`Growth Hormone in Escherichia coli:
`Expression of Different Precursors and
`Physiological Effects of Glucose, Acetate,
`and Salts
`
`E. Bech Jensen and 5. Carlsen*
`Novo Nordisk als, Novo Aile, DK-2880 Bagsvaerd, Denmark
`
`Accepted for publication September 1, 1989
`
`The constitutive cytoplasmic expression in E. coli of hu(cid:173)
`man growth hormone (hGH) with different N-terminal
`extensions (3 or 4 amino acids) has been studied. These
`hGH precursors were used for in vitro cleavage to ob(cid:173)
`tain the mature, authentic hormone. Small changes in
`the amino acid extensions of the hGH precursors led to
`three-fold differences in specific expression rates. The
`specific expression rate of the hGH precursors was in(cid:173)
`versely proportional to the ratios of the specific growth
`rates of plasmid containing and plasmid free cells
`(J.L+ IJ.L-) and also to the genetic stability. To ensure a sat(cid:173)
`isfactory genetic stability in production fermentors, an
`hGH precursor with a moderate expression efficiency
`was chosen.
`The medium composition and growth conditions
`were studied, resulting in the choice of a glucose fed
`batch fermentation process using a complex medium. In
`this process a yield of 2000 mg/L of met-ala-glu-hGH
`(MAE-hGH) was obtained. The fermentation process
`comprised a glucose-limited growth phase followed by
`a second phase with increased glucose feed and ex(cid:173)
`haustion of phosphate from the medium. The second
`phase is characterized by an MAE-hGH production,
`whereas further biomass formation is blocked. High
`concentrations of glucose led to reduced specific ex(cid:173)
`pression of MAE-hGH-the specific and total yield in
`batch glucose fermentations is only about 30% of the
`yield in optimized fed batch fermentations. The physio(cid:173)
`logical background for this was investigated. Chemostat
`experiments showed that the glucose concentration and
`the metabolic condition of the cells- i.e. with or with(cid:173)
`out formation of acetate-was not critical per se in
`order to obtain a high specific yield of MAE-hGH. There(cid:173)
`fore it is unlikely that formation of MAE-hGH is catabo(cid:173)
`lite repressed by glucose. Furthermore it was shown
`that the specific production rate of MAE-hGH was inde(cid:173)
`pendent of the specific growth rate and it was further
`demonstrated that the decrease in expression efficiency
`in glucose batch fermentation was a result of an inhibi(cid:173)
`tory effect of acetic acid. In batch fermentations this in(cid:173)
`hibitory effect was enhanced by a salt effect caused by
`increased consumption of acid and base used to control
`
`* To whom all correspondence should be addressed.
`
`pH. The identity of the acid and the base used are not
`important in this context.
`From studies of the expression of other proteins in
`E. coli with constitutive as well as inducible promoters
`we conclude that glucose fed batch processes are often
`superior to batch processes in the production of het(cid:173)
`erologous proteins in E. coli.
`
`INTRODUCTION
`
`For many years Escherichia coli has been the dominant
`procaryote organism in the genetic exploration of micro(cid:173)
`organism. On the basis of the vast knowledge accumulated
`on this bacteria it has become the most used host organism
`for expression of recombinant products and with very good
`results in laboratory cultures. At the same time though,
`there has not been the same physiological background for
`the development of a high yield fermentation process,
`where the biomass concentration is increased to high levels
`(>25 g dry cell wt/L) and at the same time the specific
`yield of the heterologous gene product is kept high.
`The glucose metabolism in E. coli has been studied ex(cid:173)
`tensively and it is well known 1
`3 that under aerobic growth
`-
`conditions with high glucose concentrations and high
`growth rates there is a production of acetate and/or lactate,
`depending on the strain. The physiological background for
`this acid production under aerobic conditions has been
`found to be a result of unbalanced rates of glycolysis and
`the oxidation of the metabolites formed due to saturation
`of this respiration capacity of E. coli 4 and/or the capacity
`of oxidative phosphorylation due to nutrient limitations
`(i.e., trace metals). 5
`Pan et al. 6 recognized that it was necessary to avoid ac(cid:173)
`cumulation of acetate in a fermentation process if very
`high concentrations of biomass (90 g dry cell wt/L) were
`to be obtained. They suggested that growth was inhibited
`by a combination of high concentrations of the metabolic
`byproducts (mainly acetate) and high pC02 levels (0.35-
`
`Biotechnology and Bioengineering, Vol. 36, Pp. 1-11 (1990)
`© 1990 John Wiley & Sons, Inc.
