The PSP model was developed to quantitatively predict the strength of promoters on the basis of the nucleotide sequence [3]. A relative strength of 0.20 was calculated with this model for the endogenous promoter of the ppc gene (ppcp) in a scale from 0 to 1.00 defined by a library of 42 synthetic promoters. Out of the library, two stronger synthetic promoters, p37 (0.82) and p55 (0.36), were chosen for the knock-in experiments.

The ppcp was chosen to be rationally replaced. Upstream of the ppc gene, there is a 91 nt leader region harboring two RBS sequences, a promoter region with -10 and -35 boxes [12], and three repeated sequences (from -62 to -506) (figure 1). The relevance to ppc expression of the repeated sequences is not known.

In a first attempt, a cassette containing the antibiotic resistance marker, the synthetic p37 promoter, a canonical RBS, a start codon and a polyhistidine sequence (tag) substituted the start codon of the ppc gene (figure 2a). The obtained mutant was grown in triplicate at flask scale to compare gene expression at transcriptomic and enzymatic level to the wild-type (wt), (table 1). The insertion of the artificial promoter resulted in a 2-fold reduction of expression of the ppc gene. Therefore, a second strategy was designed to replace the sequence from position -506 nt to the start codon (+94) by the p37 promoter with an artificial RBS and a start codon followed by a polyhistidine sequence (figure 2b) to eliminate the influence of a putative transcription factor binding site [13] which is located upstream of the endogenous promoter on the artificial promoter. Interestingly, an even stronger reduction in the mRNA level (5- to 6-fold) compared to the wt was observed (table 1). To exclude that the observed reduction of expression was due to the presence of the polyhistidine sequence, the latter approach was repeated by inserting the p37 promoter with its RBS but without the polyhistidine sequence (not shown in the diagram). However, a similar reduction of expression was observed (0.3 (0.2-0.4)). A possible explanation is that the synthetic leader region was not adequate to ensure mRNA stability [14]. Consequently, a third strategy was designed, respecting the natural RBS region, with introducing neither the polyhistidine sequence nor the synthetic RBS. The cassette with the p37 promoter replaced the sequence from -506 to -1 (figure 2c). This approach resulted in a successful 3- to 4-fold increase in expression of the ppc gene (3.9 compared to 1.0). A similar increase was observed at the enzymatic level (3.33 versus 1.00) (table 1). These results are in agreement with what was predicted by the PSP model.

Two synthetic promoters (p37 and p55) were chosen on the basis of their relative strength to replace the natural ppc promoter. After optimizing the promoter knock-in procedure with the p37 promoter, the same approach was followed with the p55 promoter. The expression of the ppc gene in the wt and the mutant was measured as before. A 2.5-fold increase of ppc expression was observed (table 2). As shown in figure 3, there is a good correlation between the relative expression from the synthetic promoters at both transcriptomic and enzymatic activity levels and the predicted promoter strengths using the PSP model.

Next, the influence of the genetic background on the expression of ppc from the synthetic promoter was investigated. The p37 and p55 promoters were knocked-in in an E. coli MG1655 mutant in which three genes coding for enzymes involved in the pyruvate oxidation pathway were knocked-out. The expression of the ppc gene in the mutant strains was measured relative to the wt strain at transcriptomic and enzymatic activity level (table 2). The expression of the ppc gene in the 3KO strain seems to be down regulated compared to the wt strain (0.7 versus 1.0) despite the fact that no regulatory elements have yet been identified in the ppc operon. According to Lesnik and coworkers (2005) [13], this may be due to a putative transcription factor binding site that is located upstream of the ppc promoter. These results seem to indicate that the introduction of additional knock-outs (blocking the acetate pathway and the pyruvate oxidizing pathway) may create a novel phenotype of which the ppc expression is down regulated compared to the wt. Hence, this new phenotype probably influences a regulator which binds to this putative transcription factor binding site. Furthermore, these results confirm that the relative strength of both p37 (3.9 and 3.7) and p55 (2.5 and 2.3) promoters was maintained in both the wt and the 3KO strains, respectively. Moreover, the results at enzymatic level had the same tendency (3.3 - 3.6 for promoter p37 and 1.8 - 2.0 for promoter p55 in the wt - 3KO strains).

