ABSTRACT
The biosynthesis of alginate has been studied extensively due to the importance of this polymer in medicine and industry. Alginate is synthesized from fructose-6-phosphate and thus competes with the central carbon metabolism for this metabolite. The alginate-producing bacterium Pseudomonas fluorescens relies on the Entner-Doudoroff and pentose phosphate pathways for glucose metabolism, and these pathways are also important for the metabolism of fructose and glycerol. In the present study, the impact of key carbohydrate metabolism enzymes on growth and alginate synthesis was investigated in P. fluorescens. Mutants defective in glucose-6-phosphate dehydrogenase isoenzymes (Zwf-1 and Zwf-2) or glucose dehydrogenase (Gcd) were evaluated using media containing glucose, fructose, or glycerol. Zwf-1 was shown to be the most important glucose-6-phosphate dehydrogenase for catabolism. Both Zwf enzymes preferred NADP as a coenzyme, although NAD was also accepted. Only Zwf-2 was active in the presence of 3 mM ATP, and then only with NADP as a coenzyme, indicating an anabolic role for this isoenzyme. Disruption of zwf-1 resulted in increased alginate production when glycerol was used as the carbon source, possibly due to decreased flux through the Entner-Doudoroff pathway rendering more fructose-6-phosphate available for alginate biosynthesis. In alginate-producing cells grown on glucose, disruption of gcd increased both cell numbers and alginate production levels, while this mutation had no positive effect on growth in a non-alginate-producing strain. A possible explanation is that alginate synthesis might function as a sink for surplus hexose phosphates that could otherwise be detrimental to the cell.
INTRODUCTION
Knowing how carbon sources are channeled through different metabolic pathways to provide the energy and precursors needed to sustain growth is crucial for optimizing an organism for industrial bioproduction purposes. It is known that in pseudomonads, glucose is not channeled to the tricarboxylic acid (TCA) cycle via glycolysis, since these bacteria lack the phosphofructokinase gene. Instead, glucose is metabolized via the Entner-Doudoroff (ED) pathway (see Fig. 1) (1). Previous studies have also indicated strongly that fructose and glycerol are mainly converted to glucose-6-phosphate (G6P) via fructose-1,6-bisphosphate (FBP). G6P is then metabolized through the ED pathway (2).
Pseudomonas fluorescens is a diverse species that occupies many niches in nature (3). Like several other species of this genus, it has the ability to produce the linear polymer alginate. This industrially important, nonrepetitive polysaccharide, containing 1,4-linked α-l-guluronic acid and β-d-mannuronic acid, is currently commercially manufactured from brown algae. Bacterial alginate bioproduction is interesting, however, especially because bacteria may be engineered to produce alginates with properties that are well suited for medical applications. By using a nonpathogenic wild-type strain of P. fluorescens, we have been able to obtain mutant strains, for example, strain Pf201, with industrial-scale alginate production levels (4).
Alginate is synthesized from fructose-6-phosphate (F6P) and thus competes with the central carbon metabolism for FBP. This motivated us to investigate how carbon flow through the various early pathways of the central carbohydrate metabolism would influence cell growth and alginate yield. Figure 1 shows the different metabolic pathways for the degradation of glucose, fructose, and glycerol in P. fluorescens. We wanted to study the contributions of glucose dehydrogenase (Gcd) and glucose-6-phosphate dehydrogenase (G6PDH; also called Zwf) to these processes. Lessmann et al. (5) have described two different G6PDHs from P. fluorescens and have shown that the activity of one of these isoenzymes did not vary much with the carbon sources tested, while the other was induced by glucose (5). It is not clear, however, whether the two P. fluorescens Zwf enzymes have strictly defined roles, one being catabolic and the other anabolic, or if either one would be sufficient for growth. The P. fluorescens zwf-1 gene is the first gene in an operon that also contains pgl (encoding 6-phosphogluconolactonase) and eda (encoding 2-keto-3-deoxy-6-phosphogluconate [KDPG] aldolase), while zwf-2 is found downstream of the 6-phosphogluconate (6PGA) dehydrogenase gene (gnd) in another operon. These differing genomic contexts indicate that Zwf-1 functions primarily in the ED pathway, while Zwf-2 is more important in the pentose phosphate (PP) pathway (Fig. 1). Lessmann et al. (5) proposed that the Zwf enzyme linked to the ED pathway was the one that was inhibited by ATP, since this would enable metabolic control of the enzyme (6, 7). They also found that the enzymes differed with regard to cofactor preference (NAD+ or NADP+). A more thorough understanding of these enzymes and the extent to which their functions in the cell are overlapping is needed for correct modeling of the metabolism of the cell.
