ANNALS OF THE NEW YORK ACADEMY OF SCIENCES
`Volume646
`
`RECOMBINANT DNA TECHNOLOGY I
`
`I (
`
`Edited by Ales Prokop and Rakesh K Bajpai
`
`The New York Academy of Sciences
`New York, New York
`1991
`
`/
`
`BEQ 1005
`Page 1
`
`

`
`. ·.
`
`:
`
`-··
`
`Copyright c 1991l!y the New YorlcAcad~my of SCiences. All rights reserv~d. Under the proviswns of the United
`States Copynglil Act of 1976, individual readers .of the A,n.nals are peimitted to make fair use of the material in
`them for teaching and researi:h. Permiiswn is granJed to quote from _the_ Annals provided that the customary
`acknowledgment is made of the sourci. Materiol in the Annals may be republished only by permission of the
`Academy. Address inq~iries to the &ecutive Editor at the New York Academy of Scipu;es.
`Copying cee8: For eill:h copy_ ofan article made /;Jeyo!la t}Je free copyjng pdrmitted under Section 107 or 108 of
`the 1976 CofJY!ight Act, a fee should be paid through the Copyright Clearance Center Inc.~ _27 Coligrqs SL, Salem,
`MA 01970. For articles of mare than 3 page,s; the copying fee is·$1. 75.
`.
`.
`
`@ The paper used in this pubiication m«ts the minimum reJJ¢n!lll!!rUS of American N(Jtional SUJ11Jiard for
`Infortn_ation Sciences....:...Permanence of Paper for .Prilued Libr9ry 1,1aterjals, ANSI Z39.48-19IJ.!I.
`. ~
`.
`' .
`·.
`.
`
`.
`.
`library or Congress Cala'l_ogiog·itr·Publication l)ata
`..
`.
`.
`. .
`.
`. .
`Reco~bjnant bNA.t~.cbnology I / edited py Ale5 Prokol? and· Rakesh K..
`· BaJpat.
`.
`.
`.
`· · ,
`·
`.
`·
`:p , . em. --:--(Annal~ o( the New York Aeademy of Sciences, ISSN
`0077-8923; v. 646).
`. . .
`.
`.
`.
`. .ISBN 0-8976~-~73-9 (cloth: a1k.: paper).- ISBN.0-89766-674-7
`{pbk;.:'_alk. paper)
`.·
`.
`.
`· .
`. . .
`·
`:
`· · · . - ·1: Genetic engineering. 2. RccombioantD~A. 1: Prokep, f\.le~.
`II. Bajpai, Rakesh K. III. Title: Reco~bin~t PJ:'lA tec~nology I. .
`·IY.' Title: R C}conibinant DNAtecb~ology one.: · V. Series. . ·
`·[DNLM: 1. BiotecQ.nol<>g}'--.;..congresses. 2. -pNA, Recbmbinant-
`-3. Genetic Engineering-congresses-; W l AN626YL v:
`·. oopgre.sSes.
`646 ! QU 58 R3124Pl
`.
`.
`Qll.NS' vol. 646 ·
`[TP248.6] ...
`500 5--'-dc20
`· r 600'_.6sf :
`·. · .·
`bNLMt:QLC
`for. Li!>raiy-of Congr~ss
`. .
`:
`...
`... (
`.. '
`
`. ·
`...
`
`..
`
`'
`
`.
`...
`. · . . . . ..
`
`. . _ .
`
`-- . ·.
`
`'
`
`..
`. ..
`
`. . ...
`
`·.
`
`. : ,. ..
`
`.. . .. - ~
`
`.
`
`..
`
`. ,,
`..... .
`
`..-.
`
`,. -
`
`·. : . .... . . , : .. : .
`. . ·.··
`..
`
`.
`·sp
`. .
`, ...
`'Printed in the Uiz.iied States. of ~tka . · · . . . ·
`·isBN ~66;-673.9 ( clo.tli). ·. . · · .. ::_·. ,:
`: ·:
`·· ·. . : I~BN o;s97664J7-a.? (gaper): :·
`.
`.
`ISSN.4"rZ7 ~89~ ·::
`. ·. ··
`
`.·· ... · .
`
`. .. .
