under the control of known E. coli promoters, it has not been possible to produce type II polyketides, due predominantly to the production of insoluble KS-CLF heterodimer [22], [33]. Because σ54 over-expression leads to production of oxytetracycline, some soluble, functional KS (OxyA) and CLF (OxyB) must be expressed. To evaluate if σ54 over-expression constituted an effective method to produce soluble KS-CLF heterodimer, we examined the soluble fraction of oxytetracycline producing strains by SDS-PAGE (Figure S4). While a band consistent with σ54 (54 kDa) could be easily detected, there was no evidence for high-level production of OxyA (45 kDa) or OxyB (44 kDa). Thus while σ54 over-expression must lead to some functional soluble KS-CLF heterodimer, there is not substantial accumulation of these key proteins.

σ54 directly regulates heterologous production in E. coli

σ54 can positively regulate transcription directly or indirectly. For direct regulation, the σ54-RNAP complex binds to a σ54 consensus promoter upstream of the operon under σ54 control. The σ54 consensus promoter is unique among bacterial promoters with highly conserved residues −12 and −24 from the transcriptional start site (TGGCACG-N4-TTGC(T/A)) [34] and can reliably be identified from sequence data. If σ54 is directly regulating transcription of oxyB, a σ54 promoter consensus sequence must be present upstream of the operon containing oxyB. To determine if σ54 promoters were present in the oxytetracycline biosynthetic pathway the entire gene cluster was examined for σ54 promoter consensus sequences using the web-based tool PromScan (http://molbiol-tools.ca/promscan/) [35]. A high scoring consensus promoter was located upstream of the operon containing oxyB in a location that would allow direct transcription of the putative operon (Figure 3a, Table S2). The location and consensus sequence of the identified promoter provides strong evidence for direct regulation by σ54 of the heterologous production of oxytetracycline in E. coli.

thumbnail
Download:
Figure 3. Putative σ54 promoters are found in the majority of polyketide and non-ribosomal peptide biosynthetic gene clusters.

(a) The oxyABCDEF operon from the S. rimosus oxytetracycline biosynthetic pathways contains a putative σ54 promoter. The promoter shows high homology to the σ54 promoter consensus sequence especially at the key −12 and −24 positions. A mutation of the highly conserved GG residues at −24 (−24*) and 5′ RACE were used to confirm that this promoter is responsible for direct, positive transcriptional regulation of the oxyB gene by σ54 over-expression. (b) Transcription of the oxyABCDEF operon is directly controlled by the σ54 promoter upstream of oxyA. Mutation of the highly conserved GG residues −24 from the transcriptional start site to TT leads to a greater than 40 fold decrease in oxyB transcript levels as determined by qPCR. qPCR data was obtained from BAP1 transformed with pDCS11 and either pDCS61 or pDCS62. (c) Results of a bioinformatics analysis of 58 bacterial genomes containing 180 polyketide synthase (PKS) and non-ribosomal peptide synthetase (NRPS) biosynthetic gene clusters show that the majority of gene clusters contain putative σ54 promoters. It is particularly intriguing that Actinobacteria possess gene clusters with σ54 promoters since they lack the gene encoding σ54. See also Tables S2, S3, S4.

https://doi.org/10.1371/journal.pone.0064858.g003

To demonstrate that the predicted σ54 promoter is functional and involved in the transcriptional regulation of oxyB, the highly conserved GG −24 from the transcriptional start site was mutated to TT. The −24 region plays a major role in anchoring the σ54-RNAP complex to the promoter and mutation of the conserved GG is known to decrease transcription of genes under σ54 control by ten to one hundred-fold [36], [37]. qPCR analysis showed that oxyB transcript levels from the TT mutant σ54 promoter were greater than forty-fold lower as compared to the wild type (Figure 3b, Table S1). To identify promoter elements involved in transcription of the operon containing oxyB, we performed 5′ Rapid Amplification of cDNA Ends (5′ RACE). Our 5′ RACE data identified a transcriptional start site consistent with the proposed σ54 promoter, confirming that this promoter is functional and that oxyB transcription is directly regulated by σ54.

σ54 promoters are common in polyketide and non-ribosomal peptide biosynthetic pathways

If σ54 promoters are commonly found in polyketide and non-ribosomal peptide biosynthetic pathways, σ54 over-expression may be a general route to heterologous expression of polyketides and non-ribosomal peptides. To evaluate this possibility, we examined the genomes of 58 sequenced bacteria for polyketide and non-ribosomal peptide biosynthetic gene clusters and σ54 promoters. The 58 species included major secondary metabolite producers such as 10 Actinobacteria, 20 Proteobacteria, 9 Firmicutes, as well as 19 species from diverse phyla. This selection of species contains representative examples of most of the sequenced bacterial phylogenic diversity.

Manual examination of all genes annotated as polyketide synthase, 3-oxoacyl acyl carrier protein synthase, and non-ribosomal peptide synthetase (COG3321, COG0304, and COG1020 respectively) [38] and their adjacent genes, identified 180 polyketide and non-ribosomal peptide biosynthetic gene clusters (Figure 3c, Table S3).

Each bacterial genome was analyzed to identify all putative σ54 promoters. To identify σ54 promoters, a positional weighted matrix (PWM) describing the σ54 promoter consensus sequence [36] was aligned with all sites on the forward and reverse strands of the genome. The fit at each position was scored using the published algorithm [35] from the bioinformatics tool PromScan. Sequences with a fit scoring 75 or greater and on the same strand, less than 500 bases upstream of a start codon were identified as putative σ54 promoters. All experimentally validated σ54 promoters in the E. coli genome had scores ranging from the 80 s to high 90 s [30]. In GC rich organisms such as M. xanthus a number of previously identified σ54 promoters had scores in the low to mid 70 s [39], [40]. Based on these data, a score of 75 is expected to provide the required low level of false negatives across a broad range of genomes.

