generating controlled reducing environments in aerobic...

Generating Controlled ReducingEnvironments in Aerobic RecombinantEscherichia coli Fermentations: Effectson Cell Growth, Oxygen Uptake, HeatShock Protein Expression, and In VivoCAT Activity

Ryan T. Gill,1 Hyung Joon Cha,1 Alok Jain,1 Govind Rao,2

William E. Bentley1

1Center for Agricultural Biotechnology, University of MarylandBiotechnology Institute and Department of Chemical Engineering,University of Maryland, College Park, Maryland 20742; telephone:301-405-4321; fax: 301-314-9126; e-mail: [email protected] Biotechnology Center, University of Maryland BiotechnologyInstitute and Department of Chemical and Biochemical Engineering,University of Maryland Baltimore County, Baltimore, Maryland 21228

Received 2 August 1996; accepted 26 November 1997

Abstract: The independent control of culture redox po-tential (CRP) by the regulated addition of a reducingagent, dithiothreitol (DTT) was demonstrated in aeratedrecombinant Escherichia coli fermentations. Moderatelevels of DTT addition resulted in minimal changes tospecific oxygen uptake, growth rate, and dissolved oxy-gen. Excessive levels of DTT addition were toxic to thecells resulting in cessation of growth. Chloramphenicolacetyltransferase (CAT) activity (nmoles/µg total proteinmin.) decreased in batch fermentation experiments withrespect to increasing levels of DTT addition. To furtherinvestigate the mechanisms affecting CAT activity, ex-periments were performed to assay heat shock proteinexpression and specific CAT activity (nmoles/µg CATmin.). Expression of such molecular chaperones asGroEL and DnaK were found to increase after addition ofDTT. Additionally, sigma factor 32 (s32) and several pro-teases were seen to increase dramatically during addi-tion of DTT. Specific CAT activity (nmoles/µg CAT min.)varied greatly as DTT was added, however, a minimumin activity was found at the highest level of DTT additionin E. coli strains RR1 [pBR329] and JM105 [pROEX-CAT].In conjunction, cellular stress was found to reach a maxi-mum at the same levels of DTT. Although DTT additionhas the potential for directly affecting intracellular pro-tein folding, the effects felt from the increased stresswithin the cell are likely the dominant effector. That theeffects of DTT were measured within the cytoplasm ofthe cell suggests that the periplasmic redox potentialwas also altered. The changes in specific CAT activity,molecular chaperones, and other heat shock proteins, inthe presence of minimal growth rate and oxygen uptakealterations, suggest that the ex vivo control of redox po-tential provides a new process for affecting the yield and

conformation of heterologous proteins in aerated E. colifermentations. © 1998 John Wiley & Sons, Inc. BiotechnolBioeng 59: 248–259, 1998.Keywords: Escherichia coli; Chloramphenicol Acetyl-transferase (CAT); Culture Redox Potential (CRP); Dithio-threitol (DTT); reducing agents, molecular chaperones;proteases; heat shock; stress response; protein folding

INTRODUCTION

The expression of eukaryotic proteins inEscherichia colihas been problematic, particularly when properly foldednon-aggregated proteins with disulfide bonds are desired.The highly reducing environment of theE. coli cytoplasm isuntenable for disulfide bond formation. Specifically, thethioredoxin and glutaredoxin systems are responsible for thereduction of disulfide bonds in the cytoplasm and mutationswithin these systems allow proper disulfide bond formation(Derman et al., 1993; Prinz et al., 1997). However, cyto-plasmic folding events are essential as evidenced by thecytoplasmic molecular chaperones, GroEL and DnaK,which are critical for protein folding. The periplasmic spaceis more oxidizing and has been targeted by many research-ers for ensuring disulfide bond formation (Georgiou andBowden, 1990; Walker and Gilbert, 1994). Note, however,the redox potential and the folding environment in theE.coli periplasm is substantially different than in the eukary-otic endoplasmic reticulum (ER), which has been cited as aprincipal reason for low yields (Walker and Gilbert, 1994).In the E. coli periplasm, disulfide bond formation is cata-lyzed oxidatively by DsbA, which, in turn, is regenerated byDsbB (Kamitani et al., 1992). Coexpression of DsbA doesfacilitate higher yield in some cases, although there is nogeneral trend (Wunderlich and Glockshuber, 1993). The ad-

Correspondence to:William E. BentleyContract grant sponsors: National Science FoundationContract grant numbers: BCS-9010756; BES-9319366

© 1998 John Wiley & Sons, Inc. CCC 0006-3592/98/020248-12

dition of reducing agents (e.g., glutathione,N-acetylcys-teine) to lower the periplasmic redox potential has beenpartially successful in shake flasks (Wunderlich and Glock-shuber, 1993), however, a high level (20 mM) of the reduc-ing agent Dithiothreitol (DTT) is lethal (Missiakas et al.,1993).

The heat shock stimulon defines a class of proteins whoseexpression is activated by heat and whose presence en-hances the survival ofE. coli at high temperature (Chuangand Blattner, 1993; Yamamori et al., 1978). Sigma factor 32(s32) controls expression of most of the heat shock genesand mutations in this factor result in decreased expression ofheat shock proteins and diminished proteolysis of abnormalproteins (Baker and Grossman, 1984). Cytoplasmic molecu-lar chaperones, including heat shock protein families 70(hsp70) and 60 (hsp60), are known to facilitate protein as-sembly, disassembly, and folding (Parsell and Lindquist,1993). InE. coli, DnaK (hsp70) and GroEL (hsp60) act aschaperones during protein assembly but are also known topromote proteolysis of abnormal proteins (Martin et al.,1991; Strauss et al., 1988). DnaK is typically referred to asan unfoldase which at low levels, stabilizes proteins. Also,DnaK has been implicated as a feedback control regulator ofsigma 32 (s32) activity via formation of inactive complexes(Gamer et al., 1996) and eventual degradation by the mem-brane bound protease FtsH/HflB (Tomoyasu et al., 1995). Inaddition, the heat shock proteases Clp and Lon are respon-sible for much of the cytoplasmic degradation of abnormalproteins (Maurizi, 1992). A wide variety of additional en-vironmental stimuli are known to promote heat shock pro-tein synthesis, including; anoxia, amino acid starvation, UVlight, abnormal recombinant protein appearance, viral infec-tion, chemical shock, and reactive oxygen species amongothers (Ashburner and Bonner, 1979; Goff and Goldberg,1987; Neidhardt et al., 1984; Plesset et al., 1982). The pro-duction of such heat shock proteins indicates the relativestress level as well as the protein folding and unfoldingenvironment within the cell.

