In comparison with the KO11 strain, PDC_Zm and ADH_Zm enzymatic specific activities were on average, 5 and 4-fold larger, respectively, in CCE14 during both exponential and stationary phases (Fig. 1). Even though the heterologous pathway was integrated into the chromosome of strains KO11 and CCE14 using the same method (see materials and methods section), strong differences were encountered between PDC_Zm and ADH_Zm enzymatic activities, as well as in transcript levels (as shown below). This behavior might be due to lineage differences between KO11 and CCE14. It is noteworthy that strain KO11 was submitted to high chloramphenicol pressure selection to increase ethanol productivity 3, in spite of this additional strategy, PDC_Zm and ADH_Zm activities were higher in CCE14, wich was not further selected on high chloramphenicol.

Fermentation performance in mineral medium with 40 g/L glucose is presented in Fig. 2. Fig. 2A–B shows results obtained for cell mass formation and glucose consumption. Table 2 summarizes rates obtained during the exponential and stationary phases. The growth rates of E. coli C, CCE14, and KO11 were similar. During exponential growth, specific glucose consumption rates of the two ethanologenic strains (KO11 and CCE14) were 28 and 34% higher, respectively, than that obtained for E. coli C, indicating that the glycolytic flux increased as a result of pdc_Zm and adh_Zm expressions. This behavior correlates well with previously reported results for KO11 fermenting xylose 26 or glucose 27 in Luria Broth, where the maximum sugar consumption rate was 50% higher for KO11 than for strain W. In spite of the fact that the growth rate was similar for the three strains, the maximum cell mass obtained at the onset of the stationary phase (after 20 hours of fermentation) was similar for KO11 and E. coli C, and lower for CCE14 (Table 2). These results, and data presented in Fig. 1, indicate that CCE14 directs more carbon to ethanol production than to biomass biosynthesis.

Fig 2C shows that no pyruvate was secreted by ethanologenic strains, but E. coli C produced a significant amount (> 3 g/L) of this metabolite during the stationary phase (Fig 2C). Furthermore, formate production by KO11 and E. coli C were similar, reaching up to 9 g/L when glucose was exhausted (Fig. 2D). However, formate production was lower than 2 g/L for CCE14. PDC_Zm was originally selected by Ohta et al 3, largely because it has a very high affinity for pyruvate (Km for pyruvate 0.4 mM) 2829 in comparison with all competing fermentation enzymes 24. This fact and our results indicate that competition occurs at the pyruvate node level, and that higher levels of PDC_Zm and ADH_Zm apparently allow more efficient carbon channeling through the heterologous ethanol pathway.

Although glycolytic fluxes for CCE14 and KO11 were similar, the ethanol specific formation rate in the exponential phase (first 6 hours of fermentation elapsed time) was 21% higher for CCE14 as compared to KO11 (Table 3). During this phase, the ethanol formation rate in E. coli C was negligible (Fig. 2E), whereas formate (Fig. 2D) and acetate (Fig. 2F) were the main products. The increase in ethanol production rate was obtained at the expense of acid production (Table 3, Fig. 2). Pyruvate and lactate (Fig. 2C, 2G) were not produced during this phase in CCE14, whereas formate, acetate, (Fig. 2D, 2F), and succinate (Fig. 2H) production were significantly lower than for E. coli C and KO11. Balances for this phase indicate carbon recoveries very close to 100% for the three strains evaluated (Table 3). These results indicate that the glycolytic flux can be controlled by the efficient conversion of pyruvate into ethanol through the enzymatic activity levels of PDC_Zm and ADH_Zm in ethanologenic E. coli. It is hypothesized that other fermentative pathways that allow the efficient regeneration of NAD⁺ could have the same effect.

During the stationary phase, CCE14 produced ethanol 80% faster than KO11 (Table 4). Accordingly, specific enzyme activities of PDC_Zm and ADH_Zm were 6 and 4.7-fold higher for CCE14 (Fig. 1). Ethanol yields at 30 hours were 15, 70 and 90% of theoretical yield for E. coli C, KO11, and CCE14, respectively.

