`
`http://aem.asm.org/
`
` on May 30, 2016 by guest
`
`APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Apr. 1990, p. 1004-1011
`0099-2240/90/041004-08$02.00/0
`Copyright © 1990, American Society for Microbiology
`
`Vol. 56, No. 4
`
`Comparison of Growth, Acetate Production, and Acetate Inhibition
`of Escherichia coli Strains in Batch and Fed-Batch Fermentations
`GREGORY W. LULit AND WILLIAM R. STROHL*
`Department of Microbiology, The Ohio State University, 484 West 12th Avenue, Columbus, Ohio 43210
`
`Received 18 October 1989/Accepted 11 January 1990
`
`The growth characteristics and acetate production of several Escherichia coli strains were compared by using
`shake Oasks, batch fermentations, and glucose-feedback-controlled fed-batch fermentations to assess the
`potential of each strain to grow at high cell densities. Of the E. coli strains tested, including JMIOS, B, W3110,
`W3100, HBIOI, DHI, CSHSO, MC1060, JRG1046, and JRG1061, strains JMIOS and B were found to have the
`greatest relative biomass accumulation, strain MC1060 accumulated the highest concentrations of acetic acid,
`and strain B had the highest growth rates under the conditions tested. In glucose-feedback-controlled fed-batch
`fermentations, strains B and JMIOS produced only 2 g of acetate· liter- 1 while accumulating up to 30 g of
`biomass. liter- 1 • Under identical conditions, strains HBIOI and MC1060 accumulated less than 10 g of
`biomass -liter- 1 and strain MC1060 produced 8 g of acetate -liter- 1 • The addition of various concentrations
`of sodium acetate to the growth medium resulted in a logarithmic decrease, with respect to acetate
`concentration, in the growth rates of E. coli JMIOS, JM105(p0S4201), and JRG1061. These data indicated th~t
`the growth of the E. coli strains was likely to be inhibited by the acetate they produced when grown on media
`containing glucose. A model for the inhibition of growth of E. coli by acetate was derived from these
`experiments to explain the inhibition of acetate on E. coli strains at neutral pH.
`
`Media used for the high-density growth of Escherichia coli
`(20, 22), as well as for the production of recombinant DNA
`products by E. coli strains (4, 25), typically include substan(cid:173)
`tial concentrations of glucose, since this is an inexpensive
`and readily utilizable carbon and energy source. Growth of
`E. coli on excess glucose under aerobic conditions, how(cid:173)
`ever, causes the formation of acidic by-products, of which
`acetate is the most predominant (1, 4, 9, 17, 20, 23, 25). This
`glucose-mediated aerobic acidogenesis, known as aerobic
`fermentation (4, 11, 16) or the bacterial Crabtree effect (10,
`25), is most readily observed when E. coli is grown at high
`growth rates in the presence of high glucose concentrations
`(10, 11, 16). These conditions are prevalent in fed-batch
`cultures in which glucose is fed in a nonlimiting manner to
`aerobic fermentation cultures to obtain high cell densities
`(17, 20, 22). The production of acidic by-products, especially
`acetate, is a major factor in the limitation of high-cell-density
`growth (1, 4, 17, 20, 26). Moreover, the accumulation of
`acetate during recombinant E. coli fermentations has been
`correlated with a reduced production of recombinant protein
`(8, 19), demonstrating the importance of choosing host
`strains and growth conditions which minimize acetate accu(cid:173)
`mulation. Although several studies have been carried out to
`determine the effect of fermentation conditions on the accu(cid:173)
`mulation of acetate and other by-products (1, 4, 8, 10, 12, 17,
`19, 20, 23, 25-27), no previous studies have compared the
`sensitivity of various E. coli strains to the bacterial Crabtree
`effect. In this study we compare several commonly used
`strains of E. coli with respect to growth rate, biomass yield,
`and acetate formation, with the goal of determining the
`strains most (or least) likely to yield good productivity in
`initial scale-up conditions. Such data should be helpful in
`choosing strains for rDNA fermentations.
`
`* Corresponding author.
`t Present address: Microlife Technics, Sarasota, FL 34230.
`
`MATERIALS AND METHODS
`Strains and medium composition. E. coli strains and plas(cid:173)
`mids used in this study are listed in Table 1. For mainte(cid:173)
`nance, cultures were grown for 24 to 48 h on plates contain(cid:173)
`ing solidified (1.5% agar) SD-7 medium (see below) at 37oC
`and then stored at 4oc for about 2 weeks before subculture to
`fresh media. A glucose-yeast extract medium (SD-7 medium
`[Table 2]), based on mass balance with respect to cell
`composition (31) and biomass yields from major elements
`(23), was developed for the growth of the E. coli strains.