`
`• • • ..
`
`.
`
`4.
`
`.·
`
`BEQ 1005
`Page 2
`
`

`
`High Cell Density Fermentation of
`Recombinant Escherichia coli with
`Computer-Controlled Optimal
`Growth Rate
`
`W. A. KNORRE,0 W.-D. DECKWER,b D. KORZ,b
`H.-D. POHL,0 D. RIESENBERG,0 A. ROSS,b E. SANDERS,b
`AND V. SCHULZ0
`0 Zentralinstitut for Mikrobiologie und experimentelle 1herapie (ZIMET)
`0 6900 lena, 1huringen, Germany
`hGesellschaft fiir Biotechnologische Forschung mbH (GBF)
`W 3300 Braunschweig Niedersachsen, Germany
`
`In recent years interest in the production of recombinant DNA products, such as
`enzymes and pharmaceuticals, has been growing. High cell density cultivation has
`been one of the most effective ways to increase cell as well as product yield. Basic
`work in this field was carried out by Bauer and coworkers (1974-1981) with
`nonrecombinant strains of Escherichia coli. 1
`2
`•
`A variety of process strategies have been developed in fed-batch fermentation of
`E. coli. 3 Common goals are to control the oxygen demand within the oxygen transfer
`capabilities of the fermentor and to avoid the accumulation of acetate and ethanol.4
`These goals can be met by feeding the carbon source to achieve C-limited growth,
`reducing temperature to decrease the growth rate, or increasing the oxygen transfer
`capability of the fermentor using oxygen as the sparging gas. Cell densities of about
`110 g/L have been reported for a process-controlled fed-batch plus oxygen mode for
`cultivating nonrecombinant E. coli in a 2.5-L fermentor.5 An impressive application
`of rDNA products is the production of alpha-consensus interferon by recombinant
`E. coli. Fieschko and Ritch6 reported product concentrations of 5.5 g/L from 65 g/L
`cell dry mass.
`
`ADVANTAGES OF HIGH CELL DENSI1Y FERMENTATION
`
`In general, the main advantages of high-density cultivation are: reduced fermen(cid:173)
`tor and closed system volume, improved space time yield (volumetric productivity),
`reduced medium costs, reduced volume in primary downstream processing, frequent
`omission of concentration steps, and reduced plant and operating costs.
`
`HOST STRAIN E. COU
`
`For decades, E. coli has played an important role in molecular biological work,
`which explains its use as the host strain for the majority of protein productions from
`rDNA. For example, insulin, hGH, alpha2-interferon, alph~-interferon, and rennin
`300
`
`BEQ 1005
`Page 1
`
`BEQ 1005
`Page 3
`
`

`
`KNORRE et al.: IDGH CELL DENSI1Y FERMENTATION
`
`301
`
`are available from cultivations of recombinant E. coli. For none of them has a high
`cell density process been used.
`Of great interest is the control of growth rate because it strongly influences the
`formation of both products and inhibitory metabolites. Acetate formation is re(cid:173)
`ported to increase drastically when the growth rate exceeds 0.35 h- 1 (defined
`medium) and 0.2 h- 1 (complex medium). 7
`Seo and Bailey8 showed the existence of an optimum dilution rate (growth rate)
`for 13-lactamase production in continuous cultivation. This finding was not confirmed
`in batch experiments when growth rate control was changed by altering growth
`medium. Despite this, Riesenberg et a/. 9 reported an optimum growth rate in batch
`production of alpha,-interferon.9
`
`I SPECIFIC
`I GROWTH
`RATE
`
`ITU£
`I
`I
`
`I I
`I I
`I I
`I I,
`1 I
`I I
`I I
`I
`
`STIRRER SPEED
`
`0
`
`FIGURE 1. Scheme of the ZIMET HDF 30/450: t, = end of lag-phase; t2 = start of p02-control
`via stirrer speed; t3 = exhaustion of glycerol; t4 = start of glucose feeding; t5 = start of
`exponential shift in stirrer speed; and t6 = end of exponential shift in stirrer speed.
