INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Hemoglobins are a specific group of oxygen-binding proteins that can be found in mammals, plants, and microorganisms (15). The homodimeric hemoglobin of Vitreoscilla (VHb) is the best characterized bacterial hemoglobin. The expression of VHb is up-regulated by oxygen limitation (hypoxia) in Vitreoscilla (50), but its physiological functions have not yet been entirely elucidated. Nonetheless, heterologous expression of VHb has been used to alleviate physiologically unfavorable effects of oxygen limitation and to improve growth properties and productivity of various microorganisms, plants, and mammalian cells that yield industrially important metabolites (7, 18, 21, 28-30, 32, 35).

Previous research on the effects arising from heterologous VHb expression has mainly focused on the operation of the respiratory chain. Expression of VHb in Escherichia coli cells lacking either cytochrome o (aerobic terminal oxidase) or cytochrome d complexes (microaerobic terminal oxidase) revealed a 5-fold increase in cytochrome o (in a cyd mutant, cyo⁺ strain) and a 1.5-fold increase in cytochrome d (in cyd⁺, cyo mutant strain) complexes relative to wild-type cells (48). These results have led to the hypothesis that VHb is able to increase the effective intracellular oxygen concentration with concomitant increase of the amount of cytochrome o complexes (20): the proton translocation activity of cytochrome o complexes is characterized by a higher H/O ratio than cytochrome d complexes and is able to generate a larger proton gradient across the cell membrane (25, 36, 38). VHb-expressing cells are indeed able to generate a larger proton flux per reduced oxygen molecule than control cells and have a 30% higher ATPase activity and a 65% higher ATP turnover rate (8, 20).

Furthermore, it has also been reported that the steady-state NAD(P)H level is 1.8-fold lower in a VHb-expressing strain than in control cells grown under nearly anoxic conditions (47), i.e., cells are in a more reduced state under oxygen-limited conditions (16). Anoxia is known to reduce electron flow through the respiratory chain so that NAD(P)H is consumed more slowly. Therefore, one may hypothesize that, under low oxygen tension, the presence of VHb in E. coli increases the electron flux through the respiratory chain. Moreover, Tsai et al. (47) postulated that this shift in the NAD⁺/NADH concentration ratio might have implications on key steps of the central carbon metabolism in VHb-expressing E. coli. Therefore, metabolic flux analysis (MFA) was performed and a metabolic model suggested that VHb-positive cells direct a higher fraction of glucose through the pentose phosphate pathway (PPP) and channel less acetyl coenzyme A (AcCoA) through the tricarboxylic acid cycle (TCA) than wild-type E. coli (49). Direction of additional glucose through PPP generates an excess amount of NADPH and results in a transhydrogenation reaction, which creates an H⁺ flux from NADPH to NAD⁺. VHb-expressing cells also displayed strongly reduced formate and D-lactate excretion levels relative to those of controls (49).

Recently, we constructed a novel set of expression systems for native and engineered hemoglobin proteins (12). The native proteins used were VHb and the flavohemoprotein (FHP) of Alcaligenes eutrophus (9). FHP is a member of the FNR-like proteins (23) and contains an N-terminal hemoglobin (FHPg) and a C-terminal redox-active (Red) domain. The Red domain was fused with VHb to generate the VHb-Red fusion protein. In addition, the hemoglobin domain (FHPg), which shares high sequence homology with VHb, was truncated from the reductase domain and was functionally expressed in E. coli. The expression of the hemoglobins showed various interesting features, such as significantly improved growth rates and changes in activity of metabolic pathways under hypoxia (12).

