UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Use of the cellulose-binding domain of a cellulase from cellulomonas fimi for affinity purification of… Greenwood, Jeffrey M. 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1993_fall_phd_greenwood_jeffrey.pdf [ 9.43MB ]
JSON: 831-1.0098859.json
JSON-LD: 831-1.0098859-ld.json
RDF/XML (Pretty): 831-1.0098859-rdf.xml
RDF/JSON: 831-1.0098859-rdf.json
Turtle: 831-1.0098859-turtle.txt
N-Triples: 831-1.0098859-rdf-ntriples.txt
Original Record: 831-1.0098859-source.json
Full Text

Full Text

USE OF THE CELLULOSE-BINDING DOMAIN OF ACELLULASE FROM CELLULOMONAS FIMI FOR AFFINITYPURIFICATION OF FUSION PROTEINSbyJEFFREY MICHAEL GREENWOODB.Tech. (Hons), Massey University, 1985A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESDepartment of MicrobiologyWe accept this thesis as conformingTHE UNIVERSITY OF BRITISH COLUMBIAMarch 1993© Jeffrey M. Greenwood, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of ^The University of British ColumbiaVancouver, CanadaDate ^?nay 79. 3LDE-6 (2/88)AbstractThis study describes the use of a cellulose-binding domain (CBD) from a bacterialcellulase as an affinity tag for purification of heterologous proteins. The CBD ofendoglucanase A (CenA) from Cellulomonas fimi is an N-terminal domain comprising 111amino acids. CenA binds strongly to cellulose, and the CBD retains this function whenseparated from its cognate catalytic domain by proteolysis or genetic manipulation. A series offusions between CenA and alkaline phosphatase from Escherichia coli (PhoA) were generatedusing TnphoA, and were screened for binding to cellulose. CenA-PhoA fusion proteinscontaining an intact CBD bind to cellulose, while those missing 28 or 68 C-terminal aminoacids from the CBD do not. Similarly, deletions of 18 or 44 amino acids from the N-terminusof the CBD abolish binding to cellulose. This is the first demonstration of a CBD retaining itsfunction when fused to a heterologous polypeptide. Just one CBD is sufficient to bind dimericalkaline phosphatase to cellulose. Engineered CenA-PhoA fusion proteins are purified in asingle step by affinity chromatography on cellulose, with distilled water elution. The CBD wasremoved by specific proteolytic cleavage with Factor X a or by a C. fimi serine protease. CBDfusions with human interleukin 2 (IL2) were constructed, but are predominantly insoluble andextensively degraded on expression in E. coli. However, the fusion polypeptides can still bepurified by binding to cellulose.Table of ContentsPageAbstract^  iiTable of Contents^  iiiList of Tables  viiList of Figures^  viiiList of Abbreviations^  xiAcknowledgements  xiii1. Introduction^ 11.1 Affinity tags for protein purification^ 11.2 Proteolytic cleavage of fusion proteins 141.3 Cellulose^ 161.4 Cellulose-binding^domains^ 181.5 Alkaline^phosphatase 211.6 Human^interleukin^2 231.7 Objectives 242. Materials^and^Methods^ 252.1 Chemicals, enzymes and buffers^ 252.2 Bacterial strains, cell lines, plasmids, phage and transposons^ 252.3 Media and growth conditions^ 252.4 Recombinant DNA techniques 302.5 Detection of gene expression; enzyme assays^ 322.6 Cell^fractionation^ 34iii2.7^Protein determination^  362.8^Protein gel electrophoresis/Western blots ^372.9^Cellulose preparation; binding analysis^  382.10 Protein purification; alkaline phosphatase fusion proteins ^402.10.1^CenA'-'PhoA IX-8 heterodimers: heterodimer dissociation andreassociation^  402.10.2 Medium scale purification of CBDPT-'PhoA: single columnloading and elution with guanidine-HC1 or distilled water^ 422.10.3^Purification of CenA-PhoA fusion proteins: multiple column loadingand distilled water elution^  442.10.4^Removal of CBDCenA following protease digestion^ 462.11 Protein purification: IL2 fusion proteins^  462.11.1^Direct binding of CenAl-IL2 to Avicel  462.11.2^CenA'-IL2 purification: direct binding to Avicel^ 482.11.3^CenA'-IL2 solubilization trials^  482.11.4^CenA'-IL2 purification: alkaline solubilization^ 492.11.5^CBDPT-IL2 purification: alkaline solubilization  502.12 Protease digestion of fusion proteins^  502.12.1^Factor Xa digestion of alkaline phosphatase fusion proteins^ 502.12.2^C. fimi protease digestion of alkaline phosphatase fusion proteins.... 522.12.3^Factor Xa digestion of interleukin 2 fusion proteins^ 523. Results^  543.1^CenA-PhoA fusion proteins^  543.1.1^Construction of cenA'-'phoA gene fusions with TnphoA ^ 543.1.2^Filter paper binding assay for CenA'-'PhoA fusion proteins^ 573.1.3^Avicel binding of CenA'-'PhoA fusion proteins^ 57iv3.1.4^CBDCenA N-terminal deletions: construction of pTZ18R-CBDPT-'phoAA1 and pTZ18R-CBDPT-'phoAA2^ 613.1.5^Cellulose binding of CBDPT-'PhoAAl and CBDPT-'PhoAA2^ 643.1.6^CenA'-'PhoA IX-8 heterodimers: binding to Avicel^ 643.1.7^CenA'-'PhoA IX-8 heterodimers: heterodimer dissociation^ 703.1.8^CenA'-'PhoA IX-8 heterodimers: association of heterodimers oncellulose^  723.1.9^Precise fusion of CBDPTCenA to alkaline phosphatase: constructionof pTZ18U-CBDPT-'phoA and pTUg10*-CBDPT-'phoA ^ 743.1.10^High level production of CBDPT-'PhoA^ 773.1.11^Purification of CBDPT-'PhoA: single column loading and elutionwith guanidine-HC1 or distilled water^ 813.1.12^Purification of CBDPT-'PhoA: multiple column loading studies^ 883.1.13^Cellulose binding of CBDPT-'PhoA at different bufferconcentrations^  943.1.14^Proteolytic cleavage of CBDPT-'PhoA^ 943.1.15^Fusion proteins designed for proteolysis: constructionof pTUg10*-CBDPT-phoA and pTUg10*-CBD-'phoA ^ 993.1.16 Purification of CBDPT-PhoA and CBD-'PhoA^ 1023.1.17 Proteolytic cleavage of CBDPT-PhoA and CBD-'PhoA^ 1023.1.18^Removal of affinity tag by cellulose adsorption^ 1083.2^CenA-IL2 fusion proteins^  1103.2.1^Construction of pUC18-cenA '-IL2^  1103.2.2^Detection of cenA'-1L2 expression in E. coli JM109^ 1103.2.3^cenA '-IL2 expression in different E. coli strains  1203.2.4^Cell fractionation studies^  1223.2.5^CenA'-IL2 purification: direct binding to Avicel^ 126v^3.2.6^CenA'-IL2 purification: alkaline solubilization ^1293.2.7^Elution of CenA'-IL2 from Avicel^  1333.2.8^Factor Xa digestion of CenA'-IL2  1383.2.9^Construction of pTZ18U-CBDPT-IL2^ 1383.2.10^Purification of CBDPT-IL2^  1423.2.11^Factor Xa digestion of CBDPT-IL2  1444. Discussion^  1464.1^General  1464.2 TnphoA mutagenesis^  1464.3^Heterodimer studies  1474.4^High expression and purification of alkaline phosphatase fusion proteins ^ 1494.5^Proteolytic cleavage of CenA-PhoA fusion proteins^ 1524.6^Expression of CenA-IL2 fusion proteins^  1554.7^Purification and proteolytic cleavage of CenA-IL2 fusion proteins^ 1574.8 The nature of CBD binding and elution^  1594.9^CBDs as affinity tags for protein puriification: how do they measure up? ^ 1615. Bibliography^  164viList of TablesTable^ Page1.1^Affinity tags for protein purification^  51.2^Specific cleavage methods for fusion proteins^ 152.1^List of E. coli strains^  262.2^List of plasmids, phage and transposons^  282.3^Operating conditions for CF1 cellulose affinity purifications with multiplecolumn loading^  453.1^Molecular mass determinations for CenA'-'PhoA IX-8 polypeptides^ 693.2^Purification of CBDPT-'PhoA from E. coli CC118/pTUg10*-CBDPT-'phoA culture supernatant: single column loading with guanidine-HC1elution^  833.3^Purification of CBDPT-'PhoA from E. coli CC118/pTUg10*-CBDPT-'phoA culture supernatant: single column loading with distilled water elution 873.4^Purification of CBDPT-'PhoA from E. coli CC118/pTUg10*-CBDPT-phoA culture supernatant: multiple column loading with distilled waterelution^  913.5 Purification of CBDPT-'PhoA from E. coli CC118/pTUg10*-CBDPT-'phoA clarified cell extract: multiple column loading with distilled waterelution  923.6^Purification of CBDPT-PhoA from E. coli CC118/pTUg10*-CBDPT-'phoA culture supernatant: multiple column loading with distilled waterelution^  1053.7^Purification of CBD-'PhoA from E. coli CC118/pTUg10*-CBDPT-'phoAculture supernatant: multiple column loading with distilled water elution^ 1063.8^CenA'-IL2 solubilization trials^  1313.9^Elution -of Cen =ILI^at A-ViceF   137viiList of FiguresFigure Page1.1 Affinity^tag^purification^ 21.2 Cellulose^structure 171.3 Domain structure of CenA and Cex^ 193.1 Construction of cenA'-'phoA gene fusions 553.2 Binding of Cen'-'PhoA fusion proteins to filter paper^ 583.3 Binding of CenA'-'PhoA fusion proteins to Avicel 593.4 Construction of pTZ18R-CBDPT-sphoAA1 and pTZ18R-CBDPT-'phoAA2 623.5 Binding of CBDPT-'PhoAA1 and CBDPT-'PhoAA2 fusion proteins toAvicel^ 653.6 Schematic diagram of fusion polypeptide CenA'-'PhoA IX-8^ 663.7 Binding of cenA'-'phoA IX-8 encoded polypeptides to Avicel^ 683.8 Dissociation of CenA'-'PhoA IX-8 heterodimers^ 713.9 Formation of CenA'-'PhoA heterodimers on Avicel 733.10 Construction of pTZ18U-CBDPT-'phoA^ 753.11 Construction of pTUg10*-CBDPT-'phoA 783.12 Expression of CBDPT-'phoA in E. coli CC118^ 803.13 Purification profile for CBDPT-'PhoA: single column loading withguanidine-HC1^elution^ 823.14 Purification of CBDPT-'PhoA:^SDS PAGE analysis^ 853.15 Purification profile for CBDPT-'PhoA: single column loading with distilledwater^elution^ 863.16 10-cycle purification of CBDPT- 1PhoA from culture supernatant:purification^yield 893.17 5-cycle purification of CBDFT-'PhoA from clarified cell extract:93Furth-cart-an^yield^viii3.18 Effect of loading buffer concentration on water elution yield of CBDPT-'PhoA^from^CF1^cellulose^ 953.19 Factor Xa and C. fimi protease digestion of CBDPT-'PhoA^ 973.20 Cleavage sites for Factor Xa and C. fimi protease in CBDPT-'PhoA,CBDPT-PhoA and CBD-'PhoA^ 983.21 Construction of pTUg10*-CBDPT-phoA^ 1003.22 Constuction of pTUg10*-CBD-'phoA 1033.23 Factor Xa digestion of CBDPT-PhoA^ 1073.24 C. fimi protease digestion of CBD-ThoA 1093.25 CBD removal by adsorption to Avicel^ 1113.26 Construction of pUC18-cenA'-IL2 1123.27 IL2 bioassay of osmotic shock fluids from E. coli JM109/pUC18-cenir-IL2^ 1153.28 IL2 bioassay of fractions from Avicel binding of E.coli JM109/pUC18-cenA '-IL2^clarified^extract^ 1163.29 Western blot of fractions from Avicel binding of E. coli JM109/pUC18-cenA'-1L2 clarified cell extract ^ 1173.30 Western blots of crude cell extracts from E. coli JM109/pUC18-cenA'-IL2grown at 30°^and 37°^ 1183.31 Growth curves for E. coli JM109/pUC18-cenA'-IL2 at 30° and 37°^ 1193.32 Expression of cen/V-IL2 in different strains of E. coli ^ 1213.33 Partial fractionation of E. coli BL21/pUC18-cenA'-1L2 crude extracts^ 1233.34 Membrane location of CenA'-IL2^ 1243.35 Direct binding of CenA'-IL2 to Avicel: effect of sonication^ 1273.36 Wash conditions for Avicel-bound CenA'-IL2^ 1283.37 CenA'-IL2 purification:^direct binding to Avicel 130 CenA'-IL2 purification:^alkaline solubilization^ 1343.38ix3.39 CenA'-IL2 binding to Avicel at pH 11.5^ 1353.40 Western blot of recombinant human IL2 incubated with Avicel at pH 7.0and^11.5 1363.41 Factor Xa digestion of CenA'-IL2^ 1393.42 Construction of pTZ18U-CBDPT-IL2 1403.43 Purification of CBDPT-IL2^ 1433.44 Factor Xa digestion of CBDPT-IL2 145xList of Abbreviationsamp^ Ampicillinkan KanamycinCenA'^ C. fimi endoglucanase A truncated at its carboxyl terminusQ10 Temperature coefficient (reaction velocity ratio for 10°temperature change)IMDM^ Iscove's Modified Dulbecco's mediumPEG Polyethylene glycolEDTA^ Ethylenedriamine tetra-cetic acidEGTA Ethyleneglycol-bis-(f3-aminoethyl ether) N, N, N', N'-tetra-acetic acidTE^ Tns EDTACMC Carboxymethyl celluloseDCPIP^ 2, 6-DichlorophenolindophenolSDH Succinate dehydrogenaseKDO^ 2-keto-3-deoxyoctonateELISA Enzyme-linked immunosorbent assayPBS^ Phosphate-buffered salineIPTG Isopropyl-0-D-thiogalactosideXP^ 5-bromo-4-chloro-3-indolyl phosphate p-toluidine saltLB Luria-Bertanikb^ Kilobase pairsSDS-PAGE^Sodium dodecyl sulfate-polyacrylamide gel electrophoresisPMSF PhenylthiohydantoinMT!'^ 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyl tetrazolium bromide-IL2 Th-Te-rle akin 2-xihIL2^ Human interleukin 2NK Natural killer cellCTLL-2^ Cytotoxic T Lymphocyte cell lineCBDPT Cellulose-binding domain and Pro-thr linker of CenAPhoA^ E. coli alkaline phosphatase'PhoA E. coli alkaline phosphatase truncated at its amino terminusTYP^ Tryptone, yeast extract, phosphate mediumrpm Revolutions per minuteIg^ ImmunoglobulinDEAE DiethylaminoethylMalE^ E. coli maltose-binding protein'MAC Immobilized metal ion affinity chromatographyxiiAcknowledgementsI was supported by a Canadian Commonwealth Scholarship during the course of thiswork and I extend my appreciation to the Canadian Commonwealth Scholarships andFellowships Association.In particular, I wish to thank my supervisor, Tony Warren, for his enthusiasm andongoing guidance and support. I also acknowledge the continued interest and ideas from DougKilburn, Bob Miller and Neil Gilkes.On the technical side I thank Emily Kwan for preparation of C. fimi protease, SandyKielland for timely N-terminal amino acid analysis, Soo-Jeet Teh for assistance with IL2assays, and the lab of Bob Hancock for help with KDO assays. For provision of bacterialstrains, plasmids and transposons, I acknowledge Carol Gross, Jon Beckwith, Vern Petkau,Don Trimbur, Roger Graham, Shen Hua, Neena Din, and Colin Manoil.The following people deserve special thanks:My parents, Mary and Michael, for always writing when I needed it most.Edgar, for useful discussions and for keeping me motivated during difficult times.The cellulase crowd, for providing a second-to-none working environment.Pat and Helen, for home away from home support.Bruce, for remaining green throughout.Zahra and Neena, for helping us out during a housing crisis.Lastly, I wish to thank my wife Lisa, for her love and friendship both in good times and bad; Idedicate this thesis to her.1. Introduction1.1 Affinity tags for protein purification.Protein purification has been and will continue to be a cornerstone of the biologicalsciences. Biospecific affinity chromatography has held an important place in the repertoire ofprotein purification techniques since its introduction proper by Campbell et al. (1951). Affinitychromatography is applicable to almost any biological molecule for which an affinity ligandexists. Coupling of the ligand to a solid support facilitates the specific adsorption of the moleculeto the support and hence its purification from a complex mixture. The expansion of molecularbiology has placed new demands on protein purification, particularly with regard to purificationof the products of cloned genes, and recovery of wild type and mutant polypeptides generated byrandom or specific mutagenesis procedures. In the latter case, the physical and chemicalcharacteristics of the mutant polypeptides may be quite different from the wild type, and a generalpurification procedure applicable to all the variants may be difficult to achieve. Molecular biologyhas also provided a solution to such concerns, however, in the form of gene fusion technology.Gene fusions encoding hybrid polypeptides have been widely used to increase in vivo proteolyticstability, to modulate solubility and to control cellular localization of target polypeptides in theexpression host (reviewed in Uhler' and Moks, 1990), as well as to facilitate the subsequentpurification of the target polypeptides. The term "affinity tag" has been coined to refer to apolypeptide fusion partner which can itself be affinity purified and which can facilitate the affinitypurification of the target protein (Figure 1.1). Normally the fusion is produced in a recombinanthost and recovered from the cells or culture medium in an impure form. The crude extract ispassed over the affinity matrix and the fusion protein is specifically adsorbed while contaminatingproteins pass through. The fusion protein is then desorbed in pure form by competition with freeligand or by nonspecific elution, such as a change in pH or ionic strength. If necessary, theaffinity tag is specifically removed by chemical or enzymatic cleavage, and separated from thetarget protein by a second passage over the affinity matrix. In theory this approach can achieve1B 4'Figure 1.1^Affinity tag purification.A: Specific binding of fusion protein from crude extract to affinity column; B: Elution offusion protein in pure form; C: Removal of affinity tag by specific chemical or enzymaticcleavage; D: Removal of detached affinity tag by readsorption to affinity column;E: Purified target protein .2high purification factors and improved yields in comparison to an equivalent series ofconventional chromatographic steps. The cleavage step can be designed to release the targetprotein with its authentic primary sequence, an important consideration when the target protein isintended for therapeutic use, or when detailed structural or functional analyses will be performed.The affinity tag approach does not demand intimate knowledge of the target protein and canpotentially be applied to a wide range of different target proteins, which is a clear advantage in thesituations mentioned above. Requirements of affinity tags include: i) small size, insofar as thisincreases the effective target protein yield and prevents interference with the biological activity ofthe target protein; ii) resistance to denaturation and internal proteolysis, both in vivo and duringthe specific proteolysis step;^versatility of cellular location, i.e. potential for intracellularexpression or direction of export from the cell. iv) specificity of the affinity interaction, tominimize non-specific binding to the matrix; v) strength and stability of the affinity interactionunder a variety of conditions, to allow efficient and thorough removal of contaminants; vi) mildconditions for desorption from the affinity matrix; and vii) facility for N- or C-terminal fusion tothe target protein.The use of affinity tags for protein purification dates back to 1983, when Uhlen et al.(1983) described vectors for heterologous gene fusion with the staphylococcal protein A gene,and demonstrated purification of a protein A-P-galactosidase fusion protein by IgG affinitychromatography. Germino et al. (1983) first described the use of P-galactosidase as an affinitytag for protein purification. They followed a method for P-galactosidase purification developedby Steers et al. (1971), which utilized the substrate analogue inhibitor, p-aminophenyl-P-D-thiogalactoside, as an affinity ligand. Ullman (1984) showed the utility of the approach bypurifying a number of P-galactosidase fusion proteins. Germino and Bastia (1984) used the sameapproach, but engineered a linker peptide between the target protein and P-galactosidase whichwas susceptible to cleavage by collagenase. This allowed the removal of the affinity tag byproteolytic cleavage. At much the same time, the first use of a polyamino acid affinity tag wasreported by Sassenfeld (1984), who fused five arginine residues to the C-terminus of humanurogastrone. After purification of the polypeptide by cation exchange chromatography, the3polyarginine tail was removed by carboxypeptidase B digestion, and the modified behaviour ofthe polypeptide during a second cation exchange chromatographic step facilitated its furtherpurification. Nagai and Thogersen (1984) made use of the blood coagulation protease, FactorXa, for site specific cleavage of fusion proteins containing the tetrapeptide Ile-Glu-Gly-Arg.These studies, along with the chemical cleavage of fusion proteins reported earlier by Itakura etal. (1977) and Goeddel et al. (1979), and the use of trypsin to separate fused polypeptides (Shineet al., 1980), heralded the introduction of affinity tag purification technology. The developmentand use of affinity tags have grown rapidly and have been the subject of a number of reviews(Sassenfeld, 1990; Uhler' and Moks, 1990; Sherwood, 1991; Arnold, 1991; Nilsson et al.,1992), including a comprehensive account by Ford et al. (1991). To date, over thirty affinity tagshave been developed for purification of fusion proteins (Table 1.1). These vary from wholeproteins and protein domains to poly- or single amino acid residues.Staphylococcal protein A has seen widespread use as an affinity tag for proteinpurification. It demonstrates affinity for the Fc portion of mammalian IgG molecules andcontains five distinct IgG-binding domains of about 58 amino acids each (Moks et al., 1986).Based on these IgG binding domains a two domain variant, designated ZZ, has been engineeredas an affinity tag for protein purification on IgG-Sepharose (Nilsson et al., 1987). Sequences inthe IgG-binding domains sensitive to cleavage by cyanogen bromide and hydroxylamine werealtered in ZZ, to allow use of these agents to release the affinity tag from the target protein.Protein A fusions can be exported to the periplasm of Escherichia coli (Nilsson et al., 1985), andhave been shown to leak into the culture medium (Abrahmsen et al., 1986). Desorption fromIgG-Sepharose occurs at low pH, a potential disadvantage for purification of acid sensitivepolypeptides. This situation also applies to the albumin-binding domains from streptococcalprotein G (Nygren et al., 1990), and to a number of antibody-antigen affinity tag interactions.However, the FLAGTm system is a notable exception, as the antibody-FLAG peptide interactionis calcium-dependent and can be easily reversed with low concentrations of EDTA (Hopp et al.,1988).4cyclomaltodextringlucanOtransferasemonoclonalantibody FabMaltos-bindingprotein75 kDa a-cyclodextrin-Sepharosehapten40 kDa cross-linkedamyloseTable 1.1^Affinity tags for protein purification.Affinity tag^Size^Affinity^Elution conditions^ Referencematrix/ligandHanada et al., 1988; Levens and Howley, 1985; Hickman et al., 1990Germino et al., 1983; Ullman, 1984; Germino and Bastia, 1984;Offensperger et al., 1985; Phalipon and Kaczorek, 1987; Hirel et al.,1988; Scholtissek and Grosse, 1988; Grewal et al., 1989; Sieg et al.,1989; Scherf et al., 1990; Markmeyer et al., 1990; Hickman et al., 199013-Gal4ctosidase^116 kDa polyclonalantibodyP-Gal*tosidase^116 kDa p-aminophenyl-P- pH 10D-galactoside10 mM a-cyclo- Hellman and Mantsala, 1992dextrin0.5-1 mM hapten Neuberger et al., 1984; Williams and Neuberger, 198610 mM maltose^Bedouelle and Duplay, 1988; Di Guan et al., 1988; Maina et al., 1988;Blondel and Bedouelle, 1990; Clement et al., 1991; Lauritzen et al.,1991; Louis et al., 1991; Davie et al., 1992; Derbyshire and Grindley,1992; Li et al., 1992; Masuta et al., 1992; Wynn et al., 1992Protein A 31 kDa IgG-Sepharose pH 2.8-3.0 or 10mg/ml protein AUhlen et al., 1983; Nilsson et al., 1985 and 1985a; LOwenadler et al.,1986; Moks et al., 1987; Monaco et al., 1987; Dahlman et al., 1989;Lundeberg et al., 1990; Lydon et al., 1990; Stahl et al., 1990; Stringeret al., 1990; Sano and Cantor, 1991; Stirling et al., 1992continued...tnProteit Galbumin-bindingdomaiiisTrpETrpE28 kDa human serumalbumin-Sepharose27 kDa C8 reverse phase27 kDa monoclonalantibodypH 2.8^Jansson et al., 1989; Hammarberg et al., 1989; Stahl et al., 1989;Nygren et al., 1990; Murby et al., 1991a60% formic acid, Hummel et al., 198920% 2-propanolpH 2.8^Nurse et al., 1991Table 1.1^continuedAffinity tag^Size^Affinity^Elution conditions^ Referencematrix/ligandKolmar et al., 1992Wolber et al., 1992Smith and Johnson, 1988; Johnson et al., 1989; Gearing et al., 1989;Koland et al., 1990; Fikrig et al., 1990; Brissette et al., 1991; Lin andGreen, 1991; Guan and Dixon, 1991; Sankar and Porter, 1991; Baksh etal., 1992; Baratte et a1., 1992; Derbyshire and Grindley, 1992; Donaldsonand Capone, 1992; Frorath et al., 1992; Hakes and Dixon, 1992;Hartman et al., 1992; Haun and Moss, 1992; Knudsen et al., 1992;Menendez-Arias et al., 1992; Olsen and Mohapatra, 1992; Rahman et al.,1992; Roehrborn et al., 1992; Thomas et al., 1992continued...P-lactamase^30 kDa phenylboronate- 0.5 M sodiumSepharose^boratelight nieromyosin 30 kDa 20 mM KC1^0.6 M KClprecipitation^solubilizationGlutatli ione S-^26 kDa glutathione-^5-10 mM reducedtransfeluse agarose^glutathioneChloramphenicolacetyltransferaseCholine-bindingdomainsDihydrofolatereductaseIManmise-bindingproteinRecA antigenicepitopeGalactc?se-bindingdomainProtein's A doubleIgG-bindingdomains24 kDa chloramphenicol-Sepharose16 kDa DEAE-cellulose19 kDa18 kDa methotrexate-agarose17 kDa mannose-Sepharose16 kDa monoclonalantibody15 kDa galactose-Sepharose14 kDa IgG-Sepharose5 mMchloramphenicolpH 7.8140 mM choline3 mM folate, pH9.52.5 mM EDTApH 2.52 mM EDTA or0.2 M galactosepH 2.8-3.3Tablei 1.1^continuedAffinity tag^Size^Affinity^Elution conditions^ Referencematrix/ligandKnott et al., 1988; Dykes et al., 1988Sanchez-Puelles et al., 1992Iwakura et al., 1992 and 1992aTaylor and Drickamer, 1992Krivi et al., 1985Taylor and Drickamer, 1991Abrahmsen et al., 1986; Moks et al., 1987 and 1987a; Hammarberg etal., 1989 and 1991; Stahl et al., 1989; Forsberg et al., 1990, 1991 and1992; Boutelje et al., 1990; Murby et al., 1991 and 1991a; Samuelssonet al., 1991; Nilsson et al., 1991; Sano and Cantor, 1991; Wadensten etal., 1991; Waldenstrom et al., 1991continued...T^le 11.1^continuedA inity tag Size Affinitymatrix/ligandElution conditions ReferenceCellulose-bindingdomain14 kDa cellulose distilled water or6 M guanidine-Greenwood et al., 1989 and 1992; Ong et al., 1989, 1989a, 1991 and1993; Kilburn et al., 1992HC1Starch-bindingdomain13 kDa starch granules boric acid-boraxpH 8.2Chen et al., 1991 and 1991aStrepttvidin 13 kDa biotin 6 M urea pH 4.0 Sano and Cantor, 1991Tight llysine-bindindomainkringle 182 aa Sepharose-lysine 200 mM c-amino-caproic acidDeSerrano et al., 1992Biotintion 75 aa monomeric avidin 20 mM biotin Cronan, 1990sequenoeCalmoiulin-binding peptide36 aa calmodulin-agarose2 mM EGTA Carr et al., 1991; Stofko-Hahn et al., 1992YPYD1VPDYA... 20 aa monoclonal free peptide, 30° Field et al., 1988antigenic epitope antibodypeptideX di N4erminal 13 aa monoclonalantibodyZweig et al., 1987continued...00Table 11.1^continuedAffinity tag^Size^Affmity^Elution conditions^ Referencematrix/ligandHopp et al., 1988 and 1988a; Prickett et al., 1989; Brewer et al., 1991;Sassenfeld et al., 1991; Wels et al., 1992; Power et al., 1992Hochuli et al., 1988; Gentz et al., 1988 and 1989; Kilgus et al., 1989;Nardin et al., 1989; Abate et al., 1990; Stiiber et al., 1990 and 1990a;LeGrice and Griininger-Leitch, 1990; Sassenfeld et al., 1991; Caspers etal., 1991; Traunecker et al., 1991; Hoffman and Roeder, 1991; Skerraet al., 1991; Basu et al., 1991; Bush et al., 1991; Janknecht et al., 1991;Berthold et al., 1992; Boon et al., 1992; Janknecht and Nordheim,1992; Lotter et al., 1992Smith et al., 1987 and 1988; Ljungquist et al., 1989; Jansson et al.,1989; Evans et al., 1991; Sharma et al., 1991; Suh et al., 1991; Lilius etal., 1991; Brewer et al., 1991; Chattopadhyay et al., 1992; Dudler et al.,1992; Vosters et al., 1992; van Dyke et al., 1992Lilius et al., 1991Sassenfeld, 1984; Smith et al., 1984; Brewer and Sassenfeld, 1985;Brewer et al., 1991Persson et al., 1988FLAdrm peptide 8 aa^monoclonal^pH 3.0 or 2-5(DYKPDDDK)^antibody^mM EDTAPoly(His)^2-10 aa Ni2+-NTAa^low pH or 250mM imidazoleHis-containing^2-48 aa Cu2+an2-FiNi2+_ low pH or S 300peptides IDAb^mM imidazolePoly(His)^5 aa^EGTA(Zn)2^10 mM EDTAprecipitation^solubilizationPoly(Arg)^12 aa cation exchange 0.8-2.2 M NaC1Poly(Phe)^11 aa^phenyl-Superose 4.5 M ethylene1 glycolcontinued...Table 1.1^continuedAffinity tag^Size^Affinity^Elution conditions^ Referencematrix/ligandPoly(Asp)^5-11 aa polyelectrolyte^Parker et al., 1990precipitationPoly(ys)^4 aa^thiopropyl-^10 mM DTT^Persson et al., 1988SepharoseCys^1 aa^thiopropyl-^20 mM DTT^Carter and Wells, 1987; Persson et al., 1990SepharoseArginiie (C-^laa^anhydrotrypsin^10 mM^Hirabayashi and Kasai, 1992terminal) agarose^benzoylglycyl-arginine or 5 mMHClPartitiolning^4 aa^PEG phase^KOhler et al., 1991peptide AWWPa nitri1otriacetic acidb iminodiacetic acidc aqucious two phase extraction.0Glutathione S-transferase from Schistosoma japonicum was originally used as an affinitytag for purification of polypeptide antigens (Smith and Johnson, 1988), but has since been usedextensively for purification of a number of target proteins. Glutathione S-transferase fusions arepurified on immobilized glutathione and specifically desorbed with reduced glutathione. Manyglutathione S-transferase fusions are produced in the E. coli cytoplasm in a soluble form, whichaids subsequent purification. Efforts have also been made to apply the technique to fusionproteins produced in insoluble inclusion bodies (Hartman et al., 1992). Recent development ofvectors for expression of glutathione S-transferase gene fusions has focused on optimization ofthrombin cleavage of the encoded fusion polypeptides (Guan and Dixon, 1991; Hakes and Dixon,1992).Different approaches have been taken to develop purification techniques for existing E.coli gene fusion expression systems, such as those based on trpE (Yansura, 1990) and X dI(Nagai and Thogersen, 1987). Monoclonal antibodies directed against TrpE (Nurse et al., 1991)and the N-terminal region of the X cII protein (Zweig et al., 1987) have been used to facilitateaffinity purification of fusion proteins containing these polypeptides. In addition, thehydrophobic nature of TrpE fusion proteins has been exploited in a reverse-phase HPLCpurification procedure (Hummel et al., 1989).Antibody-antigen interactions can be readily applied in affinity tag purification. Theaffinity tag can be an intact protein (Hanada et al., 1988), an antigenic protein fragment, such as a12-amino acid antigenic determinant from influenza virus hemagglutinin (Field et al., 1988), or asynthetic peptide (Hopp et al., 1988a). The affinity tag/ligand order can be reversed and theantibody used as an affinity tag, as has been demonstrated by Neuberger et al. (1984).Polyamino acid affinity tags, particularly polyhistidine, have become very popularbecause they are small and in many cases do not compromise the enzymatic or biological activityof the fusion partner. Histidine residues in proteins are capable of interacting with immobilizedtransition metal ions to form coordination complexes, a property which forms the basis ofimmobilised metal ion affinity chromatography or IMAC (Porath et al., 1975). IMAC is anestablished procedure for protein fractionation and purification, and can be applied to purification11of polypeptides with engineered metal ion affinity (Smith et al., 1987; Hochuli et al., 1988).Usually this involves genetic fusion of a polyhistidine or histidine-containing peptide to the N- orC-terminus of the target protein, but there are examples of engineering chelation sites within thetarget protein itself (Suh et al., 1991; Brewer et al., 1991). The strength of the interactiondepends on the number and orientation of the histidine residues, the type of metal and the natureof the immobilized metal complex (Hochuli et al., 1987; Hochuli et al., 1988; Sulkowski, 1989;Arnold, 1991). Desorption is facilitated by reducing the pH or by competition with bufferscontaining imidazole. An advantage of polyhistidine affinity tags is the stability of binding in thepresence of protein denaturants (Hochuli et al., 1988). Polycysteine affinity tags have been usedfor protein purification through reversible disulfide formation with thiopropyl-Sepharose(Persson et al., 1988). A single free cysteine residue is sufficient for such an interaction to occur,provided that it is exposed to the solvent. Such single amino acid changes have been made withinthe primary sequences of proteins, such as subtilisin and glucose dehydrogenase, to facilitate theirpurification (Carter and Wells, 1987; Persson et al., 1990).The binding interactions of antibiotics and antibiotic resistance gene products have led tothe development of a number of affinity tag approaches. Chloramphenicol acetyltransferase and13-lactamase fusions have been used for affinity purification of target polypeptides on immobilizedchloramphenicol and the reversible P-lactamase inhibitor, phenylboronate, respectively (Dykes etal., 1988; Knott et al., 1988; Kolmar et al., 1992). Fusion proteins containing dihydrofolatereductase can be purified on immobilised methotrexate and desorbed by competition with folicacid (Iwakura et al., 1992 and 1992a).The strong binding interaction of avidin and biotin has been exploited (Cronan, 1990) byusing an amino acid sequence that is biotinated in vivo as an affinity tag. Immobilized monomericavidin has a lower affinity for biotin than the native tetrameric form, and desorption from thisaffinity matrix can be effected with buffer containing 20 mM biotin.More recent developments in affinity tag purification have included use of the tight lysine-binding kringle 1 domain of human plasminogen (DeSerrano et al., 1992). Although thispolypeptide was specifically used for purification on Sepharose-lysine of a related domain lacking12affinity for lysine, the authors emphasized the potential utility of the approach. Hirabayashi andKasai (1992) reported the purification of polypeptides containing a C-terminal arginine residue byadsorption to anhydrotrypsin, an inactive trypsin derivative with binding specificity directed atproduct-type compounds. The use of a calmodulin-binding peptide affinity tag for purification oncalmodulin-agarose has been reported (Stofko-Hahn et al., 1992). Calmodulin binding iscalcium-dependent and desorption is effected by calcium chelation with EGTA. Affinity tagpurification strategies have traditionally involved chromatographic procedures, but recentdiversifications include the use of affinity tags to direct specific protein precipitation (Parker et al.,1990; Lilius et al., 1991; Wolber et al., 1992) and partitioning in aqueous two-phase systems(Kohler et al., 1991).A number of carbohydrate-binding proteins and domains have been used as affinity tags,the most widely used of which is the E. coli maltose-binding protein, MalE. MalE fusions forpurification were first reported almost simultaneously by Bedouelle and Duplay (1988) and diGuan et al. (1988). N-terminal MalE fusions were prepared, and both cytoplasmic expressionand periplasmic export of the fusion polypeptides were demonstrated in E. coli, in the latter caseunder the direction of the MalE or alkaline phosphatase leader peptides. Fusion polypeptideswere purified on cross-linked amylose and desorbed with 10 mM maltose. Although MalE is oneof the larger affinity tags and therefore limits to some extent the level of target protein production,its advantages