Expression, Purification and Characterization ofFungal and Viral Recombinant Proteins

Helena Vihinen

Department of Chemical TechnologyHelsinki University of Technology

andInstitute of Biotechnology

University of HelsinkiFinland

Dissertation for the degree of Doctor of Science in Technology to be presented with duepermission of the Department of Chemical Technology for public examination and debatein Auditorium KE1 at Helsinki University of Technology (Espoo, Finland) on the 22nd

of March, 2001, at 12 noon.

SUPERVISED BY:

Professor Leevi KääriäinenInstitute of Biotechnology, University of Helsinki

andProfessor Marja Makarow

Institute of Biotechnology, University of Helsinki

REVIEWED BY:

Docent Anu JalankoDepartment of Human Molecular Genetics

National Public Heath InstituteHelsinki

andDocent Johan Peränen

Institute of Biotechnology, University of Helsinki

OPPONENT:

Docent Pertti MarkkanenBiotechnology Program

Espoo-Vantaa Institute of Technology

ISSN 1239-9469ISBN 951-45-9874-1

ISBN 951-45-9875-X (pdf)Gummerus Kirjapaino Oy

Saarijärvi 2001

To my brother Harri

CONTENTSAbbreviationsOriginal publicationsABSTRACT...........................................................................................................................1INTRODUCTION.................................................................................................................2

1. PRODUCTION SYSTEMS FOR RECOMBINANT PROTEINS ......................... 21.1. BACTERIAL EXPRESSION UTILIZING E. COLI ..........................................................3

1.1.1. Expression systems ......................................................................................31.1.2. Solubility......................................................................................................41.1.3. Secretion ......................................................................................................5

1.2. YEAST (SACCHAROMYCES CEREVISIAE) ...............................................................51.2.1. Episomal vectors and chromosomal integration ..........................................51.2.2. Promoter strength.........................................................................................61.2.3. Secretion of heterologous proteins...............................................................7

1.2.3.1. Yeast secretory proteins as carriers .....................................................71.2.3.2. The secretion pathway of S. cerevisiae ...............................................8

1.2.4. Optimizing expression .................................................................................91.3. INSECT CELLS (BACULOVIRUS SYSTEM)..............................................................10

1.3.1. Systems utilizing polyhedrin and p10........................................................111.3.2. Recombination ...........................................................................................111.3.3. Optimizing expression ...............................................................................12

1.4. MAMMALIAN EXPRESSION SYSTEMS UTILIZING VACCINIA AND ALPHAVIRUS VECTORS .........................................................................................131.4.1. Vaccinia virus vector .................................................................................131.4.2. Alphavirus vectors .....................................................................................14

1.4.2.1. Semliki Forest virus ..........................................................................141.4.2.1.1. Nsp1..........................................................................................151.4.2.1.2. Nsp2..........................................................................................161.4.2.1.3. Nsp3..........................................................................................161.4.2.1.4. Nsp4..........................................................................................17

2. PURIFICATION STRATEGIES FOR RECOMBINANT PROTEINS.....................172.1. CONVENTIONAL CHROMATOGRAPHY ..................................................................17

2.1.1. Ion exchange chromatography...................................................................182.1.2. Reversed phase chromatography ...............................................................182.1.3. Gel permeation chromatography................................................................19

2.2. AFFINITY CHROMATOGRAPHY.............................................................................192.2.1. Affinity tags ...............................................................................................202.2.2. Cleavage.....................................................................................................23

3. ANALYSIS OF POSTTRANSLATIONAL MODIFICATIONS ..............................243.1. GLYCOSYLATION.................................................................................................253.2. ACYLATION.........................................................................................................253.3. PHOSPHORYLATION.............................................................................................26

AIMS OF THE STUDY .....................................................................................................28MATERIALS AND METHODS .......................................................................................29

1. MATERIALS................................................................................................................291.1. STRAINS, CELLS AND MEDIA................................................................................291.2. PLASMID CONSTRUCTIONS ..................................................................................30

1.2.1. Yeast (I-III) ................................................................................................301.2.2. E. coli (IV, unpublished)............................................................................301.2.3. Baculovirus system (unpublished).............................................................30

1.2.4. Vaccinia infection/DNA-transfection (IV- IV)..........................................302. METHODS ...................................................................................................................30

2.1. DELETION CONSTRUCTS AND POINT MUTATIONS (IV-VI) ...................................302.2. IMMUNOPRECIPITATION, IMMUNOBLOTTING AND SDS-PAGE ...........................302.3. PROTEIN EXPRESSION AND PURIFICATION ...........................................................31

2.3.1. Yeast (I-III) ................................................................................................312.3.2. E.coli (IV, unpublished).............................................................................312.3.3. Sf9 cells (unpublished) ..............................................................................322.3.4. HeLa cells (IV-VI).....................................................................................32

2.4. METABOLIC LABELING (I-III,V,VI).....................................................................342.5. IMMUNOFLUORESCENCE (VI)..............................................................................342.6. FLOTATION ANALYSIS (VI) .................................................................................342.7. PHOSPHOAMINO ACID ANALYSIS (V)...................................................................342.8. PHOSPHOPEPTIDE MAPPING (V)...........................................................................352.9. OTHER SPECIFIC METHODS ..................................................................................35

RESULTS ............................................................................................................................36

1. HSP150� AS A CARRIER FOR SECRETION OF RECOMBINANT PROTEINS IN S. cerevisiae .............................................................................................................36

1.1. CHARACTERIZATION OF HSP150�-CARRIER (I)...................................................361.1.1. Expression of truncated Hsp150 (I) ....................................................................... 371.1.2. Purification of truncated and authentic Hsp150 (I) ................................................ 371.1.3. Structural features (I) ............................................................................................. 38

1.2. SECRETION AND AUTHENTICITY OF HSP150�-NGFRe (II)...................................381.2.1. Expression of Hsp150�-NGFRe (II) ...................................................................... 381.2.2. Purification of Hsp150�-NGFRe (II) ...................................................................... 39

1.3. GLYCOSYLATION OF NGFRe (III) ....................................................................... 392. EXPRESSION, PURIFICATION AND PROPERTIES OF SFV Nsps .......................41

2.1. EXPRESSION AND PURIFICATION OF NSPS (UNPUBLISHED) ..................................412.1.1. Expression of Nsp1 in E. coli (unpublished) .......................................................... 412.1.2. Expression of N-terminal fragment of Nsp2 in E. coli (unpublished)..................... 422.1.3. Expression of Nsp3 in E. coli (unpublished) .......................................................... 422.1.4. Expression of Nsp4 in Sf9 cells (unpublished) ....................................................... 44

2.2. CHARACTERIZATION OF NSP1 AND NSP3 (IV-VI)...............................................452.2.1. Critical residues for enzymatic activities of Nsp1 (IV)........................................... 452.2.2. Phosphorylation of Nsp3 (V and VI) ...................................................................... 46

2.2.2.1. Determination of phosphorylation sites (V) .................................................. 462.2.2.2. Phosphorylation of Nsp3 derivatives (VI) ..................................................... 472.2.2.3. Elimination of phosphorylation sites (V and VI) ........................................... 482.2.2.4. Effect of phosphorylation on membrane association (VI) ............................. 482.2.2.5. Phosphorylation deficient Nsp3 in the context of SFV (VI) .......................... 48

DISCUSSION ......................................................................................................................491. HSP150� AS A CARRIER FOR PROTEIN PRODUCTION .....................................502. EXPRESSION AND CHARACTERIZATION OF SFV Nsps....................................50

2.1. EXPRESSION AND PURIFICATION OF Nsps............................................................492.2. GT AND MT ACTIVITIES OF Nsp1 .......................................................................522.3. PHOSPHORYLATION OF Nsp3...............................................................................52

2.3.1. Determination of phosphorylation sites .....................................................532.3.2. Kinases.......................................................................................................532.3.3. Effects on the virus cycle and neurovirulence ...........................................54

3. CONCLUSIONS AND FUTURE ASPECTS ..............................................................55ACKNOWLEDGEMENTS ...............................................................................................57REFERENCES....................................................................................................................58

ACN acetonitrileAcMNPV Autographa californica

multiple nuclear polyhedrosisvirus

AdoHcy S-adenosyl-L-homocysteineAdoMet S-adenosyl-L-methionineBVS baculovirus expression systemBHK baby hamster kidneyBSA bovine serum albuminCD circular dichroismCHX cycloheximideCIAP calf intestine alkaline

phosphataseCID collision-induced dissociationCPVI cytopathic vacuoles type IDOC deoxycholateDTT dithiothreitolECL enhanced chemiluminescenceER endoplasmic reticulumESI electrospray ionizationFSC fetal calf serumGnd-HCl guanidinium hydrochlorideGPC gel permeation chromatographyGST glutathione-S-transferaseGT guanylyltransferaseHPLC high performance liquid

chromotographyHsp heat shock proteinIEC ion exchange chromatographyIMAC immobilized metal affinity

chromatographyIPTG isopropyl-ß-D-thiogalacto-

pyranosidekDa kilodaltonMALDI matrix-assisted laser desorp-

tion/ionization

MBP maltose binding proteinm.o.i. multiplicity of infectionMS mass spectrometryMT methyltransferaseMVA modified vaccinia virus AnkaraNGFRe ectodomein of rat nerve growth

factor receptorNMR nuclear magnetic resonanceNsp nonstructural proteinNTA nitrilotriacetic acidODV occlusion derived virusPBS phosphate buffered salinePDI protein disulfide isomerasePFU plaque forming unitp.i. post infectionPMSF phenylmethylsulfonyl fluoridePPM phosphopeptide mappingPVDF polyvinylidene difluorideRPC reversed phase chromatographySC synthetic completeSf9 Spodoptera frugiperda insect

cell lineSFV Semliki Forest virusSIN Sindbis virusSRP signal recognition particleTCA trichloroacetic acidTGN trans Golgi networkTLC thin layer chromatographyTLE thin layer electrophoresisTM tunicamycinTOF time-of-flightTrx thioredoxinwt wild type

Abbreviations

Jämsä, E., Holkeri, H., Vihinen, H., Wikström, M., Simonen, M., Walse, B.,Kalkkinen, N., Paakkola, J. and Makarow, M. (1995). Structural features of apolypeptide carrier promoting secretion of a ß-lactamase fusion protein in yeast.Yeast 11, 1381-1391.