Theculture redox potential (CRP) of a bacterial fermen-tation is a numerical measure of its oxidizing or reducingnature and has been defined as the ‘‘activity’’ of the elec-trons in the medium (Kjaergaard and Jorgensen, 1979). Thecellular redox state is monitored by such sensor systems asFnr, NarX/NarL, and ArcB/ArcA (Iuchi and Lin, 1993).Each of these sensors is an important factor in the regulationof respiration, for example, activation of the ArcB/ArcAsystem due to oxygen limitations results in the inhibition ofaerobic enzymes (Iuchi et al., 1988). In addition,cultureredox potential gradients were shown to elicit taxis inE.coli, providing evidence of the importance of redox sensingtowards optimal growth and energy production (Bespalov etal., 1996). Several active redox reactions in the fermentationbroth influence the CRP level, however, the water/oxygenredox couple is predominant in highly aerobic environments(Kwong and Rao, 1992a; Srinivas et al., 1988). Correspond-ingly, the measured CRP cannot be assigned to a particular

intracellular or extracellular redox reaction (Srinivas et al.,1988). As a result, CRP has historically been utilized inmicroaerobic and anaerobic cultures where independentoxygen detection is difficult (Akashi et al., 1978; Srinivas etal., 1988). Recent studies, however, indicate CRP is usefulfor monitoring changes in the metabolic state of aerobiccultures (Kwong and Rao, 1991) and for optimizing yield ofaerobic and anaerobic fermentation products (Kwong andRao, 1992b).

It is well known that addition of dithiothreitol (DTT, amembrane-permeating reducing agent) generates an intra-cellular reducing environment that affects protein folding,in particular disulfide bond formation, in eukaryotic cells(Gelman and Prives, 1996; Lodish and Kong, 1993; Sawyeret al., 1994). In addition, at levels greater than 20 mM, DTThas been shown to be lethal toE. coli (Missiakas et al.,1993). However, the effects of DTT on protein folding inaerobicE. coli fermentations have not appeared. This articleexamines the effects of DTT addition on the activity ofchloramphenicol acetyl transferase (CAT), a cytoplasmi-cally expressed model protein. Moreover, the independentcontrol of the CRP in highly aerobic cultures is demon-strated. Depending upon the operation, very little responsein specific growth rate and specific oxygen uptake was ob-served. There was no significant impact on acetate produc-tion. Additional measures including the relative levelssigma 32 (s32), GroEL, DnaK, ClpB, and other proteasesindicated a significantly altered intracellular environmentdue to DTT addition. Importantly, specific activity of amodel heterologous cytoplasmic protein, CAT, was stronglyaffected. It is hypothesized that a controlled culture redoxpotential can facilitate the intracellular folding apparatussuch that expression of properly folded eukaryotic proteinsis achieved inE. coli.

MATERIALS AND METHODS

Microorganisms

Escherichia coliRR1 (F-supE44 lacY1 ara-14 gal K2 xtk-5ntk-1 leuB6 proA2 hsd20 (r−m−) rpsL20 thi-1) and JM105(Dlac-pro thi strA endA sbcB15 hspR4 F8 tra36 proAB+

lacIq-ZDM15) bearing plasmids [pBR329] and [pROEX-CAT], respectively, were used for all fermentations exceptas noted. Three additional strains were used in this work as‘‘stress probes’’ (Cha et al., 1998).E. coli JM105 was thehost for these three transcriptional fusion plasmids whereinthe gene for the green fluorescent protein (GFP) is fused inframe with the promoter/operator regions of the heat shockproteins: s32 (P3), ClpB, and DnaK (P2), respectively.Thus, the plasmid constructs, denoted [pGFPuv-Sigma],[pGFPuv-ClpB], and [pGFPuv-DnaK], contain heatshock stress protein promoters fors32, ClpB, and DnaK,respectively, and these promoters drive expression of theGFPuv structural gene upon conditions that endogenouss32, ClpB, and DnaK would be expressed. These plasmids

GILL ET AL.: CONTROLLED REDOX IN E. COLI CULTURES, EFFECTS ON YIELD 249

are referred to as ‘‘stress probes’’ in this analysis becausethe fluorescence is easily measured within the cells, therebyserving as a ‘‘probe’’ for cellular stress. Transformations ofthese strains and construction of stress probe plasmids isdescribed elsewhere (Cha et al, 1998).

Fermentations and Media

All E. coli RR1 [pBR-329] experiments employed M9 mini-mal media supplemented withL-Leucine,L-Proline, vitaminB1 and ampicillin for final concentrations of 41, 164, 0.166,and 0.2 mg/mL (Rodriguez and Tait, 1983) All JM105[pROEX-CAT] experiments were performed in M9 mini-mal media supplemented with vitamin B1 and ampicillin forfinal concentrations of 0.166 and 0.05 mg/mL respectively.Sterile media were prepared using autoclaved salt solutionsand filtered glucose, amino acids, and antibiotics. The glu-cose concentration was between 6–12 g/L, as specified.

E. coli RR1 [pBR329] and JM105 [pPROEX-CAT]shake flasks were performed in 250 mL Erlenmeyer flasks(100 mL working volume) at 37°C in either a reciprocatingwater bath shaker at 250 rpm (New Brunswick Scientific),or in an air shaker at 200 rpm (New Brunswick Scientific).1.5 mL of frozenE. coli was added to M9 media (Rodriguezand Tait, 1983), and cultivated overnight at 30°C to anoptical density (@600 nm) of 0.7. A 5% innoculum wastransferred to shake flasks containing 37°C M9 minimalmedia and grown to mid-exponential phase (ODJM105 40.7, ODRR1 4 0.25) at which point DTT was added atvarious levels. Cells were harvested 20 min post-DTT ad-dition.

Stress probe cells were grown to early log phase at 30°Cin 100 mL of LB medium containing 0.05 mg/mL ampicil-lin (Sigma Chemical Co., St. Louis, MO) from 30°C over-night cultures in the same medium. After harvesting, cellswere washed with phosphate buffered saline (PBS; 1.44 g/LNa2HPO4, 0.24 g/L KH2PO4, 0.2 g/L KCl, and 8 g/L NaCl,pH 7.4). Cells were resuspended in 20 mL PBS bufferand were in a ‘‘resting’’ cell condition (Cha et al., 1997).Fluorescence data are reported relative to the control plas-mid not containing the GFP gene. The relative fluorescentintensity was calculated as (RFI/OD)4 RFIsample/ODsample

− RFIcontrol/ODcontrol. Normalization relative to the stressprobe plasmids in an unstressed condition was equivalent(Cha et al., 1998).