`
`CCC 0006-3592/90/01 0001-11 $04.00
`
`BEQ 1037
`Page 1
`
`
`
`0.4 atm). Their experiments were carried out with a wild(cid:173)
`type strain of E. coli B, which has a much better growth
`performance than manipulated laboratory strains such as
`E. coli K12. 5
`Gleiser and Bauer7 showed that a fed batch process
`where the carbon feed rate was controlled by the oxygen
`level, was useful in obtaining high biomass concentrations
`(42 g dry cell wt/L) of E. coli W.
`There have been only few publications on the relation(cid:173)
`ship between culture methods for E. coli and the amount
`produced of a recombinant product, except for those deal(cid:173)
`ing with the specific physiological impact on promoter in(cid:173)
`duction, plasmid copy number, etc. Meyer et al. 3
`8 have
`'
`reported that high acetate concentrations, during cultiva(cid:173)
`tion of an interferon-producing E. coli strain, indicated
`that the interferon titer was low. The production of inter(cid:173)
`feron was growth associated and the decrease in interferon
`titer might be correlated to the reduced growth rate in the
`presence of acetate.
`In this article we describe the development of an E. coli
`fermentation process for the production of recombinant
`amino acid extended human growth hormone (X-hGH)
`with high yields. The hGH precursors produced can be
`converted to authentic human growth hormone by in vitro
`treatment with the proteolytic enzyme DAP 1. 9 In this way
`the aminoacid extension including the methionine corre(cid:173)
`sponding to the translation start codon is removed.
`The study of expression and genetic stability of different
`hGH precursors has been combined with physiological in(cid:173)
`vestigations leading to an optimized two-stage fed batch
`fermentation process. This article discusses the physiologi(cid:173)
`cal basis for the fact that glucose-limited fermentations can
`result in substantially higher yields of hGH precursors than
`glucose batch fermentations.
`
`MATERIALS AND METHODS
`
`Organism
`E. coli MC1061 10 -a derivative of E. coli K12-was
`used as the host organism. Different expression plasmids,
`corresponding to different amino-extensions of hGH, were
`constructed as described elsewhere. 9
`11 The plasmids were
`•
`based on the vector pAT153 (derived from pBR-322), con(cid:173)
`taining the ampicillin resistance gene. The X-hGH genes
`were controlled by a synthetic constitutive promoter, and
`X-hGH was produced as a cytoplasmic protein. In all the
`constructs described, the produced X-hGH was hGH ex(cid:173)
`tended at the amino terminal with 3 or 4 amino acids.
`
`Culture Media
`
`The complex medium HKSII consisted of yeast extract,
`5 g/L; tryptone, 10 g/L; acid hydrolyzed casein, 2 g/L;
`salts (Mg, K, and Ca); trace metals (Fe, Zn, Mn, Cu, Co,
`B, Mo, and I); and antifoam (Glanapon), 0.2 mL/L.
`NaOH and H2S04 were used to adjust pH to 7.2. HKSO
`was a four-fold concentrate of HKSII.
`
`Plasmid Stability
`
`The stability of X-hGH expression, for cells grown for a
`number of generations, was determined by subcultivation
`in shake flasks: 0.098 mL inoculum (seed lot) was added
`to 100 mL medium (HKSII) followed by incubation for
`24 hat 30°C. A sample was analyzed for X-hGH concen(cid:173)
`tration and OD, and a new 100-mL HKSII shake flask was
`inoculated with 0.098 mL of the culture. By repeating this
`procedure, an expression level of X-hGH was established
`after every growth period, corresponding to 10 additional
`generations.
`
`Batch Fermentations
`
`Equipment consisted of a 2.5-L MBR, and 7-L CHEMAP
`fermentors. The pH, temperature, DOT, agitation, aera(cid:173)
`tion, and foam level were automatically controlled.
`Propagation consisted of the following. A single colony
`of X-hGH producing E. coli was picked from a Petri dish
`containing ampicillin and transferred to a shake flask con(cid:173)
`taining 100 mL HKSII and 50 mg/L ampicillin. Following
`an incubation for 20 h at 30°C and 180 rpm, a fermentor
`was inoculated with 20 mL of the culture per L fermenta(cid:173)
`tion volume.
`Fermentation consisted of the following. Temperature was
`30°C. pH was maintained at 7.2 (controlled by the addition
`of H2S04 and NaOH). Aeration was 1 VVM. Agitation
`was :;:::500 rpm, adjusted to ensure DOT :;:::20%. Medium
`was HKSO. As carbon source, glucose was autoclaved
`separately and added to the medium before inoculation.