`SD-7 medium was titrated to pH 7.0 with 5 M NH40H before
`autoclaving, and the glucose and MgS04 were autoclaved
`separately from the remaining components and added after
`cooling. The trace-element solution contained the following
`(in grams per liter of 5 M HCI): FeS04 • 7H20, 40.0;
`MnS04 • H 20, 10.0; Al2(S04h, 28.3; CoCl · 6H20, 4.0;
`ZnS04 · 7H20, 2.0; Na2Mo04 • 2H20, 2.0; CuC12 · 2H20,
`1.0; and H3B04 , 0.5.
`For fermentation experiments, SD-7 medium was modi(cid:173)
`fied by reducing the ammonium and magnesium content to
`prevent precipitation (SD-8 base medium [Table 2]). SD-8
`medium was used for batch fermentations and as the initial
`medium for fed-batch fermentations. The remaining nutri(cid:173)
`ents added in the fed-batch fermentations were supplied in
`the various feed solutions by a computer-controlled scheme
`(described below).
`Shake Oask experiments. Growth characteristics and the
`effects of varying the acetate concentrations on these char(cid:173)
`acteristics were evaluated by using a final volume of 27 ml in
`300-ml flasks. The flasks were autoclaved empty and dried
`overnight at 85°C. Before the experiment, 25 ml of sterile
`SD-7 medium (containing 2 to 5 g of glucose · liter- 1 as
`designated) were pipetted into each of the shake flasks,
`which were equilibrated to 37oc by shaking in an orbital
`shaker (250 rpm) for 2 h. A seed flask containing 25 ml of
`SD-7 medium (with 2 g of glucose . liter- 1) was inoculated
`with 1.0 ml of an overnight culture which had been grown in
`5 ml of tryptic soy broth (Difco Laboratories, Detroit,
`
`1004
`
`BEQ 1011
`Page 1
`
`
`
`VOL. 56, 1990
`
`E. COLI FERMENTATIONS
`
`1005
`
`TABLE 1. E. coli strains and plasmid used in this study
`
`Strain or
`plasmid
`
`Genotype
`
`Source"
`
`Downloaded from
`
`http://aem.asm.org/
`
` on May 30, 2016 by guest
`
`Once the flasks were equilibrated and the initial amounts
`of acetate were added, the turbidity was measured in a
`Klett-Summerson photometer to determine the background
`values (with a distilled water blank). Each flask was then
`inoculated, the contents were mixed, and 1.0 ml was re(cid:173)
`moved for later analysis of initial conditions. A turbidity
`measurement was taken to begin the experiment. Growth of
`the strains in these shake flask experiments was monitored
`every 30 min until early stationary phase. Another 1.0-ml
`sample was taken to determine final conditions, and the
`remainder of the culture was harvested for final dry-weight
`determinations. Samples were either stored on ice and
`assayed the same day or frozen at - 20°C and assayed the
`next day.
`Fermentations. Seed cultures for fermentations were
`started by picking an isolated colony from a plate and
`transferring by loop to a 1-liter flask containing 500 ml of
`SD-7 medium with 2 g of glucose · liter- 1 • Flasks were
`incubated overnight at 37°C at 200 rpm on an orbital shaker.
`The entire contents of the flask were used to inoculate the
`fermentor, which for batch culture conditions contained
`SD-8 base medium.
`The fermentation hardware has been extensively de(cid:173)
`scribed elsewhere (18, 28). Fermentors (MF-214; New Brun(cid:173)
`swick Scientific Co., Inc., Edison, N.J.) equipped with
`14-liter vessels (working volume, 10 liters) were used
`throughout this study. Control units (MRR-1; B. Braun
`Biotech, Bethlehem, Pa.) were used to monitor dissolved
`oxygen and pH in the fermentations with polarographic
`dissolved-oxygen probes (model 40180-02; Ingold Elec(cid:173)
`trodes, Inc., Andover, Mass.) and autoclavable pH probes
`(model 465; Ingold), respectively. All fermentations were
`carried out under the following growth conditions: temper(cid:173)
`ature, 37°C; pH, direct-digital controlled at 7 .0; dissolved
`oxygen, controlled by using the increase in agitation from
`300 to 600 rpm, followed by addition of mixtures of air and
`pure oxygen to 1 voUvoUmin (total flow) to maintain relative
`dissolved oxygen above 20% of saturation (18, 28).