lie in the versatility of cellular targetting and in particular the mild conditions ofbinding to and elution from the affinity matrix. A cyclomaltodextrin glucanotransferase affinitytag from Bacillus circulans var. alkalophilus also exhibits reversible binding to its affinity matrix,a-cyclodextrin-Sepharose, under mild conditions (Hellman and Maritsala, 1992). Galactose- andmannose-specific carbohydrate-recognition domains from animal lectins have recently been usedfor affinity purification of fusion proteins, and have the advantage of calcium-dependent bindingto their respective ligands (Taylor and Drickamer, 1991 and 1992). This allows the use ofchelating agents as well as competition with monosaccharides to effect the desorption of thedomains from their affinity matrices. In addition, these carbohydrate recognition domains can beproduced in prokaryotic and eukaryotic hosts; expression of gene fusions in E. coli, mammalian13cells, and a baculovirus expression system have been reported. The starch binding domain fromAspergillus glucoamylase has been used to bind p-galactosidase to native starch granules, but theadsorption of the fusion protein to starch was not readily reversible (Chen et al., 1991). Deletionstudies revealed a 119 amino acid starch-binding polypeptide which, when used as an affinity tag,exhibited higher binding affinity and higher resistance to proteolysis than the original 133 aminoacid polypeptide (Chen et al., 1991a). The use of cellulose-binding domains of cellulases asaffinity tags for protein purification has been reported by this laboratory (Greenwood et al., 1989and 1992; Ong et al., 1989, 1989a and 1993), and is the subject of this thesis.1.2 Proteolytic cleavage of fusion proteins.A requirement of affinity tag purification technology is the facility of affinity tag removalthrough proteolytic cleavage. This is not always necessary, particularly when the affinity tag issmall and does not interfere with the catalytic or biological activity of the target protein, or whenthe target protein is used for the production of antibodies. In the latter case, the presence of theaffinity tag can sometimes enhance the immune response against the target protein (Lowenadler etal., 1986), particularly if it is a small peptide antigen. However, in many cases, removal of theaffinity tag is necessary, and this generally calls for proteolytic cleavage at a precise locationbetween the affinity tag and its fusion partner. No single cleavage method can be applied for allfusion proteins, and often the optimization of affinity tag removal will be a systematic andinteractive process (Forsberg et al., 1992).Site-specific cleavage of fusion proteins is achieved by either chemical or enzymaticmeans, for which some excellent reviews have been published (Bornstein and Balian, 1977;Landon, 1977; Fontana and Gross, 1986; Carter, 1990). Chemical methods have the advantageof high efficiency and scalability, but are relatively non-specific and require harsh reactionconditions which can lead to protein denaturation. In addition, the treated protein is often subjectto a certain degree of amino acid side chain modification during chemical cleavage procedures.Enzymatic cleavage procedures, on the other hand, exhibit greater specificity and can proceed14Method^Cleavage Position^ReferenceChemicalCNBrHydroxylamineAcidBNPS-skatoleo-iodosobenzoic acidN-chlorosuccinimideEnzymatic TrypsinChymotrypsinFactor XaThrombinEnterokinaseReninCollagenaseKallikreinClostripainEndoproteinase Arg-CEndoproteinase Lys-CH64A Subtilisin BPN'Ubiquitin protein peptidaseNeisseria type 2 IgA proteaseCathepsin CKEX2PlasminUrokinaseV8 proteaseCarboxypeptidase BCarboxypeptidase AAminopeptidase MAminopeptidase P-RI or -K1'-WI , -Y1 or -F1'-I-E-G-R 1'-L-V-P-R1'-D-D-D-K-1-A-P--P-F-H-LIL-V-Y--P-XIG-P--R-F-0'-RI-RI-K1'-A-A-H-YI-Ubiquitinl-P-13-1•X-P-, X=T, S, or AXi-X2 1A3-, Xi#R or K;X2,X3#P-K-R1' or -R-R1'G-A-R1P-G-R4'-DI or -El-X1R or -XII(-17-1-X, X=aromatic;X#R. or KX1Y-, Y#13X113-Itakura et al., 1977Moks et al., 1987aSzoka et al., 1986Dykes et al., 1988Villa et al., 1988Nilsson et al., 1991Shine et al., 1980Dahlman et al., 1989Nagai and Thogersen, 1987Smith and Johnson, 1988Hopp et al., 1988Haffey et al., 1987Germino and Bastia, 1984Tonouchi et al., 1988Lowe et al., 1987Taylor and Drickamer, 1992Allen and Henwood, 1987Forsberg et al., 1992Sabin et al., 1989Pohlner et al., 1992Hsiung and MacKellar, 1987Thomas et al., 1988Forsberg et al., 1992Forsberg et al., 1992Gentz et al., 1989Sassenfeld, 1984Hochuli et al., 1988Nakagawa et al, 1987Tonouchi et al., 1988Table 1.2^Specific cleavage methods for fusion proteins.15under physiological conditions, but are often inefficient and expensive to scale up. Furthermore,contamination of pure preparations with other protease activities can lead to non-specific cleavage.Some proteases, such as Factor X a, have specificity determinants only on the N-terminal side ofthe scissile bond, an obvious advantage when releasing a C-terminally-fused target protein withits correct N-terminus. A list of different methods of fusion protein cleavage is given in Table1.2, together with cited examples. For further information and references the reader is referred tothe above reviews.1.3 Cellulose.Cellulose is the most abundant organic polymer in nature, being the major component ofthe cell walls of plants. It is eminently suitable as a structural material, as in the native state itforms compact crystalline microfibrils of associated cellulose molecules (B6guin et al., 1987).These crystalline regions in cellulose microfibrils are normally interspersed with paracrystalline oramorphous regions, and in plant cell walls the microfibrils are embedded in a matrix ofhemicellulose and, in woody tissues, lignin (Preston, 1986). The cellulose molecule is a linearpolymer of up to 10,000 13-1,4-linked glucose residues, in which successive residues are rotatedby 180° relative to each other (Figure 1.1A). In native crystalline cellulose (cellulose I) the chainsare oriented in a parallel fashion and are stabilized by intra- and inter-chain hydrogen bonds(Figure 1.1D; Blackwell, 1982; Kremer and Tabb, 1990). The hydrogen bond network in nativecellulose is entirely in one plane; hydrogen-bonded sheets of cellulose chains are held together byvan der Waals contacts between the faces of the sugar rings (Figure 1.1B-D; Blackwell, 1982).The interior of the crystalline matrix is hydrophobic, in contrast to the hydrophilic exterior.The physical nature of cellulose is more complex than its native crystal structure wouldsuggest. Cellulose from different sources exhibits variable degrees of crystallinity and differentphysical characteristics (Young, 1986; Kulshreshtha and Dweltz, 1973). Of the two kinds ofcellulose used in this study, CF1 cellulose is a fibrous cellulose derived from cotton. It has anaverage fibre length of 200 gm and diameter of 13.5 gm (E. Heilweil, Whatman Inc., personal16BC^DFigure 1.2 Cellulose structure.A: Chemical structure of cellulose; B: Structure of cellulose I, looking along the chainaAcs (ulb projection); . Structure of cellulose I, ac projection; D Cellulose 1, hydrogen bondnetwork in the sheet parallel to the ac plane. Adapted from Blackwell (1982).17communication). Cotton is a highly crystalline form of cellulose (Wood, 1988) and CF1cellulose is estimated to be 75-85% crystalline. Avicel, a microcrystalline cellulose, was alsoused in this study. Avicel is a product of wood cellulose which has undergone partial acidhydrolysis to a levelling off degree of polymerisation of 100-250. Following hydrolysis thecellulose is neutralised, washed, and spray dried (Wood, 1988; Marshall and Sixsmith, 1974).The resulting material is a mixture of rod-shaped and irregular particles, and exhibitsinterparticulate voids (-20 pm) and intraparticulate pores (-2 nm; Marshall and Sixsmith, 1974).Reports on the crystallinity of Avicel vary, but an average value of 50% has been quoted (Gilkeset al., 1992). Our laboratory uses CF1 cellulose for column affinity chromatography, as it hasgood packing and flowrate characteristics. Avicel, on the other hand, performs poorly incolumns but has a higher adsorption capacity (Ong et al., 1993a) and a lower wet specificvolume, making it more suitable for small scale binding experiments and batch affinitypurification (Coutinho et al., 1992).1.4 Cellulose-binding domains.The insolubility and inert nature of cellulose provide a unique challenge formicroorganisms that catalyze its hydrolysis to simple sugars. Such organisms usually produce abattery of different enzymes, called cellulases, which act synergistically to hydrolyze the insolublesubstrate. A number of cellulases exhibit affinity for cellulose in addition to catalytic potential,and this property has frequently been used to facilitate their purification (Beguin and Eisen, 1978;Halliwell and Griffin, 1978; Nummi et al., 1981; Mart'yanov et al., 1984; Boyer, 1987; Au andChan, 1987; Owolabi et al., 1988; Gilkes et al., 1988; Ghangas and Wilson, 1988; Gilbert et al.,1990; Coutinho, 1992a; Meinke et al., 1993).Cellulomonas fimi, a gram positive mesophilic coryneform bacterium, produces a numberof cellulases, the genes for five of which have been cloned, sequenced, and expressed in E. coli,and the gene products characterized (Gilkes et al., 1984; Wong et al., 1986; O'Neill et al., 1986;Warren et al., 1986; Moser et al., 1989; Coutinho et al., 1991; Meinke et al., 1991 and 1993).18The first genes to be cloned and sequenced encode an endoglucanase, CenA, and anexoglucanase/xylanase, Cex (Wong et al., 1986; O'Neill et al., 1986), both of which bindstrongly to cellulose. CenA and Cex have distinct catalytic and cellulose-binding domains, eachseparated by short linkers consisting of 20-23 alternating proline and threonine residues, termedPro-Thr linkers (Figure 1.3; Gilkes et al., 1988). The Pro-Thr linker in CenA appears tomaintain the separation and orientation of the cellulose-binding domain and catalytic domainnecessary for normal enzyme function (Shen et al., 1991); however, this property is not generallyapplicable to linker peptides in other13-1,4-glycanases (Ferreira et al., 1990). The cellulose-binding domains of CenA and Cex share 50% amino acid homology and are positioned at the N-terminus of CenA and the C-terminus of Cex; these orientations presumably arose by shuffling ofconserved domains (Warren et al., 1986). CenA and Cex are exported from C. fimi and haveleader peptides which are functional in E. coli (Wong et al., 1986; O'Neill et al., 1986a; Guo etal., 1988).CenA1^111 134^418PTA1, Catalytic DomainCex1^315 335^443Catalytic Domain^PTFigure 1.3^Domain structure of CenA and Cex.Mature CenA and Cex are shown, with the cellulose-binding domain (CBD, shaded),Pro-Thr linker (PT), and catalytic domain indicated. Numbers refer to amino acid residues at theN-terminus of each polypeptide and the C-terminus of each domain structure. The position ofprimary cleavage of each polypeptide by C.fimi protease is marked  by an arrow.19Cellulose-binding domains (CBDs) were first identified as conserved regions of cellulaseenzymes which retained cellulose-binding activity when separated from their cognate catalyticdomains by limited proteolysis (van Tilbeurgh et al., 1986; Langsford et al., 1987; Tomme et al.,1988; Gilkes et al., 1988). In the case of recombinant CenA and Cex, a serine protease isolatedfrom C. fimi culture supernatants processes both enzymes at exactly analogous positions at the C-termini of the Pro-Thr linkers (Figure 1.3), releasing CBD fragments (CBDCenA, CBDCex)which bind to cellulose (Gilkes et al., 1988). In addition, subcloning and expression of thecoding regions of CBDCenA and CBDCex yields polypeptides which also bind to cellulose (Din etal., 1991; Gilkes et al., 1992; Ong et al., 1993a). DNA sequencing of cloned (3-1,4-glycanasegenes has permitted the grouping of CBDs of both bacterial and fungal origin into families on thebasis of similarities in amino acid sequence (Coutinho et al., 1992; Gilkes et al., 1991). FamilyII, of which CBDcenia, and CBDCex are members, comprises polypeptides of about 100 aminoacids in length which feature i) low contents of charged amino acids; ii) high contents ofhydroxyamino acids; and iii) conserved tryptophan, asparagine, glycine and cysteine residues.Many of the CBDs have been shown experimentally to bind to cellulose (Coutinho et al., 1992and references therein), but the nature of the binding interaction is not well understood. InCBDCenA and CBDCex (and by analogy other bacterial CBD family members) the conservedcysteine residues form a disulfide bond (Gilkes et al., 1991a). The conserved tryptophanresidues in bacterial CBDs have been implicated in binding interactions with cellulose, astryptophans are also conserved within other groups of polysaccharide-binding proteins(Drickamer, 1988; Svensson et al., 1989) and have been shown in a number of cases toparticipate in protein-carbohydrate interactions (Quiocho, 1986; Johnson et al., 1988; Martineauet al., 1990). More recently the importance of tryptophan residues in cellulose-binding of CBDshas been shown by site-directed mutagenesis approaches (Poole et al., 1993; N. Din, submittedfor publication). At present no detailed 3-dimensional structures of bacterial CBDs have beenreported, but small-angle x-ray scattering analyses of CenA and Cex from C. fimi indicate that theenzymes have a tadpole-like structure, with the CBD, in CenA at least, forming an extended tail region (Pilz et al., 1990; Shen et al., 1991; N.R. Gilkes, unpublished results). Similar studies on20C. fimi endoglucanase B (CenB), which contains a C-terminal CBD, also indicate the presence ofan extended tail region (Meinke et al., 1992). The structure of a fungal CBD from Trichodermareesei cellobiohydrolase I (CBDcbta) has been determined by NMR spectroscopy of thesynthetically produced polypeptide (Kraulis et al., 1989; Reinikainen et al., 1992). CBDcbhicomprises 33 amino acids and contains two disulfide bonds. It is a wedge-shaped molecule withthree conserved tyrosine residues positioned along one face, in an orientation that would allowtheir stacking against the six-membered rings of the repeating cellobiose units in cellulose.Cellulose-binding domains can target their cognate catalytic domains to the insolublesubstrate to establish a high local concentration, but the exact role of CBDs in cellulose hydrolysisis unclear. The binding phenomenon has been associated with the ability of cellulases tohydrolyze crystalline cellulose (Klyosov, 1990). Some studies on proteolytic removal andgenetic addition of CBDs have shown that the CBD modulates the activity of the catalytic domaintowards insoluble substrates (Gilkes et al., 1988; Maglione et al., 1992), while other studiesreport no such effect (Poole et al., 1991). There is also evidence that CBDs cause physicaldisruption of cellulose fibres, an effect that may facilitate subsequent hydrolysis (Din et al., 1991;Nordquist, 1992).1.5 Alkaline phosphatase.Alkaline phosphatase of E. coli (EC is a dimeric phosphate monoester hydrolasewith a subunit molecular mass of 47 kDa (Coleman and Gettins, 1983). Alkaline phosphatase isencoded by the phoA gene and is exported to the periplasmic space of E. coli under the directionof a signal peptide (Inouye and Beckwith, 1977; Michaelis et al., 1983). The native enzyme has apH optimum of 8.0 (Garen and Levinthal, 1960), and coordinates two zinc ions and onemagnesium ion per subunit, the zinc ions being required for enzymatic activity (Sowadski et al.,1985; Reid and Wilson, 1971). Alkaline phosphatase is resistant to chemical and thermaldenaturation, but can be reversibly dissociated into subunits by heating at 95° for 15 minutes(Schlesinger and Barrett, 1965), reducing the pH below 3 (Schlesinger and Levinthal, 1963), by21thiol reduction in the presence of 6 M urea (Levinthal et al., 1962), or by treatment with 6 Mguanidine-HCI (Schlesinger and Barrett, 1965). Methods for purification of alkaline phosphatasehave been reviewed extensively (Torriani, 1966; Reid and Wilson, 1971; McComb et al., 1979).Periplasmic proteins are typically released from harvested cells by osmotic shock (Neu andHeppel, 1965), lysozyme (Malamy and Horecker, 1964a), or low pH (Schlesinger and Olsen,1970). Purification is usually based on ion exchange chromatography on DEAE-cellulose or-Sepharose (Garen and Levinthal, 1960; Schlesinger and Olsen, 1970; Harris and Coleman,1968; McCracken and Meighen, 1980) and other steps such as heating (Garen and Levinthal,1960), precipitation (Applebury and Coleman, 1969), crystallization (Malamy and Horecker,1964), and size exclusion chromatography (Csopak et al., 1972a).The N-terminus of alkaline phosphatase is relatively insensitive to the addition or removalof amino acids (Hoffman and Wright, 1985), a fact which has contributed to its widespread useas an enzyme marker for gene expression and protein export in E. coli (Manoil et al., 1990).Manoil and Beckwith (1985) developed a transposon system for creating protein fusions betweenalkaline phosphatase and the N-terminal sequences of other proteins. TnphoA is a derivative oftransposon Tn5 which contains a phoA gene lacking its promoter, its translation initiation site,and the DNA encoding the leader peptide and the first five amino acids of the mature polypeptide.Transposition into a gene in the correct orientation and reading frame generates a gene fusionencoding a hybrid polypeptide. As alkaline phosphatase is only active when exported to theperiplasm (Michaelis et al., 1983; Hoffman and Wright, 1985) fusion to an exported polypeptidecan be screened for by the presence of alkaline phosphatase activity in a negative background.Tn5 is reported to demonstrate low insertional specificity (De Bruijn and Lupski, 1984), so theuse of TnphoA can be expected to yield a range of transposon insertions throughout the targetgene. As CBDCenA is located at the N-terminus of CenA, this approach was used in this study toproduce CenA-PhoA fusion proteins and to probe the C-terminal extent of the cellulose-bindingdomain in CenA.221.6 Human interleukin 2.Interleukin 2 (IL2) is a 15 kDa protein released by T lymphocytes upon antigenstimulation, and is a key component in the amplification of the immune response (Smith, 1988).1L2 stimulates the proliferation and differentiation of lymphoid cells, including natural killer (NK)cells and antigen-stimulated B and T cells (reviewed in Kaplan et al., 1992); the stimulatorysignals occur through binding to specific cell-surface 1L2 receptors (Smith, 1988). Because of itspivotal role in the immune response, IL2 has potential for a number of therapeutic uses (Smith,1988; Kaplan et al., 1992; Smith, 1992), including treatment of cancer (Stein et al., 1991;Caligiuri et al., 1991), use as an adjuvant to vaccination (Ramshaw et al., 1987; Nunberg et al.,1989), and augmentation of the immune response in intracellular infectious diseases (Kaplan etal., 1991), In addition, linkage of 1L2 to bacterial toxins such as diphtheria toxin (Williams et al.,1987) and Pseudomonas exotoxin (Lorberboum-Galski et al., 1988) has opened the way forspecific immunosuppression applicable in organ transplants and treatment of some autoimmunedisorders. The gene for human IL2 was cloned by Taniguchi et al. (1983) and first expressed inE. coli by Devos et al. (1983). Expression of the IL2 gene in E. coli typically leads to theformation of insoluble cytoplasmic inclusion bodies (Devos et al., 1983; Liang et al., 1985; Weirand Sparks, 1987; Bailon et al., 1987), which must be dissolved in denaturing solvents and theprotein subsequently refolded. The presence of a disulfide bond and one free cysteine residue inthe native polypeptide (Wang et al., 1984) complicates the refolding process, and yields ofcorrectly folded protein are often low (Weir and Sparks, 1987; Weigle et al., 1989). In addition,cytoplasmic expression of the mature form of IL2 results in two molecular species, one with anadditional N-terminal methionine residue (Nakagawa et al., 1987). Purification of recombinanthuman 1L2 is usually achieved by multiple chromatography steps, including gel permeation(Liang et al., 1985; Koths et al., 1986; Weir and Sparks, 1987), cation exchange (Kato et al.,1985; Tsuji et al., 1987; Weigle et al., 1989), and reverse phase HPLC (Liang et al., 1985; Katoet al., 1985; Koths et al., 1986; Tsuji et al., 1987; Weir and Sparks, 1987; Weigle et al., 1989).Affinity purification of human IL2 on immobilized IL2 receptors has also been reported (Bailon et23al., 1987). In this study the affinity tag approach was used to purify human IL2 with a cellulose-binding domain affinity tag. It was anticipated that the CenA leader peptide would direct exportof the fusion protein to the E. coli periplasm, where correct folding and formation of disulfidebonds could take place. This would allow affinity purification of the soluble fusion protein andproteolytic release of mature IL2 by specific proteolysis. Periplasmic export and correct foldinghas been achieved for a number of eukaryotic proteins, including human growth hormone(Hsiung et al., 1986), superoxide dismutase (Takahara et al., 1988), antibody fragments(reviewed in Pllickthun, 1991), and murine IL2 (Le et al., 1987). Human IL2 can retain itsbiological activity when heterologous polypeptides are fused to its N- (Hopp et al., 1988;Williams et al., 1987) or C-terminus (Lin et al., 1988; Lorberboum-Galski et al., 1988).1.7 Objectives.The overall objective of this study was to investigate the potential of CBDc enA as anaffinity tag for purification of heterologous proteins. The specific objectives were as follows:i) to use TnphoA to generate a range of fusions between CenA and alkaline phosphatase, both todetermine if CBDCenA retains its function when fused to a heterologous polypeptide and to probethe C-terminal end of the CBD for the minimum primary sequence requirements for cellulosebinding; ii) to develop a purification procedure for alkaline phosphatase based on celluloseaffinity chromatography, using CBDcemek as an affinity tag; iii) specific proteolytic removal ofthe CBD from the purified alkaline phosphatase fusion protein using Factor Xa and C. firniprotease; iv) to purify human interleukin 2 and remove the affinity tag using similar procedures.242. Materials and Methods2.1 Chemicals, enzymes, and buffers.All chemicals were of analytical or HPLC grade. Water for solutions and buffers wassingle glass distilled. Avicel PH-101 was from FMC International and CF1 cellulose was fromWhatman. Media components were from Difco. E. coli alkaline phosphatase (EC wasfrom Sigma. Buffers were prepared according to Sambrook et al. (1989) and Perrin andDempsey (1974).2.2 Bacterial strains, cell lines, plasmids, phage and transposons.Escherichia coli was the host organism used throughout this study for DNA manipulationand gene expression. The bacterial strains used are described in Table 2.1. The CTLL-2 cell line(Gillis and Smith, 1977) was provided by H.S. Teh. The plasmids, phage and transposons usedand/or constructed in this study are described in Table Media and growth conditions.Growth media for E. coli used in this study were LB medium (Miller, 1972) and TYPbroth (16 g tryptone, 16 g yeast extract, 5 g NaC1, 2.5 g K2HPO4 per litre). Cultures weregrown either at 30° or at 37°, as specified. Shaker speed for shake flask cultures was 250 rpmunless indicated otherwise. Induction of expression from lac and tac promoters was with• isopropyl-P-D-thiogalactoside (IPTG, Sigma) at concentrations ranging from 0.1 to 1 mM, asdescribed in the text. Antibiotics used were ampicillin (amp, Boehringer Mannheim) at 75 or 100pg/ml, for strains containing pUC, pTZ and pTUG plasmids, and kanamycin (kan, BoehringerMannheim) at 50 or 300 gg/ml, for TnphoA mutagenesis and single-stranded DNA  rescue Solidmedia contained 1.5% (w/v) agar.25Suppressor host foramplication of A,TnphoA-1De Bruijn and Lupski, 1984Alkaline phosphatase deletion Manoil and Beckwith, 1985mutantRelevant characteristicsRecombination deficientSource or Reference Yanisch-Perron et al., 1985Yanisch-Perron et al., 1985-lihk41^List of E. coli strains.E. coli strain GenotypeJM101^supE thi A(lac-proAB)IF' traD36 proAB lacIqZA1V115JM109^endAl recAl gyrA96 thi hsdR17 (rK- me) relAl supE44A(lac-proAB)IF traD36 proAB laclqZA1VI15LE392^F- hsdR574 (11C me) supE44 supF58 lacYl galK2 galT22metal trpR55CC 118^araD 139 A(ara leu)7697 AlacX74 phoAA20 galE galK thirpsE rpoB argE(am) recA 1RZ1032^HfrKL16 PO/45 [lysA(61-62)] dutl ungl thil relAl Zbd-279::Tn10 supE44BL21^hsdS ompTRW193^F- ara-14 leuB6 azi-6 lacYl proC14 tsx-67 entA403X - trpE381^IfbDl rpsL109 x371-5 mt1-1 thi-1UT5600i^RW193 A (ompTiepC)266Produces uracil substitutedsingle-stranded DNA for site-directed mutagenesisOmpT protease mutant;deficient in Ion proteaseOmpT protease mutantKunkel et al., 1987Studier and Moffatt, 1986,Grodberg and Dunn, 1988Earhart et al., 1979Earhart et al., 1979continued...continuedE. coli strain^ Genotype^ Relevant characteristics^Source or referenceKS272 F- AlacX74 galE galK thi rpsL (strA) AphoA (Pvuil)1pp-5508KS476^KS272 degP41 (APstI-Kanr) DegP protease mutantJon BeckwithStrauch et al., 1989Jon BeckwithStrauch et al., 1989AB 189i)^F- thr-1 ara-14 leuB6 A(gpt-proA)62 lacYl tsx-33 lon-1^Ion protease mutant^Howard-Handers et al., 1964supE44 galK2X- rac- hisG4 rfbDl rpsL31 kdgK51 xyl-5md-1 argE3 thi-1CAG627* lac(am) trp(am) pho(am) supC ipsL mal(am)^ Carol GrossCooper and Ruettinger, 1975CAG62^CAG627 lon -CAG597^CAG627 hipR165-Tn.10CAG62^CAG627^htpR165-Tn10Ion protease mutant^Carol Grossheat shock sigma factor mutantion protease/heat shock sigmafactor double mutant* also own as E. coli SC122Table 2.2^List of plasmids, phage and transposons.Plasmid^Promoter^Relevant characteristics^Source or ReferencepUC18-1.6cenApUC18-1.6cenA::TnphoAp'FZ18U-CBDCenApTZ18R-cenA-LNhepTZ18R-CBDPT-'phoAA1pTZ18R-CBDPT-'phoAA2pIX-8Alac Complete coding sequence of cenA,expressed as lacZ fusionlac Family of plasmids encoding cenA'-'phoA gene fusionslac Encodes CBDCenA without the Pro-Thr linkerlac cenA gene has Nhe I site separatingleader peptide and CBD codingregionslac Encodes CBDPT-'PhoA with deletionof CenA amino acids 2-19lac Encodes CBDPT-'PhoA with deletionof CenA amino acids 2-45lac Derivative of pUC18-1.6cenA::TnphoA LX-8 withtransposon sequences deletedlac Subclone from pUC18-1.6cenA::TnphoA #1-37 into pTZ18Rlac Derivative of pTZ18R-cenA'-'phoA#1-37A with Stu I site at 5' end ofphoA genelac Encodes CBDPTcenA with Factor Xacleavage site after Pro-Thr linkerlac Encodes CBDPTCenA fused to Pro-6of alkaline phosphatase throughFactor Xa cleavage sitetac Vector for high expression of cenAfragments and fusions in E. coliGuo et al., 1988This study;Greenwood et al.,1989N. Din, submittedfor publicationNordquist, 1992This studyThis studyThis study;Greenwood et al.,1992This studyThis studyGilkes et al., 1992;Shen Hua,unpublished resultsThis studyThis study;R.W. Graham andD.E. Trimbur, inpreparationpTZ18R-cenA'-'phoA#1-37ApTZ18R-#1-37AStu IpTZ18U-CBDPT-IEGRpTZ18U-CBDPT- 'phoApTUg 1 0*N18continued...28Phage Genotype^Source or ReferenceM13K07^kanr Vieira and Messing, 1987DescriptionTransposon Source or referenceXTnp hoA- 1 b221 c1857 Pam3 with TnphoA in or near rex^Gutierrez et al., 1987;C. ManoilTn5 derivative with alkaline phosphatase gene Manoil and Beckwith, 1985-inserted-into IS 50L    TnphoATable 2.2^continuedPlasmid^Promoter^Relevant characteristics^Source or ReferencepTUg10*-CBDPT-'phoApTUg10*-CBDPT-phoApUC18-1.6cenA4Pro-ThrpTUg10*-CBD- 'phoApUC13-1L2pUC18-cenA '-IL2pTZ18U-CBDPT-IL2tac CBDPT-'phoA subcloned into^This studypTUglO*N18tac Encodes CBDPTcenA fused to Thr-2 This studyof alkaline phosphatase throughFactor Xa cleavage sitelac Encodes CenA with deletion of the^Shen et al., 1991Pro-Thr linkertac Encodes CBDCenA fused to Pro-6 of This studyalkaline phosphatase through FactorXa cleavage sitelac* cDNA for human 1L2 cloned in^Taniguchi et al.,pUC13^ 1983;V. Petkaulac Encodes CenA'-IL2 fusion with Pro- This study213 of CenA fused to Ala-1 of IL2through Factor Xa cleavage sitelac Encodes CBDPTCenA fused to Ala-1 This studyof IL2 through Factor Xa cleavagesite* gene cloned in reverse orientation.29Cell density was determined by measuring the absorbance at 600 nm (A600) against agrowth medium blank in a Hitachi U-2000 spectrophotometer.Cell growth for single strand DNA rescue from phagemid vectors was in TYP brothcontaining ampicillin and 5 x 108 pfu/ml of M13K07 (D.E. Trimbur, personal communication).2 ml cultures were inoculated with a growing colony from an LB agar plate containing ampicillin(LB amp agar) and incubated at 37° until visible growth occurred. At this point, kanamycin wasadded to 50 tg/m1 and incubation was continued overnight at 37°.Amber suppressor strain E. coli LE392 was the host for amplification of phage XTnphoA-1, performed by the plate lysate method (Sambrook et al., 1989).The CTLL-2 cell line used in IL2 bioassays was maintained by S.J. Teh. Cells weregrown in IMDM medium (Gibco) supplemented with penicillin (100 U/ml) and streptomycin (100ilg/m1), 5 x 10 -5 M 2-mercaptoethanol and 10% foetal calf serum (Gibco). For maintenance ofthe cell line the medium was also supplemented with 50 mM a-methyl mannoside and 5-10%supernatant from ConA-activated rat spleen cells (Gillis and Smith, 1977). Cells were incubatedat 37° in a 5% CO2 atmosphere and split every 2-3 days at 1:10 dilution.Details of growth conditions not described here are given in the text.2.4 Recombinant DNA techniques.Standard techniques were used for all DNA manipulations (Sambrook et al., 1989). DNArestriction and modification enzymes were used according to manufacturer's instructions. DNAfragments were purified from excised agarose gel bands using the Gene Clean® kit (BIO 101, LaJolla, CA). Plasmid constructions are described in the text. Oligonucleotides were synthesizedon an Applied Biosystems automated DNA synthesizer model 380A (University of BritishColumbia Oligonucleotide Synthesis Lab) and purified by reverse phase chromatography on C18Sep Pak® columns (Millipore). Single strand DNA templates containing uracil, foroligonucleotide-directed mutagenesis, were prepared in E. coli RZ1032 (Kunkel et al., 1987).Oligonucleotide-directed mutagenesis was carried out according to Zhou et al. (1990), but using30Sequenase version 2.0TM (United States Biochemical) for primer extension. Single and doublestrand DNA sequencing was carried out by the dideoxy method using the Sequenase version2.0TM kit (United States Biochemical). Double strand DNA sequencing templates were purifiedfrom alkaline cell lysates either by Qiagen tip columns, by CsClisopycnic centrifugation, or byprecipitation in polyethylene glycol (PEG)(Kraft et al., 1988).Single strand phagemid DNA was purified by the following procedure (D.E. Trimbur,personal communication): 1.5 ml of a 2 ml culture, grown as described in Section 2.3, weretransferred to a 1.7 ml Eppendorf tube and centrifuged for 2-5 minutes at room temperature topellet cells. The supernatant was transferred to a new tube, mixed with 200 pl of 3.75 Mammonium acetate, 25% PEG 6000 and incubated on ice for 30 minutes. After centrifugation for10 minutes at 4° the supernatant was removed and the phage pellet was resuspended in 20 IA of10 mM Tris-HC1, 1 mM EDTA pH 8.0 (FE). 200 ill of 4 M NaC1O4 were added and the mixturewas incubated at room temperature for 5 minutes to dissociate the phage particles. The DNA wasthen bound to a glass fibre filter (GF/C, Whatman) by vacuum filtration: filter disks were cut tothe well size of a 96-well microtitre plate and placed in the wells of a flat-bottomed plate, whichhad been pierced with a hot wire. The plate was then fitted to a vacuum manifold and the samplewas filtered through slowly under vacuum. The filter was washed thoroughly with 70% ethanoland air dried for 5 minutes. The filter was transferred to a 0.5 ml Eppendorf tube (pierced at thebottom with a hot 21-gauge needle) using sterile tweezers, and 201110.1x TE were added to elutethe DNA. After 5 minutes incubation at room temperature, the tube was placed inside a 1.7 mlEppendorf tube and given a quick pulse in the centrifuge to transfer the liquid containing thesingle strand DNA to the larger tube. The DNA yield was estimated by agarose gelelectrophoresis. Typically, 5-7 gl of the preparation were used for sequencing or primerextension reactions.Competent cells were prepared and transformed according to Miller (1987).Transposon mutagenesis with TnphoA was carried out in E. coli CC118, a phoA deletionstrain (Table 2.1). Transposition was mediated by infection of plasmid-carrying cells with a  X suicide phage, XTnphoA-1 (Gutierrez et al., 1987). Cells grown to early stationary phase in LB31amp medium supplemented with 10 mM MgSO4 were infected with A,TnphoA-1 at a multiplicityof infection of approximately 1. After incubation at 30° for 15 minutes aliquots were diluted 1:10into LB amp medium and incubated for 4-15 hours at 30° to allow outgrowth. 