E. coli RR1 [pBR329] batch fermentations were per-formed in a 3 L Applikon fermentor with a 2 L workingvolume and sparged at 1 vvm. One mL of frozenE. coliRR1[pBR329] was added to LB media (5 g/L yeast extract(Sigma Chemical Co.), 10 g/L bactotryptone (Difco Lab,Detroit, MI.), and 10 g/L NaCl) (Rodriguez and Tait, 1983),and cultivated overnight. A 2% inoculum was then trans-ferred to another shake flask with M9 medium and the cellswere grown overnight to an optical density of about 0.6(@600nm). This procedure was repeated twice, then a 20–40 mL inoculum was transferred to the fermentor with thesame M9 medium. In the fermentor, the pH and the tem-

perature were controlled (pH 7.0 and 37°C) by a ApplikonBiocontroller ADI 1030. The agitation rate was maintainedat 600 rpm by an Applikon controller ADI 1012, exceptwhen controlled by an external signal. The pH was mea-sured by a glass pH electrode (Ingold) and was controlledby addition of 5N NaOH and 2N HCl. The dissolved oxygen(DO) was measured by a polarographic probe (Ingold, Wil-mington, MA). The DO probe was polarized for 6 h beforeeach fermentation and was calibrated by saturating the me-dium overnight with air. The CRP was measured using aplatinum electrode (Phoenix, Houston, TX). The CRP read-ings were reported with respect to the standard hydrogenelectrode; the probe was calibrated before each run usingtwo redox standards (0.22 V and 0.46 V).

Dithiothreitol (DTT, Sigma) was prepared in bottles(with deoxygenated water) and kept under nitrogen atmo-sphere during all experiments. A signal from a micropro-cessor was used to control the addition of DTT according tothe prevailing culture redox potential in the fermentor. Inthis way, the CRP was maintained at a specific level bycontroling the addition of the reducing agent.

Data Acquisition and Control System

The data acquisition system consisted of a terminal board(#DT707, Data Translation, Inc., Marlboro, MA), an ana-log/digital card (#DT1028, DT), a PC/XT (IBM), and Note-book software (Labtech Inc., Wilmington, MA). This con-figuration was used to monitor and control both the CRPand the DO. Analog signals were inputs to the A/D cardusing a terminal board interface and Notebook software wasprogrammed to monitor and control the two parameters.Analog signals were sent out through the board to the agi-tation rate controller ADI 1012 for DO control and to thepump connecting the reducing agent tank to the fermentorfor the CRP control (see Fig. 1). Unless otherwise noted, theCRP and DO levels were maintained at −0.2 to −0.05 V and10% (air sat.), respectively. Data was recorded every 30 sec.The control parameters for the system (PI controller) wereobtained by trial and error. The complete data acquisitionand control system is shown schematically in Figure 1.

Analytical

Optical Density (OD) was measured at 600 nm (Milton RoySpec. 21DV). Samples were diluted with deionized water toobtain OD in the linear range (0–0.25 OD units). The OD ofthe sterile medium was used as a correction for all readings.Samples (6 mL) were extracted from the fermentor every1–2 h. The cells were harvested by centrifugation at 12krpm for 15 min. (Eppendorf 5415). After separating thesupernatant (which was stored for further analyses), cell pel-lets were washed (50 mM Tris, pH4 7.5) twice, disrupted,and stored at −20°C until assayed for enzyme activity andtotal protein (Bentley et al., 1990). Chloramphenicol acet-yltransferase (CAT) and protein assays were performed us-ing a UV/VIS spectrophotometer (Rodriquez and Tait,

250 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 59, NO. 2, JULY 20, 1998

1983). The CAT assay was performed at 37°C; sampleswith high activity were diluted with deionized water to ob-tain readings in the linear range. The protein assay wasperformed using BIO-RAD protein assay dye reagent. CATactivity (nmoles/mg total protein − min) and specific CATactivity (nmoles/mg CAT − min) were calculated, respec-tively, by dividing the CAT activity on a volume basis(nmoles/mL − min) by the total protein concentration (mg/mL) or by the total CAT concentration (mg/mL) as obtainedby Western Blot analysis. Acetate assays were performedusing an enzymatic analysis kit (Boehringer Mannheim#148261, Indianapolis, IN). All assay results are reported asaverages of repeated measurements with standard error asindicated.

Protease activity was assayed by SDS-GPAGE (sodiumdodecyl sulfate-gelatin polyacrylamide gel electrophoresis)(Harcum and Bentley, 1993). The SDS-GPAGE techniquedifferentiates proteases by molecular weight and activity.Gels were cast at 12.5% (w/v) polyacrylamide and 0.1%(w/v) gelatin and run at 200 V for approximately 1 h at4°C.The sample preparations do not containb-mercaptoethanolor DTT because nondenatured proteases are required afterelectrophoresis. Purified CAT, gelatin, and glycerol wereobtained from Sigma. Tris-Base, acrylamide,N-N8-Methylene-bis-Acrylamide, sodium dodecyl sulfate, ammo-nium persulfate, glycine, molecular weight markers, bromo-phenol blue, andN,N,N,N1-tetramethyl ethylenediamine(TEMED) were obtained from BIO-RAD. Equal amounts oftotal protein (15mg) were loaded into each well. Substrategels were washed in 2.5% (v/v) Triton X-100 for 1 h atroom temperature on an orbital shaker to remove SDS andwere rinsed twice in deionized water prior to incubation.The gels were incubated at 37°C for 24 h in an incubationbuffer (0.1M Glycine, 2 mM ATP, 2 mM MgCl2, pH 4 7.5)

then stained with 0.2% amido black in water:methanol:ace-tic acid (20:5:2) for 1 h and destained in water:methano-l:acetic acid (47:5:2) for 2 h to reveal clear zones of pro-teolytic activity. Gels were immediately imaged (Eagle EyeII, Stratagene, La Jolla, CA) and proteins were quantifiedwith N I.H. Image analysis software. Differences in theclear zones, measured by size and intensity, reflects thequantity of active protease. Again, protease amounts wereaverages of multiple readings and error limits are as indi-cated.

Western Blot protein analyses were performed followingTowbin et al. (1979). Primary antibodies were obtained forGroEL (Stressgen, British Columbia, Canada), DnaK(Stressgen, Canada), and Chloramphenicol Acetyltransfer-ase (5 prime—3 prime, Boulder, CO.). Secondary antibod-ies (Alkaline phosphatase labeled IgG) were obtained fromKirkegaard and Perry, Inc., Gaithersburg, MD. SDS-PAGEgels were cast in a Mini-Protean II 1-D apparatus (BIO-RAD). The gel (12.5% Polyacrylamide) was run at 200 Vand 4°C for∼1 h. Preparations of the samples were boiledat 100°C for 3–5 min prior to electrophoresis. Equalamounts of total protein, 15mg, were loaded into each well.Reference levels of 0.5mg, 0.5 mg, and 0.75mg, respec-tively, of pure GroEL, DnaK, and CAT were included asstandards. The separated proteins were blotted using aTrans-Blot Semi-Dry Transfer cell (BIO-RAD) onto sup-ported nitrocellulose membrane (BIO-RAD) for 20 min at10 V and 20 min at 20 V. Transferred proteins were imme-diately quenched in Tris-buffered saline (TBS, 20 mM Tris-HCL, 500 mM NaCl, pH4 7.5) and subsequently blocked(5% (w/v) non-fat dry milk in TBS) overnight at 4°C.Blocking solution was removed with TTBS (0.05% (v/v)Tween 20 in TBS) washing and primary antibody (1/2000(w/v) dilution in 1% (w/v) non-fat dry milk in TTBS) wasfixed for 2 h under gentle agitation. Primary antibody wasremoved with multiple washes of TTBS and secondary an-tibody (Alkaline Phosphatase labeled IgG Rabbit antibody1/5000 (w/v) dilution in 1% (w/v) non-fat dry milk inTTBS) was fixed for 2 h under gentle agitation. Secondaryantibody was removed with sequential washes of TTBS andTBS. Substrate (1 BCIP/NBT (Sigma) tablet in 10 mLdeionized water) was added. Proteins were visualized andthe reaction was quenched by rinsing with deionized water.Membranes were immediately imaged (Eagle Eye II) andproteins were quantified with N I.H. Image analysis soft-ware. Protein amounts used for analyses were averages ofduplicate readings. CAT Western Blots produced fourbands at molecular weights of 21 kDa, 25 kDa, 30 kDa, and40 kDa. All results were based on analysis of the 25 kDaband due to its agreement with the molecular weight andmigration proximity of the CAT standard.