To evaluate the level of false positives, a sequence randomized E. coli genome was analyzed [41]. 25.6% of genes in the native E. coli genome and 18.4% of the genes in the sequence randomized genome were predicted to have σ54 promoters. 18.4% thus provides a baseline for the level false positive predictions. This high false positive rate is typical of PWM based promoter prediction tools [42], [43].

As false positives are not expected to be conserved across species, a comparative genomic approach that examines the predicted promoters across a wide range of bacterial genomes should decrease the false positive rate and increases the specificity [44]. Our data shows that >50% of the bacterial genomes possessing a glnA ortholog (COG0174), the archetypical gene under direct σ54 transcriptional control, had one or more predicted σ54 promoters upstream of this gene. In comparison only 10% of the genomes examined had predicted σ54 promoters upstream of the well-characterized non-σ54 regulated ribosomal proteins encoding genes (COG0093, COG0049) [45], transcription elongation factor encoding genes (COG0264, COG0231) [46], [47], and the ATP synthase encoding genes (COG005, COG0224) [48]. Thus while individual promoter predictions have a high false positive rate, correlating promoter predictions across multiple species for individual orthologs substantially decreases the false positive predictions and increases the ability to accurately predict orthologs with σ54 promoters.

Of the 180 polyketide and non-ribosomal peptide biosynthetic gene clusters identified, 124 (69%) contained one or more σ54 promoters appropriately positioned to regulate transcription (Table S3, Table S4). 24 gene clusters possessed a single σ54 promoter, however the vast majority had two or more, with some pathways possessing up to two dozen σ54 promoters. Additionally 75 of the 124 (60%) clusters had a putative σ54 promoter appropriately placed to directly regulate either a polyketide synthase or a non-ribosomal peptide synthetase encoding ortholog. These results demonstrate that a majority (69%) of polyketide and non-ribosomal peptide biosynthetic gene clusters possess putative σ54 promoters and that these promoters are appropriately positioned to regulate transcription of at least one operon in these gene clusters. These results are consistent with the hypothesis that σ54 over-expression may be a general method for ensuring transcription of polyketide and non-ribosomal peptide biosynthetic gene clusters in heterologous hosts, such as E. coli.

Discussion

Heterologous expression of polyketide and non-ribosomal peptide biosynthetic gene clusters will play a major role in the discovery of new natural products from metagenomic and environmental DNA samples. Currently no general heterologous expression method appropriate for screening DNA libraries exists. Herein we describe a new mechanism for heterologous expression of a polyketide biosynthetic pathway, where a high-level, pleiotropic alternative sigma factor from the heterologous host positively regulates transcription of the biosynthetic gene cluster. In contrast, known methods for heterologous expression rely on either replacing each native promoter with known, well-characterized promoter from the heterologous host, such as the T7 promoter in E. coli [16]–[18], or rely on the heterologous host to constitutively express each gene from the native promoters [10]–[15]. Our approach, which actively induces transcription of gene clusters by over-expression of alternative sigma factors, may provide a general solution to the heterologous expression problem that is compatible with screening DNA libraries.

Our results demonstrate that over-expression of the sigma factor, σ54, enables efficient heterologous expression of oxytetracycline biosynthetic gene cluster in E. coli. Using qPCR, we have demonstrated that no transcript is detectable under standard culture conditions for the key oxytetracycline biosynthetic gene oxyB. LC-MS/MS analysis of these cultures showed no detectable oxytetracycline production in the absence of detectable oxyB transcripts. Over-expression of σ54 positively regulates transcription of oxyB, generating detectable levels of oxyB transcripts, enabling production of oxytetracycline. Bioinformatics analysis of the oxytetracycline biosynthetic gene cluster revealed a σ54 promoter sequence appropriately positioned to directly regulate oxyB transcription. Site directed mutagenesis and 5′ RACE strongly support that this promoter is functional. Further bioinformatic analysis of 180 polyketide and non-ribosomal peptide biosynthetic pathways shows that the majority of these pathways possess putative σ54 promoter sequences (69%), suggesting that σ54-mediated heterologous expression may be a general phenomenon.

Substantial evidence exists to support the hypothesis that over-expression of alternative sigma factors, such as σ54, can positively regulate diverse polyketide biosynthetic pathways. Nutrient limitation, which is one of the only known general regulators of polyketide biosynthesis [21], initiates the stringent response in bacteria. During stringent response the levels of the global bacterial regulator (p)ppGpp increases. Increases in (p)ppGpp have been shown to correlate with polyketide production in diverse bacterial species [25], [26]. For example relA, the gene encoding the (p)ppGpp synthetase, in S. coelicolor is required to trigger polyketide biosynthesis26. (p)ppGpp disrupts the interaction between the RNA polymerase (RNAP) and the principle sigma factor, σ70, enabling RNAP to interact with alternative sigma factors, which increases alternative sigma factor-mediated transcription [24]. Thus nutritional limitation and (p)ppGpp exert their regulatory effects by increasing alternative sigma factor-mediated transcription, suggesting that polyketide biosynthetic pathways may be under either direct or indirect alternative sigma factor transcriptional control.