Simonen, M., Vihinen, H., Jämsä, E., Arumäe, U., Kalkkinen, N. and Makarow,M. (1996). The hsp150�-carrier confers secretion competence to the rat nervegrowth factor receptor ectodomain in Saccharomyces cerevisiae. Yeast 12, 457-466.

Holkeri, H., Simonen, M., Pummi, T., Vihinen, H. and Makarow, M. (1996).Glycosylation of rat NGF receptor in the yeast Saccharomyces cerevisiae. FEBSLett. 383, 255-258.

Ahola, T., Laakkonen, P., Vihinen, H. and Kääriäinen, L. (1997). Critical Residuesof Semliki Forest virus RNA capping enzyme involved in methyltransferase andguanylyltransferase-like activities. J. Virol. 71, 392-339.

Vihinen, H. and Saarinen, J. (2000). Phosphorylation site analysis of SemlikiForest virus nonstructural protein 3. J. Biol. Chem. 275, 27775-27783.

Vihinen, H., Ahola, T., Tuittila, M., Merits, A. and Kääriäinen, L. (2001).Elimination of phosphorylation sites of Semliki Forest virus replicase proteinnsP3. J. Biol. Chem. 276, in press.

I

II

III

IV

V

VI

Original publications

This thesis is based on the following original publications which are referred to in the text bytheir Roman numerals. Additional unpublished data will also be presented in the text.

All articles are reprinted by kind permission from the publishers.

1

This work reports the production of recom-binant yeast and viral proteins in a numberof diverse in vivo model systems for enzy-matic and structural studies. In the first partHsp150� peptide, a derivative of the yeast(Saccharomyces cerevisiae) secretory heat-shock protein Hsp150, was investigated forits ability to act as a carrier in transportingthe ectodomain of rat nerve growth factor(NGFRe) out from the yeast cell. TheHsp150�-NGFRe fusion protein was effi-ciently secreted into the growth medium,where it constituted the majority of totalsecreted proteins. Inhibition experiments withpurified Hsp150�-NGFRe showed thatHsp150� did not prevent NGFRe from fold-ing into a ligand-binding conformation. Cir-cular dichroism (CD) analysis revealed thatthe Hsp150�-carrier did not have any spe-cific secondary structure, which was also sug-gested by NMR analysis of a syntheticpolypeptide corresponding to the repetitiveconsensus sequence of subunit II of Hsp150.These findings suggest that Hsp150� can suc-cessfully act as a carrier for foreign proteins,such as NGFRe, made and secreted by S.cerevisiae.

The second part of this study involved theexpression and purification of an RNA ani-mal virus, Semliki Forest virus (SFV),nonstructural proteins (Nsp1-4) using a num-ber of in vivo protein expression systems. Toensure quantities large enough for structuraland enzymatic studies of the Nsps, each ofthem was expressed either in bacteria (Es-cherichia coli) or in insect cells (Sf9). Allthe proteins were expressed in high quanti-ties (10-100 mg/l), and purified by affinity

and size exclusion chromatography undernondenaturing or denaturing conditions. In-dependent of the expression system used, allthe partially purified Nsps aggregated andprecipitated either upon concentration, dialy-sis, storing or thawing. No detergents werefound that could alleviate the aggregationproblem or assist in the purification process.

Despite the unsuccessful purification ofNsps for structural studies, the expression andpartial purification of Nsp1 and Nsp3 permit-ted biochemical characterization of their en-zyme activities and posttranslational modifi-cations. Point mutational analysis of the Nsp1methyltransferase domain revealed that resi-due His38 was essential for the guanylyl-transferase activity of Nsp1. Furthermore,residues Asp64 and Asp90 were found to beimportant for the methyltransferase activityof Nsp1. Phosphorylation sites in Nsp3 weredeterminated by point mutational analysis,electrospray ionization (ESI) and matrix as-sisted laser desorption ionization (MALDI)mass spectrometry (MS) as well as byphosphopeptide mapping and Edman se-quencing. A phosphorylated domain (aa 320-368) was located in the C-terminal, non-con-served region of Nsp3, where 12 serines and4 threonines could be modified by phos-phates. The phosphorylation of Nsp3 seemednot to affect the membrane association or thelocalization of Nsp3 in either transfected orinfected cells. Furthermore, Nsp3 phosphory-lation deficient mutant viruses were capableof replication in infected mammalian cells asimilar manner to the wild type SFV, but theirneuropathogenicity in adult mice was greatlyreduced.

ABSTRACT

Abstract

Introduction

2

Molecular biology offers technologieswhereby proteins can be produced and puri-fied easier and more efficiently than everbefore. Using recombinant DNA techniquessuch as gene fusion it is possible to generatechimeric proteins, which are novel in struc-ture and function. At the DNA level, aminoacid coding regions are now routinely recom-bined to incorporate new functional domains,proteolytic cleavage sites, intra- and extra-cellular localization or targeting signals,stabilizing signals, and even amino acidsequences that facilitate protein purification.All of these directed mutations can be used tosubtly or dramatically alter the properties (e.g.,solubility, electric charge, hydrophobicity,conformation, substrate binding and activesites) of the protein. Thus, protein engineer-ing has become a powerful tool in molecularbiology to investigate protein structure andfunction, in addition to production and purifi-cation of useful proteins (Sassenfeld, 1990).The main applications of recombinant proteinsobtained by genetic engineering are in themedical therapeutic fields (e.g., production ofrecombinant vaccines, and therapeutic proteinsfor human diseases), and medical diagnosis(e.g., antigen engineering for poly- and mono-clonal antibody production used in diseasetesting). Other areas where recombinant pro-teins are commonly utilized include enzymesfor food and fiber production, testing food formicrobial contamination and veterinary medi-cine (Nilsson et al., 1992).

Besides producing valuable proteins forpractical applications, the production ofproteins using recombinant technology offersunprecedented possibilities in the basicresearch of protein function, structure, inter-action, turnover, domain shuffling, and manyother areas, which could ultimately lead to thedevelopment of novel protein applications.Most proteins are expressed in infinitesimal

amounts in their native cells and tissues, andit is only by recombinant techniques that it ispossible to produce amounts great enough forbasic research or for practical uses. Therefore,the expression of engineered proteins in effi-cient heterologous protein expression systemis integral to the production, and purificationof many proteins of interest. Moreover, withthe vast amounts of predicted protein se-quences being generated by genomic research,the application of protein engineering will mostcertainly be used to better characterize manyunknown protein functions, interactions,cellular locations and potential practical uses.

1. PRODUCTION SYSTEMS FORRECOMBINANT PROTEINS

Although much of recombinant proteinengineering is performed in vitro, the actualprotein synthesis usually utilizes livingprokaryotic or eukaryotic cells. Proteins pro-duced in vivo can be expressed transiently,constitutively or by induction, with the last tworequiring cell lines which stably maintain thegenetic material coding for the protein.Another means consist of using live bio-reactors, i.e., live transgenic plants andanimals, as recombinant protein productionsystems. The choice of production systemdepends on many factors including the origi-nal source of the protein and codon usage (e.g.,prokaryotic, mammalian, plant), posttransla-tional processing (e.g., phosphorylation, pro-teolytic processing, glycosylation, acylation),its future use (e.g., human therapy, plant pestcontrol) and economic factors (Hodgson,1993). However, production of any protein isa complex multistep cellular process involv-ing gene transcription, translation, proteinprocessing and localization. Thus, the expres-sion of a recombinant protein in a heterolo-gous host cell or organism is often problem-atic. For example, protein expression levels

INTRODUCTION

Introduction

3

can be low, or the purified protein may beinsoluble or unstable. And although, certainprecautions can be taken (e.g., altering codonusage to match the host’s cells), often trial anderror is the only real method of operation.

1.1. BACTERIAL EXPRESSION UTILIZING E. COLI

Escherichia coli (E. coli) has long been theprimary prokaryotic host for heterologousprotein expression, therefore there is muchinformation and technology regarding thissystem with many different proteins (Baneyx,1999; LaVallie et al., 1995). The advantagesof E. coli include its relatively easy, rapid andinexpensive methods of growth, transforma-tion and maintenance. Furthermore, most ofthe biochemical pathways of E. coli areunderstood in great detail, and its entiregenome has been sequenced. Foreign proteinscan be produced in E. coli in large amounts(5-50% of total protein). Nevertheless,prokaryotic cells such as E. coli are unable toperform some posttranslational modifications,which occur in eukaryotic cells (e.g., glyco-sylation, phosphorylation and disulfidebondformation), and which may be crucial for aforeign protein to obtain an active form. Theexpression and accumulation of a foreign pro-tein in E. coli may also cause aggregates ofthe protein to form (Lilie et al., 1998; DeBerbardez Clark, 1998). Sometimes theprotein in these inclusion bodies can berefolded in vitro to produce functional protein,but this is not always possible and renaturationcan be expensive and time consuming. Fur-thermore, proteolytic degradation of heterolo-gous protein expression may be a potentialobstacle (Murby et al., 1996; Matsuo et al.,1999). Even though E. coli may not be usefulfor all foreign protein production, it has beensuccessfully utilized to produce many func-tional human proteins such as human growthhormone, proinsulin, interferon-gamma andantibody fragments (Patra et al., 2000; Cowleyand Mackin, 1997; Davis et al., 1999;Plückthun, 1992).