`
`Fed Batch Fermentation
`
`Equipment, propagation method, medium, and culture
`conditions were similar to batch fermentations. As carbon
`source, the glucose feed was started at a biomass concen(cid:173)
`tration corresponding to OD = 5. The feeding rate was ad(cid:173)
`justed to the desired feed rate and maintained constant
`until the end of the fermentation. In some experiments a
`stepwise increase in the feed rate during the fermentation
`was introduced.
`
`Chemostat Culture
`
`Equipment consisted of a 2.5-L Chemap fermentor (work(cid:173)
`ing volume of 1.3 L). A fixed dilution rate was obtained
`by maintaining a constant working volume, based on
`weight control of the fermentor. The propagation method
`and culture conditions were similar to batch fermentation.
`Medium was HKSII + 5 g glucose/L + 50 mg ampi(cid:173)
`cillin/L.
`
`Analytical Methods
`
`hGH
`X-hGH was analyzed by an hGH-ELISA assay. 12 The pre(cid:173)
`treatment was as follows. The fermentation samples were
`
`2
`
`BIOTECHNOLOGY AND BIOENGINEERING, VOL. 36, JUNE 1990
`
`BEQ 1037
`Page 2
`
`
`
`frozen at -20°C. After thawing, 100-pL sample was
`mixed with 750 ILL lysis buffer (50 mM Tris/HCl, pH 8.0,
`50 mM EDTA, 15% sucrose, 1% Tween 80, 1 mg/mL
`lysozyme) and incubated at 0°C for 60 min. Thereafter
`150 ILL DNase buffer (50 mM Tris/HCl, pH 7.5, 280 mM
`MgC1 2 , 4 mM CaCl2, and 70 ILg/mL DNasei from Sigma)
`was added and the incubation continued at 0°C for 5 min.
`The sample was then frozen at -20°C until the ELISA
`analysis was carried out.
`
`100 % AmpR
`
`50
`
`Biomass Concentration, OD
`
`The optical density was measured spectrophotometrically
`at 525 nm. One absorbance unit (1 OD unit) corresponds
`to 0.25 ±0.05 g dry wt/L (depending on growth phase
`and cell size). The samples were diluted with 0.9% NaCl
`until the measured absorbance was in the interval from 0.1
`to 0.6.
`
`Acetate
`
`Acetate was determined with an enzymatic test kit from
`Bohringer Mannheim (No. 148.261). Concentration mea(cid:173)
`surements by gas chromatography analysis were done on a
`few of the samples and a good correlation was found with
`the enzymatic test.
`
`Glucose
`
`Glucose was determined with an enzymatic test kit from
`Sigma (No. 115).
`
`RESULTS AND DISCUSSION
`
`Genetic Stability and Specific Expression Rate of
`hGH Precursors
`
`In the X-hGH fermentation process the E. coli was grown
`under nonselective conditions; antibiotics such as ampi(cid:173)
`cillin are unwanted in the production medium, because of
`environmental and health-care considerations. Furthermore
`ampicillin is degraded rapidly in suspension cultures by re(cid:173)
`leased ~-lactamase.
`Under nonselective conditions problems with the expres(cid:173)
`sion stability could be observed. In Figure 1, one sees that
`the decline in expression level was closely correlated to the
`proportion of ampicillin resistant cells. The 100% MEAE(cid:173)
`hGH-expression was defined as the expression in the first
`shake flask inoculated with the working cell bank, which
`was also defined as 10 generations. This strongly indicates
`that the decrease in MEAE-hGH expression observed after
`80-100 generations (Fig. 2) resulted from plasmid free
`cells overgrowing plasmid containing cells in the culture.
`A number of hGH precursors, with different amino acid
`extensions were produced mainly with the aim of optimiz(cid:173)
`ing the in vitro enzymatic conversion of DAP 1 to hGH.
`Figure 2 and Table 1 demonstrate that small changes in the
`amino acid extensions, in otherwise identical constructs,
`
`50
`
`%
`100
`Relative MEAE·hGH Expression
`
`Figure 1. Correlation between the relative MEAE-hGH expression and
`the proportion of ampicillin resistant cells. Samples were taken from sub(cid:173)
`cultivations in shake flasks (see Materials and Methods section). %
`AMPR was determined by plate count on plates with and without
`50 mg/L AMP.
`
`result in large differences in genetic stability and expres(cid:173)
`sion efficiency.