Uptake and utilization of glucose, fructose, and glycerol in P. fluorescens. Abbreviations: G6P, glucose-6-phosphate; 6PGA, 6-phosphogluconate; 2KGP, 2-keto-6-phosphogluconate; KDPG, 2-keto-3-deoxy-6-phosphogluconate; F6P, fructose-6-phosphate; FBP, fructose-1,6-bisphosphate; F1P, fructose-1-phosphate; GAP, glyceraldehyde-3-phosphate; Pyr, pyruvate; DPGA, 1,3-diphosphoglycerate; 3PGA, 3-phosphoglycerate; 2PGA, 2-phosphoglycerate; PEP, phosphoenolpyruvate; OAA, oxaloacetate; G3P, glycerol-3-phosphate; DHAP, dihydroxyacetone phosphate; Gad, gluconate dehydrogenase; Gcd, glucose dehydrogenase; Kgk, 2-ketogluconate kinase; Gnk, gluconokinase; Glk, glucokinase; Kgr, 2KGP reductase; Gnd, 6PGA dehydrogenase; GlpK, glycerol kinase; GlpD, G3P dehydrogenase; Zwf1/2, G6P dehydrogenases; Pgi, phosphoglucose isomerase; Pgl, 6-phosphogluconolactonase; Edd, 6PGA dehydratase; Eda, KDPG aldolase; Fpk, 1-phosphofructokinase; Fdp, fructose-1,6-bisphosphatase; Fda, fructose-1,6-bisphosphate aldolase.
In most studies on the carbon flow in Pseudomonas spp., glucose has been used as the carbon source. However, for industrial bioproduction purposes, cheaper carbon sources, such as glycerol, are more attractive, because they could make the cost of microbial alginates more competitive, leading to a broader potential range of applications. It is not common to modify the central carbohydrate metabolism for bioprocess development, but since alginate is made from F6P, we hypothesized that inactivation of key genes such as gcd, zwf-1, or zwf-2 might lead to stimulation of alginate biosynthesis. To make the study directly relevant for bioprocess development, we used strain Pf201 as a starting point. Our results show that although alginate synthesis in this strain is generally not very sensitive to the gene inactivations investigated, disruption of zwf-1 nevertheless has a very clear stimulatory effect on polymer synthesis when glycerol is used as the carbon source.