RESULTS

Growth Rate and Dissolved Oxygen Levels

Six shake flasks were inoculated identically and at 6.5 h,DTT was added at varying concentration. As seen in Figure

Figure 1. Fermentation system for on-line control of culture redox po-tential (CRP) by the addition of dithiothreitol (DTT). Reducing agent pumpflow rate is regulated according to CRP measurement and setpoint usingproportional-integral (PI) controller. Dissolved oxygen is controlled bychanging agitation rate. pH is regulated by addition of acid and base. DO,pH, and temperature are regulated independently by biocontroller (Appli-kon).


2, cell growth was increasingly affected by the increasingDTT concentration. Shake flasks with 0.25 g/L and 2.0 g/Lwere not significantly different than those with 0.5 g/L and1.0 g/L, respectively (not shown). Below 0.5 g/L there wasminimal effect on cell growth rate and final yield. At higherconcentrations the DTT was toxic, as expected, stunting thegrowth rate.

Figure 3A depicts the cell growth profiles of three aeratedfermentation experiments: run I (control; no DTT addition),run II (CRP controlled to −0.05 V by DTT addition), andrun III (addition of 20 g/L DTT at 12 mL/h). There was littledifference in the growth rate and final OD of the control andCRP-controlled experiments. In run III (high-DTT), cellgrowth was completely stopped after a 5 h lag[accumulated(DTT) 4 0.6 g/L]. Interestingly, the cessation in cellgrowth was relatively sudden when the apparent DTT con-centration was 0.6 g/L. In Figure 3B, the dynamic specificoxygen uptake rate is depicted. Again, there was little dif-ference between runs I and II, the uptake rate decreasedsignificantly upon DTT addition for run III, which ulti-mately stopped growing due to excess DTT. These experi-ments demonstrate that when used moderately, DTT servesas a CRP-controling agent without disrupting the pheno-typic cell growth or oxygen uptake rates of theE. coli.

As noted earlier, CRP is a function of the water/oxygenredox couple and is highly dependent on the dissolved oxy-gen (DO) level (Kjaergaard and Jorgensen, 1979; Srinivas etal., 1988). However, under highly aerobic conditions, addi-tion of a reducing agent (e.g., DTT) affects CRP but not DO(Kwong and Rao, 1992a). Figure 4 depicts the profiles ofDO and CRP for the fermentor experiments shown in Figure3. In the control experiment (Fig. 4A), the CRP and DOprofiles were similar, illustrating the dominating influenceof DO on the CRP measurement. In Figure 4B, the CRP

data for all batch fermentation experiments are depicted. Inrun II, CRP was regulated to −0.05 V by our controllerwhile the DO was unaffected (compare DO traces for twoexperiments, Fig. 4C). In run III, the addition rate of DTTwas 12 mL/h (20 g/L) for∼10 h, which dropped the CRP to−0.28 V, and was toxic to the cells after 5 h. Correspond-ingly, the DO trace mirrored the previous experiments untilthe cessation in cell growth, at which time the DO graduallyrose due to the nonviability of the remaining cells. Acetateproduction was low (<1 g/L) and did not vary significantlyamong these fermentations (not shown).

Chloramphenicol Acetyltransferase Activity

In Figure 5, the CAT activities (nmoles/mg total protein-min) are shown for the same fermentation experiments. Inall cases, CAT activity was reduced by the presence of

Figure 2. Effect on cell growth and yield from DTT addition at 6.5 h insix shake flask cultures: 0, 0.25 (not shown), 0.5, 1.0, 2.0 (not shown), 3.0g/L. The flasks with 0.25 and 2.0 were not significantly different than 0.5and 1.0 g/L experiments, respectively.

Figure 3. (A) Optical density vs. time in aerated recombinantE. coli(RR1 [pBR329]) fermentations. Duration of DTT addition is indicated bythe arrows. Experiments I, II, and III represent the control (no DTT) andCRP-controlled (−0.05 V and −0.28 V) runs, respectively.(B) Specificoxygen uptake rate in same fermentations. Rates are calculated dynami-cally with assumed mass transfer coefficient, kLa (400 h), and saturated DOconcentration (1 mM).


DTT. Decreases in CAT activity as measured by this assaycan result from two factors: a reduction in the fraction ofCAT in a properly folded and active configuration (referredto as configurational effects), or a decline in the expressionof CAT as a fraction of the total protein (referred to asexpression effects). The decrease in activity for run II(∼40%) was significant, while there was an insignificanteffect on cell growth as described earlier. Importantly, therewas no significant difference in total protein content (notshown). In run III, the decrease in CAT activity was likelydue to the effects on both expression (through cell toxicity)and configuration (through improper folding or otherwiseimpaired activity).

Continuous cultures can be more revealing than batchbecause they can be operated at steady state and subse-quently perturbed systematically, generating a response dueentirely to the imposed stress and no other factors (e.g.,varying media composition found in batch cultures). In Fig-ure 6, results of a sterile-feed continuous stirred tank reactor(CSTR) operating at 0.1 h−1 with DO controlled to 10% aredepicted. In this case, DTT (0.1 g/L, 12.6 mL/h [DTT]46.3 mg/L) was added at 46 h and increased to 25.2 mL/h([DTT] 4 12.6 mg/L) at 58 h (second arrow). As in run IIabove, there was no change in cell growth during the periodof DTT addition (indicated by a constant OD in this case).There was a nearly 20% increase in glucose uptake, whichwas similar to earlier work withCorynebacter(Kwong andRao, 1991). Additionally, the dissolved oxygen control wasless accurate in this experiment during the DTT addition, inpart due to the influence of CRP on the DO measurement atlow DO (Kwong and Rao, 1992a). As in batch experiments,specific CAT activity dropped steadily during the period ofDTT addition. Thus, results of the CSTR were very similarto the batch experiments in that there was little change incell growth but a significant change in CAT activity.