`The constructs MTEE-hGH and MFEE-hGH resulted in
`very efficient expression, but due to poor genetic stability,
`high yields could not be obtained in a large-scale fermenta(cid:173)
`tion process. On the other hand, the MLE-hGH construct
`gave a stable expression for more than 100 generations.
`The loss of plasmids was not followed by a measurable
`decline in expression until the plasmid free part of the cell
`population constituted more than about 3%. The number of
`generations before a decline in expression was observed
`(the horizontal part of the curves in Fig. 2) could be corre-
`
`X·hGH
`mg/00·1
`
`20
`
`10
`8
`
`6
`5
`4
`
`3
`
`2
`
`1.0
`
`MEAE
`
`80
`
`100
`
`140
`120
`no. of gen.
`
`Figure 2. Expression stability of different hGH precursors as a function
`of the number of generations the cells were grown. Subcultivations were
`performed in shake flasks as described in materials and methods.
`
`JENSEN AND CARLSEN: HUMAN GROWTH HORMONE IN£. COLI
`
`3
`
`BEQ 1037
`Page 3
`
`
`
`Table I. Correlations between the specific growth rate, the specific product expression rate, and
`the genetic stability.
`
`Amino Maximum specific
`extensions
`growth rate
`(h-')
`in X-hGH
`
`Specific
`expression rate
`(mg X-hGH/OD h L)
`
`Plasmid
`stability
`(No. gen.')
`
`Expression
`decline rate a
`(%expression/generation)
`
`MTEE
`MFEE
`MAE
`MEAE
`MEEE
`MLE
`Host
`
`0.69
`0.70
`0.82
`0.79
`0.82
`0.80
`0.83
`
`4.1
`3.9
`1.1
`1.2
`1.0
`1.4
`0
`
`:510
`:510
`85
`70
`80
`110
`
`11.5
`11.0
`2.8
`6.8
`4.6
`2.6
`
`' This is the number of generations with expression >90%.
`The maximum specific growth rate was determined in shake flasks using a semilogarithmic plot
`of the biomass concentration versus time in the beginning of the exponential growth phase. The
`hGH analysis of the same samples was used to calculate the specific expression rate.
`
`lated to the actual expression decline rate a for higher gen(cid:173)
`eration numbers [see Fig. 3(a)]. a (percent decline of
`expression per generation) can be calculated from the slope
`of the declining part of the curves in Figure 2. This means
`that a gives a good quantitative basis for a description of
`plasmid stability.
`There is as expected a good correlation between a and
`the synthetic burden on the cells (synthesis of X-hGH),
`measured as the relative growth rates of the cells contain(cid:173)
`ing plasmids to the plasmid free cells (f.L + / f.L -), as shown
`in Figure 3(b). The differences in the relative specific
`growth rates and in the plasmid stabilities can be explained
`by variations in the specific expression rates of the differ(cid:173)
`ent X-hGH's, as shown in Figure 3(c).
`One might find it surprising that these small changes in
`amino acid extensions of X-hGH lead to such marked dif(cid:173)
`ferences in expression efficiency. This phenomenon is
`most likely caused by differences in the rate of protein
`synthesis rather than in protein stability. If higher yields
`due to a better stability were observed with one type of
`hGH precursor compared to another, the burden of X-hGH
`synthesis put on the cells would still be similar, if the rate
`of synthesis of the two hGH precursors were identical. In
`that case, the correlation shown in Figure 3(c) would not
`be seen. The reason for the differences in the rate of syn(cid:173)
`thesis of the different hGH precursors can possibly be
`found in differences in the overall translation rate of
`X-hGH encoding mRNA. This could be caused either by
`different X-hGH mRNA concentrations (differences in
`mRNA stability) or by different rates of initiation of trans(cid:173)
`lation (different secondary mRNA structure).
`A good genetic stability of the X-hGH producing strain
`is required for production purposes and full expression
`must be maintained for at least 40 generations. For this
`reason, the high producing, but unstable MTEE-hGH and
`MFEE-hGH strains are excluded (as long as a strong con(cid:173)
`stitutive promoter is used). The strains MEAE-hGH and
`MAE-hGH seem to be more useful for large-scale processes.
`
`Development of a Fermentation Process for
`Production of MAE-hGH
`
`The fermentation process was developed on the basis of
`data obtained from shake flasks and batch and fed batch
`fermentation experiments. By comparing growth in differ(cid:173)
`ent defined and complex media we found that the highest
`biomass concentrations could be obtained in complex me(cid:173)
`dia (data not shown). Unless otherwise stated an optimized
`complex medium HKSO (see materials and methods) was
`used in the following studies.