1.1.1. Expression systems

Various E. coli expression systems areavailable commercially and in the publicdomain, which is shared within the scientificcommunity. Each system offers differentbenefits for protein expression, detection, andpurification, and should be consideredaccording to the specific criteria and require-ments each protein poses. The expression unitor cassette in which the foreign protein codingregion will be cloned into should consist of astrong promoter (e.g., T7, trc, lac, tac, PL, PR,phoA, ara, xapA, cad, recA), ribosomalbinding site(s), and an efficient transcriptionterminator (LaVallie et al., 1995). Tightlycontrolled transcription coupled with a strongand rapid transcription inducer (e.g., IPTG,tryptophan, temperature, phosphatase starva-tion, arabinose, xanthosine, pH, nalidixic acid)are crucial for high protein expression. Otherfactors to be considered include convenientmultiple cloning sites, appropriate codonusage, and plasmid copy number. Usually, highprotein expression levels are desirable, but incases where the protein may form insolubleinclusion bodies, or may be toxic to the bacte-rial host, lower and slower expression may bepreferred. The ideal promoter for expressiondirects efficient transcription to allow high-level production, and is tightly regulated tominimize the metabolic burden and toxiceffects of a foreign protein. Regulatable pro-moters are used to drive foreign proteinexpression to avoid the selection of non-expressing mutant cells during the cell growthphase. Promoters such as T3, SP6 and T7 arewidely used and efficient promoters derivedfrom E. coli bacteriophages.

The T7 RNA polymerase is very specificfor its own promoter, and can transcribe RNAtemplates approximately five times faster thanthe E. coli RNA polymerase (Mertens et al.,1995). The gene coding for T7 RNA poly-merase is either lysogenic or it is transfectedinto the cell by the bacteriophage CE6. In DE3E. coli cells the T7 RNA polymerase gene is

Introduction

4

under the control of the LacUV5 promoter,which is inducible by IPTG. The addition ofIPTG to a culture of DE3 cells will induce thetranscription of T7 RNA polymerase, whichin turn transcribes any gene under the controlof the T7 promoter. However, the basal levelof T7 RNA polymerase activity will promotesome transcription of the target gene in non-induced cells. More strict control mechanismshave been engineered to depress basal T7 RNApolymerase expression. One method is basedon the addition of a regulatory lacI gene, whichcodes for a strong repressor of the LacUV5promoter (Dubendorff and Studier, 1991). Verylow basal expression levels of target gene canbe reduced even more by adding T7 lysozyme,yet high level expression can be achieved uponinduction with IPTG. Moreover, the T7lacOexpression vectors pBAT, pHAT and pRAThave been reported to be suitable for expres-sion of toxic proteins in E. coli (Peränen etal., 1996). In these vectors tight control isachieved by centering the lac operator 14 basepairs downstream from the major RNA tran-scription start point.

1.1.2. Solubility

The majority of foreign proteins producedin E. coli are cytosolic, but as mentioned abovesome proteins accumulate as aggregates calledinclusion bodies. In these aggregates only asmall fraction of the protein is folded correctly,and the rest is in a denatured and usually non-functional form (Geisow, 1991; Valax andGeorgiou, 1993). The probability of any het-erologous protein forming inclusion bodiescorrelates greatly with its charge average andturn forming residue fraction, whereas hydro-philicity and total number of residues do notappear to correlate with inclusion body for-mation (Wilkinson and Harrison, 1991).Inclusion bodies may consist of not only theforeign protein, but also of some host compo-nents such as T7 RNA polymerase, membranecomponents, 16S and 23S RNA and even plas-mid DNA (Mitraki and King, 1989). In somecases the formation of inclusion bodies can be

advantageous, since the expressed protein canbe purified simply by washing and pelletingthe aggregated bodies (Georgiou and Valax,1999). However, often the aggregation of theprotein is detrimental, therefore expressionconditions must be optimized so that thepossibility of producing soluble protein isincreased. The maximized solubility of aprotein can be achieved by optimizing severalfactors such as the expression temperature, E.coli strain used, pH of the growth medium,fusion partner the protein is fused with, rateand level of expression and co-expression withchaperones and foldases (Murby et al., 1996;Zhang et al., 1998; Weickert et al., 1996). Forexample, fusions to ZZ (two domains derivedfrom staphylococcal protein A) and thio-redoxin have been reported to increase thesolubility of insulin-like growth factor I andcytokines (Samuelsson et al., 1994; LaVallieet al., 1993). Chaperone proteins mediate boththe folding and rehabilitation of proteins (vanDyk et al., 1989; Bukari and Zipser, 1973).The most extensively studied chaperons/foldases in E. coli are GroEL/GroES, whichhave been reported to assist in folding proteinsinto active forms (Goloubinoff et al., 1989;Lee and Olins., 1992; Yasukawa et al., 1995;Amrein et al., 1995; Wall and Plückthun,1995). Moreover, co-overproduction ofthioredoxin has been reported to increase thesolubility of insulin-like growth factor I as wellas other vertebrate proteins expressed in E. coli(Yasukawa et al., 1995). However, even if co-overexpression of chaperons/foldases (e.g.,GroES, GroEL, DnaK) is likely to be usefulfor some proteins, it is unlikely to provide auniversal solution to the problem of inclusionformation (Hockney, 1994). Other factors af-fecting protein solubility include regulation ofthe expression vector and bacterial cell lysismethods. Protein inclusion bodies are usuallysolubilized using urea or guanidine hydro-cloride (Gdn-HCl). After denaturation the pro-tein must be renatured under conditions thatfavor their proper folding and native functions.Conditions that must be considered when

Introduction

5

trying to refold a denatured protein involveoptimizing the renaturation temperature(usually around 10°C), redox conditions,protein concentration, and addition of solubi-lizing additives such as polyethylene glycol.Soluble and active proteins have also beenobtained with treatment with sarkosyl (N-dodecanoyl-sarcosinate, Frangioni and Neel,1992).

1.1.3. Secretion

Secretion of foreign proteins to the mediumcan greatly facilitate their purification.However, secretion of a protein to the mediumby Gram-negative bacteria, such as E. coli,requires translocation of the protein throughthe bacterial cell wall consisting of inner andouter membrane bilayers surrounding the peri-plasmic space. Only a limited number of E.coli strains are capable of secreting proteinsfrom the cell (reviewed in Pugsley, 1993).Nevertheless, some genetically modifiedproteins with inserted secretion sequence sig-nals, have been successfully secreted into themedia (H�gset et al., 1990; Wadensten et al.,1991; Weiss et al., 1994; Samuelsson et al.,1994). More importantly, E. coli is capable ofsecreting genetically altered proteins into theperiplasm, which may represent 20-40% of thetotal cellular volume. With this strategy thetarget protein can be separated from the cyto-plasmic proteins and accumulate in a moreoxidizing environment where disulfide-bondformation can occur (Missiakas et al., 1993).Secretion into the periplasmic space has beensuccessfully utilized in large scale productionof biologically active antibody fragments(Plückthun, 1992).

1.2. YEAST (SACCHAROMYCES CEREVISIAE)Saccharomyces cerevisiae (bakers and brew-

ers yeast) is generally considered as a safe or-ganism for the production of foreign proteins,since it has been used in food production forthousands of years (Müller et al., 1998). Thisunicellular eukaryote has been extensively

studied at the biochemical, molecular andgenomic levels, and its entire genome has beenrecently sequenced (Goffeau et al., 1997). LikeE. coli, yeast grows rapidly in relatively in-expensive and simple media, is easy to trans-form, and maintain (Buckholz and Gleeson,1991). But unlike E. coli, yeast is an eukaryoticcell, and therefore expresses and processes pro-teins more similarly to higher eukaryotes. (LinCereghino and Cregg, 1999). The secretorypathway of yeast closely resembles that ofmammalian cells, thus yeast is capable of manyposttranslational modifications such as pro-teolytic processing, disulfide bond formation,and glycosylation (Eckart and Bussineau,1996). Thus, yeast offers a eukaryotic proteinproduction system that can retain the foreignprotein intracellularly or the product can besecreted to the medium if fused to a carrierpeptide. However, S. cerevisiae is unable toperform certain complex eukaryotic posttrans-lational modifications such as prolylhydroxy-lation and amidation. In addition, the glyco-sylation of yeast can differ from that of highereukaryotes (Sudbery, 1996). Nevertheless,many foreign proteins have been produced ona large scale using S. cerevisiae, includinghuman serum albumin, the hepatitis B viralsurface antigen, insulin and hirudin (Goodly,1993; Valenzuela et al., 1982; Thim et al.,1986; Ladisch and Kohlmann, 1992; Mendoza-Vega et al., 1994).

1.2.1. Episomal vectors and chromosomalintegration

Recombinant DNA which encodes foreignproteins can be expressed in S. cerevisiae asan episomal plasmid or integrated into itsgenome. Extra-chromosomal replicons arebased either on plasmids containing auto-nomously replicating sequences, or on the na-tive 2µ circle of S. cerevisiae (Romanos et al.,1992). Most yeast expression vectors havebeen constructed from the multi-copy 2µplasmid, which has also been engineered toreplicate and be selectable in both E. coli and

Introduction

6

yeast. Because the native 2µ circle is presentin most S. cerevisiae strains at approximately60-100 copies per haploid genome, and theplasmid is stable inherited, 2µ-based proteinproduction can be very efficient (Futcher,1988; Unternährer et al., 1991). The most com-mon selection markers are LEU2, TRP1, URA3and HIS3 in corresponding amino acid (aa)uptake deficient mutant strains, which areauxotrophic for leucine, tryphophan, uracil andhistidine, respectively. Selection can also beperformed using e.g., the antibiotic resistanceelement G418, a 2-deoxy-streptamine antibi-otic (Jimenez and Davies, 1980). There arealso yeast strains engineered to ensure plasmidmaintenance irrespective of the culture condi-tions, for example, ura3 fur1 -strains are non-viable on uracil minus media since they areblocked both in the de novo and salvage path-ways of uridine 5’-monophosphate synthesis(Loison et al., 1986; Napp and Da Silva, 1993).These strains can grow on uracil minus mediaonly if they maintain a plasmid that containsthe URA3 gene, which complements themutant allele.

