.
`(cid:14)
`Journal of Immunological Methods 216 1998 165]181
`
`Antibody engineering: Comparison of bacterial, yeast,
`insect and mammalian expression systems
`
`R. Vermaa,1, E. Boletia,b,1, A.J.T. Georgea,U
`aDepartment of Immunology, Di¤ision of Medicine, Imperial College School of Medicine, Hammersmith Hospital,
`Du Cane Road, London W12 0NN, UK
`bDepartment of Clinical Oncology, Di¤ision of Medicine, Imperial College School of Medicine, Hammersmith Hospital,
`Du Cane Road, London W12 0NN, UK
`
`Abstract
`
`Engineered antibody molecules, and their fragments, are being increasingly exploited as scientific and clinical
`tools. However, one factor that can limit the applicability of this technology is the ability to express large amounts of
`active protein. In this review we describe the relative advantages and disadvantages of bacterial, yeast, insect and
`mammalian expression systems, and discuss some of the problems that can be encountered when using them. There
`is no ‘universal’ expression system, that can guarantee high yields of recombinant product, as every antibody-based
`molecule will pose its own problems in terms of expression. As a result the choice of system will depend on many
`factors, including the molecular species being expressed, the precise sequence of the individual antibody and the
`preferences of the individual investigator. However, there are general rules with regards to the design of expression
`vectors and systems which will help the investigator to make informed choices as to which strategy might be
`appropriate for their application. Q 1998 Elsevier Science B.V. All rights reserved.
`
`Keywords: Antibody engineering; Expression systems; Escherichia coli; Yeast; Insect cells; Mammalian cells;
`Single-chain F
`
`1. Introduction
`
`The successful development of hybridoma tech-
`(cid:14)
`.
`¨
`nology by Kohler and Milstein 1975 , and the
`
`U Corresponding author. Tel.: q44 181 3831475; fax: q44
`181 3832788; e-mail: ageorge@rpms.ac.uk
`1R. Verma and E. Boleti have contributed equally towards
`the completion of this review.
`
`resulting ability to produce monoclonal anti-
`(cid:14)
`.
`bodies MAbs
`initiated a new era for science.
`Subsequently, the use of recombinant DNA tech-
`nology, and the increasing knowledge of the ge-
`netics and structure of the immunoglobulins, has
`permitted the genetic manipulation of antibody
`molecules. This allows their properties to be al-
`tered, creating novel improved molecules. In or-
`der to do this, various expression systems have
`
`0022-1759r98r$19.00 Q 1998 Elsevier Science B.V. All rights reserved.
`(cid:14)
`.
`P I I S 0 0 2 2 - 1 7 5 9 9 8 0 0 0 7 7 - 5
`
`BEQ 1009
`Page 1
`
`

`
`166
`
`R. Verma et al. r Journal of Immunological Methods 216 1998 165]181
`)
`(
`
`.
`1985; Better et al., 1988 as well as humanised
`antibodies in which just the CDRs are of rodent
`(cid:14)
`.
`origin Jones et al., 1986; Riechmann et al., 1988a .
`The second class of molecules consists of frag-
`ments of antibody molecules. These include frag-
`ments that are accessible through proteolysis, such
`(cid:14)
`.
`as Fab, Fab9, F ab9
`, as well as other fragments,
`2
`(cid:14)
`.
`such as Fv based molecules Fig. 2 . These
`. (cid:14)
`(cid:14)
`molecules include sFv single-chain Fv Bird et
`.
`(cid:14)
`al., 1988; Huston et al., 1988 , and the dsFv dis-
`. (cid:14)
`.
`ulphide stabilised Fv Glockshuber et al., 1990 .
`Small antibody fragments have advantages over
`whole immunoglobulins for some clinical applica-
`tions, such as good penetration of solid tumours
`(cid:14)
`.
`and rapid clearance Huston et al., 1993, 1996 . In
`addition they can be produced by phage display
`(cid:14)
`.
`libraries McCafferty et al., 1990 .
`These fragments can be endowed with new
`properties by fusion with other molecules, such as
`(cid:14)
`.
`metal-binding proteins George et al., 1995 , cy-
`(cid:14)
`(cid:14)
`.
`tokines Boleti et al., 1995 , toxins or drugs Hus-
`.
