Development of a Continuous Bioconversion System Using a Thermophilic Whole-Cell Biocatalyst

Abstract

The heat treatment of recombinant mesophilic cells having heterologous thermophilic enzymes results in the denaturation of indigenous mesophilic enzymes and the elimination of undesired side reactions; therefore, highly selective whole-cell catalysts comparable to purified enzymes can be readily prepared. However, the thermolysis of host cells leads to the heat-induced leakage of thermophilic enzymes, which are produced as soluble proteins, limiting the exploitation of their excellent stability in repeated and continuous reactions. In this study, Escherichia coli cells having the thermophilic fumarase from Thermus thermophilus (TtFTA) were treated with glutaraldehyde to prevent the heat-induced leakage of the enzyme, and the resulting cells were used as a whole-cell catalyst in repeated and continuous reactions. Interestingly, although electron microscopic observations revealed that the cellular structure of glutaraldehyde-treated E. coli was not apparently changed by the heat treatment, the membrane permeability of the heated cells to relatively small molecules (up to at least 3 kDa) was significantly improved. By applying the glutaraldehyde-treated E. coli having TtFTA to a continuous reactor equipped with a cell-separation membrane filter, the enzymatic hydration of fumarate to malate could be operated for more than 600 min with a molar conversion yield of 60% or higher.

INTRODUCTION

Thermophilic enzymes are promising tools for biotransformation owing to their high operational stability, cosolvent compatibility, and low risk of contamination (1–3). The direct use of recombinant mesophiles (e.g., Escherichia coli) having heterologous thermophilic enzymes at high temperatures results in the denaturation of indigenous enzymes and the elimination of undesired side reactions; therefore, highly selective whole-cell catalysts comparable to purified enzymes can be readily prepared. Honda et al. have demonstrated that the rational combination of these thermophilic whole-cell catalysts enables the construction of in vitro artificial biosynthetic pathways for the production of value-added chemicals (4). Recently, Ye et al. have successfully constructed a chimeric Embden-Meyerhof pathway with the balanced consumption and regeneration of ATP and ADP, using nine recombinant E. coli strains, each of which overproduces a thermophilic glycolytic enzyme (5).

The membrane structure of E. coli cells is partially or entirely disrupted at high temperatures, and thus thermophilic enzymes, which are produced as soluble proteins, leak out of the cells (6–9). Although the heat-induced leakage of thermophilic enzymes results in better accessibility between the enzymes and substrates, it limits the applicability of thermophilic whole-cell catalysts to continuous and repeated-batch reaction systems. This limitation prevents us from exploiting the most advantageous feature of thermophilic biocatalysts, namely, their excellent stability. To overcome this limitation, one potential strategy is the integration of thermophilic enzymes to the membrane structure of cells. In our previous work, we found that the heat-induced leakage of a thermophilic glycerol kinase from recombinant E. coli cells could be prevented by fusing the enzyme to an E. coli membrane-intrinsic protein, YedZ (8). However, the specific enzyme activity of the recombinant E. coli having the YedZ-fused enzyme decreased to 6% of that of the recombinant with the nonfusion enzyme. A tight integration of the glycerol kinase to the E. coli membrane structure might have prohibited the conformational change of the enzyme, resulting in a decreased specific activity. Thus, the screening for a suitable membrane-anchoring protein would be essential to mitigate the loss of the specific activity.

An alternative approach to preventing the heat-induced leakage is the use of protein cross-linking reagents for the consolidation of the cell membrane as well as for the linkage of enzymes to the membrane structure. In this approach, unlike in the integration via membrane-anchoring proteins, cross-linkage level can be readily controlled by changing the conditions for the cross-linking reaction, and thus the best compromise between the prevention of the heat-induced leakage and the maintenance of the specific enzyme activity can be achieved. Glutaraldehyde (GA) and related dialdehydes are some of the most effective protein cross-linking reagents and have been widely used for biocatalyst immobilization (10–13). GA is mainly used to immobilize enzymes to carriers such as activated charcoal, anion-exchanging resin, and glass beads. Generally, for the cross-linkage of enzymes to these carriers, the enzyme has to be isolated from cells, purified to a certain level, attached to carriers in a suitable way, and then cross-linked with GA.