Chromosomal integration offers a morestable alternative to episomal maintenance ofplasmid DNA coding for foreign proteins. Themost widely utilized technique to obtain inte-gration is homologous recombination (re-viewed in Romanos et al., 1992). Integrationvectors are similar to the above mentionedyeast plasmids with both E. coli and yeastselectable markers and bacterial replicationsequences, but lack yeast replication se-quences. Once the plasmid has been trans-formed into a yeast cell, usually in a linearform, it can not replicate, and is maintainedonly if it integrates into the yeast genome.Homologous recombination can occur by asingle or double crossover, either as a multi-copy integration into ribosomal DNA, or bytransposable elements such as Ty or gamma(Fleer, 1992). Double cross-over vectors con-tain the foreign DNA and selection markerflanked by yeast DNA homologues to the 5’and 3’ regions of the chromosomal DNA to

be replaced. The frequency of transformationby this method is lower than episomal plasmidtransformation, but it results in very stable,usually single copy transformants (Romanoset al., 1992). Multicopy integration of heter-ologous genes usually employs integration intoreiterated chromosomal DNA sequences thatmay be present from 20 to 140 repeats perhaploid genome. Chromosomal integrationcan also be obtained by using Ty transposi-tion vectors, which are analogous to theretroviral vectors of mammalian cells (Boekeet al., 1988).

1.2.2. Promoter strength

Many different yeast transcriptional promot-ers are available for foreign protein expres-sion in S. cerevisiae (Fleer 1992). Yeastpromoters consist of at least three elementswhich regulate the efficiency and accuracy ofthe initiation of mRNA transcription. Theseelements are upstream activation sequences(UAS), the TATA-box and initiator elements(Romonos et al., 1992). Numerous strongyeast promoters to be utilized are either in-ducible (e.g., GAL1, GAL7, GAL10, MET25),or constitutive (e.g., CYC1, ADH1, TEF2,GPD and MF�) (Mumberg et al., 1994 and1995). In the case of large scale proteinproduction in yeast, it is crucial to be able toseparate the cell growth phase from the proteinproduction phase (Da Silva and Bailey, 1991).The promoter strength of three different pro-moters i.e., SUC2, PGK and GAL7 were stud-ied when �-amylase was produced in 30 hourfed-batch yeast culture (Park et al., 1993).According to the results the efficiency of thepromoters appeared to be in the followingorder PGK> GAL7> SUC2, while plasmidstability and promoter strength appeared to beinversely correlated. Also expression vectorswith a tetracycline and copper ion regulatablepromoter systems have been developed (Gariet al., 1997; Labbé and Thiele, 1999). Anotherapproach to optimize gene expression in yeastutilizes transactivators such as mammaliansteroid hormone receptors, which have been

Introduction

7

reported to increase transcriptional activity(Fleer, 1992), as well as regulate the overpro-duction of the GAL4 protein (Schultz et al.,1994).

1.2.3. Secretion of heterologous proteins

S. cerevisiae is capable of secreting heter-ologous proteins, but most of them have to befused to an appropriate pre-pro-leader(Romanos et al., 1992; Shuster, 1991). Thesecarriers are relatively short amino acid se-quences and include pre-pro-leaders from suchproteins as �-factor, killer-toxin, bar protease,gp37, and Hsp150 (reviewed by Simonen,1994). The carrier peptide may have a crucialrole in the transport process from the ER tothe Golgi, which has been suggested to be re-ceptor- mediated (Balch et al., 1994). The car-rier may thus provide a positive secretion sig-nal for the capture of fusion protein into vesiclebudding sites at the ER membrane (Suntio etal., 1999). The carrier may also assist the fu-sion partner to acquire a secretion-competentconformation (Simonen et al., 1994; Simonenet al., 1996; Hammond and Helenius, 1995).

Some proteins can be externalized in S.cerevisiae without a carrier. Human lipo-cor-tin-1 was first expressed as a fusion with the

�-factor pre-pro-region but only 10% of theprotein was secreted (Chung et al., 1996). Anexpression system utilizing the inulinase sig-nal peptide increased secreted lipocortin-1 5-fold, resulting up to 95% of the protein to besecreted into the medium (Chung et al., 1996).Furthermore, fed-batch fermentation of theheterologous protein produced 2.1 g/l, repre-senting more than 80% of the total extracellu-lar protein (Chung et al., 1999). Also, it hasbeen observed that culture growth conditionscan affect protein secretion (Rossini et al.,1993). For example, ß-galactosidase, which isnormally refractile to secretion, is secretedfrom yeast grown in a rich medium. ß-galac-tosidase secretion could be further increasedby elevated growth temperatures.

1.2.3.1. Yeast secretory proteins ascarriers

In S. cerevisiae the most commonly used car-rier is the pre-pro-region of the �-factor (Fig.1A and 1B). The pre-sequence (signal se-quence) is cleaved upon translocation into theER and the pro-region is cleaved by the Kex2protease in the Golgi compartment (Julius etal., 1984). However, the mechanisms whichthe �-factor pre-pro-region uses to confer

Figure 1. A Schematic presentation of primary translation products of S. cerevisiae mating �-factor(A), pre-pro-region of �-factor used as a carrier (B), killer toxin (C) and pre-pro-region of killer toxinused as a carrier (D). Pre-pro-killer toxin carrier comprises killer toxin residues 1-34, which are joinedby alanine to the gamma-region residues 177-233. The Kex2 cleavage and N-glycosylation sites areindicated by arrows and asterisks, respectively. The signal sequence is marked by a horizontally stripedarea and the repetitive region of �-factor is colored gray.

Introduction

8

secretion competence to the fusion partners,as well as its structure, are not known. Syn-thetic leaders have also been used as carriers(Kjeldsen et al., 1997 and 1998b). Interest-ingly, it appears that three N-linked oligosac-charide chains are necessary for secretion com-petence of the �-factor pre-pro-leader, whereassynthetic pre-pro-leaders lacking the consen-sus N-glycosylation sites confers secretioncompetence of correctly folded insulin precur-sors (Kjeldsen et al., 1998a and 1998b).

Another polypeptide used as a carrier inyeast is derived from the killer toxin (Fig. 1D).The K1 killer toxin is a heterodimer of twoprotein subunits (� and ß), maturated from the316 residue pre-pro-toxin (Fig. 1C) (Bostianet al., 1984). As in the pre-pro-region of �-factor, the signal sequence of the pre-pro-toxinis cleaved upon translocation into the ER andthe pro-region is subsequently cleaved by theKex2 protease in the Golgi compartment(Redding et al., 1991). The ability of differentfragments of the killer toxin pre-pro-region toconfer secretion competence for heterologousprotein was studied by fusing ß-lactamase tovarious fragments of pre-pro-toxin (Cartwrightet al., 1992). The most efficiently secreted con-struct had a carrier in which the genes of theß-subunit and the control region of � weredeleted (Fig. 1D). Secreted ß-lactamasereached level of several µg/ml which is equalto the level produced by the pre-pro-region ofthe �-factor (Cartwright et al., 1994).

1.2.3.2. The secretion pathway of S.cerevisiae

The secretory pathway in eukaryotic cellsis essential for posttranslational processing,targeting and transporting proteins to intra- andextracellular membranes and for proteins des-tined for export out of the cell. The secretorypathway of yeast is depicted in Figure 2. Itconsists of several membrane-bound compart-ments and vesicle-trafficing systems betweenthese compartments. Protein secretion in yeasthas been studied using temperature-sensitivesecretion (sec) mutants, in which the transport

of proteins can be reversibly blocked in dis-tinct compartments (Kaiser and Schekman,1990). Secretery proteins have a signal pep-tide, which is an endoplasmic reticulum (ER)targeting sequence present usually at their N-terminus, which allows the protein to enter theER. Most posttranslational modifications, suchas the initiation of glycosylation, disulfidebond formation and protein folding occur inthe ER. Transport to the next compartment,the cis Golgi, occurs by vesicular transport,which is also the way proteins are transportedthrough the Golgi (intra-Golgi) to the transGolgi network (TGN) (Rexach and Sheckman,1991). Transport, docking, and fusion of trans-port vesicles are regulated by proteins gener-ally refered to as SNAREs (reviewed by Gerst,1999). The anterograde and/or retrogrademembrane traffic between the ER and Golgioccurs in vesicles coated with COPI and COPIIproteins (Bednarek et al., 1995; Schekman andMellman, 1997; Orci et al., 1997). In the TGNproteins are sorted into distinct vesicles, which

Figure 2. A schematic presentation of the secre-tory route of yeast S. cerevisiae. The nascentsecretory proteins are translocated into the lumenof the endoplasmic reticulum (ER), from wherethey are transported by carrier vesicles via theGolgi to the vacuole, the plasma membrane, thecell wall or to the growth medium. Reproductionfrom Saris (1998) with permission.

Introduction

9

are destined to the plasma membrane or tovacuoles (Griffiths and Simons, 1986;Conibear and Stevens, 2000). The sorting tovacuoles occurs via a late endosome-likeprevacuolar compartment (Horazdovsky et al.,1995; Gerrard et al., 2000; Götte and Lazar,1999).

The ER signal sequence consists of a shorthydrophobic peptide with hydrophilic residuesat both ends (von Heijne, 1985; Rapoport etal., 1996). Polypeptides are translocated intothe lumen of the ER usually cotranslationally,although some proteins translocate post-translationally. During co-translational trans-location, the signal sequence of a nascentpolypeptide is first recognized by the signalrecognition particle (SRP) (Walter andJohnson, 1994). After binding to the SRP,translation is arrested until the complex isbound to the ER membrane via the SRPreceptor. The translocation of proteins occursthrough translocation channels containinghetero-trimeric complexes of the Sec61 protein(Hanein et al.,1996; Hamman et al., 1997).The signal peptides of proteins which trans-locate posttranslationally are less hydrophobicand are thus independent of SRP (Ng et al.,1996; Rapoport et al., 1996). Posttranslationaltranslocation requires a complex of Sec62,Sec63, Sec71 and Sec72 proteins in additionto the Sec61p translocon channel (Panzner etal., 1995; Brodsky and Scheckman, 1993). TheBiP chaperone has a critical role in trans-location. It binds to the translocating poly-peptide, acting as a molecular ratchet thuspreventing the backsliding of the protein(Matlack et al., 1999).