`Expression and growth were relatively unsensitive to pH
`in the interval 6.8-7.6 and therefore pH 7.2 was used as
`the setpoint. It was very important to avoid oxygen limita(cid:173)
`tion, because otherwise the specific yield of the hGH pre(cid:173)
`cursor decreased significantly. All fermentations were
`carried out with an aeration rate of l VVM and the agita(cid:173)
`tion was automatically adjusted to ensure that DOT was
`greater than 20%. Fermentations were performed at 30°C,
`since the dominant part of the produced precursor re(cid:173)
`mained in a soluble form at this temperature in contrast to
`fermentations at 37°C where inclusion bodies were formed
`(data not shown).
`In the following, we focus mainly on the influence of
`the carbon source. Batch fermentations with glucose as the
`carbon source showed that high glucose concentrations led
`to reduced specific MAE-hGH yields as seen in Table II.
`The decrease in specific yields of MAE-hGH is very pro(cid:173)
`nounced with increasing glucose concentrations. An ex(cid:173)
`ample of a batch fermentation with 50 g glucose/L is
`shown in Figure 4 where yield of about 400 mg/L MAE(cid:173)
`hGH is obtained. The influence of glucose on MAE-hGH
`expression was further studied in fed batch fermentations.
`A balanced glucose addition with a constant glucose feed
`rate resulted in low glucose concentrations throughout the
`whole fermentation process, resulting in increased MAE(cid:173)
`hGH yields. A specific expression level comparable to that
`of optimized shake flask cultures (without glucose addi-
`
`4
`
`BIOTECHNOLOGY AND BIOENGINEERING, VOL. 36, JUNE 1990
`
`BEQ 1037
`Page 4
`
`
`
`Expression decline rate a
`• %per gen.
`
`r = -0.97
`
`10
`
`5
`
`0+----------r----------r-----------~
`100 Stable expression:
`0
`50
`no. of gen.
`
`(a)
`
`Expression decline rate a
`%per gen.
`
`r= -0.97
`
`10
`
`5
`
`O+----------r----------r-------------
`,..+!,..-
`1.0
`0.9
`0.8
`Relative specific growth rates
`
`(b)
`
`r = -0.98
`
`..
`
`1.0
`
`0.9
`
`0.8 +-------,-------.-------.----,.-
`3 mg X·hGH/I·h·OD
`0
`2
`Specific expression rate
`
`(c)
`
`(A) Relative expression decline rate a versus number of gen(cid:173)
`Figure 3.
`erations with stable expression. Stable expression is defined as >90% of
`the initial expression level. (B) Relative expression decline rate a as a
`function of the relative growth rate fJ. + / fJ.- of plasmid containing (fA.+)
`and of plasmid free cells (fA.-). (C) Decrease in growth rate as a function
`of X-hGH expression rates. (Data from table 1).
`
`Table II. Batch fermentations with different amounts of glucose.
`
`Carbon source
`added to the
`complex medium
`
`None
`30 g g1ucose/L
`50 g glucose/L
`
`MAE· hGH
`(mg/L)
`
`190
`420
`426
`
`OD
`
`22
`68
`111
`
`Specific
`MAE-hGH yield
`(mg/OD L)
`
`8.6
`6.2
`3.8
`
`tion) was obtained even at high biomass concentrations.
`An example of a fed batch fermentation is shown in
`Figure 5. An MAE-hGH yield of about 1000 mg/L was
`obtained.
`The glucose feed rate is essential to obtain high specific
`expression levels. In Figure 6 the results of fed batch fer(cid:173)
`mentations with varying constant feed rates of glucose are
`shown. Feed rates of more than about 4.2 g/L h resulted
`in specific yields of MAE-hGH which were similar to
`those obtained with glucose batch fermentations. The spe(cid:173)
`cific yields also decreased at low feed rates. This reduction
`could be the consequence of an insufficient amount of car(cid:173)
`bon source, which is necessary for generating energy for
`both maintenance metabolism and peptide synthesis. It
`could be speculated that an exponentially increasing glu(cid:173)
`cose feed rate, where the specific growth rate was kept
`constant, would lead to even higher yields than the con(cid:173)
`stant feed rate. This could however not be demonstrated
`(data not shown).