`
`ZIMET IDGH CELL DENSI'IY FERMENTATION 30/450
`
`In the ZIMET a high cell density fermentation (HDF) process for a glucose/
`mineral salt medium allows growth of a recombinant E. coli strain (TG 1, pBB 210)
`up to a cell density of 60 g/L in a 30- and 450-L Chemap fermentor. 9 Except for the
`feeding of glucose as a carbon source and of aqueous ammonia for pH control, there
`was no need for the feeding of other nutrients and for the supply of oxygen-enriched
`air. FIGURE 1 schematically illustrates this process with a batch phase with glucose
`
`BEQ 1005
`Page 2
`
`BEQ 1005
`Page 4
`
`

`
`302
`
`ANNALS NEW YORK ACADEMY OF SCIENCES
`
`consumption and a feed-batch phase with glucose feeding. In the fed-batch phase the
`p02 was kept at 20% of saturation via closed-loop controls with two variables,
`namely, stirrer speed and feeding rate of glucose. The fed-batch mode prevented
`significant accumulation of acetate and other metabolic byproducts. The recombi(cid:173)
`nant E. coli expressed alpha1-interferon constitutively with a higher efficiency at a
`lower specific growth rate(~,= 0.17 h- 1
`) than at the maximal specific growth rate
`(ltmu = 0.45 h-') (FIG. 2). Therefore, after reaching a suitable cell density with
`growth at ltmu• the culture was forced to grow at the optimal specific growth rate, It .,
`by open-loop control for agitation directing the input of oxygen and hence the sup.Pfy
`of glucose. The stirrer speed was increased according to an e-function profile.
`
`ZIMET/GBF-HIGH CELL DENSITY FERMENTATION 70/1500
`
`To overcome the well-known disadvantages of high density processes, GBF and
`ZIMET in 1989 developed a special HDF process for E. coli that produces more than
`
`CH£HAP 30
`
`lUll
`toto
`
`INTERfERON ALPHA 1
`
`100
`
`x.s
`
`g/l
`
`80
`
`70
`
`60
`
`so
`
`40
`
`30
`
`20
`
`10
`
`0
`
`0
`
`4
`
`8
`
`12
`
`16
`
`20
`
`24 h
`
`FERIEHTATI<Jf Til£
`
`FIGURE 2. ZIMET HDF 30 process: Kinetics of glucose, biomass, and alpha1-interferon.
`
`BEQ 1005
`Page 3
`
`BEQ 1005
`Page 5
`
`

`
`KNORRE et aL: HIGH CELL DENSI1Y FERMENTATION
`
`303
`
`BATCH
`
`FED-BATCH
`PHASE 1
`
`PHASE 2
`
`log BI<J4ASS X..x
`
`116 g/1
`
`100
`~
`p02
`
`50
`
`0
`
`0
`
`5
`
`10
`
`15
`fEIMNT A TI lW TI 1£
`
`20
`
`25
`
`30 h
`
`FIGURE J. ZIMET /GBF HDF process. Kinetics of biomass, glucose, and p02 in the 70-L pilot
`fermentor.
`
`100 g/L cell dry mass. The special cultivation strategy prevents oxygen limitation and
`hence the accumulation of inhibitory metabolites such as acetate and ethanol. (See
`FIG. 5 for acetate kinetics.)
`FIGURE 3 shows a typical time course of the HDF process with two exponential
`growth phases. The first is characterized as a batch phase under maximum growth
`rate corresponding to the media and process parameter, that is, temperature and pH.
`The subsequent fed-batch operation can be subdivided into IJ.-controlled phase 1,
`implemented as a dissolved oxygen control loop via nutrient feed and a J.L control
`loop via agitation rate, and a second period (phase 2) with decreasing growth rate at
`the maximum attainable oxygen transfer rate.
`In the example shown, growth rate was switched from IJ.max .,. 0.45 h- 1 after 12
`hours to a lower constant value of IJ. .,. 0.11 h-', which corresponded to an optimal
`growth rate for rONA product formation. Over more than 18 hours, respectively, of
`three doublings of biomass (12-95 g/L), growth rate was kept constant (FIG. 4).