The ratios of intracellular metabolic fluxes in the central carbon metabolism of E. coli MG1655 cells grown under oxygen-limited conditions have previously been determined using biosynthetically directed fractional ¹³C-labeling of amino acids and two-dimensional (2D) [¹³C, ¹H]-correlation nuclear magnetic resonance (NMR) spectroscopy (11). This investigation, which serves as a reference for the presently studied hemoglobin-expressing cells, showed that, under microaerobic conditions, the topology of the active pathways is characteristic of anaerobic metabolism, as was primarily evidenced by the activity of the pyruvate-formate lyase and the suppression of 2-oxoglutarate dehydrogenase activity. This result is at variance with previously assumed aerobic configurations and illustrates the importance of experimental characterization of the topology of active pathways.

Further studies of the central carbon metabolism of hemoglobin-expressing cells promise to lead to new insights into the cellular physiology, which in turn may support the design of improved strains for biotechnology. Therefore, we applied the recently developed approach of biosynthetically directed fractional labeling of proteinogenic amino acids (43-46) for the assessment of intracellular carbon flux ratios by using a variety of hemoglobin-expressing E. coli strains.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains and plasmids. The plasmids pPPC1 (which carries the Vitreoscilla vhb gene), pAX1 (which carries the A. eutrophus hemoglobin gene domain [fhpg]), pAX5 (which contains the native flavohemoglobin gene [fhp] of A. eutrophus), and pAX4 (which carries the gene fusion between the vhb gene and the FHP reductase domain coding sequence of A. eutrophus [vhb-red]) were used to study physiological consequences of globin expression in E. coli MG1655 (, F) (Cold Spring Harbor Laboratory). The construction of the plasmids has been described previously (12, 22).

Microaerobic bioreactor cultivations. Cultivations of E. coli MG1655 were performed under microaerobic conditions (concentration of O₂, 0.02 mmol/liter) in a SixFors bioreactor unit (Infors, Bottmingen, Switzerland) in a fed-batch mode using a minimal medium containing glucose as the sole carbon source (4 g of glucose/liter) (22). Based on previous experiments, which showed lower glucose consumption for E. coli cells expressing the A. eutrophus hemoglobin genes (fhpg and fhp), the glucose concentration of the medium was reduced to 2 g/liter for these strains. This modification prevents both glucose accumulation and dilution of the ¹³C-labeled substrate with unlabeled glucose. The bioreactor parameters were as follows: working volume, 300 ml; stirrer speed, 300 rpm; temperature, 37°C; aeration rate, 120 ml of air/min; and pH of 7.0 ± 0.2, adjusted with addition of either 2 M NaOH or 2 M H₃PO₄.

Biological activity of expressed hemoglobin proteins. The biological activity of the expressed hemoglobin proteins was confirmed by a CO-binding activity assay (17).

Quantitative analysis of by-products and intracellular metabolites. By-product concentrations were measured enzymatically with a Beckman SYNCHRON CX5CE autoanalyzer at 1-h intervals. Ethanol and residual glucose concentrations were determined using Beckman alcohol (no. 445900) and glucose (no. 442640) kits, respectively. The acetate kit (no. 148261) was purchased from Roche Diagnostics. Quantifications of D-lactate, succinate, and formate concentrations were performed using enzymatic assays adapted for the Beckman autoanalyzer system (5, 49).