`Glucose-feedback-controUed fed-batch fermentations. A
`computer-assisted on-line glucose analyzer was developed
`for the combined model-based and glucose feedback control
`offed-batch fermentations, as described previously (18). The
`system consisted of the prototype model 2000 glucose ana(cid:173)
`lyzer (Yellow Springs Instruments, Inc., Yellow Springs,
`Ohio), a filtration system (Megaftow TM-10; New Brun(cid:173)
`swick) containing a 10-in2 , 0.2-~J.m-pore-size filter, and high(cid:173)
`speed circulation pump (model7520-25; Cole-Parmer Instru(cid:173)
`ment Co., Chicago, Ill.). Modifications to the previously
`described system for this study included a higher sampling
`rate (2 min) and a 0.2-(.l.m-pore-size filter (Gelman Sciences,
`Inc., Ann Arbor, Mich.) between the glass tee union and the
`two-way acrylic valve. The setpoint for glucose control in
`the fed-batch fermentations described herein was 1.0 ± 0.2
`g · liter- 1 • When the glucose concentration remained within
`this window, the pump rate remained the same.
`For fed-batch fermentations, three feed solutions were
`used. (i) Feed no. 1 contained the following (in grams per
`liter): glucose, 200; MgS04 • 7H20, 0.85; (ii) feed no. 2
`contained the following (in grams per liter): glucose, 780;
`MgS04 • 7H20, 8.58; and (iii) feed no. 3 contained the
`following (in grams per liter): NH4Cl, 280; yeast extract, 150;
`it also contained 32 ml of the trace-element solution. All
`components of the feed solutions dissolved during autoclav(cid:173)
`ing and remained in solution upon cooling. A feeding strat(cid:173)
`egy was devised which began with SD-8 base medium in the
`vessel, with the exception that the initial glucose concentra-
`
`Strains
`B
`CSH50
`DH1
`
`HB101
`
`Wild type
`F- il.(lac proAB) strA thi ara
`F- recAJ endAJ gyrA96 thi-1
`hsdR17 supE44 A-
`F- hsd-20 recA13 ara-14 proAl lac-
`41 galK2 Str xyl-5 mtl-1 supE44
`A-
`F- il.(lac proAB) lacJ'I thi repsL
`endAJ slcB15 hadR4 traD36
`proABil.(ZM15)
`JRG1046 F- pta-39 trpR80 iclR17 A-
`JRG1061 F- ack-11 trpA9761 trpR72 iclR7
`gal-25 A-
`MC1060 F- il.(lacl-lacY)74 galEJ5 galK16
`relAJ rpsL150 spoTJ hsdR2 A-
`F- gal Ahft
`F- [r- m-]
`
`JM105
`
`W3100
`W3110
`
`osu 333
`C. J. Daniels
`J. S. Lampe!
`
`J. S. Lampe!
`
`Pharmacia
`
`CGSC 5992
`CGSC 5993
`
`CGSC 6648
`osu 395
`osu 389
`
`Plasmid
`p0S4201 pKK223-3 with denC at Smal siteh
`
`D. H. Dean
`
`a Abbreviations: OSU, The Ohio State University culture collection;
`CGSC, E. coli Genetic Stock Center (B. J. Bachmann).
`b The denC gene encodes delta endotoxin from Bacillus thuringiensis.
`
`Mich.). The culture in the seed flask was grown to mid(cid:173)
`exponential phase (Klett values of 50 to 100), and then 1.0 ml
`of this seed culture was used to inoculate each experimental
`flask.
`Acetate stock solutions were prepared in concentrated
`form so that 0.1 to 1.0 ml of the stock solution could be
`added to the experimental shake flask to give the desired
`final concentration. The acetate stock solution contained
`380.0 g of sodium acetate . liter- 1 (270.2 g of acetate
`ion · liter- 1) so that 0.1, 0.2, 0.5, and 1.0 ml yielded final
`concentrations of 1, 2, 5, or 10 g of acetate anion. liter- 1 ,
`respectively, in 27-ml (final volume) cultures. The volumes
`were equalized to 27 ml by addition of sterile water. In
`experiments in which the acetate was added at mid-loga(cid:173)
`rithmic phase, both the acetate additions and the water
`blanks were added after the experimental cultures had been
`grown to mid-exponential phase.