`An interesting relationship between the batch and fed
`batch results could be demonstrated. The specific yields of
`MAE-hGH were inversely correlated with the amount of
`base (NaOH), that was necessary to control pH during the
`fermentation, see Figure 7. In the fermentation process,
`the base (NaOH) was used mainly to neutralize the pro(cid:173)
`duced acetic acid, and therefore there was also an inverse
`correlation between the specific MAE-hGH yield and the
`amount of acetic acid produced. Furthermore, in fermenta(cid:173)
`tions where the produced acetic acid was reassimilated, the
`amount of acid (H2S04) used for pH control was nearly
`proportional to the amount of base that was used. The data
`obtained from a glycerol batch fermentation (50 g/L),
`which in contrast to a corresponding glucose batch fermen(cid:173)
`tation led to a lower acetic acid production, fited well in
`the overall picture as shown in Figure 7.
`The yield of approximately 1000 mg MAE-hGH/L, ob(cid:173)
`tained in a fed batch fermentation process, was still lower
`than the specific yields that could be obtained with the
`MTEE-hGH and MFEE-hGH strains, although, only in
`small volumes due to genetic instability. The genetic sta(cid:173)
`bility of the MTEE-hGH or MFEE-hGH strains could be
`improved by using an inducible promoter. Alternatively,
`one might improve the yield of the MAE-hGH strain even
`further, by finding a substrate component that by limitation
`would reduce cell growth but not MAE-hGH synthesis. As
`previously mentioned, the medium used in the process was
`complex, which lowered the degree of freedom with which
`to choose the second limiting substrate component after the
`
`JENSEN AND CARLSEN: HUMAN GROWTH HORMONE IN E. COLI
`
`5
`
`BEQ 1037
`Page 5
`
`
`
`MAE-hGH
`mg/1
`
`~-·----+- 00
`---~
`
`/
`
`/
`
`/
`
`I
`
`MAE-hGH
`
`I
`I
`I
`t
`I
`I
`I
`
`/ -I
`
`1000
`
`500
`
`/.
`
`/
`
`0 f'-';;_-=----,------ ---.--------.----
`10
`0
`20
`30 hours
`
`Figure 4. Production of MAE-hGH in a glucose batch fermentation (50 g
`glucose/L).
`
`00.
`I
`I
`I
`I
`I
`I
`I
`100 1
`I
`I
`I
`I
`I
`I
`50-1
`
`0
`
`00 t
`I
`I
`I
`I
`100 ~
`I
`I
`I
`I
`I
`50 i
`I
`I
`I
`I
`I
`ol
`
`MAE-hGH
`mg/1
`
`1000
`
`500
`
`/+
`
`A
`
`0
`
`0
`
`- - - - -+ - -+ - -+ - 00
`---~
`
`MAE-hGH
`
`/
`
`/
`
`/
`
`/
`
`I
`
`'
`
`I
`I
`I
`Start of I I
`Glucose
`I
`feed
`I
`t/
`
`10
`
`20
`
`30 hours
`
`Figure 5. Fed batch fermentation of MAE-hGH (glucose feed rate 3.6 g/L · h).
`
`MAE-hGH
`mg/1·00
`
`12
`
`10
`
`8
`
`6
`
`0
`
`2
`
`3
`
`4
`
`6 g/l·h
`5
`Glucose feed rate
`
`Figure 6. Specific yield of MAE-hGH in fed batch fermentations with differ(cid:173)
`ent constant feed rates of glucose.
`
`6
`
`BIOTECHNOLOGY AND BIOENGINEERING, VOL. 36, JUNE 1990
`
`BEQ 1037
`Page 6
`
`
`
`(i) fed batch glue. dos. = 2.6 g!l·h
`® fed batch glue. dos. = 3.6 g!l·h
`fed batch glue. dos. = 3.7 g!l·h
`@
`® fed batch glue. dos. = 4.2 g/l·h
`•
`batch glycerol, 50 g/1
`A batch glucose, 50 g/1
`
`MAE-hGH
`mg/I·OD
`
`G)
`
`0
`G)
`
`10
`
`8
`
`6
`
`4
`
`2
`
`NaOH
`0+-------,-------~------~-----------
`0
`100
`200
`300
`mmol/1
`
`Figure 7. Correlation between specific yields of MAE-hGH and the
`amount of base used for pH control in different fermentations.
`
`carbon source. By reducing the amount of yeast extract
`and omitting any phosphate addition, the medium was ex(cid:173)
`hausted for phosphate during the fermentation, and conse(cid:173)
`quently the biomass formation was blocked. By increasing
`the glucose feed rate after the phosphate limiting stage had
`been reached a prolonged MAE-hGH synthesis was ob(cid:173)
`tained (see Fig. 8). Using this cultivation procedure it was
`possible to obtain MAE-hGH yields of 2000 mg/L. The
`correct glucose feed rates and the correct timing of initia(cid:173)
`tion of the second step was critical in obtaining this yield.