Although the batch and continuous culture results dem-

Figure 4. (A) CRP and DO traces in control experiment. (CRP:—;DO:L). Note CRP closely tracks DO curve indicating influence of water/oxygen redox couple.(B) CRP traces for three fermentations noted inFigure 3. Run II was controlled to −0.05 V by DTT addition during periodindicated by arrows.(C) DO traces for same experiments. In run III,increase in DO after 15 h indicates toxicity from excess DTT. In runs I andII, DO curves are similar until stationary phase.

Figure 5. Chloramphenicol acetyltransferase (CAT) activity per mg totalprotein during addition of DTT for the same three experiments. Totalprotein results from runs I and II were similar, run III was strongly dimin-ished by DTT.


onstrated a change in CAT activity, an altered CRP, and thatCRP could be independently controlled in highly aerobiccultures, they did not provide sufficient information to dis-tinguish which of the competing effects (configurational orexpression) controlled the reduction in CAT activity. Toobtain this information, a series of shake flask experimentswere run in which the relative stress level within the cellwas assayed by GFP transcriptional fusion ‘‘stress probe’’plasmids along with the levels of molecular chaperonesGroEL and DnaK, transcription factors32, and several in-tracellular proteases.

Effects of Cellular Stress on Folding and SpecificCAT Activity

Shake flask experiments involving strains RR1 [pBR329]and JM105 [pROEX-CAT], respectively, were inoculatedidentically and at mid-exponential phase, DTT was added atvarying concentration. GroEL and DnaK concentrationswere measured for each sample and quantified by WesternBlotting. Figures 7A and B depict levels of GroEL andDnaK (normalized by the control value, 0 g/L DTT) withrespect to added DTT for strains RR1 [pBR329] and JM105[pROEX-CAT], respectively. The levels of GroEL andDnaK varied dramatically in both strains. For RR1 (upperpanel), DnaK was uniformly greater than the control,whereas GroEL was uniformly less than the control. Inter-estingly, both chaperones were at a minimum level at 1.0g/L DTT. For JM105 [pROEX-CAT] (lower panel), thelevels of both molecular chaperones increased due to DTTaddition, after an initial decrease at 0.5 g/L DTT.

Gelatin PAGE analysis was performed for evaluation andcharacterization of several proteases active at physiologicalpH (≈7.5). Figures 8A and B display the relative levels ofthe three predominant proteases for the same samples. The

major proteases visualized in the GPAGE had molecularweights of approximately 41, 55, and 89 kDa. As was evi-dent, significant metabolic changes occurred with respect toprotease production upon addition of various levels of DTT.In strain RR1 [pBR329], the combined amount of the threepredominant proteases increased with increasing levels ofDTT. At 3 g/L of DTT, the combined protease activity was170% of the control. In strain JM105 [pROEX-CAT], ad-dition of 0.5 g/L DTT resulted in a three fold increase in thetotal protease. As DTT levels increased, total protease con-centration decreased but remained elevated compared tocontrol levels. The increased protease levels associated withDTT addition also suggest a stress response had occurred.

To further investigate the changes in stress levels withadded DTT, shake flask experiments were performed using

Figure 6. Results from CSTR operating at 0.1 h−1 with DTT addedduring periods indicated by arrows. DO and CRP are indicated in upperpanel. Glucose (d), CAT (j), and OD (h) are depicted in the lower panel.During first addition period, DTT (0.1 g/L) flow rate was 12.6 mL/h; lateraddition rate was doubled.

Figure 7. GroEL and DnaK (mgsample/mgcontrol) concentrations and west-ern blots for various levels of DTT addition in shake flask cultures,(A)strain RR1[pBR329],(B) strain JM105 [pROEX-CAT]. Individual chap-erone levels changed with respect to levels of DTT.


E. coli JM105 with plasmids [pGFPuv-Sigma], [pGFPuv-dnaK], and [pGFPuv-clpB] as described in Methods. Toreiterate, the appearance of fluorescence is due to the ex-pression of GFPuv which was placed under the control ofs32 (P3), ClpB, and DnaK (P2) promoters. Figures 9A–Cdisplay the amount of fluorescence relative to a controlplasmid with no GFP. A value of 0 indicates no increase instress.s32 controlled fluorescence reached a maximum 3 hafter DTT addition at 3.0 g/L. Further,s32 promoter activityincreased directly with the level of DTT added. Analysis ofs32 promoter activity over the course of the experimentindicated the strong effect of DTT on elicitation of the heatshock stress response within JM105. DnaK measurements,Figure 9B, indicated a maximum at 2 h after addition ofDTT at a level of 3.0 g/L. Further, the relative quantity ofDnaK promoter-regulated fluorescence increased with ad-dition of DTT. The results for ClpB controlled fluorescenceagreed qualitatively with the protease analysis of the samestrain. At 1 h post-DTT injection, the maximum ClpB ex-pression was with 0.5–1.0 g/L DTT which was similar for

each of the proteases detected in the GPAGE after1⁄2 h. TheClpB transient profile, however, suggested a different de-pendence on DTT. After 4 h post-DTT addition, the maxi-mum ClpB fluorescence activity was found at 3.0 g/L DTTand decreased with decreasing DTT level. In panel D, themaxima are depicted as a function of DTT level. In general,there was an increasing trend in fluorescence maxima withDTT concentration, demonstrating that overall the level ofthe three stress proteins increased with DTT.

Figure 10 depicts the CAT activity quantified three ways,as a function of DTT level. Again, these were from samplestaken from1⁄2 to 1 h after DTT addition to shake flasks. Theactivities are expressed as: (1) CAT activity per mg totalprotein; (2) CAT mass of protein per mg of total protein;and (3) CAT activity per mg of CAT. In the upper panel(RR1), CAT activity (nmoles/mg total protein − min)changed minimally with respect to levels of added DTT,however, the fraction of CAT expressed was significantlylower in the presence of DTT. The ratio of these two quan-tities was denoted ‘‘specific CAT activity’’ (nmoles/mgCAT − min), which was uniformly higher than the control,but decreased with respect to increasing DTT level. Forstrain JM105 [pROEX-CAT] an almost identical specificCAT activity profile was observed, although the location ofthe maxima shifted to a higher DTT level (1 g/L). In bothstrains, there was a strong effect on specific CAT activitywith addition of DTT. With the zero DTT level taken asunity, in both strains there was an increase, to a levelroughly 2 times the control, followed by a decrease with theincreasing DTT level. Interestingly, in all cases except thehighest DTT levels in JM105, the specific CAT activity inthe presence of DTT was higher than the control. This quan-tity was used as an indicator that the specific conformationof CAT, which resides in the cytoplasm, was affected by theaddition of DTT. Overall, the CAT activity per mg totalprotein did not change significantly while the total CATquantity (mass CAT/mg total protein) decreased uniformlywith DTT relative to the control.

