Biosensor-guided improvements in salicylate production by recombinant Escherichia coli

Construction of AraC-based sensor system for agar plate-based high-throughput screening

AraC variant “AraC-SA” (P8G, T24L, H80G, Y82I and H93R) responds to salicylate, as described previously [20]. To assess the specificity of this sensor, we compared the response (fold activation of GFP expression under the control of promoter P_BAD) to salicylate with that of several other similar molecules (substituted benzoates) (Fig. 2b). A concentration of 5 mM was used to compare compounds, as this amount is relevant to screening salicylate production, without major growth inhibition. Salicylate induces GFP expression over 200-fold, while there is essentially no response to the other compounds tested. The complete AraC-SA salicylate dose response is provided as Additional file 1.

To confirm that AraC-SA does not also respond to compounds upstream in the salicylate biosynthetic pathway (namely chorismate and isochorismate), strains expressing the full or partial salicylate pathway with the AraC-SA/GFP module were constructed (Fig. 2c). Only the strain expressing the full pathway produced salicylate (~ 4.5 mM after 14 h post-induction), resulting in 15-fold increased GFP fluorescence. Strains without the complete pathway exhibited only basal fluorescence signal. The AraC-SA sensor/reporter thus responds specifically to salicylate, at least in the context of salicylate biosynthesis in recombinant E. coli.

In using our sensor/reporter system to screen libraries for mutants with enhanced salicylate production, a colony-based screen was developed [21]. This approach provides a simple means of spatially isolating individual mutants and preventing “cheaters” from responding to salicylate produced and secreted by other mutants. To that end, a single copy of lacZ under control of P_BAD was integrated into the salicylate production host strain (QH4) [14], resulting in reporter strain SQ22 for library screening. As shown in Fig. 2d, in the presence of X-gal (5-bromo-4-cholo-3-indolyl β-d-galactopyranoside), the colony’s blue color intensity resulting from LacZ-catalyzed X-gal hydrolysis correlates with the salicylate concentration. Further validation of this colony screen for isolating clones with enhanced salicylate production is provided in the (Additional file 2).

Salicylate biosynthetic pathway RBS library construction and screening

Given the known metabolic burden imposed by gene overexpression, and potential toxicities or enzyme inhibition resulting from accumulation of pathway intermediates [22, 23], we reasoned that one approach to improving salicylate production in recombinant E. coli is to fine-tune the relative expression levels of the six key biosynthetic pathway genes (described above). The RBS calculator [7] was used to design a series of RBS sequences resulting in a range of predicted translation initiation strengths for the six targeted genes (entC, pchB, aroL, ppsA, tktA, aroG^fbr) (refer to Additional file 3 for details). These six genes with varied RBS sequences were assembled combinatorially into plasmid pPCC1244, leading to pathway library “QSAlib3” composed of theoretically 38,880 unique members (Fig. 3a). Strain SQ22 containing a chromosomal P_BAD-lacZ reporter was transformed with the QSAlib3 library and spread onto large, X-Gal-containing plates for screening. Strain SQ22 is derived from strain QH4 [14] having pheA and tyrA gene deletions (Table 1), which prevents conversion of chorismate to phenylalanine and tyrosine. Yeast extract (5 g/L) was therefore added to our modified minimal medium to support growth.

Plasmids	Relevant features	References
pZE-EP	entC, pchB, and luc genes under control of Plac; AmpR	[14]
pCS-APTA	aroL ppsA, tktA, aroGfbr under control of Plac; KanR	[14]
pPCC1115-5	CRIM plasmid with PBAD-lacZ;;AprR	[21, 29]
pPCC1244	araC-SA gene under the control of Ptac; AprR	In this study
pPCC1250	entC, pchB genes under control of Ptac; AmpR	In this study
pPCC1251	araC-SA, entC, pchB genes under control of Ptac; AprR	In this study
pPCC1252	aroL ppsA, tktA, aroGfbr genes under control of Ptac; AprR	In this study
pPCC1253	araC-SA, entC, pchB genes under control of Ptac; aroL ppsA, tktA, aroGfbr genes under control of Ptac; AprR	In this study
pPCC1253-aroL	araC-SA, entC, pchB genes under control of Ptac; ppsA, tktA, aroGfbr genes under control of Ptac; AprR	In this study
pPCC507	From vector pJA1. Transposable element containing a chloramphenicol resistance gene flanked by IS10 inverted repeat sequences; R6 K replication origin; tnp gene encoding a mutant Tn10 transposase with 100-fold lower frequency of insertion at hot spots; AmpR	[21, 31]
pFG1	araC gene under the control of Ptac, tac; AprR	[18]
pFG29-SA	araC-SA gene under the control of Ptac; gfpuv under control of PBAD promoter, AprR	[20]
Strains
HF19	BW27786 (∆araFGH, ∆araBAD, ∆lacZ), with araC deleted	[16]
QH4	ATCC31884 with pheA and tyrA disrupted	[14]
SQ18	QH4 with PBAD-lacZ reporter integrated into chromosome at HK022 site	In this study
SQ22	SQ18 with LacZ in the lac operon disrupted	In this study
QH4Δrnd	QH4 with rnd gene disrupted	In this study
SQ22Δrnd	SQ22 with rnd gene disrupted	In this study