`
`The Influence of Acetate on hGH Expression
`
`The kinetics of E. coli growth, acetate formation, and
`MAE-hGH expression in chemostat experiments are shown
`
`in Figure 9. Acetate was produced at high growth rates
`(#-'- > 0.4 h- 1
`) as described by Hollywood 1 and others. The
`concentration of MAE-hGH declined, when the dilution
`rate increased. At high dilution rates a rapid increase in
`glucose concentration was seen, indicating that the maxi(cid:173)
`mum growth rate of the E. coli strain under these condi(cid:173)
`tions had almost been reached. At D = 1.1 h -t the cells
`were washed out.
`The calculated specific production rate of MAE-hGH
`(qhoH) and of acetate (qHAc) is shown in Figure 10 as a
`function of the dilution rate. It was demonstrated that the
`production rate of MAE-hGH was independent of the
`growth rate of E. coli and furthermore it was not influ(cid:173)
`enced by the shift in metabolic state from nonacetate pro(cid:173)
`duction to acetate production. The lack of dependence of
`qhGH on 1-'- could be due to the expression system, which
`combined an efficient transcription (due to the strong pro(cid:173)
`moter) with an efficient translation (because the mRNA
`has an efficient ribosome binding siten). In such a system,
`a high rate of hGH synthesis could be upheld at lower
`growth rates even though the overall capacity of the pro(cid:173)
`tein synthesizing system was reduced. These results indi(cid:173)
`cate that production of MAE-hGH is not glucose catabolite
`repressed, which is further illustrated in Figure 11. As
`demonstrated the specific production rate on MAE-hGH
`was independent of the glucose concentration. These re(cid:173)
`sults are in accordance with the results from glucose batch
`fermentations with and without the addition of indole ace(cid:173)
`tic acid, a c-AMP analog that eliminates catabolite repres(cid:173)
`sion. 13 The MAE-hGH yield was found to be independent
`of the addition of indole acetic acid (data not shown). The
`fact that the growth rate did not influence the MAE-hGH
`production is most likely important in achieving the non(cid:173)
`growth-associated product formation in the second step of
`the fed batch fermentation (Fig. 8).
`These chemostat studies demonstrate that the glucose
`concentration and the metabolic state with or without ace-
`
`OD~
`I
`I
`150 1
`I
`I
`I
`I
`I
`100 ~
`I
`I
`I
`I
`I
`I
`50-j
`I
`I
`I
`I
`I
`ol
`
`MAE-hGH
`mg/1
`
`1500
`
`MAE-hGH
`
`.... •--+ OD
`
`__ .., __ .
`t
`2. step
`
`/
`
`/
`
`/
`
`I
`I
`I
`I
`
`10
`
`20
`
`30 hours
`
`1000
`
`500
`
`/
`Start of
`Glucose1
`feed 1
`l//
`
`0 _ _±_..
`0
`
`Figure 8. Fed batch fermentation with production of MAE-hGH in a two step
`process with (I) glucose limitation, and (2) phosphate exhaustion (glucose feed
`rate l. step = 3.6 g/L · h; glucose feed rate 2. step = 5.4 g/L · h).
`
`JENSEN AND CARLSEN: HUMAN GROWTH HORMONE IN£. COLI
`
`7
`
`BEQ 1037
`Page 7
`
`
`
`OO(e)
`glucose (.A.)
`HAc(\7)
`g/1
`
`hGH(O)
`ml/1
`
`e' .... -
`
`-
`
`- - ; - ..... -
`
`-
`
`100
`
`10 10
`
`50 0.5
`
`0
`
`0
`
`0
`
`•• 0.2
`
`/
`
`~ =+
`0.4
`
`0.6
`
`Glucose
`1-
`-..."'"' /:-~ /
`// ..,
`\
`I
`"'"'-...
`
`?
`
`\1 "'"'r\ HAc
`
`y ""00
`I
`I
`I
`~~hGH
`.---~-0
`hours-1
`0.8
`1.0
`
`Figure 9. Chemostat studies of the concentration of MAE-hGH, biomass, ace(cid:173)
`tate and glucose at different growth rates (see Materials and Methods section). The
`samples were taken at steady state conditions (p, = D).
`
`qhGH
`mg/I·OD·h
`
`•
`
`qHAc
`
`/
`
`/
`
`'i7
`/
`
`/
`
`'i7
`
`/
`\7/
`/
`
`?