The ER is the main location where secre-tory proteins fold and it contains severalfolding enzymes and chaperone proteinsessential for proper protein folding and trans-port (Hong, 1996). The lumen of the ER is anoxidizing environment and contains highlevels of calcium, which provides properconditions for chaperones and disulfide bondformation (Lodish et al., 1992; Hwang et al.,1992). Signal peptidase and glycosyl-

transferase complexes are functionally asso-ciated, and signal sequence cleavage andglycosylation, together with protein folding oc-cur cotranslationally in the ER. In yeast cells,proteins are glycosylated with N-linked andO-linked glycans. Glycosylation is assumedto promote correct folding, protect againstproteolysis and thermal denaturation, as wellas to regulate intracellular trafficking (Lis andSharon, 1993). In O-linked oligosaccharides,the first mannose residue is attached to a serineor threonine in the ER, which is unlike mam-malian cells where mucin-type O-glyco-sylation is initiated in the Golgi by adding anN-acetyl-galactosamine to the fully foldedprotein (Tanner and Lehle, 1987; Lussier etal., 1995; Van den Steen et al., 1998). O-linkedoligosaccharide chains in yeast can be elon-gated up to 5 mannoses in the Golgi (Fig. 3A)whereas O-glycans in mammalian cells areelongated by adding mannose, galactose, N-acetyl glucosamine, fucose, sialylate residuesand/or polylactosamine-extensions (Varki,1998, Van den Steen et al., 1998). N-glyco-sylation of secreted glycoproteins in yeast isinitiated in the ER (Helenius, 1994, Heleniuset al., 1992). In yeast, the core oligosaccha-ride consisting of two N-acetyl glucosamine,nine mannose residues and three glucose resi-dues is similar to that of higher eukaryotes (Fig.3B) (Kukuruzinska et al., 1987). Unlikemammalian cells, the mannose residues are notreplaced by other sugars (N-acetyl glu-cosamine, galactose, fucose and sialic acid)in the Golgi, but the outer chains are elongatedby mannose residues only (Fig. 3C) (Byrd etal., 1982; Tanner and Lehle, 1987).

1.2.4. Optimizing expression

The genetic background of both natural andrecombinant yeast strains have been found toaffect the quantity and structure of heterolo-gous proteins (Eckart and Bussineau, 1996).Because different genetic backgrounds caninfluence transcription and translation efficien-cies, the secretory pathway, protein quality,plasmid stability and plasmid copy number, it

Introduction

10

may be necessary to screen a number ofdifferent host strains with varying geneticbackgrounds when optimizing heterologousprotein expression in yeast (Schultz et al.,1994). Some genetic background consider-ations that may aid in foreign protein produc-tion in yeast include: the use of protease-deficient strains, which can decrease heterolo-gous protein degradation thus increasing yields(Van den Hazel et al., 1996), and the use ofstrains lacking hyperglycosylation of N-linked

sites (Schultz et al., 1994; Nakanishi-Shindoet al., 1993). However, these strains usuallygrow slower than normal strains, but there arealso yeast strains that require relatively fewgenerations to obtain high production yields(Bussineau and Shuster, 1994). An N-termi-nus sequence of human interleukin 1ß couldbe utilized as an enhancer for heterologousprotein expression in the same way as has beenreported for two human growth hormones (Leeet al., 1999). And, as in E. coli, the co-expression of chaperones have been reportedto assist in the proper folding and secretion offoreign proteins in S. cerevisiae (Langer et al.,1992; Chen et al., 1994; Robinson et al., 1995).Furthermore, the overexpression of disulfideisomerase (PDI) resulted in a 10-fold increasein secreted human platelet-derived growthfactor, a four fold increase in acid phosphatase(Robinson et al., 1994), a two to eight foldincrease for five single chain antibody frag-ments (Shusta et al., 1998), and a 15 to 24-fold increase in antistasin (Schultz et al., 1994).

1.3. INSECT CELLS (BACULOVIRUS SYSTEM)The baculovirus expression system (BVS)

is commonly used to express heterologousproteins in insect cells (Kost and Condreay,1999). Since insect cells are eukaryotic,proteins expressed in them will be post-translationally modified in a manner similarto that of mammalian cells (Miller, 1988).Insect cells can be grown as suspension cul-tures, which permits the use of large-scale bio-reactors for easier production scale-up. Fur-thermore, unlike mammalian cell lines insectscells do not require CO2-incubators. The mostcommonly used baculovirus system utilizedAutographa californica Multiple NuclearPolyhedrosis virus (AcMNPV) (Jones andMorikawa, 1996) in a cell line derived fromLepidopteran Spodoptera frugiperda ovariancells (Sf9). The AcMNPV’s genome is adouble-stranded, circular DNA, approximately130 kilobases in length, and has been fullysequenced (Ayres et al., 1994). The produc-

Figure 3. Carbohydrate structures of S. cerevisiaeglycoproteins. (A) The first mannose of an O-linked oligosaccharide is attached to a serine (Ser)or threonine (Thr) residue in the ER and the chainis elongated up to pentamannosides in the Golgicompartment. (B) An N-acetyl glucosamine isattached to an asparagine (Asp) residue to fromthe core structure of N-linked oligosaccharide inthe ER and (C) the outer chains are elongated inthe Golgi. The length of the outer chain variesfrom 2 to 15 mannose residues.

Introduction

11

tion systems are commonly regulated by strongpromoters like polyhedrin and p10 but alsoimmediate early promoters (for example ie1)have been utilized (Jarvis et al., 1990). Latterexpression systems gave a continuous andstable expression of human glycoprotein inboth infected and transformed Lepidopterancells (Jarvis et al., 1996).

As an alternative to baculovirus expressionsystems, an expression cassette for continu-ous protein expression by transformed insectcells have also been developed (Farrell et al.,1998). The system utilizes the promoter of thesilkmoth cytoplasmic actin gene, the ie1transactivator gene and the HR3 enhancer re-gion of Bombyx mori MNPV to stimulate geneexpression. Levels of produced proteins in thissystem are comparable to the ones producedin BVS (Lu et al., 1997; Keith et al., 1999).

1.3.1. Systems utilizing polyhedrin andp10

In general, wild type baculoviruses exhibitlytic and occluded life cycles, producing ex-tra-cellular and occlusion derived virus(ODV), respectively. ODV is embedded inproteinaceous viral occlusions called poly-hedra (Rohrmann, 1986). Polyhedrin, a 29 kDprotein, is the major component of polyhedra,which in turn are crystalline occlusion bodiesvisible in the nuclei of infected cells by lightmicroscopy. The expression of viral genes isregulated in a successive cascade consistingof 4 distinguishable phases: early, delayedearly, late and very late (Summers and Smith,1988). During the very late phase of viralreplication two very abundant mRNAs are pro-duced, one of which codes for the polyhedrinprotein and other codes for the p10 protein (10kDa). The p10 polypeptide is involved in theformation of fibrillar structures found both inthe nucleus and in the cytoplasm of baculo-virus infected cells. The role of polyhedra isto protect occluded baculoviruses after thedeath of infected insects, and its presence isessential for the maintenance of the virus innature. However, viral propagation in cell

cultures is based on virions which bud frominfected cells, not on ODV. Thus, baculovirusexpression vectors have been engineered inwhich the polyhedrin and p10 coding regionsare deleted, but their strong promoters are re-tained to express foreign proteins in largeamounts in the late and very late phases ofinfection (Smith et al., 1983b; Kost andCondreay, 1999).

In the baculovirus expression system utiliz-ing the p10 promoter, the sequence encodingthe p10 protein as well as the entire polyhedrinlocus, including its promoter, have beendeleted (Vlak et al., 1990; Roelvink et al.,1992; Bonning et al., 1994; Naggie andBentley, 1998). The remaining p10 promoteris used to express recombinant protein se-quences. In this way, expression of both thenative polyhedrin and p10 proteins are abol-ished, which reduces background problemscaused by the expression of large amounts ofvery late native proteins while increasing theexpression of foreign recombinant proteins.Furthermore, the p10 promoter is activatedearlier in the infection cycle than the poly-hedrin promoter (Roelvink et al., 1992;Bonning et al., 1994). The absence of the p10protein results in delayed cell lysis, and therebyan extended period of recombinant protein syn-thesis. The yields of two reporter enzymes, ju-venile hormone esterase and ß-galactosidasewere higher under the p10 promoter comparedto polyhedrin controlled expression (Bonninget al., 1994).