DISCUSSION

The expression of eukaryotic proteins inE. coli has beendifficult due to the untenable environment for disulfidebond formation of theE. coli cytoplasm. A variety of tech-niques have been investigated for altering this environmentto enhance protein expression and facilitate active confor-mation. Molecular chaperones define a set of proteins activein protein assembly, disassembly, and folding. Proteolysisaffects the stability of expressed proteins and is elevatedduring stress conditions. The heat shock family of proteins,among many others, includes the molecular chaperonesGroEL and DnaK and proteases Lon and Clp. Many pro-teins in this family are under the control of sigma factor(s32). Addition of DTT to E. coli fermentations was inves-tigated as a method for altering the redox potential in thecell so that the folding environment could potentially bemanipulated by changes to the bioreactor.

Figure 8. Relative intensities (mgsample/mgcontrol) of proteases as deter-mined by GPAGE analysis and GPAGE gels of samples from shake flaskcultures; (A) RR1 [pBR329], (B) JM105 [pROEX-CAT]. The resultsshown are for the most prominent proteases of approximately 41 kDa, 55kDa, and 89 kDa. Protease levels increased dramatically upon addition ofDTT.


A simple experimental configuration that will potentiallyprove useful for regulating the in vivo protein folding en-vironment in situ for aerated recombinantE. coli fermenta-tions was demonstrated. That is, the addition of a CRPprobe, peristaltic pump, and simple PI controller was usedwith minimal phenomenological effect on cell growth,while dramatically affecting the quantity of actively config-ured recombinant protein. Batch fermentations with con-trolled CRP (Fig. 5) indicated a consistent decrease in CATactivity with respect to CRP. This result, however, wasinsufficient for properly interpreting the effects of DTT onthe intracellular folding environment. To be specific, thedecrease in CAT activity was likely the result of eitherdecreased expression (denoted expression effects) or a de-

Figure 9. Time profiles of relative fluorescence (RFIsample/ODsample −RFIcontrol/ODcontrol) for ‘‘stress probe’’ shake flask fermentations usingJM105 [pROEX-CAT] with DTT added at 0.5, 1.0, and 3.0 g/L. Thefigures represent relative levels of(A) s32, (B) DnaK, and (C) ClpB,respectively, as measured by production of GFP. Higher values of RFIrepresent greater levels of stress experienced by the cell. Addition of DTTat 3.0 g/L produced the strongest responses fors32 and DnaK, while lowerlevels elicited greater ClpB activity. These results matched those displayedin Figures 7B and 8B respectively. Figure(D) indicates peak values fors32, DnaK, and ClpB over the entire sample period.s32 and DnaK peakvalues increased with DTT addition.

Figure 10. CAT activity (nmoles/mg total protein − min), CAT (mg/mgtotal protein), and specific CAT activity (nmoles/mg CAT-min) for shakeflasks at different levels of DTT addition(A) RR1 [pBR329],(B) JM105[pROEX-CAT]. Shown below each graph are the respective CAT westernblots.


creased specific activity of the molecule CAT (denoted con-formational effects). Also, degradation of CAT could haveplayed a role. Additional investigation demonstrated thatDTT addition affected both expression and conformation,confounding the issue. In both strains evaluated, a maxi-mum in specific CAT activity (U/mg CAT) was measured atan intermediate DTT level and, in general, the specific CATactivity increased with DTT relative to control. Also, thequantity of CAT apparently decreased with DTT relative tocontrol.

It is understandable that the specific activity of CATwould be altered due to changes in its local redox environ-ment because reactive thiols, histidine, lysine, and othercharged amino acids, as well as interactions regulating theoverall structure (e.g., hydrophilic, hydrophobic) are influ-enced by redox potential (Kwong and Rao, 1992b; Wim-penny, 1969; Wimpenny and Necklen, 1971). Chloram-phenicol acetyltransferase is widely recognized as verystable and quickly folded at expression rates up to c.a. 30%total cell protein (Robben et al., 1993; Shaw, 1983). It is,therefore, one of the most common reporter proteins in pro-karyotic and eukaryotic systems. Although early evidencesuggested cysteine (Cys) involvement at the active site,more recent evidence has demonstrated that critical Cysresidues (e.g., Cys214) influence the positioning of anahelix domain (a), conferring properly folded and activeCAT (Robben et al., 1993). Instead of a cysteine sulfhydryl,a proton associated with Ser148 at the active site stabilizesthe ternary complex (CAT, acetyl-coA, and chlorampheni-col) and modulates the binding of acetyl-coA (Ellis et al.,1995). Notably, the addition of detergents was shown tocompetitively inhibit CAT by disrupting the binding ofchloramphenicol, through the disruption of hydrophobic in-teractions (Lu and Jiang, 1993). Thus, the local redox po-tential has a direct effect on the conformation and activity ofCAT.

Based on the CAT-DTT results, it was hypothesized thatthe cytoplasmic redox potential was altered by the con-trolled addition of DTT. However, measured levels of heatshock proteins also varied with added DTT. Alterations inthe intracellular levels of these proteins will also affect thefolding environment and potentially alter the expression andactivity of CAT. Levels of molecular chaperones GroELand DnaK varied in coordination with each other but did notchange consistently with respect to levels of DTT. The pro-tease activity measured by GPAGE analysis indicated, par-ticularly in JM105, that a significant intracellular proteolyt-ic response resulted due to DTT. Interestingly, JM105 alsoexhibited the most significant increase inspecificCAT ac-tivity. However, we have recently demonstrated that thelower MW protease assayed in the GPAGE gels does nothave appreciable activity towards CAT (Pulliam, 1996). In-deed, none of the proteases depicted on these gels revealedspecificity towards CAT (Pulliam, 1996). Instead, their totalactivity serves as an indicator of cellular stress (Pulliam,1996). The dominant factor influencing overall CAT activ-

ity is most likely the stress response elicited by DTT insteadof a specific alteration of the cytoplasmic reducing potentialor a specific increase in proteolysis.

The severity of the DTT-induced stress response wasconfirmed bys32 fluorescence measurements indicating asharp increase in stress levels as DTT was added. Theseresults suggest that DTT influences the levels of hsps, in-cluding DnaK and GroEL, which in turn modulates thefolding environment and may have increased the specificactivity of CAT. The proposed role of DnaK as a feedbackregulator ofs32 activity (Gamer et al., 1996) was supportedby comparing Figures 9A and B, where thes32 increasedmore rapidly and to a greater extent than DnaK, whichlagged behind thes32. Because there has been no evidenceto suggest that the association of chaperones with CATresults in an improved molecule, our interpretation of theseresults is that the distribution of CAT in active and poten-tially active conformations vs. an inactive conformation wasshifted by the alterations in chaperone and hsp levels.