`/
`. / ~ •
`/
`
`(
`
`•
`qhGH
`
`•
`
`•
`
`•
`
`/
`
`/
`
`/
`
`/
`
`/
`
`/
`
`/
`/\7
`
`'i7
`
`qHAc
`mg/I·OD·h
`~
`I
`100~ 1.0
`I
`I
`I
`I
`I
`I
`sol 0.5
`I
`I
`I
`I
`I
`ol
`
`0
`
`'\7"----
`0.2
`
`0
`1.0 hours-'
`Figure 10. The specific production rate of MAE-hGH (qhoH) and acetate (qHA,) as a func(cid:173)
`tion of the growth rate. Chemostat results (data from Fig. 9).
`
`/
`
`/
`
`0.4
`
`0.6
`
`0.8
`
`qhGH
`mg/l·h·OD
`
`1.0
`
`0.5
`
`o+-------------~----.-----------~----~---
`10
`50
`500
`100
`1000
`
`Glucose
`cone. mg/1
`
`Figure 11. The specific production rate of MAE-hGH (qhGH) as a function of the
`glucose concentrations in chemostat experiments (data from Fig. 9).
`
`8
`
`BIOTECHNOLOGY AND BIOENGINEERING, VOL. 36, JUNE 1990
`
`BEQ 1037
`Page 8
`
`
`
`tate formation per se are not critical to obtaining a high
`specific yield of MAE-hGH. On the other hand, the ace(cid:173)
`tate concentration could itself be important. In the chemo(cid:173)
`stat studies (Fig. 9), the absolute concentration of acetate
`even at high growth rates was low (1.6 g/L) compared to
`the values (approximately 10 g/L) measured during batch
`fermentations.
`To investigate qhGH at higher acetate concentrations, a
`study in glucose-limited chemostats at fixed dilution rates
`of 0.49 h- 1 (just above the dilution rate where acetate is
`produced) was performed. Sodium acetate was added in
`different concentrations to the medium. The results of the
`chemostat experiments are shown in Table III. As seen in
`Table III, acetate has an inhibiting effect on the growth of
`the MAE-hGH producing E. coli in concentrations 2::
`6 g/L, but already at a concentration of 2.4 g/L a signifi(cid:173)
`cant decrease in the specific production rate of MAE-hGH
`can be demonstrated. It should be noted that the effect of
`Na was studied in another experiment (see the data in
`Table V, 50 mM Na 2S04). Here it was demonstrated that
`Na in an amount equal to 8 g sodium acetate/L did not
`significantly influence the MAE-hGH yield in a fed batch
`fermentation. The negative effect of acetate on the growth
`of E. coli is likely to be an effect on the cell that also
`reduces the production rate of MAE-hGH. There is no evi(cid:173)
`dence for the mechanisms behind the effect and it is uncer(cid:173)
`tain whether the effect is caused by the undissociated as
`well as the dissociated acetic acid. By decreasing the pH,
`the inhibitory effect increased (data not shown), indicating
`that the undissociated acid had the most pronounced im(cid:173)
`pact on growth and product formation. This phenomenon
`could be due to enhanced energy uncoupling by proton
`leakage out of the cells as outlined by Jones. 14
`The concentration of acetate reached a maximum of
`about 10 g/L during a batch fermentation, whereas the
`acetate concentration was kept below 2-3 g/L in a fed
`batch fermentation with a constant glucose feed rate of
`3.5 g/L h. We would expect a specific yield of MAE-hGH
`in batch fermentations of about 50% of the yield in fed batch
`fermentations based on the decrease in qhGH (Table Ill).
`Earlier, it was shown that the addition of 50 g glucose/L
`in batch fermentations only resulted in yields of about 30%
`of the specific yield obtained in fed batch fermentations
`(without the second step with phosphate limitation). There(cid:173)
`fore, it is likely that other factors are important in deter(cid:173)
`mining the MAE-hGH yield as well.
`
`Influence of the Acid and Base Used for pH
`Control during the Fermentation Process
`
`With reference to Figure 7, a correlation between the
`amount of acid or base used during the fermentation and
`the specific MAE-hGH yield could be found. We have in(cid:173)
`vestigated whether this demonstrates any direct correlation
`between production of the hGH precursor and the presence
`of neutralized acid or base in the medium. NaOH was used
`as the base and H 2S04 as the acid, to control pH in all fer(cid:173)
`mentations. To test if the very nature of the acid or base
`had any influence on MAE-hGH-production, other acids
`and