`The special advantages of the ZIMET/GBF HDF process (FIG. 5) include:
`defined nutrient medium without turbidity; the simple composition of the one-feed
`medium; a lowered risk of contamination; aeration with air instead of oxygen; if
`desired, N, P or other limitations are possible; a controlled growth rate over a long
`period of time; a low concentration of inhibitory metabolites; and no pep tides from
`medium components in downstream processing.
`In contrast to the work of Seo and Bailey,8 it is possible to study product
`formation and plasmid content of the cells at different growth rates without changing
`the medium. This enables the researchers to determine the optimum growth rate in
`
`BEQ 1005
`Page 4
`
`BEQ 1005
`Page 6
`
`

`
`304
`
`ANNALS NEW YORK ACADEMY OF SCIENCES
`
`0
`N
`
`-N
`.., -~
`~
`~
`<
`"'
`s
`
`'N
`0
`
`~
`
`.r:::
`
`~ :::: en
`
`~ .... ...
`~
`.....
`l:f ,_
`~
`"'
`
`•
`
`t!!
`~
`""
`::1:
`~~ <:>
`•
`~ ........
`~"'
`
`(J
`
`N -
`
`0
`
`0
`
`0
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30 h
`
`f"ERKNTATION Til£
`
`FIGURE 4. ZIMET /GBF HDF process. Kinetics of oxygen uptake and concentration in the
`exhaust gas. Time course of the specific growth rate.
`
`batch experiments instead of the chemostat operation which may lead to conclusions
`that are not always easily applicable to subsequent batch cultivations.
`
`SCALE-UP
`
`Process development was carried out in a three-stage bioreactor cascade of 70,
`300, and 1,500 L. The reactors are equipped with multichannel control microcomput(cid:173)
`ers (MICON P200) that act as standard controllers for basic parameters such as
`temperature, pH, gas flow, and pressure. Furthermore, a second P200 calculates the
`actual growth rate from exhaust gas analysis and controls it by appropriate alteration
`of the oxygen transfer to the culture.
`Any operation of the user such as input of setpoints and switching between
`cultivation phases and therefore different modes of the controllers can be performed
`at a terminal connected to a process computer, which provides for data transfer
`between P200 microcomputers, exhaust gas analysis, and host computer link.
`With data from the 70-L cultivation experiments it was no problem to scale-up
`the process to 1,500 L. The only critical problem that arose with increasing scale was
`the removal of heat.
`
`SCALE-DOWN
`
`Further investigations on HDF and its applications on rONA protein production
`are under way. Actually, variations of HDF are being analyzed and appropriate
`
`BEQ 1005
`Page 5
`
`BEQ 1005
`Page 7
`
`

`
`KNORRE et al.: HIGH CELL DENSflY FERMENTATION
`
`305
`
`host/vector systems are being prepared. This work can be carried out in laboratory
`scale fermentors, so that scale-down to 5-L working volume was performed in an
`early phase of process development. The major difference from the original scale
`( ~ 70 L) HDF is operation at atmospheric pressure with the use of glass fermentors.
`In this case the only way to ensure sufficient oxygen transfer into the broth is to
`aerate with oxygen-enriched air.
`In 5-L scale a slightly modified HDF process was successfully applied to produc(cid:173)
`tion of recombinant ~-galactosidase using a chemically inducible system. Specific
`activities (U I g) obtained in standard cultivations were also achieved under dense
`cultivation conditions at a cell mass 10-50 times higher, that is, up to 50 times higher
`volumetric activity.
`
`FUTURE PROSPECTS
`
`In the last decade, most scientific attention has focused on genetics-based
`solutions to productivity issues, and progress has been impressive. Ultimately, the
`returns from a purely genetics approach will diminish, and further optimization will
`depend on understanding the relationship between the microbial environment and
`synthesis of the desired protein. More recently, Pliickthun 10 reported an expression
`system with which fully functional antibody F. or F.b fragments can be expressed in E.
`coli. Both chains are co-expressed and co-secreted into the periplasm of E. coli with
`correct signal processing, disulfide formation, and chain association. Such expression
`systems should also be suitable for other similar proteins. Therefore, it is necessary
`
`x.nax = 116 g/1
`
`0
`
`"' 0
`
`""'
`.....
`< .....