Sample preparation and 2D [¹³C, ¹H]-COSY NMR measurements. Fractionally labeled cells were hydrolyzed in 6 N HCl at 110°C, and the NMR spectra of the amino acid mixtures were recorded at 40°C and at a proton resonance frequency of 400 MHz, using a Varian Inova 400 spectrometer. Two proton-detected 2D [¹³C, ¹H] heteronuclear single quantum coherence spectra were recorded for each sample (6). Pulsed-field gradients were used for coherence pathway rejection (4, 52), and the decoupling scheme WALTZ (42) was applied during detection. The spectra containing the aliphatic resonances were recorded in 12.5 h (1,400 by 256 complex points; t_1max = 412 ms; t_2max = 128 ms; 1.5-s relaxation delay between scans) with the carrier frequency set to 42.5 ppm and a spectral width of 33.8 ppm. The spectra for the aromatic resonances were recorded in 7.5 h (950 by 512 complex points; t_1max = 412 ms; t_2max = 109 ms; 1.5-s relaxation delay between scans) with the ¹³C-carrier frequency set to 123.8 ppm and a spectral width of 22.8 ppm. Spectra were transformed using the program PROSA (14). The digital resolution after zero-filling was 0.9 Hz/point along ₁ and 2.0 Hz/point along ₂ for the spectrum with the aliphatic resonances and was 0.6 Hz/point along ₁ and 4.6 Hz/point along ₂, respectively, for the spectra containing the aromatic resonances. The overall degree of the ¹³C labeling of the amino acids, P₁, was measured from 1D ¹H-NMR spectra (t_1max = 1.024 s; 8-s interscan delay), as shown in Fig. 1. The resulting value of P₁ was in good agreement with the value calculated from the fraction of [¹³C₆]glucose in the minimal medium and the fraction of ¹³C-labeled biomass assessed via first-order wash-out kinetics (40). The value for P₁ was independently confirmed from analysis of the scalar coupling fine structure of Leu- (43). Forty-two individual ¹³C-¹³C coupling fine structures were analyzed in the correlation spectra, and the relative abundances of intact carbon fragments present in eight principal intermediates of the central carbon metabolism were derived according to Szyperski (43), using the program FCAL (46). Flux ratios through several key pathways in the cellular central metabolism were then calculated from the abundances of fragments as described previously (40, 43).

View larger version (20K): [in this window] [in a new window]   FIG. 1.   Determination of the degree of overall 13C labeling of the biomass from a 1D 1H-NMR spectrum. (a) Region of the 1D 1H-NMR spectrum recorded for the hydrolyzed biomass comprising the aromatic resonances. Several well-resolved peaks allow observation of the 13C satellites. (b) Expansion showing a well-resolved resonance (unassigned). The satellite doublet arising from protons bound to 13C and the corresponding 12C-H peak are indicated by arrows. The ratio of the sum of the integrals of the two satellites to the total integral of all three peaks yields the overall labeling degree of the biomass (4.5% in this case).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Physiological characterization of hemoglobin-expressing strains. The E. coli strains MG1655:pPPC1, MG1655:pAX1, MG1655:pAX4, and MG1655:pAX5 displayed CO-binding activities that are indicative of the expression of biochemically active hemoglobin proteins. The absorption spectra of the various hemoglobins corresponded to the values reported previously (data not shown) (12).

1

1

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Growth trajectories of the four hemoglobin-expressing E. coli MG1655 strains used for this study, when grown in a minimal medium under hypoxic conditions. The expression of the proteins was induced at an A600 of 1, and the fractional 13C-labeling was started at an A600 of 4.5. The cells were harvested for NMR analysis at the end of the cultivations, with an A600 of 7. , MG1655:pPPC1, which expresses the VHb protein; , MG1655:pAX5, which expresses the FHP protein; , MG1655:pAX1, which expresses the FHPg protein; ×, MG1655:pAX4, which expresses the VHb-Red protein.

Validation of physiological data. We performed a carbon balance analysis for all hemoglobin-expressing strains, taking into account the glucose uptake, production of biomass, CO₂ evolution, and excretion of by-products. The carbon recovery calculated as the ratio of C output to C uptake was larger than 93% in all cases, i.e., the carbon balance could be closed to within a few percentage points (Table 3).

Analysis of the 2D [¹³C, ¹H]-correlation spectroscopy (COSY) spectra and derivation of metabolic flux ratios. The ¹³C-labeling experiments revealed major differences in the central carbon metabolism when comparing VHb- and VHb-Red-expressing cells to FHPg- and FHP-expressing cells. VHb- and VHb-Red-positive cells yielded labeling patterns similar to those previously observed for anaerobically grown wild-type E. coli cells (43) and microaerobic MG1655 cells (11), whereas for FHPg- and FHP-expressing cells, a more aerobic fluxome was found (Fig. 3 and Table 4).