`
`TABLE 2. Components of media used to grow E. coli in batch
`and fed-batch fermentations
`
`Amt component (g/liter) added to medium listed
`for purpose mentioned
`
`Component
`
`NH4Cl
`KH2P04
`Na2HP04
`K 2S04
`MgS04 • 7H20
`Trace elements•
`Yeast extract
`Glucose
`Batch
`Fed-batch
`
`SD-7
`medium
`
`7.0
`1.5
`1.5
`0.35
`0.17
`0.8
`5.0
`
`1.0-5.0
`
`SD-8 medium
`
`Base and
`initial
`medium
`
`Components
`added in feed
`solutions
`
`7.0
`7.5
`7.5
`0.85
`0.17
`0.8
`10.0
`
`20.0
`1.0
`
`28.0
`0.0
`0.0
`0.0
`0.86
`3.2
`15.0
`
`99.0
`
`a Milliliters of trace elements solution, prepared as described in Materials
`and Methods, added per liter of medium or feed solution.
`
`BEQ 1011
`Page 2
`
`
`
`Downloaded from
`
`http://aem.asm.org/
`
` on May 30, 2016 by guest
`
`1006
`
`LULl AND STROHL
`
`APPL. ENVIRON. MICROBIOL.
`
`tion was 1.0 g. liter-1 (Table 2), and feed no. 1 was added
`based on glucose feedback control set at 1.0 g · liter-1 until
`the culture reached about 9 g of dry cell weight
`(DCW). liter-1. At that point, feeds no. 2 and 3 were added
`in place offeed no. 1 to keep up with the requirements of the
`high-density fermentations. In all cases, the glucose solu(cid:173)
`tions were fed at a rate based on glucose requirements
`calculated from previously run fermentations. Glucose feed(cid:173)
`back control was based on the on-line glucose concentration
`measurements as previously described (18).
`Analytical procedures. Cell growth was monitored by
`measuring culture turbidity with a Klett-Summerson color(cid:173)
`imeter using a red filter and by dry weight of biomass,
`determined as described previously (28). Off-line glucose
`analysis was carried out by using an analyzer (model 27;
`Yellow Springs Instruments) calibrated with either 2.0 or 5.0
`g of glucose standards. liter-1. Samples were clarified by
`centrifugation in a microcentrifuge and diluted in distilled
`water if necessary.
`Fermentation broth samples were prepared for acetic acid
`analysis by precipitation of macromolecules at pH 2.0 by the
`addition of 50 fLl of 70 mM H2S04 per 1.0 ml of sample at
`room temperature. The precipitate was pelleted for 2 min in
`a microcentrifuge, and the supernatants were filtered
`through a 0.2-fLm-pore-size, 13-mm-diameter filter (Gelman).
`The acetic acid produced in fermentation and shake flask
`experiments was quantitated by high-pressure liquid chro(cid:173)
`matography. The system consisted of Beckman/ Altex model
`100A pumps, model 420 pump controller, and model 400
`solvent mixer; an injector with a 20-fLl loop (model 7125;
`Rheodyne Inc., Cotati, Calif.); and Teflon tubing (inner
`diameter, 0.18 mm; Alltech Associates Inc., Deerfield, Ill.)
`was used throughout the system. An organic acid analysis
`Aminex ion-exchange column (7 .8 by 300 mm; model HPX-
`87H; Bio-Rad Laboratories, Richmond, Calif.) was used,
`and organic acids were separated by using 3.5 mM H2S04 at
`pH 2.0 at a flow rate of0.7 ml. min- 1 at room temperature.
`The elution was monitored with a Hitachi model 160-40
`variable-wavelength spectrophotometer set at 210 nm, and
`A210 peaks were recorded and integrated with either a model
`3390A or model3396 integrator (Hewlett-Packard Co., Palo
`Alto, Calif.). Organic acid standards were prepared from
`reagent-grade chemicals dissolved in high-pressure liquid
`chromatography-grade water. Standards were injected under
`the same conditions as the fermentation samples, and the
`retention times were compared. The identification of acetate
`in samples of fermentation broth was confirmed by cochro(cid:173)
`matography.