0.2 ml of eachculture was plated on LB amp agar containing 300 gg/mlkanamycin and 401.1g/m15-bromo-4-chloro-3-indoly1 phosphate p-toluidine salt (XP, Sigma; section 2.5) and the plates wereincubated at 30° for 2-3 days. Plasmid was prepared from blue colonies by alkaline lysis(Sambrook et al., 1989) and was transformed into E. coli CC118. Transformants were selectedon LB amp agar containing XP and 50 ptg/mlkanamycin. Blue colonies were purified on LBagar containing XP and 50 tg/ml kanamycin, and in some cases were screened for 0-1,4-endoglucanase activity (section 2.5). Plasmid was prepared from purified colonies and screenedby restriction analysis to determine the positions of TnphoA insertion. The exact insertionpositions were identified by DNA sequencing.2.5 Detection of gene expression; enzyme assays.Screening for 13-1,4-endoglucanase activity in TnphoA insertion clones was carried out byreplica plating colonies onto LB amp medium containing 501.1g/mlkanamycin and 1% (w/v)carboxymethyl cellulose (CMC, Sigma), and detecting subsequent zones of CMC hydrolysis bystaining with Congo Red (Sigma)(Teather and Wood, 1982).Screening for alkaline phosphatase (PhoA) activity on agar plates was facilitated byadding XP from a 20 mg/m1 stock in N,N-dimethyl formamide to a concentration of 40 tg/m1 tothe cooled medium prior to pouring. Alternatively, 40 Ill of the XP stock was spread onto eachplate prior to inoculation. Alkaline phosphatase expression resulted in the formation of bluecolonies.Alkaline phosphatase activity was quantified by the rate of release of p-nitrophenol fromp-nitrophenyl phosphate (Sigma 104® phosphatase substrate) by following the change inabsorbance at 410 nm (Garen and Levinthal, 1960). Absorbance measurements were takenina_Hitachi U-2000 spectrophotometer. The cuvette holder was maintained at 25° by a refrigerated32circulating water bath (Forma Scientific). An extinction coefficient for p-nitrophenol in 1.0 MTris-HC1, pH 8.0 of 1.62 x 104 M -1 .cm -1 was used (Applebury and Coleman, 1969). One unitof alkaline phosphatase activity was defined as the amount of enzyme liberating 1 pmolep-nitrophenol in one minute under the described conditions. Normalization of published specificactivities determined at temperatures other than 25° was achieved using a Qic, value of 1.56(Garen and Levinthal, 1960).Screening for human interleukin-2 production in E. coli was by colony immunoblot.Colonies on agar plates were transferred to nitrocellulose (Schleicher & Schuell) or filter paper(Whatman No. 541), by overlaying the filters on the plates. Filters were transferred colony-side-up to LB amp agar plates supplemented with 0.1 mM IPTG, and were incubated at 37° for 4hours to allow expression from the lac promoter. Colonies were lysed according to Helfman andHughes (1987), and blots were processed in the same fashion as Western blots, using anti-humaninterleukin-2 antibody as a primary probe (section 2.8).IL2 biological activity was measured by a bioassay based on the proliferation of an IL2-dependent murine cell line, CTLL-2 (Gillis and Smith, 1977), according to a variation of themethod of Gillis et al. (1978) as described by Mossman (1983). Serial two-fold dilutions ofsample (0.21.1m filter sterilized) were added in duplicate to aliquots of 1.5 x 104 cells in a 96 wellflat-bottomed microtitre plate. After 24 hours incubation the tetrazolium salt MIT (344,5-dimethylthiazol-2-y1)-2,5-diphenyl tetrazolium bromide, Sigma) was added at 500 jig/ml. Duringthe subsequent 4 hours incubation metabolically active cells converted the MIT to a blueformazan product which was quantitated after dissolution by measurement of the absorbance at600 nm in a Bio-Rad microtitre plate reader, model 2550. The IL2 activity of the sample wascalculated relative to a murine 1L2 standard, provided by H.S. Teh. The dilution of the 1L2standard for which the A600 reading was 50% of the maximum was defined as having an activityof 1 U/ml.Succinate dehydrogenase (SDH), an E. coli cytoplasmic membrane marker, was assayedby the rate of reduction of 2,6-dichlorophenolindophenol (DCPIP, Sigma) by following thedecrease in absorbance at 600 nm (Kasahara and Anraku, 1974). Absorbance measurements33were taken using a Varian DMS 100 spectrophotometer connected to a chart recorder. Thecuvette holder was maintained at 37° by a Haake circulating water bath. An extinction coefficientfor DCPIP of 2.2 x 104 M -1 .cm-1 (pH 8.0) was used (Armstrong, 1964). One unit of SDHactivity was defined as the amount of enzyme catalyzing the reduction of 1 psnole of DCPIP in 1minute at 37° under the described conditions.The 8-carbon saccharide 2-keto-3-deoxyoctonate (KDO), in E. coli, is exclusively foundin the lipopolysaccharide and was assayed as an outer membrane marker using the method ofWeissbach and Hurwitz (1959) with some modifications (R.E.W. Hancock, personalcommunication). 10 p1 membrane samples were mixed with an equal volume of 0.5 N H2SO4and heated at 100° for 15 minutes. After cooling the sample to room temperature, 10 pl of 0.1 MH5I06 were added and the mixture incubated at room temperature for 10 minutes. 40 p1 ofNaAsO2 (0.3 M in 0.5 N HC1) and 160 p1 of 0.6 % thiobarbituric acid (prepared fresh) wereadded sequentially with vortex mixing. The mixture was then heated at 100° for 10 minutes andcooled to room temperature. 300 pl of butanol reagent (5 ml concentrated HC1 added to 95 ml n-butanol) were added and mixed well. The two phases were separated by centrifugation for 5minutes in a microfuge, and 200 pl of the upper butanol layer was transferred to a microtitreplate. The absorbance at 550 nm was measured in an ELISA plate reader (SLT Lab InstrumentsEasy Reader, model EAR400AT). A standard curve was prepared using pure KDO (0-200pg/m1), for which the heating step following H2SO4 addition was omitted. Pure KDO was a giftfrom R.E.W. Hancock.C. fimi protease was assayed by the hydrolysis of hide powder azure (HPA, Sigma), byfollowing the increase in absorbance at 585 nm (Gilkes et al., 1988). One unit of HPAse activitygives a change in A585 of 1.0 in 1 hour under the described conditions.2.6 Cell fractionation.Osmotic shock was carried out as described in Nossal and Heppel (1966). Chloroformextracts of periplasmic proteins for filter paper and Avicel binding assays were made from 2 ml34cultures grown overnight at 30° (Ferro-Luzzi Ames et al., 1984). For filter paper binding assayscultures were grown in LB medium without induction. For Avicel binding assays cultures weregrown in TYP broth and induced with 0.1 mM IPTG at early to mid log phase. Media containedeither no antibiotic, 50 p.g/mlkanamycin, or 100 pg/mlampicillin, depending on the antibioticresistance expressed by the cells.For preparation of cell extracts, samples were kept on ice and centrifugation was at 4°unless indicated otherwise. Cells were harvested by centrifugation at 6500-11,000 x g for 5-10minutes. Unless otherwise indicated, resuspended cells were ruptured by passage through aFrench pressure cell (Aminco) at 15,000-17,000 psi. Phenylmethylsulphonyl fluoride (PMSF,Sigma), a serine protease inhibitor, was dissolved in anhydrous acetone at 20 mg/ml and added tocrude cell extracts at 20-100 µg/ml. Where indicated, the aspartyl protease inhibitor pepstatin Awas added to crude cell extracts at 4 x 10 -5 M. To reduce the viscosity of cell extracts wherenecessary, DNA was either precipitated by the addition of streptomycin sulfate (Sigma) to 1.5%followed by incubation overnight at 4°, or degraded by the addition of DNase I (BoehringerMannheim)(up to 50 tg/ml) in the presence of magnesium ions (up to 5 mM) with incubation onice for up to 1 hour. Clarified cell extracts were produced by centrifugation of crude cell extractsat 42,000 x g for 30 minutes. Further details on the preparation of crude and clarified cell extractsare given in the text.Partial fractionation of E. coli BL21/pUC18-cenA'-IL2 cell extracts (Results, section3.2.4) was performed according to a method from R.E.W. Hancock (personal communication)with modifications. 200 ml of LB amp medium were inoculated with 2 ml of an overnight culturegrown at 30° then incubated at 30° until A600 = 1.4. IPTG was added to 0.5 mM and incubationwas continued for 2 hours. Harvested cells were resuspended in 4.5 ml 10 mM sodiumphosphate, 5 mM MgSO4 pH 7.4. DNase I was added to 50 µg/ml and PMSF to 100 lag/m1prior to rupture with a French press (3 passages at 17,000 psi). PMSF was again added to 100µg/ml after rupture, and the crude cell extract was centrifuged at 1000 x g for 10 minutes at 4° topellet cell debris and any protein inclusion bodies. The pellet (low speed pellet) was resuspendedin 5 ml distilled water by vortex mixing. The supernatant (low speed supernatant) was35centrifuged at 140,000 x g for 1 hour at 4° to pellet the cell envelopes. The supernatant (highspeed supernatant) was removed and the pellet (high speed pellet) was resuspended in 2.5 mldistilled water by vortex mixing. An equal volume of 20 mM Tris-HC1, 10 mM MgSO4, 4%Triton X-100 pH 6.8 was added, and the suspension was incubated overnight at 4°. The highspeed centrifugation step was repeated and the supernatant (Triton supernatant) and pellet (Tritonpellet) were recovered. The Triton pellet was resuspended in 5 ml distilled water by vortexmixing.Separation of E. coli cytoplasmic and outer membranes was by sucrose isopycniccentrifugation (Osborn and Munson, 1974). The bottom of each gradient tube was punctured and1 ml fractions were collected. Fractions were diluted 10-fold with 1 mM EDTA and centrifugedat 250,000 x g for 2 hours at 4° to pellet membranes. Membranes were resuspended in 50 mMTris-HC1 pH 8.0 by vortex mixing prior to protein and membrane marker assays.Insoluble cell fractions were recovered by centrifugation at 105,000 x g for 1 hour at 4°unless indicated otherwise. Pellets were resuspended either by vigorous vortex mixing, bysonication with a Branson Sonifier® cell disrupter 350 (5x10 pulses at output 1.5, 20% dutycycle), or by homogenization in a Wheaton hand-held homogenizer.2.7 Protein determination.Protein concentrations were usually measured by the Bio-Rad dye-binding assay(Bradford, 1976), using bovine serum albumin (BSA) as a standard. Purified CBDPT-'PhoAwas the standard used for assay of alkaline phosphatase fusion protein preparations prior topurification. A detergent-compatible modification of the Lowry protein assay (Sandermann andStrominger, 1972) was used for assay of cell fractions during purification of interleukin-2 fusionproteins, with BSA as a standard. For assay of membrane fractions containing sucrose, amodification of the Bio-Rad assay was used (Simpson and Sonne, 1982), again with BSA as astandard. Purified alkaline phosphatase fusion proteins were quantified by measuring theabsorbance at 280 nm (A280) in a Hitachi U-2000 spectrophotometer. The extinction coefficients,36calculated according to Mach et al. (1992), were: CBDPT-'PhoA, E280 = 1.15 ml.mg - l.cm-1 ;CBDPT-PhoA, 6280 = 1.14 ml.mg - l.cm-1 ; CBD-'PhoA, 6280 = 1.20 ml.mg- l.cm-1 . Empiricallydetermined extinction coefficients (Scopes, 1974) were within 12% of these values. Theextinction coefficient for CBDPT-'PhoA was also used for determination of the monomeric formof the fusion protein (section 2.12.1). For determination of pure alkaline phosphatase, anextinction coefficient of 0.77 ml.mg- l.cm-1 at 280 nm was used (Rothman and Byrne, 1963).2.8 Protein gel electrophoresis/Western blots.Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS PAGE) was carried outaccording to Laemmli (1970). Molecular weight standards (Sigma) were, with molecular massesin daltons: myosin, 212,000; p-galactosidase, 130,000; phosphorylase B, 97,400; bovine serumalbumin, 68,000; catalase, 57,000; glutamate dehydrogenase, 53,000; alcohol dehydrogenase,45,000; ovalbumin, 41,000; glyceraldehyde-3-phosphate dehydrogenase, 36,000; carbonicanhydrase, 29,000; bovine trypsin inhibitor, 20,100; cytochrome C, 12,400. Prestainedmolecular weight standards were from BRL and Bio-Rad. In all figures the sizes of molecularweight standards are indicated (molecular weight x10 -3). Protein gels were stained withCoomassie Brilliant Blue R250 (Bio-Rad)(Sambrook et al., 1989). For alkaline phosphataseactivity staining, gels were incubated in 2.5% Triton X-100 for 1 hour to remove SDS, andequilibrated in 50 mM Tris-HC1, 2 tM ZnC12 pH 8.0 for 1 hour before staining for 15-30minutes as described in section 2.9.Scanning densitometry of Coomassie blue-stained SDS polyacrylamide gels was carriedout using a Molecular Dynamics Computing Densitometer.For N-terminal amino acid sequencing, polypeptides separated on SDS polyacrylamidegels were electrophoretically transferred onto ImrnobilonTm-P membrane (Millipore Corp.) andstained with Coomassie blue as described (Matsudaira, 1990). Excised bands were sequenceddirectly by automated Edman degradation using an Applied Biosystems 470A gas-phase37sequenator with on-line PTH analyzer and 900A system controller and data analyzer (ProteinSequencing Facility, University of Victoria, Victoria, B.C.).Western blots were produced according to Towbin et al. (1979), with 0.1% SDS added tothe transfer buffer. Blocking and antibody binding were in phosphate-buffered saline (PBS), 3%BSA (Harlow and Lane, 1988) and washes were in PBS, 0.05% Tween 20. Primary antibodiesused were: i) rabbit polyclonal antiserum raised against CenAm (Wong et al., 1986; Whittle etal., 1982), used at a 1/3000 dilution;^Affinity purified goat polyclonal antibody raised againstrecombinant human interleukin-2 (hIL2), used at a dilution of 1/10,000 (R&D Systems).Primary antibodies were preadsorbed with E. coli lysates (Helfman and Hughes, 1987). Alkalinephosphatase-conjugated secondary antibodies were used (Blake et al., 1984): i) goat anti-rabbitIgG (BRL), used at a 1/3000 dilution; ii) rabbit anti-goat IgG (Sigma), used at a 1/1000 dilution.Colour development was as described (Blake et al., 1984), with 50 mM Tris-HC1 pH 9.6 as thefinal wash and assay buffer.2.9 Cellulose preparation; binding analysis.The binding of CenA'-'PhoA fusion proteins to filter paper was carried out usingchloroform extracts (section 2.6). 75 IA of chloroform extract were applied to each of two discs(1.6 cm diameter) of Whatman 541 filter paper, which had been preincubated with 50 mM Tris-HC1 pH 7.5, 5% BSA to prevent non-specific binding of proteins. One filter from each pair waswashed exhaustively with 50 mM Tris-HC1 pH 7.5 and with 0.5 M NaCl. Both filters werestained for PhoA activity (Avrameas et al., 1971) with 1 mg/ml naphthol AS-MX phosphate(Sigma), 2 mg/ml Fast Red TR salt (Sigma) in 50 mM Tris-HC1 pH 8.0.Avicel was washed three times in distilled water to remove fines and was equilibrated inthe same buffer as the sample prior to binding. Avicel was recovered from suspension bycentrifugation for 1 minute in a microfuge. Centrifugation of larger Avicel samples was at 3000 xg for 5 minutes in a Beckman JA20 rotor or at 1000 x g for 1 minute in a Sorvall GLC-1swinging bucket centrifuge.38Binding of proteins to Avicel, unless indicated otherwise, was for 1 hour on ice withmixing every 5-10 minutes. For small scale Avicel binding assays, carried out in 1.5 mlEppendorf tubes, the supernatant (Avicel supernatant) was recovered by centrifugation, and theAvicel pellet was subjected to various buffer washes, as described in the text. Proteins bound tothe Avicel were recovered (Avicel extract) by boiling the Avicel directly in SDS PAGE samplebuffer.For binding of CenA'-'PhoA fusion proteins from chloroform extracts (section 2.6) toAvicel, small columns were prepared from P1000 pipette tips. Tips were plugged with glasswool and packed with 10 mg of Avicel using a peristaltic pump (Holter 909) operating at 0.11ml/minute. The peristaltic pump was operated continuously at 0.11 ml/minute during eachbinding assay. 100 tl of chloroform extract were applied to the column and drawn throughcompletely, before separate washes of 250 tl 50 mM Tris-HC1, 1 M NaCl pH 7.5 and 250 pl 50mM Tris-HC1 pH 7.5 were applied. To elute bound protein, 133 pl of SDS PAGE sample bufferwhich had been heated on a boiling water bath were applied to the column. Column flowthroughand Avicel extract fractions were collected for SDS PAGE analysis.CF1 cellulose was washed several (10-20) times with distilled water to remove fines;after each wash the cellulose was allowed to settle and the supernatant was removed byaspiration. The cellulose was equilibrated with 50 mM Tris-HC1 pH 7.5, 0.02% NaN3 andstored in this buffer at 4°. Samples of the evenly suspended material were washed several timeswith distilled water and dried to constant weight at 42° to determine the cellulose concentration.Glass columns were packed with measured volumes of the suspension using a peristaltic pumpoperating at 2 ml/minute. Columns used were from Pharmacia (XK 26/20 and XK 50/30;medium scale purifications) and Bio-Rad (5 cm x 1 cm I.D. Econo-Column®; small scale bufferconcentration/binding experiments).Buffer concentration/binding experiments with purified CBDPT-'PhoA were carried outon columns containing 150 mg CF1 cellulose, connected to an FPLCTM system (Pharmacia LKBBiotechnology, Uppsala, Sweden), and equilibrated in the same buffer as the sample to be  loaded. Purified CBDPT-'PhoA was diluted to a concentration of 0.1 mg/ml in Tris-HC1 pH 7.539at a final Tris concentration of 1, 5, and 50 mM. Protein solution was applied to the column at0.5 ml/minute using a 1 ml sample loop. Loading was followed with a 5 ml wash with samplebuffer and elution with 10 ml distilled water. 8 ml flowthrough/wash and water elution fractionswere collected and assayed directly for alkaline phosphatase activity.2.10 Protein purification: Alkaline phosphatase fusion proteins.2.10.1 CenA'-'PhoA IX-8 heterodimers: heterodimer dissociation and reassociation.For CenA'-'PhoA IX-8 heterodimer dissociation and reassociation experiments, aclarified cell extract of E. coli CC118/pIX-8.6, was used. A 2 ml culture grown at 30°overnight inLB amp medium was inoculated into 25 ml LB amp medium and incubated at 30° and 200 rpmuntil A600 = 0.8; IPTG was then added to 0.2 mM and incubation was continued for 6 hours.Harvested cells were resuspended in 2 ml 50 mM Tris-HC1 pH 7.5, and ruptured with a Frenchpress. No protease inhibitors were added after cell rupture. The clarified cell extract wasproduced by centrifugation at 105,000 x g for 30 minutes at 4°.The heterodimer dissociation experiment was divided into three parts, each involvingbinding of 25 p1 of E. coli CC118/pIX-8A clarified cell extract to 10 mg Avicel:i) 25 ill of clarified cell extract were mixed with 75 ill 50 mM Tris-HC1 pH 7.5, and wereincubated with 10 mg Avicel for 30 minutes on ice (section 2.9). The Avicel supernatant wasrecovered by centrifugation and the Avicel was washed once with 500 pl 50 mM Tris-HC1, 1 MNaC1 pH 7.5 and twice with 500 ml 50 mM Tris-HC1 pH 7.5. Avicel-bound polypeptides wereextracted into SDS PAGE sample buffer.ii) 25 pl of clarified cell extract were bound to Avicel and the Avicel was washed as described ini), and the Avicel supernatant was discarded. The Avicel was resuspended in 49 µ1 50 mM KC1-HC1 pH 2.0 and the pH was adjusted to 2 with (1 p,1) 1N HC1. After incubation for 15 minuteson ice the Avicel supernatant was recovered and the Avicel was resuspended in 50 IA 50 mMKC1-HC1 pH 2.0 and the 15 minute incubation was repeated. The Avicel supernatants were40pooled and the Avicel was washed once with 500 µl50 mM KC1-HC1 pH 2.0 and once with 500pl 50 mM Tris-HC1 pH 7.5 before extraction of bound polypeptides into SDS PAGE samplebuffer.iii) 50 pl of clarified cell extract were mixed with 141 pl 50 mM KCl-HC1 pH 2.0 and adjusted topH 2 with (3 p.1) 1 N HC1. After incubation on ice for 30 minutes, precipitated proteins wereremoved by centrifugation at 105,000 x g for 10 minutes at 4°. The supernatant (acidified cellextract) was split and one half (97 pl) was mixed with 1 p.1 0.5 M EDTA and 2 pi 2 N NaOH,and retained for SDS PAGE analysis. The other 97 pl was incubated with 10 mg Avicel(equilibrated with 50 mM KC1-HC1 pH 2.0) on ice for 30 minutes. The Avicel supernatant wasrecovered and mixed with 1 pl 0.5 M EDTA and 2 pl 2 N NaOH. The Avicel was washed oncewith 500 pi 50 mM KC1-HC1 pH 2.0, once with 500 pl 50 mM Tris-HC1, 1 M NaC1 pH 7.5, andtwice with 500 pl 50 mM Tris-HC1 pH 7.5, and bound proteins were extracted with SDS PAGEsample buffer. The clarified and acidified cell extracts, Avicel supernatants and Avicel extractsfrom i), ii) and iii) were analyzed by SDS PAGE.For reassociation of heterodimers on Avicel, Avicel-bound monomers of CenA'-'PhoAIX-8 were generated by following procedure ii), above, as far as the wash with 500 µ15O mMKC1-HC1 pH 2.0. 5 µl0.5 M EDTA were added to the Avicel during this wash. The Avicel wasthen washed once with 500 µ15O mM KC1-HC1, 5 mM EDTA pH 2.0 and once with 500 pl 50mM Tris-HC1, 5 mM EDTA pH 7.5. Six such preparations of Avicel-bound CenA'-'PhoA IX-8monomers were prepared for the reassociation experiment; each 10 mg Avicel sample hadapproximately 6 tg fusion protein monomer bound.Purified E.coli alkaline phosphatase was also converted to the monomer form byincubation at low pH. Samples from an approximately 2 mg/ml suspension of the protein in 2.5M (NH4)2SO4 were adjusted to pH 2 with 1 N HC1 and diluted either 20 or 40 fold in 50 mMKCl-HC1 pH 2.0. After incubation on ice for 30 minutes, EDTA was added to 5 mM and thesamples were incubated a further 5 minutes on ice. The protein samples were neutralized with 2N NaOH before addition to the Avicel pellets as described below.41Six preparations of CenA'-'PhoA IX-8 monomers bound to 10 mg Avicel wereresuspended as follows:a) and b) in 100 pl 50 mM Tris-HC1, 5 mM EDTA pH 7.5;c) and d) in 100 pl of alkaline phosphatase monomer solution at —0.1 mg/ml;e) and in 100 pi of alkaline phosphatase monomer solution at —0.05 mg/ml.10 pl distilled water were added to tubes a), c), and e), and 10 IA 0.1 M ZnC12 wereadded to tubes b), d), and f). The Avicel suspensions were incubated for 1 hour at 37°, withregular mixing. The Avicel in tubes a), c), and e) (without zinc) was washed twice with 500 IA50 mM Tris-HC1, 5 mM EDTA pH 7.5, and the Avicel in tubes b), d), and f) (with zinc) waswashed twice with 500 pi 50 mM Tris-HC1 pH 7.5. Bound proteins were then extracted intoSDS PAGE sample buffer and analyzed by SDS PAGE.2.10.2 Medium scale purification of CBDPT-'PhoA: single column loading andelution with guanidine-HC1 or distilled water.CBDPT-'PhoA was purified from a culture supernatant of E. coli CC118/pTUg10*-CBDPT-'phoA by CF1 cellulose affinity chromatography. 8 ml of a TYP amp culture grownovernight at 30° were inoculated into 400 ml TYP amp broth (2% inoculum). After 6 hoursgrowth at 30° and 250 rpm, IPTG was added to 0.1 mM, and growth was continued for 24hours. The culture supernatant was recovered by centrifugation and samples were taken forprotein and alkaline phosphatase activity assays. The culture supernatant was cooled to 4° prior topurification on a Pharmacia XK 50/30 column containing 20 g CF1 cellulose (section 2.9), with abed volume of approximately 100 ml. The column was maintained at 4° and was connected to anFPLC system, operating at a flowrate of 1 ml/minute. 90 ml of culture supernatant (containingapproximately 20 mg of fusion protein) were loaded onto the column at 1 ml/minute with aPharmacia P-1 peristaltic pump. The column was washed with 300 ml of 50 mM Tris-HC1, 1 MNaCl pH 7.5, 0.02% NaN3 (high salt buffer), followed by a 100 ml linear gradient to 50 mMTris-HC1 pH 7.5, 0.02% NaN3 (low salt buffer), and 150 ml of low salt buffer. The procedure42to this point was followed for both guanidine-HC1 elution and distilled water elution runs. FreshCF1 cellulose was packed into the column for each run.For the guanidine-HC1 elution run, CBDPT-'PhoA was eluted from the column with a200 ml linear gradient of 0-6 M guanidine-HC1 in 50 mM Tris-HCJ pH 7.5, followed by 200 mlof 6 M guanidine-HC1 in the same buffer. This was followed by a 50 ml linear gradient to lowsalt buffer and a final wash of 150 ml of low salt buffer. 15 ml fractions were collectedthroughout the run, and for each fraction the A280 was measured in a Hitachi U-2000spectrophotometer and the conductivity was measured with a digital conductivity meter (VWRScientific model 604). The fractions were divided into column flowthrough, high salt wash, lowsalt wash and guanidine-HC1 elution peak pools, and the alkaline phosphatase activity of the firstthree pools was determined. The pooled guanidine-HC1 elution peak was filtered through a 0.45polysulphone membrane filter (Gelman) and concentrated to 50 ml against a PM30 membrane(Amicon) in a 50 ml Amicon ultrafiltration unit, operated at room temperature. The guanidine-HC1 concentration was reduced to 10 p.M by diafiltration with 650 ml of 50 mM Tris-HC1, 0.1mM ZnC12, 0.1 mM MgC12 pH 7.5 at room temperature, and the renatured protein solution wasconcentrated at 4° to about 1 ml final volume. Any precipitated protein was removed bycentrifugation at 105,000 x g for 15 minutes at 4°, and the supernatant was stored at 4°.For the distilled water elution run, CBDPT-'PhoA was eluted from the column with a 200ml linear gradient of low salt buffer to distilled water, followed by 200 nil of distilled water. Thiswas followed by a 95 ml linear gradient to low salt buffer, and any remaining elutable protein wasdesorbed with a 100 ml linear gradient of 0-6 M guanidine-HC1 in 50 mM Tris-HC1 pH 7.5, and100 ml of 6 M guanidine-HC1 in the same buffer. Finally the column was washed with a 50 mllinear gradient to low salt buffer and 150 ml of low salt buffer. 15 ml fractions were collectedthroughout the run, and A280 and conductivity readings were made as described above. Inaddition the alkaline phosphatase activity was measured for fractions around the water elutionpeak. The water elution peak was filtered through a 0.45 pm polysulphone membrane filter, andthe filtrate was adjusted to 50 mM Tris-HC1 pH 7.5, 0.1 mM ZnC12, and 0.1 mM MgC12 prior tomeasurement of A280 and alkaline phosphatase activity. The protein solution was concentrated to43less than 2 ml by ultrafiltration (UF) at 4° against a PM30 membrane in a 50 ml Amiconultrafiltration unit, and precipitated proteins were removed by centrifugation as described above.The guanidine-HC1 elution peak was treated as described above, except that the operating volumefor diafiltration was 25 ml instead of 50 ml.2.10.3 Purification of CenA-PhoA fusion proteins: multiple column loading anddistilled water elution.CF1 cellulose affinity chromatography with multiple column loading and distilled waterelution was used to purify CBDPT-'PhoA from a culture supernatant and clarified cell extract ofE. coli CC118/pTUg10*-CBDPT-'phoA, and CBDPT-PhoA and CBD-'PhoA from culturesupernatants of E. coli CC118/pTUg10*-CBDPT-phoA and /pTUg10*-CBD-'phoA respectively.Cultures were grown and culture supernatants were prepared as described in section 2.10.2,except that the culture volumes for E. coli CC118/pTUg10*-CBDPT-phoA and /pTUg10*-CBD-phoA were both 100 ml. For cell extract preparation of E. coli CC118/pTUg10*-CBDPT-sphoA, cells were resuspended in 5 ml low salt buffer and ruptured with a French press. PMSFwas added to 100 µg/ml immediately following rupture, and the crude cell extract was clarified bycentrifugation at 105,000 x g for 1 hour at 4°. For all purifications the CF1 cellulose was packedinto a Pharmacia XK 26/20 column. The column was maintained at 4° and was connected to anFPLC system. Culture supernatants were loaded onto the column using a Pharmacia P-1peristaltic pump, and the clarified cell extract was loaded using a 10 ml superloop (Pharmacia).For each purification cycle the sample loading was followed by a high salt buffer wash, low saltbuffer wash, water elution, and low salt buffer re-equilibration. All the different steps wereseparated by linear gradients. All flowrates were 1 ml/minute, except in the purification ofCBDPT-PhoA and CBD-'PhoA, where the high salt wash, gradient to low salt, low salt wash,and low salt re-equilibration were at 1.5 ml/minute. Details of the various purifications are givenin Table 2.3. Pooled water elution fractions for each cycle were filtered through a 0.45 j.tmpolysulphone membrane filter and adjusted to 50 mM Tris-HCJ pH 7.5, 0.1 mM ZnC12, and 0.144Protein CBDPT-'PhoA CBDPT-'PhoA CBDPT-PhoA CBD-'PhoASource culture clarified cell culture culturesupernatant extract supernatant supernatantWeight of CF1cellulose (mg)3 6 6 6No. of purificationcycles10 5 5 3Loading per cycle(ml)9 0.6 20.7 25.7High salt washvolume (ml)46 60 85 90High-low saltgradient (ml)15 30 40 30Low salt washvolume (ml)25 15 45 45Low salt wash-->water gradient (ml)30 60 45 45Water elutionvolume (ml)30 60 75 75Water-->low saltgradient (ml)10 15 15 15Low salt re-equilibration (ml)20 45 30 30Table 2.3^Operating conditions for CF1 cellulose affinity purifications with multiplecolumn loading.45mM MgC12 prior to measurement of alkaline phosphatase activity. The fractions were thenconcentrated by ultrafiltration and clarified as described in section 2.10.2. For the 10-cyclepurification of CBDPT-'PhoA, fusion protein remaining bound to the column after the finalpurification cycle was eluted with a 15 ml linear gradient of 0-6 M guanidine-HC1 in low saltbuffer, and 15 ml of 6 M guanidine-HC1 in the same buffer. This was followed by an 8 ml lineargradient to low salt buffer and a final column wash of 22 ml of low salt buffer. The guanidine-HC1 elution peak was treated as described in section 2.10.2, with a diafiltration volume of 25 ml.2.10.4 Removal of CBDCenA following protease digestion.Factor Xa-digested CBDPT-PhoA (section 2.12.1) was passed over an Avicel column toremove the CBDcenA fragment. 20 mg of Avicel was equilibrated in 50 mM Tris-HC1 pH 8.0,0.1 M NaCl, 1 mM CaC12 and packed into a Mobicol column (Mobitec). 5014 of the digestedprotein in 50 g.t1 volume were applied to the column and allowed to pass through under gravity.Any entrained liquid was forced through with a syringe. This was followed with 50 41 of thesame buffer, and the two flowthrough fractions were pooled. The Avicel was suspended in 500of the same buffer and transferred to a 1.5 ml Eppendorf tube. The Avicel was recovered bycentrifugation and the wash fluid discarded. Bound polypeptides were then extracted into SDSPAGE sample buffer.C. Jlmi protease-digested CBD-'PhoA (section 2.12.2) was also passed over an Avicelcolumn as described above. In this case, 75 i_tg of the digested protein in 150 IA volume wereapplied to the column, and low salt buffer was used to equilibrate the Avicel and for column andAvicel washes. Column flowthrough and Avicel-bound fractions were collected for SDS PAGEanalysis.2.11 Protein purification: IL2 fusion proteins.2.11.1 Direct binding of CenA'-IL2 to Avicel.46To study the direct binding to Avicel of CenAt-IL2 from the insoluble fraction of E. coli,a crude cell extract was prepared from a 50 ml 20 hour uninduced culture of E. coliBL21/pUC18-cenitli-IL2 (section 2.6). Cells were resuspended in 5 ml 50 mM sodiumphosphate pH 7.0 prior to rupture. 0.6 ml of crude cell extract was adjusted to 2% Triton X-100and 20 mM MgC12 at a protein concentration of 7 mg/ml and incubated 10 minutes on ice. Theinsoluble fraction was recovered by centrifugation and resuspended in 0.6 ml sodium phosphatepH 7.0 by sonication. The suspension was adjusted to 2% Triton X-100 and 5 mM MgC12 andincubated with 10 mg Avicel (section 2.8). The Avicel supernatant was removed and added toanother 10 mg of Avicel, and the suspension was sonicated (50 pulses at output 1.5, 30% dutycycle, with cooling after every 10 pulses) in a Branson Sonifier cell disrupter 350. Both Avicelpellets were washed with 600 pi 50 mM sodium phosphate, 5 mM MgC12, 2% Triton X-100 pH7.0, and 600 pl 50 mM sodium phosphate, 5 mM MgC12, 2% Triton X-100, 0.5 M NaC1 pH7.0. Samples of all fractions were taken for analysis by SDS PAGE.To test different wash solutions for removal of non-specifically bound protein fromAvicel, the same crude cell extract was used; 0.6 ml of cell extract was extracted with Triton X-100 as described above. The pellet was resuspended in 200 pl 50 mM sodium phosphate pH 7.0by sonication. The suspension was added to 30 mg Avicel and sonicated as described above.The Avicel was washed with 800 pl 50 mM sodium phosphate pH 7.0, and aliquots containing 2mg Avicel were transferred to separate tubes. Each Avicel pellet was mixed with 200 pl of washsolution and incubated for 5 minutes on ice. The wash solutions (described in the legend toFigure 36) were all in 50 mM sodium phosphate pH 7.0, with the exception of the distilled waterwash. The Avicel supernatants were removed and the Avicel pellets were washed once with 100pl 50 mM sodium phosphate pH 7.0 (or distilled water). For each Avicel sample the twosupernatants were combined and 10 pl were passed through a nitrocellulose filter on a dot blotmanifold (BRL HYBRI•DOrm). The nitrocellulose was probed with anti-CenAEC2 serum asdescribed in section 2.9. The Avicel pellets were resuspended in SDS PAGE sample buffer andboiled for 10 minutes, prior to analysis by SDS PAGE.472.11.2 CenA 1-IL2 purification: direct binding to Avicel.1 1 of LB amp medium containing 0.1 mM IPTG was inoculated with 50 ml of a mid logphase culture of E. coli BL21/pUC18-cenA'-IL2 and incubated for 20 hours at 30°. The cellswere harvested by centrifugation and resuspended in 40 ml of 50 mM sodium phosphate, 3 mMEDTA pH 7.0 (phosphate/EDTA). After rupture with a French press, PMSF, MgC12 and DNaseI were added to 1001.ig/n -11, 5 mM and 1 µg/ml respectively, and the crude cell extract wasincubated on ice for 2 hours. After protein determination the crude cell extract was diluted to aprotein concentration of 5 mg/ml with phosphate/EDTA and Triton X-100 was added to 2%.After overnight incubation at 4° the insoluble fraction was collected by centrifugation and wasresuspended in 10 ml phosphate/EDTA by homogenization. The Triton X-100 extraction andcentrifugation steps were repeated and the insoluble fraction was resuspended again in 10 ml ofphosphate/EDTA. The resuspended pellet was divided in two and CenA'-IL2 was purified fromone half by the alkaline solubilization method (section 2.11.4). The other half was added to 100mg Avicel and sonicated for 5x15 pulses at output 2, 30% duty cycle in a Branson Sonifier celldisrupter 350. The Avicel supernatant was recovered by centrifugation, and the Avicel waswashed with 5x5 ml phosphate/EDTA, 3x5 ml phosphate/EDTA, 0.5 M NaC1, 3x5 mlphosphate/EDTA, 5 ml phosphate/EDTA, 0.1% SDS, and 5 ml phosphate/EDTA, 1% SDS. TheSDS washes were carried out at room temperature and the other washes were at 0°. SDS wasremoved from the 0.1% wash as follows: KC1 was added to 0.1 M and the solution wasincubated on ice for 2 hours. After centrifugation at 20,000 x g for 10 minutes, the supernatantwas adjusted to 0.5% Triton X-100 and incubated overnight at room temperature. One tenth ofthe total volume was incubated with 10 mg Avicel (section 2.9) and the supernatant and Avicelfractions were recovered for SDS PAGE analysis.2.11.3 CenA'-IL2 solubilization trials.48Solubilization trials for CenA'-1L2 were carried out as follows:All detergent solubilization trials were on the crude cell extract of E. coli BL21/pUC18-cenA'-IL2 from a 20 hour uninduced culture. Cells were resuspended in 0.1 culture volume ofphosphate/EDTA prior to rupture. 