We investigated the effect of over-expression of all the alternative sigma factors from E. coli on the transcription of oxyB and on the production of oxytetracycline. Only σ54 over-expression led to both detectable levels of the oxyB transcript and detectable levels of oxytetracycline production. σ54 is known to play a major role in response to nitrogen limitation in most bacteria [27] and sporulation in delta-proteobacteria [49]. Interestingly, both nitrogen limitation and sporulation correlate with polyketide production [21], [25]. While this suggests that σ54 may play a role in regulation of polyketide biosynthesis in some bacteria, substantial work in native hosts is required to test this hypothesis.

A unique aspect of σ54-mediated transcription is that it requires co-activation to initiate transcription [50]. Enhancer binding proteins (EBPs) allow σ54-loaded RNA polymerase to form a transcriptionally active open complex. EBPs contain a DNA binding domain that interacts with a specific DNA sequence called the upstream activating sequence (UAS), directing the EBP to a specific σ54 promoter. The ATPase domain of the EBP then catalyzes ATP hydrolysis, leading to open complex formation and transcription. Without this co-activation step, transcription cannot occur. High levels of EBPs have been shown in vivo and in vitro to enable formation of the active open complex independent of EBP binding to the UAS [51], [52].

As our data supports direct σ54-mediated transcription of the operon containing oxyB in the heterologous host E. coli, co-activation must be occurring. σ54 over-expression has been shown to positively regulate many of the twelve EBPs native to E. coli [53], including glnG, zraR, and fhlA [30]. As no EBPs are present in the oxytetracycline gene cluster (no open reading frames in the oxytetracycline gene cluster possess the key AAA+ and helix-turn-helix domains found in EBPs [53]), a potential hypothesis for co-activation in our heterologous transcription is the UAS-independent co-activation by a positively regulated native E. coli EBP.

Over-expression of σS and FecI showed an increase in oxyB transcription but did not lead to detectable levels of oxytetracycline production. oxyB transcript levels were at least 10 fold lower for σS and FecI over-expression than when compared to σ54 over-expression. Presumably the low levels of the oxyB transcript were insufficient to generate enough OxyB to produce detectable levels of oxytetracycline. A plausible explanation for the transcription of oxyB during σS and FecI over-expression is cross-talk between these alternative sigma factors and σ54, leading to low levels of σ54-mediated transcription. This is supported by the observation that σ54, σS, and FecI are known to be linked by the polyamine response [54], [55].

The identification of a functional σ54 promoter in the S. rimosus oxytetracycline biosynthetic gene cluster and 236 putative σ54 promoters in 61 gene clusters from seven different actinobacterial genomes is highly unexpected (Table S3). Actinobacteria do not contain a gene encoding σ54. Three hypotheses could account for this observation. These promoters are non-functional in Actinobacteria and are remnants of horizontal biosynthetic pathway transfer from non-Actinobacteria where these σ54 promoters are functional. Alternatively, a functional equivalent of σ54 could be present. Extensive research efforts into regulation of polyketide biosynthesis in Streptomyces has not uncovered a functional equivalent to σ54, however the vast majority of these studies have been carried out in S. coelicolor, the only Streptomyces in our genome analysis to contain no σ54 promoters in any of its 10 polyketide and non-ribosomal peptide biosynthetic gene clusters (Table S3). A third hypothesis is that these promoters represent false positives from the bioinformatics-based σ54 promoter prediction. While the prediction of promoters for orthologs across a wide range of bacterial species, as is done in this study, can decrease the false positive rate associate with PWM-based analyses and improve the selectivity of genome wide predictions, experimentation is ultimately required to evaluate the functionality of individual promoters. Understanding the origins and roles of these putative σ54 promoter sequences thus remains a wide-open question.

We have developed a new E. coli-based heterologous expression system for oxytetracycline polyketide biosynthetic gene clusters. We have demonstrated the over-expression of the alternative sigma factor σ54 directly and positively regulates heterologous expression of the oxytetracycline biosynthetic gene cluster in E. coli. Bioinformatics analysis indicates that σ54 promoters are present in nearly 70% of polyketide and non-ribosomal peptide biosynthetic pathways suggesting that σ54-mediated heterologous expression may be an effective, general approach. If sufficiently general, this approach may facilitate heterologous expression of new polyketides from metagenomic and environmental DNA samples.

This study also opens the door to characterizing and engineering polyketide biosynthetic pathways in E. coli. The vast majority of genetic experiments, such as deletion or complementation experiments, used to characterize these biosynthetic pathways [22], [56]–[61] have been carried out in either native producing organisms or in Streptomyces heterologous hosts. By developing an E. coli-based heterologous expression system, these experiments can be carried out in the highly genetically malleable E. coli, simplifying these studies.

This study represents the first example of successfully expressing functional type II polyketide synthases in E. coli [22], [33]. Type II polyketide synthases have been extremely challenging to produce in E. coli, even when placed under the control of known E. coli promoters. A major hurdle to production of type II PKS in E. coli has been the ability to generate soluble KS-CLF dimer, an essential part of the polyketide biosynthetic machinery. When expressed as standalone or fusion proteins, KS and CLF have always formed insoluble inclusion bodies [22], [33]. The prevailing hypothesis for this result is an incompatibility between the rates of protein synthesis, subunit folding, and heterodimerization [33]. Our work shows that while it is possible to generate soluble functional KS and CLF from the oxytetracycline pathway, the KS-CLF-heterodimer is expressed at very low levels, limiting this tools utility as a recombinant protein expression system. Our results will however be of use for in vivo characterization of aromatic polyketide backbone construction and the ensuing tailoring chemistries.

Finally, it is not unreasonable to suggest that over-expression of other sigma factors, such as σH, σS or FecI, may positively regulate transcription of other biosynthetic gene clusters, potentially expanding the scope of alternative sigma factor-mediated heterologous expression.