MATERIALS AND METHODS
Strains, media, and growth conditions.The bacterial strains and plasmids used in this study are listed in Table 1. P. fluorescens and Escherichia coli strains were routinely grown in L broth (10 g/liter tryptone, 5 g/liter yeast extract, 5 g/liter NaCl) in 250-ml shake flasks at 225 rpm at 30 and 37°C, respectively. Conjugations between E. coli and P. fluorescens strains were carried out on L agar (L broth containing 15 g/liter agar), and transconjugants were selected on Pseudomonas isolation agar (PIA; Difco) with appropriate antibiotics added. For growth experiments, alginate production, and G6PDH activity assays, 1% LB-grown precultures were inoculated in DEF3 minimal medium, consisting of 0.39 g/liter yeast extract, 0.65 g/liter KH2HPO4, 2.73 g/liter (NH4)2HPO4, 0.9 g/liter citric acid·H2O, 0.02 g/liter ferric citrate, 0.001 g/liter H3BO3, 0.005 g/liter MnCl2·4H2O, 0.0039 g/liter EDTA·2H2O, 0.0005 g/liter CuCl2·2H2O, 0.0008 g/liter Na2Mo4O4·2H2O, 0.0008 g/liter CoCl2·6H2O, 0.0027 g/liter Zn(CH3COO)2·2H2O, 1.56 g/liter NaCl, 0.57 g/liter MgSO4·7H2O, 5 g/liter morpholinepropanesulfonic acid (MOPS), and 20 g/liter of a carbon source (glucose, fructose, or glycerol). The pH in the medium was adjusted to 7.0 with NaOH. When appropriate, antibiotics were added at the following concentrations: triclosan (Tric), 25 μg/ml; apramycin (Am), 50 μg/ml; ampicillin (Ap), 200 μg/ml; kanamycin (Km), 50 μg/ml; tetracycline (Tc), 15 μg/ml. Proteases (0.15 ml/liter of Alcalase 2.4 L and Neutrase 0.5 L) were added to alginate production media in order to degrade alginate lyases. The Pm promoter was induced by the addition of 0.25 mM m-toluate. For liquid cultures, an inducer was added 8 h after inoculation. For the detection of LacZ activity, 60 μl of X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) solution (20 mg/ml in dimethyl sulfoxide [DMSO]) was added to each agar plate 30 min before the plating of bacteria.
Bacterial strains and plasmids used in this study
DNA standard techniques.All routine DNA manipulation methods, transformation, and cloning were performed as described previously (8). E. coli strains S17.1 and S17.1 λpir were used as hosts for standard cloning procedures and the replication of suicide plasmids. The sequences of all amplified DNA fragments were confirmed by DNA sequencing.
Construction of strains.To generate P. fluorescens Δzwf-1, Δzwf-2, and Δgcd mutants, homologous recombination procedures were performed using pMG48 derivatives (pMM7, pMM6, and pMM5, respectively) as described previously (9) (see Table 1 for details on cloning). All mutant strains constructed were verified by PCR and restriction digestion.
Genes were overexpressed by constructing transposon vectors where the genes in question were placed under the control of the inducible Pm-G5 promoter (4). These plasmids were then transferred to the relevant strains by conjugation, and transposon insertion mutants were selected using Km or Am. Several different transposon insertion mutants were evaluated in order to avoid choosing mutant strains in which the phenotype observed was caused by the transposon insertion site and not by induced gene expression.
Alginate quantification.P. fluorescens cell culture samples were harvested, diluted using 0.2 M NaCl to decrease viscosity when necessary, and centrifuged at 13,000 rpm for 10 min to remove cells. The alginate in the supernatants was deacetylated by a mild alkaline treatment as described previously (10). Alginate was quantified using M-specific lyase (AlxM; Pseudomonas sp.) and G-specific lyase (AlyA; Klebsiella aerogenes) as described previously (11). All alginate quantifications were performed in triplicate.
G6PDH activity assays.Cell extracts containing G6PDH were prepared from overnight cultures of P. fluorescens NCIMB 10525 Δzwf-2::TnMM12 and P. fluorescens NCIMB 10525 Δzwf-1::TnSM1 using B-PER II bacterial protein extraction reagent (Pierce). Cell extracts were stored at −80°C. Total protein concentrations in the samples were estimated using a colorimetric protein assay (Bio-Rad) based on the Bradford method (12). Bovine serum albumin was used to generate the standard curve.
G6PDH activity was determined by measuring NAD(P)+ reduction spectrophotometrically at 340 nm. The enzyme assay was performed according to the protocol described previously by Lynch et al. (13) but modified for measurements in 96-well UV plates. Ten microliters of the cell extract and 190 μl of the assay solution were mixed to start the reaction. In order to investigate the inhibitory effect of ATP on the enzymes, activity was assayed in the presence and absence of 3 mM ATP. Assay mixtures without G6P were used to measure nonspecific reduction of NAD(P)+. A340 was measured every 10 s for 60 min, and values were converted to micromolar concentrations of NAD(P)H, i.e., reduced NAD(P)+ (6). One unit of enzyme is defined as the amount of total soluble protein that catalyzes the reduction of 1 μmol of NAD(P)+ per min.