The interpretation of these results is supported by theintroduction of intracellular dynamics. In the shake flaskwork, the responses occurred within1⁄2 to 1 h of DTT ad-dition. The cellular indicators most dramatically affectedwithin this time frame were the protease levels and the levelof s32, clearly demonstrating a rapid stress response. Thecorresponding levels of GroEL and DnaK did change sig-nificantly but as we observed from the DnaK dynamic result(Fig. 9B), the maximum change in DnaK level occurredseveral hours post-addition. More specifically, the data inFigure 9B indicated an increase in fluorescence intensity of∼0.06 for DnaK within1⁄2 h, but after 2–3 h the intensityincreased by a factor of 4–5. In the case of GroEL, the onlydata obtained were from 1 h post-DTT and the levelchanged in a varied manner. It was clear that GroEL andDnaK were modulated by DTT at a slower rate thans32 andproteases.

When comparing the fermentations with the 1 h post-DTT addition shake flask cultures several observations arenoteworthy. First, in fermentor Run III (toxic DTT levelswere observed), the toxicity became readily apparent after 5h of DTT addition. By calculating the addition rate of DTT,assuming no oxidation, the equivalent DTT concentration inthe reactor was [0.6 g/L] at the time of the cessation ofgrowth. In the shake flasks, the protease activity, GroELlevels, and DnaK levels were all significantly different thanthe 0 g/L DTT control at all levels tested, even at the lowerlimit (0.5 g/L DTT). Therefore, it was understandable thatfermentor Run III had an altered CAT activity. The CATactivity in RR1 shake flasks was not as significantly altered,however, they were sampled only 1/2 h after DTT addition,vs. 2 h for fermentor Run III. It is likely that the shake flaskresults would have diverged significantly from the control atlater times (as was shown by the optical density resultspresented in Fig. 2). In Figure 5, therefore, the pronounceddeviation in CAT activity from the control was most likelydue to the severe stress response and altered expression as


evidenced by the dramatic change in cell growth and ap-parent toxicity.

In fermentor Run II, there was significantly less DTTadded. In this fermentation, the CAT activity over time wasstable and typically 40% less than the control fermentor.Based on our shake flask results at low DTT, in all cases theseverity of responses in ClpB, DnaK, GroEL, and proteaseswas less than at higher DTT. At the lowest levels of DTTaddition, the most significant changes vs. the control oc-curred for DnaK, the mass of CAT, and the specific CAT.Notably, the DnaK and specific CAT activity increasedwhile the level of CAT expressed decreased. Again, thesechanges were observed shortly after DTT addition. Thus, inthe fermentor, we suggest the decrease in total CAT activitywas the combination of a higher specific CAT activity, asindicated by a configurational effect due to redox or chap-erone level, and a lower CAT expression level. Because thetotal activity decreased overall, the decrease in expressionof CAT was dominant. However, that the specific CAT washigher with DTT suggests the ‘‘quality’’ of a recombinantprotein could be improved, but expressed at a lower level,by adopting this CRP control methodology.

These conclusions are corroborated by the results fromstrain JM105. In this strain, a stronger response was ob-served in the protease and GroEL levels and a similar re-sponse was observed for DnaK relative to RR1. The mea-sured CAT mass decreased more excessively with DTT ad-dition relative to RR1, suggesting a link between stressresponse and decreased expression levels. Also, the specificCAT activity was at times two-fold higher in JM105 thanfound in RR1, suggesting a linkage between cell stress andincreased specific CAT activity.

In summary, the results shown in this work unequivocallydemonstrated that the yield of active chloramphenicol acet-yltransferase, which is expressed in the cytoplasm, can becontrolled in a stable manner by altering the redox potentialin the fermentor through the addition of DTT at non-lethallevels (<3.0 g/L). In particular, CRP control was accom-plished in a highly aerated fermentor independent of theoxygen level. Moreover, the work demonstrated that a cel-lular stress response was observed upon DTT addition inshake flasks. However, the affects of this response could be‘‘regulated’’ in the fermentor, as noted by a consistent, al-tered level of CAT activity in the absence of significantchanges in cell growth and oxygen uptake. Although CATwas present in reduced quantity, its specific activity wasincreased. Also, the increased activity was stable because itwas measured after sonication in cell extracts. We cannotcomment, however, whether DnaK or GroEL had any directeffect on the specific activity. There was no general trend inCAT activity relative to either chaperone. It is likely that theeffects seen here are, in part, protein specific, hence theintroduction of CRP control by DTT addition would benefitfrom similar analysis of other model proteins. Finally, thatthe effects of DTT were evident in the cytoplasm suggeststhat additional effects would be seen in the periplasm forperiplasmically-targeted proteins.

References

Akashi, K., Ikeda, S., Shibai, H., Kobayashi, K., Hirose, Y. 1978. Deter-mination of redox potential levels for cell respiration and suitable forL-leucine production. Biotechnol. Bioeng.20: 27–41.

Ashburner, M., Bonner, J. 1979. The induction of gene activity in Dro-sophila by heat shock. Cell17: 241–254.

Baker, T., Grossman, A., Gross, C., 1984. A gene regulating the heat shockresponse inEscherichia colialso affects proteolysis. Proc. Natl. Acad.Sci. 81: 6779–6783.

Bentley, W. E., Mirjalili, N., Andersen, D. C., Kompala, D. S., Davis,R. H. 1990. Plasmid encoded protein: The principal factor in the‘‘metabolic burden’’ associated with recombinant bacteria. Biotech-nol. Bioeng.35: 668–681.

Bespalov, V., Zhulin, I., Taylor, B. 1996. Behavioral responses ofEsch-erichia coli to changes in redox potential. Proc. Natl. Acad. Sci.93:10084–10089.

Cha, H. J., Srivastana, R., Vakharia, V. N., Rao, G., Bentley, W. E. 1998.Green fluorescent protein as a non-invasive ‘‘stress probe’’ in recom-binantEscherichia coliculture. (Submitted).

Chuang, S., Blattner, F. 1993. Characterization of twenty six new heatshock genes ofEscherichia coli.J. Bacteriol.175: 5242–5252.

Derman, A., Prinz, W., Belin, D., Beckwith, J. 1993. Mutations that allowdisulfide bond formation in the cytoplasm ofEscherichia coli.Science262: 1744–1747.

Ellis, J., Bagshaw, C. R., Shaw, W. V. 1995. Kinetic mechanism of chlor-amphenicol acetyltransferase: The role of ternary complex intercon-version in rate determination. Biochem.34: 16852–16859.

Gamer, J., Multhaup, G., Tomoyasu, T., McCarty, J., Schonfeld, H. J.,Schirra, C., Bujard, H., Bukau, B. 1996. A cycle of binding and releaseof the DnaK, DnaJ, and GrpE chaperones regulates activity of theEscherichia coliheat shock transcription factors32. EMBO J. 15:607–617.

Gelman, M. S., Prives, J. M. 1996. Arrest of subunit folding and assemblyof nicotinic acetylcholine receptors in cultured muscle cells by dithio-threitol. J. Biol. Chem.271: 10709–10714.