Next, the influence of growth parameters (oxygen supply, pH and culture volume (scaling up)) on the strength of the knocked-in synthetic promoters in both the wt and the 3KO strains was investigated using a bioreactor. Again, a similar relative expression of the ppc gene was observed at both transcriptomic and enzymatic activity levels, showing that scaling up the culture conditions affect the strength of the synthetic promoters keeping the same tendency (6.7 - 5.7 for promoter p37 and 5.0 - 4.2 for promoter p55 in the wt - 3KO strains) (table 2).

The wild-type (wt) strain Escherichia coli MG1655 [λ^-, F^-, rph-1, rfb-50, ilvG^-, fnr^-] was obtained from the Netherlands Culture Collection of Bacteria (NCCB, Utrecht, The Netherlands). The mutant strain E. coli MG1655 ΔackA-pta, ΔpoxB [λ^-, F^-, rph-1, rfb-50, ilvG^-, fnr^-, ΔackA-pta, ΔpoxB] was constructed using the method of Datsenko & Wanner (2000) [20] and it is referred to as 3KO. The plasmids pKD46 (Red helper plasmid, Ampicillin resistance), pKD3 (containing an FRT-flanked chloramphenicol resistance (cat) gene), pKD4 (containing an FRT-flanked kanamycin resistance (kan) gene), and pCP20 (expressing FLP recombinase activity) were obtained from Prof. Dr. J-P Hernalsteens (Vrije Universiteit Brussel, Belgium). The artificial promoter library was constructed by De Mey et al. (2007) [3]. The plasmid pBluescript (Fermentas, St. Leon-Rot, Germany) was used to create the promoter delivery constructs.

The culture medium Luria Broth (LB) consisted of 1% tryptone-peptone (Difco, Erembodegem, Belgium), 0.5% yeast extract (Difco) and 0.5% sodium chloride (VWR, Leuven, Belgium). The pH of the medium was 6.7.

For flask cultures, minimal medium (MM-flask) consisted of 18 μM FeCl₂.4H₂O (Merck, Leuven, Belgium), 34 μM CaCl₂.2H₂O (Merck), 8.3 μM MnCl₂.2H₂O (Merck), 2.2 μM CuCl₂.2H₂O (Sigma, Bornem, Belgium), 2,1 μM CoCl₂.6H₂O (Merck), 6.9 μM ZnCl₂ (Merck), 0.4 μM H₃BO₄ (Merck), 40.3 μM Na₂EDTA.2H₂O (Fluka, Bornem, Belgium), 3 μM thiamine HCl (Sigma), 0.4 μM Na₂MoO₄.2H₂O (Fluka), 37.4 mM NH₄Cl (Merck), 37.8 mM (NH₄)₂SO₄ (Merck), 22 mM KH₂PO₄ (Acros, Geel, Belgium), 42 mM K₂HPO₄ (Acros), 40 mM MOPS (Sigma), 2 mM MgSO₄.7H₂O (Fluka), 8.6 mM NaCl (VWR) and 83.3 mM glucose.H₂O (Stop, Dendermonde, Belgium). The pH was set at 7.0 with a 1 M K₂HPO₄ (Acros) solution.

For batch cultures, the minimal medium (MM-batch) composition was identical to MM-flask, except for the concentration of KH₂PO₄ 14.7 mM, and the absence of MOPS. The pH was not set to 7.0, but left at approx 5.4.

A preculture from a single colony was grown in 5 ml MM-flask medium overnight and 2 ml was transferred to 100 ml MM-flask medium in a 0.5 l flask. Incubation was performed at 37°C in a rotary shaker (160 rpm) for 16 hours. The inoculum was set at OD₆₀₀ = 0.5, and 75 ml was used to inoculate 1.5 l MM-batch in a Biostat M fermentor (Sartorius Stedim Biotech S.A., Melsungen, Germany). In batch cultures, the pH (7.0) was kept constant using 4 N KOH and 1 N H₂SO₄, the temperature, agitation and air supply were set at 37°C, 1000 rpm, and 1.5 l/min, respectively. The pH, pO₂, temperature, agitation, used acid and used base were followed online using the MFCS/WIN software of Sartorius Stedim Biotech S.A. Samples were taken using a rapid sampling loop. Each hour, a sample for OD₆₀₀ and extracellular measurements was taken using the stainless bead sampling method as described by Mashego et al. [21], followed by cold centrifugation. During exponential growth, a sample was taken every 30 min. At OD₆₀₀ = 1, 1 ml samples were taken for total RNA extraction and enzymatic activity measurements.

