See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/223457363
`
`Minimizing inclusion body formation during
`recombinant protein production in Escherichia
`coli at bench and pilot plant scale
`
`Article in Enzyme and Microbial Technology · March 2004
`
`Impact Factor: 2.32 · DOI: 10.1016/j.enzmictec.2003.10.011
`
`CITATIONS
`72
`
`4 authors, including:
`
`READS
`149
`
`Frank Hoffmann
`Martin Luther University Halle-Wittenberg
`
`Joop van den Heuvel
`Helmholtz Centre for Infection Research
`
`25 PUBLICATIONS 825 CITATIONS
`
`57 PUBLICATIONS 1,514 CITATIONS
`
`SEE PROFILE
`
`SEE PROFILE
`
`All in-text references underlined in blue are linked to publications on ResearchGate,
`letting you access and read them immediately.
`
`Available from: Joop van den Heuvel
`Retrieved on: 12 April 2016
`
`BEQ 1036
`Page 1
`
`

`
`Enzyme and Microbial Technology 34 (2004) 235–241
`
`Minimizing inclusion body formation during recombinant protein
`production in Escherichia coli at bench and pilot plant scale
`∗
`Frank Hoffmann1, Joop van den Heuvel, Nadine Zidek, Ursula Rinas
`Biochemical Engineering Division, GBF German Research Center for Biotechnology, Mascheroder Weg 1, 38124 Braunschweig, Germany
`
`Received 14 April 2003; accepted 14 October 2003
`
`Abstract
`
`Many recombinant proteins partially aggregate into inclusion bodies during production in Escherichia coli in batch culture on defined
`medium. Production on complex medium, however, effectively prevented inclusion body formation of a ␤-galactosidase-HIVgp41 fusion
`◦
`C under control of a temperature-inducible expression system. Cells
`protein for detection of anti-HIV antibodies which is produced at 42
`pre-conditioned by cultivation on complex medium before induction showed faster growth, higher product concentration and reduced
`inclusion body formation even when producing on defined medium. In contrast, for human basic fibroblast growth factor (hFGF-2)
`◦
`C. Here,
`produced under control of the phage T7-promoter, medium composition could not reduce inclusion body formation even at 30
`slow production in high-cell density fed-batch mode using a defined medium with limited glucose feeding enabled the accumulation of
`50 mg product per gram cell dry mass exclusively in the soluble cell fraction, resulting in a volumetric concentration of more than 4 g per
`litre hFGF-2. With the ␤-galactosidase fusion protein produced in fed-batch, over 100 mg of product per gram cell dry mass accumulated
`in the soluble cell fraction. With a cell density of 100 g cell dry mass per litre, this resulted in a volumetric concentration of 10 g per litre of
`soluble ␤-galactosidase-HIVgp41 fusion protein. Thus, two approaches to balance heterologous protein production and host physiology
`are presented, which fit the needs of lab bench or pilot plant, respectively.
`© 2003 Elsevier Inc. All rights reserved.
`
`Keywords: Inclusion bodies; Recombinant protein production; Escherichia coli; High-cell density; Cultivation; ␤-galactosidase fusion protein; hFGF-2
`
`1. Introduction
`
`Determinants of successful recombinant protein produc-
`tion, such as rate or duration of production and quality or
`stability of the product, depend on the physiology of the pro-
`ducer cell. This can be manipulated by metabolic engineer-
`ing of the host cell, by genetic engineering of the expression
`vector, and, last but not least, by process engineering.
`The factors that determine production rate, protein syn-
`thesis capacity and gene doses, vary with growth rate in op-
`posite directions. On the one hand, higher specific growth
`rates during production [1] or even before induction [2,3]
`can increase recombinant protein production, presumably
`via higher ribosome content [4], but on the other hand de-
`crease the cellular plasmid content [5]. Thus, the growth rate
`at which the production rate reaches a maximum depends
`on the system under study [5–10]. Generally, a fast produc-
`
`Corresponding author. Tel.: +49-531-6181-126;
`∗
`fax: +49-531-6181-111.
`E-mail address: URI@gbf.de (U. Rinas).
`1 Present address: Institut für Biotechnologie, Martin-Luther-Universität
`Halle-Wittenberg, Kurt-Mothes-Str. 3, 06120 Halle/Saale, Germany.
`
`0141-0229/$ – see front matter © 2003 Elsevier Inc. All rights reserved.