Materials and Methods

General

Escherichia coli XL1 Blue (Stratagene) and E. coli TOP10 (Invitrogen) cells were used for routine cloning and plasmid preparations. E. coli BAP1 was used in all heterologous production experiments. E. coli cells were grown in LB media (Fisher) supplemented with the appropriate antibiotics when necessary. Oxytetracyline standard and IPTG were purchased from Sigma-Aldrich. Antibiotics and media components were purchased from Fisher. All strains were chemically or electronically transformed using standard transformation protocols.

Plasmid Construction

PrimeSTAR HS polymerase and the Thermocycler Mastercycler personal were used for all PCR reactions. Pfu Ultra II polymerase and the Thermocycler Mastercycler personal were used for mutagenesis reactions. Primers can be found in the expanded experimental procedures in the supplementary information (File S1). E. coli MG1655 genomic DNA was used as the template. Each sigma factor was cloned into pCR-Blunt and confirmed by sequencing. Genes were sub-cloned using NdeI and EcoRI restriction sites into pET28b or pKH22 (a pET21c derivative with an AvrII site engineered after the native EcoRI site) [62], except for rpoE and rpoD, where HindIII instead of EcoRI and NheI instead of NdeI were used to due presence of native restriction sites. To construct pDCS61 the oxyTA1ABCD cassette was amplified from pMRH08 [12] using forward primer 5′ – TAATACGACTCACTATAGGG – 3′ and the reverse primer 5′ – CAGTGAATTCTCATAGCTCCAGGCTG - 3′. The oxyTA1ABCD cassette was then digested with EcoRI and inserted into pET28b. Site directed mutagenesis was then performed on pDCS61 using forward primer 5′ – CCTGCGTCCCCTAAAACGGCGGTGGC – 3′ and the reverse primer 5′ – GCCACCGCCGTTTTAGGGGACGCAGG – 3′ following the QuikChange mutagenesis protocol to construct pDCS62. The mutant was confirmed by sequencing. To construct pLRVP09, pBR322 was digested with ScaI and SspI to remove the β-lactamase marker and ligated closed. Plasmid names and descriptions can be found in the expanded experimental procedures in the supplementary information (File S1).

qPCR Analysis

Strains examined were BAP1+pDCS11, BAP1+pMRH08, BAP1+pMRH08+either pDCS11, pDCS57, pDCS58, pDCS59, pNDP6, or pNDP7, and BAP1+pDCS11+either pDCS61 or pDCS62. All strains were grown in 25 mL LB in shake flasks at 37°C. At O.D.600 = 0.4, cultures were induced with 1 mM IPTG and incubated at 20°C for 24 hours. Cells were harvested for RNA isolation. In concert with RNA isolation the broth from each culture was extracted and analysed for oxytetracycline production as described below. The SV Total RNA Isolation System was used to isolate RNA. cDNA was generated from RNA using AMV Reverse Transcriptase. qPCR experiments were performed in triplicate for each strain with a 10-fold dilution series of the 16 s RNA acting as an internal standard. Primers were designed to provide 180–210 bp amplicons of the rpoN, oxyB, oxyF, oxyK, oxyP, and oxyT transcripts. qPCR primer sequences can be found in the expanded experimental procedures in the supplemental information (File S1). The qPCR reactions contained 1.0 µM of each primer, 12.5 µL Absolute SYBR Green QPCR mix, 0.5 µL prepared cDNA, and 10.5 µL dH2O to give a total reaction volume of 20 µL. A Mx3000 qPCR Thermocycler was used with the following conditions: 1 cycle 95°C for 15 min, 40 cycles of 95°C for 15 s, 55°C for 30 s, and 72°C for 30 s, followed by denaturation for 1 cycle at 95°C for 1 min, 55°C for 30 s, and 95°C for 30 s. Standard curve R2 values and amplification efficiency values ranged from 0.991–1.0 and 90.7%→100% respectively. The amplification efficiency was calculated using the formula A = 10(−1/slope) in which the slope was calculated by regression analysis obtained from the Ct values versus calculating log number of cells in serial dilutions (See Table S1).

Oxytetracycline Production, Isolation, Characterization

The negative control strains were prepared by electroporating E. coli BAP1 with either pDCS11, pDCS57, pDCS58, pDCS59, pNDP6, or pNDP7. These strains lacked the oxytetracycline pathway and thus cannot produce oxytetracycline. Test strains were prepared by electroporation of E. coli BAP1 with pMRH08, the oxytetracycline gene cluster, and either no additional plasmid, pDCS11, pDCS57, pDCS58, pDCS59, pNDP6, or pNDP7. Tetracycline resistant strains were generated by transforming electrocompetent BAP1/pMRH08 with pDCS11 and pLRVP09. All strains were grown in 25 mL LB medium with appropriate antibiotics in shake flasks at 37°C. At O.D.600 = 0.4, cultures were induced with 1 mM IPTG and grown at 20°C for 48 hours. Cultures were treated with acetone (1 mL) and vigorously vortexed. Cell debris was removed by centrifugation and Amberlite XAD-16 resin (0.50 g) was added to the clarified media. The resin was incubated with the media overnight and collected by filtration. The organic extracts were eluted from the resin with 20 mL MeOH over 2 hours and concentrated in vacuo. Extracts were then resuspended in 100 µL MeOH for ESI-LC-MS/MS analysis. All ESI-LC-MS/MS analysis was performed on an API2000 LC/MS/MS System equipped with a turbo-ion spray ESI probe interfaced with a Prominence UFLC. A reverse phase BDS Hypersil C18 column (100 mm×2.1 mm I.D., 3 µm particle size,) was employed. Mobile phases throughout experimental were A: 5% MeCN∶95% H2O with 0.05% formic acid and B: 95% MeCN∶5% H2O with 0.05% formic acid. For initial determination of oxytetracycline production a gradient program (3 min 100% A, 40 min linear gradient to 100% B, 10 min 100% B) was developed with a flow rate of 0.250 mL/min into the mass spectrometer (Ion spray 5,500 V, mass range 250–550 m/z). For oxytetracycline MS2 fragmentation scans a gradient program (3 min 100% A, 30 min linear gradient to 100% B, 2 min 100% B) was developed and the mass spectrometer settings were optimized for oxytetracycline (Ion spray 5,500 V, collision energy 50 eV, Q3 scan 400–500 m/z, MS2 parent ion 461.3 m/z, MS2 product ion range 198–462 m/z). For MRM quantification of oxytetracycline production, a gradient program identical to that of the MS2 program was use with the following mass spectrometer settings: Ion spray 5,500 V, collision energy 33 eV, MRM Q1 461.3 m/z, MRM Q3 483.2 m/z. Quantification was performed in triplicate using a standard curve generated from authentic oxytetracycline.