RESULTS
Properties of the two P. fluorescens G6PDH isoenzymes.G6PDH (encoded by zwf) is required in both the ED and PP pathways. Knowledge of the properties of the two Zwf isoenzymes is therefore important for understanding the carbohydrate metabolism of P. fluorescens. The zwf-1 and zwf-2 genes were first separately inactivated by in-frame deletions in the wild-type P. fluorescens strain NCIMB 10525, and then a Δzwf-1 Δzwf-2 double mutant was constructed. Cell extracts from the single mutant strains, especially NCIMB 10525 Δzwf-1, showed low levels of G6PDH activity. As expected, no G6PDH activity was detected in cell extracts from the double mutant. In order to analyze the properties of each enzyme separately in vitro, each functional gene was cloned into a transposon and was placed under the control of the inducible Pm-G5 promoter. The transposons were transferred to the respective deletion strains, generating strains NCIMB 10525 Δzwf-2::TnMM12 (expressing only zwf-1) and NCIMB 10525 Δzwf-1::TnSM1 (expressing only zwf-2). Extracts from these strains were then used to analyze the properties of each enzyme separately. The strains were grown in LB overnight, and a 1% inoculum was transferred to DEF3 minimal medium containing glucose as the carbon source. The cultures were grown for 8 h before Pm-G5 was induced by the addition of m-toluate, and cultivation was continued overnight. G6PDH activity was measured in cell extracts from these strains as described in Materials and Methods. Both NAD+ and NADP+ were used as coenzymes, in separate experiments.
The results showed that both enzymes display the highest activity with NADP+ as the coenzyme, but the difference was much smaller for Zwf-1 than for Zwf-2 (Fig. 2). Furthermore, addition of 3 mM ATP resulted in no detectable Zwf-1 activity with either coenzyme. Zwf-2 also lost its activity when assayed with NAD+ and ATP but retained 60% activity in the presence of ATP when NADP+ was used as the coenzyme. These data indicate that Zwf-2 is active even when the energy status of the cell is good, probably taking part in the pentose phosphate pathway operating in the anabolic direction. Lessmann et al. (5) reported that the most highly expressed isoenzyme, now annotated as Zwf-1, showed a slight preference for NAD+ (5), while our results indicate that both enzymes prefer NADP+ to NAD+ as a cofactor. The reasons for this discrepancy are not known but could, for instance, be due to strain-dependent differences.
Coenzyme preference and ATP-induced inhibition of the glucose-6-phosphate dehydrogenases Zwf-1 (shaded bars) and Zwf-2 (open bars). Enzyme activities are given as average values measured in cell extracts obtained from three independently grown cultures of strains expressing only one of the two G6PDH isoenzymes. Extracts were assayed using NAD+ or NADP+ as the coenzyme, and with or without the addition of ATP. (See Materials and Methods for details.) One unit of enzyme is defined as the amount of total soluble protein that catalyzes the reduction of 1 μmol NAD(P)+ per min.
The phenotypes of gcd, zwf-1, and zwf-2 mutants with respect to growth depend on the carbon source.In P. fluorescens, glucose is channeled into the central carbohydrate metabolism either via glucose dehydrogenase (Gcd) or via glucokinase (Glk) and G6PDH/Zwf (1). Both pathways converge in the common product KDPG (Fig. 1). It has been shown that this metabolite binds to and releases the repressor HexR from promoters controlling the expression of Zwf-1 and other key enzymes in the ED pathway (14). We wanted to investigate the effect of inactivating the three genes gcd, zwf-1, and zwf-2 by assessing the impact on growth when glucose, fructose, or glycerol was used as the carbon source. Since alginate production imposes a drain on F6P that can also affect growth (15), these mutants were generated from the non-alginate-producing wild-type strain NCIMB 10525. Strains carrying single and double mutations were constructed.