Transcript levels of 49 genes from the CCE14 strain were analyzed in the exponential growth phase and normalized for values obtained with E. coli C. Levels of pdc_Zm and adh_Zm were normalized with KO11 values and analyzed for both exponential and stationary phases. A Student' t-test with a p value of ≤ 0.05 was applied to each set of normalized values in order to determine statistical significant differences in expression levels.

Interestingly, the comparison of CCE14 and KO11 gave a direct relationship between increases in both pdc_Zm and adh_Zm transcript values (4.51 and 9.06-fold, respectively) (Fig. 3) and specific enzyme activity levels (5.1 and 3.8-fold, respectively) (Fig. 1). It has been reported that pdc_Zm and adh_Zm mRNA are more stable than other transcripts in Z. mobilis 30. A correlation was also found between CCE14 and KO11 during the stationary phase for transcripts (3.66-fold and 4.78-fold for pdc_Zm and adh_Zm, respectively) and specific enzyme activity levels (6-fold and 4.7-fold, respectively) (Fig. 1). Nevertheless, absolute values of the enzyme activity for these enzymes decreased around 50% during the stationary phase. The increase in specific ethanol formation rate (Table 4) correlated with higher transcript (Fig. 3) and specific enzyme activity (Fig. 1) levels for CCE14 in comparison to KO11. The requirements of higher PDC_Zm enzymatic levels were demonstrated by Huerta-Beristain and co-workers 25. These authors reported that when PDC_Zm activity increased 7-fold, using a multicopy plasmid, the yield of ethanol from glucose increased from 70 to 85%, whereas organic acid formation rates were reduced in the KO11 strain. Accordingly, results in Fig. 1 clearly show that increases in the specific activities of the enzymes participating in the ethanologenic pathway boost the carbon flow to ethanol in strain CCE14. These results also suggest that high levels of both enzymes are essential to increase the ethanol production rate. Pyruvate pools in ethanologenic E. coli are substantially reduced by high-level expression of these Z. mobilis genes, but cells adjust their metabolism to various levels along the glycolytic pathway to fulfill the carbon flux. Direct increments between transcripts and specific enzyme activities demonstrate the correct translation of pyruvate decarboxylase and alcohol dehydrogenase when they are over expressed.

Enzyme activity levels for CCE14 strain were analyzed in exponential and stationary phases and normalized for values obtained with E. coli C (Fig. 4). Enzymatic levels for pdc_Zm and adh_Zm were normalized with KO11. A t-student test with a p value of ≤ 0.05 was applied to each set of normalized values to determine statistical differences in enzyme activity levels.

In the CCE14 strain, PTS and PGK enzymatic activities were higher, whereas TPI and PFL activities were lower during the exponential growth phase. These data correlate with the fact that ethanologenic strains consume glucose at a higher rate. In addition, increases in PGK activity suggest that the over expression of the genes coding for PDC_Zm and ADH_Zm modify the ATP/ADP balance. It is known that in aerobic conditions, a high ATP demand causes an increase in glycolytic flux in E. coli 59. For Lactococcus lactis, the control of glycolytic flux resides to a large extent in processes outside the pathway, such as ATP consuming reactions and glucose transport 60. As discussed above, the heterologous ethanologenic pathway increases glycolytic flux with the subsequent increases in ATP production and consumption. Therefore, it appears that cells tend to increase ATP formation through an increase in PGK synthesis; although no increase in PYK activity was found. ATP is also produced when acetate is formed; however, CCE14 does not produce acetate during the exponential phase. A decrease in PFL activity in CCE14 correlates with a strong reduction in formate production (Fig. 2D). On the other hand, the observed 50% reduction in the TPI specific enzyme activity does not reduce glycolytic flux. A large increase in ZWF specific activity was found, and was discussed above.

A comparison of enzymatic activity data between CCE14 and E coli C during the stationary phase, indicates higher values of PTS, FDP, GAPDH, PGK, and PYK for the ethanologenic strain. The LDH enzymatic activity was lower in CCE14. It is known that lactate dehydrogenase is allosterically activated by pyruvate 61. Lactate production was found only during the stationary phase of two cultures. Pyruvate formation in E. coli C correlates with lactate production (Fig. 2C–G), and lower levels of this metabolite in CCE14 correlate with a 40% lower LDH specific activity (Fig. 4). These results suggest that a higher ethanol formation rate, i.e., higher PDC_Zm and ADH_Zm specific activities originate higher rates of glucose transport, ATP synthesis, and NAD-NADH+H turnover.