`doi:10.1016/j.enzmictec.2003.10.011
`
`tion terminates earlier than a slower production, sometimes
`giving higher final product yields with the slower systems.
`This was observed comparing strong versus moderate pro-
`moter systems [11], high versus low growth temperatures
`[12], and fast versus slow growth in batch and fed-batch
`systems [13]. Possible reasons include early attainment of
`a steady state of product concentration, or stress affecting
`the cells to an extent that prevents further production. Thus,
`taking sustainability into account, conditions that maximise
`the production rate do not necessarily optimise the overall
`process. Moreover, fast production of heterologous protein
`may exceed the capacity of the host cell to handle the protein
`properly. Many recombinant proteins tend to form inclusion
`bodies upon production in E. coli. Thus, process conditions
`must be found which balance heterologous protein produc-
`tion and host physiology to optimise the overall yield of
`active product.
`Medium composition influences recombinant protein
`production, giving higher growth rates and lower plasmid
`content in complex medium [14]. Addition of amino acids
`can reduce product degradation [15] by proteases induced
`during amino acid limitation [16]. Likewise, addition of
`casamino acids or peptone can enhance the stability or the
`
`BEQ 1036
`Page 2
`
`

`
`236
`
`F. Hoffmann et al. / Enzyme and Microbial Technology 34 (2004) 235–241
`
`synthesis of the recombinant protein [17]. Disadvantages of
`complex compounds include reduced solubility of glucose
`in peptone-containing feeding solution [18], or can result
`in reduced plasmid stability in case of amino acid sup-
`plementation [19]. Moreover, only a defined medium may
`be acceptable to guaranty the reproducibility and safety
`required for the production of protein pharmaceuticals.
`Thus, means to improve recombinant protein production
`on the bench scale may not fit the needs of large-scale
`processes.
`the production of two proteins, a big
`In this study,
`tetrameric ␤-galactosidase-fusion protein of about 480 kDa
`and the small monomeric human basic fibroblast growth
`factor (hFGF-2) of 18 kDa is optimized. The fusion pro-
`tein shows the same overall folding as the homologous
`␤-galactosidase of E. coli but displays an antigenic pep-
`tide from the HIV gp41 envelope protein on the surface
`[20]. Binding of anti-peptide-antibodies modulates the en-
`zymatic activity of the fusion protein, giving a sensitive
`homogenous assay for detection of HIV infection in blood
`samples [20]. The all-␤-sheet protein hFGF-2 structurally
`resembles interleukin-1␤ and shows very slow folding ki-
`netics [21]. Several therapeutic applications are suggested
`for hFGF-2, including wound healing or therapeutic angio-
`genesis [22]. We examined the influence of medium (com-
`plex versus synthetic) and cultivation regime (batch versus
`fed-batch) on production rate, sustainability of production,
`and prevention of in vivo aggregation of these two diverse
`proteins.
`
`2. Materials and methods
`
`2.1. Strains and plasmids
`
`Escherichia coli BL26 (Novagen, Madison, WI, USA), a
`lacZ-deleted derivative of BL21, was used as host for the
`plasmid pNF795gpC, coding for ␤-galactosidase carrying an
`antigenic peptide from HIV gp41 that is inserted at amino
`acid position 795 of ␤-galactosidase [20]. Expression of the
`fusion gene was under the control of the pR promoter of bac-
`teriophage lambda, repressed by the temperature-sensitive
`repressor cI857 and induced by temperature shift from 30
`◦
`to 42
`C.
`The IPTG inducible T7 polymerase expression sys-
`tem pET29c(+)hFGF-2 was constructed by cloning the
`Nde1-BamH1 hFGF-2 fragment of pJHLbFGF [23] into
`pET29c(+), which was also digested with Nde1 and BamH1.
`The strain BL21(DE3) and plasmid pET29c(+) for expres-
`sion were obtained from Novagen (Madison, WI, USA).
