Ferricyanide Reduction by Escherichia coli: Kinetics, Mechanism

Ferricyanide reduction was studied by flow injection analysis (FIA) and chronoamperometry (CA) using two host strains and one recombinant strain of E...
0 downloads 0 Views 108KB Size
Download PDF
Anal. Chem. 2000, 72, 4949-4956

Ferricyanide Reduction by Escherichia coli: Kinetics, Mechanism, and Application to the Optimization of Recombinant Fermentations Peter Ertl,† Birgit Unterladstaetter,‡ Karl Bayer,‡ and Susan R. Mikkelsen*,†

Department of Chemistry, University of Waterloo, Waterloo, Ontario, Canada, N2L 3G1, and Institute of Applied Microbiology, State University of Agricultural Sciences, Nussdorfer Laende 11, A-1190 Vienna, Austria

Ferricyanide reduction was studied by flow injection analysis (FIA) and chronoamperometry (CA) using two host strains and one recombinant strain of E. coli. Samples taken from batch cultures of E. coli JM105 and HB101 showed maximal specific ferricyanide reduction rates in the late exponential phase of growth, with values (µmol/min‚g) of 24 (FIA) and 17 (CA) for JM105, and 36 (FIA) for HB101, when shake-flask cultures were sampled, and 70 for HB101, when a chemostat was used to control pH and dissolved oxygen concentration throughout the cultivation. Remarkably higher ferricyanide reduction rates were obtained with HB101 cells cultivated continuously at very slow growth rate, when chilled, resuspended cell samples were incubated for 5 min in solutions containing 10 mM succinate or formate. These compounds are substrates for primary, membrane-bound dehydrogenases that transfer electrons via ubiquinone to the cytochrome oxidase complexes. Apparent MichaelisMenten kinetics were observed with respect to ferricyanide concentration when 10 mM succinate was included in the assay buffer; apparent Km values of 10.1 ( 0.6 mM and 14.4 ( 1.2 mM ferricyanide were obtained for exponential- and stationary-phase E. coli JM105, respectively. Cyanide inhibition studies show that ferricyanide is reduced mainly by cytochrome o oxidase in exponentially growing cells. The large difference in ferricyanide reduction rates observed in the absence and presence of succinate and formate were used to signal stationary-phase entry 5 h after induction of recombinant human Cu/Zn superoxide dismutase expression in a batch fermentation of E. coli HMS174(DE3)(pET3ahSOD). This new method can be used as an adjunct to the quantitation of medium components for the optimization of recombinant fermentations. In research and industrial settings, the growth of microorganisms is monitored and controlled to maximize the yield of native or recombinant proteins. Chemostats are used to control pH and dissolved oxygen levels through measurements from in situ * To whom correspondence should be addressed: (fax) (519) 746-0435; (email) [email protected]. † University of Waterloo. ‡ State University of Agricultural Sciences. 10.1021/ac000358d CCC: $19.00 Published on Web 09/16/2000

© 2000 American Chemical Society

sensors, and off-line measurements of biomass and medium composition commonly form the basis for timing the addition of medium components and the induction of recombinant protein expression.1,2 In aerobic fermentations, the growth rate of the culture is proportional to the specific respiration rate, the rate of oxygen consumption divided by cell density or biomass,1-3 and strong evidence points to terminal oxygen reduction as the ratelimiting step in the overall growth of batch cultures.4,5 Respiration in aerobic fermentations is commonly controlled by varying agitation and air flow rates to maintain the dissolved oxygen concentration at a constant value, regardless of cell density.1,3 Aerobic respiration in Escherichia coli results from a number of parallel paths that converge in the cytoplasmic membrane just after the reactions catalyzed by membrane-bound, primary dehydrogenases (Scheme 1).6-10 These dehydrogenases do not rely on the nicotinamide cofactors (NADH and NADPH) that their soluble counterparts require; instead they transfer electrons to membrane-soluble quinones (mainly ubiquinone-8) that transport reducing equivalents to membrane-bound b-type cytochromes, which then react with a cytochrome oxidase complex that ultimately reduces molecular oxygen to water. A great deal has been learned within the past decade about the components of the E. coli respiratory chain6-8 and their regulation,9,10 and it is now known that the transition from exponential- to stationary-phase growth triggers significant changes in the protein distribution within the cells. When grown in batch culture, E. coli eventually depletes its surrounding medium of at least one essential nutrient. The limiting (1) Omstead, D. R., Ed. Computer Control of Fermentation Processes; CRC Press: Boca Raton, FL, 1990. (2) Georgiou, G. AIChE J. 1988, 34, 1233-1248. (3) Crueger, W.; Crueger, A. Biotechnology: A Textbook of Industrial Microbiology, 2nd ed.; Sinauer (Science Tech Publishers): Madison, WI, 1990; pp 8692. (4) Andersen, K. B., von Meyenburg, K. J. Bacteriol. 1980, 144, 114-123. (5) Sun, I. L., Crane, F. L. Biochim. Biophys. Acta 1985, 811, 233-264. (6) Gennis, R. B.; Stewart, V. In Escherichia coli and Salmonella: Cellular and Molecular Biology; Niedhardt, F. C., Ed.; American Society for Microbiology: Washington, DC, 1997; Vol. 1, Chapter 17, pp 217-261. (7) Unden, G.; Bongaerts, J. Biochim. Biophys. Acta 1997, 1320, 217-234. (8) Sturr, M. G.; Krulwich, T. A.; Hicks, D. B. J. Bacteriol. 1996, 176, 17421749. (9) Hengge-Aronis, R. In Escherichia coli and Salmonella: Cellular and Molecular Biology; Niedhardt, F. C., Ed.; American Society for Microbiology: Washington, DC, 1997; Vol.ume 1, Chapter 93, pp 1497-1512. (10) Loewen, P. C.; Hu, B.; Strutinsky, J.; Sparling, R. Can. J. Microbiol. 1998, 44, 707-717.