`~
`
`0 ....,
`
`~
`
`iii
`
`o,
`
`0
`
`0
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30h
`
`FERMENTATION TIME
`
`FIGURE 5. ZIMET /GBF HDF process. Typical time course of biomass and acetate.
`
`BEQ 1005
`Page 6
`
`BEQ 1005
`Page 8
`
`

`
`306
`
`ANNALS NEW YORK ACADEMY OF SCIENCES
`
`to develop HDF processes for E. coli with such an expression system and to optimize
`the accumulated periplasmatic proteins.
`The research and commercial forces that are guiding the development of
`recombinant DNA technology will certainly motivate the growing effort in HDF
`process research devoted to recombinant microorganisms, especially as products
`move from research into development.
`
`SUMMARY
`
`In recent years recombinant DNA technology has enabled us to produce various
`proteins of therapeutic importance with microorganisms. As an appropriate host
`organism, E. coli plays a dominant role. Yields of E. coli dry cell mass in shaker flask
`culture range from 1-2 g/L, whereas in fermentors up to 10 g dry cells/L can be
`achieved. ZIMET and GBF have developed a high cell density fermentation process
`that produces E. coli (on a glucose/mineral salt medium) up to more than 100 g dry
`cells/L in a special fed-batch mode. This cultivation strategy prevents oxygen
`limitation and hence the accumulation of acetate and other metabolic byproducts.
`The specific growth rate can be adjusted so that product formation reaches its
`optimum value. An example of the production of alpha1- interferon is presented. The
`high cell density fermentations were realized in 30- and 450-L Chemap fermentors
`(ZIMET) and in a three-stage bioreactor scale-up system (72, 300, and 1,500 L)
`developed in cooperation with GBF and B. Braun Melsungen AG. Multiloop
`controllers were used to control the process variables.
`
`REFERENCES
`
`1. BAUER, S. & J. SHILOACH. 1974. Maximal exponential growth rate and yield of E. coli
`obtainable in a bench-scale fermentor. Biotechnol. Bioeng. 16:933-941.
`2. GLEISER, I. E. & S. BAUER. 1981. Growth of E. coli W to high cell concentration by oxygen
`level linked control of carbon source concentration. Biotechnol. Bioeng. 23: 1015-1021.
`3. ZABRISKIE, D. W. & E. J. ARCURI. 1986. Factors influencing productivity of fermentations
`employing recombinant microorganisms. Enzyme Microbiol. Technol. 8: 706-717.
`4. PAN, J. G., J. S. RHEE & J. M. LEBEAULT. 1984. Physiological constraints in increasing
`biomass concentration of E. coli B in fed batch culture. Biotechnol. Lett. 9: 89-94.
`5. EPPSTEIN, L., J. SHEVf1Z, X.-M. YANG & S. WEISS. 1989. Increased biomass production in
`a benchtop fermentor. Bio/Technology 7: 1178-1181.
`6. FIESCHKO, J. & T. RITCH. 1986. Production of human alpha consensus interferon in
`recombinant E. coli. Chern. Eng. Commun. 4S: 229-240.
`7. MEYER, H.-P., C. LEIST & A. FIECHTER. 1984. Acetate formation in continuous culture of
`E. coli Kl2 D1 on defined and complex media. J. Biotechnol. 1: 355-358.
`8. SEo, J.-H. & J. E. BAILEY. 1986. Continuous cultivation of recombinant E. coli: Existence
`of an optimum dilution rate for maximum plasmid and gene product concentration.
`Biotechnol. Bioeng. 28: 1590-1594.
`9. RIESENBERG, D., K MENZEL, V. SCHULZ, K. SCHUMANN, G. VEITH, G. ZUBER & W. A.
`KNoRRE. 1991. High-cell-density-fermentation of recombinant E. coli expressing hu(cid:173)
`man interferon alpha-1. Appl. Microbiol. Biotechnol. 34:77--82.
`10. PLiJC.tCTHUN, A. 1990. Recombinant antigen binding fragment of an antibody expressed in
`E. coli: folding in vivo, properties and catalytic activity. E ngineering Foundation
`Conference: Progress in Recombinant DNA Technology and Application. June 3-8,
`Potosi, Missouri.
`
`BEQ 1005
`Page 7
`
`BEQ 1005
`Page 9