Approximately 100,000 colonies were screened by eye. Rescreening of clones representing the 75 darkest blue colonies was performed by quantifying salicylate production in liquid culture (using strain SQ22 harboring the selected mutant plasmids). Among these, the top six mutant plasmids based on salicylate titer were sequenced (Fig. 3c) and salicylate production using these constructs was further characterized with strain QH4 (Fig. 3b). Two clones produced about 13.8 ± 1.2 mM salicylate, which is > 120% improved relative to strain QH4 overexpressing all six target genes, each with the same “original” RBS sequence (AGGAGA) (plasmid pPCC1253). The time courses of salicylate production and cell density for QH4 + pPCC1253 and QH4 + pQSA-50 (provided in Additional file 4) show that salicylate production occurs primarily after growth has already slowed and culture densities are near maximum.

As shown in Fig. 3c, the RBS sequences appearing in the top six selected clones encompass the full range of weak to strong, for all six genes. This gene expression library is represented on polycistronic operons (Fig. 3a), so translational coupling [24] may at least partly explain the general lack of patterns to RBS strengths for the selected clones. While there are no prominent expression patterns for these genes, we note that most clones contain a relatively strong RBS upstream of tktA, and relatively weak RBS sequences upstream of aroL. SDS-PAGE analysis of cell lysates (using Coomassie blue staining) confirms these observations, in that strong bands are present for transketolase (though less prominent for the weaker RBSs in pQSA-78 and pQSA-94), and the two relatively stronger RBS sequences for aroL (pQSA-43 and pQSA-94) show higher aroL expression than the others (Additional file 5). We tested whether further lowering aroL expression by removing this gene from the plasmid (construct pPCC1253-aroL), and relying on only chromosomal expression of this gene, would further improve salicylate production. This however significantly reduced salicylate production (Fig. 3b). These results demonstrate the challenges of rationally tuning pathway gene expression levels, and the value in using combinatorial approaches to identify improvements.

Screening a transposon insertion library for further improvements in salicylate production

We next sought to mutate the host genome and screen for further improvements in salicylate production. A suicide plasmid vector containing a transposable DNA fragment and IPTG-inducible transposase was used to generate the transposon insertion library in strain SQ22 harboring plasmid pQSA-50. The SQ22-transposon insertion library was screened using our AraC-P_BAD-lacZ sensor-reporter system. After screening ~ 70,000 colonies on plates, 50 selected clones were re-screened in liquid culture for HPLC analysis. Among these, clone QSA#228 consistently produced ~ 16.5 mM salicylate after 48 h induction, which is ~ 27% improved over parent strain SQ22 harboring mutant plasmid pQSA-50 (Fig. 4). Curing the originally selected strain of plasmid, and re-transforming with fresh plasmid pQSA-50 resulted in the same improvement in salicylate titer. Self-formed adaptor PCR (SEFA-PCR) analysis [25] located the transposable element within the rnd gene, encoding RNase D. RNase D is an exonuclease involved in the 3′ ribonucleotide processing of tRNA [26]. Direct deletion of rnd from the genomes of host strains SQ22 and QH4 resulted in similar improvements in salicylate production. It is not known how changes in protein expression levels due to the rnd mutation are beneficial to salicylate biosynthesis.