Growth curves for wild-type and mutant strains cultivated in minimal medium with glucose as the carbon source are shown in Fig. 3a. The three single gene deletion strains displayed different growth kinetics. The Δzwf-2 strain grew very similarly to the wild type, while the Δgcd strain grew slightly slower. The Δzwf-1 mutant displayed a prolonged lag-phase but nevertheless reached a much higher final optical density (OD) than all the other strains. The growth phenotypes of the Δzwf-1 Δzwf-2 and Δzwf-2 Δgcd double mutants indicated moderate additive negative effects of combining the two deletions. The most drastic negative effect on growth was observed for the strain where both gcd and zwf-1 were inactivated. As a control, TnHE418, where gcd is controlled by the inducible Pm-G5 promoter, was transferred to strain NCIMB 10525 Δzwf-1 Δgcd. As expected, the complemented strain grew similarly to strain NCIMB 10525 Δzwf-1 (results not shown). Our results demonstrate that the activity of Zwf-2 alone is not sufficient to sustain normal growth rates on glucose and suggest that most of the carbon flow from glucose is channeled through the reactions catalyzed by the gcd and zwf-1 gene products.
Growth of P. fluorescens NCIMB 10525 and its mutant derivatives on various carbon sources. The strains were cultivated in minimal medium using glucose (a), fructose (b), or glycerol (c) as the carbon source, and growth was measured as the absorbance at 660 nm. The following strains were investigated: P. fluorescens NCIMB 10525 (◆) and its Δzwf-1 (■), Δzwf-2 (▲), Δgcd (●), Δzwf-1 Δzwf-2 (□), Δzwf-1 Δgcd (△), and Δzwf-2 Δgcd (○) mutants. Each cultivation was repeated at least three times, each time resulting in growth profiles essentially the same as those displayed here.
In cultivations on fructose (Fig. 3b) the Δzwf-1 mutant displayed a strongly reduced growth rate and did not recover at late stages of growth, as it did on glucose. This difference may occur because glucose, but not fructose, can be metabolized by Gcd. As a consequence, inactivation of gcd has no additional negative growth effect on a Δzwf-1 mutant when grown on fructose. The Δgcd and Δzwf-2 single mutants, as well as the Δzwf-2 Δgcd double mutant, grew similarly to the wild type. In contrast, the Δzwf-1 Δzwf-2 double mutant hardly grew at all on fructose. This indicates that a carbon flow through the reactions catalyzed by the Zwf enzymes, and mostly by Zwf-1, is necessary for the cell under these conditions. This result was unexpected, since fructose enters the central carbon metabolism as fructose-1,6-bisphosphate in P. fluorescens, and the bacterium expresses a functional fructose-1,6-biphosphatase (Fig. 1).
As a control, transposons TnMM12 (expressing zwf-1) and TnSM1 (expressing zwf-2) were transferred separately to the Δzwf-1 Δzwf-2 double mutant. The phenotypes of the new strains were similar to those of the zwf-2 and zwf-1 single mutants, respectively, confirming that the inability of the double mutant to grow on fructose was caused by the absence of G6PDH enzymes (data not shown). These data also demonstrate that the in frame zwf-1 deletion introduced by using pMM10 does not have a strong (if any) polar effect on the downstream 6-phosphogluconolactonase gene (pgl). The genome contains only one copy of this gene, and Pgl activity is necessary to convert the G6PDH product to 6-phosphogluconate (Fig. 1).
When the strains were grown on glycerol, the only strain exhibiting a severely reduced growth rate was the Δzwf-1 Δzwf-2 double mutant (Fig. 3c), demonstrating that G6PDH activity is important even when carbon enters the cells as a 3-carbon compound, although it is less important than for fructose-grown cells.