In this study, E. coli cells having a thermophilic fumarase were treated with GA. GA-treated cells were heated at 70°C to inactivate the intrinsic enzymes, and then directly used for the conversion of fumarate to malate. Through this simple procedure, many steps required in conventional procedures for the preparation of immobilized enzymes, such as protein extraction, enzyme purification, and the preparation of immobilizing carriers, could be entirely skipped, and a highly stable and selective immobilized enzyme, of which heat-killed E. coli cells served as carriers, could be prepared.

MATERIALS AND METHODS

Bacterial strain and culture conditions.

The expression vector for the Thermus thermophilus fumarase (TtFTA) was obtained from the RIKEN Thermus thermophilus HB8 expression plasmid library (14) and designated pET-TtFTA. The expression vector for the malic enzyme of Thermococcus kodakarensis KOD1 (TkME) was constructed as described elsewhere (15). E. coli Rosetta 2 (DE3) pLysS (Novagen, Madison, WI) was used as the host cell for gene expression. Recombinant E. coli was cultured in a 500-ml Erlenmeyer flask containing 200 ml of Luria-Bertani broth supplemented with 100 mg/liter ampicillin and 34 mg/liter chloramphenicol. Cells were cultivated at 37°C with orbital shaking at 180 rpm. Isopropyl-β-d-1-thiogalactopyranoside (IPTG) was added to the culture at a final concentration of 0.4 mM in the late-log phase. After a 3-h induction, the cells were harvested by centrifugation and washed once with 0.1 M sodium phosphate buffer (pH 7.0).

Glutaraldehyde treatment.

Two hundred milligrams of wet cells was suspended in 1 ml of 0.1 M sodium phosphate buffer (pH 7.0). GA solution (25% in water; Nacalai Tesque, Kyoto, Japan) was added to the cell suspension to give final concentrations of 0.03% to 0.15% (vol/vol). The mixture was gently stirred at 4°C for 1 h. The treated cells were harvested by centrifugation at 8,000 × g for 10 min and then washed once with the same buffer.

Enzyme assay.

TtFTA activity was assessed at 70°C by coupling with the Thermococcus malic enzyme, which catalyzes a NADP⁺-dependent decarboxylation of malate to pyruvate. E. coli cells having either TtFTA or TkME were suspended in 0.1 M sodium phosphate buffer (pH 7.0) and disrupted using an ultrasonicator (UD-201; Kubota, Osaka, Japan) at 40 W for 3 min. The resulting lysate was heated at 70°C for 20 min. After the removal of cell debris and denatured proteins by centrifugation, the resulting supernatant was used as the enzyme solution. Enzyme activity was measured spectrophotometrically by monitoring the increase in the absorbance of NADPH at 340 nm. A molar absorption coefficient for NADPH of 6.22 × 10³ M⁻¹ cm⁻¹ was used. One unit of enzyme was defined as the amount of enzyme that catalyzes the production of 1 μmol of NADPH per 1 min. The standard reaction mixture contained 1.25 mM sodium fumarate, 50 mM MnCl₂, 0.1 mM NADP⁺, 0.1 U TkME, and 0.1 M phosphate buffer (pH 8.0), in a total volume of 1 ml. The reaction was started by adding an appropriate amount of TtFTA solution.

For the determination of the reaction rate catalyzed by whole cells, free or GA-treated E. coli cells having TtFTA were suspended in 0.1 M phosphate buffer at a concentration of 8 mg (wet weight)/ml. Cells were used as catalysts after the heat treatment at 70°C for 20 min. The reaction was initiated by adding 2.5 ml of the heat-treated cell suspension into a screw-cap cylindrical vessel (inner diameter, 35 mm) containing 2.5 ml of 0.4 M sodium fumarate in the same buffer, which was preheated at 70°C. The reaction was performed at 70°C with stirring. Aliquots (0.5 ml) of the mixture were withdrawn after the reaction for 5, 10, 30, and 60 min, acidified with HCl to stop the reaction, and then subjected to high-performance liquid chromatography (HPLC).

Heat-induced leakage of TtFTA.