General

Restriction enzymes, Phusion High-Fidelity DNA polymerase, and T4 DNA ligase were purchased from New England Biolabs (Ipswich, MA). Oligonucleotides were synthesized by Integrated DNA Technologies (Coralville, IA). DNA sequencing was performed at SeqWright (Houston, TX) or Genewiz (South Plainfield, NJ). All chemicals were purchased from Sigma-Aldrich (St. Louis, MO). Molecular biology techniques for DNA manipulation were performed according to standard protocols [27] and all cultures were grown in lysogeny broth (LB). Antibiotics and IPTG were prepared as a 1000× stock solution in purified water and sterile filtered with EMD Millipore Millex-GP syringe driven filters (EMD Millipore, Cat. No. SLGP033RS). The modified NBS medium contains MOPS (50 mM, pH 7.4) and per liter: 10 g glycerol; 2.5 g glucose; 5 g yeast extract; 3.4 g KH2PO₄; 5.2 g K₂HPO₄; 3.3 g (NH₄)₂HPO₄, 0.25 g MgSO₄·7 H₂O, 15 mg CaCl₂·2 H₂O, 0.5 mg thiamine, and 1 mL of trace metal stock (described by Chin and Cirino [28]). The concentrations of antibiotics used for maintaining the plasmids are as follows: apramycin 50 µg/ml, chloramphenicol 25 µg/ml, ampicillin 100 µg/ml.

Plasmids

Plasmids and strains used in this study are listed in Table 1. All primers used in this study are listed in Additional file 7. Primers pPCC1244-gib-for and pPCC1244-gib-rev were used to amplify lacI-Ptac-araC-SA DNA fragment from pFG29-SA, and the PCR product was assembled with PciI- and XmnI-digested pFG1, resulting in plasmid pPCC1244.

To create plasmid pPCC1250, pZE-EP was double digested with Bsu36I and SphI, and the digested DNA fragment was subjected to a DNA-blunting reaction. The blunt-ended DNA fragment was self-ligated, resulting in plasmid pPCC1250. A DNA fragment containing entC and pchB was amplified from pPCC1250 using primers pPCC1251-gib-for and pPCC1251-gib-rev2, and assembled with XmnI-digested pPCC1244, resulting in plasmid pPCC1251. A DNA fragment containing aroL ppsA, tktA, aroG^fbr genes was amplified from pCS-APTA using primers pPCC1252-gib-for and pPCC1251-gib-rev2, and the amplified DNA fragment was assembled with NdeI- and XmnI-digested pPCC1244, resulting in plasmid pPCC1252. Then, a DNA fragment of Ptac-araC-SA-entC-pchB was amplified from pPCC1251 using primers pPCC1253-NheI-for and pPCC1253-NotI-rev, and digested with NheI and NotI. The digested DNA fragment was ligated into pPCC1252 digested with the same restriction enzymes, resulting in plasmid pPCC1253. To remove aroL from plasmid pPCC1253, primers pPCC1251-gib-for and pPCC1253-Ptac-rev were used to amplify the entC-pchB DNA fragment from pPCC1253, and primers pPCC1253-ppsA-for and pPCC1253-tktA-rev were used to amplify the ppsA-tktA section from pPCC1253. The two amplified PCR products were assembled into KpnI/SphI double-digested pPCC1253, resulting in plasmid pPCC1253-aroL.

Construction of pathway RBS library

The RBS calculator [7] was used to design six RBS sequences having different translation initiation rates (TIRs) for each gene (see Additional file 1 for the RBS sequences and calculated TIRs). For each gene, the six designed upstream primers containing different RBS sequences were mixed equimolar, resulting in primer mixtures entC-RBSs-for, pchB-RBSs-for, etc., and each primer mixture was used to construct the RBS library. Primers pFG29_araC_GS_fwd_1 and AraC-gib-rev were used to amplify Ptac-AraC-SA DNA fragment from pPCC1244. Primers entC-RBSs-for and entC-RBS-rev were used to amplify entC from pZE-EP, and primers pchB-RBSs-for and pPCC1251-gib-rev2 were used to amplify pchB from pZE-EP. Plasmid pPCC1244 was double digested by BstAPI and XmnI, and the linearized vector was assembled with the three PCR products, resulting in QSALib1, which contains the RBS libraries for entC and pchB genes.

Primers pFG29_araC_GS_fwd_1 and Ptac-gib-rev were used to amplify a DNA fragment containing promoter Ptac. The aroL gene was amplified from pCS-APTA by primers aroL-RBSs-for and AroL-RBS-rev. Similarly, ppsA was amplified from pCS-APTA using primers ppsA-RBSs-for and ppsA-RBS-rev. Overlap extension PCR (OE-PCR) was performed to assemble these three PCR products using primers pFG29_araC_GS_fwd_1 and QSAlib2-OE123-rev, resulting in DNA fragment QSAlib2-f123. Next, primers tktA-RBSs-for and tktA-RBS-rev were used to amplify tktA from plasmid pCS-APTA, and primers AroG-RBSs-for and pPCC1251-gib-rev2 were used to amplify aroG. tktA and aroG were assembled by OE-PCR using primers QSAlib2-OE45-for and QSAlib2-OE45-rev, resulting in DNA fragment QSAlib2-f45. pPCC1252 was then double digested with BstAPI/BamHI, QSAlib2-f123 was double digested with BstAPI/SpeI, and QSAlib2-f45 was double digested with SpeI/BamHI. Ligation of these three digest fragments resulted in library QSAlib2.