`
`RESULTS
`Shake flask experiments. The growth rate, yield, final
`biomass, and acetate production by eight E. coli strains were
`compared after growth of the strains under identical condi(cid:173)
`tions in shake flasks containing SD-7 medium plus 5 g of
`glucose . liter- 1 (Table 3). Under the conditions tested, the
`growth rates of the strains ranged from 0.85 h-1 (generation
`time, 49 min) to 1.42 h- 1 (generation time, 29 min), the final
`dry cell weights ranged from 1.16 to 1.87 g · liter-1, the
`yields (grams of biomass per gram of glucose utilized) ranged
`from 0.54 to 0.77, and 0.30 to 0.92 g of acetate · liter- 1 was
`produced (Table 3). The strains could be divided into two
`categories based on their growth rates in SD-7 medium:
`fast-growing strains (strains B, W3110, W3100, and CSH50)
`and slow-growing strains (strains JM105, DH1, MC1060, and
`HB101). The faster-growing strains generally produced more
`
`TABLE 3. Comparison of the growth parameters of several
`E. coli strains grown in shake flasks containing SD-7 medium•
`
`Strain
`
`B
`W3110
`W3100
`CSH50
`JM105
`DH1
`MC1060
`HB101
`
`Initial specific
`growth rate
`(h-1)
`
`DCW
`(g/liter)
`
`Yield (g of cells/
`g of glucose
`consumed)
`
`Amt of acetate
`produced
`(g/liter)
`
`1.42
`1.29
`1.13
`1.00
`0.93
`0.91
`0.90
`0.85
`
`1.76
`1.34
`1.61
`1.45
`1.69
`1.16
`1.42
`1.87
`
`0.63
`0.57
`0.54
`0.60
`0.71
`0.64
`0.60
`0.77
`
`0.79
`0.50
`0.79
`0.62
`0.35
`0.51
`0.92
`0.30
`
`• SD-7 medium in these experiments contained 5 g of glucose · liter 1
`(autoclaved separately and added after cooling). The data represent the
`average of three separate experiments, and in no case was the standard error
`above 5%.
`
`acetate and had lower biomass yields from glucose than did
`the slower-growing strains. The major exceptions to these
`generalized relationships were found with strain MC1060,
`which, although slow-growing, produced the largest amount
`of acetate of any strain tested, and strain B, which main(cid:173)
`tained both high growth yields and high growth rates (Table
`3).
`Effect of acetate additions to shake flask cultures. The
`addition of sodium acetate to shake flask cultures of the
`acetate kinase mutant, strain JRG1061, the low-acetate(cid:173)
`producing strain, JM105, and a recombinant strain, JM10?
`(p0S4201), in SD-7 medium (containing 2 g of glucose · h(cid:173)
`ter-1) reduced the growth rates of the organisms logarithmi(cid:173)
`cally (Fig. 1). The ack (acetate kinase minus) mutant strain,
`JRG1061, which should have a reduced ability to assimilate
`acetate (9), was inhibited by acetate to the same extent as
`was strain JM105. Therefore, the ability with which a strain
`
`2 .0 1 r - - - - - - - - - - - - - - - - ,
`•
`
`.c
`i 0.2
`
`2 "
`
`0jL-~--*2--i3--~4--~5~~6~~7~~8~~9~~~
`Na-Acetate lg/LI
`
`57
`
`45.6
`34.2
`22.8
`11.4
`lmg/L; pH7.01
`ICH3COOHI
`FIG. 1. Effect of exogenously added acetate concentration in
`SD-7 medium (containing 2 g of glucose. liter- 1 [pH 7.0]) on the
`growth rates of E. coli strains. The acetate was added just before
`inoculation. Symbols: •· JM105; A, strain JM105(p0S4201); e,
`strain JRG1061; 0, data showing the same effect on strain JM105 of
`acetate added at mid-logarithmic growth phase. Each datum point
`represents triplicate values with less than 2% error. The line was
`drawn by calculated linear regression for data from strain JM105 (r
`= 0.9997). The concentration of the protonated form of acetate at
`pH 7.0 was calculated by using the Henderson-Hasselbalch equa(cid:173)
`tion.