Effects of zwf-2 and gcd inactivation on growth and alginate production.P. fluorescens strain Pf201 is a stable high-level alginate-producing mutant (8). We decided to use this strain to study how the channeling of carbon through the different metabolic pathways shown in Fig. 1 affects alginate production. Pf201 mutants with in-frame deletions in zwf-2 or gcd were easily obtained and formed mucoid colonies on PIA, showing that the mutants had retained the ability to produce alginate. The same strategy was applied for zwf-1, but unsuccessfully, so for this gene, a more complicated procedure was needed (see below). The Δzwf-2 and Δgcd strains were cultivated in defined medium containing glucose, fructose, or glycerol in order to assess growth and alginate production. Since alginate production sequesters carbon and imposes a drain on some of the F6P that could otherwise have been used for generating biomass, the biomass reached was, as expected, lower than that for the isogenic non-alginate-producing strains. Neither of the mutations affected alginate production significantly when the strains were grown on glycerol or fructose (results not shown).
On glucose, the Δgcd mutant produced significantly more alginate than Pf201, while inactivation of zwf-2 resulted in slightly increased alginate production (Fig. 4). The apparent cell density of Pf201 Δgcd was also higher than that for the other strains (Fig. 4a), but this was not reflected in a higher number of CFU after 24 h (Fig. 4c), indicating that Pf201 Δgcd cultures contain something aside from cell mass that contributes to their high OD at 660 nm (OD660). In order to confirm the phenotype of Pf201 Δgcd, the strain was complemented by transposon TnHE418, carrying gcd under the control of the Pm-G5 promoter. As expected, the complemented strain showed lower OD values and produced less alginate than Pf201 Δgcd, proving that lack of Gcd was indeed causing the phenotype observed.
Effects of zwf-2 and gcd inactivation on growth and alginate production in cells cultivated with glucose as the carbon source. (a) Growth, expressed as OD660. Data are averages for three biological replicates, and error bars indicate standard deviations. (b) Alginate production, expressed as grams per liter of culture. Data are averages for three biological replicates. (c) Growth, expressed as CFU per milliliter of culture. Data are averages for at least two biological replicates. Strains: Pf201 (◆, open bars), Pf201 Δzwf-2 (▲, dark shaded bars), Pf201 Δgcd (●, striped bars), Pf201 Δgcd::TnHE418 (△, light shaded bars), and the non-alginate-producing wild-type (wt) strain NCIMB 10525 (stippled bars).
A combination of alginate and gluconate production induces cell death.While it is not unusual for the number of viable cells to decrease somewhat in late-stationary phase, the CFU counts for NCIMB 10525, Pf201, and derivatives of Pf201 indicated that when cultivated on minimal medium with glucose as the carbon source, Pf201 exhibited clear signs of extreme stress (Fig. 4c), and in two of three cultures, <1,000 viable cells were present after 48 h. The third culture also displayed a decreased CFU count. No comparable decreases in CFU were detected in cultures cultivated on carbon sources other than glucose (not shown). The CFU count of strain Pf201 Δgcd increased slightly from 24 to 48 h, while the strain complemented with TnHE418 behaved similarly to Pf201. This strongly indicates that simultaneous production of gluconate and alginate resulted in a stress situation that the cells had difficulty coping with, while they were able to thrive when challenged with the production of only one of the acids.
Inactivation of zwf-1 increases alginate production when glycerol is used as the carbon source.As mentioned above, the frequency of recombinant mutants with an inactivated zwf-1 gene was found to be much lower in Pf201 than in the wild-type non-alginate-producing strain. Furthermore, all Pf201 Δzwf-1 mutants obtained formed nonmucoid colonies on PIA, indicating a deficiency in alginate biosynthesis. Heterologous recombination was used to replace the Δzwf-1 gene in one of these mutants with wild-type zwf-1, using plasmid pMM27. Even though the genotype of zwf-1 was confirmed to be restored to wild type by PCR, alginate production was not recovered. We therefore considered it likely that the original mutant strain had acquired a secondary mutation that suppresses alginate synthesis. This would indicate that a lack of Zwf-1 during alginate synthesis imposes a selective growth disadvantage that makes it difficult to obtain zwf-1 mutants from an alginate-producing strain. A similar problem had been encountered earlier in studying the effect of AlgL on alginate production and was then solved by first constructing a Pf201 derivative with m-toluate-inducible algC (required for alginate biosynthesis) inserted on a transposon in a ΔalgC background. An in-frame deletion in algL could then be constructed under noninducing conditions (8). Correspondingly, we constructed strain Pf201 ΔalgC Δzwf-1::TnKB60 (see Table 1 for details). We further used plasmid pMM27 to repair the deletion in zwf-1, and the resulting strain (Pf201 ΔalgC::TnKB60 MM27) was used as an additional control strain. Both these strains were mucoid on PIA in the presence of m-toluate, demonstrating that a zwf-1 deletion can be made with no loss of alginate synthesis.