`ton et al., 1993 . They show particular promise for
`in vivo imaging applications, and radiolabelled
`sFv have been successfully used in the clinic to
`(cid:14)
`.
`image colorectal carcinoma Begent et al., 1996 .
`In addition bispecific and bivalent antibodies can
`(cid:14)
`.
`be made, such as diabodies Holliger et al., 1993 .
`
`3. Expression of antibody molecules
`
`Recombinant antibody fragments have been
`produced in various expression systems, such as
`(cid:14)
`bacterial Better et al., 1988; Skerra and
`¨
`Pluckthun, 1988; Huston et al., 1988; Bird et al.,
`.
`(cid:14)
`1988 , mammalian Jost et al., 1994; Dorai et al.,
`.
`(cid:14)
`.
`(cid:14)
`1994 insect Bei et al., 1995 , yeast Davis et al.,
`.
`(cid:14)
`1991; Ridder et al., 1995b , plant Whitelam et al.,
`1994 and in ¤itro translation systems Nicholls et
`.
`(cid:14)
`.
`al., 1993 .
`In order to achieve a desirable expression, the
`cloned gene must be transcribed and translated
`efficiently. The yields and biological activity of
`recombinant proteins differ greatly, and depend
`on a large number of factors, such as solubility,
`stability and size of the protein.
`Every protein poses unique problems in its
`expression because of its unique amino acid se-
`quence. Although general conclusions can be
`
`Fig. 1. Structure of the antibody molecule and its fragments.
`This figure shows the common antigen binding fragments of
`an IgG molecule. The IgG molecule is shown with the con-
`stant domains in dark shading and the variable domains of
`(cid:14)
`.
`both the chains in lighter shading. The F ab9
`fragment can
`2
`be made by pepsin digestion; following mild reduction this
`yields the Fab9 molecule. Fab fragments can be made by
`papain digestion. In some molecules it is possible to generate
`the Fv fragment by enzymatic approaches. Expression of the
`relevant gene segments also permits expression of recombi-
`nant versions of these molecules.
`
`been developed with the aim of producing, at
`reasonable cost and effort,
`functional antigen
`binding molecules.
`
`2. Recombinant antibodies
`
`There are two main classes of recombinant
`antibodies. The first is based upon the intact
`(cid:14)
`.
`immunoglobulin molecule Fig. 1 and is designed
`to reduce the immunogenicity of
`the murine
`molecule. Thus chimeric molecules, which consist
`of the murine V regions and human constant
`(cid:14)
`regions, have been developed Boulianne et al.,
`1984; Morrison et al., 1984; Neuberger et al.,
`
`Fig. 2. Recombinant molecules based on the Fv fragment.
`The Fv fragment is the smallest antibody fragment that retains
`an intact antigen binding site. However, it is unstable, as the
`V and V domains are free to dissociate. Two strategies
`H
`L
`have been adopted to overcome this. The first is to link the
`(cid:14)
`.
`domains with a peptide to generate a single-chain Fv sFv .
`The second is to introduce cysteines at the interface between
`the V and V domains, forming a disulphide bridge that
`H
`L
`w
`.x
`(cid:14)
`holds them together a disulphide stabilised Fv dsFv . The
`location of the bond shown in the figure is for illustrative
`purposes only.
`
`BEQ 1009
`Page 2
`
`

`
`R. Verma et al. r Journal of Immunological Methods 216 1998 165]181
`)
`(
`
`167
`
`drawn from the study of one protein, expression
`has to be optimised for every new protein. While
`the term ‘antibody’ covers one class of protein,
`each antibody has a different sequence. There-
`fore the expression of each antibody or its frag-
`ments has its own problems. One expression sys-
`tem which may be suitable for the expression of
`one antibody may not be suitable for another.
`The optimal system will depend on the type of
`(cid:14)
`molecule being expressed IgG, Fab, sFv, dia-
`.
`body , the individual antibody and also other fac-
`tors, such as the required quantity and purity of
`the final product.
`
`4. Gene expression using Escherichia coli cells
`
`Immunoglobulin fragments are commonly ex-
`pressed in E. coli. One advantage of this system is
`the ability to produce protein in large quantities.
`E. coli grow at a very fast rate in comparison to
`mammalian cells, giving the opportunity to purify,
`analyse and use the expressed protein in a much
`shorter time. In addition, transformation of E.