100 [1.1 of cell extract, adjusted to 5 mg protein/ml, were mixedwith 100 pi detergent solution at 2x concentration and incubated 1 hour at 4°. After centrifugationat 105,000 x g for 1 hour at 10°, supernatants were removed and pellets were resuspended in anequal volume of 50 mM sodium phosphate pH 7.0 by sonication. Supernatant and pellet sampleswere analyzed by Western blotting with anti-CenAEC2 serum. All other solubilization trials wereon membrane pellets from a Triton X-100-extracted crude cell extract of E. coli BL21/pUC18-cen/1 1-1L2. Cells from a 24 hour culture were resuspended in 0.04 culture volume ofphosphate/EDTA prior to rupture. The crude cell extract was diluted to a protein concentration of10 mg/ml and Triton X-100 was added to 2%. Following incubation for 18 hours at 4° the extractwas centrifuged as above and the pellet resuspended in phosphate/EDTA by homogenization(section 2.6). Samples containing 1.5 mg protein were recentrifuged and the pellets resuspendedin 200111 of the solubilization buffer by sonication. Following incubation for 18 hours at roomtemperature the samples were treated in the same fashion as the detergent-treated samples.2.11.4 CenA 1-IL2 purification: alkaline solubilization.The Triton X-100-extracted insoluble fraction of E. coli BL21/pUC18-cenA '-IL2 from a 11 culture was prepared as described in section 2.11.2 and resuspended in 10 ml phosphate/EDTA.One half of the resuspended pellet was centrifuged again and the insoluble pellet resuspended to aprotein concentration of 5 mg/ml in 25 mM Na2HPO4-NaOH pH 11.5. After 2 hours incubationat room temperature the insoluble material was pelleted by centrifugation. The supernatant (20ml) was added to 100 mg of Avicel and diluted 5-fold with phosphate/EDTA buffer at roomtemperature, to a final pH of 7.1; the buffer was added to the mixed Avicel suspension at 0.5ml/minute using a peristaltic pump. The Avicel supernatant was recovered by centrifugation  and 49the Avicel was washed with 2x5 ml phosphate/EDTA, 0.5 M NaCl and 2x5 ml phosphate/EDTA.Samples of all fractions were taken for SDS PAGE analysis.2.11.5 CBDPT-IL2 purification• alkaline solubilization.1 1 of LB amp medium was inoculated with 50 ml of a culture of E. coli BL21/pTZ18U-CBDPT-IL2 which had been grown at 30° until A600 = 1.5. The inoculated culture was incubatedat 30° for 21 hours. Harvested cells were resuspended in 10 ml phosphate/EDTA buffer. Afterrupture with a French press, PMSF, MgC12 and DNase I were added to 100 µg/ml, 10 mM and 51.1g/m1 respectively. After incubation on ice for 30 minutes, EDTA was added to 10 mM and thecrude cell extract was centrifuged to pellet the insoluble fraction. The pellet was resuspended in 5ml phosphate/EDTA buffer by sonication and adjusted to a protein concentration of 5 mg/ml andmade 2% in Triton X-100. After incubation overnight at 4° the insoluble fraction was againpelleted by centrifugation and the pellet resuspended in 25 mM Na2HPO4-NaOH pH 11.5 to aprotein concentration of 5 mg/ml by homogenization. After 2 hours incubation at roomtemperature, the suspension was again centrifuged, and the supernatant was recovered. Thesupernatant was incubated with 500 mg Avicel overnight at 4° on a Labquake® Shaker(Labindustries, Inc.). The Avicel supernatant was recovered by centrifugation and the Avicelpellet was washed with 2x10 ml 25 mM Na2HPO4-NaOH pH 11.5, then repeatedly with 10 mlvolumes of phosphate/EDTA, 1 M NaC1, until the A280 of the supernatant was less than 0.01.The Avicel was washed finally with 3x 10 ml phosphate/EDTA and resuspended in 5 ml of thesame buffer. Samples of all fractions were taken for SDS PAGE and Western blot analysis.2.12 Protease digestion of fusion proteins.2.12.1 Factor Xa digestion of alkaline phosphatase fusion proteins.50All Factor Xa digestions of alkaline phosphatase fusion proteins were carried out at roomtemperature in 50 mM Tris-HC1 pH 8.0, 0.1 M NaC1, 1 mM CaC12 (Factor Xa digestion buffer),at a fusion protein concentration of 0.5 mg/ml, unless otherwise indicated.Factor Xa digestion of CBDPT-'PhoA was carried out on both monomeric and dimericforms of the fusion protein. CBDPT-'PhoA monomers were produced by buffer exchange with50 mM KC1-HC1, 1 mM EDTA pH 2.2 in a BIOSPIN-6 gel filtration column (BIO RAD). 350pg of CBDPT-'PhoA in 100 of 50 mM Tris-HC1 pH 7.5, 0.1 mM ZnC12, 0.1 mM MgC12were applied to the column, which had been equilibrated in the low pH buffer. The column wascentrifuged for 4 minutes at 1100 x g in a Sorvall GLC-1 swinging bucket centrifuge, and theflowthrough was incubated for 30 minutes on ice. The low pH buffer was then exchanged with50 mM Tris-HCI pH 8.0, 1 mM EDTA, 0.1 M NaCl by the same method. The A280 and alkalinephosphatase activity of the dissociated fusion protein were measured prior to setting up theprotease digestion. CBDPT-'PhoA was incubated for 18 hours with bovine Factor X a(Boehringer Mannheim), at 1:50 protease:protein ratio in 50 pi volume. The monomeric form ofthe fusion protein was digested in 50 mM Tris-HC1 pH 8.0, 0.1 M NaCl, 1 mM EDTA. CenA-IEGR, a derivative of CenA which has a Factor X a cleavage site between the Pro-Thr linker andthe catalytic domain (Shen Hua, unpublished results), was used as a positive control. Allproteins were also incubated for 18 hours without protease, and a control with Factor Xa onlywas carried out. Digestions were terminated by addition of PMSF to 0.4 mg/ml, and sampleswere taken for SDS PAGE analysis.Factor Xa digestion of CBDPT-PhoA was carried out under the conditions describedabove, except that samples were taken at 6, 12 and 18 hours, and the fusion protein was alsotreated for 18 hours at a protease: protein ratio of 1:10. The Factor X a protease-only control wascarried out at the higher protease concentration. E. coli alkaline phosphatase was also included asa negative control.Factor Xa digestion of CBDPT-PhoA (100 p,g) for use in the affinity tag removalexperiment was for 18 hours at 25°, at a fusion protein concentration of 1 mg/m1 and aprotease:protein ratio of 1:10. The digestion was stopped by addition of PMSF to 0.2 mg/ml.512.12.2 C. fimi protease digestion of alkaline phosphatase fusion proteins.C. fimi protease was prepared from culture supernatants as described (Gilkes et al., 1988)and assayed as described in section 2.5. All C. fimi protease digestions were carried out at 37° in50 mM Tris-HC1 pH 7.5, 0.02% NaN3, with a fusion protein concentration of 0.5 mg/ml and aprotease concentration of 10 U/mg fusion protein. Digestions were terminated by addition ofPMSF to 0.4 mg/ml. Digestions were incubated for up to 18 hours, and samples were taken atthe time points indicated. Recombinant CenA, produced in E. coli, was used as a positivecontrol, and E. coli alkaline phosphatase was used as a negative control. All proteins were alsoincubated for 18 hours without protease, and a control with C. fimi protease only was carried out.Samples of all digestions were taken for SDS PAGE.For use in the affinity tag removal experiment (section 2.10.4), 10014 of CBD-'PhoAwas digested for 18 hours with C. fimi protease as described above. The digestion was stoppedby addition of PMSF to 0.1 mg/ml.2.12.3 Factor Xa digestion of interleukin-2 fusion proteins.Factor Xa digestion of CenA'-IL2 was carried out on Avicel-bound fusion protein purifiedby the direct binding method (section 2.11.2). 10 mg Avicel, to which approximately 10 tg ofCenA'-IL2 were bound, was washed several times with phosphate/EDTA buffer and twice with500 'al Factor Xa digestion buffer. 5 mg of Avicel was incubated with Factor X a, at 1:25protease:protein ratio in 10 volume, for 18 hours at room temperature. The remaining Avicelwas incubated in buffer alone. The positive control, CenA-IEGR (1 mg/ml), was digested withFactor Xa at a 1:50 protease:protein ratio at room temperature for 18 hours, and a sample withoutprotease was incubated in parallel. Samples were mixed with SDS PAGE sample buffer, boiledfor 10 minutes and analyzed by SDS PAGE.52Factor Xa digestion of CBDPT-IL2 was carried out on Avicel-bound fusion proteinpurified by the alkaline solubilization method (section 2.11.5). Approximately 130 gg ofCBDPT-1L2 bound to 100 mg Avicel were first washed with 3x5 ml Factor X a buffer and thendigested with Factor Xa at a protein concentration of 0.15 mg/ml and a 1:30 protease:protein ratio.Digestion was for 18 hours at room temperature with continuous agitation on a vortex mixer, andsamples were also taken at 6 and 12 hours for analysis by SDS PAGE. 10 mg Avicel sampleswere centrifuged and the supernatant was recovered. The Avicel was washed once with 500 IllFactor Xa buffer before resuspension in SDS PAGE sample buffer. Additional controls wereCBDPT-IL2 with no protease, Factor Xa with Avicel but no CBDPT-IL2, and CenA-IEGR insolution (not shown).533. Results3.1 CenA-PhoA fusion proteins.3.1.1 Construction of cenA'-'phoA gene fusions with TnphoA.To determine if the CBD of CenA retained its function when linked to a foreignpolypeptide, a range of fusions between cenA and the gene for alkaline phosphatase (phoA) fromE. coli were generated using TnphoA (Manoil and Beckwith, 1985; Materials and Methods,section 2.4). Transposition of TnphoA in the correct orientation and reading frame into a geneencoding an exported protein, such as CenA, can result in a hybrid protein with alkalinephosphatase activity. Figure 3.1A shows the construction of a cenA'-'phoA fusion by insertionof TnphoA into cenA. The target plasmids used, pUC18-1.6cenA and pTZ18U-CBDC enA (Table2.2), contained the coding regions for CenA and CBDCenA respectively. The latter plasmid wasused to increase the probability of TnphoA insertion within the CBD coding region. Bothplasmids were expressed in the phoA deletion strain E. coli CC118 (Table 2.1), and TnphoA wasdelivered to the cells on a suicide A phage, ATnphoA-1 (Gutierrez et al., 1987). Transposoninsertions into the target plasmid were selected for by inclusion of a high level (300 tg/m1) ofkanamycin in the plating medium, as well as by retransforming plasmids isolated from kanamycinresistant colonies from the initial infection. At each stage colonies expressing alkalinephosphatase activity were identified by including XP in the plating medium (Sarthy et al., 1981).A set of fusions was obtained in which TnphoA was inserted progressively further downstreamof the translational start site in cenA (Figure 3.1B and C). EcoR I restriction analysis revealed thepositions of transposon insertion (Figure 3.1B), which were determined exactly by DNAsequencing. Insertion of TnphoA into cenA effectively disrupted CenA catalytic activity, evenwhen the encoded fusion polypeptide only lacked the 32 C-terminal amino acids of CenA (resultsnot shown). This supports the assertion that the C-terminus of CenA is essential for enzymatic activity (Wong et al., 1986). The isolation of several independent TnphoA insertions at the same54Figure 3.1^Construction of cenA'-'phoA gene fusions.A. Example of TnphoA transposition into cenA to create cenA'-'phoA gene fusion.For XTnphoA-1, dashed lines represent X DNA, and the positions of the Tn5 insertionsequences, 'phoA gene (dark shaded region within IS5OL), and kanamycin resistance gene(kan) are shown. Target plasmid pUC18-1.6cenA contains the complete CenA codingsequence (shaded). The EcoR I sites used for restriction mapping, lac promoter (Plac), andampicillin resistance gene (amp) are also shown.B. 1% agarose gel of EcoR I digested plasmid preparations from cenA'-'phoA fusionclones. Clone designations are: lane 2, D30; lane 3, 33-10; lane 4, #1-37; lane 5, IX-8; lane6, VIII-6; lane 7, C9; lane 8, D31; lane 9, VII-2; lane 10, XII-lb; lane 11, Al. The targetplasmid for 33-10 was pTZ18U-CBDc enA; for all other fusions the target plasmid waspUC18-1.6cenA.Lanes 1 and 12 show EcoR I/Hind III digested X DNA size markers—sizesin kb: 21.2, 5.15, 4.97, 4.27, 3.53, 2.03, 1.90, 1.58, 1.38, 0.95, 0.83, 0.56.C. Positions of TnphoA insertion in cenA. The cenA gene is shown, with translationalstart and stop codons and regions coding for the leader peptide (black), CBD, Pro-Thr linker(PT), and catalytic domain indicated. The cross-hatched box identifies the coding region for 10amino acids from LacZ fused to the N-terminus of CenA. The arrows show positions ofTnphoA insertion into cenA, determined by DNA sequencing, =I the numbers correspond to--lanes in panel B.55pUC18-1.6cenA4.28 kbcenA'-'phoAB.1 2 3 4 5 6 7 8 9 10 11 12IS5OLATG TGAA. UnphoA-1'phoA-1S5OL^ IS50RcenACBD^PTT^12^3 4Catalytic Domain1^TT^1115^6 7^8 9 10^1156location in the coding region for the Pro-Thr linker indicated the presence of a "hot spot" fortransposition. Such hot spots have been reported for Tn5 transposition (Berg et al., 1984) andhave been loosely correlated with the presence of the motif GC(N)6GiC at hot spot insertionpoints; however, this can not be the only determinant of such specificity, as this motif is notpresent at the observed insertion hot spot in cenA, but does occur at 28 other places in the codingsequence.3.1.2 Filter paper binding assay for CenA'-'PhoA fusion proteins.A qualitative cellulose binding assay for CenA'-'PhoA fusion proteins was developed,based on the binding of alkaline phosphatase activity from periplasmic extracts to filter paperdisks (Greenwood et al., 1989; Materials and Methods, section 2.9). Fusion polypeptides couldbind to filter paper if they contained the entire CBD of CenA, but not if they only contained 43(39%) or 83 (75%) N-terminal amino acids of the CBD (Figure 3.2). The Pro-Thr linker isobviously not required for cellulose binding, as a fusion containing only the first two amino acidsof the linker could still bind to cellulose.3.1.3 Avicel binding of CenA'-'PhoA fusion proteins.In addition to the filter paper binding assay described above, the binding of CenA'-'PhoAfusion proteins to microcrystalline cellulose (Avicel) was investigated. Periplasmic extracts fromcells treated with chloroform were passed through small Avicel columns (Materials and Methods,section 2.9) and the flowthrough and Avicel-bound fractions were analyzed by SDS PAGE withstaining for alkaline phosphatase activity (Figure 3.3, Materials and Methods, section 2.8). Therewas a significant level of proteolysis of most of the fusion proteins, and in some cases activitybands representing the intact fusion protein were not visible (Figure 3.3, lanes B8-13, C2-4).However, where intact fusion polypeptides were present, the Avicel binding behaviour matchedthat with filter paper, i.e. fusion polypeptides containing a complete CBD bound to Avicel and57Figure 3.2^Binding of CenA'-'PhoA fusion proteins to filter paper.Periplasmic extracts from cells expressing cenA'-'phoA gene fusions were applied toeach of two filter paper disks, and the disks were stained for alkaline phosphatase activityeither before or after extensive buffer washing (Materials and Methods, section 2.9). MatureCenA is shown, with the CBD (hatched box), Pro-Thr linker (PT), and catalytic domain (openbox) indicated. The first and last amino acids of CenA and the Pro-Thr linker are numbered.The arrows indicate the junctions of CenA with PhoA for the different fusions, as determinedby DNA sequencing. Control disks shown were incubated with purified E. coli alkalinephosphatase or a periplasmic extract from the PhoA - host, E. coli CC118.58Figure 3.3^Binding of CenA'-'PhoA fusion proteins to Avicel.Periplasmic extracts from cells expressing cenA'-'phoA gene fusions were applied tosmall columns containing 10 mg of Avicel; the columns were washed with buffer and boundproteins were eluted with SDS PAGE sample buffer (Materials and Methods, section 2.9).Samples of periplasmic extract, column flowthrough, and Avicel-bound fractions wereelectrophoresed through 10% SDS polyacrylamide gels. The gels were stained for alkalinephosphatase activity (Materials and Methods, section 2.8). The gels were loaded in order ofincreasing fusion protein Mr, and the three lanes for each clone contained (from left to right)periplasmic extract, column flowthrough, and Avicel-bound fractions. Lanes: 1 (panels A-C),prestained molecular weight standards; A2-4, D30; A5-7, 33-10; A8-10, #1-37; A11-13, IX-8;B2-4, VIII-6; B5-7, C9; B8-10, D31; B11-13, VII-2; C2-4, XII-lb; C5-7, Al; C8-10,purified E. coli alkaline phosphatase control; C11-13, E. coli CC118 control. Equivalentvolumes were loaded in all lanes. The correspondence of clone designation to TnphoAinsertion position in cenA is given in Figure 3.1.59A. 1 2 3 4 5 6 7 8 9 10 11 12 13106-80-49.5-32.5-B.1 2 3 4 5 6 7 8 9 10 11 12 13106-80-49.5— 1111 US OP.32.5-C-1 2 3 4 5 6 7 8 9 10 11 12 13106-80-49.5-32.5-60those containing only a portion of the CBD did not. It was significant that the binding of fusionpolypeptides to Avicel occurred very rapidly; at a flow rate of 1.1 ml/min the fluid residence timein the small columns was on the order of 1-2 seconds. Both CenA and CBDPTcenA (CBDcenAplus the Pro-Thr linker) have been reported to reach binding equilibrium on bacterialmicrocrystalline cellulose within 12 seconds (Gilkes et al., 1992).No proteolysis products were evident for fusion polypeptide #1-37 (Figure 3.3, lanes A8-10). The fusion point for #1-37 was within the Pro-Thr linker, between the cellulose binding andcatalytic domains; the fusion points for all the other polypeptides were within one or otherdomain, potentially leading to incomplete or incorrect polypeptide folding and proteolyticsensitivity.For the E. coli alkaline phosphatase control, a small fraction of the activity was Avicel-associated (Figure 3.3, lane C10); this was most likely due to incomplete washing of the Avicel,caused by entrapment of air bubbles in the Avicel prior to the washing step. Batchwise bindingexperiments have shown that E. coli alkaline phosphatase does not bind to Avicel (Greenwood etal., 1989).3.1.4 CBDCenA N-terminal deletions: construction of pTZ18R-CBDPT-'phoAA1and pTZ18R-CBDPT-'phoAA2.The TnphoA mutagenesis approach generated two CBDc eoc 'PhoA fusions withdeletions from the C-terminus of the CBD. Two additional CenA'-'PhoA fusions were madewith N-terminal deletions in the CBD to determine the effect of such deletions on cellulosebinding. pTZ18U-CBDPT-'phoA (construction described in section 3.1.9) and pTZ18R-cenA-LNhe (Table 2.2) were the parent plasmids used for these deletion constructs. pTZ18U-CBDPT-'phoA encoded the CBD and Pro-Thr linker of CenA fused through a Factor X a cleavage site toPro-6 of mature alkaline phosphatase. pTZ18R-cenA-LNhe contained the cenA gene with anengineered Nhe I site immediately following the coding region for the leader peptidase cleavagesite. Construction of the deletions utilized the Nhe I site of pTZ18R-cenA-LNhe and two61Figure 3.4^Construction of pTZ18R-CBDPT-'phoAAl and pTZ18R-CBDPT-'phoAA2.pTZ18R-cenA-LNhe was digested with Nhe I and the ends were filled in with theKlenow fragment of DNA polymerase I (kPol). Subsequent digestion with Pst I released a2.93 kb vector fragment. CBDPT-'phoA insert fragments were generated by digestingpTZ18U-CBDPT-'phoA, i) with Nar I (filled in with kPol) and Pst I, and ii) with BamH I(filled in with kPol) and Pst I. Ligation of the vector and insert fragments regenerated the NheI site for pTZ18R-CBDPT-'phoAA1 and the BamH I site for pTZ18R-CBDPT-'phoAA2, andfused the truncated CBD coding sequences in frame with the intact CenA leader peptide codingsequence. The positions of the plasmid (ori) and filamentous phage (fl ori) replication originsand ampicillin resistance gene (amp) are indicated by light filled boxes, and the leader peptidecoding sequence of CenA is indicated by darker filled boxes. The orientation of cenA andCBDPT-'phoA gene fusions are indicated by arrows. The lac _promoter (Plac), truncated lacZ'gene and relevant restriction sites are also shown.62Nar IkPol fill inPst I\BamH IkPol fill inPst IPst I Pst INhe IkPol fill inPst Ilac%Pst IPst IPst IPlacpTZ18R-CBDPT- 'phoA625.76 kb0pTZ18R-CBDPT-'pha4015.83 kbT4 DNALigaseBamH I00121T4 DNALigase Nhe IBamH I63convenient restriction endonuclease cleavage sites within the CBD coding region of pTZ18U-CBDPT-'phoA (Figure 3.4). Plasmid was isolated from ampicillin resistant, PhoA+transformants of E. coli CC118, and was screened by restriction endonuclease digestion. TheDNA sequence across the deleted region was confirmed for both constructs. pTZ18R-CBDPT-phoAA I encoded a fusion polypeptide (CBDPT-'PhoAAl) lacking amino acids 2-19 of matureCenA, and the polypeptide (CBDPT-PhoAA2) encoded by pTZ18R-CBDPT-'phoAA2 had adeletion of amino acids 2-45. Both the encoded polypeptides had an additional amino acid at thesite of deletion, a serine for CBDPT-'PhoAA1 and an arginine for CBDPT-'PhoAA2.3.1.5 Cellulose binding of CBDPT-IPhoAA1 and CBDPT-'PhoAA2.Periplasmic extracts containing CBDPT-'PhoAAl and CBDPT-'PhoAA2 were assayedfor binding to filter paper and to Avicel (Materials and Methods, section 2.9). Neitherpolypeptide bound to filter paper (results not shown), nor was there any significant binding toAvicel (Figure 3.5). CBDPT-'PhoAAl was subject to proteolysis (Figure 3.5, lanes 2 and 3),with a major band of activity running at a similar position to the CBDPT-'PhoAA2 polypeptide.The parent polypeptide, CBDPT-'PhoA, was completely bound to Avicel (lanes 8-10).3.1.6 CenA'-'PhoA IX-8 heterodimers: binding to Avicel.It was evident from Avicel binding studies on periplasmic extracts containing CenA'-'PhoA fusion polypeptide IX-8 (Figure 3.3, lanes A11-13) that some proteolysis products boundto Avicel along with the intact fusion polypeptide. This was investigated further using clarifiedcell extracts of E. coli CC118/pIX-8A. Plasmid pIX-8A was constructed from pUC18-1.6cenA::TnphoA IX-8 by deleting 2.8 kb and 2.0 kb Bgl II fragments containing thetranspositionally active IS5OR sequence and religating the resulting 7.2 kb fragment. Thisreduced the plasmid size significantly and protected the plasmid from further transposition events.The encoded fusion polypeptide, CenA'-'PhoA IX-8A, contained the CBD, Pro-Thr linker, and64Figure 3.5^Binding of CBDPT-'PhoAAl and CBDPT-'PhoAA2 fusion proteins to Avicel.Periplasmic extracts from cells expressing CBDPT-'phoAA1 and CBDPT-phoAA2gene fusions were applied to small Avicel columns, as described in Materials and Methods,section 2.9, and the legend to Figure 3.3. Samples of periplasmic extract, columnflowthrough, and Avicel-bound fractions were electrophoresed through a 10% SDSpolyacrylamide gel; the gel was stained for alkaline phosphatase activity (Materials andMethods, section 2.8). The three lanes for each fusion contained (from left to right)periplasmic extract, column flowthrough, and Avicel-bound fractions. Lanes: 1, prestainedmolecular weight standards; 2-4, CBDPT-ThoAA1; 5-7, CBDPT-'PhoAA2; 8-10, purifiedCBDPT-ThoA control; 11-13, E. coli CC118 control. Equivalent volumes were loaded in all lanes.65CenA111 134^418PT Catalytic Domain CenA'-'PhoA IX-8111 134 174 191^636II 170^ 175^ 180I ' - AL T P Q A DS Y T QATC GCG CTC ACC CCG CAG GCT GAC TCT TAT ACA CAAcenA^TnphoAFigure 3.6^Schematic diagram of fusion polypeptide CenA'-'PhoA IX-8.The mature polypeptide is shown, with the CBD (hatched box), Pro-Thr linker (PT),catalytic domain (open box), Tn5-encoded amino acids (black box) and 'PhoA domain (shadedbox) indicated. Mature CenA is shown for comparison. Numbers refer to amino acid residuesat the mature N-terminus of each polypeptide and the C-terminus of each domain structure.The DNA sequence and deduced amino acid sequence of the fusion junction in CenA'-'PhoAIX-8 are shown. cenA DNA is underlined and amino acid residues are numbered. Theposition of cleavage to generate the 51 kDa polypeptide is indicated by a vertical arrow.6640 amino acids of the catalytic domain of CenA, with the fusion point after Gln-174 of matureCenA (Figure 3.6).Avicel-bound polypeptides from a clarified cell extract of E. coli CC118/pIX-8A wereseparated by SDS PAGE and stained with Coomassie blue (Figure 3.7). N-terminal amino acidsequence analysis showed that the 74 kDa and 24 kDa polypeptides both had the N-terminalsequence of mature CenA, and therefore represented the intact fusion protein and a N-terminalfragment containing CBDc enA respectively. The 51 kDa polypeptide was probably thecorresponding C-terminal 'PhoA fragment, as it had the N-terminal sequence Ala-Leu-Thr-Pro-Gln, identical to amino acids 170-174 of CenA'-'PhoA IX-8 (Figure 3.6), and its Mr was closeto that expected for such a fragment (Table 3.1). A band of similar M r was also observed withalkaline phosphatase activity in Figure 3.3 (lanes A11-13). The 51 kDa fragment probably aroseby proteolysis of CenA'-'PhoA IX-8 between amino acids 169 and 170, within the CenAcatalytic domain segment. E. coli alkaline phosphatase is a dimeric enzyme which does not bindto cellulose (Figures 3.2, 3.3, and Greenwood et al., 1989). Since the 51 kDa polypeptidelacked a CBD, it must have bound to Avicel by forming a dimer with an intact CenA'-'PhoA IX-8fusion polypeptide.N-terminal sequencing of the 49 and 44 kDa polypeptides yielded ambiguous sequencedata; however, it is likely that the 49 kDa polypeptide was formed by further truncation of the 51kDa polypeptide, as a corresponding band exhibiting alkaline phosphatase activity was observedin Figure 3.3 (lanes A11-13). The origin of the 44 kDa polypeptide was unclear, but theappearance of a similar band in Avicel bound fractions from various E. coli periplasmic extractssuggested that this polypeptide was an E. coli protein which bound non-specifically to cellulose(Greenwood et al., 1989 and results not shown).The apparent molecular masses of the 74 and 24 kDa polypeptides were about 7 kDagreater than the molecular masses predicted from amino acid sequence data (Table 3.1). There ismounting evidence that the Pro-Thr linker adopts an extended structure during SDS PAGE whichcauses this anomalous migration (Gilkes et al., 1989 and 1992; Shen et al., 1991). The 51 kDapolypeptide, which did not contain a Pro-Thr linker, had an apparent molecular mass similar to67Figure 3.7^Binding of cenA'-'phoA IX-8-encoded polypeptides to Avicel.E. coli CC118/pIX-8A was grown overnight at 30° in LB amp medium. Harvestedcells were resuspended in 50 mM Tris-HC1 pH 7.5, 0.02% NaN3 prior to rupture with aFrench press. PMSF and pepstatin A were added immediately following rupture, and nucleicacids were precipitated with streptomycin sulfate and removed by centrifugation. The clarifiedcell extract was incubated with Avicel (Materials and Methods, section 2.9). The Avicel waswashed once with 50 mM Tris-HC1, 1 M NaC1, pH 7.5 and twice with 50 mM Tris-HC1 pH7.5. Bound polypeptides were electrophoresed on a 10% SDS polyacrylamide gel andelectroblotted onto Immobilon-P. Lanes 1 and 2 show different loadings of the same sample,stained with Coomassie blue. Sizes of major bands are shown in kilodaltons. UnambiguousN-terminal amino acid sequences were obtained for bands marked with an asterisk.68Polypeptide Total aminoacid residuesbApparent^Predictedmolecular masse molecular massd(kDa)^(kDa)Apparent minuspredictedmolecular mass(lcDa)CenA'-'PhoA IX-8 636 (639)e 74 66.4 (66.7) 7.6 (7.3)'PhoA 467 51 49.0 2.0CenA' 169 (172) 24 17.4 (17.7) 6.6 (6.3)Table 3.1^Molecular mass determinations for CenA'-'PhoA IX-8 polypeptides.aa Molecular mass data for monomer forms of dimeric polypeptides.b From the deduced amino acid sequence.c Determined by SDS PAGE.d Calculated from the deduced amino acid sequence.e Values in parentheses refer to N-terminal variant of the polypeptide (Guo et al., 1988;Gilkes et al., 1988).69that predicted from amino acid sequence data (Table 3.1). Consistent with previous studies in E.coli (Guo et al., 1988; Gilkes et al., 1988), the N-terminal sequences of the 74 and 24 kDapolypeptides indicated that the CenA leader peptide had been processed at two related sites threeamino acids apart.3.1.7 CenA'-'PhoA^heterodimers: heterodimer dissociation.To further study the binding of CenA'-'PhoA IX-8 heterodimers to cellulose, the dimericalkaline phosphatase portion of the fusion was dissociated by low pH (Schlesinger and Barrett,1965; Materials and Methods, section 2.10.1) either before or after binding to Avicel. Figure 3.8shows SDS PAGE analysis of Avicel supernatants and extracts following i) incubation of aclarified cell extract of E. coli CC118/pIX-8A with Avicel under standard binding conditions; ii)Avicel binding of the clarified cell extract followed by buffer washing at pH 2.0; and iii)acidification of the clarified cell extract prior to Avicel binding. Under standard binding andwashing conditions the intact fusion protein and a small proportion of the 51 and 49 kDapolypeptides bound to Avicel (Figure 3.8, lanes 2-4). The majority of the 51 and 49 kDapolypeptides remained in the Avicel supernatant, and probably were 'PhoA/'PhoA homodimerswith no affinity for cellulose. When the Avicel was washed with buffer at pH 2.0, all of theproteolytic alkaline phosphatase fragments were removed from the Avicel while most of the intactfusion protein remained bound (lanes 5 and 6), consistent with the dissociation of Avicel-boundCenA'-'PhoA/'PhoA heterodimers. It was not known whether the intact fusion polypeptide thatwas eluted from Avicel was part of the heterodimer or homodimer form of the fusion, but itseemed likely that under conditions favouring dissociation of alkaline phosphatase dimers themain determinant of elution would be the heterogeneity of CBD binding to Avicel.When the clarified cell extract of E. coli CC118/pIX-8A was preincubated at pH 2.0,many of the cellular proteins precipitated but the CenA'-'PhoA IX-8 fusion protein and relatedproteolytic fragments remained in solution (Figure 3.8, lane 7). When the acidified cell extractwas incubated with Avicel, the majority of the intact fusion protein bound, but none of the 'PhoA70Figure 3.8^Dissociation of CenA'-'PhoA IX-8 heterodimers.Avicel binding assays were carried out with a clarified cell extract of E. coli CC118/pIX-8A, i) under standard binding and washing conditions;^under standard bindingconditions followed by buffer washing at pH 2.0; and iii) with acidification of the clarified cellextract prior to Avicel binding (Materials and Methods, section 2.10.1). Samples of cellextract, Avicel supernatant and Avicel extract fractions were electrophoresed through two 10%SDS polyacrylamide gels. One gel was stained with Coomassie blue (panel A) and the othergel was stained for alkaline phosphatase activity (panel B; Materials and Methods, section 2.8).Lanes: 1, molecular weight standards (prestained for panel B); 2, E. coli CC118/p1X-8Aclarified cell extract 3, Avicel supernatant; 4, Avicel extract; 5, Avicel pH 2.0 wash; 6, Avicelextract after pH 2.0 wash; 7, acidified cell extract; 8, Avicel supernatant from acidified cellextract; 9, Avicel extract from acidified cell extract. Equivalent volumes were loaded in alllanes.71proteolytic fragments were Avicel-associated (lanes 8 and 9). This was also consistent with the51 and 49 kDa polypeptides having no intrinsic affinity for cellulose and being unable to bindwhen dissociated from an intact CenA'-'PhoA polypeptide. The ability of CBDCenA to bind toAvicel at pH 2.0, albeit with reduced affinity, was striking, and suggests that protonation of thethree acidic residues in the CBD does not curtail its function to any significant extent. Figure 3.8also conveys the significant degree of purification of CenA'-'PhoA IX-8 achieved in a singleaffinity step on Avicel.3.1.8 CenA'-'PhoA IX-8 heterodimers: association of heterodimers on cellulose.CBD affinity tags could potentially be used as tools for probing protein-proteininteractions by facilitating the purification of multisubunit proteins and protein complexes. TheCenA'-'PhoA IX-8 heterodimer system was used to address this possibility, by reconstitutingCenA'-'PhoA/PhoA heterodimers bound to Avicel (Greenwood et al., 1992). The acid-induceddissociation of alkaline phosphatase into monomer subunits involves the loss of the coordinatedzinc ions, and is reversible on increasing the pH, but reformation of the dimer at neutral pH canbe prevented by chelation of zinc ions with EDTA (Schlesinger and Barrett, 1965). CenA'-'PhoA IX-8 monomers bound to Avicel were prepared by washing the Avicel-bound fusionpolypeptide at pH 2.0 and increasing the pH to 7.5 in the presence of 5 mM EDTA (Materials andMethods, section 2.10.1). The immobilized fusion protein monomer was incubated withsimilarly prepared monomers of E. coli alkaline phosphatase, either under conditions whichfavoured dimerization (pH 7.5, excess zinc) or prevented dimerization (pH 7.5, excess EDTA).The polypeptides which bound to Avicel were subsequently analyzed by SDS PAGE (Figure3.9). When zinc was present, reassociation of alkaline phosphatase subunits took place, and thepresence of a 48 kDa alkaline phosphatase polypeptide together with the 74 kDa CenA'-'PhoApolypeptide showed that CenA'-'PhoA/PhoA heterodimers had been formed which bound toAvicel. When EDTA was present in excess, no heterodimer formation was observed. Increasingthe amount of added alkaline phosphatase monomer increased the yield of bound heterodimer.72A.B.1^4 5 6106 —80 —49.532.5 —Figure 3.9^Formation of CenA'-PhoA/PhoA heterodimers on Avicel.Acid-dissociated CenA'-'PhoA IX-8 bound to Avicel was resuspended in buffercontaining acid-dissociated alkaline phosphatase monomer subunits or buffer alone (control).Duplicate samples were incubated with or without 10 mM ZnC12 (Materials and Methods,section 2.10.1). Polypeptides binding to Avicel were extracted into SDS PAGE sample bufferand electrophoresed through two 10% SDS polyacrylamide gels. One gel was stained withCoomassie blue (panel A) and the other gel was stained for alkaline phosphatase activity (panelB; Materials and Methods, section 2.8). Lanes: 1, molecular weight standards (prestained forpanel B); 2 and 3, incubation with 50 mM Tris-HC1, 5 mM EDTA pH 7.5 (lane 2, no ZnC12;lane 3, 10 mM ZnC12); 4-7, incubation with acid-dissociated alkaline phosphatase (lanes 4 and6, no ZnC12; lanes 5 and 7, 10 mM ZnC12). Approximate molar ratios of PhoA:CenA'-'PhoAwere 2.5:1 (lanes 4 and 5) and 1:1 (lanes 6 and 7). Equivalent volumes were loaded in alllanes.73It is