The strains were cultivated using different carbon sources, and growth and alginate production were monitored (Fig. 5). On glucose, both the growth rate (Fig. 5a) and the alginate production (Fig. 5b) of the strain lacking zwf-1 (Pf201 ΔalgC Δzwf-1::TnKB60) were similar to those observed for the control strains (Pf201 ΔalgC::TnKB60 [not shown], Pf201, and Pf201 ΔalgC Δzwf-1::TnKB60 MM27).
Effects of zwf-1 disruption on growth and alginate production. P. fluorescens strains Pf201 (◆), Pf201 ΔalgC Δzwf-1::TnKB60 (■), and Pf201 ΔalgC::TnKB60 MM27 (▲) were grown in minimal medium containing glucose (a and b), fructose (c and d), or glycerol (e and f). Growth and alginate data are shown as averages for at least two biological replicates. Error bars indicate standard deviations.
As expected from the experiments using the wild-type non-alginate-producing strain, the growth rate on fructose was lower for the Pf201 Δzwf-1 strain than for Pf201 or the repaired strain containing wild-type zwf-1 (Fig. 5c). The amounts of alginate produced per unit of culture volume were similar for all three strains (Fig. 5d), but per OD unit, production was somewhat higher for the Δzwf-1 strain. Note that the amount of alginate produced with fructose as the carbon source is much higher than that with glucose, as is also the case for the original Pf201 strain.
For glycerol-grown cultures, the initial growth rates were similar for all three strains, but the final cell mass was lower for the Δzwf-1 strain (Fig. 5e). Interestingly, this strain still produced significantly more alginate than the two control strains (Fig. 5f). The polymer production per unit of cell mass is therefore even higher. This observation is potentially of biotechnological significance, as discussed below.
DISCUSSION
The results of this study show that among the genes evaluated, inactivation of the G6PDH gene zwf-1 led to the most pronounced phenotypes with the carbon sources tested. Our in vitro data show that Zwf-2 probably has an anabolic role in the cell, tolerating a higher level of ATP than Zwf-1. Still, zwf-2 is dispensable in vivo in the presence of a functional zwf-1 gene. Furthermore, the data show that for the wild-type strain, G6PDH activity is required for growth on fructose, and also for a normal growth rate when glycerol is used as the carbon source. Results similar to ours have been obtained with Pseudomonas aeruginosa, where a zwf mutant grew very poorly on fructose (16). P. fluorescens and Pseudomonas putida seem to have similar fructose metabolisms, and it has been shown that when P. putida grows on fructose, 52% of the sugar is channeled through the ED pathway, 34% through glycolysis, and 14% through the PP route (1). Thus, even if the flux through glycolysis is too low to allow for full growth, it is puzzling that hardly any growth was observed (Fig. 3b). The remaining part of the PP pathway should enable the synthesis of metabolites needed for growth (e.g., purines, pyrimidines, and histidine).
NADPH is one important metabolite resulting from G6PDH activity and is known to be important for surviving oxidative stress in P. fluorescens (17). Our in vitro data show that both Zwf-1 and Zwf-2 are able to reduce NADP+. When the bacteria are cultivated on glucose, some NADPH is produced in the reaction catalyzed by 6-phosphogluconate dehydrogenase, while no NADPH is produced by the reactions required for the uptake and utilization of glycerol and fructose to produce glyceraldehyde-3-phosphate (GAP) (Fig. 1). This might explain some of the growth deficiencies of the zwf-1 zwf-2 double mutant when it is cultivated on fructose, but not the differences observed between growth on fructose and growth on glycerol.