`coli cells with the foreign DNA is easy and re-
`quires minimal amounts of DNA. Antibody engi-
`neering using E. coli tends to be inexpensive.
`These reasons explain the popularity of bacterial
`systems. However, E. coli are not capable of
`glycosylating proteins. Therefore if whole anti-
`body molecules are required, which are glycosy-
`lated in the C 2 domain, it is necessary to use
`H
`other expression systems.
`In order to achieve successful expression, the
`gene encoding the antibody molecule must be
`placed in the context of appropriate sequences
`that allow transcription and translation of the
`protein. Inducible promoters are normally used to
`control expression of the protein. This is vital to
`prevent loss or mutation of the gene in situations
`where its production might be toxic to the bacte-
`ria. Commonly used promoters include the lac
`promoter, the trp promoter and their hybrid, the
`tac promoter that is regulated by the lac repres-
`sor and is induced by isopropyl-b-galactosidase
`(cid:14)
`. (cid:14)
`.
`IPTG Amann et al., 1983; de Boer et al., 1983 .
`Another popular promoter is the lP promoter,
`L
`responsible for the transcription of the l DNA
`molecule, which is regulated by a temperature-
`
`sensitive repressor. The T7 RNA promoter can
`also be used to obtain tightly controlled, high
`level, expression which involves two levels of am-
`(cid:14)
`plification Tabor and Richardson, 1985; Studier
`.
`and Moffatt, 1986 . A second important factor for
`efficient translation in E. coli is the existence of a
`(cid:14)
`prokaryotic ribosome-binding site Gold et al.,
`.
`(cid:14)
`.
`1981 . It consists of an initiation codon ATG
`(cid:14)
`.
`and the Shine]Dalgarno sequence SD , formed
`by 3]9 nucleotides and located 3]11 base pairs
`(cid:14)
`.
`(cid:14)
`bp upstream from the initiation codon Shine
`.
`and Dalgarno, 1975; Steitz, 1979 . The last impor-
`tant control element is the transcription termina-
`tor which prevents transcription beyond the de-
`sired gene and adds stability to the DNA.
`The expression of recombinant antibody frag-
`ments in the reducing environment of the cyto-
`plasm leads to the formation of insoluble inclu-
`sion bodies, which contain unfolded protein. This
`necessitates the development of refolding proto-
`cols to recover active material. There are a num-
`ber of refolding strategies that can be employed
`(cid:14)
`.
`Fig. 3 , and they need to be optimised for each
`molecule. Most strategies include the isolation of
`inclusion bodies, the solubilisation of the recom-
`binant proteins, and their renaturation in an envi-
`ronment
`that promotes the correct disulphide
`bond formation and adoption of the appropriate
`three-dimensional shape. This review will not at-
`tempt to provide a detailed discussion of different
`refolding protocols, but rather will concentrate on
`the main principles of the process. Further detail
`(cid:14)
`.
`can be found in Huston et al. 1995 .
`Solubilisation of the inactive proteins is typi-
`cally done using denaturing agents, such as guani-
`dine]HCl or urea. However, mild detergents,
`which do not bind too strongly to the protein
`(cid:14)
`.
`(cid:14)
`Tanford, 1968 , can also be used Lacks and
`.
`Springhorn, 1980; Kurucz et al., 1995 . In addition
`reducing agents, such as b-mercaptoethanol or
`(cid:14)
`.
`dithiothreitol DTT can be used to reduce inter-
`and intra-chain disulphide bonds that might have
`formed during the lysis of the bacteria and solu-
`bilisation of the protein.
`The formation of disulphide bonds can be per-
`(cid:14)
`formed by simple air oxidation Anfinsen et al.,
`.
`1961 , in some cases promoted by the presence of
`(cid:14)
`.
`metal ions Saxena and Wetlaufer, 1970 . ‘Shuf-
`
`BEQ 1009
`Page 3
`
`

`
`168
`
`R. Verma et al. r Journal of Immunological Methods 216 1998 165]181
`)
`(
`
`Fig. 3. Refolding pathways for sFv molecules. This figure illustrates the three major pathways by which denatured and reduced sFv
`(cid:14)
`.