`
`2.2. Shake flask experiments
`
`Shake flask experiments with BL26 pNF795gpC were
`−1 bactotryp-
`done in 10 ml complex LB medium (10 g l
`−1 yeast extract, 7 g l
`−1 NaCl) supplemented with
`tone, 5 g l
`
`−1 ampicillin, or in a defined glucose mineral salt
`50 ␮g ml
`−1 glucose and supplemented
`medium [24] containing 10 g l
`−1 ampicillin. Cultures were inoculated with
`with 50 ␮g ml
`1% (v/v) of overnight preculture in the respective medium,
`grown to OD600 of 0.4–0.6 and induced by transfer from
`◦
`30 to 42
`C. For change of medium, cells were collected by
`centrifugation (15 min at 4000 × g), and the pellets were
`resuspended in fresh medium. For supplementation exper-
`iments, ten times concentrated solutions, i.e. 10× LB or
`−1 glucose, were added to 10% (v/v) of the culture
`100 g l
`volume.
`
`2.3. High-cell density cultivation and on-line analysis
`
`The preparation of the defined medium using glucose
`as carbon source and the high-cell density fed-batch cul-
`tivation strategy has been described before [24]. After
`termination of the batch growth phase, an exponentially
`increasing feeding rate of the concentrated glucose/salt so-
`lution aimed at a specific growth rate of µset = 0.12 h
`−1
`C and µset = 0.08 h
`◦
`−1 after temperature shift to
`at 30
`◦
`C for induction of BL26 pNF795gpC. For production
`42
`of hFGF-2 with BL21(DE3) pET29c(+)hFGF-2, the tem-
`◦
`perature was 30
`C, the set growth rate was kept constant
`at µset = 0.12 h
`−1, and induction was achieved by addition
`−1. Induction
`of IPTG to a final concentration of 0.5 mmol l
`of recombinant protein synthesis was started in both cul-
`tures when the OD600 reached 100. Details of the cultiva-
`tions are specified in the figure captions. Biomass, specific
`growth rate, and yields of biomass and carbon dioxide were
`estimated from on-line measured ammonia and glucose
`consumption rates and off-gas analysis data as described
`[25].
`
`2.4. Off-line analysis
`
`Cell growth was also monitored by turbidity at a wave-
`length of 600 nm of samples appropriately diluted with
`medium; cell dry mass was determined from washed cell
`pellets collected in preweight Eppendorf tubes after drying
`◦
`at 40
`C under vacuum. The specific growth rate was also
`calculated off-line from the change in time of the natural
`logarithm of the cell dry mass.
`For protein determination, cell pellets were collected
`stored at −70
`◦
`by centrifugation,
`C,
`resuspended in
`−1 sodium phosphate buffer (pH 7) to OD600 =
`50 mmol l
`4.5 and disrupted by sonication on ice. Soluble and insolu-
`ble cell fractions were separated by centrifugation (45 min at
`38,000× g and 4
`◦
`C), analyzed by SDS–PAGE, and specific
`protein concentrations quantified by densitometry assuming
`a constant content of cellular proteins of 550 milligram pro-
`tein per gram cell dry mass [26]. Relative productivity qP
`−1 h
`−1) was calculated as the change of the specific
`(mg g
`−1) of fusion protein multiplied with
`concentration c (mg g
`the optical density OD at the respective sampling times t
`(h), and normalized with the optical density at the time of
`
`BEQ 1036
`Page 3
`
`

`
`F. Hoffmann et al. / Enzyme and Microbial Technology 34 (2004) 235–241
`
`237
`
`induction ODind Eq. (1).
`ct2 · ODt2 − ct1 · ODt1
`qP = 1
`t2 − t1
`ODind
`
`(1)
`
`3. Results and discussion
`
`3.1. Influence of medium composition on the
`production of a β-galactosidase fusion protein
`in shake flask experiments
`
`A fusion protein of ␤-galactosidase carrying an antigenic
`peptide from gp41 of HIV was produced in E. coli BL26
`pNF795gpC from a temperature inducible expression sys-
`tem on complex medium containing yeast extract and protein
`hydrolysate, or on defined mineral salt medium with glucose
`as sole carbon source (Fig. 1). The specific product concen-
`tration obtained on defined medium was considerably higher
`than on complex medium (Fig. 1B), whereas the cell growth
`was slower (Fig. 1A). While the relative productivity was
`initially similar on both media, on defined medium growth
`and production was sustained for a longer period with a con-
`siderably higher final productivity (Fig. 1C). However, up
`to 40% of the fusion protein produced on defined medium
`accumulated in the insoluble cell fraction, compared to less
`than 5% final accumulation as inclusion body on complex
`medium (Fig. 1D).