Georgiou, G., Bowden, G. A. 1990. Inclusion body formation and therecovery of aggregated proteins, pp. 333–351. In: C. H., A. Prokop,and R. Bajpai (eds.),Recombinant DNA technology and applications.McGraw-Hill, New York.

Goff, S., Goldberg, A. 1987. An increased content of Protease La, the longene product, increases protein degradation and block growth inEsch-erichia coli. J. Biol. Chem.262: 4508–4515.

Harcum, S., Bentley, W., 1993. Detection, quantification, and character-ization of proteases in recombinantEscherichia coli.Biotechnol. Tech.7: 441–447.

Iuchi, S., Lin, E. C. C. 1993. Adaption ofEscherichia colito redox envi-ronments by gene expression. Mol. Microbiol.9(1): 9–15.

Iuchi, S., Cameron, D. C., Lin, E. C. C. 1988. A second glutal regulatorgene (arcB) mediating repression of enzymes in aerobic pathways ofEscherichia coli.J. Bacteriol.171(2):868–873.

Kamitani, S., Akiyama, Y., Ito, K. 1992. Identification and characterizationof an Escherichia coligene required for the formation of correctlyfolded alkaline phosphatase, a periplasmic enzyme. EMBO J.11:57–62.

Kjaergaard, L., Jorgensen, B. B. 1979. Redox potential as a state variablein fermentation systems. Biotechnol. Bioeng. Symp.9: 85–94.

Kwong, S. C. W., Rao, G. 1991. The utility of culture redox potential foridentifying metabolic state changes in the amino acid fermentation.Biotechnol. Bioeng.38: 1034–1040.

Kwong, S. C. W., Rao, G. 1992a. On-line assessment of metabolic activi-ties based on culture redox potential and dissolved oxygen profilesduring aerobic fermentation. Biotech. Prog.8: 576–579.

Kwong, S. C. W., Rao, G. 1992b. Effects of reducing agents in an anaero-bic amino acid fermentation. Biotechnol. Bioeng.40: 851–857.

Lodish, H. F., Kong, N. 1993. The secretory pathway is normal in dithio-threitol-treated cells, but disulfide-bonded proteins are reduced and


reversibly retained in the endoplasmic reticulum. J. Biol. Chem.268:20598–20605.

Lu, J., Jiang, C. 1993. Inhibition kinetics of chloramphenicol acetyltrans-ferase by selected detergents. Biochem. Biophys. Res. Comm.196:12–17.

Martin, J., Langer, T., Boteva, R., Schramel, A., Horwich, A., Hartl, F.1991. Chaperonin-mediated folding at the surface of GroEL through a‘‘molten globule’’-like intermediate Nature352: 36–42.

Maurizi, M. 1992. Protease and protein degradation inEscherichia coli.Experientia48: 178–199.

Missiakas, D., Georgopoulos, C., Raina, S. 1993. Identification and char-acterization of theEscherichia coligenedsbB,whose product is in-volved in the formation of disulfide bonds in vivo. Proc. Natl. Acad.Sci. USA90: 7084–7088.

Neidhart F., VanBogelen R., Vaughn V. 1984. The genetics and regulationof heat-shock proteins. Ann. Rev. Genet.18: 295–329.

Parsell, D. A., Lindquist S. 1993. The function of heat-shock proteins instress tolerance: Degradation and reactivation of damaged proteins.Ann. Rev. Genet.27: 437–496.

Plesset, J., Palm, C., McLaughlin, C. 1982. Induction of heat shock pro-teins and thermotolerance by ethanol inSaccharomyces cerevisiae.Biochem. Biophy. Res. Comm.108: 1340–1345.

Prinz, W., Aslund, F., Holmgren, A., Beckwith, J. 1997. The role of thethioredoxin and glutaredoxin pathways in reducing protein disulfidebonds in theEscherichia coli cytoplasm. J. Biol. Chem.272:15661–15667.

Pulliam-Holloman, T. R. 1996. Stress induced proteolysis inE. coli. PhD.Dissertation. University of Maryland, College Park, MD.

Robben, J., Van der Schueren, J., Volckaert, G. 1993. Carboxyl terminusis essential for intracellular folding of chloramphenicol acetyltransfer-ase. J. Biol. Chem.268: 24555–24558.

Rodriguez, R. I., Tait, R. C. 1983. Recombinant DNA techniques: An in-troduction. Benjamin/Cummings Publishing, Menlo Park, CA.

Sawyer, J. T., Lukaczyk, T., Yilla, M. 1994. Dithiothreitol treatment in-

duces heterotypic aggregation of newly synthesized secretory proteinsin HepG2 cells. J. Biol. Chem.269: 22440–22445.

Shaw, W. V. 1983. Chloramphenicol acetyltransferase: Enzymology andmolecular biology. CRC Crit. Rev. Biochem.14(1): 1–39.

Srinivas, S. P., Rao, G., Mutharasan, R. 1988. Redox potential in aerobicand microaerobic fermentation, p. 1187 In: L. E. Erickson and D. Y. C.Fung (eds.), Handbook on anaerobic fermentations. Marcel Dekker,New York.

Straus, D., Walter, W., Gross, C. 1988.Escherichia coliheat shock genemutants are defective in proteolysis. Genes Develop.2: 1851–1858.

Tomoyasu, T., Gamer, J., Bukau, B., Kanemori, M., Mori, H., Rutman, A.,Oppenheim, A., Yura, T., Yamanaka, K., Niki, H., Hiraga, S., Ogura,T. 1995.Escherichia coliFtsH is a membrane-bound, ATP-dependentprotease which degrades the heat-shock transcription factors32.EMBO J.14: 11 2551–2560.

Towbin, H., Staehelin, T., Gordon, J. 1979 Electrophoretic transfer ofproteins from polyacrylamide gels to nitrocellulose sheets: Procedureand some applications. Proc. Natl. Acad. Sci.76: 4350–4354.

Walker, K. W., Gilbert. 1994. Effect of Redox environment on the in vitroand in vivo folding of RTEM-1 b-Lactamase andEscherichia colialkaline phosphatase. J. Biol. Chem.269: 28487–28493.

Wimpenny, J. W. T. 1969. The effect of Eh on regulatory processes infacultative anaerobes. Biotechnol. Bioeng.11: 623–629.

Wimpenny, J. W. T., Necklen, D. K. 1971. The redox environment andmicrobial physiology. I. The transition from anaerobiosis to aerobiosisin continuous cultures of facultative anaerobes. Biochim. Biophys.Acta 253: 352–359.

Wunderlich, M., Glockshuber, R. 1993. In vivo control of redox potentialduring protein folding catalyzed by bacterial protein disulfide-isomerase (DsbA). J. Biol. Chem.268: 24547–24550.

Yamamori, T., Ito, K., Nakamura, Y., Yura, T. 1978. Transient regulationof protein synthesis inEscherichia coliupon shift-up of growth tem-perature. J. Bacteriol.134: 1133–1140.


generating controlled reducing environments in aerobic...

Documents