E. coli strains used in this work are listed in Table 1. With the purpose to have an ethanologenic strain with different PDC_Zm and ADH_Zm enzymatic levels, a new strain CCE14 was constructed integrating into the chromosome the pyruvate decarboxylase (pdc) and the alcohol dehydrogenase (adh) genes of Zymomonas mobilis under the control of the pfl native promoter. This chromosome integration was made with pLOI510 as described previously, 3. The vector pLOI510, was constructed to allow direct selection for the integration of pdc and adhB of Z. mobilis genes into the pfl region of the chromosome by using a DNA fragment which lacks a replicon 3. Transformants were screened for chloramphenicol resistance (20 μg/ml) and CO₂ production in tubes, and subsequently tested for ethanol production in mini-fermentors with 20 g/l of glucose.

All stock cultures were stored at -70°C in Luria Broth (LB) medium 62 containing 40% glycerol. To develop inocula, cells were transferred twice on LB-agar plates supplemented with 20 g/L of glucose-chloramphenicol (20 μg/ml), and no chloramphenicol for E. coli C. Single colonies were transferred to overnight cultures in shake flasks (35°C, 120 rpm), containing glucose (20 g/L) in mineral M9-medium 62. M9-medium contains: 6 g/L Na₂HPO₄, 3 g/L KH₂PO₄, 1 g/L NH₄Cl, 0.5 g/L NaCl. The following components were sterilized by filtration, and then added (per liter of final medium): 2 ml of 1 M MgSO₄7H₂O, 1 mL of 0.1 M of CaCl₂, 1 mL of 1 mg/mL thiamine-HCl. These cells were harvested in exponential growth phase by centrifugation and used to inoculate non-aerated mini-fermentors 63, containing 200 ml of M9-medium with 40 g/L of glucose. Starting OD₆₀₀ was 0.1, chloramphenicol (0, 40 and 20 μg/ml) was included for strains C, KO11 and CCE14, (respectively), cells were cultivated at 35°C, 100 rpm, and the pH was maintained at 7 by the automatic addition of 2 N KOH. All cultures were carried out in triplicate.

Samples were periodically taken from cultures to measure optical density at 600 nm using a spectrophotometer (Beckman DU-70, Palo Alto, CA) and the dry cell weight was calculated using a previously determined conversion factor of 1 OD₆₀₀ = 0.37 g/L. An HPLC system (600E quaternary bomb, 717 automatic injector, 2410 refractive index, and 996 photodiode array detectors, Waters, Milford, MA) and an Aminex HPX-87H column (300 × 7.8 mm; 9 μm) (Bio-Rad, Hercules, CA) were used to separate and quantify D-glucose, formate, acetate, succinate, and lactate concentrations. Running conditions were: mobile phase, 5 mM H₂SO₄, flow 0.5 ml/min, and temperature 50°C. Under these conditions glucose was detected by refractive index, and organic acids were identified by photodiode array at 210 nm. Ethanol was quantified by gas chromatography (Agilent 6850, Wilmington, D.E.), using 1-butanol as internal standard.

Samples were taken at the mid-exponential and stationary phases. All operations were carried out at 4°C. 1 mL of cell culture was harvested by centrifugation at 10,000 × g for 10 min, washed twice with 1 mL of 100 mM Tris-HCl (pH 7.0) containing 20 mM KCl, 5 mM MnSO₄, 2 mM DTT and 0.1 mM EDTA, and then suspended in 1 mL of the same buffer. Cells were disrupted by four sonication steps (15 s each) in an ultrasonic disrupter (Soniprep 150, UK). The cell debris was removed by centrifugation; 10 min at 10,000 × g. The resulting crude extracts were used immediately for determination of enzymatic activities and protein, or stored at -20°C.