`Addition of a concentrated glucose solution to the com-
`plex medium upon induction did not increase the specific
`product concentration (Fig. 1B), indicating that the low prod-
`uct concentration in complex medium is not due to depletion
`
`of carbon sources. This was also concluded by MacDonald
`and Neway [27], who replaced spent medium in perfusion
`cultures of recombinant E. coli. Addition of a concentrated
`solution of the complex medium components to the defined
`medium accelerated culture growth to the rate observed in
`complex medium (Fig. 1A), but strongly reduced the spe-
`cific concentration of the fusion protein (Fig. 1B). While
`this treatment hardly changed the relative productivity of
`total fusion protein (Fig. 1C), the final volumetric yield of
`soluble fusion protein increased by 30% (data not shown).
`The components of the complex medium were individu-
`ally added to defined medium upon induction, and bactotryp-
`tone was as efficient as the complete medium in promoting
`growth and reducing aggregation. Yeast extract was a little
`less effective, whereas NaCl or several inorganic nitrogen
`sources had no effect (data not shown).
`the
`To check the effect of pre-induction conditions,
`medium was changed by centrifugation of the culture and
`resuspension of the cells in fresh medium before transfer
`to the induction temperature. Cells producing on complex
`medium grew two to three times faster than cells produc-
`ing on defined medium, and had 40% lower product level
`(Table 1) similar to the unperturbed culture described above.
`The pre-induction growth medium had a strong effect on
`the product quality, and the effect was identical with both
`production media: “Skimped” cells,
`i.e. cells cultivated
`on the defined medium before induction, are compared to
`“coddled” cells that were cultivated on complex medium
`before induction, comparing those cultures that were af-
`terwards resuspended in the same medium for production.
`After 1.5 h of production, the fusion protein concentrations
`and the growth rates of the “skimped” cells were 20–25%
`
`Fig. 1. Profiles of growth, production of the ␤-galactosidase-HIVgp41 peptide fusion protein, and aggregation on medium of different compositions.
`(A) Growth after induction was followed by the optical density at 600 nm, (B) specific concentration of the ␤-galactosidase-HIVgp41 peptide fusion
`as determined from densitometry of Coomassie stained SDS–PAGE gels, (C) relative productivity, and (D) percentage of aggregated fusion protein,
`determined from SDS–PAGE analysis of disrupted cells separated by centrifugation. Medium: (䊐) defined medium, (䊏) defined medium plus 10% (v/v)
`10× concentrated solution of complex LB medium upon induction, (䊊) complex LB medium, (䊉) complex LB medium plus 10% (v/v) glucose solution
`−1) upon induction.
`(100 g l
`
`BEQ 1036
`Page 4
`
`

`
`238
`
`F. Hoffmann et al. / Enzyme and Microbial Technology 34 (2004) 235–241
`
`Medium
`
`Table 1
`Effect of medium before and after induction on growth, production and aggregation of the ␤-galactosidase-HIVgp41 peptide fusion protein
`−1)
`Specific growth rate (␮ h
`◦
`
`◦
`
`C
`
`Fusion protein
`−1a)Total (mg g
`
`
`
`Before induction
`
`After induction
`
`Before induction 30
`
`C
`
`After induction 42
`
`Insoluble
`−1a
`mg g
`19 ± 6
`1.51 ± 0.01
`ndd
`0
`≈10
`2.8 ± 1
`31 ± 9
`0.64 ± 0.05
`14 ± 3
`1.17 ± 0.04
`ndd
`0
`≈20
`4.6 ± 2
`23 ± 5
`0.50 ± 0.03
`The specific growth rate was determined from the optical density at 600 nm, specific concentration of fusion protein from densitometry of SDS–PAGE
`of samples taken 1.5 h post-induction.
`a ␤-Galactosidase-HIVgp41 peptide fusion protein as milligram per gram cell dry mass.
`b Insoluble ␤-galactosidase-HIVgp41 peptide fusion protein as percentage of total fusion protein.
`c 95% confidence interval from three parallel experiments.
`d Below detection limit.
`
`Complex
`
`Glucose salt
`
`Complex
`Glucose salt
`
`Complex
`Glucose salt
`
`0.72 ± 0.02c
`
`0.38 ± 0.01
`
`%b
`
`lower than those of “coddled” cells (Table 1), using both
`defined or complex media for resuspension in the produc-
`tion phase. Moreover, despite lower total concentration of
`fusion protein, “skimped” cells showed higher tendency to
`form inclusion bodies when producing the fusion protein
`on defined medium (Table 1).