Analytical Chemistry, Vol. 72, No. 20, October 15, 2000 4949

Scheme 1. Reactions of the Respiratory Chain of Aerobically Grown E. coli, Including the Primary Dehydrogenases (PDH) and the Cytochrome Oxidase Complexes (Ox) (Summarized from Refs 6-10)

nutrient may be the source of carbon (such as glucose or glycerol), nitrogen, or phosphate, and in-shake flask cultures where dissolved oxygen is not controlled, the rapid growth of exponential-phase cultures can cause oxygen depletion. Depletion of any one of these nutrients triggers the production of a different sigma subunit (σS) for the enzyme RNA polymerase; this subunit is believed to change the binding selectivity of the enzyme for promoter sequences in the E. coli genome and thereby causes a change in the relative expression levels of the promoted genes.11 The transition from the normal σD subunit, present during exponential growth, to ∼30% σS (as has been found in stationaryphase cultures11) has been shown to cause increased production of certain respiratory enzymes, including pyruvate oxidase, glycerol-3-phosphate dehydrogenase, and the recently discovered cytochrome bd II oxidase, as well as various outer membrane and periplasmic proteins, to allow cells to survive nutrient limitation and to change the physiology of the cell to provide maximal protection against diverse stress conditions.9,10 E. coli and other Gram-negative microorganisms can reduce non-native, hydrophilic oxidants such as ferricyanide directly.12-14 The terminal components of the Gram-negative respiratory chain are associated with the inner (cytoplasmic) membrane,6 but are accessible to small, hydrophilic oxidants such as ferricyanide, because the outer membrane contains channel proteins (porins) that allow free diffusion of low-molecular-weight species (MW e ∼1000) into the periplasmic space.15 Respiration has been quantitated in aerobic, anaerobic, prokaryotic, and eukaryotic cells as well as specialized tissues, using a variety of non-native oxidants such as ferricyanide,12-14,16-20 ferric (11) Jishage, M.; Ishihama, A. J. Bacteriol. 1995, 177, 6832-6835. (12) Ramsay, G.; Turner, A. P. F. Anal. Chim. Acta 1988, 215, 61-69. (13) Ding, T.; Schmid, R. D. Anal. Chim. Acta 1990, 234, 247-251. (14) Gaisford, W. C.; Richardson, N. J.; Haggett, B. G. D.; Rawson, D. M. Biochem. Soc. Trans. 1991, 19, 15-18. (15) Atlas, R. M. Principles of Microbiology, 2nd ed.; Wm. C. Brown: Dubuque, IO, 1997; pp 109-113. (16) Hadjipetrou, L. P.; Gray-Young, T.; Lilly, M. D. J Gen. Microbiol. 1966, 45, 479-488. (17) Crane, F. L.; Sun, I. L.; Clark, M. G.; Grebing, C.; Loew, H. Biochim. Biophys. Acta 1985, 811, 233-264. (18) Kulys, J.; Wang, L.; Razumas, V. Electroanalysis 1992, 4, 527-532.