`molecules as might be found after solubilisation and reduction from inclusion bodies can be refolded. In all the figures, the fully
`denatured and reduced molecule is shown at the left, with the cysteine residues reduced. The linker residue is indicated by a
`(cid:14) .
`crenulated line. The fully refolded molecules are shown to the right of the diagram. In dilution refolding 1 the denatured molecule
`is allowed to renature, usually by rapid dilution into a buffer that lacks the chaotropic agent responsible for denaturation. As the
`molecule has adopted the correct three-dimensional shape, oxidation of the cysteine residues leads to formation of the appropriate
`(cid:14) .
`disulphide bonds. In redox refolding 2 the renaturation and reoxidation occur at the same time, forming an equilibrium with the
`denatured form of the molecule and other partially folded or incorrectly folded species. This approach relies on the correctly
`refolded molecule being the energetically preferred species, so becoming the predominant form. In disulphide restricted refolding
`(cid:14) .3 the disulphide bonds are allowed to form in a random manner. The molecule is then renatured. As two of the intermediate
`species will have incorrect disulphide bonds, they are unable to form an active sFv. An alternative form of disulphide restricted
`refolding can be performed on insoluble material obtained from the periplasm, where the disulphide bonds are assumed to be
`correctly formed, but the molecule has precipitated out of solution. Which of these schemes work is dependent on the properties of
`the individual sFv. For example, in some cases the correctly refolded molecule is not the most energetically stable, and so scheme 3
`(cid:14)
`.
`must be adopted. Figure adapted from Huston et al. 1995 .
`
`.
`(cid:14)
`fling’ breaking and reforming of the disulphide
`bonds to increase the chance of obtaining the
`correct configuration can be promoted by use of
`(cid:14)
`enzymes
`such as disulphide isomerase Car-
`.
`michael et al., 1977 or the inclusion of a redox
`couple made by a mixture of reduced and oxi-
`(cid:14)
`.
`dised thiol groups Saxena and Wetlaufer, 1970 ,
`.
`(cid:14)
`as provided, for example, by glutathione Fig. 4 .
`
`One of the major problems that needs to be
`overcome during the refolding process is the for-
`mation of aggregates. These are a consequence of
`the interaction of hydrophobic patches on the
`surface of malfolded or partially folded proteins.
`This process can be minimised by lowering the
`concentration of the refolding mixture.
`Refolding can also be promoted by addition of
`
`BEQ 1009
`Page 4
`
`

`
`R. Verma et al. r Journal of Immunological Methods 216 1998 165]181
`)
`(
`
`169
`
`Fig. 4. Formation of disulphide bonds during refolding. One of the major problems with refolding is to ensure that the cysteine
`residues in the molecule oxidise to form the correct disulphide bonds. At the bottom is shown the number of combinations of
`different disulphide bonds that can be formed in a molecule, depending on the number of bonds in the native molecule. As sFv
`fragments normally have two disulphide bonds, there are three different ways in which these bonds can be formed. If the bonds are
`formed at random only a third of the molecules will have their cysteines joined in the correct configuration. As can be seen the
`number of combinations rapidly rises with the number of disulphide bonds. One way to encourage the formation of the correct
`bonds is to set up a redox couple, for example by using a mixture of oxidised and reduced glutathione. These form equilibrium
`reactions, as shown at the top of the figure, which leads to the formation and breaking of disulphide bonds. This ‘shuffling’ can
`allow the incorrect combinations of disulphide bonds to be changed, and, assuming that the correct conformation is energetically
`(cid:14)
`.
`favourable, lead to an increase in the correct combination. Figure adapted from Jaenicke and Rudolph 1989 .
`
`ately surrounding the protein, either due to steric
`(cid:14)
`.
`hindrance as is seen with polyethylene glycol or
`by perturbation of the surface tension of water at
`the interface between the protein and the solvent
`(cid:14)the mechanism by which arginine acts as a co-
`.
`solvent , or by chemical interactions between the
`(cid:14)
`co-solvent and the protein such as charge repul-
`. (cid:14)
`.
`sion Timasheff and Arakawa, 1997 . In mal-
`(cid:14)
`folded or unfolded proteins which tend to be
`.
`.
`(cid:14)
`asymmetric the area or zone in which the co-
`solvents are excluded is greater than the native
`(cid:14)
`.