not surprising that an excess of alkaline phosphatase was necessary for significantheterodimer formation, as there would be direct competition from PhoA/PhoA homodimerformation in solution and possibly also CenA'-'PhoA/CenA'-'PhoA homodimer formation onAvicel.3.1.9 Precise fusion of CBDPTCenA to alkaline phosphatase: construction ofpTZ18U-CBDPT-'phoA and pTUg10*-CBDPT-'phoA.For detailed studies on fusion protein purification and affinity tag removal, a precisefusion between the sequence coding for CBDPTCenA and the 'phoA gene from TnphoA wasgenerated, as outlined in Figure 3.10. pTZ18R-cenA'-'phoA #1-370, the plasmid used for theconstruction, was a derivative of pUC18-1.6cenA::TnphoA #1-37 (Figure 3.1), from which a3.55 kb Sac I/Hind III fragment containing the cenA'-'phoA gene fusion was subcloned into SacI/Hind III-digested pTZ18R. This facilitated the production of single strand plasmid DNA foroligonucleotide-directed mutagenesis (Materials and Methods, section 2.4). A Stu I site wasintroduced at the 5'-end of the 'phoA gene in pTZ18R-cenA'-'phoA #1-37A (Figure 3.10).Plasmid from transformants of E. coli CC118 expressing alkaline phosphatase activity wasscreened by Stu I digestion and the incorporation of the Stu I site in pTZ18R-#1-370Stu I wasconfirmed by DNA sequencing. Construction of the Stu I site allowed the subcloning of a StuI/Hind III 'phoA fragment into pTZ18U-CBDPT-IEGR, a plasmid which coded for CBDPTCenAfused to a Factor Xa cleavage site. Plasmid from transformants of E. coli CC118 expressingalkaline phosphatase activity was screened by restriction endonuclease digestion and the DNAsequence across the fusion junction was confirmed. The resulting plasmid, pTZ18U-CBDPT-'phoA, coded for the cellulose-binding domain and Pro-Thr linker of CenA fused through aFactor Xa cleavage site to Pro-5 of mature alkaline phosphatase. The predicted molecular mass ofthe encoded fusion protein, CBDPT-'PhoA, was 60.5 kDa.The E. coli expression vector pTUglO*N18, a derivative of pTrc99A (Amann et al.,1988), was developed in our laboratory for high level expression in E. coli of genes and gene74Figure 3.10 Construction of pTZ18U-CBDPT-'phoA.A Stu I site was introduced into pTZ18R-cenA'-'phoA #1-37A, at the junction betweenthe Tn5 IS5OL repeat sequence and the truncated 'phoA gene, by oligonucleotide-directedmutagenesis (Materials and Methods, section 2.4). The mutagenic oligonucleotide (3 lmer) isshown hybridized to the single stranded template strand, with the two base mismatch creatingthe Stu I site indicated. The 5'-end of the 'phoA gene is shown underlined and the direction ofprimer extension is indicated by the arrow. The resulting plasmid, pTZ18R-#1-37AStu I, wasdigested with Stu I and Hind III to produce a 3.06 kb 'phoA fragment . This was ligated intopTZ18U-CBDPT-IEGR, which had been digested with the same restriction enzymes, toproduce pTZ18U-CBDPT-'phoA. The nucleotide sequence and corresponding amino acidsequence across the fusion junction is shown, with the N-terminal amino acids of alkalinephosphatase underlined. Amino acids are numbered relative to the mature N-termini of CenAand alkaline phosphatase. The Stu I site used in constructing pTZ18U-CBDPT-'phoA isshown both underlined and in bold type. The positions of the plasmid (ori) and filamentousphage (fl ori) replication origins and ampicillin resistance gene (amp) are indicated by lightfilled boxes, and the leader peptide coding sequence of CenA is indicated by darker filledboxes. The orientation of CBDCenA and cenA'-'phoA gene fusions are indicated by arrows.The lac promoter (Plac), truncated lacZ' gene and relevant restriction sites are also shown.75Primer extensionT4 DNA ligaseprimer^Stu I 5AA' CCT TTC CCG TTT IAG G CCT1 GTT CTG GAA AA3 --II.. .0 CTT GGA AAG GGC AAA ACG GGA CAA GAC CTT TTG G...templatepTZ18U-CBDPT-IEGR3.39 kbii, / T4 DNA ligase132 133^ 5 6^7Thr Pro Ile Glu Gly Arg Pro Val Leu ACG CCG ATC GAG GGC AGG CCT GTT CTGStu IHind IIIStu IHind IIIHind IIIStu IHind IIIStu IHind IIIStu IHind III76fusions containing cenA translational start signals (R.W. Graham and D.E. Trimbur, inpreparation). pTUg10*N18 contains the tac promoter, the leader and Shine-Dalgarno sequencesfrom bacteriophage T7 gene 10 (Rosenberg et al., 1987), and the pUC18 polylinker, with theShine-Dalgarno sequence fused directly to the Sac I site of the polylinker. This permitssubcloning of Sac I-digested cenA gene fragments into the plasmid and utilization of the cenAtranslational start codon (Figure 3.11). The plasmid also carries transcriptional terminators fromthe ribosomal RNA rrnB operon, and the /ac/q allele of the lac repressor gene (Calos, 1978), totightly regulate transcription from the tac promoter. The CBDPT-'phoA gene fusion frompTZ18U-CBDPT-'phoA was subcloned into pTUg10*N18 on a 3.57 kb Sac I/Hind III fragment(Figure 3.11). Plasmid from transformants of E. coli CC118 expressing alkaline phosphataseactivity were screened by restriction endonuclease digestion. The resulting plasmid, pTUg10*-CBDPT-'phoA, utilized the cenA ATG, unlike pUC and pTZ vectors which coded for N-terminalfusions of the CenA leader peptide with amino acids from LacZ (Guo et al., 1988). Expressionof CBDPT-'phoA in this vector yielded a polypeptide which bound to cellulose, both in a filterpaper binding assay (not shown) and an Avicel binding assay (Figure 3.5).3.1.10 High level production of CBDPT-'PhoA.TYP broth was used for growth of E. coli CC118/pTUg10*-CBDPT-'phoA as morefusion protein was produced in this medium than in LB medium (results not shown). Thekinetics of fusion protein production from pTZ18U-CBDPT-'phoA and pTUg10*-CBDPT-phoA were compared under identical growth and induction conditions (Figure 3.12). Bothcultures accumulated alkaline phosphatase activity over extended growth periods and leakedsignificant amounts of activity into the growth medium; however, E. coli CC118/pTUg10*-CBDPT-'phoA yielded almost three times the activity in the culture supernatant as the straincarrying pTZ18U-CBDPT-'phoA, and periplasmic levels in the former culture were also greater.The more rapid onset of fusion protein accumulation for CC118/pTUg10*-CBDPT-'phoAreflected the difference in promoter strength between the tac and lac promoters. The highest77Figure 3.11 Construction of pTUg10*-CBDPT-'phoA.pTZ18U-CBDPT-'phoA was digested with Sac I and Hind III to isolate a 3.57 kbCBDPT-'phoA fragment. This was ligated into pTUglO*N18 which had also been digestedwith Sac I and Hind III. The DNA sequence around the Sac I site in pTUg10*-CBDPT-'phoAis shown, with the Sac I site underlined and the Shine-Dalgarno sequence in bold type. Theinitiation and subsequent codons of cenA are shown together with the encoded amino acids.The positions of the plasmid (ori) and phage (fl ori) replication origins, ampicillin resistancegene (amp), and laclq gene (laclq) are indicated by light filled boxes, and the leader peptidecoding sequence of CenA is indicated by darker filled boxes. The orientation of the CBDPT-phoA gene fusion is indicated by arrows. The lac (Plac) and tac (Ptac) promoters, T7 gene 10leader (g10), transcriptional terminators (rrnB T1T2), truncated lacZ' gene, and relevantrestriction sites are also shown.78sio Sac IPtio^Hind IIISac IHind IIISac IHind IIIT4 DNA ligasemet ser^thrAAGAAGGAGCTCCTTG ATG TCC ACCSac I97 Sac IHind IIISac IHind III796.05.0 —4.0 ——0— pTZ18U — periplasm- pTZ18U — culture supernatant—0— pTUg10* — periplasm- pTUg10* — culture supernatantAlkaline^3.0 -phosphataseactivity (Wm!)2.0 —1.0 —0. 00^5^10^15^20^25^30^35^40^45^50Time after induction (hours)Figure 3.12 Expression of CBDPT-'phoA in E. coli CC118.0.5 ml inocula from cultures of E. coli CC118/pTZ18U-CBDPT-'phoA and /pTUg10*-CBDPT-'phoA grown at 30° in TYP amp broth to A600 = 1.5 were added to 50 ml TYP ampbroth. The cultures were incubated at 30°and 250 rpm, and IPTG was added to 0.1 mM atAgo = 0.8. At the times indicated, 1 ml samples were removed and centrifuged to separate thecells from the culture supernatant. Periplasmic proteins were extracted from the cells withchloroform (Ferro-Luzzi Ames et al., 1984) Alkaline phosphatase activities of the culturesupernatant and periplasmic fractions were determined (Materials and Methods, section 2.5) as m ni s^o culture..^S .180alkaline phosphatase activity levels in the periplasm and culture supernatant combinedcorresponded to approximately 300 mg/1 of fusion protein. The accumulation of enzyme activityin the culture supernatant may have been partly due to cell lysis, but the maintained periplasmiclevels of activity suggested that much of the activity was leaked from intact cells. The leakage ofCenA and Cex from E. coli have been reported previously (Guo et al., 1988).3.1.11 Purification of CBDPT-'PhoA: single column loading and elution withguanidine-HC1 or distilled water.CBDPT-'PhoA was purified from culture supernatants of E. coli CC118/pTUg10*-CBDPT-'phoA by affinity chromatography on CF1 cellulose, using a method similar to thatreported by Shen et al. (1991), in which the column was washed with 50 mM Tris-HC1, 1 MNaCl pH 7.5, 0.02% NaN3 (high salt buffer) and 50 mM Tris-HC1 pH 7.5, 0.02% NaN3 (lowsalt buffer) and the bound protein was eluted with guanidine-HC1 (Materials and Methods, section2.10.2). Guanidine-HC1 has been reported to reversibly dissociate alkaline phosphatase intomonomer subunits (Schlesinger and Barrett, 1965), so removal of the guanidine-HC1 bydiafiltration was expected to allow reassociation of CBDPT-'PhoA monomers to reform the activedimeric species. Diafiltration was carried out at room temperature to facilitate the renaturation ofthe fusion polypeptide. The purification profile for this experiment is shown in Figure 3.13. Amajor A280 peak representing the fusion protein was eluted from the column with guanidine-HC1;the absorbance started to increase as soon as the guanidine-HC1 gradient was initiated and theabsorbance peak coincided with a guanidine-HC1 concentration of approximately 3 M. Thepooled elution peak fractions (fractions 51-68) were passed through a 0.451.tm membrane filter,as this was found to reduce protein precipitation during the subsequent diafiltration andconcentration steps. The final purified protein had a specific activity of 26 U/mg, whichrepresented a 3.5-fold purification factor, and the purification yield was 84% (Table 3.2). Thespecific activity corresponded to a value of 33 U/mg for the alkaline phosphatase portion of the fusion protein. This was somewhat lower than reported values for purified E. coli alkaline81—•-• A280—1:1— Conductivity---- [Guanidine-HCI]... ..ameo•o".orriritilirtiviritool^I-I-r-rt-rrAt10^20^30^40^50Fraction No.60^70^80^1.00 ^0.90 -0.80 -0.70 -0.60 -A2800.50 -0.40 -10.30 -0.20 -0.10 -0.00 ^0100- 90- 80- 7060- 50^-Conductivity_(mmho)- 40- 30- 20- 1010.^(M)7 4.0- 3.012.0-1 1. 00^_ 0.0Figure 3.13 Purification profile for CBDPT-'PhoA: single column loading with guanidine-HC1 elution.CBDPT-'PhoA was purified from a culture supernatant of E. coli CC118/pTUg10*-CBDPT-'phoA (Materials and Methods, section 2.10.2). The fractions were divided intocolumn flowthrough (fractions 5-12), high salt wash (fractions 13-34), low salt wash(fractions 35-50), and elution peak (fractions 51-68) pools. The guanidine-HC1 concentrationwas estimated from the conductivity readings using a standard curve.82Purification Volume PhoA Total Protein Total Specific Yield PurificationStep (ml) Activity Activity (mg/ml) Protein Activity (%) Factor(U/ml) (U) (mg) (U/mg)Culturesupernatant90 6.7 600 0.90 81 7.4 100G-HCl elution, 1.15 440 510 17.0 19.6 26 84 3.5OF retentateaCF1 columnflowthrough120 0.0445 5.34 0.89High salt wash 330 0.003 0.99 0.16Low salt wash 240 0.0015 0.36 0.06Table 3.2^Purification of CBDPT-'PhoA from E. coli CC118/pTUg10*-CBDPT-'phoAculture supernatant: single column loading with guanidine-HC1 elution.a After final centrifugation.83phosphatase in the range of 34-59 U/mg (Garen and Levinthal, 1960; Rothman and Byrne, 1963;Plocke et al., 1962; Applebury and Coleman, 1969; McCracken and Meighen, 1980; Schlesingerand Olsen, 1970; Csopak et al., 1972a; Harris and Coleman, 1968; Malamy and Horecker, 1964;Roberts and Chlebowski, 1984). For a single purification step, however, the cellulose affinitypurification was very effective, and the 84% purification yield was comparable to or better thanany published values. Less than 1% of the activity loaded onto the column appeared in theflowthrough, and the activity losses during the wash steps were minimal. The small A280 peakappearing during the initiation of the low salt wash (Figure 3.13, fractions 34-43) probablyrepresented the elution of non-specifically bound protein from the column. Analysis of theculture supernatant and purified polypeptide by SDS PAGE revealed CBDPT-'PhoA as a 68 kDaband, together with a minor 57 kDa degradation product which retained alkaline phosphataseactivity (Figure 3.14,lanes 2 and 3). The Mr of the intact fusion polypeptide was 7500 more thanthe predicted value of 60,500, a discrepancy which could be attributed to the presence of the Pro-Thr linker (Gilkes et al., 1989; Shen et al., 1991).Preliminary purification studies indicated that some CenA'-'PhoA fusion proteins couldbe eluted from cellulose with distilled water (Greenwood et al., 1989, and results not shown), ashas been reported for C. fimi Cex, CenB, endoglucanase C (CenC), and endoglucanase D(CenD) (Gilkes et al., 1988; Owolabi et al., 1988; Coutinho, 1992a; Meinke et al., 1993). Thisproperty also applied to CBDPT-'PhoA. The fusion polypeptide was purified from a culturesupernatant of E. coli CC118/pTUg10*-CBDPT-'phoA by CF1 cellulose affinitychromatography, under the same loading and washing conditions as described above, but withdistilled water elution (Materials and Methods, section 2.10.2). The purification profile (Figure3.15) shows an A280 peak eluting from the column with distilled water. This peak coincided withan alkaline phosphatase activity peak, confirming the elution of CBDPT-'PhoA from CF1cellulose at low ionic strength. The conductivity at which the peak started to appear correspondedto a Tris concentration of about 10 mM. The water elution peak was smaller than thecorresponding guanidine-HC1 elution peak from Figure 3.13, and subsequent washing of thecolumn with guanidine-HC1 eluted a further A280 peak. The yield of activity from the water84A.B.Figure 3.14 Purification of CBDPT-'PhoA: SDS PAGE analysis.CBDPT-'PhoA was purified from the culture supernatant (lanes 2-6) and clarified cellextract (lanes 7 and 8) of E. coli CC118/pTUg10*-CBDPT-iphoA, as described in the text.Samples corresponding to 2.5 lig CBDPT-PhoA were electrophoresed through two 12.5%SDS polyacrylamide gels. One gel was stained with Coomassie blue (panel A) and the othergel was stained for alkaline phosphatase activity (panel B; Materials and Methods, section 2.8).Lanes: 1, molecular weight standards; 2, culture supernatant; 3, single column loading-guanidine-HC1 elution; 4, single column loading—water elution; 5, single column loading-guanidine-HC1 elution following water elution; 6, multiple column loading—water elution; 7,clarified cell extract; 8, multiple column loading—water elution.850.40 -00.30 -1 01 .00.00 0. 0^o10. PhoAActivity5.0 (U/mI)900^10^20^30^40^50^60Fraction No.70^80201.00 ^0.900.80 -0.70 -—10— A280—0— Conductivity- PhoA Activity0.60 -A280 •0.50 -90807060Conductivity(mmho)504030- 4.0- 3.0- 2.00.20 -0.10 -Figure 3.15 Purification profile for CBDPT-'PhoA: single column loading with distilledwater elution.CBDPT-'PhoA was purified from a culture supernatant of E. coli CC118/pTUg10*-CBDPT-'phoA (Materials and Methods, section 2.10.2). Elution with distilled water wasfollowed up with 6 M guanidine-HC1 as described. The peak fractions were divided into waterelution peak (fractions 61-77), and guanidine-HCI elution peak (fractions 84-95) pools.86Purification Volume PhoA Total Protein Total Specific Yield PurificationStep (ml) Activity Activity (mg/ml) Protein Activity (%) Factor(U/ml) (U) (mg) (U/mg)Culturesupernatant90 6.7 600 0.90 81 7.4 100H2O elution,before UP266 1.40 374 0.070 19 20 62 2.7H2O elution, 1.80 202 365 7.8 14 26 60 3.5UF retentatebG-HC1 elution, 0.985 177 174 6.7 6.6 26 29 3.6UF retentatebTable 3.3^Purification of CBDPT-'PhoA from E. coli CC118/pTUg10*-CBDPT-'phoAculture supernatant: single column loading with distilled water elution.a After membrane filtration and buffer reconstitution.b After final centrifugation.87elution was about 60% (Table 3.3), and that for the subsequent guanidine-HC1 elution was 29%,so the total yield of 89% was comparable to that for guanidine-HC1 elution alone (Table 3.2).This suggested a degree of binding heterogeneity, in that some of the CBDPT-'PhoA wasresistant to elution by distilled water but could be removed from cellulose under denaturingconditions. Purified CBDPT-'PhoA from both the water elution and subsequent guanidine-HCTelution had a specific activity of 26 U/mg, the same as for the guanidine-HC1-eluted protein inTable 3.2. Interestingly, the specific activity of the water-eluted protein increased by about 30%during the ultrafiltration step; this was either due to precipitation of denatured or contaminatingproteins in the ultrafiltration cell or activation of the concentrated enzyme in the Tris/ZnC12/MgC12buffer. The purified polypeptides were virtually indistiguishable on analysis by SDS PAGE(Figure 3.14, lanes 4 and 5).3.1.12 Purification of CBDPT-'PhoA: multiple column loading studies.It was hypothesized that a fraction of the potential CBD binding sites on CF1 cellulosebound CBDPT-'PhoA with higher affinity, and that if these sites were saturated then the fusionprotein could be bound to the remainder of the binding sites and desorbed with distilled water athigher yields. In other words, a CF1 cellulose column used repeatedly with cycles of sampleloading and water elution could allow higher purification yields of CBDPT-'PhoA than a columnused only once. This was shown to be true for CBDPT-'PhoA purified from the culturesupernatant of E. coli CC118/pTUg10*-CBDPT-iphoA. The same culture supernatant as used insection 3.1.11 above was applied to a 3 g CF1 cellulose column in 10 aliquots of 9 ml each(Materials and Methods, section 2.10.3). After each loading the column was washed with highand low salt buffer, eluted with distilled water, and re-equilibrated with low salt buffer. Thealkaline phosphatase activities of the column flowthrough/wash and water elution fractions areshown in Figure 3.16. For the first purification cycle the activity yield was relatively low, at36%, but thereafter it rose to 70%, and remained relatively stable at that level throughout theremaining cycles. This result was consistent with initial saturation of high affinity binding sites88100908070Yield of alkalinephosphatase soactivity (%)50403020100 ^ Flowthrough/wash■ H2O elutioni 1 1 I 1 1 I I I I1^2^3^4^5^6^7^8^9^10Cycle No.Figure 3.16 10-cycle purification of CBDPT-'PhoA from culture supernatant: purificationyield.CBDPT-'PhoA was purified from a culture supernatant of E. coli CC118/pTUg10*-CBDPT-'phoA (Materials and Methods, section 2.10.3). Column flowthrough and washfractions were pooled for activity measurement. The alkaline phosphatase yield is presented asa percentage of the activity loaded for each purification cycle.89on the CF1 cellulose and subsequent binding to and elution from lower affinity sites. The activitylost in the flowthrough/wash fraction increased gradually throughout the purification to a level of8.5% by the last cycle. Following the last purification cycle the column was washed with 6 Mguanidine-HC1 to elute any remaining fusion protein. The purification data for the 10 waterelution cycles combined and the guanidine-HC1 elution are shown in Table 3.4. The overall waterelution yield was 81%, compared to 60% for the single column loading in Table 3.3. As theactivity yields for individual cycles were around 70% (Table 3.4) the enzyme underwentactivation during the concentration step in the ultrafiltration cell. It is possible that some enzymeactivity was lost during the water elution and this was recovered after buffering andconcentration. This activation effect was observed in other similar purification experiments(results not shown). The specific activity of the water-eluted fusion protein (28 U/mg) wascomparable to the values obtained for the single loading purifications, but the fusion proteineluted with guanidine-HC1 had a specific activity of only 20 U/mg. This could have been due toincomplete reactivation of the enzyme by diafiltration or the accumulation of non-specificallybound contaminants on the cellulose which eluted during the guanidine-HC1 wash. The watereluted protein appeared identical to previously purified material on analysis by SDS PAGE(Figure 3.14, lane 6).An increased purification yield of CBDPT-'PhoA by multiple column loading was alsoachieved from a clarified cell extract of E. coli CC118/pTUg10*-CBDPT-'phoA (Table 3.5). At81% and 28 U/mg, the purification yield and specific activity of the purified protein were thesame as for the corresponding purification from the culture supernatant (Table 3.4). In additionthe purification yield for individual cycles increased over the first four cycles (Figure 3.17). Theloading levels for all multiple loading purification experiments were approximately 0.67 mgfusion protein/g CF1 cellulose; operation of small CF1 cellulose columns with purified CBDPT-'PhoA showed that higher loading levels resulted in increased activity losses in theflowthrough/wash fraction (results not shown). The purification factor for CBDPT-'PhoApurified from the clarified cell extract was 5.8 compared to 3.7 for purification from the culturesupernatant, the difference being due to the lower specific activity of the clarified cell extract. The90Purification Volume PhoA Total Protein Total Specific Yield PurificationStep (ml) Activity Activity (mg/ml) Protein Activity (%) Factor(U/ml) (U) (mg) (U/mg)Culturesupernatant90 6.7 600 0.90 81 7.4 100H2O elution, 2.105 230 490 8.3 18 28 81 3.7UF retentateaFinal G-HC1elution, 0.880 68 59 3.35 3.0 20 9.9 2.7UF retentateaTable 3.4^Purification of CBDPT-'PhoA from E. coli CC118/pTUg10*-CBDPT-'phoAculture supernatant: multiple column loading with distilled water elution.a After final centrifugation.91Purification Volume PhoA Total Protein Total Specific Yield PurificationStep (ml) Activity Activity (mg/ml) Protein Activity (%) Factor(U/ml) (U) (mg) (U/mg)Cell extract 2.85 182 517 38 108 4.8 100H2O elution, 1.94 216 420 7.8 15 28 81 5.8OF retentateaFlowthrough/wash1080 0.0325 35 6.8Table 3.5^Purification of CBDPT-'PhoA from E. coli CC118/pTUg10* -CBDPT - iphoAclarified cell extract: multiple column loading with distilled water elution.a After final centrifugation.921 2 54100908070Yield of alkalinephosphatase 60activity (%)504030201003Cycle No.Figure 3.17 5-cycle purification of CBDPT-'PhoA from clarified cell extract: purificationyield.CBDPT-'PhoA was purified from a clarified cell extract of E. coli CC118/pTUg10*-CBDPT-'phoA (Materials and Methods, section 2.10.3). The alkaline phosphatase yield ispresented as a percentage of the activity loaded for each purification cycle.93purified protein appeared identical to other purified preparations of CBDPT-'PhoA on SDSPAGE analysis (Figure 3.14, lane 8)..3.1.13 Cellulose binding of CBDPT-'PhoA at different buffer concentrations.Further evidence for the presence of low affinity binding sites for CBDPT-'PhoA on CF1cellulose was obtained from small scale CF1 column studies with purified CBDPT-'PhoA loadedin different concentrations of Tris-HC1 pH 7.5 (Figure 3.18; Materials and Methods, section 2.9).At a Tris concentration of 50 mM only about 1% of the activity loaded onto the column waspresent in the flowthrough fraction, and about 40% of the loaded activity was eluted with distilledwater. As the Tris concentration decreased, a trend of increased activity in the columnflowthrough with a corresponding decrease in water elution yield was observed. This trend wasconsistent with a proportion of the potential binding sites on the CF1 cellulose being able to bindCBDPT-'PhoA at higher ionic strength and release it at low ionic strength. Under conditions oflow ionic strength which would favour desorption, one would expect fewer of the "low affinity"sites to be occupied and therefore an increase of activity in the flowthrough and a correspondingdecrease of activity desorbed with distilled water. The remainder of the activity was notaccounted for, but would represent irreversibly bound activity, and to a lesser extent non-specifically bound activity and activity lost through inactivation.3.1.14 Proteolytic cleavage of CBDPT-'PhoA.The utility of the affinity tag approach to protein purification is dependent on theproteolytic removal of the affinity tag from the purified fusion protein. With this in mind, theCBDPT-'phoA fusion had been constructed with a sequence encoding a Factor X a recognitionsite separating the CBDPT and alkaline phosphatase coding sequences (Figure 3.10). PurifiedCBDPT-'PhoA, both in the dimeric and monomeric form, was subjected to proteolysis withFactor Xa (Materials and Methods, section 2.12.1), and the products were analyzed by SDS941 5 5040 -Yield of alkalinephosphataseactivity (%) 30 _Tris concentration(mM)Figure 3.18 Effect of loading buffer concentration on water elution yield of CBDPT-'PhoAfrom CF1 cellulose.0.1 mg samples of purified CBDPT-'PhoA were applied to 150 mg CF1 cellulosecolumns in different concentrations of Tris-HC1 pH 7.5 (Materials and Methods, section 2.9).The columns were washed, first with the loading buffer and then with water. Activities of thecolumn flowthrough/wash and water elution fractions were determined. Mean values fromduplicate experiments are presented.95PAGE (Figure 3.19A). The dimeric form of CBDPT-'PhoA was not cleaved by Factor X a, evenafter 18 hours digestion with a protease:fusion protein ratio of 1:50 (lane A3). Under the sameconditions the control protein, CenA-IEGR, was completely digested (lane A10). The lack ofproteolysis of CBDPT-'PhoA could be explained by the presence of Pro-6 of alkalinephosphatase in the P1'position, immediately following the Ile-Glu-Gly-Arg recognition sequence.It has been noted that trypsin-like proteases (including Factor X a) do not cleave substrates with aproline in the Pi' position (Nagai and Thogersen, 1984). What was surprising was the observedcleavage of the monomeric form of CBDPT-'PhoA by Factor X a, releasing a 47 kDa fragmentand a 20 kDa fragment (Figure 3.19A, lane 7). The N-terminal sequence of the 47 kDa fragmentwas Ala-Ala-Gln-Gly-Asp-Ile, which corresponded to amino acids 12-17 of mature alkalinephosphatase. This meant that Factor X a had cleaved CBDPT-'PhoA after Arg-11 of alkalinephosphatase, with the non-consensus recognition sequence Leu-Glu-Asn-Arg (Figure 3.20), andthat this site had not been accessible to the protease in the dimeric form of the fusion protein. Theincomplete digestion of monomeric CBDPT-'PhoA may have been due to the presence of somedimeric fusion protein, as following the acid dissociation and neutralization steps there was 14%residual enzymatic activity, attributable only to the dimeric species (McCracken and Meighen,1980).Cleavage of CBDPT-'PhoA with C.fimi protease was also investigated (Materials andMethods, section 2.12.2), as this protease had been shown to cleave CenA at both ends of thePro-Thr linker (Gilkes et al., 1989). CBDPT-'PhoA was cleaved by C. fimi protease (Figure3.19B, lanes 3-5), with digestion going almost to completion after 18 hours incubation at 37°(Figure 3.19B, lane 9). A major band of 52 kDa was released, as well as two or more minorbands of about 48 kDa. The N-terminal sequence of the 52 kDa band was Thr-Ser-Pro-Thr-Pro,which corresponded to amino acids 110-114 of mature CenA (Figure 3.20). This is the sameplace that C.fimi protease cleaves recombinant CenA, to release the transient polypeptide p36(Gilkes et al., 1989). Two N-terminal amino acid sequences were obtained for the 48 kDapolypeptides: Ala-Ala-Gln-Gly-Asp and Leu-Glu-Asn-Arg-Ala. These sequences correspondedto amino acids 12-16 and 8-12, respectively, of mature alkaline phosphatase (Figure 3.20). This96Figure 3.19 Factor Xa and C. fimi protease digestion of CBDPT-'PhoA.CBDPT-'PhoA was digested with Factor X a and C. fimi protease as described inMaterials and Methods, sections 2.12.1 and 2.12.2. A: Factor X a digestion of CBDPT-'PhoA. Lanes: 1, molecular weight standards; 2, CBDPT-'PhoA before digestion; 3,CBDPT-'PhoA + Factor Xa-18 hours; 4, CBDPT-'PhoA - Factor Xa-18 hours; 5, FactorXa only-18 hours; 6, CBDPT-'PhoA monomer before digestion; 7, CBDPT-'PhoAmonomer + Factor Xa-18 hours; 8, CBDPT-'PhoA monomer - Factor Xa-18 hours; 9,CenA-0-,:GR before digestion; 10, CenA-IEGR + Factor Xa-18 hours; 11, CenA-IEGR -Factor Xa-18 hours. B: C. fimi protease digestion of CBDPT-'PhoA. Lanes: 1, molecularweight standards; 2, CBDPT-'PhoA before digestion; 3, CBDPT-'PhoA + C. fimi protease-6 hours; 4, CBDPT-'PhoA + C. fimi protease-11 hours; 5, CBDPT-'PhoA + C. fimiprotease-18 hours; 6, CBDPT-'PhoA - C. fimi protease-18 hours; 7, C. fimi proteaseonly-18 hours; 8, CenA before digestion; 9, CenA + C. fimi protease-18 hours; 10, CenA- C. fimi protease-18 hours. Loadings were all equivalent to 2.514 of fusion protein.97108 109 4110 111Pro Thr Thr SerCBDPT-'PhoACBD^P 'PhoA131 132 133^ 6^7 4 8^9^10^11.1, 412^13Pro Thr Pro Ile Glu Gly Arg Pro Val Leu Glu Asn Arcr Ala Ala CBDPT-PhoACBD^P PhoA132 133^ 2^3^4^5^6^7^8^9^10^11Thr Pro Ile Glu Gly Arg Thr Pro Glu Met Pro Val Leu Glu Asn ArgCBD-'PhoA CBD 'PhoA106 107 108 109 4110 4111^6^7^8^9Thr Val Pro Thr Thr Ser Val Pro Val Leu GluFigure 3.20 Cleavage sites for Factor Xa and C. fimi protease in CBDPT-'PhoA,CBDPT-PhoA and CBD-'PhoA.The three fusion polypeptides are shown schematically with the CBD, Pro-Thr linker(PT), and PhoA domain indicated. The CenA leader peptide (shaded box) and Factor X arecognition site (black box) are also shown. Relevant amino acid sequences are shown, withalkaline phosphatase amino acids underlined. Amino acids are numbered from the matureN-termini of CenA and alkaline phosphatase. Positions of cleavage by Factor X a (1-) andC. fimi protease (.0.) are indicated. m—monomer form of protein.98showed that the C. fimi protease also cleaved CBDPT-'PhoA near the C-terminal end of the Pro-Thr linker. The p36 fragment of CenA is only a transient product of C. fimi protease digestion,as cleavage at the C-terminal end of the Pro-Thr linker converts it to p30, which is stable tofurther proteolysis (Gilkes et al., 1989). In CBDPT-'PhoA the 52 kDa polypeptide remained asthe major proteolysis product after 18 hours, in which time the CenA control was completelyconverted to p30 (Figure 3.19B, lanes 5 and 9). This indicated that the C-tenninal end of thePro-Thr linker was not as good a protease substrate as it is in CenA.3.1.15 Fusion proteins designed for proteolysis: construction of pTUg10*-CBDPT-phoA and pTUg10*-CBD-'phoA.Variants of the fusion protein, CBDPT-'PhoA, were designed as improved substrates forFactor Xa and C. fimi protease-catalyzed release of alkaline phosphatase from CBDc enA•CBDPT-PhoA contained amino acids 2-5 of mature alkaline phosphatase inserted betweenthe Factor Xa recognition sequence and Pro-6 of alkaline phosphatase in CBDPT-'1 3hoA (Figure3.20). This placed a threonine residue in the P1' position following the arginine of the Factor X acleavage site; the same amino acid is present in the Pit position in one of the Factor Xa cleavagesites in prothrombin, a natural substrate of Factor Xa (Nagai and Thogersen, 1984), so it wasexpected that CBDPT-PhoA would be a better substrate for Factor X a cleavage than itspredecessor. In addition, the released alkaline phosphatase would only be lacking Arg-1 of themature polypeptide, an amino acid that normally undergoes partial processing in vivo to producethe three isozymes characteristic of E. coli alkaline phosphatase (Nakata et al., 1987). CBDPT-PhoA was coded for by plasmid pTUg10*-CBDPT-phoA, the construction of which is shown inFigure 3.21. Plasmid from transformants of E. coli CC118 expressing alkaline phosphataseactivity were screened by digestion with BsiW I, an enzyme for which a unique cleavage siteexisted in the inserted linker. The presence of the linker was verified by DNA sequencing. Theencoded fusion protein had a predicted molecular mass of 61.0 kDa.99Figure 3.21 Construction of pTUg10*-CBDPT-phoA.pTUg10*-CBDPT-'phoA was digested with Mlu I and Stu Ito isolate a 6.14 kb'phoA/vector fragment. The same plasmid was also digested with Mlu I and Pvu Ito isolatethe corresponding 1.58 kb CBDCenA fragment. A linker coding for a Factor Xa cleavage siteand amino acids 2-5 of mature alkaline phosphatase was inserted between the CBDPT andPhoA coding regions in a three fragment ligation. The ratio of linker:CBDPT:'phoA/vectorwas 20:1:1. The double stranded linker is shown with its encoded amino acid sequence. Thenucleotide sequence and corresponding amino acid sequence across the fusion junction isshown, with the phoA gene underlined. Amino acid residues are numbered relative to themature N-termini of CenA and alkaline phosphatase. The regenerated Pvu I site and the BsiWI site used for screening are indicated. The positions of the plasmid (ofi) and phage (fl ori)replication origins, ampicillin resistance gene (amp), and /ac/q gene (bag) are indicated bylight filled boxes, and the leader peptide coding sequence of CenA is indicated by darker filledboxes. The orientation of CBDPT-phoA gene fusions is indicated by arrows. The lac (Plac)and tac (Ptac) promoters, transcriptional terminators (rrnB T1T2), and relevant restriction sitesare also shown.100Linkerglu gly^arg^thr pro glu^metC GAA GGT CGT ACG CCA GAG ATGTAG CTT CCA GCA TGC GGT CTC TACT4 DNA ligase132 133^ 2 3^4 5 6 7thr pro ile glu gly arg thr pro glu met pro valACG CCG ATC GAA GGT CGT A_ CGS CCA GAG ATG CCT GTTI Pvu I^BsiW I101CBD-'PhoA was a derivative of CBDPT-'PhoA which lacked the Pro-Thr linker of CenA(Figure 3.20). As the C. fimi protease cleaved CBDPT-'PhoA at both ends of the Pro-Thr linker,it was anticipated that the cleavage pattern of CBD-'PhoA would be more uniform. PlasmidpTUg10*-CBD-'phoA, which