`
`BEQ 1011
`Page 3
`
`
`
`Downloaded from
`
`http://aem.asm.org/
`
` on May 30, 2016 by guest
`
`VoL. 56, 1990
`
`E. COLI FERMENTATIONS
`
`1007
`
`TABLE 4. Comparison of the growth characteristics of several
`E. coli strains grown in batch cultures of SD-8 medium
`in 10-liter stirred-tank fermentors
`
`Strain
`
`Specific Glucose
`growth
`consump-
`rate
`tion rate
`(h-')
`(h-')"
`
`Yield
`(g of cells/
`g of glu-
`cose con-
`sumed)
`
`Final
`DCW
`(g/liter)
`
`Acetate
`production
`
`Initial
`rate
`(h-')"
`
`Final
`concn
`(g/liter)
`
`B
`JRG1046
`JRG1061
`JM105
`MC1060
`JB101
`
`1.14
`1.02
`0.89
`0.79
`0.77
`0.64
`
`1.09
`1.38
`0.56
`0.44
`0.89
`0.85
`
`1.02
`0.61
`0.33
`0.64
`0.56
`0.73
`
`9.3
`9.6
`8.3
`10.8
`8.0
`5.4
`
`0.50
`0.71
`0.39
`0.62
`0.78
`0.57
`
`1.75
`1.60
`3.03
`1.20
`5.12
`0.88
`
`a Calculated as slope= log [(grams of acetate per liter at time point T2 -
`grams of acetate per liter at time point T1)/(time point T2 -
`time point T1)],
`which yields h- 1•
`
`metabolized acetate did not apparently influence the inhibi(cid:173)
`tory effect of acetate on that strain. Similarly, the growth
`rate of the recombinant strain, JM105(p0S4201), was inhib(cid:173)
`ited to the same extent by the added acetate, indicating that
`the presence of the plasmid did not alter the level of toxicity
`of acetate to strain JM105 (Fig. 1). Moreover, the inhibition
`of growth of E. coli by acetate was independent of the age of
`the culture (Fig. 1). When sodium acetate was added at
`mid-exponential phase to make 10 g. liter-1 (final concen(cid:173)
`tration), both the growth rate and yield (grams of DCW per
`gram of glucose utilized) of strain JM105 were reduced by
`more than 50%, compared with control cultures in which
`only buffer was added at the same time (data not shown).
`Batch fermentations. Four strains were further compared
`in 10-liter batch fermentations with SD-8 medium: (i) a slow(cid:173)
`growing, high-acetate producer (strain MC1060); (ii) a slow(cid:173)
`growing, low-acetate producer (strain HB101); (iii) a fast(cid:173)
`growing, high-acetate producer (strain B); and (iv) a fast(cid:173)
`growing, relatively low-acetate producer (strain JM105). In
`addition to these four strains, two E. coli mutant strains
`defective in acetate metabolism, JRG1046 and JRG1061 (15),
`were included in these batch fermentation experiments.
`With the six strains of E. coli that were grown in SD-8 base
`medium in batch fermentations, wide ranges of growth rates
`(0.79 to 1.14 h-1), glucose consumption rates (0.44 to 1.09
`h-1), acetate production rates (0.39 to 0.78 h-1), biomass
`yield on glucose (0.33 to 1.02 g . g-1), and amount of acetate
`produced (0.88 to 5.12 g. liter-1) were observed. The
`growth rates of the six strains grown in SD-8 base medium in
`batch fermentation (Table 4) were 15 to 25% lower than the
`growth rates for the same strains grown in SD-7 medium in
`shake flasks (Table 3). There was generally an inverse
`relationship between the growth rates and the yields from
`glucose, which was similar to the shake flask data (Table 3)
`and data obtained in other investigations (10). One exception
`to these trends was that strain B maintained a high biomass
`yield from glucose during rapid growth. Unexpectedly,
`strains MC1060 and JRG1061 produced large amounts of
`acetate while growing relatively slowly.
`Fed-batch fermentations. The growth parameters and ace(cid:173)
`tate production of four strains were evaluated under glucose
`feedback-controlled conditions in which the glucose setpoint
`was 1.0 ± 0.2 g. liter-1. The actual glucose control was not
`quite as tight as that set by the computer and was usually
`within 1.0 ± 0.5 g. liter-1; this was due to the changes in
`growth rates by the strains used during the fermentations.
`Under these glucose-controlled conditions, strain JM105
`
`grew to 31 g · liter of DCW liter-1 (Fig. 2A) with a biomass
`yield of 0.42 g of DCW · g of glucose consumed-1 and a
`productivity of 3.47 g of DCW. liter-1 . h- 1. The initial
`growth rate was 0.69 h-1 for the first 5 h; however, the
`growth rate continually declined for the last 4 h of the
`experiment. Acetate was accumulated by JM105 to a final
`concentration of about 2 g . liter-1.
`In fed-batch cultures, strain B grew at a rate of 1.03 h- 1
`during the initial growth phase, and growth continued to
`about 30 g of DCW · liter-1 (Fig. 2B). Because of the rapid
`growth, the proportional control of the feed pumps could not
`maintain the feed rate to meet the glucose demand of the
`culture, and therefore the culture became glucose limited by
`3.5 h. Biomass productivity was 3.75 g of DCW . li(cid:173)
`ter-1 · h-1, and about 2 g of acetate · liter-1 was accumu(cid:173)
`lated. However, during the period of glucose limitation, the
`acetate produced at the beginning of the experiment was
`consumed. Once relieved from glucose limitation, after
`about 6 h of growth, acetate again began to accumulate.