1.3.2. Recombination

In baculovirus expression systems the inser-tion of genes coding for foreign proteins canbe accomplished by homologous recombina-tion, site-specific transposition (Bac-to-Bac)or insertion directly into the viral genome invitro. In homologous recombination, the geneof interest is cloned into a transfer vectorcontaining a baculovirus promoter flanked bybaculovirus DNA derived from the polyhedringene. After transfection into insect cells, thegene is inserted into the genome of the parent

Introduction

12

virus by homologous recombination, at a rateof approximately 0.1 to 1% (Smith et al.,1983a). Recombinants are identified by theiraltered plaque morphology, which appears as‘occlusion minus’ plaques. A higher rate ofrecombination can be achieved when the par-ent virus genome is linearized at site(s) nearthe target site of foreign gene insertion (Kitts

et al., 1990). Kitts and Possee (1993) reportedrecombination rates approaching 100% whenthey used linearized viral DNA that is miss-ing an essential portion of the baculovirus ge-nome downstream from the polyhedrin locus.Nevertheless, by this method it can take morethan a month to purify the plaques, and con-firm the desired recombinants. A faster methodto generate recombinant baculovirus can beachieved by utilizing baculovirus Ac-omega,which contains an unique restriction site down-stream of the polyhedrin promoter (Ernst etal., 1994). By this method recombinant viruseswere obtained 8 days posttransfection andaccording to PCR analysis the non-recombi-nant background was 25 fold lower than thatof the recombinant viral DNA. Anothermethod utilizing site-specific transposition isalso relatively fast (7 - 10 days) because plaquepurification or virus amplification are notneeded (Luckow et al., 1993). In this method,the foreign gene is cloned into a baculovirusshuttle vector (bacmid) that can replicate in E.coli, but can also infect susceptible lepi-dopteran insect cells (Fig. 4). Bacmid is a re-combinant virus that contains a mini-F repli-con, a kanamycin resistance marker, andattTn7, the target site for the bacterialtransposon Tn7 (Leusch et al., 1995). Expres-sion cassettes comprising a baculoviruspromoter driving expression of a foreign genethat is flanked by the left and right ends ofTn7, can transpose to the target bacmid in E.coli when Tn7 transposition functions are pro-vided by a helper plasmid.

1.3.3. Optimizing expression

The density at which cells are infected, andthe multiplicity of infection (m.o.i.) greatlyaffects the expression of heterologous proteins(Power et al., 1994; Licari and Bailey, 1992;Wong et al., 1995; Klaassen et al., 1999). Forinstance, when the integral membrane proteinbovine rhodpsin was expressed in largeamounts, the highest volumetric yields wereobtained with an m.o.i. of 0.01 during early tomid-exponential growth (Klaassen et al.,

Figure 4. Schematic outline of the generation ofrecombinant baculovirus with a bac-to-bac expres-sion system. A donor vector carrying the gene ofinterest flanked by a Tn7-element is transformedto E. coli. The Tn7 element can integrate into theattTn7 target site in the presence of transpositionproteins provided by a helper plasmid. Recombi-nant bacmids are selected using their antibioticresistance and recombinant bacmid DNA is usedto transfect insect cells.

Introduction

13

1999). For easier purification, and cost savingreasons, it is possible to produce heterologousproteins in Sf9 cells in serum-free and pro-tein-free media. The cells were graduallyadapted to serum-free media in monolayercultures and reached a doubling time ofapproximately 25 h, compared with 18 h inserum containing medium. Furthermore, thecontrol of proteolysis in insect cells has a im-portant role when optimizing the yield ofrecombinant protein (Naggie and Bentley,1998). Like in S. cerevisiae, unwanted man-nose residues can be added to secreted glyco-proteins and they lack penultimate galactoseas well as terminal sialic acid residues (Jarvisand Finn, 1996). The immediate earlypromoters (for example ie1), have been shownto be beneficial for protein production e.g.,when highly glycosylated or otherwise modi-fied proteins are to be produced (Bonning etal., 1994; Chazenbalk and Rapoport, 1995).

1.4. MAMMALIAN EXPRESSION SYSTEMSUTILIZING VACCINIA AND ALPHAVIRUSVECTORS

In some cases, mammalian cell systems arethe only possible way to produce properly pro-cessed and active foreign proteins. Gene trans-fer into mammalian cells may be performedeither by infection with a virus carrying therecombinant gene of interest (reviewed byMakrides, 1999) or by direct transfer of plas-mid DNA (Geisse and Kocher, 1999).

1.4.1. Vaccinia virus vector

Recombinant vaccinia virus vectors havegenerally been acknowledged as versatile toolsfor the expression of foreign genes (reviewedby Moss, 1996). Since vaccinia is infectiousto man, safety aspects must be taken into con-sideration, especially when working withlarge-scale preparations. To circumvent safetyproblems, an avian host-restricted vaccinia,modified vaccinia Ankara (MVA), can be used(Sutter et al., 1994; Wyatt et al., 1995). Fur-thermore, MVA does not produce the rapid

cytophatic effect accompanied by destructionof the cell monolayer seen with the standardreplication strains of vaccinia used. MVA hasbeen shown to efficiently produce foreignproteins in several mammalian cell lines usingeither the homologous vaccinia promoter, p11(Sutter and Moss, 1992) or a hybrid vaccinia/bacteriophage T7 promoter system (Wyatt etal., 1995). In the vaccinia/T7 polymerase hy-brid system, the vaccinia virus contains the T7RNA polymerase gene, which is under thecontrol of an early vaccinia promoter. Theforeign gene to be expressed is cloned into aseparate expression vector under the controlof the T7 promoter. After the host cells areinfected with the vaccinia/T7 virus, the cellsare transfected with the expression plasmid.Problems with vaccinia systems include theircytopathic nature, and dependence on efficienttransfection rates. Mammalian cell lines canbe engineered so that the genes to be trans-fected are stably integrated into the cell´s ge-nome, but then a tightly controlled-induciblepromoter must be used. This method is espe-cially advantageous when foreign protein pro-duction must be scaled-up. Furthermore, ho-mologous vaccinia promoter systems can bemore effective in producing secreted foreignproteins as compared to the heterologous T7promoter system (Pfleiderer et al., 1995).Another vaccinia virus based expression sys-tem utilizes defective vaccinia virus lackingthe D4R open reading frame, and a comple-mentary cell line providing the D4R gene prod-uct (Himly et al., 1998). Experiments donewith human secreted proteins (i.e., factors VIIand XI) showed that the defective systemproduces more secreted proteins than the wildtype vaccinia. Surprisingly, recombinanthuman factor VII was more efficientlyproduced using the defective vaccinia recom-binant under non-complementing conditions(Himly et al., 1998). This suggests that thepersistence of early phase vaccinia replication,combined with a delay in the shutoff of hostprotein synthesis, can be advantageous forforeign protein production.

Introduction

14

1.4.2. Alphavirus vectors

Semliki Forest (SFV) and Sindbis (SIN)viruses together with Venezuelan equineencephalitis virus are used as vectors for theexpression of heterologous proteins in manydiverse cells types (Makrides, 1999; Garoffand Li, 1998; Agapov et al., 1998). Alphavirusreplicon vectors are self-replicating RNAmolecules which include the genes for non-structural proteins (Nsps), while lacking thegenes of the viral structural proteins. Insteadof the structural genes, cloning sites have beenengineered to accept foreign genes for expres-sion. The replicons are introduced into hostcells by transfection either as RNA, synthe-sized in vitro, or as DNA under the control ofan eukaryotic promoter. Alternatively, the re-combinant alphavirus RNA can be packagedinto viral particles in cells that have beencotransfected with replication competent, butpackaging incompetent helper RNA coding forthe alphavirus structural proteins (Bredenbeeket al., 1993). Although, recombinationbetween vector and helper genomes can gen-erate a fully infectious virus, this usuallyhappens at very low frequencies (Berglund etal., 1993). One way to prevent this recombi-nation is to use conditional mutations to limitthe infectivity of any such recombinants(Smerdou and Liljeström, 1999). In an im-proved packaging system, the helper virus hasbeen modified so that the structural proteinsare expressed in two separate plasmids.

As in vaccinia systems, alphavirus infectionsshuts off the host’s own protein synthesis,which makes protein purification easier.Compared to vaccinia, SFV is less cytopathicto the host, and thereby potentially more effi-cient in foreign protein production (Liljeströmand Garoff, 1991). Furthermore, non-patho-genic alphavirus vectors have been developed.For example, a Sindbis virus derivative, whichhas low host cell pathogenicity due to a singlepoint mutation in Nsp2, demonstrates apersistent infection, while its replication rateis as high as the one of the wild type SIN(Dryga et al., 1997). Although expression of

foreign proteins using alphaviruses derivedvectors can generate very high yields (up to25% of total cellular proteins), cloning of largeforeign genes into replicons can be problem-atic (Liljeström and Garoff, 1991). Foreignsequences of about 2 kb are usually stable inSIN vectors, whereas various inserts over 3kb were rapidly deleted suggesting that thepackaging capacity of the virion had beenexceeded (Pugachev et al., 1995). Besidestissue and suspension culture applications ofalphavirus replicons, in situ infections andforeign protein expression have been accom-plished in the neurons of rat hippocampalslices (Schlesinger and Dubensky, 1999). Thissystem has been used to produce ß-galactosi-dase, tissue plasminogen, chloramphenicolacetyl transferase, hemagglutinin andHepadnavirus proteins. Furthermore, deriva-tives of alphavirus vectors have been used todirect foreign protein expression in specificcell types in a tissue (Huang, 1996; Ohno etal., 1997).

1.4.2.1. Semliki Forest virus

Semliki forest virus (SFV) is an envelopedRNA virus with a single strand genome ofpositive polarity. SFV belongs to the Alpha-virus genus which in turn is a member of thefamily Togaviridae. The prototype virus ofSFV, used in laboratories, is considered to benonpathogenic to humans and has been longstudied as a model virus for the Alphavirusgenus. SFV enters the cell via receptor medi-ated endocytosis, resulting in nucleocapsidliberation of its RNA genome into the cyto-plasm (Helenius et al., 1980). Two-thirds ofthe genomic 42S RNA (total length ca. 11.5kb) is translated into a polyprotein, which isthen autoproteolytically cleaved into four non-structural proteins: Nsp1 - Nsp4 (Fig. 5) (re-viewed in Kääriäinen et al., 1987; Strauss andStrauss, 1994). In the initial stages of SFVinfection, a RNA replication complex, isformed from the intermediate translation prod-ucts Nsp123 and Nsp4 and is needed torecognize the 42S RNA plus strand for tran-

Introduction

15

scription of the negative strand 42S RNA(Lemm and Rice, 1993a, 1993b). Later ininfection the intermediate Nsp123 is pro-teolytically processed into the final products,Nsp1, Nsp2 and Nsp3, after which the tran-scription of the minus strand 42S RNA willbe terminated. According to the geneticcriteria, all nonstructural proteins are neededto form the replication complex. Replicationoccurs in association with modified endo-somes and lysosomes called cytopathic vacu-oles type I (CPVI) (Froshauer et al., 1998;Peränen and Kääriäinen, 1991). The structuralproteins: capsid protein and envelope glyco-proteins E1, E2 and E3 are transcribed from asubgenomic 26S RNA (reviewed in Straussand Strauss, 1994). The functions of individualNsps are briefly reviewed below.