Analysis of protein over-expression

BAP1, BAP1/pDCS11, and BAP1/pDCS11/pMRH08 were each grown in 25 mL LB medium with appropriate antibiotics in shake flasks at 37°C. At O.D.600 = 0.4, cultures were induced with 1 mM IPTG and grown at 20°C for 24 hours. Cells were harvested from 5 mL of cell culture by centrifugation at 4000 g and resuspended in 200 µL of lysis buffer (100 mM sodium

Phosphate pH 8.0, 300 mM NaCl, 10% (v/v) glycerol, 1 mg/mL lysozyme, 1 µg/mL pepstatin A, and 1 µg/mL leupeptin). The cells were disrupted by sonication on ice, and cell debris was removed by centrifugation at 15000 g and 4°C. The resulting soluble fraction was analyzed by 4–20% gradient SDS-PAGE followed by staining with Coomassie Brilliant Blue.

Gene cluster identification

To identify polyketide and non-ribosomal peptide biosynthetic gene clusters from bacterial genomes, all genes with cluster of orthologous groups (COG) annotations [35] of polyketide synthase (COG3321), 3-oxoacyl acyl carrier protein synthase (COG0304), and non-ribosomal peptide synthetase (COG1020) were identified from 58 bacterial genomes. A list of genomes used can be found in the expanded experimental procedures in the supplementary information (File S1). COG0304 annotated genes with gene names containing fab were discarded as they are involved in fatty acid biosynthesis. Operons containing genes commonly associated with secondary metabolite biosynthesis were included in the gene cluster. Transposons and housekeeping genes were considered to occur outside of the biosynthetic gene cluster. All identified gene clusters can be found in the supplementary information (Table S2).

σ54 promoter identification

To identify putative σ54 promoter sequences in bacterial genomes the PromScan Perl script [32] was modified to output all hits, including intragenic hits, with a score of 65 or higher. The complete script can be found in the expanded experimental procedures in the supplementary information (File S1). The script was run on Windows using Strawberry Perl (http://strawberryperl.com/). The inputs were bacterial genome DNA sequence files (FNA files from NCBI's Genome Database) and the σ54 promoter positional weighted matrix file, which is based on 186 known σ54 promoter sites and can be found in the expanded experimental procedures in the supplementary information [33].

The results were uploaded to an SQL database created using SQL Server 2008 Express (http://www.microsoft.com/express/). PTT files (from NCBI's Genome Database) containing annotation data were uploaded to the database. Using LINQ to SQL, hits generated from the modified PromScan algorithm were linked to genes located within 500 bp upstream and in the same coding direction as the gene start site. Promoter prediction data is available on the web at http://www.sigma54.ca.

5′ RACE

Total RNA was isolated from BAP1 with pMRH08 and pDCS11 as described for qPCR analysis. cDNA was generated using AMV Reverse Transcriptase and the oxyA specific primer 5′ – AGGCTCATCGTCATGCCGCA – 3′, Poly(dC) tails were added using terminal deoxynucleotidyl transferase (Fermentas). The tailed cDNA was amplified by PCR by PrimeStar HS DNA polymerase (Takara) using the oxyA specific primer and the abridged anchor primer 5′ – GGCCACGCGTCGACTAGTACGGGIIGGGIIGGGII – 3′. The diluted PCR reaction was used as the template for a second round of PCR amplification using the oxyA nested primer 5′ – CCGCCTCGCCCGTCACAC – 3′ and the abridged universal anchor primer 5′– GGCCACGCGTCGACTAGTAC – 3′. The PCR products were gel purified and reamplified using the oxyA nested primer and the abridged universal anchor primers. The PCR products were then cloned into pCR-Blunt and sequenced.

Supporting Information

Figure S1.

Transcript levels, determined by qPCR, for E. coli transformed with the oxytetracycline gene cluster (pMRH08) compared to a null strain lacking the oxytetracycline gene cluster.

https://doi.org/10.1371/journal.pone.0064858.s001

(TIF)

Figure S2.

Transcript levels, determined by qPCR, for E. coli transformed with the oxytetracycline gene cluster (pMRH08) and over-expressing σ54 compared to a null strain lacking the oxytetracycline gene cluster and over-expressing σ54.

https://doi.org/10.1371/journal.pone.0064858.s002

(TIF)

Figure S3.