The free and GA-treated E. coli cells having TtFTA were suspended in 0.1 M phosphate buffer (pH 7.0) at 200 mg (wet weight)/ml and incubated at 70°C. After the incubation for 5, 10, 30, and 60 min, the cells and their debris were removed by centrifugation at 15,000 × g for 10 min at 4°C. The supernatants were resubjected to heat treatment at 70°C for 20 min. Denatured proteins were removed by centrifugation, and the TtFTA activity in the supernatant was determined. Leakage level was expressed as the percentage of TtFTA activity in the cell extract, prepared by subjecting the same concentration (200 mg/ml) of the free-cell suspension to ultrasonication, followed by heat treatment at 70°C for 20 min.

Electron microscopic analysis.

Scanning electron microscopy (SEM) (S-5200; Hitachi, Tokyo, Japan) was used to take images of the free and GA-treated E. coli cells before and after the heat treatment at 70°C for 20 min. The suspensions of the free and GA-treated cells were centrifuged at 5,000 × g for 10 min. The pellets were fixed with 2.5% (wt/wt) GA–0.2 M phosphate buffer for 1 h, and then stained with 5% (wt/wt) osmic acid in the same buffer for 30 min. The samples were dehydrated in a graded ethanol series, lyophilized for 2 h, and then coated with osmium (HPC-1S hollow cathode plasma chemical vapor deposition, vacuum device; Mito, Japan).

For transmission electron microscopy (TEM), the dehydrated samples were mixed with an epoxy resin solution and then polymerized at 45°C for 5 days. The resin-embedded samples were cut with a glass knife on a Reichert-Nissei ultramicrotome, mounted on carbon-coated copper grids (Nisshin EM, Tokyo, Japan), and poststained with uranyl acetate and lead citrate. TEM was carried out with a JEOL JEM 1200 EX (80-kV) electron microscope.

Effect of heat treatment on the membrane permeability of GA-treated E. coli cells.

Fluorescein-labeled dextrans with average molecular masses of 3 kDa (Dex3) and 40 kDa (Dex40) (Invitrogen, Carlsbad, CA) were used to evaluate the membrane permeability of the GA-treated cells. The GA-treated cells were suspended in 5 ml of 0.1 M sodium phosphate buffer (pH 7.0) at 200 mg (wet weight)/ml and then heated at 70°C for 20 min. The heated-cell suspension was centrifuged at 8,000 × g for 10 min, and the cell pellet was washed once with the same buffer. The cells were then resuspended in 5 ml of the buffer containing either Dex3 or Dex40 (0.05 μM each) and incubated in the dark at room temperature for 30 min with gentle shaking. After incubation, the cells were pelleted by centrifugation at 8,000 × g for 10 min and then resuspended in 5 ml of the fresh buffer without dextrans. The resuspended solution was shaken with a vortex mixer for 1 min and then centrifuged at 8,000 × g for 10 min. The fluorescence intensity of the supernatant was measured at excitation and emission wavelengths of 494 and 521 nm, respectively. A control experiment was carried out using the GA-treated cells without the heat treatment.

Reusability of GA-treated cells.

The reaction mixture containing 4 mg/ml wet cells having TtFTA and 0.2 M sodium fumarate–0.1 M sodium phosphate buffer (pH 7.0) was placed in a screw-cap cylindrical vessel (inner diameter, 27 mm) and incubated at 70°C with stirring. After being allowed to react for 30 min, the cells were removed by centrifugation at 8,000 × g for 10 min. The supernatant was acidified with HCl, recentrifuged to remove denatured proteins, and then subjected to HPLC to determine the concentrations of malate and fumarate. The reaction was repeatedly performed in the same manner to assess the reusability of the free and GA-treated E. coli cells. After each reaction cycle, the cells were harvested by centrifugation at 8,000 × g for 10 min, washed with the buffer, and then resuspended in a fresh reaction mixture.

HPLC was performed using a Shimadzu Prominence LC-20 system (Kyoto, Japan) equipped with a Cosmosil 5C18-AR-II packed column (Nacalai Tesque) (4.6 by 250 mm) at 50°C. The UV detector was set at 210 nm, and a 0.1% (vol/vol) phosphoric acid solution at a flow rate of 1 ml/min was used as the mobile phase.