Enzyme activities were measured spectrophotometrically at 340 nm in a thermostatically controlled (30°C) spectrophotometer (BioMate 5, ThermoSpectronic, NY). All compounds of the reaction mixtures were pipetted into 1 cm light path cuvettes, reactions were initiated by adding the cell extract or substrate to give a final volume of 1 mL. The millimolar extinction coefficient for NAD⁺, NADH, NADP⁺ and NADPH is 6.22 cm^-1. mM^-1.

The assay conditions for glucose:PEP phosphotransferase (PTS), 6-phosphofructosekinase (PFK), fructose-1,6-bisphosphatase (FDP), glucose phosphate isomerase (PGI), fructose bisphosphate aldolase (FDP aldolase), glyceldehyde-3-phosphate dehydrogenase (GAPDH), triose phosphate isomerase (TPI), 3-phosphoglycerate kinase (PGK), pyruvate kynase (PYK), 6-phosphogluconate dehydrogenase (ZWF), pyruvate-formate lyase (PFL), and lactate dehydrogenase (LDH) were measured based on the methods used by Peng et al 64. The assay conditions used for alcohol dehydrogenase (ADH_Zm) and pyruvate decarboxylase (PDC_Zm) were based on the methods reported by Conway et al 65. Phosphoglycerate mutase was measured based on the method of Maitra et al 66. Protein concentration was estimated by the Bradford method, 67, with bovine serum albumin used as the standard. Each assay was performed three times for the same culture, from three independent experiments. A t-student test with a p value of ≤ 0.05 was applied to each set of normalized values in order to determinate statistical differences in enzyme activity levels.

Total RNA extraction was performed using hot-phenol equilibrated with water, precipitated with 3 M sodium acetate and ethanol, and treated with DNase kit (DNA-free™, Ambion; 33. RNA integrity was tested by densitometry in 1.2% agarose gels. RNA quantification was performed by absorbance at 260/280 nm. cDNA was synthesized using RevertAid™ H First Strand cDNA Synthesis kit (Fermentas Inc.) and a mixture of specific DNA primers. The sequences of the primers used for cDNA synthesis were those reported by Flores et al 33, except for pdc_Zm (5'-GACAAAGTTGCCGTCCTCGT and 5'-ATGGTAGCAACTGCGCCAC) and adh_Zm (5'-TTACCCCGATGGTTTCCGT and 5'-TTCAAATGCGTGGGTCAGAG) genes. cDNA obtained in this way was used as template for RT-PCR assays. Reproducibility of this procedure was determined by performing two separate cDNA synthesis experiments from the RNA extracted for each strain. Similar results were obtained for the transcription of all genes that were measured.

Real-time PCR (RT-PCR) was performed with the ABI Prism 7000 Sequence Detection System (Perkin Elmer/Applied Biosystems, Foster City, CA), using the SYBR Green PCR Master Mix (Perkin Elmer/Applied Biosystems, Foster City, CA) and amplification conditions described by Flores et al 33. The primers for specific amplification were designated using the Primer Express software (Perkin Elmer/Applied Biosystems, Foster City, CA). The size of all amplimers was 101 bp. The final primer concentration, in a total volume of 15 μl, was 0.2 μM. Five nanograms of target cDNA for each gene was added to the reaction mixture. All experiments were performed in triplicate for each gene of each strain, obtaining very similar values. A non-template control reaction mixture was included for each gene. The quantification technique used to analyze data was the 2−ΔΔCT MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xH8viVGI8Gi=hEeeu0xXdbba9frFj0xb9qqpG0dXdb9aspeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaGaeGOmaiZaaWbaaSqabeaacqGHsislcqqHuoarcqqHuoarcqqGdbWqdaWgaaadbaGaeeivaqfabeaaaaaaaa@3316@ method described by Livak and Shmittgen 68. The data were normalized using the ihfB gene as an internal control (housekeeping gene). We detected the same expression level of this gene in all the strains in the conditions in which the bacteria were grown. For each analyzed gene in all strains, the transcription level of the wild type gene, considered as one, was used as the control to normalize the data. Data is reported as relative expression levels compared to the expression levels of E. coli C. A t-student test with a p value of ≤ 0.05 was applied to each set of normalized values in order to determinate statistical differences in expression levels.