`
`3.2. Production of the β-galactosidase fusion protein in
`high-cell density cultivation
`
`For pilot-scale production of the ␤-galactosidase fusion
`carrying the antigenic peptide from gp41 of HIV, a fed-batch
`protocol established for the production of recombinant pro-
`teins was used. After a batch phase on defined medium, a
`glucose solution was fed with an exponentially increasing
`rate aiming at a specific growth rate of µset = 0.12 h
`−1,
`which corresponds to about 25% of the maximum growth
`◦
`rate at 30
`C. When reaching an OD600 of 100, the feeding
`was restarted with µset = 0.08 h
`−1 and the temperature was
`◦
`raised to 42
`C to induce the fusion protein synthesis.
`After induction, the specific product concentration in-
`creased permanently, finally reaching 120 mg of total fusion
`protein per gram cell dry mass, corresponding to 20% of the
`total cell protein (Fig. 2A). These high amounts of fusion
`protein were mainly maintained in the soluble cell fraction,
`as less than 15% of the fusion protein produced was aggre-
`gated at the end of the cultivation (Fig. 2A). In this high-cell
`density cultivation, the maximum specific relative produc-
`tivity, calculated as described in Section 2 for comparison
`with productivities obtained in shake flask cultures, reached
`about 70 mg per hour and gram cell dry mass four hours after
`induction and declined afterwards (Fig. 2A). This was about
`six times lower than in the shake flask experiments (Fig. 1C)
`due to the lower growth rate and hence lower protein pro-
`duction rate in the glucose limited fed-batch cultivation.
`Growth was monitored off-line by cell dry mass and
`on-line by balancing the alkali consumed for pH mainte-
`nance (Fig. 2B). The specific growth rate was close to the
`set value of µset = 0.12 h
`−1 during the fed-batch phase
`before induction. After induction, the specific growth rate
`
`Fig. 2. Production of the ␤-galactosidase-HIVgp41 peptide fusion protein
`in high-cell density cultivation. After batch (initial glucose concentration
`−1, 13.25 h) and first fed-batch phase (µset = 0.12 h
`−1 for 10.75 h)
`30 g l
`C, the feeding rate was adjusted (µset = 0.08 h
`◦
`−1) starting 25 min
`at 30
`before induction. When the new feeding rate was reached 15 min later,
`the temperature shift was initiated after 3 min. The induction temperature
`◦
`of 42
`C was reached after 7 min. The time is given relative to this time
`point of induction tind. (A) Product accumulation: The specific fusion
`protein concentration was estimated densitometrically from Coomassie
`stained SDS–PAGE in the total cell extract (䊏) and in the insoluble cell
`fraction (䊐), separated by centrifugation of disrupted cells. The relative
`productivity ((cid:10)) was determined as in Fig. 1C from Eq. (1). (B) Growth
`profile: cell density (䊏, —) and specific growth rates (䊐, . . .) were
`determined off-line (䊏, 䊐) from cell dry mass and culture volume, and
`estimated on-line (—,. . .) from alkali consumption and bioreactor weight
`as described in [25]. (C) Yields of biomass (—) and carbon dioxide on
`glucose (. . ., in mol CO 2 per mol carbon in glucose) calculated from the
`rates of alkali consumption, carbon dioxide evolution and glucose feeding
`as described in [25].
`
`BEQ 1036
`Page 5
`
`

`
`F. Hoffmann et al. / Enzyme and Microbial Technology 34 (2004) 235–241
`
`239
`
`was initially high due to overfeeding, and then gradually
`declined to the set value of µset = 0.08 h
`−1 (Fig. 2B). The
`biomass yield with respect to the glucose dropped from 0.43
`◦
`in the fed-batch phase at 30
`C to 0.35 g per gram glucose
`after induction (Fig. 2C). During the production phase, the
`−1 ten
`biomass yield decreased further, reaching 0.22 g g
`hours after induction. Conversely, the yield of carbon diox-
`−1
`ide from glucose, YCO2/S, increased from 0.34 mol mol
`−1
`(mol CO2 per mol carbon in glucose) to 0.6 mol mol
`upon induction (Fig. 2C). Later, YCO2/S decreased in par-
`allel to the specific growth rate. Suddenly at four to five
`hours after induction, the YCO2/S rose considerably from
`−1, concomitant to the decline of the
`0.46 to 0.76 mol mol
`fusion protein productivity (Fig. 2A).