Malate production in a continuous reactor using GA-treated cells.

The reaction mixture (10 ml) comprising 4 mg (wet weight)/ml of the GA-treated cells having TtFTA, 0.2 M sodium fumarate, and 0.1 M sodium phosphate buffer (pH 7.0) was put in a stirred ultrafiltration cell (model 8010; Millipore, Bedford, MA). A polyvinylidene fluoride (PVDF) membrane filter (Toray, Kamakura, Japan) (diameter, 25 mm) (16) was set at the bottom of the reactor to separate cells from the solution. The reaction was performed at 70°C with stirring by a magnetic stirrer. After the reaction was performed for 40 min to achieve a steady state, a fresh substrate solution (0.2 M sodium fumarate in the same buffer) was added and the product solution was removed using peristaltic pumps (model 3385; Fisher Scientific, Fair Lawn, NJ). To determine product yield, aliquots (0.1 ml) of the eluent were taken at 40-min intervals and subjected to HPLC. The feeding and removal rates of the solution were maintained at 0.25 ml/min. Residence time and conversion rate were calculated as follows:

• Residence time τ (min)

  $$\tau =F/V$$
• Conversion rate γ (mmol/liter/min)

  $$\left(S_{\text{in}}-S\right)=\tau \times \gamma$$

where S_in = substrate concentration in the feeding solution, S = substrate concentration in the reactor, F = flow rate, and V = total volume of reaction mixture.

RESULTS

Effects of GA treatment on heat-induced leakage of TtFTA.

Fig. 1 shows the time course of the heat-induced leakage of TtFTA from the recombinant E. coli cells. When the free cells were incubated at 70°C, more than 80% of the activity was released from the cells by 5 min. This observation was in good agreement with previous reports on the thermolysis of E. coli cells (7–9). Owing to thermolysis, the boundary surface between the supernatant and the pelleted free cells was unclear after the centrifugation at 15,000 × g for 10 min. However, the heated GA-treated cells could be clearly separated by centrifugation. The level of TtFTA leakage from the GA-treated cells was considerably lower than that from the free cells and obviously decreased with the increasing GA concentration used for the cross-linkage. The leakage from the cells pretreated with 0.11% (vol/vol) or higher GA concentrations was kept under a detectable level after the incubation at 70°C for at least 60 min. However, it was also possible that the decrease in the apparent level of enzyme leakage arose from the inhibitory effect of GA on the enzyme as well as the lowered substrate permeability due to the consolidation of the membrane structure caused by the cross-linkage. We then assessed the performance of the free and GA-treated cells for catalyzing fumarate hydration. As predicted, the catalytic performance of GA-treated cells decreased in proportion to the GA concentration used for their preparation (Fig. 2). The apparent initial reaction rate was estimated on the basis of the product concentration after the reaction for 5 min. The reaction rates obtained using the cells pretreated with 0.03, 0.07, 0.11, and 0.15% GA were 94, 80, 77, and 64% of that obtained using the free cells, respectively. However, no significant difference was observed in the product concentration after the reaction for 30 min. On the basis of these observations, we employed a GA concentration of 0.11% for the cross-linkage as the best compromise between the prevention of heat-induced leakage and the maintenance of enzyme activity.

Fig 1.

Heat-induced leakage of TtFTA from free cells (filled diamonds) and cells pretreated with 0.03% (circles), 0.07% (open diamonds), 0.11% (triangles), and 0.15% (squares) GA. The leakage level was expressed as the percentage of the TtFTA activity in the cell extract prepared by ultrasonication. The data are shown as averages ± standard deviations (n = 3).

Fig 2.

Time course of malate production catalyzed by free cells (filled diamonds) and cells pretreated with 0.03% (circles), 0.07% (open diamonds), 0.11% (triangles), and 0.15% (squares) GA. The data are shown as averages ± standard deviations (n = 3).