View larger version (66K): [in this window] [in a new window]   FIG. 3.   Flux ratios and interaction of glycolysis and the TCA cycle in the central carbon metabolism of hemoglobin-expressing microaerobic E. coli MG1655 cells. (A) Flux ratios of wild-type E. coli cells (11), VHb-expressing cells, and VHb-Red-expressing cells (from top to bottom of the boxes). (B) Flux ratios of FHPg- and FHP-expressing E. coli cells (from top to bottom of the boxes). Glycolysis (left) and the TCA cycle (right) are highlighted. Enzymatically measured by-products and glucose are given in bold italics. Carbon fragment patterns of the boxed metabolites were directly determined by [13C, 1H]-COSY of proteinogenic amino acids. The fractions of molecules given in boxes are synthesized via the reactions pointing at them, and numbers in ellipses indicate the amount of reversible interconversion of the molecules. Enzyme names are given in italics. Active pathways are shown with solid arrows. Dashed arrows indicate that the represented enzymatic conversion is inactive. Abbreviations: CIT, citrate; MAL, malate; OAA, oxaloacetate; OGA, oxoglutarate; PEP, phosphoenolpyruvate; PYR, pyruvate; ME, malic enzyme; PFL, pyruvate-formate lyase; PEP-CARB, phosphoenolpyruvate carboxylase; CIT-SYN, citrate synthase; GOX shunt, glyoxylate shunt; OGA-DH, oxoglutarate dehydrogenase.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The expression of VHb, VHb-Red, FHPg, and FHP was previously shown to improve hypoxic growth properties of E. coli MG1655 cells in bioreactor cultivations (12). Although the expression of these hemoglobin proteins is able to positively modulate the microaerobic growth of E. coli, little is known about the underlying changes of metabolic configurations induced by the expression of these heterologous globins. Therefore, we have studied the microaerobic carbon metabolism of E. coli MG1655 cells expressing VHb, VHb-Red, FHPg, or FHP by using fractional ¹³C-labeling and 2D NMR spectroscopy (43-46). In addition, we analyzed the by-product excretion and intracellular metabolite content in order to unravel more accurately the changes of carbon metabolism. Our results revealed major differences in the metabolic states between VHb- and VHb-Red-producing E. coli strains on the one hand and FHPg- and FHP-expressing E. coli strains on the other hand.

The genetic approach to alleviate adverse effects of oxygen limitation by VHb-expression has been extensively studied in E. coli. Previously, a mathematical model conjectured that VHb is able to increase the intracellular oxygen tension and to activate both terminal oxidases in E. coli, albeit to different extents (20, 49). Thus, the model indirectly suggested that VHb expression is able to shift the global microaerobic regulatory network to a more aerobic regime. This hypothesis was supported by results of Tsai et al. (48) revealing a 5-fold increase in cytochrome o and a 1.5-fold increase in cytochrome d content in VHb-expressing cells relative to those of controls. In addition, the specific activity of cytochrome o was enhanced by 50% relative to those of VHb-negative controls. The cytochrome o complex is able to extrude two protons per oxygen molecule reduced, whereas the cytochrome d complex does not function as a proton pump (38). VHb-expressing cells do actually display a higher transmembrane proton gradient and a higher yield of membrane-translocated protons per oxygen molecule reduced when compared to controls (20, 48). These findings are in qualitative agreement with previous ³¹P-NMR results showing that VHb-expressing cells have a 65% higher ATP turnover rate than controls. The reentry of the proton flow into the cell via ATPase yields up to 30% higher ATPase activity and ATP production than controls (8, 20).