coded for CBD-'PhoA, was constructed as outlined in Figure3.22. Plasmid from transformants of E. coli CC118 expressing alkaline phosphatase activity wasscreened by restriction endonuclease digestion. The encoded fusion protein had a predictedmolecular mass of 58.0 kDa.3.1.16 Purification of CBDPT-PhoA and CBD-'PhoA.CBDPT-PhoA and CBD-'PhoA were purified from culture supernatants by affinitychromatography on CF1 cellulose, using multiple column loading and distilled water elution(Materials and Methods, section 2.10.3). This procedure also proved satisfactory for thesefusion proteins; CBDPT-PhoA and CBD-'PhoA were purified to specific activities of 26 and 29U/mg and with yields of 82 and 74% respectively (Tables 3.6 and 3.7). The slightly lower yieldof CBD-'PhoA compared to the other fusion proteins purified by this technique may have beendue to the use of only three loading cycles for the purification.3.1.17 Proteolytic cleavage of CBDPT-PhoA and CBD-'PhoA.Purified CBDPT-PhoA was digested with Factor Xa (Materials and Methods, section2.12.1), and the products analyzed by SDS PAGE (Figure 3.23). The purified protein appearedas a 67 kDa band, which was cleaved by Factor X a to produce 48 and 20 kDa bands. The N-terminal amino acid sequence of the 48 kDa band was Thr-Pro-Glu-Met-Pro, which correspondedto amino acids 2-6 of mature alkaline phosphatase; therefore Factor X a cleaved the fusion proteinat the expected position, after the arginine of the consensus recognition sequence (Figure 3.20).The 20 kDa band was the corresponding CBDCenA fragment, as it appeared in proportion to the 48 kDa band and ran at the same position as the CBDCenA fragment released from CenA-IEGR102Figure 3.22 Construction of pTUg10*-CBD-'phoA.pTUg10*-CBDPT-'phoA was digested with Sac I and Stu Ito remove a 511 byCBDPTcenA fragment. This was replaced with a 436 by CBDCenA fragment isolated frompUC18-1.6cenAApro_Thr. The positions of the plasmid (ori) and phage (fl on) replicationorigins, ampicillin resistance gene (amp), and /acil gene (laclq) are indicated by light filledboxes, and the leader peptide coding sequence of CenA is indicated by darker filled boxes.The orientation of cenAAPT and the CBDPT-phoA gene fusions is indicated by arrows. Thelac (Plac) and tac (Ptac) promoters, transcriptional terminators  rrrLAIIIIT2)_,Anincated_laaL-------gene, and relevant restriction sites are also shown.103I-1pa IT4 DNAligasepUC1 8-1 .6cenAAp ro _Thr4.20 kbHpa ISac I104Purification Volume PhoA Total Protein Total Specific Yield PurificationStep (ml) Activity Activity (mg/ml) Protein Activity (%) Factor(U/ml) (U) (mg) (U/mg)Culturesupernatant104 5.0 519 0.64 66 7.8 100H2O elution, 0.920 460 423 17.5 16 26 82 3.4OF retentateaFlowthrough/wash1300 0.031 40 7.8Table 3.6^Purification of CBDPT-PhoA from E. coli CC118/pTUg10*-CBDPT-phoAculture supernatant: multiple column loading with distilled water elution.a After final centrifugation.105Purification Volume PhoA Total Protein Total Specific Yield PurificationStep (ml) Activity Activity (mg/ml) Protein Activity (%) Factor(U/ml) (U) (mg) (U/mg)Culturesupernatant77.1 4.03 311 0.64 4.9 6.3 100H2O elution, 1.053 218 229 7.4 7.8 29 74 4.7OF retentateaFlowthrough/wash795 0.0195 15.5 5.0Table 3.7^Purification of CBD-'PhoA from E. coli CC118/pTUg10*-CBD-PhoA culturesupernatant: multiple column loading with distilled water elution.a After final centrifugation.106Figure 3.23 Factor X a digestion of CBDPT-PhoA.CBDPT-PhoA was digested with Factor X a as described in Materials and Methods,section 2.12.1. The ratio of protease to protein was 1:50 in lanes 3-5, 10, and 13, and 1:10 inlane 6. Lanes: 1, molecular weight standards; 2, CBDPT-PhoA before digestion; 3, CBDPT-PhoA + Factor Xa-6 hours; 4, CBDPT-PhoA + Factor X a-12 hours; 5, CBDPT-PhoA +Factor Xa-18 hours; 6, CBDPT-PhoA + Factor X a-18 hours; 7, CBDPT-PhoA - FactorXa-18 hours; 8, Factor Xa only-18 hours; 9, alkaline phosphatase before digestion; 10,alkaline phosphatase + Factor Xa-18 hours; 11, alkaline phosphatase - Factor Xa-18 hours;12, CenA-IEGR before digestion; 13, CenA-IEGR + Factor Xa-18 hours; 14, CenA-IEGR -Factor Xa^ 18 hours. Loadings were equivalent to 2.514 of fusion protein.107(lane 13). The rate of cleavage of CBDPT-PhoA was slow relative to the CenA-IEGR control,suggesting that the cleavage site wasn't freely available to the protease; however, CBDPT-PhoAwas a much improved substrate for Factor X a over its predecessor, CBDPT-'PhoA.The corresponding digestion of purified CBD-'PhoA with C. fimi protease was alsocarried out (Materials and Methods, section 2.12.2), and the products were analyzed by SDSPAGE (Figure 3.24). The fusion protein,which appeared as a 60 kDa band, was processed byC. fimi protease to produce a 48 kDa band. Processing was about 90% complete after 18 hours,by which time time some lower molecular weight degradation products were starting to appear(lane 5). The 48 kDa band gave two N-terminal sequences: Thr-Ser-Val-Pro-Val-Leu and Ser-Val-Pro-Val-Leu-Glu. These corresponded to cleavage of CBD-'PhoA between Thr-109—Thr-110 and Thr-110—Ser-111 of mature CenA, immediately preceding the deleted Pro-Thr linker(Figure 3.20). The former of these two positions matched the processing site in CenA to producep36 (Gilkes et al., 1989). CenAAPT, a CenA derivative with the Pro-Thr linker deleted, was notprocessed at this position but at a site four amino acids downstream, the processing point forrelease of the p30 catalytic domain fragment of CenA (Shen et al., 1991). The rate of processingof CBD-'PhoA was less than that of CenA (Figure 3.24, lane 12) and CenAAPT (Shen et al.,1991), which, together with the C. fimi protease and Factor Xa digestion results presented above,suggested that the region around the Pro-Thr linker was more sensitive to proteolysis in CenAand CenAAPT than in corresponding alkaline phosphatase fusion proteins. From the point ofview of homogeneity of proteolysis products, CBD-'PhoA was a better C. fimi protease substratethan its predecessor, CBDPT-'PhoA,3.1.18 Removal of the affinity tag by cellulose adsorption.Some affinity tag purification schemes incorporate a second affinity column stepfollowing proteolysis, to remove the cleaved affinity tag from the purified target protein (Maina etal., 1988; Smith and Johnson, 1988; Taylor and Drickamer, 1991). The feasibility of thisoperation was examined on the small scale for CBDPT-PhoA and CBD-'PhoA cleaved with1081 2 3 4 5 6 7 8 9 10 11 12 1397.4— es68— •53 oupp am dim* mi".45— is ium•■■■■3629—20.1---4•4111■4111Pero MOP Figure 3.24 C. fimi protease digestion of CBD-'PhoA.CBD-'PhoA was digested with C. fimi protease as described in Materials andMethods, section 2.12.2. Lanes: 1, molecular weight standards; 2, CBD-'PhoA beforedigestion; 3, CBD-'PhoA + C. fimi protease-6 hours; 4, CBD-'PhoA + C. fimi protease-12 hours; 5, CBD-'PhoA + C. fimi protease-18 hours; 6, CBD-'PhoA - C. fimi protease-18 hours; 7, C. fimi protease only-18 hours; 8, alkaline phosphatase before digestion; 9,alkaline phosphatase + C. fimi protease-18 hours; 10, alkaline phosphatase - C. fimiprotease-18 hours; 11, CenA before digestion; 12, CenA + C. fimi protease-18 hours; 13,CenA - C. fimi protease-18 hours. Loadings were equivalent to 2.5 Kg of fusion protein.109Factor Xa and C. finii protease respectively (Materials and Methods, section 2.10.4). Samplesfrom the protease digestions were passed through small Avicel columns and the flowthrough andAvicel bound fractions were analyzed by SDS PAGE (Figure 3.25). The alkaline phosphatasefragment of both fusion proteins passed through the column (lanes 6 and 13) while the CBDfragments and undigested fusion protein were adsorbed (lanes 7 and 14). About 80% of theactivity was recovered in the column flowthrough for both of the experiments. The alkalinephosphatase activity remaining bound to the Avicel appeared to be in the heterodimer form, ascomparable amounts of intact fusion polypeptide and alkaline phosphatase fragment were present(lanes 7 and 14). The effect of the Pro-Thr linker on the gel mobility of CBDCenA was clearlyevident (panel A, lanes 4 and 7 c.f. lanes 11 and 14).3.2 CenA-IL2 fusion proteins.3.2.1 Construction of pUC18-cenA'-IL2.To further test the potential of CBDc enA as an affinity tag for protein purification, a fusionbetween cenA and the gene for human interleukin 2 was constructed as outlined in Figure 3.26.Plasmid was isolated from ampicillin resistant transformants of E. coli JM109 and was screenedby restriction endonuclease digestion. The fusion junction between cenA and IL2 was confirmedby DNA sequencing. The resulting plasmid, pUC18-cenA'-IL2, encoded the CenA CBD, Pro-.Thr linker and 79 amino acids of the catalytic domain fused through a Factor X a cleavage site tothe mature N-terminus of IL2. The predicted molecular mass of the encoded fusion protein,CenA'-IL2, was 37.9 kDa.3.2.2 Detection of cenA'-1L2 expression in E. coli JM109.It was anticipated that the CenA leader peptide would direct the secretion of the CenA'-IL2fusion protein across the E. coli cytoplasmic membrane into the periplasm, from where it could be110Figure 3.25 CBD removal by adsorption to Avicel.CBDPT-PhoA and CBD-'PhoA were digested with Factor X a and C. fimi proteaserespectively, as described in Materials and Methods, sections 2.12.1 and 2.12.2. Samples ofeach digest were passed over 20 mg Avicel columns and the flowthrough and Avicel-boundfractions were sampled for SDS PAGE (Materials and Methods, section 2.10.4). Samplesfrom each digestion and binding experiment were electrophoresed through two 15% SDSpolyacrylamide gels. One gel was stained with Coomassie blue (panel A) and the other gelwas stained for alkaline phosphatase activity (panel B). Lanes: 1 and 8, molecular weightstandards (prestained for panel B); 2, CBDPT-PhoA before digestion; 3, CBDPT-PhoA -Factor Xa-18 hours; 4, CBDPT-PhoA + Factor Xa-18 hours; 5, Factor Xa only-18 hours;6, column flowthrough; 7, Avicel-bound fraction; 9, CBD-'PhoA before digestion; 10, CBD-'PhoA - C. fimi protease-18 hours; 11, CBD-'PhoA + C. fimi protease-18 hours; 12, C.fimi protease only-18 hours; 13, column flowthrough; 14, Avicel-bound fraction. Thepositions of CBDPT and CBD fragments are indicated. Loadings were all equivalent to 2.5i.tg of fusion protein.111Figure 3.26 Construction of pUC18-cenA'4L2.pUC18-1.6cenA was digested with Pst I and Sma Ito remove an 860 by fragment encodingthe C-terminal 205 amino acids of the CenA catalytic domain. This fragment was replaced witha 703 by ApaL I/Pst I fragment derived from pUC13-1L2 encoding mature human IL2. TheSma I end of the vector/cenA' fragment and the ApaL I end of the IL2 fragment were joined bya linker encoding a Factor Xa protease cleavage site. The DNA fragments were ligated at avector:insert:linker ratio of 1:1:14 and transformed into E. coli JM109. The positions of thereplication origin (ori.) and ampicillin resistance gene (amp) in the plasmids are indicated bylight filled boxes, and the leader peptide coding sequences of cenA and the IL2 gene areindicated by darker filled boxes. The orientations of cenA and the IL2 gene are indicated byarrows. The nucleotide sequence and corresponding amino acid sequence of the linker areshown. The lac promoter (Plac), truncated lacZ' gene and relevant restriction sites are alsoshown.112PstlSmalSaclSmalPst ISma IPst IPvu IIT4 DNA LigaseLinkerIle Glu Gly ArgG ATC GAA GGT CGC TAG CTT CCA GCA CGTApaL IApaL I113recovered in a soluble and active form. Indeed, initial studies indicated that osmotic shock fluidsof E. coli JM109/pUC18-cenA'-1L2 contained IL2 biological activity relative to a JM109/pUC18control (Figure 3.27). However, the calculated IL2 activities were low, in the order of 10 U/ml,and as specific activities of around 10 7 U/mg have been reported for purified recombinant IL2(Liang et al., 1985), this corresponded to IL2 concentrations of less than 1 ng/ml of culture.An experiment was conducted to determine if the IL2 activity from a clarified cell extractof JM109/pUC18-cenA'-IL2 could bind to Avicel, and if bound activity could be eluted withdistilled water (Figure 3.28). Although the range of dilutions was not extensive enough toaccurately determine the IL2 activities, it was apparent that much of the IL2 activity in the cellextract did not bind to Avicel, as judged by the proximity of the cell extract and Avicel supernatantcurves to each other at high dilutions. This suggested that either the fusion protein did not bindwell to cellulose or there were biologically active proteolysis products which could no longerbind. In addition, any activity which did bind to Avicel was not eluted with distilled water.Analysis of the same samples by Western blot probed with rabbit antiserum raised againstCenAEC2, a truncated form of CenA which lacked 45 N-terminal amino acids of the CBD,revealed one major immunoreactive band in the clarified cell extract which was removed fromsolution after incubation with Avicel (Figure 3.29, lanes 2 and 3). However, as this band ran atMr 23,000 relative to prestained standards, it probably represented a proteolytic degradationproduct of CenA'-IL2 which retained cellulose binding ability. This was supported by theappearance of a band of similar Mr in the cell extract of JM109/pUC18-1.6cenA, which alsobound to Avicel (lanes 8 and 9). No proteins of the expected M r (37,900) were detected.The fact that soluble 1L2 activity could not be correlated with the presence of intact CenA'-IL2 on Western blots prompted a more general approach to the detection of the fusion protein.The kinetics of CenA'-IL2 production in E. coli JM109/pUC18cenA '-IL2 were monitored byWestern blot analysis of crude cell extracts from cultures grown at 30° and 37°. Identical Westernblots were probed with anti-CenAEC2 serum and anti-human 1L2 (hIL2) antibody (Figure 3.30Aand B). Culture growth kinetics at 30° and 37° were also followed by measurement of absorbanceat 600 nm (Figure 3.31). The CenA'-1L2 fusion protein appeared as an immunoreactive band of1140.20—0— 1L2 standardJM109/pUC18-cenA'-1L2 #4JM109/pUC18-cenA'-IL2 #12JM109/pUC18I^1^I^I2^3^4^5^6^7^8^9log2 (dilution factor)0.001Figure 3.27 IL2 bioassay of osmotic shock fluids from E. coli JM109/pUC18 -cenA'4L2.The osmotic shock fluids from two positive clones screened by restriction endonucleasedigestion were assayed for IL2 biological activity (Materials and Methods, section 2.5). 30 mlcultures were grown at 37° until A600 = 0.3-0.5. IPTG was added to 1 mM and incubationwas continued until A600 = 0.7. Osmotic shock fluids were prepared as described in Materialsand Methods, section 2.6, and PMSF was added to 2014/ml.1150.50IL2 standardJM109/pUC18-cenN-1L2 cell extractJM109/pUC18-cenA'-1L2 Avicel supernatantJM109/pUC18-cenN-1L2 Avicel water washJM1 09/pUC1 8 cell extractJM1 09/pUC1 8-1 .6cenA cell extract0.45 —0.40 —0.35 —0.30 —A6000.25 —0.20 —0.15 —0.10 —0.05 —0.00^ 7^I 7^72^3^4^5^6^7^8^9log2 (dilution factor)Figure 3.28 IL2 bioassay of fractions from Avicel binding of E. coli JM109/pUC18-cenA'-IL2 clarified cell extract.25 ml cultures were grown at 37° until A600 = 0.4. IPTG was added to 1 mM andincubation was continued until A600 = 1.4. Harvested cells were resuspended in 5 nil 50 mMsodium phosphate pH 7.0 prior to rupture. After protein determination (Bio-Rad) a sample ofthe clarified cell extract of E. coli JM109/pUC18cenA'-IL2 containing 500 j_tg protein wasincubated with 10 mg Avicel (Materials and Methods, section 2.9). Washes were: 200 ill 0.5M NaC1, 2x200 ill 50 mM sodium phosphate pH 7.0, and 200 ill and 2x100 p1 distilled water.Water washes were pooled and adjusted to 0.15 M in NaCl prior to IL2 bioassay (Materialsand Methods, section 2.5). All samples were normalized to an equivalent volume of clarifiedcell extract.116Figure 3.29 Western blot of fractions from Avicel binding of E. coli JM109/pUC18-cenA'-112 clarified cell extract.Clarified cell extract preparation and Avicel binding were carried out as described in thelegend to Figure 3.28 and Materials and Methods, section 2.6. Equivalent volumes of clarifiedcell extract (lanes 2, 5, and 8), Avicel supernatant (lanes 3, 6 and 9) and Avicel water wash(lanes 4, 7 and 10) fractions were run on a 10% SDS polyacrylamide gel and blotted ontonitrocellulose. The blot was probed with anti-CenAEC2 serum. Lanes: 1, prestainedmolecular weight standards; 2-4, JM109/pUC18-cenA '-IL2; 5-7, JM109/pUC18; 8-10,JM109/pUC18-1.6cenA.117Figure 3.30 Western blots of crude cell extracts from E. coli JM109/pUC18-cenA'4L2grown at 30° and 37°.Cultures were grown at 30° (lanes 1-5) and 37° (lanes 7-11), as described in the legendto Figure 3.31. At 3, 5, 7, 9 and 11 hours after inoculation, 1 ml samples were taken and cellswere pelleted by centrifugation. Cells were resuspended in SDS sample buffer, boiled for 10minutes and electrophoresed through 10% SDS polyacrylamide gels. The gels were blottedonto nitrocellulose and the blots were probed with anti-CenAm serum (panel A) and anti-hIL2 antibody (panel B). Control cultures were sampled in stationary phase. Lanes: 1 and 7,3 hours; 2 and 8, 5 hours; 3 and 9, 7 hours; 4 and 10, 9 hours; 5 and 11, 11 hours; 6,prestained molecular weight standards; 12, JM109/pUC18; 13, JM109/pUC18-1.6cenA. Allloadings were equivalent to 5011.1 of culture.1181^I^I^1^I^I^1^I^I^I^I0^1^2^3^4^5^6^7^8 9^10 11Time (h)Figure 3.31 Growth curves for E. coli JM109/pUC18-cenA'-IL2 at 30° and 37°.0.5 ml of a 30° overnight culture of JM109/pUC18-cen/V-IL2 was inoculated into 2x50ml LB amp medium containing 0.2 mM IPTG. The cultures were incubated, one at 30° and theother at 37°, for 11 hours and the absorbance at 600 nm (A600) was measured every hour.119Mr 43,000 relative to the prestained standards. This was 5,000 higher than the predicted M r of37,900, a discrepancy which could be attributed to the presence of the Pro-Thr linker (Gilkes etal., 1989; Shen et al., 1991). More CenA 1-IL2 accumulated per unit culture volume at 30°. The30° culture also reached a higher cell density during the measured growth period, as indicated byA600 measurements (Figure 3.31). The rapid levelling off of absorbance of the 37° culture in thelate log phase of growth suggested that expression of cenA'-IL2 restricted cell growth at thistemperature. Extensive proteolysis of the fusion protein occurred at both growth temperatures,but more proteolytic intermediates were apparent at 30°. It is likely that higher proteolytic activityat 37° would prevent the accumulation of such intermediates. All subsequent expression ofcenA'-IL2 was with cultures grown at 30°.3.2.3 cenA'-IL2 expression in different E. coli strains.The extensive proteolysis of CenA'-IL2 observed in E. coli JM109 prompted the searchfor an E. coli strain which produced higher levels of the fusion protein with less proteolyticdegradation. The plasmid pUC18-cenA'-IL2 was transformed into a number of different E. colistrains, some of which carried mutations in genes for cellular proteases (Table 2.1). For eachrecombinant strain growth kinetics were monitored and Western blot analyses performed asdescribed above for JM109/pUC18-cenA'-IL2, except that 30° was the only growth temperatureused. Cell extracts from 11 hour cultures of different strains were compared on Western blotsprobed with anti-CenAEC2 serum and anti-hIL2 antibody (Figure 3.32), with sample loadingsnormalized for culture optical density. None of the strains tested yielded a significant reduction inthe level of proteolysis of CenA'-IL2, and in some cases the protease mutant strains yielded lessintact CenA'-1L2 than their wild type parents (e.g. lanes 3 and 4). Only E. coli BL21, deficient inthe outer membrane protease, OmpT, gave a significant increase in the level of intact fusionprotein, but also gave correspondingly higher levels of degradation products (lane 5). BL21 andCAG626, a Ion mutant strain, were the strains used for subsequent expression of cenA'-IL2._____120Figure 3.32 Expression of cenA'-1L2 in different strains of E. coli.Strains containing pUC18-cenA'-IL2 were grown at 30° as described in the legend toFigure 3.31. Crude cell extract samples prepared after 11 hours growth were analysed byWestern blot as described in the legend to Figure 3.30. Blots were probed with anti-CenAEC2serum (panel A) and anti-hIL2 antibody (panel B). All loadings were normalized for cultureoptical density. The following lane designations give the E. coli strain carrying pUC18-cenA'-1L2 and the relevant protease gene mutation (if any). Lanes: 1, prestained molecular weightstandards; 2, JM109; 3, RW193; 4, UT5600 (ompT); 5, BL21 (ompT); 6, KS272; 7, KS476(degP); 8, AB1899 (Ion); 9, CAG627; 10, CAG626 (Ion); 11, CAG597 (htpR); 12, CAG629(Ion, htpR); 13, CAG629/pUC18; 14, CAG629/pUC18-1.6cenA.1213.2.4 Cell fractionation studies.To determine the cellular location of the CenA 1-IL2 fusion protein expressed in E. coli,partial fractionation was carried out on harvested cells of BL21/pUC18-cenA'4L2 ruptured with aFrench press (Materials and Methods, section 2.6). An initial low speed centrifugation step at1000 x g for 10 minutes was performed to pellet unbroken cells as well as any protein inclusionbodies. This was followed with a high speed centrifugation step at 140,000 x g for 1 hour, topellet components of the cell envelope. The pelleted cell envelope fraction was extracted withbuffer containing 2% Triton X-100, a non-ionic detergent commonly used for solubilization ofmembrane proteins. Analysis of the generated fractions by Western blot (Figure 3.33) showedthat the intact fusion protein was located exclusively in the high speed pellet fraction (lane 5),suggesting localization in or association with the E. coli cell envelope. Formation of proteininclusion bodies is unlikely, as no fusion protein was evident in the low speed pellet (lane 3), andobservation of induced cultures by phase contrast microscopy gave no indication of the presenceof refractile inclusions. Some low molecular weight degradation products appeared in the solublefraction (lane 6), and the yield of these soluble products was enhanced by Triton X-100 extraction(lane 8); however, the intact fusion protein was not solubilised by Triton X-100 (lane 7). ThusTriton X-100 extraction provided a means of enriching the insoluble fraction for intact CenA 1-IL2. A faint immunoreactive band was present immediately above the 43,000 molecular weightband; this could represent the unprocessed form of the fusion protein. Surprisingly, inductionwith IPTG had little effect on the accumulated level of CenA 1-1L2; when a parallel culture wasgrown for partial fractionation but not induced, the level of intact fusion protein in the crude cellextract was comparable to that for the induced culture (not shown).To further determine the cellular location of CenA'-IL2, cytoplasmic and outer membraneswere separated by sucrose density gradient centrifugation. Gradient fractions were assayed forprotein, the cytoplasmic membrane marker succinate dehydrogenase, and the outer membranemarker 2-keto-3-deoxyoctonate (KDO), as shown in Figure 3.34A. In addition, the gradientfractions were analyzed by Western blot to determine the distribution of CenA'-1L2 (Figure1227144 —28 —15—Figure 3.33 Partial fractionation of E. coli BL21/pUC18-cenA'-IL2 crude cell extracts.Culture growth and the fractionation procedure are described in Materials and Methods,section 2.6. Samples of the various fractions were mixed with SDS sample buffer, boiled for10 minutes, and electrophoresed through 10% SDS polyacrylamide gels. The gels wereblotted onto nitrocellulose and the blots were probed with anti-CenAm serum (panel A) andanti-hIL2 antibody (panel B). Sample loadings were normalized for the volume of crude cellextract. Lanes: 1, prestained molecular weight standards; 2, crude cell extract 3, low speedpellet; 4, low speed supernatant 5, high speed pellet; 6, high speed supernatant; 7, Tritonpellet 8, Triton supernatant; 9, BL21/pUC18 crude cell extract.123Figure 3.34 Membrane location of CenA'-IL2.Sucrose isopycnic centrifugation fractions from E. coli CAG626/pUC18-cenA'-IL2crude membranes were assayed for protein content, succinate dehydrogenase activity(cytoplasmic membrane marker) and KDO content (outer membrane marker) (panel A).Samples of the same fractions were mixed with SDS PAGE sample buffer, boiled for 10minutes, and electrophoresed through 10% SDS polyacrylamide gels. The gels were blottedonto nitrocellulose and the blots were probed with anti-CenAEc2 serum (panel B) and anti-hIL2 antibody (panel C). Lanes: S, prestained molecular weight standards; T,CAG626/pUC18-cenA'-IL2 total membranes; 1-11, sucrose density gradient fractionscorresponding to panel A; 12, CAG626/pUC1a crude cell extract 13, CAC7626/pU€18-1.6cenA crude cell extract.124A.2.0 0.20 1000-•-• Protein--0- KDO/ProteinSDH/Protein9001.5Protein(mg/ml)^2BottomB.0.15KDO/Protein,(mg/mg)0.100.059^10^11Top800SDH/Protein700 (Units/mg)6005004003002001004 5 6 7 8Fraction #0.00^- 0S T 1 2 3 4 5 6 7 8 9 10 11 12 13106-71-44-28- -18-15-C.S T 1 2 3 4 5 6 7 8 9 10 1112 13106-71-44-28-18-15-1253.34B and C). The fusion protein was most concentrated in fractions 6 and 7, whichcorresponded to the cytoplasmic membrane peak in panel A. Therefore, intact CenA'-IL2 waseither localized within or associated with the E. coli cytoplasmic membrane. The fusion proteincould be identified in a Coomassie blue-stained SDS polyacrylamide gel of the same membranefractions (not shown) as a faint band of Mr 43,000 which was absent in membrane fractions fromthe negative control, E. coli CAG 626/pUC18.3.2.5 CenA'-IL2 purification: direct binding to Avicel.The possibility of binding the cytoplasmic membrane-associated CenA'-IL2 directly toAvicel was investigated using the resuspended high-speed centrifugation pellet from a crude cellextract of BL21/pUC18-cenA'-1L2 which had been pre-extracted with Triton X-100 (Figure3.35). When the resuspended membrane pellet was incubated with Avicel only a smallproportion of the fusion protein bound (lanes 5 and 6); however, when the Avicel supernatantcontaining unbound material was reincubated with Avicel and subjected to sonication, all thepreviously unbound CenA'-1L2 now bound to the Avicel (lanes 12 and 13). The sonicationprocedure probably facilitated contact of the CBD with the Avicel by disrupting membranevesicles as well as by reducing the particle size of the Avicel. Although this step resulted insignificant purification of CenA'-1L2, there was a considerable level of non-specific proteinbinding to the Avicel. A variety of wash conditions was investigated for the Avicel-bound protein(Figure 3.36), and only 0.1% SDS appeared to remove non-specifically bound protein whileleaving a substantial amount of CenA'-IL2 still bound (lane 6). 3 M guanidine-HC1 also reducednon-specific binding but completely eluted the fusion protein from Avicel. 6 M urea, lowerconcentrations of guanidine-HC1 and distilled water had similar but lesser effects.From the above findings, a purification scheme for CenA'-1L2 was developed based onthe following steps: i) the high speed centrifugation of crude cell extracts which had beenincubated with Triton X-100, to selectively solubilize and remove proteolytic degradationproducts; ii) sonication-aided binding of CenA'-1L2 from the resuspended pellet fraction to126Figure 3.35 Direct binding of CenA 1-IL2 to Avicel: effect of sonication.The Triton X-100-extracted insoluble fraction prepared from a crude cell extract of E.coli BL21/pUC18-cenA'-IL2 was incubated with Avicel, first without and then with sonication(Materials and Methods, section 2.11.1). Samples of all fractions were mixed with SDSPAGE sample buffer, boiled for 10 minutes, and electrophoresed through 10% SDSpolyacrylamide gels. Gels were stained with Coomassie blue (panel A) or blotted ontonitrocellulose and probed with anti-CenAEC2 serum (panel B). Lanes: 1 and 10, molecularweight standards; 2, crude cell extract; 3, Triton X-100-extracted supernatant; 4 and 11, TritonX-100-extracted pellet; 5, Avicel supernatant-without sonication; 6 and 7, Avicel extract-without sonication; 8, BL21/pUC18 crude cell extract; 9, BL21/pUC18-1.6cenA crude cellextract; 12, Avicel supernatant-with sonication; 13 and 14, Avicel extract-with sonication.Loadings are normalized to the crude cell extract in lane 2, except lanes 7 and 14, which areoverloaded by 15-fold.127Figure 3.36 Wash conditions for Avicel-bound CenAt-IL2.CenA 1-IL2 was bound to Avicel as described in Materials and Methods, section 2.11.1,and the Avicel was washed with various solutions in an effort to remove non-specificallybound protein. Avicel pellets after washing were extracted with SDS PAGE sample buffer andelectrophoresed through a 10% SDS polyacrylamide gel. The gel was stained with Coomassieblue (panel A). 10 ill samples of the corresponding wash supernatants were applied tonitrocellulose in a dot blot manifold and probed with anti-CenAm serum (panel B). Lanes:1, molecular weight standards; 2, 50 mM sodium phosphate pH 7.0; 3, 2% Triton X-100; 4,distilled H20; 5, 0.01% SDS; 6, 0.1% SDS; 7, 0.5 M guanidine-HC1; 8, 1 M guanidine-HC1;9, 1.5 M guanidine-HC1; 10, 2 M guanidine-HC1; 11, 3 M guanidine-HC1; 12, 1 mM DTT; /3,1 M urea; 14, 2 M urea; 15, 6 M urea.128Avicel ; iii) washing of the Avicel with buffer and 0.1% SDS, to remove non-specifically boundprotein, followed by elution of the fusion protein with higher concentrations of SDS. Figure3.37 shows SDS PAGE analysis of samples from this purification method. Although 0.1% SDSeffectively removed non-specifically bound protein from Avicel, this and other experimentsshowed that this step often eluted significant amounts of the fusion protein also (lanes 9 and 10).Attempts to remove SDS from these partially purified CenA'-IL2 preparations resulted inprecipitation of much of the fusion protein, but the combined use of cold and KC1 addition toprecipitate the detergent and detergent exchange with Triton X-100 allowed the recovery of smallquantities of soluble fusion protein which could rebind to Avicel (Figure 3.37, lanes 11-13). N-terminal sequencing of the purified polypeptide yielded two sequences: Ala-Pro-Gly-Cys-Argand Ala-Gln-Ala-Ala-Pro. These sequences correspond respectively to the mature N-terminus ofCenA and a leader peptidase processing variant; both forms are characteristic of CenApolypeptides expressed in E. coli (Guo et al., 1988; Gilkes et al., 1988).3.2.6 CenA'-IL2 purification: alkaline solubilization.An alternative approach to purification of CenA'-1L2 involved solubilization of the fusionprotein from the membrane fraction of E. coli and binding the solubilized form to Avicel. Theeffect of a number of different conditions on solubilization of CenA'-1L2 was assessed by highspeed centrifugation and Western blot analysis. The results, shown in Table 3.8, indicated that ofthe detergents tested, only 3% SDS had a solubilizing effect. Chaotropic agents such as Nal andNaSCN had a minor effect, but more significant was the effect of high pH on solubilization of thefusion protein. pH 11.5 was the lowest pH tested which effected significant solubilization ofCenA'-1L2. Further studies showed that by reducing the protein concentration to 5 mg/ml,almost complete solubilization of CenA'-1L2 could be achieved at this pH. Also, when thissolubilized material was neutralized and incubated with Avicel, or alternatively neutralized duringincubation with Avicel, the fusion protein bound to the Avicel (not shown). These findingsformed the basis of a purification scheme for CenA'-IL2, of which the major steps were:129Figure 3.37 CenA'-IL2 purification: direct binding to Avicel.CenA'-IL2 was purified from a crude cell extract of E. coli BL21/pUC18-cenA'-1L2(Materials and Methods, section 2.11.2): i) the insoluble fraction was extracted twice withTriton X-100 to remove contaminating proteins and proteolytic degradation products of CenA'-IL2;^CenA'-IL2 in the resuspended insoluble fraction was bound to Avicel with the aid ofsonication; iii) the Avicel was washed exhaustively with buffer and CenA'-IL2 was elutedwith buffer containing SDS; iv) the SDS was partially removed by precipitation and detergentexchange and the remaining soluble CenA'-1L2 was rebound to Avicel. Samples from variouspurification steps were mixed with SDS PAGE sample buffer and boiled for 10 minutes, thenelectrophoresed through 10% SDS polyacrylamide gels. The gels were stained withCoomassie blue. Lanes: 1, molecular weight standards, 1 lag per band; 2, crude cell extract;3, supernatant from first Triton X-100 extraction; 4 and 5, pellet from second Triton X-100extraction; 6, Avicel supernatant; 7, Avicel bound fraction before buffer washes; 8, bufferwash 1; 9, 0.1% SDS wash; 10, 1% SDS wash; 11, soluble fraction after SDS removal from0.1% SDS wash; 12, Avicel supernatant; 13, Avicel extract; 14, molecular weight standards,0.4 mg per band. Lanes 2-4 were loaded with equivalent volumes of sample, as were lanes 5-10  and lanes 11-13. Lanes 5-10 were overloaded 10-fold relative to lanes 2-4, and lanes 11-13were overloaded 2.5-fold relative to lanes 5-10.130Table 3.8^CenA'-1L2 solubilization trials.a All detergent treatments were in 50 mM sodium phosphate pH 7.0, and contained 0.15 MKC1, except the SDS treatment which contained no KC1. For chaotropic agent trialsmembrane pellets were resuspended in equal volumes of 50 mM sodium phosphate pH 7.0and an 8 M solution of the solubilization agent.b All detergent treatments were on crude cell extracts adjusted to the stated proteinconcentration. All other treatments were on membrane pellets from crude cell extracts whichhad been pre-extracted with 2% Triton X-100; the pellets were resuspended to the statedprotein concentration.c -1,-evels-c•f-sOtubilization: - none, +1- less than 10%, + 10-50%, ++ over 50%, +++ 100%.131Treatmenta Protein (mg/m1)b Solubilizatione3% Lubrol PX 53% CHAPS 53% octyl glucoside 53% SDS 5 +++3% sodium cholate 51% Triton X-100, 10 mM guanidine-HCJ 51% Triton X-100, 100 mM guanidine-HC1 51% Triton X-100, 0.5 M guanidine-HC1 50.75% Triton X-100, 0.75% CHAPS 51% Triton X-100, 1mM DTT 51% Triton X-100, 1 M urea 550 mM glycine-HCJ pH 2.6 7.5 +1-0.1 M NaHCO3-Na2CO3 pH 9.5 7.5 +/-25 mM NaHCO3-NaOH pH 10.5 7.5 +/-25 mM Na2HPO4-NaOH pH 11.5 7.5 ++50 mM KC1-NaOH pH 12.5 7.5 +++4 M Nal 7.5 +4 M NaSCN 7.5 +4 M NaC104 7.5132i) selective solubilization of CenA'-IL2 proteolytic degradation products from a crude cell extractwith Triton X-100, as described for the direct binding method in section 3.2.5; ii) thesolubilization of CenA 1-1L2 from the high-speed pellet at pH 11.5; and iii) binding of the fusionprotein to Avicel. A flow diagram of the purification scheme is shown in Figure 3.38 togetherwith an SDS polyacrylamide gel of samples from the various purification steps. Avicel-boundpurified CenA 1-IL2 appeared as a single band of Mr 43,000 (lane 8). Compared against themolecular weight standards in lane 9, the purified CenA'-IL2 band in lane 8 represented a yield ofapproximately 1.5 mg of fusion protein/litre culture.Further study of binding conditions for solubilized CenA'-IL2 showed that the fusionprotein could bind to Avicel at pH 11.5, and that there was in fact less non-specific binding at thispH than at pH 7.0 (Figure 3.39). Although not used here, this property was incorporated into thepurification scheme for a modified version of the fusion protein, described in Materials andMethods, section 2.11.5. A control experiment on recombinant hIL2 expressed in E. colishowed that this polypeptide does not bind to Avicel either at pH 7.0 or at pH 11.5 (Figure 3.40).3.2.7 Elution of CenA'-IL2 from Avicel.The study of various buffer washes on Avicel-bound CenAt-IL2 (Figure 3.36) showedthat the fusion protein was resistant to elution from Avicel, and only protein denaturants such asguanidine-HC1, SDS, and urea could effect significant elution. The effect of distilled water isdifficult to assess in batchwise washes, as residual buffer trapped within the Avicel pelletincreases the ionic strength of the wash solution; however, distilled water did not elute CenA'-IL2from CF1 cellulose in a column format (not shown). Additional elution trials were carried out onAvicel-bound CenA'-IL2 purified by the alkaline solubilization procedure. The amount of fusionprotein eluted from the Avicel was assessed by SDS PAGE. The compiled results from elutionexperiments are shown in Table 3.9. The fusion protein remained bound to Avicel under mostconditions tested, including extremes of pH. It is evident that at least partial denaturation ofCenA'-IL2 is necessary to break its association with cellulose.133Figure 3.38 CenA'-1L2 purification: alkaline solubilization.CenA'-1L2 was purified from a crude cell extract of E. coli BL21/pUC18-cenA'-1L2according to the above flow diagram and as described in Materials and Methods, section2.11.4. Samples from various purification steps were mixed with SDS PAGE sample bufferand boiled for 10 minutes, then electrophoresed through a 10% SDS polyacrylamide gel. Thegel was stained with Coomassie blue. Lanes: 1, crude cell extract; 2, supernatant from firstTriton X-100 extraction; 3 and 4, pellet from second Triton X-100 extraction; 5, supernatantafter pH 11.5 extraction; 6, pellet after pH 11.5 extraction; 7, Avicel supernatant; 8, Avicelextract; 9, molecular weight standards. All loadings were normalized to the volume of crudecell extract. Lanes 4-8 were overloaded 10-fold relative to lanes 1-3.134Figure 3.39 CenA'-IL2 binding to Avicel at pH 11.5.A crude cell extract of E. coli BL21/pUC18-cenits-IL2 was extracted with Triton X-100 and the insoluble fraction extracted with buffer at pH 11.5, as described in Materials andMethods, section 2.11.4. 