`However, the production process could be sustained for
`ten hours with this protocol concomitant to cell growth. Fi-
`−1
`nally, i.e. after 34 h of cultivation, a cell density of 100 g l
`was obtained (Fig. 2B). Continued growth and production
`−1.
`resulted in a volumetric product concentration of 12 g l
`Moreover, most of the fusion protein was soluble with
`this protocol, even on defined medium. Thus, the high-cell
`density process is the cultivation regime of choice for the
`large-scale production of soluble ␤-galactosidase fusion
`protein carrying the antigenic peptide from the HIV gp41
`surface protein.
`
`3.3. Production of hFGF-2 in high-cell density
`cultivation
`
`Using a similar temperature-inducible expression system,
`the human basic fibroblast growth factor (hFGF-2) was suc-
`cessfully produced up to 50 mg per gram cell dry mass in
`the same high-cell density cultivation procedure as described
`above for the production of the ␤-galactosidase fusion pro-
`tein [28]. As much as 60% of the product, however, were
`found in the insoluble cell fraction even with the high-cell
`density fed-batch protocol [28].
`To circumvent the higher propensity for inclusion body
`formation of recombinant proteins at high temperatures,
`the temperature-independent T7 expression system was em-
`ployed for production of hFGF-2. Using the T7-system, up
`to 140 mg hFGF-2 per gram cell dry mass were obtained
`◦
`in shake flask experiments at 30
`C. However, about 50%
`were still found in the insoluble cell fraction. Moreover, all
`hFGF-2 produced was in the insoluble cell fraction upon pro-
`◦
`duction at 37
`C using the T7-system (data not shown). The
`medium composition (glucose/salt versus complex medium)
`had no influence on the partitioning of hFGF-2 in the sol-
`◦
`uble or insoluble cell fraction even when grown at 30
`C
`in shake flask experiments. However, inclusion body forma-
`tion could be effectively prevented during production in a
`◦
`carbon-limited fed-batch procedure at 30
`C and a reduced
`growth rate of µset = 0.12 h
`−1 throughout the entire cul-
`tivation. Using this protocol, 50 mg hFGF-2 per gram cell
`dry mass was present as soluble protein and no hFGF-2 was
`detected in the insoluble cell fraction (Fig. 3A).
`
`Fig. 3. Production of hFGF-2 in high-cell density cultivation. After batch
`◦
`−1, 14 h), the feeding
`C (initial glucose concentration 30 g l
`phase at 30
`was started aiming at a specific growth rate of µset = 0.12 h
`−1. The
`◦
`C and 12 h after feeding start the production of
`temperature was kept at 30
`−1 IPTG without
`hFGF-2 was initiated through the addition of 0.5 mmol l
`changing the feeding protocol. (A) Product accumulation: the specific
`protein concentration was estimated densitometrically from Coomassie
`stained SDS–PAGE in the total cell extract (䊐) and in the soluble cell
`fraction (䊊), separated by centrifugation of disrupted cells. Growth profile:
`cell density (䊏) was determined off-line from cell dry mass measurements
`and estimated on-line from alkali consumption and bioreactor weight (—).
`(B) Yields of biomass (thin lines) and carbon dioxide on glucose (thick
`lines, in mol CO2 per mol carbon in glucose) calculated from the rates
`of glucose feeding, alkali consumption and carbon dioxide evolution of
`two independent cultivations (—,- - - - -).
`
`The yield of carbon dioxide on glucose increased lin-
`early during the fed-batch phase with this construct as a
`function of the cell density (Fig. 3B). Deviation from the
`linear trend coincide with addition of antifoam, therefore
`reflecting physical rather than physiological changes. Al-
`though the biomass yield on glucose started to decrease
`−1
`immediately after induction and approached 0.2–0.3 g g
`−1
`four hours later (Fig. 3B), a cell density of about 85 g l
`was finally reached (Fig. 3A). This resulted in a volumetric
`−1, which was exclusively
`hFGF-2 concentration of 4.2 g l
`soluble.
`With temperature-induced production of hFGF-2, a vol-
`−1 of hFGF-2 was reached.