Keeping the above-mentioned findings in mind and taking into account the present data of the VHb-expressing E. coli which showed an interrupted TCA cycle, we tried to integrate all the information into a model for microaerobic growth of VHb-expressing E. coli. Interruption of the TCA cycle is normally regarded as a characteristic of anaerobic growth. However, it has also been shown for aerobically growing E. coli cells that the TCA cycle can operate in a branched fashion once the cellular energy demands are satisfied with energy derived from glycolysis (1). A study by Tsai et al. (49) showed that the carbon fluxes through the TCA cycle are concomitantly decreased with increasing VHb concentration. As a consequence, the expression of increasing amounts of VHb reduces the carbon flux entering the TCA cycle via AcCoA but slightly increases the carbon flux through the anaplerotic reaction connecting glycolysis to the TCA cycle. Therefore, we may hypothesize that the TCA cycle is interrupted above a threshold VHb concentration, as shown in this study for VHb- and VHb-Red-expressing E. coli cells. The absence of cyclic TCA activity leads to less efficient oxidation of carbon sources and concomitantly to a decreased generation of reducing equivalents and ATP. Higher glycolytic activity in VHb-expressing cells may then satisfy the ATP and NAD(P)H demands for growth and maintenance. To support this hypothesis, we included some previously published NMR data on hemoglobin-negative control cells in Table 4 (11). Comparison of the data on VHb- and VHb-Red-expressing cells with the VHb-negative control reveals similar flux ratios. However, additional data on wild-type E. coli MG1655 cells grown with the same experimental conditions reveal a higher production of formate and an increased consumption of glucose but a lower growth rate relative to the hemoglobin-expressing strain (Table 1). These findings may confirm our hypothesis that VHb and VHb-Red expression may enable the cells to operate their metabolism more efficiently. Therefore, the lack of functional TCA cycle activity in these strains might not result from anaerobiosis but simply from the fact the cellular energy demands can be satisfied by a more efficient ATP generation system.

Unfortunately, the physiological role of FHP either in its native host, A. eutrophus, or in other heterologous expression systems is poorly characterized. FHP expression, like that of VHb, is induced under microaerobic conditions and may interact with gas metabolism during denitrification in A. eutrophus (9, 37). However, fhp mutants did not show any difference relative to wild-type cells when grown anaerobically with nitrite as a sole electron acceptor, but did not transiently accumulate nitrous oxide. Preliminary attempts to demonstrate nitric oxide reduction with purified FHP and NADH as an electron donor were unsuccessful (9), and the role of flavohemoglobins in the detoxification of nitric oxide is still under discussion (24, 31). Since our previous results have shown that FHP can help to support growth of E. coli under microaerobic conditions (12), it would nonetheless appear that FHP is able to increase the free intracellular oxygen concentration in E. coli. This hypothesis is supported by our data showing that FHP and FHPg expression leads to a higher oxygen uptake rate than in VHb-expressing cells and inactivation of the highly O₂-sensitive enzyme Pfl, which was, rather unexpectedly, found to be active in VHb- and VHb-Red-expressing cells.

Recently, Gardner et al. (13) purified FHP, determined the kinetic properties for oxygen binding and release (k_on = 50 µM¹ s¹, k_off = 0.2 s¹), and found that the dissociation constant differs strongly from the values reported for VHb (k_on = 78 µM¹ s¹, k_off = 5,600 s¹) (34, 51). The association equilibrium constants (K) determined for FHP and VHb attribute high oxygen affinity to FHP (K = 250 µM¹) and low oxygen affinity to VHb (K = 0.014 µM¹). Due to the very high values for k_off, VHb-expressing cells have previously been proposed to scavenge oxygen and possibly provide O₂ rapidly to respiratory complexes in E. coli (20, 47-49). The mechanism of biochemical action of FHP globin on the cellular metabolism of E. coli is still unknown.

In this study, we have shown that inverse metabolic engineering (3) can be used to improve metabolic efficiency of microaerobic cells by expression of heterologous globin genes. Through introducing FHP and FHPg into E. coli cells, we were able to reduce glucose consumption by approximately 50% relative to that of VHb- or VHb-Red-expressing cells without affecting cell growth. This has direct implications for economical aspects of potential biotechnological applications, promising lowered production costs.