4950

Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

chelates,21 2,6-dichlorophenolindophenol (DCIP),19,20,22,23 phenazine and phenoxazine derivatives,18,24,25 benzoquinone18,19,26 and quinone derivatives,14,27 and methyl viologen.28 These compounds possess appropriate reduction potentials and are readily quantitated in their reduced forms. The oxidants are sometimes used in combination, where a hydrophobic, membrane-permeable mediator at low concentration is combined with a hydrophilic terminal acceptor present at higher concentration; for example, DCIP and ferricyanide have been used with E. coli O15720 and beer yeasts.29 Electrochemical measurements of direct ferricyanide reduction by suspensions of E. coli have been reported.12-14,20,30 The reduction potential of the ferri-/ferrocyanide redox couple (+0.418 V vs NHE31) is positive of the terminal oxidases (see Scheme 1), suggesting that any sequence of preceding primary dehydrogenase reactions may contribute to observed ferricyanide reduction rates. Studies with E. coli have shown that additions of glucose, succinate, lactate, and formate can increase observed ferricyanide reduction rates, and similar effects have been observed with other microorganisms.30 Electrochemical transducers have also been used in cell-based biosensors for metabolizable substrate quantitation18,23,26,30,32 and in living fuel cells, where respiration generates a spontaneous current.21,22,25 When recombinant protein expression is induced in E. coli, the effect is similar to entrance into the stationary growth phase, since the cellular metabolic machinery shifts from its normal reproductive pathways toward the synthesis of a metabolically irrelevant product.11,33-35 As a result of this shift, and especially with a strong promoter attached to the inserted gene, the cell effectively experiences starvation.33 A recent study has shown that overexpression of recombinant proteins causes accumulation of two proteins normally produced during the heat-shock response;34 furthermore, a separate study showed that E. coli subjected to heat shock (30-42 °C) produced the same σ subunit for RNA polymerase (σS) as is normally produced upon entry into the stationary growth phase.11 The ultimate effect of this transition (19) Ikeda, T.; Kurosaki, T.; Takayama, K.; Kano, K.; Miki, K. Anal. Chem. 1996, 68, 192-198. (20) Perez, F. G.; Mascini, M.; Tothill, I. E.; Turner, A. P. F. Anal. Chem. 1998, 70, 2380-2386. (21) Tanaka, K.; Vega, C. A.; Tamamushi, R. Bioelectrochem. Bioenerg. 1983, 11, 135-143. (22) Nishikawa, S.; Sakai, S.; Karube, I.; Matsunaga, T.; Suzuki, S. Appl. Environ. Microbiol. 1982, 43, 814-818. (23) Takayama, K.; Kurosaki, T.; Ikeda, T.; Nagasawa, T. J. Electroanal. Chem. 1995, 381, 47-53. (24) Turner, A. P. F.; Ramsay, G.; Higgins, I. J. Biochem. Soc. Trans. 1983, 11, 445-448. (25) Bennetto, H. P.; Stirling, J. L.; Tanaka, K.; Vega, C. A. Biotechnol. Bioeng. 1983, 25, 559-568. (26) Takayama, K.; Kurosaki, T.; Ikeda, T. J. Electroanal. Chem. 1993, 356, 295301. (27) McManus, D. C.; Josephy, P. D. Anal. Biochem. 1993, 304, 367-370. (28) Tatsumi, H.; Takagi, K.; Fujita, M.; Kano, K.; Ikeda, T. Anal. Chem. 1999, 71, 1753-1759. (29) Chunxiang, X.; Gang, L.; Haobin, C.; Yue, X. Sens. Actuators B 1993, 12, 45-48. (30) Richardson, N. J.; Gardner, S.; Rawson, D. M. J. Appl. Bacteriol. 1991, 70, 422-426. (31) Szentrimay, R.; Yeh, P.; Kuwana, T. In Electrochemical Studies of Biological Systems; Gould, R. F., Ed.; American Chemical Society: Washington, DC, 1977; p 143. (32) Miki, K.; Tsuchida, T.; Kawagoe, M.; Kinoshita, H.; Ikeda, T. Denki Kagaku 1994, 62, 1249-1250. (33) Kurland, C. G.; Dong, H. Mol. Microbiol. 1996, 21, 1-4. (34) Dong, H.; Nilsson, L.; Kurland, C. G. J. Bacteriol. 1995, 177, 1497-1504. (35) Atlung, T.; Bronsted, L. J. Bacteriol. 1994, 176, 5414-5422.

on the recombinant organisms was cell death due to ribosome destruction,34 but its effect on the quality and quantity of recombinant proteins has not yet been investigated. We now report that electrochemical measurements of ferricyanide reduction rates made on host and recombinant strains of E. coli can be used to indicate entry into the stationary growth phase. Cultures were examined using electrochemical flow injection analysis (FIA) and chronoamperometry (CA) at varying growth rates and under different assay conditions. We have found that brief incubations of chilled, resuspended cell samples with the primary dehydrogenase substrates succinate and formate lead to information complementary to that provided by standard oxygen-uptake measurements. Measurements made during the batch production of recombinant human Cu/Zn superoxide dismutase (SOD)36 suggest stationary-phase entry 4 h after induction of SOD biosynthesis. This new method can be applied to the optimization of recombinant protein production in industrial fermentations. EXPERIMENTAL SECTION Materials and Instrumentation. Potassium hexacyanoferrate(II), potassium hexacyanoferrate(III), cobalt(II) chloride hexahydrate, D-(+)-glucose monohydrate and sodium L-lactate were obtained from Fluka. Anthraquinone-1-sulfonate, 1,4-benzoquinone, Janus Green B, and 2,3-dimethoxy-5-methyl-1,4-benzoquinone were obtained from Aldrich. L-Proline, di(monocyclohexylammonium)L-glycerol-3-phosphate and isopropyl β -D- thiogalactopyranoside (IPTG) were obtained from Sigma. Potassium cyanide was purchased from BDH. Sodium pyruvate, disodium malate, succinic acid, sodium formate, dipotassium hydrogen phosphate trihydrate, potassium dihydrogen phosphate, trisodium citrate dihydrate, magnesium sulfate heptahydrate, calcium chloride dihydrate, ammonium chloride, ammonium sulfate, extract of yeast powder (No. 3753), sodium hydroxide, potassium hydroxide, iron(II) sulfate heptahydrate, manganese(II) sulfate monohydrate, aluminum chloride hexahydrate, zinc(II) sulfate heptahydrate, disodium molybdate dihydrate, copper(II) chloride dihydrate, sodium chloride, and boric acid were supplied by Merck. Difco provided bacto-tryptone powder, and Dow Corning provided industrial grade antifoam RD emulsion. Solutions and media were prepared using distilled, deionized water. E. coli HB101 and HMS174 (DE3)(pET3ahSOD) were obtained from the Institute of Applied Microbiology strain collection and maintained on Luria Broth medium (tryptone, 10 g/L, yeast extract, 5 g/L, NaCl, 5 g/L, pH 7.2 at 37 °C). E. coli JM105 was obtained from the strain collection of the Department of Biology, University of Waterloo. Fermentations conducted in shake flasks were maintained at 37 °C and 200 rpm. Optical density measurements were made at 600 nm using either a Novaspec II single-beam visible spectrophotometer (Pharmacia) or a Cary 1 double-beam UV-visible spectrophotometer. Three CMF minifermentors (Chemap AG, Switzerland) were used in this work: two had total and working volumes of 2.0 and 1.2 L, while the third had total and working volumes of 4.0 and 2.4 L, respectively. These fermentors were equipped with computer(36) Kramer, W.; Elmecker, G.; Weik, R.; Mattanovich, D.; Bayer, K. Ann. N. Y. Acad. Sci. 1996, 782, 323-327.