`molecule Fig. 5 . As the creation of zones of
`exclusion are thermodynamically unfavourable
`this encourages
`the correct
`folding of
`the
`(cid:14)
`.
`molecules Timasheff and Arakawa, 1997 .
`A recurring theme in this review is that every
`antibody or antibody fragment is unique. There is
`no one universal refolding protocol that can be
`
`Fig. 5. Co-solvents. The addition of a stabilising co-solvent
`encourages the formation of the native structure of a protein
`(cid:14)
`.
`during refolding shifts the equilibrium to the left , as the
`zone of exclusion of the co-solvent is increased for asymmet-
`ric, denatured molecules. Figure adapted from Timasheff and
`(cid:14)
`.
`Arakawa 1997 .
`
`a stabilising co-solvent, such as arginine. These
`co-solvents are excluded from the area immedi-
`
`BEQ 1009
`Page 5
`
`

`
`170
`
`R. Verma et al. r Journal of Immunological Methods 216 1998 165]181
`)
`(
`
`adopted. Some molecules appear to rapidly and
`easily adopt the correct conformation and can be
`refolded simply by diluting denatured material
`into a simple buffer and allowing the disulphide
`bonds to air oxidise. Other molecules can be
`refolded using protocols similar to these de-
`scribed above. However, some antibody fragments
`do not readily denature, rather appearing to adopt
`an energetically stable state that is not that of the
`(cid:14)
`.
`antigen binding molecule Huston et al., 1995 .
`There are many parameters that can be op-
`timised in any refolding protocol, these include
`factors such as the temperature of refolding, the
`time, the concentration of protein, the presence
`of a co-solvent or redox couple and the pH of the
`reaction. Given that several different refolding
`strategies may be needed to find the one best for
`any particular antibody, this can represent a con-
`siderable investment in time and energy. How-
`ever, the yields can be worth it. Using fermented
`cultures up to 100]130 mgrl of active sFv or sFv
`(cid:14)
`fusion proteins have been prepared Huston et
`.
`al., 1995 .
`An alternative approach is to use leader se-
`quences to direct secretion of the antibody to the
`(cid:14)
`. (cid:14)
`periplasmic space of bacteria Fig. 6 Skerra and
`.
`¨
`Pluckthun, 1988 . This lies between the inner and
`outer membrane of Gram negative bacteria, and
`is an oxidising environment. In addition there are
`a number of chaperonin-like molecules and disul-
`phide isomerases which may help the refolding of
`the recombinant antibody. This approach was first
`(cid:14)
`used in the expression of Fv
`Skerra and
`.
`¨
`Pluckthun, 1988 . A number of leader sequences
`can be used, including the pelB leader from the
`
`pectate lyase gene of Erwinia caroto¤ora Lei et
`(cid:14)
`.
`al., 1987 and the leader sequence derived from
`alkaline phosphatase gene. These sequences are
`cleaved by the signal peptidases
`inside the
`.
`(cid:14)
`periplasm Ferenci and Silhavy, 1987 .
`In some cases the recombinant antibody mate-
`rial
`in the periplasm ‘leaks’ through the outer
`(cid:14)
`membrane into the culture medium Ward et al.,
`.
`1989 . The extent to which this occurs is depen-
`dent on the bacterial strain, the induction condi-
`tions and, perhaps most importantly, the individ-
`(cid:14)
`ual amino acid sequence of the antibody Knap-
`.
`¨
`pik and Pluckthun, 1995 , rather than a result of a
`(cid:14)
`.
`signal sequence Suominen et al., 1987 . This can
`have a number of advantages;
`in particular,
`it
`permits rapid screening for antibody secretion
`and, when the yield is high, it may allow for direct
`purification of material
`from the supernatant.
`However, it should be stressed that not all anti-
`bodies are secreted in this way.
`If the material is not found in the supernatant,
`it may be necessary to isolate the protein from a
`periplasmic extract, frequently obtained by os-
`motic lysis. This has the advantage that the pro-
`tein is often present at high concentrations, in a
`(cid:14)
`.
`reasonably pure form Skerra, 1994 .
`In many cases the material precipitates out of
`solution in the periplasm. In these cases it is
`necessary to solubilise the protein, and renature
`(cid:14)
`.