40111 aliquots of the resulting supernatant were diluted twofold witheither 50 mM sodium phosphate, 3 mM EDTA pH 7.0 (phosphate/EDTA) or with 25 mMNa2HPO4-NaOH pH 11.5 and incubated with 10 mg of Avicel The Avicel supernatants wererecovered and the Avicel pellets were washed once with 100 of the dilution buffer. Sampleswere boiled in SDS PAGE loading buffer and electrophoresed through a 10% SDSpolyacrylamide gel. The gel was stained with Coomassie blue. Lanes: 1, molecular weightstandards; 2 and 5, supernatant after pH 11.5 extraction; 3, phosphate/EDTA dilution-Avicelsupernatant; 4, phosphate/EDTA dilution-Avicel extract; 6, phosphate pH 11.5 dilution-Avicelsupernatant; 7, phosphate pH 11.5 dilution-Avicel extract.135Figure 3.40 Western blot of recombinant human IL2 incubated with Avicel at pH 7.0 and11.5.1 lig samples of recombinant human IL2 (BRL # 3238SA) were lyophilized andresuspended in 20111 of either 50 mM sodium phosphate, 3 mM EDTA pH 7.0 or 25 mMNa2HPO4-NaOH pH 11.5. After centrifugation at 20,000 x g for 10 minutes at 4°, 10 ill ofeach supernatant were incubated with 1 mg Avicel for 1 hour on ice. The remaining 10111were incubated without Avicel under the same conditions. Avicel pellets were washed twicewith 100111 of the respective buffers before sampling for SDS PAGE. Lanes: 1, prestainedmolecular weight standards; 2, pH 7.0, no Avicel; 3, pH 7.0, Avicel supernatant; 4, pH 7.0,Avicel extract; 5, pH 11.5, no Avicel; 6, pH 11.5, Avicel supernatant 7, pH 11.5, Avicelextract.136Treatment^ Elutione50 mM sodium phosphate pH 7.0a2% Triton X-100a^ -distilled H20a +/-0.1% SDSa^ +/++3 M guanidine-HCla^ +++1 mM DTTa -6 M ureaa^ +0.1 M KCl-HC1 pH 1.5b^ -50 mM KC1-NaOH pH 12.5b +/-10% cellobioseb4% carboxymethyl celluloseb^ -5% n-butanolb^ -10% n-propanolb -1 M NaSCNb^ +/-1 M NaC104b +/-Table 3.9^Elution of CenA'-IL2 from Avicel.a Results taken from Figure 3.36.b Tested on Avicel-bound CenA 1-IL2 purified by the alkaline solubilization procedure(Materials and Methods, section 2.11.4). 1 mg Avicel samples, which had approximately7.5 pg CenA'-IL2 bound, were resuspended in 50 t1 of test solution and incubated for 30minutes at room temperature. The Avicel supernatant and extract were analyzed by SDSPAGE to determine the degree of elution of the fusion protein.c Amount of CenA'-IL2 eluted: - none, +/- less than 10%, +, 10-50%, ++ over 50%, +++100%.1373.2.8 Factor Xa digestion of CenA'-IL2.The cenA'-1L2 gene fusion had been constructed with a linker coding for a Factor X acleavage site, so that digestion with Factor X a would release IL2 from the fusion protein with itscorrect N-terminus. Due to the difficulty in eluting the purified CenA'-IL2 from Avicel, theAvicel-bound fusion protein was digested with Factor X a and the products analyzed by SDSPAGE (Figure 3.41). Surprisingly, at least 4 cleavage products were observed, although thefusion protein contained only one consensus Factor X a cleavage site. Subsequent N-terminalsequencing of these fragments revealed that the smallest fragment of M r approximately 15,000represented correctly processed 1L2 (Figure 3.41, fragment 1), but two of the other fragmentsresulted from cleavage within the CenA catalytic domain portion of the fusion protein, after Arg-153 and Arg-197 of mature CenA (fragments 3 and 2 respectively). Apart from the two arginineresidues and Gly-196, the two spurious Factor X a cleavage sites shown in Figure 3.41 bore noresemblance to the consensus recognition sequence. Identical amino acid sequences were presentin the positive control protein, CenA-IEGR, but this protein was not cleaved at either of thesesites. As the catalytic domain in CenA-IEGR is correctly folded into a functional tertiarystructure, the potential Factor Xa cleavage sites are probably masked, whereas in CenA'-1L2 thetruncated catalytic domain can not attain its native tertiary structure and the same sites may beexposed to proteolytic action. This may provide one reason for the proteolytic instability ofCenA'-IL2 in E. coli.3.2.9 Construction of pTZ18U-CBDPT-1L2.A modified gene fusion between cenA and the gene for human interleukin 2 wasconstructed to delete the CenA catalytic domain coding region from cenA'-1L2, and thereforeremove the non-specific Factor X a cleavage sites from the encoded fusion polypeptide. Theplasmid pTZ18U-CBDPT-IEGR, which coded for the CBD and Pro-Thr linker of CenA fused toa Factor Xa cleavage site, was used for the construction of a precise CBDPT-IL2 fusion (Figure138Figure 3.41 Factor Xa digestion of CenA'-IL2.Avicel-bound CenA'-1L2 was digested with Factor X a as described in Materials andMethods, section 2.12.3. Samples incubated for 18 hours with and without protease wereboiled in SDS PAGE sample buffer and electrophoresed through a 12% SDS polyacrylamidegel alongside corresponding samples of the control protein, CenA-IEGR. The gel was stainedwith Coomassie blue. Lanes: 1, molecular weight standards; 2, CenA-IEGR without protease;3, CenA-IEGR with protease; 4, CenA'-IL2 without protease; 5, CenAt-1L2 with protease.Indicated bands were sequenced at their N-termini to determine the position(s) of Factor X acleavage, which are shown in the schematic diagram of CenA'-IL2. The shaded boxrepresents the leader peptide and the black box the consensus Factor X a cleavage site. Mrepresents the mature N-terminus of CenA. The deduced Factor X a recognition sites in CenA'-IL2 are also listed. A schematic diagram of CenA-IEGR is also shown, with the singlecleavage position of Factor X a indicated.139Figure 3.42 Construction of pTZ18U-CBDPT-IL2.pTZ18U-CBDPT-IEGR was digested with Sac I and Pst Ito isolate a 2.83 kb vectorfragment, and with Sac I and Pvu Ito isolate a 500 by CBDPT fragment. These were ligatedtogether with a 715 by IL2 fragment from pUC18-cenA'-1L2 in a three-fragment ligation. Thepositions of the plasmid (ori) and filamentous phage (f 1 ori) replication origins and theampicillin resistance gene (amp) in the plasmids are indicated by light filled boxes, and theleader peptide coding sequences of cenA and the IL2 gene are indicated by darker filled boxes.The orientations of cenA gene fragments and fusions are indicated by arrows. The nucleotidesequence and corresponding amino acid sequence of the fusion junction of CBDPT-IL2 isshown, with the Ile-Glu-Gly-Arg tetrapeptide separating the Pro-Thr linker of CenA frommature IL2 (underlined amino acids). The Pvu I site used for cloning is shown, bothunderlined and in bold type.140PvulPst IPvu IT4 DNA LigaseThr Pro Ile Glu Gly Arg Ala Pro ACG CCG ATC GAA GGT CGT GCA CCTPvu I1413.42). Ampicillin resistant transformants in E. coli JM101 were screened by colony immunobloton both nitrocellulose and filter paper, with anti-hIL2 antibody as the probe for CBDPT-IL2production (not shown). Positive clones were further screened by restriction digestion, and thefusion junction between CBDPT and IL2 was confirmed by DNA sequencing. The resultingplasmid, pTZ18U-CBDPT-1L2, coded for the cellulose-binding domain and Pro-Thr linker ofCenA fused through a Factor X a cleavage site to the mature N-terminus of IL2. The predictedmolecular mass of the encoded fusion protein, CBDPT-IL2, was 29.4 kDa. pTZ18U-CBDPT-IL2 was transformed into E. coli BL21 for expression and purification of CBDPT-IL2.3.2.10 Purification of CBDPT-IL2.Purification of CBDPT-IL2 from crude cell extracts was by the alkaline solubilizationprocedure developed for CenA'-IL2 (Figure 3.38) and incorporated a modified binding stepwhereby the solubilised fusion protein was bound to Avicel at pH 11.5. Samples from each ofthe purification steps were analyzed by SDS PAGE and Western blot and are shown in Figure3.43. Unlike CenA'-IL2, CBDPT-IL2 was present in both the soluble and insoluble fractions ofthe crude cell extract (lanes 3 and 4), and some of the fusion protein was also solubilised bytreatment of the pellet fraction with Triton X-100 (lane 5). It is clear that removal of the CenAcatalytic domain portion of CenA'-IL2 did not diminish the level of proteolysis of the resultingfusion protein in E. coli, as an array of immunoreactive degradation products were visible in bothWestern blots. On the Coomassie blue-stained gel the Mr of the purified protein was 32,000(lanes 11 and 12); it appeared relatively free from contaminating proteolytic fragments, althoughthe Western blots indicated that low levels of such contaminants must have been present. TheCBDPT-IL2 in the soluble fraction could also bind to Avicel but there were significant levels ofproteolysis products which copurified with the intact fusion protein (not shown). Additionalbands of Mr about 70,000 were visible on both the SDS gel and the Western blots, and probablyrepresented dimeric aggregates of the fusion protein which were resistant to dissociation in theSDS PAGE sample buffer. The yield of pure Avicel-bound protein by this procedure was142Figure 3.43 Purification of CBDPT-IL2.CBDPT-IL2 was purified by the alkaline solubilization method developed for CenA'-1L2as described in Materials and Methods, section 2.11.5. Samples from various purification stepswere mixed with SDS PAGE sample buffer, boiled 10 minutes and electrophoresed through three15% SDS polyacrylamide gels. One gel was stained with Coomassie blue (panel A) and the othertwo were blotted onto nitrocellulose and probed with anti-hIL2 antibody (panel B) and anti-CenAEC2 serum (panel C). Lanes: 1, molecular weight standards; 2, crude cell extract; 3,clarified cell extract; 4, pellet before Triton X-100 extraction; 5, supernatant after Triton X-100extraction; 6, pellet after Triton X-100 extraction; 7 and 9, supernatant after pH 11.5 extraction; 8,pellet after pH 11.5 extraction; 10, Avicel supernatant; 11 and 12, Avicel extract after bufferwashes; 13, BL21/pTZ18U crude cell extract; 14, BL21/pUC18-cenA'-1L2 crude cell extract. Allloadings were normalized to the volume of crude cell extract. On the Coomassie blue-stained gel,lanes 9-11 were overloaded 3-fold relative to lanes 2-8, and lane 12 was overloaded 5-fold relativeto lane 11. The Western blotted gels were loaded in the same way, but lanes 2-8 were 2-foldoverloaded relative to the the same lanes on the Coomassie blue-stained gel.143approximately 0.65 mg/litre of culture, as determined by densitometric scanning of the purifiedCBDPT-IL2 band (lane 12) compared against the molecular weight standards (lane 1).3.2.11 Factor Xa digestion of CBDPT-IL2.When purified Avicel-bound CBDPT-IL2 was washed repeatedly with distilled water thefusion protein remained bound to the cellulose (not shown). For this reason, CBDPT-IL2 wasdigested with Factor Xa while bound to Avicel, and the digestion products analyzed by SDSPAGE (Figure 3.44). Under the conditions used, partial digestion of CBDPT-IL2 was observed,releasing two polypeptide fragments of Mrs 20,000 and 15,000 relative to standards. Analysison Western blots (not shown) indicated that the 15,000 molecular weight fragment was IL2 andthe 20,000 molecular weight fragment was the CBD. The N-terminal amino acid sequence of thesmaller fragment was Ala-Pro-Thr-Ser-Ser, which corresponded to the N-terminal sequence ofmature IL2 (Figure 3.44). Therefore Factor X a cleaved CBDPT-IL2 specifically at the consensustarget sequence to release mature IL2. The mobility of the IL2 fragment on SDS gels wasconsistent with the predicted molecular mass for IL2 of 15.4 kDa. Surprisingly, the cleaved IL2fragment remained associated with the Avicel after digestion (Figure 3.44, lanes 6-8), suggestingeither that the polypeptide precipitated following cleavage or that it was somehow trapped in thecellulose matrix. Studies on this purification method were discontinued at this point; however,the purification itself showed that the CBD functions effectively as an affinity tag, even underextreme conditions of pH.144Figure 3.44 Factor Xa digestion of CBDPT-1L2.Avicel-bound CBDPT-1L2 was digested with Factor Xa as described in Materials andMethods, section 2.12.3. Supernatant and Avicel-bound samples taken at 6, 12 and 18 hourswere boiled in SDS PAGE sample buffer and electrophoresed through a 15% SDSpolyacrylamide gel. The gel was stained with Coomassie blue. Lanes: 1, molecular weightstandards; 2, CBDPT-IL2 at t = 0; 3, CBDPT-IL2 + Factor Xa-6 hour supernatant; 4, 12 hoursupernatant; 5, 18 hour supernatant; 6, CBDPT-IL2 + Factor Xa-6 hour Avicel bound; 7, 12hour Avicel bound; 8, 18 hour Avicel bound; 9, CBDPT-IL2 only-18 hour supernatant; 10,CBDPT-1L2 only-18 hour Avicel bound; 11, Factor Xa only-18 hour supernatant; 12, FactorXa only-18 hour Avicel bound. Equivalent volumes were loaded in all lanes. The N-terminalamino acid sequence of the arrowed band is shown underlined, and the corresponding FactorXa cleavage site is indicated by a vertical arrow. The amino acids of IL2 are numbered,starting at the mature N-terminus.1454. Discussion4.1 General.This study has shown that fusion of heterologous polypeptides to the cellulose-bindingdomain of C. fimi  endoglucanase A facilitates their binding to and purification on cellulose.When initially reported for alkaline phosphatase (Greenwood et al., 1989), this was the firstdemonstration of a cellulose-binding domain retaining its binding function when fused to aheterologous polypeptide. Together with the work of Ong et al. (1989) this has initiated thedevelopment of applications of CBDs in biotechnology (Ong et al., 1989a; Kilburn et al., 1992),a patent for which has been granted (Kilburn et al., 1992a). A cellulose affinity purificationmethod for alkaline phosphatase fused to CBDCenA was developed and optimized for increasedyield by elution with distilled water. Variations of the fusion polypeptide were also engineeredfor cleavage by Factor X a and C. fimi protease for removal of the affinity tag. Fusions betweenCBDCenA and human interleukin 2 were generated, which also facilitated purification of therecombinant polypeptides on cellulose, although proteolysis and insolubility of the polypeptidesin E. coli were limiting factors to the approach.4.2 TnphoA mutagenesis.The TnphoA mutagenesis approach used in this study provided a convenient method forproducing a range of C-terminal CenA deletion mutants fused to E. coli alkaline phosphatase andfor probing the C-terminal limits of the CBD by screening the cellulose binding of such mutants.Two qualitative assays were developed for screening fusion proteins, based on binding to filterpaper and to microcrystalline cellulose (Avicel). It was found that the Pro-Thr linker was notnecessary for cellulose binding, but deletion of 28 or 68 amino acids of the CBD completelyabolished binding function. The N-terminus of the CBD was also sensitive to deletions of 18 or44 amino acids. The stability of the various fusions reflected the domain structure of CenA, as146fusions within the Pro-Thr linker or near the C-terminus of the catalytic domain were less subjectto proteolysis in vivo or during the extraction procedure than were fusions within the CBD orcatalytic domain. The failure of either of the N-terminal deletions to bind to cellulose is contraryto results reported by Gilkes et al. (1989) who found that proteolytic fragments of recombinantCenA lacking 45, 47, or 64 N-terminal amino acids retained weak affinity for cellulose. Thereason for this difference is not known, but may stem from the different approaches taken togenerate the deletion, i.e. expression of the truncated gene in E. coli or proteolytic generation ofthe deletion from the correctly folded polypeptide.One problem encountered with the TnphoA mutagenesis approach was the high number of(3-lactamase (bla) gene fusions isolated, even when ampicillin was included in the platingmedium. Plasmid preparations from such clones usually showed the presence of more than onetype of plasmid, one of which had a TnphoA insertion in bla and the other either had a TnphoAinsertion in cenA or no apparent insertion. Such clones could be distinguished by the formationof sectored colonies on medium containing XP and ampicillin but no kanamycin.4.3 Heterodimer studies.Experiments with heterodimers of CenA'-'PhoA IX-8 generated in vivo by partialproteolytic cleavage showed that alkaline phosphatase dimers containing only one attached CBDwere still able to bind to cellulose. This fmding was mirrored by studies on CBDCex fused to theC-terminus of a dimeric P-glucosidase from an Agrobacterium species (Abg-CBDcex ;Greenwood et al., 1992). Furthermore, dissociated CenA'-'PhoA IX-8 monomers bound toAvicel could, under conditions favouring dimerization, reassociate with alkaline phosphatasemonomer in solution to form Avicel-bound heterodimers. This result has implications for the useof CBD fusion proteins bound to cellulose as affinity ligands for the detection or purification ofpolypeptides which associate with the fusion partner. This can be viewed as a form of affinitychromatography, where the affinity ligand itself can be purified and immobilized in one step.Examples of this sort of approach using CBDs as affinity tags are the fusion of CBDc ex to147protein A, which facilitates the affinity purification of antibodies and development of diagnostics(Kilburn et al., 1992), and CBDCex fusion to streptavidin, for detection or immobilization ofbiotinylated polypeptides and nucleic acids (Le, 1992). Some interesting examples of thisapproach using other affinity tags have been reported. In one study, a peptide antigen wasexpressed as a fusion with two different affinity tags, the IgG binding domains of protein A andthe human serum albumin binding domain of streptococcal protein G; one fusion protein wasused to generate an immune response against the recombinant peptide, and the other fusionprotein was used to screen and purify the peptide-specific antibodies (Stahl et al., 1989). Anotherreport outlines the use of a protein A fusion to the lac repressor of E. coli to purify on IgG-sepharose DNA fragments containing a lac operator sequence (Lundeberg et al., 1990).In addition, the heterodimer studies show that the CBD could be used for purification orimmobilization of multisubunit proteins generated in vivo, provided the CBD fused at either theN- or C-terminus did not interfere with subunit association. In one of the earliest reports onaffinity tag purification, Germino et al. (1983) purified a plasmid replication initiator proteinthrough subunit interaction with a hybrid protein fused to 13-galactosidase. A polyhistidineaffinity tag has been used to purify heterodimers of reverse transcriptase from human type 1immunodeficiency virus (LeGrice and Griininger-Leitch, 1990). A further implication from theseheterodimer studies is the use of CBD fusions as tools for probing protein-protein interactions.An elegant example of such an application involves the use of glutathione S-transferase as anaffinity tag to a eukaryotic transcriptional activator to probe activator interactions with generaltranscription factors (Lin and Green, 1991).Immobilized subunits of alkaline phosphatase have been generated previously by chemicalimmobilization of the purified polypeptide, and used to probe binding of zinc by the monomerand dimer and to demonstrate reassociation of free and immobilized monomers as has beendemonstrated here (McCracken and Meighen, 1980). However, the earlier study used acombination of 6 M guanidine hydrochloride and low pH to dissociate the immobilized dimersand dilution to facilitate reassociation, a treatment that would have resulted in desorption of theCenA'-'PhoA IX-8 fusion protein from cellulose. The conditions used for dissociation and148reassociation described here provide a new insight into the requirement for zinc ions for themonomer subunits of alkaline phosphatase to reassociate after dissociation induced by low pH(Schlesinger and Barrett, 1965; Applebury and Coleman, 1969). Applebury and Coleman (1969)have reported that dimers can reform in the absence of zinc as the pH is raised above 6. Thisstudy used dialysis in buffer containing chelating agents 1,10-phenanthroline or 8-hydroxy-5-quinolinesulfonic acid, to form the apoprotein, followed by dialysis in metal-free buffers toremove the chelating agent prior to subunit dissociation and reassociation. On the other hand,Schlesinger and Barrett (1965) found that neutralization of the acid-dissociated subunits in thepresence of EDTA prevented dimerization, but the subsequent addition of excess zinc allowedreassociation to occur. This latter finding is supported by the heterodimer reconstitutionexperiment described here. In addition, the modified Factor Xa cleavage pattern of acid-dissociated CBDPT-'PhoA neutralized in the presence of EDTA suggests a significantly differentconformation for this polypeptide preparation over the dimeric form of the fusion protein. It hasbeen reported that EDTA can bind to metal-free alkaline phosphatase (Csopak et al., 1972); it ispossible that dimerization is prevented in the presence of EDTA by EDTA binding to the subunit,rather than by the direct chelation of zinc ions. This explanation reconciles the two apparentlyconflicting reports, provided that EDTA in the presence of excess zinc no longer interferes withsubunit association.4.4 High expression and purification of alkaline phosphatase fusion proteins.With the use of pTUglO*N18 to express engineered cenA-phoA gene fusions in E. coli,the high levels of fusion protein leaked into the culture medium eliminated the need for cell extractpreparation or concentration of the culture supernatant prior to affinity purification. Leakage ofC. fimi cellulases from E. coli has been reported previously (Guo et al., 1988), and there areother reports of leakage of overexpressed polypeptides from the E. coli periplasm (Abrahmsen etal., 1986; Georgiou et al., 1988; Hellman and Mantsala, 1992). Significantly, the extracellularlevels of alkaline phosphatase activity reported here are on the order of four times that reported for149optimized extracellular production of alkaline phosphatase in periplasmic-leaky (lky) strains of E.coli (Atlan and Portalier, 1984).The fusion proteins CBDPT-'PhoA, CBDPT-PhoA and CBD-'PhoA could all be purifiedto near homogeneity by affinity chromatography on CF1 cellulose. Two elution conditions wereused, both of which had been used previously for purification of C. fimi cellulases expressed inE. coli (Gilkes et al., 1988). Guanidine-HC1 at concentrations up to 6 M eluted CBDPT-'PhoAfrom CF1 cellulose at over 80% yield, and active fusion protein could subsequently be recoveredfollowing removal of the guanidine-HC1 by diafiltration. Distilled water also effected desorptionof CBDPT-'PhoA under the same column loading and operating conditions, but the purificationyield of about 60% was less than that achieved with guanidine-HC1. It was proposed that thelower water elution yield reflected a heterogeneity of binding sites on the CF1 cellulose, some ofwhich bound the fusion protein with greater affinity than others. It was also proposed that fusionprotein bound to the lower affinity sites could be eluted from the cellulose under the milderconditions of distilled water, while guanidine-HC1 was necessary to effect elution of fusionprotein from the higher affinity sites. Evidence in support of this proposal was obtained frommultiple-loading column purification studies, where sample loading, buffer washing, and waterelution steps were repeated in a cyclic fashion. The observation that the water elution yieldincreased during the first few cycles and then remained relatively constant was consistent withsaturation of higher affinity binding sites, and subsequent utilization of the lower affinity bindingsites. This kind of behaviour is reminiscent of multiple-use polyclonal antibody affinity columns,where increases in recovery and decreases in capacity have been observed during the first fewpurification cycles, a phenomenon also attributed to saturation of high affinity binding sites (Ehleand Horn, 1990). In addition, the application of fusion protein to columns of CF1 celluloseunder conditions of decreasing ionic strength showed increasing levels of fusion protein passingthrough the column without binding, at the expense of the yield of water-eluted protein. This wasconsistent with the presence of lower affinity binding sites on the cellulose which could bindfusion protein under conditions of high ionic strength and release it at low ionic strength. Thepractical result of these studies was a CF1 cellulose affinity purification scheme for three CenA-150PhoA fusion protein variants based on multiple cycles of crude sample loading and water elution.The purification scheme used smaller column volumes than equivalent single-loadingpurifications, and achieved water elution yields of fusion protein of about 80%, approaching thatof single loading purification with guanidine-HC1 elution. Losses of enzyme activity in theflowthrough and wash fractions were always less than 10%. The purification runs took longer tocomplete than corresponding single-loading purifications, but the whole procedure could bereadily programmed on the FPLC system.Two-site binding models have been proposed previously to describe the binding ofcellulolytic enzymes to cellulose (Woodward et al., 1988; Stanlberg et al., 1991). In thissituation, further analysis of such a model is complicated by a number of factors: i) CF1cellulose is a poorly defined binding substrate, and although of high crystallinity, also containsparacrystalline or amorphous regions; ii) alkaline phosphatase is a dimeric enzyme, and it isunclear whether it binds to cellulose with one binding domain or two; iii) evidence from Din et al.(1991) and Nordquist (1992) shows that CBD binding effects physical disruption of cellulose,indicating that the nature of the CF1 cellulose may change during column purification cycles; iv)a high and low affinity site binding model for this system is not amenable to Scatchard bindinganalysis (Gilkes et al., 1992). With regard to statement iv), low angle x-ray scattering analysisindicates that the CBD is of sufficient size to cover several cellobiose lattice units on the cellulosematrix (Gilkes et al., 1992; Pilz et al., 1990; Shen et al., 1991), and therefore in binding wouldoverlap a number of potential binding sites and prevent them from being bound. This binding siteexclusion effect makes traditional Scatchard analysis inappropriate for such a system (McGheeand von Hippel, 1974). The presence of binding sites of varying affinity could be attributed tocrystalline and amorphous regions on the cellulose matrix, or binding sites on different crystalfaces of cellulose microfibrils (Henrissat et al., 1988). There is a recently reported example of aCBD which binds to amorphous cellulose but not to crystalline cellulose (Coutinho et al., 1992),so it is reasonable to assume that the strength as well as the specificity of the binding interaction isinfluenced by the nature of the cellulose matrix. The geometry of the available binding sites maydirect CenA-PhoA fusion proteins to bind with one or both CBDs. Analysis of relative affinity151binding constants for the fusion protein relative to CBDC enA on a more defined crystallinecellulose substrate, such as bacterial microcrystalline cellulose (Gilkes et al., 1992), couldprovide information on the question of single or double CBD-binding. A higher relative affinityconstant for CBDPT-'PhoA would reflect use of both of the CBDs on the fusion protein dimerfor binding (Nordquist, 1992); however, an equivalent or lesser relative affinity constant wouldnot preclude partial binding of two CBDs.4.5 Proteolytic cleavage of CenA-PhoA fusion proteins.Two strategies were used for proteolytic cleavage of CenA-PhoA fusion proteins: site-specific cleavage by Factor X a at an engineered recognition site, and conformation specificcleavage by C. fimi protease. In each case, modifications to the original form of the fusionprotein, CBDPT-'PhoA, were necessary to optimize the removal of the affinity tag. Generalapproaches using different cleavage sites or cleavage methods to optimize the release of theaffinity tag from the target polypeptide have been reported (Forsberg et al., 1991; Yasukawa andSaito, 1990; Dykes et al., 1988; Forsberg et al., 1992).Factor Xa cleaved the monomer form of CBDPT-'PhoA after Arg-11 of alkalinephosphatase, with the non-consensus recognition site Leu-Glu-Asn-Arg. Trypsin cleaves nativealkaline phosphatase after Arg-11, and the resulting protein has a slightly altered conformationand a 20% reduction in specific activity (Roberts and Chlebowski, 1984). Interestingly, only themonomer form of CBDPT-'PhoA was cleaved by Factor X a, suggesting a different conformationaround the fusion junction in the dimer and monomer forms. The Ile-Glu-Gly-Arg consensusrecognition sequence for Factor Xa comes from comparison of the sequences cleaved by thisprotease in vivo, but studies with chromogenic peptide substrates have indicated that the aminoacid recognition sequence is less specific (Lottenberg et al., 1986). Carter (1990) has compiledsome examples of non-consensus Factor X a cleavage sites in fusion proteins, and from this studythree more can be added to the list, as shown below:Leu-Glu-Asn-ArgtAla-Ala in CBDPT-'PhoA152Tyr-Thr-Gly-Arg ,IJAla-Val in CenA'-IL2Gln-Gly-Tyr-ArgllAla-Trp in CenA'-IL2Cys-Asn-Gly-Arg1lTrp-ValSer-Leu-Ser-ArgilMet-'ThrAla-Leu-Ala-Arglitys-TyrAla-Asn-Phe-Val-LystAsn-Ala-Lys-ilLys-Tyr-Asp-ProVal-Pro-Gly-Argt•in Carter, 1990Lauritzen et al., 1991The only feature these sites have in common is a basic amino acid in the P1 position.Introduction of alkaline phosphatase amino acids 2-5 into CBDPT-'PhoA, to produce CBDPT-PhoA, facilitated Factor Xa cleavage following the consensus recognition site. Modification ofsequences surrounding the protease cleavage site to optimize cleavage has been reported for otherfusion proteins (Guan and Dixon, 1991; Markmeyer et al., 1990; Ellinger et al., 1989).The cleavage of recombinant CenA and Cex by C. fimi protease at equivalent positionsfollowing the Pro-Thr linkers of the respective enzymes (Gilkes et al., 1988) led to the initialinterpretation that the protease activity was specifically directed at those sites. However,subsequent studies (Gilkes et al., 1989) showed that denatured CenA was rapidly fragmented intosmall peptides by the protease, suggesting that its activity was not sequence specific but merelydirected against proteolytically sensitive areas of the native polypeptides. The term C. fimiprotease refers to a relatively crude preparation which contains at least one protease activity(Gilkes et al., 1988). The use of this protease was justified by the fact that alkaline phosphataseis relatively resistant to proteolysis (Schlesinger, 1965). CBDPT-'PhoA was cleaved by C. fimiprotease at at least three positions, at either end of the Pro-Thr linker. Subsequent removal of thePro-Thr linker from CBDPT-'PhoA