controlled chemostat systems to maintain pH (through the addition of 10% aqueous KOH) and dissolved O2 (by controlling agitation) at preset values. Ingold electrodes were used for pH and dissolved O2 measurements. Foaming was controlled by the addition of 10% (v/v) aqueous antifoam emulsion. The flow injection system used with E. coli HB101 and HMS174(DE3)(pET3ahSOD) samples consisted of an Ismatec SA eight-cylinder low-pressure pump, a Rheodyne 7125 injector equipped with a 100-µL injection loop, a CC-5 thin-layer electrochemical flow cell with glassy carbon working, Ag/AgCl reference and stainless steel auxiliary electrodes controlled by a CV-27 potentiostat (Bioanalytical Systems), and a BBC SE 120 strip chart recorder. For E. coli JM105 samples, a Waters model 510 HPLC pump was used with a Rheodyne 7125 injector (20-µL sample loop) and a thin-layer amperometric detector (E. G. & G. Princeton Applied Research model 400 with glassy carbon working, Ag/ AgCl reference, and stainless steel auxiliary electrodes). The mobile-phase flow rate was set to 0.30 mL/min in both instruments. Chronoamperometry was performed using a PAR model 263 potentiostat with a glassy carbon rotating disk working electrode at 600 rpm (PAR model 616) at +0.8 V vs an Ag/AgCl reference electrode (Bioanalytical Systems), using a stainless steel wire auxiliary electrode. Methods. Cultivation of E. coli. A stock solution of trace elements for growth media was prepared in 5 N HCl (Merck) and contained the following compounds: FeSO4‚7H2O (40 g/L), MnSO4‚H2O (10 g/L), AlCl3‚6H2O (10 g/L), CoCl2‚6H2O (4 g/L), ZnSO4‚7H2O (2 g/L), Na2MoO4‚2H2O (2 g/L), CuCl2‚2H2O (1 g/L), and H3BO3 (0.5 g/L). The growth medium contained the following components: KH2PO4 (2.88 g/L), K2HPO4‚3H2O (5.76 g/L), tryptone (2.4 g/L), yeast extract powder (1.2 g/L), trisodium citrate dihydrate (1.2 g/L), MgSO4‚7H2O (0.48 g/L), CaCl2‚2H2O (0.048 g/L), (NH4)2SO4 (1.63 g/L), NH4Cl (1.34 g/L), glucose (13.2 g/L), and 240 µL of the trace element stock solution per liter of medium. The glucose was prepared as a concentrated aqueous solution, sterilized separately, and added to the growth medium prior to inoculation. Inocula for chemostat cultivations were prepared by the addition of 2 mL of 1:1 glycerol/cell culture mixture (that had been stored at -80 °C) to 30 mL of growth medium in a shake flask. Growth proceeded overnight, for a minimum of 9 h and a maximum of 14 h, in an incubator-shaker. Continuous cultures were allowed to stabilize until five volume exchanges of medium had occurred at the desired dilution rate. The fermenters were maintained at 37 °C, pH 6.90, and a dissolved O2 concentration of 50% of the value obtained prior to inoculation under air saturation conditions. Initial batch fermentations were allowed to proceed to late exponential phase, at which time continuous cultures were initiated. During continuous culture, growth rates were maintained by the controlled addition of fresh medium, via a low-pressure pump, and the removal of cell culture at the same rate, such that the total volume in the fermenter was constant. Sample Collection. Cultivations in shake flasks (50 mL total volume) were sampled periodically by transferring 1-2 mL to Eppendorf tubes and chilling the samples to 0 °C for a minimum of 10 min. Sampling from the fermenter (∼20 mL) was performed using a sterile syringe inserted into a T-valve that allowed steam Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

4951

sterilization following sample removal. Determination of Optical Density. A measured aliquot of the cell suspension was centrifuged at 5000 rpm for 10 min. The supernatant was removed, and the pellet was reconstituted by the addition of 1.50 mL of distilled water. Sample dilution factors were chosen such that the measured optical density (600 nm) was between 0.1 and 0.6 au, using water as a reference. Determination of Bacterial Dry Matter (BDM). Samples were centrifuged (5000 rpm, 10 min), the supernatant was decanted, and the pellets were twice resuspended in 10 mL of water and recentrifuged. Following a final resuspension, the cells were quantitatively transferred to a preweighed beaker, dried at 105 °C, cooled in a desiccator, and reweighed. Quantitation of SOD. Samples of E.coli HMS174(DE3)(pET3ahSOD) were treated with lysozyme in the presence of Triton X-100, and a sandwich ELISA was performed along with total protein quantitation as previously described.37 Flow Injection Analysis. The FIA mobile phase was identical to the E. coli growth medium, except that no glucose, trace elements, tryptone, or yeast extract was present. Cell culture samples were cooled to 0 °C in an ice bath for a minimum of 10 min prior to sample preparation. Unless otherwise indicated, 200 µL of cell suspension were centrifuged at 14 000 rpm for 1 min, the supernatant was discarded, the walls of the tube were carefully dried, and the cells were resuspended in 700 µL of mobile phase by vortexing for 30 s and incubating at ambient temperature (25 °C) for 270 s (total incubation time 5 min). Potassium ferricyanide solution (700 µL, 10.0 mM K3Fe(CN)6, dissolved in mobile phase) was then added, and the sample was briefly vortexed. With the working electrode poised at +0.80 V vs Ag/AgCl, three injections of this mixture were made, at 120-s intervals. Injections of standard 50 or 100 µM ferrocyanide were made before and after the three sample injections, and linearity up to 1 mM ferrocyanide was periodically checked. Under these conditions, the maximum signals corresponded to ferrocyanide concentrations below 500 µM. Chronoamperometry. Background current was allowed to stabilize in 100.0 mL of buffer (identical to the flow injection mobile phase, except for the addition of (unless otherwise indicated) 5 mM ferricyanide) for up to 5 min, and at t ) 0, a 1.00-mL sample of bacterial suspension was added. This suspension resulted from a cell culture sample that had been cooled to 0 °C (>10 min) and centrifuged and the pellet resuspended in buffer. Current was recorded continuously for ∼15 min following addition of the E. coli sample. Calibrations were performed using potassium ferrocyanide over the 0-1 mM concentration range. RESULTS AND DISCUSSION FIA was compared with CA for the determination of ferricyanide reduction rates during the growth of E. coli JM105. Specific rates measured by FIA result from three timed injections occurring within the first 6 min of exposure of chilled, oxygen-depleted cells to ferricyanide, while the CA method allows continuous data acquisition over a longer period, typically 15 min. Figure 1 shows growth curves obtained by (a) FIA and (b) CA for separate shakeflask cultivations performed under identical conditions. The insets show typical raw data used in the calculation of specific rates. (37) Portsmann, T.; Wietschke, R.; Schmechta, H.; Grunow, R.; Pergande, M.; Stachat, S.; Baehr, R. Clin. Chim. Acta 1988, 171, 1-10.