Two selected promoters (p37, and p55) (table 3) were amplified from an existing promoter library [3] by PCR with primers (Fw-EcoRI-p37/Rv-BamHI-p37, and Fw-EcoRI-p55/Rv-BamHI-p55, respectively) flanked with restriction site regions (EcoRI and BamHI) (table 3). The antibiotic resistance genes (for chloramphenicol or kanamycin resistance) flanked with FRT sites were amplified by PCR with primers (Fw-EcoRI-P1/Rv-HindIII-P2) carrying the restriction site regions (EcoRI and HindIII) and priming sites from pKD3 and pKD4, respectively. The PCR products were digested with the appropriate restriction enzymes and introduced in a vector (p-Bluescript) previously linearised (BamHI and HindIII). After verification of the promoter sequence, the final plasmid was used as template in the promoter knock-in procedure.

The promoter knock-in system is based on the λ Red-mediated one step recombination procedure for creating a knock-out mutant as described by Datsenko and Wanner [20], with several modifications. The basic strategy is to replace a chromosomal sequence with a cassette containing a selectable antibiotic resistance gene and the sequence to be inserted. The cassette is generated by PCR by using primers with circa 50 nt homology extension (H1 and H2).

The design of the primers (table 3) for the replacement of the endogenous ppc promoter (ppcp) was based on different strategies: (i) insertion of an artificial promoter, plus an artificial ribosomal binding site (RBS) and a polyHis-tag sequence between the ppc gene and the ppcp (Fw-ppc-HIS-RBS-37(55)/Rv-ppc-P2); (ii) replacement of ppcp by an artificial promoter, an artificial RBS and a polyHis-tag sequence (Fw-ppc-HIS-RBS-37(55)/Rv-ppc-3-P2); (iii) replacement of ppcp by an artificial promoter, but respecting the transcriptional starting point and the natural RBS (Fw-ppc-37(55)/Rv-ppc-3-P2). In strategies I and II, a polyHis-tag was included to facilitate detection and purification of the protein.

The wild-type and the mutants were grown in flasks in 20 ml MM-flask medium in triplicate. One ml samples were collected at OD₆₀₀ = 1.0 for mRNA and protein collection. Total RNA extraction was done using the RNeasy mini kit of Qiagen^® (KJ Venlo, The Netherlands). The purity of RNA was verified on a FA-agarose gel as recommended by Qiagen^® and the RNA concentration was determined by measuring the absorbance at 260 nm. 2 μg RNA was used to synthesize cDNA using a random primer and RevertAid H Minus M-MulV reverse trancriptase (Fermentas).

The strength of the promoter was determined by RT-qPCR carried out in an iCycler IQ^® (Bio-Rad, Eke, Belgium) using the primers Fw-ppc-qPCR and Rv-ppc-qPCR. SYBR GreenER qPCR supermix (Invitrogen^®) was used to perform a brief UDG (uracil DNA glycolsylase) incubation (50°C for 2 min) immediately followed by PCR amplification (95°C for 8.5 min; 40 cycles of 95°C for 15 s and 60°C for 1 min) and melting curve analysis (95°C for 1 min, 55°C for 1 min and 80 cycles of 55°C+0.5°C/cycles for 10 s) to identify the presence of primer dimers and analyze the specificity of the reaction. This UDG incubation step before PCR cycling destroys any contaminating dU-containing products from previous reactions. UDG is then inactivated by the high temperatures during normal PCR cycling, thereby allowing the amplification of genuine target sequences. Each sample was performed in triplicate. The relative expression ratios were calculated using the "Delta-delta ct method" of PE Applied Biosystems (Perkin Elmer, Forster City, CA). The gene rpoB was used as housekeeping gene using the primers Fw-rpoB-qPCR and Rv-rpoB-qPCR.

Cell lysis was performed with the EasyLyse™-kit (Epicentre^® Biotechnologies, BIOzymTC, Landgraaf, Netherlands), following the procedure recommended by the supplier. PEP carboxylase activity was assayed as described by De Maeseneire et al. [22]. The absorbance at 415 nm was measured in a microplate reader (680 XR microplate reader, Bio-Rad, Eke, Belgium). PEP carboxylase activity was measured in the knock-in mutants and in the wild-type.