ESI-LC-MS/MS Q1 ion extraction chromatograms for the oxytetracycline [M+H]+ m/z = 461. Standard: an oxytetracycline standard. Positive control: Organic extracts from a culture of BAP1 with the oxytetracycline biosynthetic gene cluster over-expressing σ54. Negative control: organic extracts from a culture of BAP1 with the oxytetracycline biosynthetic gene cluster. σE: organic extracts from a culture of BAP1 with the oxytetracycline biosynthetic gene cluster over-expressing σE. σF: organic extracts from a culture of BAP1 with the oxytetracycline biosynthetic gene cluster over-expressing σF. σH: organic extracts from a culture of BAP1 with the oxytetracycline biosynthetic gene cluster over-expressing σH. MS2 traces further confirm that no oxytetracycline can be detected in the σE, σF, and σH samples.

https://doi.org/10.1371/journal.pone.0064858.s003

(TIF)

Figure S4.

SDS-PAGE of the soluble fraction from oxytetracycline producing strains. Lane 1 is protein ladder, lane 2 is the soluble fraction from uninduced BAP1, lane 3 is the solubule fraction 24 h post induction from BAP1/pDCS11, lane 4 is the soluble fraction 24 h post induction from BAP1/pMRH08/pDCS11. A band corresponding to σ54 (54 kDa) is seen in lanes 3 and 4. The OxyA-OxyB KS-CLF heterodimer (45 kDa and 44 kDa respectively) is not obeserved in lane 4.

https://doi.org/10.1371/journal.pone.0064858.s004

(TIF)

File S1.

Expanded experimental procedures.

https://doi.org/10.1371/journal.pone.0064858.s005

(PDF)

Table S1.

Complete data set for qPCR analysis as shown in Figures 1 and 3. Standard curve, efficiencies, R2 and Ct values are given.

https://doi.org/10.1371/journal.pone.0064858.s006

(PDF)

Table S2.

List of putative σ54 promoters identified from a bioinformatics analysis of the oxytetracycline gene cluster.

https://doi.org/10.1371/journal.pone.0064858.s007

(TIF)

Table S3.

List of PKS, NRPS and NRPS/PKS gene clusters with the associated σ54 promoters identified from our bioinformatics analysis.

https://doi.org/10.1371/journal.pone.0064858.s008

(PDF)

Table S4.

Tabulated number of PKS, NRPS and NRPS/PKS gene clusters with one of more σ54 promoters identified from our bioinformatics analysis at increasing cutoff stringencies.

https://doi.org/10.1371/journal.pone.0064858.s009

(TIF)

Acknowledgments

The authors thank Philip Pelletier for helpful discussion.

Author Contributions

Conceived and designed the experiments: DCS CNB. Performed the experiments: DCS KC NP LRVP CNB. Analyzed the data: DCS KC CNB. Contributed reagents/materials/analysis tools: AGG. Wrote the paper: DCS CNB.