4952 Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

Figure 1. Growth curves of optical density at 600 nm (dashed lines) and specific ferricyanide reduction rate (solid lines) for shake-flask cultures of E. coli JM105 obtained using (a) flow injection analysis and (b) chronoamperometry. The insets show typical raw data used to determine specific reduction rates.

The five injections shown in the inset to Figure 1a, representing two standard ferrocyanide injections bracketing the three timed injections of the E. coli assay mixture, clearly show increased current as a function of injection time. The increase in current with time was used to calculate the rate of increase in ferrocyanide concentration. Reduction rates (µM/s) and the specific reduction rates shown (µmol/(min‚g), normalized for bacterial dry matter) both increase to maximum values in the late exponential phase and then slowly decrease as the stationary phase is entered. Figure 1b shows results of the corresponding experiment using CA to monitor ferrocyanide accumulation as a result of ferricyanide reduction by E. coli JM105. The inset shows typical raw data as a plot of current against time; the slope of this plot is converted to an increase in ferrocyanide concentration with time, using calibration data obtained with known concentrations of ferrocyanide in the same electrochemical cell in the absence of bacterial cells. The CA results also show an increase in specific reduction rates to a maximum in the late exponential phase, followed by a decrease as the culture enters the stationary phase. The maximums of the FIA and CA experiments occur at similar specific ferricyanide reduction rates of 24 and 17 µmol/min‚g, respectively. The shapes of these curves are typical of ferricyanide reduction rate results obtained in all shake-flask experiments, which lack control of pH and dissolved oxygen concentration. Experiments with E. coli HB101 cultivated in shake flasks and monitored using FIA showed growth curves with identical shape, but larger values for specific respiration, with the late-exponential-phase maximum

Figure 2. Growth curves of optical density at 600 nm and specific ferricyanide reduction rate obtained for a batch chemostat cultivation of E. coli HB101 using flow injection analysis.

occurring at a specific ferricyanide reduction rate of 36 ( 4 µmol/ min‚g (n ) 3). The FIA method was used to examine a batch fermentation of E. coli HB101 conducted in a fermenter with pH and dissolved oxygen control. Figure 2 shows specific respiration rates measured with this organism. It can be seen that the shape of the ferricyanide reduction growth curve is identical to that observed for E. coli JM105, with a mid-exponential-phase maximum followed by a decline in specific respiration rate as the stationary phase is entered; however, the magnitude of the maximal specific respiration rate is much higher under these conditions, with a value of 70 µmol/min‚g, compared with the value of 36 obtained with the same organism under shake-flask culture conditions. The 2-fold increase shows that control of pH and dissolved oxygen concentration significantly increases ferricyanide reduction rates with this organism. This type of curve, with a late-exponential-phase maximum, has also been observed for oxygen uptake rates4 and is attributed to the change in protein expression levels that accompany stationary-phase entry. Factors that influence the transition from exponential- to stationary-phase growth include declining sources of carbon, phosphorus, and nitrogen; initially, low concentrations of these nutrients trigger specific responses designed to allow efficient scavenging of the limiting ingredient or to facilitate the use of an alternative carbon source.9 Ultimately, the exhaustion of the medium for an essential nutrient increases the intracellular concentration of σS, the σ subunit for stationary-phase RNA polymerase.9,10 During this transition, the cell increases synthesis of certain enzymes, and pyruvate oxidase, glycerol-3-phosphate dehydrogenase, and cytochrome bd II oxidase are known to be among those preferentially synthesized.10 The genes coding for the two subunits of cyt bd II oxidase are homologous to those coding for cyt bd I oxidase, but their expression is triggered by σS, rather than low-oxygen (microaerophilic) conditions.6,9,38 Continuous cultures of E. coli HB101 were also examined, because they allow control of the growth rate of the culture for comparison of exponential conditions at high growth rates to stationary-phase-like conditions at very slow growth rates. Various respiratory substrates were tested in preliminary experiments in which 5-min incubations with 10 mM substrate immediately (38) Kita, K.; Konishi, K.; Anraku, Y. J. Biol. Chem. 1984, 259, 3375-3381.