1.4.2.1.1. Nsp1Nsp1 (537 aa) is multifunctional enzyme

catalyzing the capping of viral mRNA. Firstly,it is a methyltransferase (MT) (Mi et al., 1989,Mi and Stollar, 1991; Laakkonen et al., 1994).Nsp1, like the methyltransferase purified fromthe vaccinia virion can methylate GTP in thepresence of S-adenosyl-L-methionine(AdoMet) (Laakkonen et al., 1994, Martin andMoss, 1976). Secondly, Nsp1 has been shown

to form a covalent complex with 7-methyl-GMP thus providing at least the first reactioncatalyzed by guanylyltransferases (GT) (Aholaand Kääriäinen, 1995). The capping reactionof viral mRNA catalyzed by Nsp1 differs fromthe capping of eukaryotic mRNA and manyother viral mRNA by requiring methylationof guanine before covalent complex formationbetween the guanylyltransferase and 7-methyl-guanylate (Fig. 6) (Ahola and Kääriäinen,1995). The 7-methyl-guanylate is probablybound to Nsp1 through a phosphoamino bondeither on a lysine or on a histidine. Further-more, Nsp1 also assists in the initiation of mi-nus strand synthesis (Hahn et al, 1989b).

When Nsp1 is expressed alone in HeLa cellsby transfection, it associates with the plasmamembrane as well as with endosomes and ly-sosomes (Peränen et al., 1995). The tight mem-brane association of Nsp1 is partly due topalmitoylation on amino acids Cys418-Cys420

(Peränen et al., 1995; Laakkonen et al., 1996).When the palmitoylation sites are mutated, theprotein remains enzymatically active andbound to membranes (Laakkonen et al., 1996),though less tightly as compared to the wildtype (i.e., the nonpalmitoylated mutant formcould be released from the membranes with 1M NaCl, whereas wt Nsp1 could be releasedonly with 50 mM sodium carbonate, pH 12).

Figure 5. Proteolytic processing of the nonstructural and structural proteins of Semliki Forest virus.

Introduction

16

In addition, the morphology of cells infectedor transfected with nonpalmitoylated Nsp1differs from wt by showing fewer filopodialike structures than the wt Nsp1.

1.4.2.1.2. Nsp2According to what is presently known, Nsp2

(799 aa) is responsible for three differentenzymatic activities in SFV replication. Firstly,it is an autoproteinase containing a papin-likethiol proteinase at its C-terminal region (Hardyand Strauss, 1989). Secondly, it is an RNAhelicase (Gomez de Cedron et al., 1999),which has single stranded RNA-stimulatedATPase and GTPase activities (Rikkonen etal., 1994). Thirdly, Nsp2 is a triphosphatase,which catalyzes the first reaction in mRNAcapping (Vasiljeva et al., 2000). Moreover,Nsp2 has been reported to act on the regula-tion of negative strand synthesis, and isrequired for the synthesis of the subgenomic26S RNA (Hahn et al., 1989b). About half ofthe Nsp2 is transported into the nucleus duringinfection, and this transport is mediated bynuclear signal sequences (Peränen et al., 1990;Rikkonen et al., 1992). SFV in which thenuclear targeting signal of Nsp2 is removed,has been reported to be apathogenic in mice(Rikkonen et al., 1996).

1.4.2.1.3. Nsp3Nsp3 (482 aa) is an essential subunit for the

replication complex although its specific func-tions are still not known. Studies of SIN Nsp3have suggested that the polyprotein interme-diates have distinct essential functions in thesynthesis of negative strands during the earlyphases of RNA replication (Lemm et al., 1994;Shirako and Strauss 1994). SIN Nsp3 has beenreported to affect negative-strand RNA syn-thesis, and possibly also the synthesis ofsubgenomic mRNA (Wang et al., 1994;LaStarza et al., 1994b). When the localizationof SFV Nsp3 was studied by indirect immun-ofluorescence it was found to be mainly invacuole like structures both in infected as wellas transfected cells (Peränen and Kääriäinen,1991).

The N-terminal region of different Nsp3s arevery conserved among the alphaviruses, andpart of that sequence has homology to rubellavirus, hepatitis E virus, and coronavirusessequences (X-motif, Koonin and Dolja, 1993).Recently, through genome sequencing, it hadbecome apparent that this domain can be foundin bacteria, animals and plants and it showshomology to the nonhistone region of macrohistone 2 (Pehrson and Fuji, 1998). TheNational Center for Biotechnology Informa-tion (NCBI) has assigned this domain as

� �

��������������� ��������������

������ ���������� ������ ����������

���������������������������� �������������������������

������������� ���������������� ������������� �����������������

�������������������������� �������������������

�������������� ������������������ ����������������� ���������������

����������������������� ����������������������������

������������������ ���������������� ��������������� ���

���������������

Figure 6. Reactions involved in capping of alphavirus mRNA (A) and eukaryotic mRNA (B).

Introduction

17

DUF27 with an unknown function. Moreover,this domain has significant homologue to oneyeast ORF which has been associated withadenosine-diphosphate-ribose-1-phosphate(Appr-1"-p) processing activity (Martzen etal., 1999). Conservation of this region duringevolution suggests an important function(s) forthis domain.

The C-terminal part of alphavirus Nsp3 isnot conserved among alphaviruses, and var-ies both in sequence and in length from 134aa (Middel-burg virus) to 246 aa (O’nyong-nyong virus) (Strauss and Strauss, 1994). Thisdomain is rich in acidic residues, as well as inserine and threonine and appears to be devoidof any predicted secondary structures. Theregion also contains duplicate amino acid (aa)sequences i.e., ADVPEPA, PAPR andTFGDFD, suggesting that this region hasevolved through duplication events. Nsp3 isthe only alphavirus nonstructural protein modi-fied by phosphorylation (Peränen et al., 1988;Li et al., 1990).

1.4.2.1.4. Nsp4Even though direct evidence is still lacking,

Nsp4 (614 aa) most probably is the catalyticsubunit of the alphavirus RNA-replicationcomplex (Barton et al., 1988). Nsp4 hasconserved a GDD motif found in other viralRNA polymerases (Argos, 1988), and a SINtemperature sensitive mutant, ts6, has beenreported to fail in the synthesis of RNA due toa mutation in Nsp4 (Sawicki et al., 1981a1981b, Hahn et al., 1989). Nsp4 has also beenreported to be an autoproteinase (Takkinen etal., 1990).

2. PURIFICATION STRATEGIES FORRECOMBINANT PROTEINS

Purification is an important step in the pro-duction of recombinant proteins. The char-acteristics of industrial-scale purificationschemes, such as conventional chromatogra-phy, have a significant impact on the finalcost of production. It is often more efficient

to use one of the available tags to ‘fish out’the target protein. The purpose for which theprotein will be used determines requireddegree for its purity and authenticity. Topurify intracellularly produced proteins thecells are harvested and lysed, which naturallycontributes to the complexity of the proteinmixture. The advantage of intracellular ver-sus secreted protein is the volume to behandled, i.e., the secreted proteins usuallyexist diluted in the culture medium. Isolationof a desired protein from the medium followsthe general scheme: (1) concentration by pre-cipitation, ultrafiltration, batch adsorption orpartition in an aqueous phase system (2)enrichment by chromatography or partition(3) high resolution purification: by chroma-tography and/or immuno adsorption (4) finalconcentration (Menge et al., 1987). If fetalcalf serum (FCS) is used in cultivation,bovine serum albumin and globulins will bethe most abundant proteins in the superna-tant. The interaction of albumin especiallywith hydrophobic proteins represents asignificant problem for the effective purifi-cation of minor constituents.

Even though there are many existingexamples as well as heuristics and complexalgorithms for suggesting the potential ofpurification processes, no universal schemehas been developed which could be appliedto all proteins. There are a variety of differ-ent methods which can be used to purify pro-teins, but only chromatographic purificationmethods such as affinity, ion exchange,reversed phase and gel filtration chromatog-raphy will be briefly reviewed here.

2.1. CONVENTIONAL CHROMATOGRAPHYChromatographic methods (size exclusion,

ion exchange, reversed phase, hydrophobicinteraction and affinity based) can be utilizedeither in traditional, low pressure or high-performance liquid chromatography instru-mentation. In addition to liquid-solid chroma-tography there is liquid-liquid chromatogra-

Introduction

18

phy which is analogous to gas-liquid chroma-tography (Cichna et al., 1995). There are alsomany other types of chromatography methodse.g., liquid adsorption chromatography, frontalchromatography, displacement chromatogra-phy covalent chromatography and membranechromatography (Mao et al., 1993; Heeter andLiapis, 1998; Freitag, 1999; Caldas et al.,1998; Zeng and Ruckenstein, 1999).

2.1.1. Ion exchange chromatography

Ion exchange chromatography (IEC) hasbeen the most widely used technique for theisolation and purification of biological macro-molecules since the 1950s (Choudhary andHorváth, 1996). IEC is able to separate almostany type of charged molecules, from large pro-teins to small nucleotides and amino acids. Theion exchanger are insoluble solid matricescontaining fixed ionogenic groups which bindreversibly to sample molecules (proteins, etc.).Desorption is then brought about by increas-ing the salt concentration or by altering thepH of the mobile phase. The two major classesof ion exhangers are cation exchangers andanion exchangers, having negatively and posi-tively charged functional groups, respectively.Ion exchange containing diethyl aminoethyl(DEAE) or carboxymethyl (CM) groups aremost frequently used.