References

  1. 1. Zhang W, Ames B, Tsai S, Tang Y (2006) Engineered Biosynthesis of a Novel Amidated Polyketide, Using the Malonamyl-Specific Initiation Module from the Oxytetracycline Polyketide Synthase. Appl Environ Microbiol 72: 2573–2580.
  2. 2. Tang L, Shah S, Chung L, Carney J, Katz L, et al. (2000) Cloning and Heterologous Expression of the Epothilone Gene Cluster. Science 287: 640–642.
  3. 3. Schwecke T, Aparicio JF, Molnár I, König A, Khaw LE, et al. (1995) The biosynthetic gene cluster for the polyketide immunosuppressant rapamycin. Proc Natl Acad Sci USA 92: 7839–7843.
  4. 4. Newman DJ, Cragg GM (2007) Natural Products as Sources of New Drugs over the Last 25 Years. J Nat Prod 70: 461–477.
  5. 5. Bode H, Müller R (2006) Analysis of myxobacterial secondary metabolism goes molecular. J Ind Microbiol Biotechnol 33: 577–588.
  6. 6. Udwary DW, Zeigler L, Asolkar RN, Singan V, Lapidus A, et al. (2007) Genome sequencing reveals complex secondary metabolome in the marine actinomycete Salinispora tropica. Proc Natl Acad Sci USA 104: 10376–10381.
  7. 7. Van Lanen SG, Shen B (2006) Microbial genomics for the improvement of natural products discovery. Curr Opin Microbiol 9: 252–260.
  8. 8. Li JWH, Vederas JC (2009) Drug Discovery and Natural Products: End of an Era or an Endless Frontier? Science 325: 161–165.
  9. 9. Rodriguez E, Menzella HG, Gramajo H (2009) Heterologous Production of Polyketides in Bacteria. Methods Enzymol 459: 339–365.
  10. 10. Kao CM, Katz L, Khosla C (1994) Engineered biosynthesis of a complete macrolactone in a heterologous host. Science 265: 509–512.
  11. 11. Binz TM, Wenzel SC, Apotheke HJS, Bechthold A, Müller R (2008) Heterologous Expression And Genetic Engineering of the Phenalinolactone Biosynthetic Gene Cluster by Using Red/ET Recombineering. ChemBioChem 9: 447–454.
  12. 12. Stevens DC, Henry MR, Murphy KA, Boddy CN (2010) Heterologous expression of the oxytetracycline biosynthetic pathway in Myxococcus xanthus. Appl Environ Microbiol 76: 2681–2684.
  13. 13. Baltz RH (2010) Streptomyces and Saccharopolyspora hosts for heterologous expression of secondary metabolite gene clusters. J Ind Microbiol Biotechnol 37: 759–772.
  14. 14. Chen Y, Wendt-Pienkowski E, Shen B (2008) Identification and utility of FdmR1 as a Streptomyces antibiotic regulatory protein activator for fredericamycin production in Streptomyces griseus ATCC 49344 and heterologous hosts. J Bacteriol 190: 5587–5596.
  15. 15. Gomez-Escribano JP, Bibb MJ (2011) Engineering Streptomyces coelicolor for heterologous expression of secondary metabolite gene clusters. Microbial Biotech 4: 207–215.
  16. 16. Pfeifer BA, Admiraal SJ, Gramajo H, Cane DE, Khosla C (2001) Biosynthesis of complex polyketides in a metabolically engineered strain of E. coli. Science 291: 1790–1792.
  17. 17. Boddy CN, Hotta K, Tse ML, Watts RE, Khosla C (2004) Precursor-directed biosynthesis of epothilone in Escherichia coli. J Am Chem Soc 126: 7436–7437.
  18. 18. Watanabe K, Hotta K, Praseuth AP, Koketsu K, Migita A, et al. (2006) Total biosynthesis of antitumor nonribosomal peptides in Escherichia coli. Nat Chem Biol 2: 423–428.
  19. 19. Jørgensen H, Fjærvik E, Hakvåg S, Bruheim P, Bredholt H, et al. (2009) Candicidin Biosynthesis Gene Cluster Is Widely Distributed among Streptomyces spp. Isolated from the Sediments and the Neuston Layer of the Trondheim Fjord, Norway. Applied Environ Microbiol 75: 3296–3303.
  20. 20. Fisch KM, Gurgui1 C, Heycke N, van der Sar SA, Anderson SA, et al. (2009) Polyketide assembly lines of uncultivated sponge symbionts from structure-based gene targeting. Nat Chem Biol 5: 494–501.
  21. 21. Bibb MJ (2005) Regulation of secondary metabolism in streptomycetes. Curr Opin Microbiol 8: 208–215.
  22. 22. Zhang W, Li Y, Tang Y (2008) Engineered biosynthesis of bacterial aromatic polyketides in Escherichia coli. Proc Natl Acad Sci USA 105: 20683–20688.
  23. 23. Pickens LB, Tang Y (2010) Oxytetracycline biosynthesis. J Biol Chem 285: 27509–27515.
  24. 24. Potrykus K, Cashel M (2008) (p)ppGpp: Still Magical? Annu Rev Microbiol 62: 35–51.
  25. 25. Hesketh A, Sun J, Bibb MJ (2001) Induction of ppGpp synthesis in Streptomyces coelicolor A3(2) grown under conditions of nutritional sufficiency elicits actII-ORF4 transcription and actinorhodin biosynthesis. Mol Microbiol 150: 1485–1493.
  26. 26. Knauber T, Doss SD, Gerth K, Perlova O, Müller R, et al. (2008) Mutation in the rel gene of Sorangium cellulosum affects morphological and physiological differentiation. Mol Microbiol 69: 254–266.
  27. 27. Wösten MM (1998) Eubacterial sigma factors. FEMS Microbiol Rev 22: 127–150.
  28. 28. Britton RA, Eichenberger P, Gonzalez-Pastor JE, Fawcett P, Monson R, et al. (2002) Genome-Wide Analysis of the Stationary-Phase Sigma Factor (Sigma-H) Regulon of Bacillus subtilis. J Bacteriol 184: 4881–4890.
  29. 29. Chen G, Schellhorn HE (2003) Controlled induction of the RpoS regulon in Escherichia coli, using an RpoS-expressing plasmid. Can J Microbiol 49: 733–740.
  30. 30. Zhoa K, Liu M, Burgess RR (2010) Promoter and regulon analysis of nitrogen assimilation factor, σ54, reveal alternative strategy for E. coli MG1655 flagellar biosynthesis. Nucleic Acid Res 38: 1273–1283.
  31. 31. Chopra I, Roberts M (2001) Tetracycline antibiotics: mode of action, applications, molecular biology, and epidemiology of bacterial resistance. Microbiol Mol Biol Rev 65: 232–260.
  32. 32. Burrows GE, Morton RJ, Fales WH (1993) Microdilution antimicrobial susceptibilities of selected gram-negative veterinary bacterial isolates. J Vet Diagn Invest 5: 541–547.