Figure 3. Specific ferricyanide reduction rate by FIA as a function of continuous culture growth rate for E. coli HB101, following 5-min incubation in the absence (a) and presence of (b) 10 mM succinate, (c) 10 mM formate, and (d) 10 mM succinate plus 10 mM formate.

preceded ferricyanide addition. The tested substrates included glucose, pyruvate, succinate, formate, glycerol 3-phosphate, proline, lactate, acetate, and malate, using cultures with growth rates of 0.5 h-1 and above; of these, glucose, proline, and glycerol 3-phosphate yielded small increases above the control value obtained with no substrate present (the largest was glucose, with an 8% increase), while succinate and formate increased the control rate by 40 and 51%, respectively. Notably, NADH was not investigated due to anticipated interference with electrochemical measurements. Figure 3 shows a plot of specific ferricyanide reduction rate against growth rate for E. coli HB101, where measurements were made using FIA in the absence and presence of the respiratory substrates succinate and formate. As expected,4 in the absence of these substrates, the specific ferricyanide reduction rate increases with growth rate. Only a minimal effect of respiratory substrate incubation is observed at higher culture growth rates, with little difference seen between samples reconstituted in buffer containing no additive (curve a), 10 mM succinate (curve b), and 10 mM formate (curve c). Differences are readily apparent at lower growth rates, however, with values obtained in assays involving succinate and formate increasing as the growth rate is decreased. These results show that significant changes in protein expression are triggered when growth rates for this organism are below 0.3 h-1. A second continuous culture of E. coli HB101, with µ ) 0.30 h-1, was used to examine the concentration dependence of the ferricyanide reduction rates obtained in the presence of formate and succinate. The results (Table 1) show that specific ferricyanide reduction rates are concentration-dependent, since the small increases over control samples that are observed with 1 mM extracellular succinate or formate increase substantially for concentrations 10- and 100-fold higher. It should be noted that the intracellular concentrations of these compounds following the 5-min incubation period are not known. To further investigate the difference between exponential- and stationary-phase behavior, a separate experiment was undertaken in which E. coli HB101 was cultured in a fermenter under pHand oxygen-controlled conditions as above, and then a slow (µ) 0.08 h-1) continuous culture was established. Samples of this Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

4953

Table 1. Specific Ferricyanide Reduction Rates for E. coli HB101 (µ ) 0.30 h-1) as a Function of Respiratory Substrate Concentration substrate none formate succinate

a

concn, mM

specific ferricyanide reduction rate,a µmol/(min‚g)

1.00 10.0 100 1.00 10.0 100

43 53 59 68 52 59 67

Average of at least two measurements, RSD < 8%.

culture (3 × 15 mL) were diluted 1:1 in shake flasks with fresh medium containing no glucose, and after 3-h incubation at 37 °C and 200 rpm, samples were collected and chilled. FIA of samples resuspended in buffer, in the absence of respiratory substrates, yielded a specific ferricyanide reduction rate of 10 ( 3 µmol/ min‚g. A 5-min incubation with 10 mM formate immediately prior to ferricyanide addition increased this value to 68 ( 7, while incubation with 10 mM succinate yielded a specific reduction rate of 90 ( 7 µmol/min‚g. These results, along with those shown in Figure 3, clearly indicate that it is not the inherent respiratory activity of the cells that is measured when succinate or formate is added to slowly growing or stationary-phase cells. Ferricyanide reduction kinetics must depend on substrate uptake rates as well as the concentrations and intrinsic kinetic properties of the respiratory enzymes. Enzyme expression levels in E. coli have been shown to shift dramatically as culture growth rates slow to stationary-phase subsistence levels.9,10 For example, the levels of pyruvate oxidase, glycerol-3-phosphate dehydrogenase, and cytochrome bd II oxidase are known to increase, while the level of succinate dehydrogenase decreases due to lower expression levels of the catalytic A subunit.39 It is also becoming clear that the membrane composition also changes, to allow the cells to use any available nutrients and to protect them from diverse stress conditions. The expression of the aerobic C4dicarboxylic acid transporter (dctA) that is responsible for succinate uptake across the cytoplasmic (inner) membrane was recently shown to increase 19-fold in the stationary phase.40 A specific bidirectional transporter for formate, called focA, is also known to exist in the cytoplasmic membrane of aerobic E. coli,41 although its expression levels have not yet been investigated as a function of growth rate; the anaerobic formate transporter (focB) is known to be induced during nitrogen limitation.42 The major anion-selective porin in E. coli, called phoE,43 is induced by phosphate limitation conditions,44 while the main cation-selective porins (ompF and ompC), which are better understood, are strictly regulated by nutrient limitation (growth rate),45 osmolarity,46 and (39) Xu, J.; Johnson, R. C. J. Bacteriol. 1995, 177, 938-947. (40) Davies, S. J.; Golby, P.; Omrani, D.; Broad, S. A.; Harrington, V. L.; Guest, J. R.; Kelly, D. J.; Andrews, S. C. J. Bacteriol. 1999, 181, 5624-5635. (41) Kaiser, M.; Sawers, G. Microbiology 1997, 143, 775-783. (42) Andrews, S. C.; Berks, B. C.; McClay, J.; Ambler, A.; Quail, M. A.; Golby, P.; Guest, J. R. Microbiology 1997, 143, 3633-3647. (43) Benz, R.; Darveau, R. P.; Hancock, R. E. Eur. J. Biochem. 1984, 140, 319324. (44) Benz, R.; Schmid, A.; Hancock, R. E. J. Bacteriol. 1985, 162, 722-727. (45) Liu, X.; Ferenci, T. J. Bacteriol. 1998, 180, 3917-3922.