`umetric concentration of 2.8 g l
`This low volumetric yield was caused by the stress response
`resulting from temperature-induced production of
`the
`growth factor [29,30], which was causing a severe growth
`inhibition and consequently little biomass increase after
`temperature induction. The specific product concentration
`[28] was not lower than obtained with the T7-expression
`system presented here, however, only 40% of the product,
`
`BEQ 1036
`Page 6
`
`

`
`240
`
`F. Hoffmann et al. / Enzyme and Microbial Technology 34 (2004) 235–241
`
`−1, was soluble after temperature-
`corresponding to 1.2 g l
`induction. As all hFGF-2 was soluble with the T7 system,
`the volumetric concentration of soluble product was in-
`creased by 250% compared to the temperature-inducible
`system in fed-batch culture.
`
`4. Conclusions
`
`4.1. Pre-conditioning of producer cells
`
`Higher pre-induction growth rates can result in higher
`levels of recombinant protein [2]. Likewise, coddled cells
`pre-conditioned by cultivation in complex medium before
`induction accumulated higher levels of the ␤-galactosidase
`fusion protein than “skimped” cells cultivated in defined
`medium, even when the medium was changed upon induc-
`tion. This can be explained by the higher ribosome con-
`tent at higher growth rates [4]. Moreover, the post-induction
`growth of pre-conditioned cells was faster. The production
`of the recombinant protein hence impaired pre-conditioned
`cells to a lesser extend than “skimped” cells. Finally, the
`pre-conditioned cells formed less inclusion bodies when pro-
`ducing on defined medium than those cells cultivated in de-
`fined medium from the start, although the concentration of
`the fusion protein was higher with pre-conditioned cells.
`Thus, the performance of E. coli during temperature-induced
`production of the ␤-galactosidase-HIVgp41-fusion protein
`depends on the overall cell physiology, which in part is de-
`termined before induction.
`Also pre-conditioning with the reducing agent dithiothre-
`itol is effective in enhancing the activity of a chlorampheni-
`col acetyltransferase fusion protein [31]. This treatment,
`however, imposes additional stress on the cells, whereas cul-
`tivation on complex medium is beneficial to product and
`producer.
`
`4.2. Balancing growth and production for improved
`product quality
`
`Recombinant protein production is strictly proportional to
`growth when limited addition of the substrate controls the
`growth rate [1]. However, an inverse correlation of growth
`rate and biologically active product concentration was found
`when cells grew with the maximum growth rate as a function
`of different medium compositions, both with periplasmic
`␤-lactamase [14] and with our cytoplasmic ␤-galactosidase
`fusion protein. In contrast, some proteins show higher syn-
`thesis or stability when produced on medium allowing rapid
`growth such as complex medium [17]. This was observed
`also for temperature-induced synthesis of an insulin-B-chain
`fusion which is produced exclusively in form of inclusion
`bodies [32]. Likewise, addition of the amino acid that may
`become depleted upon recombinant protein production can
`lead to higher yield of the product than cultivation in simple
`glucose/mineral salt medium [33].
`
`The ␤-galactosidase fusion protein aggregated partially
`when produced on defined medium, whereas it was soluble
`when produced on complex medium. A stronger tendency
`to form inclusion bodies on less rich medium has also been
`reported for some other recombinant proteins [34,35]. With
`T7-controlled hFGF-2 production, no positive effect of rich
`medium for preventing inclusion body formation was ob-
`served. Thus, for some proteins, accelerated growth after ad-
`dition of complex medium compounds upon induction can
`reduce aggregation.
`
`4.3. Production in high-cell density cultivation
`
`Alternatively, with limiting addition of glucose, the frac-
`tion of aggregated ␤-galactosidase fusion proteins could be
`reduced below 15%. Active production was sustained for
`ten hours, giving a final total volumetric product concentra-
`−1 corresponding to more than 10 g l
`−1 soluble
`tion of 12 g l
`product. Likewise, 50 mg hFGF-2 per gram cell dry mass
`accumulated completely in the soluble cell fraction with the
`fed-batch protocol, in contrast to the results obtained in batch
`experiments. Moreover, cell growth was maintained during
`the production phase. This increase in biomass contributed
`considerably to the final volumetric product yield, leading
`−1 of soluble hFGF-2.