`Only 9.1 g (DCW) of strain HB101 · liter-1 was produced
`under fed-batch culture conditions similar to those described
`for strains JM015 and B (Fig. 2C). Glucose concentration
`was well controlled near the setpoint, and, similarly to
`strains JM105 and B, only about 2.0 g of acetate · liter-1 was
`produced. The accumulation of acetate continued through(cid:173)
`out the experiment at a nearly constant rate of 0. 2 h - 1.
`Strain MC1060 grew to about 10 g of DCW . liter-1 in 9 h
`(Fig. 2D). Even though good control of glucose concentra(cid:173)
`tion was achieved, 8 g of acetate · liter-1 was accumulated.
`The rate of acetate accumulation by strain MC1060 in
`fed-batch cultures was similar to that of batch fermentations
`(Table 4), but because of the longer period of growth in
`fed-batch cultures, more acetate was accumulated.
`
`DISCUSSION
`Growth of E. coli strains in batch and fed-batch fermenta(cid:173)
`tions. Of the six strains of E. coli grown in batch fermenta(cid:173)
`tions, strain JM105 had the lowest glucose consumption rate
`and a relatively low growth rate, and it produced the second
`smallest amount of acetate of all strains tested, which may
`explain why cell densities of JM105 were higher than those
`of the other K-12 derivatives tested. Strain JM105 was
`unique in that a biphasic growth curve was measured during
`batch fermentations, while the glucose consumption rate
`remained constant (18). A similar biphasic growth curve was
`observed previously in the high-cell-density growth of other
`E. coli strains (1).
`Strain B grew at the highest rate and had the highest
`biomass productivity of all strains tested. Strain B has been
`used to produce the highest recorded cell densities of E. coli
`in fed-batch fermentations (125 g . liter-1 [20]; 90 g . liter-1
`[22]), whereas there is only one report of an E. coli K-12
`derivative grown as dense as 78 g · liter-1 in fed-batch
`fermentations (1). On the other hand, in a comparison of the
`high-density growth of a few E. coli strains, Bauer and Ziv
`(7) found that E. coli W, a strain unrelated to E. coli K-12 (5),
`achieved both higher biomasses and higher growth rates than
`E. coli B grown at 30°C under identical high-cell-density
`fed-batch growth conditions.
`Strains HB101 and MC1060 did not grow to densities
`greater than 10 g of DCW . liter-1 under fed-batch condi(cid:173)
`tions similar to those which supported 30 g of strains JM105
`and B . liter-1 (Fig. 2). Since the fed-batch cultures of
`HB101, JM105, and B produced approximately the same
`amount of acetate, other genetic factors probably were
`
`BEQ 1011
`Page 4
`
`
`
`Downloaded from
`
`http://aem.asm.org/
`
` on May 30, 2016 by guest
`
`1008
`
`LULl AND STROHL
`
`APPL. ENVIRON. MICROBIOL.
`
`10
`
`B
`
`•
`
`.. u ...
`
`0
`
`3~
`~
`
`-~ •
`
`2J
`
`•
`•
`
`10
`
`Time [hl
`
`1 ..
`
`i
`
`o.:i
`12
`
`c
`
`D
`
`10
`
`4
`cj
`.:1
`
`•
`
`2
`
`0
`
`•
`
`8
`
`FIG. 2. Growth of E. coli strains in SD-8 medium in glucose feedback-controlled fed-batch fermentations. In all cases, the soluble glucose
`setpoint was controlled at 1.0 g. liter- 1 (horizontal line). Symbols: e, DCW (grams per liter); •· soluble glucose concentration (grams per
`liter); •. soluble acetate concentration (grams per liter). (A) Strain JM105; (B) strain B; (C) strain HBIOI; (D) strain MC1060.
`
`involved in the limitation of the growth of HB101. On the
`other hand, the fed-batch culture of strain MC1060 accumu(cid:173)
`lated large amounts of acetate, even though glucose concen(cid:173)
`trations were maintained at low levels. In previous studies
`with other E. coli strains, the production of acetate was
`limited by controlling glucose at concentrations slightly
`higher than the levels maintained in this work (1). The
`acetate production rate exhibited by strain MC1060, how(cid:173)
`ever, was only slightly higher than that of other strains which
`did not accumulate as much acetate (Table 4). In batch
`culture fermentations of strain MC1060, the level of acetate
`
`did not decrease once the glucose was completely consumed
`(data not shown), indicating that the accumulation of acetate
`to higher concentrations in MC1060 cultures may be due to
`reduced enzyme activity in the acetate reassimilation path(cid:173)
`way. Once glucose has been consumed, most E. coli cultures
`reutilize acetate (1, 4, 9, 12) by an activated tricarboxylic
`acid cycle (2, 10).