4954

Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

temperature.47 Regulators for nutrient limitation and stationaryphase gene expression include cyclic AMP, which increases as growth rates decrease,48 and the stationary-phase σ subunit of RNA polymerase, which is induced at growth rates of 0.1 h-1 and below, for several strains of E. coli.49 Modeling the uptake and enzyme kinetics for ferricyanide reduction in the presence of succinate and formate was not attempted due to the complex series of transport and catalytic steps. On the basis of the results shown in Figure 3, however, we propose that the rate-limiting steps involve electron transfer to ferricyanide for rapidly growing cells (because substrate addition has little effect) and substrate uptake for slowly growing cells. Previous studies have found evidence that terminal oxygen reduction is the rate-limiting step in the exponential growth of batch cultures,4,5 in agreement with our ferricyanide reduction results at high continuous cultivation rates. High transporter expression levels in slowly growing or stationary-phase cells40,42 suggest that substrate uptake is rate-limiting under these conditions, in agreement with our results at low cultivation rates (Figure 3). An attempt at kinetic modeling has recently been undertaken for late-exponential-phase Gluconobacter industrius (in the presence of glucose) and Pseudomonas fluorescens (in the presence of nicotinic acid) trapped on the surface of a glassy carbon electrode, using the oxidants DCIP, benzoquinone, or molecular oxygen with both organisms as well as ferricyanide with P. fluorescens.19 In this study, kinetic constants were defined in relation to cell concentration rather than the concentrations of specific enzymes and transporters; this resulted in a method that allows comparison between organisms. With E. coli JM105 in either the exponential or stationary phase, the addition of 10 mM succinate to the CA assay solution was necessary to observe apparent Michaelis-Menten kinetics with respect to ferricyanide concentration. Eadie-Hofstee plots for late-exponential-phase (OD600 ) 3.1) and stationary-phase (OD600 ) 4.3) shake-flask cultures are shown in Figure 4, and apparent Km values of 10.1 ( 0.6 and 14.4 ( 1.2 mM were obtained for ferricyanide, respectively. The apparent Vmax value for lateexponential-phase cells (138 ( 4 µmol/min‚g) is slightly higher than the value of 127 ( 5 µmol/min‚g obtained for stationary phase cells. These values are 7-8-fold lower than the value of 17 µM/s (normalized to 1 mg/mL dry matter, which is 1020 µmol/ min‚g) reported for ferricyanide reduction by P. fluorescens in the presence of 10 mM nicotinic acid; with this organism, the apparent Km for ferricyanide was 7.2 ( 1.5 mM.19 Interestingly, with P. fluorescens, which was grown in a medium containing nicotinic acid, 2-3-fold higher Vmax values were obtained when benzoquinone, DCIP, or molecular oxygen was used as oxidant instead of ferricyanide.19 Using a fixed initial ferricyanide concentration of 50 mM, various hydrophobic mediators were tested at low concentration (5 µM) with exponential-phase E. coli JM105 by CA in an attempt to increase the specific ferricyanide reduction rates measured. No (46) Hall, M. N.; Silhavy, T. J. J. Mol. Biol. 1981, 146, 23-43. (47) Lugtenberg, B.; Peters, R.; Bernheimer, H.; Berendsen, W. Mol. Gen. Genet. 1976, 147, 251-262. (48) Notley-McRobb, L.; Death, A.; Ferenci, T. Microbiology 1997, 143, 19091918. (49) Notley, L.; Ferenci, T. J. Bacteriol. 1996, 178, 1465-1468.

Figure 4. Specific ferricyanide reduction rate by CA as a function of ferricyanide concentration, for E. coli JM105 grown in shake-flask cultures to (a) late-exponential phase (OD600 ) 3.1) and (b) stationary phase (OD600 ) 4.1). The insets show the Eadie-Hofstee plots used to determine apparent Michaelis-Menten constants.

significant increases in specific ferricyanide reduction rates were observed when anthraquinone-1-sulfonate, DCIP, or benzoquinone was added to the assay solution, but small (17-22%) increases were observed with 2,3-dimethoxy-5-methylbenzoquinone and Janus Green (E° ′(SHE) ) +0.013 and +0.014 V, respectively, determined by cyclic voltammetry). These hydrophobic compounds are expected to be present at much higher concentrations within the membranes and may facilitate ferricyanide reduction by bypassing slower, natural electron-transfer pathways. Cyanide ion is a known inhibitor of the terminal cytochrome oxidases, with reported Ki values for 50% inhibition of 10 µM for cyt o oxidase and 2 mM for cyt bd I oxidase.38 The inhibition of ferricyanide reduction by cyanide ion was studied using E. coli JM105 and CA, since this method allows continuous data acquisition over long times. Figure 5 shows CA traces obtained with midexponential-phase cells in the absence (a) and presence (b) of 50 M KCN. Traces obtained with 50 µM (Figure 5b) and 6 mM KCN (data not shown) exhibit an initial burst of ferricyanide reduction when cells are added to the assay mixture, but after ∼3 min, the rate of ferricyanide reduction is significantly slower, as seen by the early curvature in the current-time trace in Figure 5b. This is consistent with the known binding of cyanide to the fully oxidized and partially reduced forms of the oxidases.38,50,51 Immediately prior to addition of the cell suspension to the CA assay (50) Pudek, M. R.; Bragg, P. D. Arch. Biochem. Biophys. 1974, 164, 682-693.