The conversion of fumarate to malate is a reversible reaction with the equilibrium at approximately 80 mol% (13, 17) under the standard state (25°C). In this study, the conversion yield using the series of E. coli cells having TtFTA reached about 70 mol% after 30 min and then remained constant, regardless of the difference in their specific activities. This indicates that the equilibrium of the reaction under our experimental conditions was achieved at a conversion yield of approximately 70 mol%.

Electron microscopic analysis.

The free and GA-treated cells before heat treatment showed the appearance of normal healthy cells in both SEM and TEM (Fig. 3A, C, E, and G). In contrast, after a 20-min incubation at 70°C, the SEM image of the free cells displayed cellular structure disintegration (Fig. 3B). Blebs with various sizes were formed on the cell surface. The TEM of the heat-treated free cells showed the surface subsidence of the cells, probably resulting from membrane disruption followed by the leakage of intracellular components, suggesting that the heat-induced leakage of the enzyme was attributable to the collapse of the membrane structure of the cells (Fig. 3F).

Fig 3.

SEM (A to D) and TEM (E to H) images of free and GA-treated E. coli cells before and after heat treatment at 70°C for 20 min. The open circle shows small blebs on the cell surface.

In contrast, the overall structure of the GA-treated cells still remained in good condition after the incubation at 70°C for 20 min. Both TEM and SEM showed the appearance of a cell surface comparable to the surface of healthy cells, except for the formation of some small blebs on the membrane (Fig. 3H). This observation supported the notion that the prevention of the heat-induced leakage of TtFTA from E. coli cells was, at least in part, attributable to the consolidation of the membrane structure by the GA treatment.

Membrane permeability of GA-treated cells.

Dextrans, polymers of glucose, are hydrophilic polysaccharides characterized by a high molecular mass and good water solubility. Dextrans with molecular masses higher than 2 kDa generally cannot permeate into intact cells (18). In this study, fluorescein-labeled dextrans with different average molecular masses were used to assess the membrane permeability of the GA-treated cells. The cells were first suspended in a fluorescence-labeled dextran solution and pelleted by centrifugation, and then the pellet was resuspended in a dextran-free buffer. The fluorescence intensity of the supernatant of the resulting cell suspension correlates with the amount of dextran involved in the cell pellet; therefore, a higher intensity can be obtained when dextran can penetrate into the cell. The slight fluorescence observed in the control experiments with the non-heat-treated cells was likely attributable to the dextran remaining in the intercellular space of the cell pellets (Fig. 4, gray bars). The fluorescence intensity of the supernatant of the heat-treated cell suspension was almost double that of the non-heat-treated cell suspension when Dex3 was used as an indicator of membrane permeability. On the other hand, when Dex40 was used, no significant difference was observed between the fluorescence intensities of the heat-treated and non-heat-treated cell suspensions. Taken together with the results of electron microscopic observation, these results indicated that the membrane structure of the GA-treated cells was partly disrupted by the heat treatment, while maintaining the overall cell structure; therefore, only relatively small molecules (up to at least 3 kDa) could permeate into and out of the cells.

Fig 4.

Membrane permeabilities of GA-treated cells with (white bars) and without (gray bars) heat treatment. The dextran concentration was determined using a calibration curve obtained by measuring the fluorescence intensities of the serial dilutions of the standards. The data are shown as averages ± standard deviations (n = 3).

Reusability of GA-treated cells in repeated batch reactions.

To evaluate the reusability of the GA-treated cells having TtFTA, batch reactions were repeatedly performed. As a control experiment, the reactions were also carried out using the free cells. Although a slight decrease was observed in the product concentration, probably due to the thermal inactivation of TtFTA, the product yields of the reactions with the GA-treated cells were maintained above 60 mol% when the reactions were repeated 6 times (Fig. 5). In contrast, the product yields with the free cells linearly decreased with repeated reactions. These results clearly demonstrated the superiority of the GA-treated cells in the repeated reactions.

Fig 5.

Reusabilities of free (diamonds) and GA-treated (triangles) cells. Each reaction was performed at 70°C for 30 min. The data are shown as averages ± standard deviations (n = 3).

Continuous bioconversion using GA-treated cells as whole-cell catalysts.