  33. 33. Gao X, Wang P, Tang Y (2010) Engineered polyketide biosynthesis and biocatalysis in Escherichia coli. Appl Microbiol Biotechnol 88: 1233–1242.
  34. 34. Buck M, Gallegos MT, Studholme DJ, Guo Y, Gralla JD (2000) The Bacterial Enhancer-Dependent σ54 (σN) Transcription Factor. J Bacteriol 182: 4129–4136.
  35. 35. Studholme DJ, Buck M, Nixon T (2000) Identification of potential N-dependent promoters in bacterial genomes. Microbiology 146: 3021–3023.
  36. 36. Barrios H, Valderrama B, Morett E (1999) Compilation and analysis of σ54-dependent promoter sequences. Nucleic Acids Res 27: 4305–4313.
  37. 37. Gronewold TMA, Kaiser D (2007) Mutation of the Act Promoter in Myxococcus xanthus. J Bacteriol 189: 1836–1844.
  38. 38. Tatusov RL, Galperin MY, Natale DA, Koonin EV (2000) The COG database: a tool for genome-scale analysis of protein functions and evolution. Nucleic Acid Res 28: 33–36.
  39. 39. Ossa F, Diodati ME, Caberoy NB, Giglio KM, Edmonds M, et al. (2007) The Myxococcus xanthus Nla4 Protein Is Important for Expression of Stringent Response-Associated Genes, ppGpp Accumulation, and Fruiting Body Development. J Bacteriol 189: 8474–8483.
  40. 40. Giglio KM, Eisenstatt J, Garza AG (2010) Identification of Enhancer Binding Proteins Important for Myxococcus xanthus Development. J Bacteriol 192: 360–364.
  41. 41. Mulligan ME, McClure WR (1986) Analysis of the occurrence of promoter-sites in DNA. Nucleic Acids Res 14: 109–126.
  42. 42. Gordon JJ, Towsey MW, Hogan JM, Mathews SA, Timms P (2006) Improved prediction of bacterial transcription start sites. Bioinformatics 22: 142–148.
  43. 43. McLeay RC, Leat CJ, Bailey TL (2011) Tissue-specific prediction of directly regulated genes. Bioinformatics 27: 2354–2360.
  44. 44. Qiu P (2003) Recent advances in computational promoter analysis in understanding the transcriptional regulatory network. Biochem Biophys Res Commun 309: 495–501.
  45. 45. Nomura M, Gourse R, Baughman G (1984) Regulation of the Synthesis of Ribosomes and Ribosomal Components. Ann Rev Biochem 53: 75–117.
  46. 46. Aseev LV, Levandovskaya AA, Tchufistova LS, Scaptsova NV, Boni IV (2008) A new regulatory circuit in ribosomal protein operons: S2-mediated control of the rpsB-tsf expression in vivo. RNA 14: 1882–1894.
  47. 47. Aoki H, Adams SL, Chung DG, Yaguchi M, Chuang SE, et al. (1991) Cloning, sequencing and overexpression of the gene for prokaryotic factor EF-P involved in peptide bond synthesis. Nucleic Acids Res 19: 6215–6220.
  48. 48. Kumar R, Shimizu K (2011) Transcriptional regulation of main metabolic pathways of cyoA, cydB, fnr, and fur gene knockout Escherichia coli in C-limited and N-limited aerobic continuous cultures. Microb Cell Fact 10: 3.
  49. 49. Jakobsen JS, Jelsbak L, Jelsbak L, Welch RD, Cummings C, et al. (2004) σ54 Enhancer Binding Proteins and Myxococcus xanthus Fruiting Body Development. J Bacteriol 186: 4361–4368.
  50. 50. Schumacher J, Joly N, Rappas M, Zhang X, Buck M (2006) Structures and organisation of AAA+ enhancer binding proteins in transcriptional activation. J Struct Biol 156: 190–199.
  51. 51. Dworkin J, Jovanovic G, Model P (1997) Role of Upstream Activation Sequences and Integration Host Factor in Transcriptional Activation by the Constitutively Active Prokaryotic Enhancer-binding Protein PspF. J Mol Biol 273: 377–388.
  52. 52. Janaszak A, Majczak W, Nadratowska B, Szalewska-Pałasz A, Konopa G, et al. (2007) A σ54-dependent promoter in the regulatory region of the Escherichia coli rpoH gene. Microbiology 153: 111–123.
  53. 53. Studholme DJ, Dixon R (2003) Domain Architectures of σ54-Dependent Transcriptional Activators. J Bacteriol 185: 1757–1767.
  54. 54. Terui Y, Higashi K, Taniguchi S, Shigemasa A, Nishimura K, et al. (2007) Enhancement of the synthesis of RpoN, Cra, and H-NS by polyamines at the level of translation in Escherichia coli cultured with glucose and glutamate. J Bacteriol 189: 2359–2368.
  55. 55. Terui Y, Higashi K, Tabei Y, Tomitori H, Yamamoto K, et al. (2009) Enhancement of the Synthesis of RpoE and StpA by Polyamines at the Level of Translation in Escherichia coli under Heat Shock Conditions. J Bacteriol 191: 5348–5357.
  56. 56. Lackner G, Schenk A, Xu Z, Reinhardt K, Yunt ZS, et al. (2007) Biosynthesis of Pentangular Polyphenols: Deductions from the Benastatin and Griseorhodin Pathways. J Am Chem Soc 129: 9306–9312.
  57. 57. Buntin K, Irschik H, Weissman KJ, Luxenburger E, Blöcker H, et al. (2010) Biosynthesis of Thuggacins in Myxobacteria: Comparative Cluster Analysis Reveals Basis for Natural Product Structural Diversity. Chem Biol 17: 342–356.
  58. 58. Carlson JC, Fortman JL, Anzai Y, Li S, Burr DA, et al. (2010) Identification of the tirandamycin biosynthetic gene cluster from Streptomyces sp. 307–9. ChemBioChem 11: 564–572.
  59. 59. Zaleta-Rivera K, Charkoudian LK, Ridley CP, Khosla C (2010) Cloning, Sequencing, Heterologous Expression, and Mechanistic Analysis of A-74528 Biosynthesis. J Am Chem Soc 132: 9122–9128.
  60. 60. Andexer JN, Kendrew SG, Nur-e-Alam M, Lazos O, Foster TA, et al. (2011) Biosynthesis of the immunosuppressants FK506, FK520, and rapamycin involves a previously undescribed family of enzymes acting on chorismate. Proc Natl Acad Sci USA 108: 4776–4781.
  61. 61. Yurkovich ME, Tyrakis PA, Hong H, Sun Y, Samborskyy M, et al. (2012) A Late-Stage Intermediate in Salinomycin Biosynthesis Is Revealed by Specific Mutation in the Biosynthetic Gene Cluster. ChemBioChem 13: 66–71.
  62. 62. Lundgren BR, Boddy CN (2007) Sialic acid and N-acyl sialic acid analog production by fermentation of metabolically and genetically engineered Escherichia coli. Org Biomol Chem 12: 1903–1909.