Figure 5. Chronoamperometric traces for exponential-phase E. coli JM105 obtained in the (a) absence and (b) presence of 50 µM KCN. The curvature observed in (b) indicates inhibition that takes place as turnover occurs (see text).

mixture, the cytochrome oxidase complexes are present in the fully reduced form, since the suspension is depleted of oxygen. Once respiratory turnover begins, cyanide begins to bind and inhibit turnover. The CA traces become linear after the first 2-3 min of exposure to cyanide, and measurements made between 5 and 10 min yield 16 (n ) 3) and 14% (n ) 4) remaining ferricyanide reductase activity with 50 µM and 6 mM cyanide, respectively. The similarity of these values at such different cyanide concentrations suggests that the main terminal oxidase present in these exponential-phase cells is the more cyanidesensitive cytochrome o oxidase, consistent with literature showing that the two cyt bd species are not detectable in exponential-phase cells50 Ferricyanide thus accepts electrons mainly from cytochrome o oxidase in exponentially growing cells, while the 14% ferricyanide reductase activity remaining in 6 mM cyanide must result from reactions with other components of the respiratory chain. A previous study that examined only the first 30 s of turnover in the presence of cyanide erroneously concluded that ferricyanide mediation occurs before the terminal oxidases in the respiratory pathway, since cyanide inhibition was not yet apparent.12 These results encouraged an investigation of the effect of deaeration on ferricyanide reduction rates, since the Km values of the terminal oxidases for O2 are quite low (2.9 and 0.38 µM, for cyt o oxidase and cyt bd I oxidase, respectively).38 However, with (51) Tsubaki, M.; Mogi, T.; Hori, H.; Sato-Watanabe, M.; Anraku, Y. J. Biol. Chem. 1996, 271, 4017-4022.

Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

4955

HMS174(DE3)(pET3ahSOD), which is engineered to produce recombinant human Cu/Zn superoxide dismutase. This organism was grown to exponential phase (1.5 g/L), and SOD expression was then induced by the addition of IPTG to the fermenter. Figure 6 shows the results of the FIA assay performed in the absence and presence of a succinate/formate mixture along with SOD activity and bacterial dry matter determinations. It can be seen that biomass becomes constant 2 h after induction at a value of ∼2.2 g/L, while a steady increase in SOD activity occurs, with 70 mg of SOD/g of total protein observed 5 h after induction. Specific ferricyanide reduction rates show steadily decreasing values from 1 to 4 h after induction, but then a divergence occurs at 5 h, indicating that the cells have initiated the heat-shock/stationaryphase entry response that ultimately causes ribosome destruction and cell death.11,33-35 The high specific SOD activity present at this time suggests that, under these culture conditions, optimal harvesting of SOD would take place 4 h after induction.

Figure 6. Recombinant production of human Cu/Zn superoxide dismutase by E. coli HMS174(DE3)(pET3ahSOD) as a function of time following induction: (a) specific ferricyanide reduction rate (FIA) obtained in the (i) absence and (ii) presence of 10 mM succinate and 10 mM formate; (b) SOD content per gram of total protein, from ELISA, and bacterial dry matter values obtained for samples taken during this cultivation.

50 mM ferricyanide present as an external oxidant, deaeration and blanketing of the CA assay solution with N2 had no measurable effect on the specific ferricyanide reduction rates of exponentialphase E. coli JM105 (three replicate measurements on each of two cultures with OD600 values of 3.01 and 3.11). Analytical measurements of cellular ferricyanide reduction rates during the cultivation of native or recombinant E. coli can be used to indicate the timing of stationary-phase entry that initiates the process of ribosome destruction and cell death11,33-35 and can thus indicate the optimal harvest time. Rates measured in the presence of added succinate or formate are much higher than those obtained in their absence when cells enter the stationary phase, and the difference in these values provides information that is not accessible using standard oxygen uptake measurements. To demonstrate its utility for indicating stationary-phase entry, the FIA assay was applied to a batch fermentation of E. coli strain

4956 Analytical Chemistry, Vol. 72, No. 20, October 15, 2000

CONCLUSIONS In the presence of the respiratory substrates succinate and formate, the rate of reduction of ferricyanide by viable E. coli is a fundamentally different measure of metabolic state than the rate of oxygen consumption. Cyanide inhibition studies showed that ferricyanide is reduced by the terminal cytochrome oxidase, and the rate of its reduction is influenced by the relative expression levels of transport proteins and respiratory enzymes. Despite this complexity, apparent Michaelis-Menten kinetics with respect to ferricyanide concentration were observed when the assay solution contained g10 mM succinate. Assays of continuous cultures showed that very slowly growing E. coli HB101 cells possess markedly higher ferricyanide reduction rates following a brief incubation of cell samples with succinate or formate, compared to rates observed when these substrates were absent. The difference in ferricyanide reduction rates in the absence and presence of these substrates was used to indicate stationary-phase entry in a recombinant batch fermentation, where overexpression of recombinant human Cu/Zn superoxide dismutase caused stationary-phase entry behavior 5 h after induction. The proposed method can thus be used as a tool to establish the optimal time for harvesting recombinant proteins. ACKNOWLEDGMENT Funding from the Natural Sciences and Engineering Research Council of Canada, the University of Waterloo and Bender & Co. GmbH is gratefully acknowledged. P.E. thanks the Austrian Ministry of Science for an International Scholarship.

Received for review March 27, 2000. Accepted August 9, 2000. AC000358D