`it. As the protein has hopefully been correctly
`folded, this can be done by solubilising the pro-
`tein in a chaotropic agent, such as urea or guani-
`dine, and then allowing it to renature by dialysis
`against a buffer containing a co-solvent, such as
`(cid:14)
`.
`arginine George et al., 1994 .
`
`Fig. 6. Diagram of a gram-negative cell envelope, showing the periplasmic space lying between the inner and outer membrane.
`(cid:14)
`.
`Figure adapted from Davis et al. 1990 .
`
`BEQ 1009
`Page 6
`
`

`
`R. Verma et al. r Journal of Immunological Methods 216 1998 165]181
`)
`(
`
`171
`
`Again it must be stressed that every antibody-
`based molecule is unique, and so what works for
`one molecule will not work for all. Thus the D1.3
`anti-lysozyme associated Fv fragment, which has
`been widely used as a model system for antibody
`engineering projects, can be isolated at yields of
`10 mgrl culture directly from the culture super-
`(cid:14)
`.
`nantant Ward et al., 1989 . Using carefully op-
`timised conditions up to 700 mgrl of an anti-CD3
`(cid:14)
`Fab molecule has been produced Rodrigues et
`al., 1992 . Up to 2 grl of a functional humanised
`.
`(cid:14)
`.
`Fab HuMab4D5-8 Fab9 have been expressed in
`(cid:14)
`.
`E. coli strain 25F2 Carter et al., 1992 . Other sFv
`cannot be obtained in useful yields even following
`attempts to refold from the insoluble periplasmic
`precipitates. Therefore there are several factors
`that need to be determined when expressing a
`(cid:14)
`new antibody. In which compartment medium,
`.
`soluble periplasm, insoluble periplasm is it to be
`(cid:14)
`found? What is the optimal induction time short
`induction tends to favour isolation from soluble
`periplasm, after longer induction material may be
`found in the supernatant or in the insoluble frac-
`.
`tion ? At what temperature should induction be
`(cid:14)
`carried out room temperature or 258C often pro-
`duces better results than 378C as the resulting
`slower synthesis of the protein does not over-
`.
`whelm the export pathway of the bacteria ? Other
`tactics may be useful; for example, it has been
`shown that addition of non-metabolisable sugars
`to the culture medium can increase the yield for
`(cid:14)
`.
`some antibodies Kipriyanov et al., 1997 , though
`(cid:14)
`this is not universal R. Verma, personal commu-
`.
`nication .
`As many groups are now using phage libraries
`to isolate novel antibodies, it is worth noting that
`the problems with bacterial expression appear to
`be reduced in the case of phage derived anti-
`bodies. Thus most antibodies isolated from the
`libraries can be produced in reasonable yield.
`This is probably because the phage selection
`operates not only to isolate ‘good binders’ but
`also ‘good expressors’, which will be relatively
`over-represented in the libraries.
`It is also of note that it is possible, by changing
`the sequence, to improve the production of anti-
`(cid:14)
`bodies by bacteria. Three point P40A, S63A and
`.
`A64D mutations in the framework region of the
`
`heavy chain of McPC603 Fv increased the ratio of
`(cid:14)
`soluble to insoluble protein by 60-fold Knappik
`.
`¨
`and Pluckthun, 1995 . In addition mutation of the
`linker peptide of sFv can be used to increase the
`yield. In one study a phage display library was
`used in which a single sFv was displayed, and
`6r15 of the amino acids forming the linker were
`mutated. Selection of antigen binding phage from
`the library led to isolation of new linker se-
`(cid:14)
`quences that improved secretion five fold Turner
`.
`et al., 1997 .
`Bacterial expression therefore has an impor-
`tant role to play in the production of recombinant
`antibody-based molecules,
`in particular for the
`fragments that do not require glycosylation. How-
`ever, while the yields can be very high, they are
`dependent on the individual antibody.
`
`5. Expression using yeast
`
`The main advantages of yeast over other ex-
`pression systems are related to the fact that it is
`both a microorganism and a eukaryote. Unlike E.
`coli, yeast provide advanced protein folding path-
`ways for heterologous proteins and, when yeast
`signal sequences are used, yeast can secrete cor-
`rectly folded and processed proteins. Therefore
`functional and fully folded heterologous proteins
`can be secreted into culture media. Unlike mam-
`malian expression systems, yeast can be rapidly
`grown on simple growth media. For the expres-
`sion of clinically and industrially important pro-
`teins, yeast is an attractive option as industrial
`scale fermentation technology is widely used.