Since the protein is covered by a hydrophiliclayer, electrostatic interactions have a majorrole in the retention of proteins. In addition tosize, geometric form as well as hydrophopicand van der Waals interactions affect the sepa-ration (Stålberg, 1999). Nevertheless, the ma-jor property which govern the adsorption toan ion exchanger is the net surface charge ofthe protein. Since surface charge is the resultof weak acidic and basic groups on proteins,separation is highly pH-dependent. The opti-mum pH range for IEC for many proteins iswithin 1 pH units of the isoelectric point. Manyretention models have been examined but thecomplexity of the adsorption process for a pro-tein to solid surfaces makes it difficult to con-struct physical models for the interactions

(Stålberg, 1999). Gradient elution with increas-ing salt concentration is most commonly usedin the IEC. The higher the net charge of theprotein, the higher the ionic strength neededto bring about desorption. Thus, to optimizeselectivity in ion exchange chromatography,the pH of the running buffer is chosen so thatsufficiently large net charge differences amongthe sample components are created.

2.1.2. Reversed phase chromatography

Reversed phase chromatography (RPC) canbe utilized to separate compounds accordingto their hydrophobicity (Geng and Regnier,1984). Unlike many other methods RPC isable to separate closely related and structur-ally disparate substances, even at picomolarlevels (Aguilar and Hearn, 1996; Pearson etal., 1982). In RPC, silica particles coveredwith chemically-bonded hydrocarbon chainsrepresent the lipophilic phase (C2 to C18),while an aqueous mixture of an organic sol-vent surrounding the particle represents the hy-drophilic phase (Henry, 1991; Zhou et al.,1991). Depending on the extractive power ofthe eluent, a greater or lesser part of the samplecomponent will be retained reversibly by thelipid layer of the particles. The partitioning ofthe sample components between the twophases will depend on their respective solu-bility characteristics. Less hydrophobic com-ponents will end up primarily in the hydro-philic phase while more hydrophobic ones willbe found in the lipophilic phase. This can beaffected by the addition of an organic solventwhich is soluble in the hydrophilic phase.Some commonly used organic solvents, inorder of increasing hydrophobicity are metha-nol, propanol, acetonitrile, and tetrahydrofu-ran. Separated components can be directlysubjected to further analysis such as Edmansequencing or electrospray mass spectrometry.

The ability of a stationary phase (lipophilicphase) to discriminate between two compo-nents is reflected by the volume between thepeak maxima of the corresponding zones afterpassing through the column (Aguilar and

Introduction

19

Hearn, 1996). Along with partitioning, mecha-nism adsorption operates at the interfacebetween the mobile and the stationary phases(Melander et al., 1984). Thus, the retention ofhydrophobic components will be greatly in-fluenced by the thickness of the lipid layer. AC18 layer is able to accommodate more hydro-phobic material than C8 or C2 ones. For hy-drophilic components, changing from a C18to a C2 layer influences the separation verylittle since only the surface area of the lipidlayer is active. The mobile phase can be con-sidered as an aqueous solution of an organicsolvent, the type and concentration of whichdetermines the extractive power. Moreover,according to experimental data, componentsinteract with the chromatographic surface inan orientation-specific manner (Chicz andRegnier, 1990; Regnier, 1987). RPC has beenutilized for purification of a variety of proteinsand peptides e.g., aprotinin, cytochrome C,bovine serum albumin, fibrinogen, insulin andlysosyme (Honda et al., 1992; Nimura et al.,1992). Even though some proteins have beenreported to maintain their native structureduring RPC (e.g., insulin, thyroid-stimulatinghormone, growth hormone), in most casesdenaturing conditions are required whichmight limit the use of RPC (Welinder et al.,1986; Welinder et al., 1987; Forage, 1986;Chlenov et al., 1993).

2.1.3. Gel permeation chromatography

The principle of gel permeation chromatog-raphy (GPC; size exclusion or gel filtrationchromatography) is based on molecularvolumes. Large molecules are excluded fromthe matrix, whereas intermediate size moleculecan partly enter and only small molecules canfreely enter the matrix. The porous threedimensional matrix acts as a steric barrier tosolute molecules as they attempt to equilibratewith liquid inside and outside the bead. Whilepurifying the protein GPC can also be used toestimate approximate molecular weights.Furthermore, gel filtration can be used fordesalting or buffer changing when the

fractionated size of the gel is small (e.g., Bio-Gel 6, Sephadex G-25) allowing the proteinto elute in the void volume.

The choice of the appropriate column typedepends on the molecular size and physicalproperties of the proteins to be separated. Afraction of the internal volume which is ac-cessible to a solute molecule can be describedby a constant Kd, which can be calculated withexperimentally determined elution volumes.A column should be chosen so that separa-tions occur in the linear part of the Kd vs. themolecular weight plot (van Dijk and Smit,2000). However, it is practically impossibleto adjust the pH and ionic strength so that theproteins are in an equal state, since at the iso-electric point the shape of globular and poly-mer coil proteins relieve compact and softsphere, respectively. Furthermore, the ionicstrength of the buffer used may affect the sizeof the proteins. Experiments with neutral dex-trans and charged proteins showed that theeffective protein size increases with decreas-ing ionic strength due to the reduction in elec-trostatic shielding (Pujar and Zydney, 1998).Thus, although GPC is viewed as size-basedseparation process, there is considerable evi-dence for the importance of electrostatic in-teractions as well. GPC has been used in therenaturation procedure (Müller and Rinas,1999; Batas and Chaudhuri, 1999). For ex-ample heterodimeric platelet-derived growthfactor was purified from inclusion bodies af-ter denaturation with Gdn-HCl (Müller andRinas, 1999). Renaturation of this growth fac-tor involved folding into an active hetero-dimeric form during GPC while circumvent-ing aggregation observed when refolding wascarried out by dilution.

2.2. AFFINITY CHROMATOGRAPHY

Affinity chromatography together with re-combinant DNA-technology offers a simpleand fast technique to purify proteins to highpurity with a single purification step (Scouten,1991). Fusion can be made on either side or

Introduction

20

both sides of the target gene depending on spe-cific application, but the majority of fusionproteins place the tag at the N-terminus of theprotein (Nilsson et al., 1997). Genetic manipu-lation of the protein can be used to form acleavage site, which helps to remove theaffinity tag after purification thereby resultingin an intact protein. Although affinity chro-matography can be used for laboratory-scalepurification, its utilization on a preparativescale can represent a major cost for the finalprotein product.

Successful separation by affinity chromatog-raphy requires that a biospecific ligand is avail-able, and that it is covalently attached to a chro-matographic bed material. It is important thatthe biospecific ligand (antibody, enzyme, orreceptor protein) retains its specific bindingaffinity for the substance of interest (antigen,substrate, or hormone). These interactionstypically have high affinities (Kd < 10-6 M),yet are reversible when conditions are changed(Wilchek et al., 1984). Due to the specificityof this recognition, it is often possible to ob-tain 100-, 1.000- or even 100.000-fold in-creases in purity of a protein sample (Clausenet al., 1990). The packing material used, calledthe affinity matrix, must be inert and easilymodified. Agarose is the most common sub-stance used as a matrix, in spite of its relativehigh costs. The ligands, or “affinity tails”, thatare inserted into the matrix can be geneticallyengineered to possess a specific affinity. In aprocess similar to ion exchange chromatogra-phy, the desired molecules adsorb to theligands on the matrix until desorption is car-ried out e.g., with a high salt concentration, acompetition reaction (e.g., imidazole), strongchelating agents and/or low pH. Fouling of thematrix can occur when a large number of im-purities are present, therefore, this type of chro-matography is usually implemented late in theprocess. In addition to most common affinitychromatography utilizing HPLC, there are alsoother techniques which involve affinity, suchas affinity precipitation, affinity partitioningof proteins using aqueous two-phase system,

foam fractionation and dye ligand chromatog-raphy (Hoshino et al.,1998; Birkenmeier et al.,1984; Lockwood et al., 1997; Boyer and Hsu,1993).

2.2.1. Affinity tags

To date, a large number of different fusionpartners that range in size from one amino acidto whole proteins capable of selective interac-tion with ligand immobilized onto a chroma-tography matrix, have been described (Nilssonet al., 1997). Although a multitude of systemshave been introduced, no single affinity fu-sion partner is ideal for all expression or puri-fication systems. Some of the most commonlyused tags are reviewed below and listed inTable 1.

In 1975 Porath and co-workers introduceda method based on the interaction between theside chains of certain amino acids, especiallyhistidines, on a protein surface and immobi-lized transition metal ions. Immobilized metalaffinity chromatography (IMAC) systems havethree basic components: an electron donorgroup, a solid support and a metal ion. Themetal ion (usually Ni2+, Co2+, Cu2+ or Zn2+) isrestrained in a coordination complex where itstill retains significant affinity towards mac-romolecules (Porath, 1992). The use ofpolyhistidine tags has been demonstrated ina wide range of host cells including E. coli, S.cerevisiae, insect cells as well as in mamma-lian cells (Table 1). IMAC is usually performedunder nondenaturing conditions, but it is alsocompatible with non-ionic detergents allow-ing highly denaturing conditions with urea andguanidium-HCl (Smith and Roth, 1993). Alsoan organic solvent (isopropanol) can be usedto increase the purification efficiency (Frankenet al., 2000). The elution is usually performedby competition with imidazole, lowering pH,or by adding strong chelating agents. IMACis particularly suitable for preparative groupfractionation of complex extracts and bio-fluids, but can also be used in the high-perfor-mance mode. Regardless of the relativelysmall size of the his-tag, is has been reported

Introduction

21

to affect the activity of some finally purifiedrecombinant proteins (Janssen et al., 1995;Pekrun et al., 1995; Büning et al., 1996).

Maltose binding protein (MBP) from E.coli is frequently used as a fusion partner forproteins. MBP (40 kDa) is a periplasmic pro-tein and thereby can be employed to inducethe secretion into the periplasm. A mixture ofcellulose and starch can