Finally, primers pPCC1253-NheI-for and pPCC1253-NotI-rev were used to amplify Ptac-araC-SA-RBS-entC-RBS-pchB from QSAlib1. The PCR product was digested with NheI and NotI, and ligated with QSAlib3 digested with the same enzymes, resulting in QSAlib3. QSAlib3 contains the RBS libraries for all six genes. Sanger sequencing of QSAlib3 confirmed proper library construction.

Strain construction

The strains used in this study are listed in Table 1. Plasmid pPCC1155-5 was integrated into the chromosome of QH4 at site HK022 as described [29]. Removal of FRT-flanked apramycin resistance cassette resulted in strain SQ18. A Phage λ Red disruption method [30] was used to delete lacZ from the lac operon of strain SQ18, resulting in strain SQ22. Deletion of rnd in strains QH4 and SQ22 was similarly performed, resulting in strains QH4∆rnd and SQ22∆rnd, respectively.

Sensor-reporter fluorescence assays

Essentially as described [20], 500 μl LB + apramycin in 2-ml-well, 96-well plate was inoculated with strain HF19 harboring pFG29-SA. These starter cultures were incubated for 6 h at 37 °C, 900 rpm, then diluted to OD₅₉₅ = 0.05 in 500 μl “biosensor medium” containing different concentrations of the compound of interest. After 6 h, the cultures were pelleted and washed with an equal volume of phosphate-buffered saline, before measuring OD₅₉₅ and fluorescence (400 nm excitation, 510 nm emission) using plate readers.

Salicylate production in baffled flasks

A colony of the salicylate-producing strain was used to inoculate 3 ml LB + apramycin, and grown in a test tube for 8 h at 37 °C and 250 rpm. This seed culture was then diluted to OD₅₉₅ = 0.05 into 25 ml modified NBS medium containing apramycin and 250 mM IPTG, in 125 ml baffled flasks. The flasks were shaken at 37 °C and 250 rpm for 48 h, at which time OD₅₉₅ values were measured and salicylate concentrations were analyzed by HPLC.

Screening the RBS library for improved salicylate producers

Strain SQ22 was transformed with QSAlib3. The outgrowth was transferred into LB + apramycin, and grown at 37 °C, 250 rpm for 12 h. The resulting culture was diluted and spread onto large plates containing modified NBS-agar with IPTG (250 µM), X-Gal (40 µg/ml), and apramycin. After 24 h of incubation, the top 5 blue colonies from each screening plate were picked and streaked onto fresh LB plates supplied with apramycin. The resulting colonies were tested for salicylate production in liquid culture.

Construction and screening of SQ22 transposon insertion library

Strain SQ22 harboring the highest-producing salicylate plasmid (pQSA-50) was transformed with 1 µg of plasmid pPCC507, and the outgrowth was grown in 1 ml SOB medium supplied with 20 µM IPTG at 37 °C for 1 h. The outgrowth was transferred into 500 ml LB + apramycin and 12.5 µg/ml chloramphenicol, and grown at 37 °C for 12 h. The resulting culture was diluted and plated on modified NBS-agar plates containing IPTG (250 µM), X-Gal (40 µg/ml), apramycin, and 12.5 μg/ml chloramphenicol. Totally 70,000 colonies were screened on 10 screening plates. After 24 h of incubation, the top 5 blue colonies from each screening plate were picked and streaked onto fresh LB plates containing apramycin and 12.5 µg/ml chloramphenicol. The resulting colonies were tested for salicylate production in liquid culture.

Salicylate quantification by HPLC

500 µl of cell culture was centrifuged at 17,900×g, and the supernatant was filtered through a 0.45 µm filter. The salicylate concentration in the filtrate was determined by reverse-phase HPLC using a C18 column on a Shimadzu LC-20AD HPLC system (Kyoto, Japan) equipped with a UV monitor. The elution profile was as follows: Solvent A, 1% (v/v) acetic acid in water; solvent B, 1% (v/v) acetic acid in acetonitrile; gradient: 5–95% B (0–15 min), 95–5% B (15–17 min), 5% B (17–20 min). The column temperature was set to 50 °C. Salicylate eluted around 11.2 min at a flow rate of 1 ml/min. Elution absorbance at 310 nm was monitored and peak areas were converted to sample concentrations based on calibration with pure salicylate.