`Whole antibodies and antibody fragments have
`(cid:14)
`been expressed using this system Wood et al.,
`.
`1985; Horwitz et al., 1988 . The binding activity of
`whole antibody and Fab secreted from yeast was
`similar to that of their counterparts derived from
`(cid:14)
`.
`lymphoid cells Horwitz et al., 1988 . Single-chain
`antibodies have also been successfully expressed
`in yeast systems, for example, an anti-fluorescein
`sFv has been produced in Schizosaccharomyces
`(cid:14)
`.
`pombe Davis et al., 1991 and anti-recombinant
`human leukaemia inhibitory factor sFv has been
`(cid:14)
`.
`expressed in Pichia pastoris Ridder et al., 1995b .
`Proteins which accumulate as insoluble inclusion
`bodies in E. coli are often soluble when ex-
`
`BEQ 1009
`Page 7
`
`

`
`172
`
`R. Verma et al. r Journal of Immunological Methods 216 1998 165]181
`)
`(
`
`.
`(cid:14)
`pressed in yeast Ridder et al., 1995b . In addi-
`tion, the degradation of heterologous proteins,
`often a problem in E. coli, is usually reduced in
`yeast.
`There are two types of vectors which are used
`for the expression of cloned genes in the yeast:
`(cid:14) .a episomal vectors which propagate extrachro-
`(cid:14) .
`mosomally; and b integrating vectors where
`chromosomal integration occurs by homologous
`recombination.
`It is important that the promoter is tightly
`regulated in order to separate the growth and
`induction phase, especially when it is necessary to
`express toxic proteins which would otherwise kill
`the host during the growth phase. In addition, this
`reduces the metabolic burden on the cells and
`avoids selection of low expressing variants. Het-
`erologous promoters do not function in yeast and
`therefore only yeast promoters can be used for
`the expression of the cloned genes. The most
`commonly used promoters in yeast are GAL1,
`GAL7, GAL5, which are repressed by glucose and
`induced by galactose. A variety of selectable
`markers are used for the isolation and selection
`of transformants;
`these include LEU2, TRP1,
`HIS3 and URA3, used in strains auxotrophic for
`leucine, tryptophan, histidine and uracil, respec-
`tively. The process of termination in yeast is simi-
`lar to higher eukaryotes involving termination of
`transcription, endonucleolytic processing and
`polyadenylation.
`In yeast, initiation of translation is inhibited by
`secondary structures and high G contents in the
`59 untranslated region; therefore, all non-coding
`sequences in the 59 end should be eliminated.
`(cid:14)
`.
`The initiation codon ATG is usually preceded
`by an A-rich sequence, such as AAAAAAATG
`for efficient initiation of translation. Heterolo-
`gous leader sequences do not appear to function
`in yeast; therefore for secretion to occur yeast
`derived signal peptides must be used.
`Unlike E. coli, yeast is capable of glycosylation
`of proteins at Asn-X-SerrThr motifs. However,
`this glycosylation is not the same as seen in
`hybridomas and myelomas since carbohydrates
`are not modified beyond the mannose addition
`(cid:14)
`.
`Kukuruzinska et al., 1987 .
`High levels of secreted recombinant antibody
`
`fragments have been achieved in yeast expression
`systems. Two single-chain antibodies, anti-CD7
`and anti-DMI, were expressed at 0.25 mgrl in E.
`coli, but when same fragments were expressed in
`Pichia pastoris their yields were increased to 60
`mgrl and 100]250 mgrl, respectively Eldin et
`(cid:14)
`.
`al., 1997 . Similarly, the yield of functional rabbit
`anti-recombinant human leukemia inhibitory sFv
`was 100-fold more in Pichia pastoris 100 mgrl
`(cid:14)
`.
`(cid:14)
`.
`than in E. coli Ridder et al., 1995b .
`
`6. Gene expression using insect cells
`
`Insect cell expression have systems emerged in
`the last few years as attractive choices for the
`expression of recombinant molecules. The rela-
`tively new stable transformation system, in addi-
`tion to the already well established baculovirus-
`mediated gene expression in insects, produces
`high amounts of the foreign protein of interest
`while allowing it to retain its functional activity.