`to accumulation of 4.2 g l
`Thus, two procedures to reach a balanced synthesis of
`recombinant proteins that reduce inclusion body forma-
`tion were employed in this study. Pre-conditioning cells by
`growth on complex medium or addition of complex medium
`components upon induction can fit the needs when working
`on the lab bench, while extended synthesis at reduced rates
`in fed-batch mode is adequate for large-scale production.
`
`Acknowledgments
`
`This work was supported in part be the European Commu-
`nity Grant BIO4CT98-0157. Helpful comments of Antonio
`Villaverde on the manuscript are gratefully acknowledged.
`
`References
`
`[1] Shin CS, Hong MS, Kim DY, Shin HC, Lee J. Growth-associated
`synthesis of
`recombinant human glucagon and human growth
`hormone in high-cell-density cultures of Escherichia coli. Appl
`Microbiol Biotechnol 1998;49:364–70.
`[2] Curless C, Pope J, Tsai L. Effect of preinduction specific growth rate
`on recombinant alpha consensus interferon synthesis in Escherichia
`coli. Biotechnol Prog 1990;6:149–52.
`[3] Curless CE, Pope J, Loredo L, Tsai LB. Effect of preinduction
`specific growth rate on secretion of granolocyte macrophage colony
`stimulating factor by Escherichia coli. Biotechnol Prog 1994;10:467–
`71.
`[4] Bremer H, Dennis PP. Modulation of chemical composition and
`other parameters of the cell by growth rate. In: Neidhardt FC, editor
`in chief. Escherichia coli and Salmonella—Cellular and molecular
`biology. 2nd ed. Washington, DC: ASM Press; 1996. p. 1553–69.
`
`BEQ 1036
`Page 7
`
`

`
`F. Hoffmann et al. / Enzyme and Microbial Technology 34 (2004) 235–241
`
`241
`
`recombinant
`cultivation of
`[5] Seo J-H, Bailey JE. Continuous
`Escherichia coli: existence of an optimum dilution rate for maximum
`plasmid and gene product concentration. Biotechnol Bioeng 1986;
`28:1590–4.
`[6] Kim SS, Kim EK, Rhee JS. Effects of growth rate on the production
`of Pseudomonas fluorescens lipase during fed-batch cultivation of
`Escherichia coli. Biotechnol Prog 1996;12:718–22.
`[7] Curless CE, Forrer PD, Mann MB, Fenton DM, Tsai LB. Chemostat
`study of kinetics of human lymphokine synthesis in recombinant
`Escherichia coli. Biotechnol Bioeng 1989;34:415–21.
`[8] Fu J, Wilson DB, Shuler ML. Continuous high level production
`and excretion of a plasmid-encoded protein by Escherichia coli in a
`two-stage chemostat. Biotechnol Bioeng 1993;41:937–46.
`[9] Turner C, Gregory ME, Turner MK. A study of the effect of
`specific growth rate and acetate on recombinant protein production
`of Escherichia coli JM107. Biotechnol Lett 1994;16:891–6.
`[10] Hellmuth K, Korz DJ, Sanders EA, Deckwer W-D. Effect of growth
`rate on stability and gene expression of recombinant plasmids during
`continuous and high cell density cultivation of Escherichia coli TG1.
`J Biotechnol 1994;32:289–98.
`[11] Kramer W, Elmecker G, Weik R, Mattanovich D, Bayer K. Kinetic
`studies for the optimization of recombinant protein formation. Ann
`NY Acad Sci 1996;782:323–33.
`[12] Surek B, Wilhelm M, Hillen W. Optimizing the promoter and
`ribosome binding sequence for expression of human single chain
`urokinase-like plasminogen activator
`in Escherichia coli and
`stabilization of the product by avoiding heat shock response. Appl
`Microbiol Biotechnol 1991;34:488–94.
`[13] Strandberg L, Andersson L, Enfors S-O. The use of fed batch
`cultivation for achieving high cell densities in the production of a
`recombinant protein in Escherichia coli. FEMS Microbiol Rev 1994;
`14:53–6.
`[14] Seo J-H, Bailey JE. Effects of recombinant plasmid content on growth
`properties and cloned gene product formation in Escherichia coli.
`Biotechnol Bioeng 1985;27:1668–74.
`[15] Harcum SW, Ramirez DM, Bentley WE. Optimal nutrient feed
`policies
`for heterologous protein production. Appl Biochem
`Biotechnol 1992;34/35:161–