As shown in Fig. 1, FBP is at the branching point of the two pathways for fructose utilization. Based on the P. putida data (1), one could hypothesize that 66% of the FBP is converted to F6P, which might exceed the capacity of the PP pathway running in the reverse direction and possibly lead to an accumulation of F6P and G6P. Accumulation of sugar phosphates is known to hamper growth in E. coli (18), and this might be the case for P. fluorescens as well. In E. coli, synthesis of the glucose transporter is negatively regulated when such stress is experienced. However, so far, the only known regulation of the fructose transporter system in Pseudomonas is that gene expression is turned on in the presence of F1P (19). Alginate production can be assumed to drain away surplus F6P. Thus, if an increased concentration of hexose phosphates is contributing to the slower growth, the growth defects of strain Pf201 ΔalgC Δzwf-1::TnKB60 grown on fructose should be less severe than those for NCIMB 10525 Δzwf-1, as was shown to be the case (compare Fig. 3b and 5c).
The periplasmic glucose dehydrogenase Gcd converts glucose to gluconate, resulting in acidification of the growth medium (20). Alginate production would be expected to decrease the pH further. Our results indicate that this affects both growth and alginate production negatively in the alginate-producing mutant Pf201 when grown on glucose, since the Δgcd mutant reached higher final cell densities, avoided stress-induced cell death, and produced more alginate than strain Pf201. The increased alginate production observed for the gcd strain at 48 h may thus be caused partly by a higher number of living cells. However, this does not explain the difference in alginate production at 24 h, when the CFU counts of the two strains are similar (Fig. 4). No similar positive growth effect was observed for NCIMB 10525 Δgcd, which actually grew more slowly than the wild-type strain (Fig. 3a). In a non-alginate-producing strain, gcd inactivation could result in increased levels of G6P and F6P, which may, as described above, lead to the lower growth rate observed. Inactivation of gcd would necessitate that all glucose be converted by the glucokinase system, producing G6P. G6P may then be directly isomerized to F6P, which could partly explain the increased alginate production in the glucose-grown Pf201 Δgcd mutant (Fig. 4b).
In a recent metabolome study of P. fluorescens, it was found that for cells grown on glycerol, an alginate-producing mutant had higher flux not only from F6P to alginate but also from F6P to G6P and through the ED pathway than the wild type (21). The latter data could explain why the zwf-1 mutation was beneficial for alginate production in cells using glycerol as the carbon source (Fig. 5f), since this mutation would decrease the flow of carbon through the competing pathway.
Our results indicate that the rate of alginate biosynthesis depends on the amount of F6P available. Excessive amounts of hexose phosphates seem to be detrimental to the cell and appear to be alleviated by channeling the surplus hexose phosphates to alginate production. These results should be relevant for the production of many different polysaccharides and other secondary metabolites by bacterial cells. Strain Pf201 has been tested under conditions (in bioreactors) more relevant for commercial production processes and was found to be a high-level, stable producer of alginate (4). The additional effects of the gcd and zwf-1 deletions reported here are therefore highly relevant for future development of processes for microbial alginate production.
ACKNOWLEDGMENTS
Susan Maleki and Mali Mærk were supported by grants from FUGE Midt-Norge and the Norwegian Research Council, respectively.
We thank Evy Renate Mellemsæther for constructing plasmid pEM1, Yi-Qian Sun for constructing plasmid pYQ1, and Karianne Bakkevig for constructing plasmid pKB64. We are grateful to Sergey B. Zotchev for helpful comments and discussions about this work.
FOOTNOTES
- Received 4 November 2014.
- Accepted 1 March 2015.
- Accepted manuscript posted online 6 March 2015.
- Copyright © 2015, American Society for Microbiology. All Rights Reserved.