`Both systems have their own advantages and limi-
`tations as will be discussed below.
`
`6.1. Baculo¤irus-mediated expression system
`
`Baculovirus expression systems are the most
`popular of the insect cell expression systems as
`they can produce large amounts of active pro-
`teins. The baculovirus system has been used to
`express functionally active antibody molecules
`(cid:14)Hasemann and Capra, 1990; Zu Putlitz et al.,
`.
`1990 . The expression levels can be higher than
`those found in the mammalian cell systems and,
`because the insect cells are able to perform most
`of the post-translational alterations which are
`used by higher eukaryotes, they have a significant
`(cid:14)
`advantage over the bacterial system Kang, 1988;
`Luckow and Summers, 1988; Maeda, 1989; Miller,
`.
`1988 . Furthermore, baculoviruses have a highly
`restricted host range which makes them safer
`(cid:14)
`than mammalian expression systems Groner,
`.
`1986 . Baculoviruses belong to a large group of
`circular double stranded DNA viruses which in-
`(cid:14)
`fect only invertebrates, usually insects Granados
`.
`and Federici, 1986 . Their genome ranges from 80
`kb to 200 kb. Most baculoviruses have the ability
`to produce viral polyhedra ] consisting mainly of
`
`BEQ 1009
`Page 8
`
`

`
`R. Verma et al. r Journal of Immunological Methods 216 1998 165]181
`)
`(
`
`173
`
`the protein polyhedrin ] in the nuclei of infected
`insect cells which, after the death of the infected
`cell, can transmit the infection to other insects.
`The baculovirus replication cycle is divided into
`four stages, namely, the immediate early, the de-
`layed early, the late and the very late stages
`(cid:14)
`.
`Friesen and Miller, 1986; Guarino, 1989 .
`In the typical baculovirus vector, the foreign
`gene is placed under the control of a strong
`polyhedrin promoter. This enables the gene to be
`transcribed at a high level, allowing simple selec-
`tion of recombinant viruses and causing the re-
`combinant protein to be secreted in the insect
`(cid:14)
`cell culture in large amounts Miller, 1988; Maeda,
`.
`1989 . The polyhedrin promoter is considerably
`stronger than most eukaryotic promoters. The
`most commonly used baculoviruses are the Auto-
`grapha californica nuclear polyhedrosis
`virus
`(cid:14)
`.
`AcNPV and the Bombyx mori nuclear polyhe-
`(cid:14)
`. (cid:14)
`drosis virus BmNPV Adams and McClintock,
`.
`1991; Bilimoria, 1991; Kool and Vlak, 1993 . The
`insect cell line Sf9 which derives from Spodoptera
`frugiperda is probably the most widely used host
`(cid:14)
`.
`Summers and Smith, 1987 .
`The basic steps for the construction of a bac-
`ulovirus expression vector involve the insertion of
`the foreign gene into a specific transfer vector
`which is usually formed by a bacterial plasmid ]
`frequently pUC-derived ] containing 59 and 39
`sequences, a multiple cloning site with a choice of
`restriction enzymes for the insertion of the gene
`of interest and, finally, the polyhedrin promoter.
`It has been shown that
`the highest
`levels of
`expression are achieved when the gene which is
`to be expressed is inserted immediately down-
`stream of the first nucleotide in the polyhedrin
`(cid:14)
`initiation codon Matsuura et al., 1987; Possee
`.
`and Howard, 1987; Luckow and Summers, 1989 .
`This allows preservation of the transcriptional
`the 59
`and translational control sequences of
`non-coding region of the polyhedrin gene. The
`next step is the insertion of the gene of interest
`into the baculovirus genome by co-transfecting
`the insect cells with the transfer vector plasmid
`DNA and wild type viral DNA. By using an occlu-
`sion-negativerpositive plaque assay the transfor-
`mants which contain the foreign gene sequence
`can be identified under the microscope by their
`
`(cid:14)
`occlusion-negative phenotype Hink and Vail,
`.
`1973; Volkman and Summers, 1975 .
`Glycosylation of proteins takes place in the
`(cid:14)
`.
`insect cells Bei et al., 1995 but, although the
`N-linked glycosylation sites are the same in mam-
`malian and insect cells