`Acetate production by acetate metabolism mutants. The ack
`(acetate kinase-minus) strain, JRG1061 (15), accumulated
`substantial amounts of acetate, which was expected since
`ack mutants were previously shown to accumulate at least as
`
`BEQ 1011
`Page 5
`
`
`
`Downloaded from
`
`http://aem.asm.org/
`
` on May 30, 2016 by guest
`
`VoL. 56, 1990
`
`E. COLI FERMENTATIONS
`
`1009
`
`much acetate as wild-type strains did (9). The production of
`acetate by an ack strain was probably due to the spontane(cid:173)
`ous hydrolysis of acetyl phosphate, which occurs at rela(cid:173)
`tively high rates at 37°C under physiological conditions (9).
`Similarly, the absence of acetate kinase would possibly slow
`the reassimilation of acetate by strain JRG1061. Even though
`acetyl coenzyme A (acetyl-CoA) synthetase is considered to
`be the primary route for acetate assimilation (21), the max(cid:173)
`imal velocity of acetyl-CoA formation by this enzyme is too
`low to support the growth of E. coli (9). Moreover, acetyl(cid:173)
`CoA synthetase is inhibited by glucose (9), suggesting that
`the acetate kinase-phosphotransacetylase route may be im(cid:173)
`portant in the reassimilation of acetate, albeit with low
`affinity (9).
`The production of 1.6 g of acetate · liter- 1 by the pta
`(phosphotransacetylase-minus) mutant strain, JRG1049, was
`unexpected since Brown et al. (9) differentiated pta and ack
`mutants based on the inability of the former to produce
`acetate. It is possible that reversal of the acetyl-CoA syn(cid:173)
`thetase reaction is responsible for this acetate production;
`however, since acetyl-CoA synthetase is repressed by glu(cid:173)
`cose (9), it should not have been active under the growth
`conditions used. Acetyl-CoA synthetase mutants have not
`yet been isolated (21), so the acetate production by a
`combined pta and acetyl-CoA synthetase mutant cannot be
`measured as yet.
`Inhibition of growth by acetate. It is well established that
`E. coli cultures growing aerobically in the presence of excess
`glucose produce acidic by-products, particularly acetate (1,
`3, 4, 8-11, 17, 19, 20, 22, 23, 26, 27). This glucose-mediated
`aerobic acidogenesis is known as the bacterial Crabtree
`effect (10, 25). The formation of acetate has been suggested
`to be caused by an imbalance between glucose metabolism
`and respiration (3, 13), a condition in which influx of carbon
`into the cell exceeds demands for biosynthesis (12), the
`presence of excess NADH (10), the repression of tricarbox(cid:173)
`ylic acid cycle enzymes (2, 14, 25), or uncoupled metabolism
`(26). It is likely that all of these interrelated causes are
`involved.
`The acetate produced by E. coli cultures growing on
`excess glucose has been suggested to cause an increased
`uncoupling in metabolism (26, 27). Because of its weak
`lipophilic nature, the protonated form of acetate is able to
`cross the cell membrane (6, 24) and act as an uncoupler of
`the proton motive force (specifically dpH) (24, 26, 27). This
`toxic effect has been clearly demonstrated at pH values at or
`below the pKa of the acid (24, 29); however, evidence now
`suggests that acetate is toxic to cell growth at neutral pH
`when sufficient amounts accumulate in the medium (1, 4, 17,
`20, 22, 26). Acetate exists at neutral pH in both the ionized
`(CH3Coo-) and protonated (CH3COOH) states. The more
`lipophilic protonated form can pass through the lipid mem(cid:173)
`brane to the interior of the cell, where it dissociates at the
`higher internal pH (ca. 7 .5) to CH3Coo- and H+, thereby
`decreasing the intracellular pH (24, 27). When the proto(cid:173)
`nated acid moves across the membrane into the cells,
`additional protonated acid is formed in the medium by the
`equilibrium, causing a net electroneutral hydrogen ion influx
`(24). The overall external pH would not change drastically,
`because of the large volume of buffered medium, so the
`decrease in intracellular pH would cause an uncoupling
`effect. The homeostatic mechanisms of E. coli require en(cid:173)
`ergy to adjust to this decrease in intracellular pH, even in the
`absence of an overall change in proton motive force (27).
`Thus for rapid