For a further demonstration of the applicability of the GA-treated cells to long-term bioconversion, the cells were used as a catalyst in a continuous reactor. The feeding rate of the substrate solution (0.25 ml/min) was experimentally decided to achieve a conversion yield of 60 mol%, and residence time (τ) was calculated to be 40 min. When the free cells were used as a catalyst, the filterability of the separation membrane was significantly impaired, and product recovery at a constant rate could not be achieved (data not shown). This was likely due to the formation of a cake layer on the separation membrane by debris of heat-damaged cells. On the other hand, when the GA-treated cells were used, a continuous product recovery could be maintained for at least 640 min. The production yield decreased slightly in the first 80 min but then remained stable until the end of the operation. As a result, an overall product yield of 60 mol% could be achieved during the operation for 640 min. Production rate was calculated to be 3.0 mmol/liter/min, and 19.2 mmol (2.57 g) of malate could be produced from 32 mmol (3.71 g) of fumarate.

DISCUSSION

In this study, we demonstrated that the heat-induced leakage of a thermophilic enzyme from recombinant E. coli cells could be prevented by the pretreatment of the cells with GA. Through the optimization of the GA concentration, a thermotolerant whole-cell catalyst with an acceptable catalytic ability could be prepared and successfully used for a continuous bioconversion of fumarate to malate. This approach is much simpler and more universal for the preparation of stable and selective immobilized enzymes than conventional procedures, which typically involve many steps, such as protein extraction, enzyme purification, and the preparation of immobilization carriers. The GA-treated cells exhibited sufficient catalytic stability and reusability in both repeated-batch and continuous bioconversions at 70°C. They could catalyze fumarate hydration in the continuous bioreactor over 10 h with a constant product yield of 60 mol% or higher. This productivity was considerably better than that in a previous report on the continuous conversion of fumarate using permeabilized Saccharomyces cells, in which the malate yield was about 50% and the enzyme lost 30 to 50% of its activity after 10 h (19).

The electron microscopic observation of the GA-treated cells revealed that heat treatment caused no apparent change in the overall structure of the cells except for the formation of some small blebs on their surface. GA is a five-carbon dialdehyde that can react with several functional groups of proteins, such as amine, thiol, phenol, and imidazole, and form covalent bonds with them (20–23). The GA-mediated stabilization of the membrane structure implied that GA might cross-link between protein molecules in the outer membrane and those in the peptidoglycan layer of the E. coli membrane to make it more rigid. In fact, the cells changed to pink-red after the GA treatment; this phenomenon is typically observed as a result of the interaction between membrane proteins and GA (24, 25).

The natural barrier function of the cellular membrane often hampers sufficient whole-cell bioconversion since it limits the accessibility of substrates to intracellular enzymes (26). Although no significant change was observed in the overall structure of cells, the membrane permeability test with different sizes of fluorescently labeled dextrans revealed that relatively small molecules (up to at least 3 kDa) could penetrate the membrane barrier of the GA-treated cells heated at 70°C. The formation of small blebs observed in the TEM analysis of the heated GA-treated cells was likely responsible for the increased membrane permeability. Most lipids and lipopolysaccharides comprising the cell membrane cannot be fully cross-linked to each other by GA (21). Thus, their orientation might be disrupted at high temperatures. Consequently, small blebs were formed on the membrane of the GA-treated cells, and the rupture of the blebs resulted in the formation of membrane pores, through which molecules having low and medium molecular masses can diffuse. The fact that the dextran with an average molecular mass of 40 kDa hardly penetrated the membrane was in good agreement with the observation that the leakage of TtFTA (approximately 51 kDa) could be prohibited by GA treatment.

The observation that the GA-mediated consolidation of the membrane structure was involved in the prevention of the heat-induced leakage of the enzyme raised the issue of whether the intracellular enzyme can freely diffuse in the cytosol of the GA-treated cells. To address this question, we attempted to extract the soluble form of TtFTA from the GA-treated cells by disrupting the cells by lysozyme digestion followed by ultrasonication. The TtFTA activity recovered in the resulting lysate was approximately 17% of that detected in the cell extract of the free cells, indicating that intracellular TtFTA was, at least in part, present in soluble form and freely diffused in the cytosolic fraction.