effectively condensed these proteolytically sensitive regions,to yield a more homogeneous proteolysis product. Although the use of C. fimi protease is notindicated when proteolysis at a specific site adjoining the CBD and target protein is required  it isa cheap alternative when the target protein is resistant to proteolysis and an exact N-terminal153sequence is not critical. Situations where this might apply include generation of purified antigensfor antibody production and recovery of enzymatic or biological activity of the target protein byremoval of the CBD. Relatively non-specific proteases such as trypsin or chymotrypsin havebeen used for release of affmity tags (Dahlman et al., 1989; Taylor and Drickamer, 1991), inwhich cases the target proteins were resistant to degradation. The low cost of producing C. firniprotease is relevant, as protease cleavage can be the most expensive purification step, especiallywhen high purity preparations of some of the site specific proteases, such as Factor X a, are used.In fact, there are reports in the literature where Factor X a cleavage sites have been engineered intofusion proteins, but trypsin has been used for the cleavage step (Varadarajan et al., 1985; Taylorand Drickamer, 1991).Removal of the affmity tag following proteolytic cleavage is routinely achieved by asecond affinity column passage (Mama et al., 1988; Smith and Johnson, 1988; Hammarberg etal., 1989), and was demonstrated in this study for CBDPT-PhoA digested with Factor X a and forCBD-'PhoA digested with C. fimi protease. This step does not usually remove the contaminatingprotease from the target protein, and additional purification steps may be needed to achieve thedesired level of purity. With the CBD affinity tag system, neither of the proteases examined haveaffinity for cellulose (Results, Figure 3.25; Langsford, 1988); however incorporating anotherseparation functionality into the cellulose matrix, such as ion exchange or size exclusioncapability, could facilitate affinity tag removal and further purification of the target protein in asingle column operation. Such an approach has been taken for fractionation of Aspergillus nigercellulases (Boyer, 1987); the resolving power of DEAE cellulose for different cellulasecomponents was attributed to a combination of ion exchange and biospecific binding effects.Limiting factors to this approach would be the effect of the bound CBD on the functionality of thematrix and the ability to regenerate and reuse the matrix. Another approach to separation of theprotease and target protein is the chemical immobilization of the protease and processing of theaffinity tagged fusion protein by passage through the immobilized protease column (Taylor andDrickamer, 1991; Forsberg et al., 1992). A variety of proteases have been immobilized withretention of activity and are available commercially (e.g. United States Biochemical). An154ingenious variation of this approach was used to cleave, purify and select for active mutants ofbovine pancreatic trypsin inhibitor in a single step on an immobilized chymotrypsin column(Altman et al., 1991). Incorporation of an affinity tag into the protease used for site specificcleavage will be a new chapter in the affmity tag purification technology, as this will allowremoval of both the affinity tag and the protease from the target protein, following proteolyticcleavage, by a second passage over the affmity matrix. Major challenges to this approach will bethe production of an active recombinant protease and the prevention of affinity tag loss throughautoproteolysis. Our laboratory has recently expressed a Factor X-CBDcex gene fusion inmammalian cells and has demonstrated cellulose binding and, following activation, proteolyticactivity of the recombinant polypeptide (Z. Assouline, submitted for publication).CBDCenA is susceptible to proteolysis by trypsin and chymotrypsin (Gilkes et al., 1989),but contains no internal methionine residues, and would therefore be suitable for affinity tagpurifications involving cyanogen bromide cleavage (Fontana and Gross, 1986). The CBDcontains one Asp-Pro dipeptide and five Asn-Gly dipeptides, making it unsuitable for affinity tagpurifications involving cleavage by low pH or hydroxylamine, respectively. Currently, nomutagenesis has been done on the CBD to modify its sensitivity to cleavage; such an approachhas been taken for the IgG binding domains from protein A, where an internal Asn-Gly sequencewas altered to Asn-Ala to enable subsequent use of hydroxylamine to remove the functionalaffinity tag from the target protein (Nilsson et al., 1987). In a recent report (Miller et al., 1992),the Pro-Thr linker of CenA was replaced with the hinge sequence of human IgAl, which wasspecifically cleaved by Neisseria gonorrhoeae IgA proteases.4.6 Expression of CenA-IL2 fusion proteins.The ease of expression and purification of a particular affinity-tagged fusion polypeptidedoes not mean that the system can be applied to any target protein with equal success. This wasevident with the documented attempts to express CenA-1L2 fusion proteins in E. coli. Thedirection of CenA-IL2 fusion proteins to the periplasm of E. coli by the leader peptide of CenA155was intended to facilitate correct folding of IL2 and disulphide bond formation, and was arequirement to justify use of the affinity tag approach over high level cytoplasmic expression ofthe mature polypeptide in inclusion bodies (Devos et al., 1983; Bailon et al., 1987). There is onereport of periplasmic accumulation of an IL2 fusion with diphtheria toxin (Williams et al., 1987),but there was no mention of the levels of expression of the recombinant polypeptide. CenA 1-IL2was produced at low levels in E. coli and was subject to extensive proteolysis. Although anexhaustive analysis of induction conditions was not carried out, the level of accumulation of intactCenA'-IL2 in E. coli BL21 was virtually independent of the induction conditions used. This mayhave been due in part to the lack of a constitutively expressed lac repressor in BL21, but alsosuggests that proteolytic sensitivity and not the level of expression was the limiting factor in theaccumulation of intact fusion protein. The intact fusion protein that did accumulate wasassociated with the cytoplasmic membrane. The reason for this membrane association is notknown, but there are two possible explanations: i) the IL2 polypeptide does not readily traversethe cytoplasmic membrane because of its relatively high content of basic amino acids, and ii)aggregation of the cytoplasmic portion of the polypeptide during translocation interferes with thetranslocation process. There is mounting evidence that positively charged amino acids near themature N-terminus of a polypeptide can have an inhibitory effect on membrane translocation in E.coli (reviewed in Boyd and Beckwith, 1990). Human IL2 contains 2 lysine residues at positions8 and 9 in the mature polypeptide, but replacement of these residues with an uncharged sequencefrom murine IL2 did not facilitate export (van Kimmenade et al., 1989). In addition, thenecessity for a polypeptide to be in a soluble state for translocation competency has beendocumented (Leemans et al., 1989). IL2 is subject to aggregation in the E. coli cytoplasm(Bailon et al., 1987; Liang et al., 1985), and the resistance of CenA'-IL2 to solubilizationsuggested that protein aggregation had occurred, so it is possible that aggregation prevented themembrane translocation of the polypeptide. Intact CenA'-IL2 was processed by leader peptidase,as the purified polypeptide had a mature N-terminus, so at least part of the fusion protein wastranslocated across the cytoplasmic membrane. It is possible that the CBD portion of CenA'-IL2passed through the membrane and folded to form a functional CBD, while the IL2 portion156aggregated in the cytoplasm and blocked complete secretion. This would explain the associationof CenA'-1L2 with the cytoplasmic membrane as well as the ability of the insoluble protein tobind to cellulose.The faint immunoreactive band observed on some Western blots of CenA'-M2 just abovethe 43,000 molecular weight CenA'-IL2 band was postulated to be the unprocessed form of thefusion protein. It could also represent the intact mature fusion protein and the major band belowit a C-terminal proteolysis product; however, the polypeptide released by Factor X a digestion ofCenA'-IL2 which had the mature N-terminal sequence of M2 was approximately 15,000molecular weight, the expected size for human IL2. It is most likely, therefore, that the precursorform of the fusion protein does accumulate to some extent, probably by aggregation in thecytoplasm.The mode of expression of CBDPT-1L2 was not studied in great detail, but, like CenAt-IL2, the fusion polypeptide was subject to extensive proteolysis, and the intact form was presentonly at low levels. These were the main factors which detracted from the E. coli expression of1L2 with a CBD affinity tag for purification. The difficulties experienced in the production ofthese fusion proteins emphasise the need for a flexible approach to the production of heterologousgene products in recombinant hosts. In work not presented in this thesis, removal of the CenAleader peptide facilitated intracellular accumulation of high levels of CBDPT-IL2 in E. coli. Thisfusion protein has also been produced in mammalian cells, and exhibits both biological activityand affinity for cellulose (J. Alimonti, unpublished results). Initial attempts to produce CBDPT-IL2 in yeast also look promising (E. Ong, unpublished results).4.7 Purification and proteolytic cleavage of CenA-IL2 fusion proteins.Although the nature and levels of expression of CenA'-1L2 and CBDPT-1L2 observed inthis study would realistically preclude the use of the described expression system for productionand purification of human interleukin-2, these fusion proteins nonetheless provided a usefulplatform to study the properties of the CenA cellulose-binding domain as an affinity tag. Binding157of CenA-IL2 fusion proteins to Avicel, even from the insoluble cell fraction, was shown to resultin a significant degree of purification. Binding of insoluble CenA'-1L2 to Avicel indicated that theCBD still retained its function when the fusion protein was associated with the cytoplasmicmembrane. Passage of E. coli cells through a French press has been reported to formpredominantly inverted membrane vesicles (Futai, 1974), which partly explains why sonicationof the membrane fraction, which would cause disruption of membrane vesicles (Schnaitman,1981), was necessary to effect complete binding of the fusion protein to Avicel. The reduction inparticle size and hence increase in surface area of the Avicel would also contribute to the increasedbinding.The solubilization of CenA-IL2 fusion proteins at pH 11.5 facilitated their subsequentbinding to Avicel. Solubilization of proteins at high pH has been reported (Marston et al., 1984;Fronticelli et al., 1991), but under these conditions proteins are susceptible to undesirablemodifications, such as deamidation (Liu, 1992), disulfide exchange (Marston et al., 1984), 13-elimination and partial peptide bond hydrolysis (Volkin and Klibanov, 1989). Surprisingly, theCBD retained its function at this pH and facilitated the quantitative binding of both fusion proteinsto Avicel. This finding, in combination with the binding of CenA'-'PhoA IX-8 to Avicel at pH 2,indicates an impressive range of pH over which CBDCenA can bind to cellulose.CenA'-IL2 bound tightly to Avicel, and significant elution was only effected by stronglydenaturing conditions. For this reason the Factor X a digestion of this polypeptide and CBDPT-1L2 was carried out on the Avicel-bound fusion polypeptide. The removal of affinity tags whilethe fusion protein is immobilized has been reported, either as an optimization of the purificationprocedure (Guan and Dixon, 1991; Gearing et al., 1989), or in situations where the elutionconditions caused inactivation of the target protein (Dahlman et al., 1989). The observedcleavage of CenA'-IL2 at two non-consensus sites within the CenA catalytic domain sequencehighlights the need to use intact domain structures when engineering affinity tags for proteinpurification. The variable stability of the CenA'-'PhoA fusion proteins also emphasises thispoint. The association of IL2 with Avicel following Factor X a digestion of CBDPT-IL2 did notcorrespond to the behaviour of pure recombinant IL2, which was shown to have no affinity for158Avicel. The reason for this Avicel association is not known, but it is likely that the polypeptidewas in a denatured form at pH 11.5. As mentioned above, disulfide exchange is enhanced at highpH; as IL2 normally contains one disulfide bond and one free cysteine residue, it is possible thatmismatched intra and intermolecular disulfide bonds formed at this pH and led to aggregation orprecipitation of the polypeptide as the pH was reduced. In the context of Avicel this could haveresulted in entrapment of the polypeptide in the cellulose matrix. Avicel comprises a mixture ofrod shaped particles and irregular particle aggregates with intraparticulate pores andinterparticulate voids (Marshall and Sixsmith, 1974), and could conceivably entrap an aggregatedpolypeptide. A report on immobilized glucoamylase II from Aspergillus niger observed a lowerrecovery of activity on refolding the denatured immobilized enzyme relative to the soluble formand ascribed this result to entanglement of the polypeptide in the support matrix (Gottschalk andJaenicke, 1991).4.8 The nature of CBD binding and elution.Some of the most detailed structural information on protein-carbohydrate interactionscomes from X-ray crystallographic studies on E. coli periplasmic sugar-binding proteins, mostnotably arabinose-binding protein (Quiocho, 1988), glucose/galactose-binding protein (Vyas etal., 1988), and maltose-binding protein (Spurlino et al., 1991). The primary protein-sugarbinding interactions are hydrogen bonds between charged and polar amino acid side chains andthe hydroxyl groups on the sugar. In addition, protein-sugar complexes are further stabilized byvan der Waals interactions, in particular the stacking of aromatic amino acid side chains againsthydrophobic patches on the six-membered sugar ring (Quiocho, 1988). CBDs have a number ofdistinctions from these well-characterized carbohydrate-binding proteins, most notably the factthat they are much smaller (CBDCenA contains 111 amino acids; maltose and arabinose bindingproteins contain 370 and 306 amino acids respectively) and bind to an insoluble crystallinematrix, suggesting that they present a binding face rather than a deep binding pocket or groove(Henrissat et al., 1988). The 3-dimensional structure of the CBD of T. reesei CbhI indicates the159presence of a potential binding face or faces (Reinikainen et al., 1992). In addition, CBDscontain relatively few charged amino acids, which indicates that any hydrogen bonding betweenthe CBD and cellulose is mediated primarily by polar amino acid side chains. This is supportedby the conservation of asparagine residues and the high contents of hydroxyamino acids inbacterial CBDs (Gilkes et al., 1991), as well as the relative pH independence of CBDCenAbinding shown in this study. Furthermore, the water elution phenomenon reported here andelsewhere (Owolabi et al., 1988; Ong et al., 1989; Coutinho et al., 1992; Gilbert et al., 1990) issuggestive of some form of hydrophobic interaction (Kennedy, 1990), the contribution of whichto binding of the E. coli periplasmic carbohydrate-binding proteins to their ligands is consideredto be minor (Quiocho, 1988). Preliminary evidence that conserved tryptophan residues inCBDCenA and a related CBD from Pseudomonas fluorescens subspecies cellulosa are involved incellulose binding has come from site-directed mutagenesis studies (N. Din, submitted forpublication; Poole et al., 1993). Exposure of tryptophan residues on the binding surface(s) of theCBD would contribute to the hydrophobic effect.CBDCenA was shown to retain its binding function at pH 2.0 (for CenA'-'PhoA IX-8)and pH 11.5 (for CenA'-IL2 and CBDPT-IL2). This is probably due in part to the low chargedensity on the polypeptide; with only three acidic and three basic amino acids, the CBD would beless subject to unfolding caused by mutual repulsion of like charges at extremes of pH than woulda polypeptide with higher charge density. Low charge density is a feature of CBDs from bothbacterial and fungal sources (Gilkes et al., 1991). The CBD from Cex has only four chargedamino acids, and also appears to bind to cellulose over a wide range of pH (Ong et al., 1993; P.Tomme and R. Graham, unpublished results); however, when fused to a P-glucosidase from anAgrobacterium sp.it can be desorbed from cellulose above pH 9 (Ong et al., 1991). The reasonfor this difference is not known. Desorption at alkaline pH has also been reported forTrichoderma reesei cellulases (Otter et al., 1989).CenA-PhoA fusion proteins bound stably to CF1 cellulose at high ionic strength but werepartially desorbed with distilled water. This behaviour is distinct from that of CenA andCBDCenA which do not elute from CF1 cellulose with distilled water (H. Damude, E. Ong, N.160Din, unpublished data). A similar discrepancy occurs for Cex (Gilkes et al., 1988) and Abg-CBDCex (Ong et al., 1989), which are desorbed from CF1 cellulose with water, and CBDce x(Ong et al., 1993a), which is not. Without structural information it is hard to predict the cause ofthis effect, but a simple explanation is that polypeptides which elute with water have fewerbinding interactions with the cellulose matrix due to steric interference of the accompanyingdomain with CBD binding. One could speculate that in the water-eluting polypeptides theadjoining domain causes some steric interference with the binding of the CBD, and the bindingaffinity is reduced to the extent that water elution can occur. The dimeric nature of CBD-PhoAfusion proteins is not necessarily responsible for the water elution effect, as a CBDCenA fusionwith myoglobin, a monomeric polypeptide, has been shown to elute from CF1 cellulose withwater (C. Hunter, personal communication).From the above discussion and the introductory comments it is evident that the binding ofCBDCenA and related CBDs to cellulose is mediated by a combination of hydrogen bonds, vander Waals interactions and the hydrophobic effect. Ionic interactions can be ruled out, becausethe cellulose polymer is uncharged, CBDs have a low charge density, and binding to cellulose isstable at high ionic strength and over a wide range of pH. The exact nature and extent of thesebinding interactions are not known. The insoluble and heterogeneous nature of cellulose is amajor bather to overcome in the dissection of the various binding interactions, and indeed, theinteractions of different CBDs with cellulose may be quite distinct. The determination of 3-dimensional structures by x-ray crystallographic or NMR spectroscopic techniques will be amajor step in understanding the interactions between CBDs and their substrate.4.9 CBDs as affinity tags for protein purification: how do they measure up?This study has shown that CBDCenA has significant potential for use as an affinity tag forprotein purification, particularly in situations where the fusion polypeptide can be desorbed fromcellulose at low ionic strength. The CBD, which contains 111 amino acid residues, is relativelysmall compared to other binding domains used as affinity tags, such as maltose binding protein161(370 amino acids) and glutathione S-transferase (218 amino acids); this is a desirable feature forreducing the synthetic load on the host cell and facilitating the biological activity of the fusionpartner. CBDcenA is preceded by a cleavable leader peptide which is capable of directing exportof fusion proteins to the periplasm of E. coli. Cellulose is cheap relative to most affinity matricesand is readily available in a wide variety of forms, such as powders, papers and membranes, andcotton. The inert and hydrophilic nature of cellulose also make it suitable for use in affinityadsorption where stability and lack of non-specific binding are important qualities. CBDc enA isbound strongly and specifically to cellulose over a wide range of pH and at high ionic strength,important considerations for both binding of a fusion polypeptide to cellulose and for removal ofnon-specifically bound protein. The ability to elute fusion polypeptides with water adds to theversatility of subsequent buffering steps and eliminates the need to remove low molecular weightligands from the eluted polypeptide (Maina et al., 1988; Smith and Johnson, 1988; Hoffman andRoeder, 1991). Optimized water elution of CenA-PhoA fusion proteins resulted in purificationyields of around 80%. This is comparable to yields reported for purifications with other affinitytags, such as maltose-binding protein (85%, Bedouelle and Duplay, 1988; 90%, Di Guan et al.,1988), cyclomaltodextrin glucanotransferase (90%, Hellman and Maritsala, 1992), protein A(80%, Nilsson et al., 1985: 95%, Abrahmsen et al., 1986)and polyhistidine (88%, Hochuli et al.,1988). Not all proteins are stable at low ionic strength (Bureau and Daussant, 1983), but it isbelieved that this approach would be applicable to a wide range of target polypeptides. Theresults shown here indicate that the CBD is amenable to proteolytic removal from the targetprotein, even when bound to cellulose. The absence of methionine residues in the CBD indicatesthat cyanogen bromide may also be used for affinity tag removal. The versatility of the affinitytag approach depends on the functional fusion of the tag to either the N- or C-terminus of thetarget polypeptide. The location of native CBDs at the N- and C- termini of P-1,4-glucanasesmakes them ideal candidates in this regard, and results presented here and elsewhere (Ong et al.,1989; Kilburn et al., 1992) support this fact.One disadvantage of cellulose as an affinity matrix is its relatively low operational binding capacity for fusion protein. CF1 cellulose columns in our laboratory are routinely loaded at 1 mg162fusion protein/g cellulose, and for the cycle purification technique a loading level of 0.67 mg/gwas found to be appropriate. As CF1 cellulose columns have a bed volume of about 5 mug dryweight, the operational capacity is 0.13-0.2 mg fusion protein/ml. This is much lower thancapacities reported for commercially available affinity matrices, such as cross-linked amylose formaltose-binding protein fusions (2 mg/ml; New England Biolabs), Ni 2+-NTA resin forpolyhistidine fusions (5 mg/ml; Qiagen), and glutathione agarose for glutathione S-transferasefusions (8 mg/ml; Smith and Johnson, 1988). This difference mainly translates into longerpurification times and greater wash volumes, but the extra cost of reagents required, most notablythe elution agent, is insignificant.At this point the main disadvantage of CBDcenie, as an affmity tag is the unpredictablenature of the elution process. CenA-PhoA fusion proteins could be partially eluted from CF1cellulose with distilled water, whereas CenA'-IL2 could not. Elution experiments with CenA 1-1L2 indicated that soluble cellulose analogues were ineffective at competitive elution of thepolypeptide, and a similar result has been reported for T. reesei cellulases (Reese, 1982).Specific elution methods, such as competition with free ligand, have the advantage of promotingless desorption of non-specifically bound material from an affinity matrix (Ostrove, 1990), but,as mentioned above, they also contaminate the eluted protein with competitive ligand which mayrequire subsequent removal. A better understanding of the interactions between the CBD and itssubstrate will direct mutagenesis approaches to enhance water elution potential. Weakening thebinding interaction may not in itself be sufficient. Two mutants of CBDCenA with tryptophanresidues replaced by alanine have reduced affinity for cellulose but are not eluted with water (N.Din, unpublished results). Modification of hydrogen bonding residues, once they are identified,may weaken the binding affinity but maintain the hydrophobic effect that appears to be necessaryfor elution. An approach with more potential in the short term is the use of the N1 CBD from C.Jimi endoglucanase C (Coutinho et al., 1992) as an affinity tag. This CBD has been shown tobind to amorphous cellulose and elute with distilled water, and has been purified on Avicel usingthis approach.1635. BibliographyAbate, C., Luk, D., Gentz, R., Rauscher III, F.J. and Curran, T. (1990). Expression andpurification of the leucine zipper and DNA-binding domains of Fos and Jun: both Fos and Juncontact DNA directly. Proc. Natl. Acad. Sci. USA 87, 1032-1036.Abrahms6n, L., Moks, T., Nilsson, B. and Uhlen, M. (1986). Secretion of heterologous geneproducts to the culture medium of Escherichia coli. Nucleic Acids Res. 14, 7487-7500.Allen, G. and Henwood, C.A. (1987). Purification of recombinant epidermal growth factor froma fusion protein expressed in Escherichia coli. In Protein Purification: Micro to Macro; UCLAsymposia on molecular and cellular biology, ed. R. Burgess. Alan Liss, Inc., New York; pp.367-373.Altman, J.D., Henner, D., Nilsson, B., Anderson, S. and Kuntz, I.D. (1991). Intracellularexpression of BPTI fusion proteins and single column cleavage/affinity purification bychymotrypsin. Protein Eng. 4, 593-600.Amann, E., Ochs, B. and Abel, K.-J. (1988). Tightly regulated tac promoter vectors useful forthe expression of unfused and fused proteins in Escherichia coli. Gene 69, 301-315.Applebury, M.L. and Coleman, J.E. (1969). Escherichia coli alkaline phosphatase. Metalbinding, protein conformation, and quaternary structure. J. Biol. Chem. 244, 308-318.Armstrong, J.M. (1964). The molar extinction coefficient of 2,6-dichlorophenol indophenol.Biochim. et Biophys. Acta 86, 194-197.Arnold, F.H. (1991). Metal-affinity separations: a new dimension in protein processing.Bio/Technology 9, 151-156.Atlan, D. and Portalier, R. (1984). Optimized extracellular production of alkaline phosphatase bylky mutants of Escherichia coli K12. Appl. Microbiol. Biotechnol. 19, 5-12.Au, K.-S. and Chan, K.-Y. (1987). Purification and properties of the endo-1,4-13-glucanasefrom Bacillus subtilis. J. Gen. Microbiol. 133, 2155-2162.Avrameas, S., Taudou, B. and Ternynck, T. (1971). Specificity of antibodies synthetized byimmunocytes as detected by immunoenzyme techniques. Int. Arch. Allergy 40, 161-170.Bailon, P., Weber, D.V., Keeney, R.F., Fredericks, J.E., Smith, C., Familletti, P.C. andSmart, J.E. (1987). Receptor-affinity chromatography: a one-step purification for recombinantinterleukin-2. Bio/Technology 5, 1195-1198.Baksh, S., Burns, K., Busaan, J. and Michalak, M. (1992). Expression and purification ofrecombinant and native calreticulin. Protein Expr. Pur. 3, 322-331.Baratte, B., Meijer, L., Galaktionov, K. and Beach, D. (1992). Screening for antimitoticcompounds using the cdc25 tyrosine phosphatase, an activator of the mitosis-inducingp34cdc2/cyclin Bcdc13 protein kinase. Anticancer Research 12, 873-880.Basu, M., Dharm, E., Levine, J.F., Kramer, R.A., Crowl, R.M. and Campbell, R.M. (1991).Expression of growth hormone-releasing^ "0purification and characterization of the expressed protein. Archives of Biochemistry andBiophysics 286, 638-644.164Bedouelle, H. and Duplay, P. (1988). Production in Escherichia coli and one-step purification ofbifunctional hybrid proteins which bind maltose. Eur. J. Biochem. 171, 541-549.Beguin, P. and Eisen, H. (1978). Purification and partial characterization of three extracellularcellulases from Cellulomonas sp. Eur. J. Biochem. 87, 525-531.Beguin, P., Gilkes, N.R., Kilburn, D.G., Miller, R.C., Jr., O'Neill, G.P. and Warren, R.A.J.(1987). Cloning of cellulase genes. Crit. Rev. Biotechnol. 6, 129-162.Berg, D.E., Lodge, J., Sasakawa, C., Nag, D.K., Phadnis, S.H., Weston-Hafer, K. and Carle,G.F. (1984). Transposon Tn5: specific sequence recognition and conservative transposition. InProceedings: Cold Spring Harbor Symposium on Quantitative Biology, Cold Spring Harbor,NY. Cold Spring Harbor Press; pp. 215-226.Berthold, H., Scanarini, M., Abney, C.C., Frorath, B. and Northemann, W. (1992).Purification of recombinant antigenic epitopes of the human 68-kDa (U1) ribonucleoproteinantigen using the expression system pH6EX3 followed by metal chelating affinitychromatography. Protein Expr. Pur. 3, 50-56.Blackwell, J. (1982). The macromolecular organization of cellulose and chitin. In Cellulose andOther Natural Polymer Systems. Biogenesis, Structure, and Degradation, ed. R.M. Brown Jr.Plenum Press, New York; pp. 403-428.Blake, M.S., Johnston, K.H., Russell-Jones, G.J. and Gotschlich, E.C. (1984). A rapid,sensitive method for detection of alkaline phosphatase-conjugated anti-antibody on Western blots.Anal. Biochem. 136, 175-179.Blondel, A. and Bedouelle, H. (1990). Export and purification of a cytoplasmic dimeric proteinby fusion to the maltose-binding protein of Escherichia coli. Eur. J. Biochem. 193, 325-330.Boon, K., Vijgenboom, E., Madsen, L.V., Talens, A., Kraal, B. and Bosch, L. (1992).Isolation and functional analysis of histidine-tagged elongation factor Tu. Eur. J. Biochem. 210,177-183.Bornstein, P. and Balian, G. (1977). Cleavage at Asn-Gly bonds with hydroxylamine. MethodsEnzymol. 47, 132-145.Boutelje, J., KarlstrOm, A.R., Hartmanis, M.G.N., Holmgren, E., Sjogren, A. and Levine,R.L. (1990). Human immunodeficiency viral protease is catalytically active as a fusion protein:characterization of the fusion and native enzymes produced in Escherichia coli. Archives ofBiochemistry and Biophysics 283, 141-149.Boyd, D. and Beckwith, J. (1990). The role of charged amino acids in the localization ofsecreted and membrane proteins. Cell 62, 1031-1033.Boyer, R.F. (1987). Fractionation of Aspergillus niger cellulases by combined ion exchangeaffinity chromatography. Biotechnol. Bioeng. 29, 176-179.Bradford, M.M. (1976). A rapid and sensitive method for the quantitation of microgramquantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72, 248-254.Brewer, S.J. and Sassenfeld, H.M. (1985). The purification of recombinant proteins using C-terminal_polyarginine fusion& Tre-nds-Bioteehftel. i3 119=122.165Brewer, S.J., Haymore, B.L., Hopp, T.P. and Sassenfeld, H.M. (1991). Engineering proteinsto enable their isolation in a biologically active form. In Purification and Analysis ofRecombinant Proteins, ed. R. Seetharam and S.K. Sharma. Marcel Dekker, Inc., New York;pp. 239-266.Brissette, L., Cahuzac-Bec, N., Desforges, M., Bec, J.L., Marcel, Y.L. and Rassart, E. (1991).Expression of recombinant human apolipoprotein A-I in Chinese hamster ovary cells andEscherichia coli. Protein Expr. Pur. 2, 296-303.Bureau, D. and Daussant, J. (1983). Desorption following immunoaffinity chromatography;generalization of a gentle procedure for desorbing antigen. J. Immunol. Methods 57, 205-213.Bush, G.L., Tassin, A.M., Friden, H. and Meyer, D.I. (1991). Secretion in yeast. Purificationand in vitro translocation of chemical amounts of prepro-a-factor. J. Biol. Chem. 266, 13811-13814.Caligiuri, M.A., Murray, C., Soiffer, R.J., Klumpp, T.R., Seiden, M., Cochran, K., Cameron,C., Ish, C., Buchanan, L., Perillo, D., Smith, K. and Ritz, J. (1991). Extended continuousinfusion low-dose recombinant interleukin-2 in advanced cancer: prolonged immunomodulationwithout significant toxicity. J. Clin. Oncol. 9, 2110-2119.Calos, M.P. (1978). DNA sequence for a low-level promoter of the lac repressor gene and an'up' promoter mutation. Nature 274, 762-769.Campbell, D.H., Luescher, E. and Lerman, L.S. (1951). Immunologic adsorbents. 1. Isolationof antibody by means of a cellulose-protein antigen. Proc. Natl. Acad. Sci. USA 37, 575-578.Carr, D.W., Stofko-Hahn, R.E., Fraser, I.D.C., Bishop, S.M., Acott, T.S., Brennan, R.G.and Scott, J.D. (1991). Interaction of the regulatory subunit (RII) of cAMP-dependent proteinkinase with RH-anchoring proteins occurs through an amphipathic helix binding motif. J. Biol.Chem. 266, 14188-14192.Carter, P. (1990). Site-specific proteolysis of fusion proteins. In Protein Purification: FromMolecular Mechanisms to Large-Scale Processes; ACS Symposium Series No. 427, ed. M.R.Ladisch et al. American Chemical Society, Washington DC; pp. 181-193.Carter, P. and Wells, J.A. (1987). Engineering enzyme specificity by "substrate-assistedcatalysis". Science 237, 394-399.Caspers, P., Etlinger, H., Matile, H., Pink, J.R., Stiiber, D. and Takacs, B. (1991). APlasmodium falciparum malaria vaccine candidate which contains epitopes from thecircumsporozoite protein and a blood stage antigen, 5.1. Mol. Biochem. Parasitol. 47, 143-150.Chattopadhyay, D., Evans, D.B., Deibel, M.R., Jr., Vosters, A.F., Eckenrode, F.M., Einspahr,H.M., Hui, J.0., Tomasselli, A.G., Zurcher-Neely, H.A., Heinrikson, R.L. and Sharma, S.K.(1992). Purification and characterization of heterodimeric human immunodeficiency virus type 1(HIV-1) reverse transcriptase produced by in vitro processing of p66 with recombinant HIV-1protease. J. Biol. Chem. 267, 14227-14232.Chen, L., Ford, C. and Nikolov, Z. (1991). Adsorption to starch of a 13-galactosidase fusionprotein containing the starch-binding region of Aspergillus glucoamylase. Gene 99, 121-126.Chen., L., Ford, C., Kusriadi, A. and Nikolov-, Z.L. (1991a). Improved adsorptiort ter starch -of aP-galactosidase fusion protein containing the starch-binding domain from Aspergillusglucoamylase. Biotechnol. Prog. 7, 225-229.166Clement, J.M., Charbit, A., Martineau, P., O'Callaghan, D., Szmelcman, S., Leclerc, C. andHofnung, M. (1991). Bacterial vectors to target and/or purify polypeptides: their use inimmunological studies. Annales De Biologie Clinique 49, 249-254.Coleman, J.E. and Gettins, P. (1983). Alkaline phosphatase, solution structure, and mechanism.Advan. Enzymol. 55, 381-452.Cooper, S. and Ruettinger, T. (1975). A temperature sensitive nonsense mutation affecting thesynthesis of a major protein of Escherichia coli K12. Mol. Gen. Genet. 139, 167-176.Coutinho, J.B. (1992a). Structure-function relationships of endoglucanase C (CenC) fromCellulomonas fimi. Ph.D. thesis, University of British Columbia.Coutinho, J.B., Gilkes, N.R., Warren, R.A.J., Kilburn, D.G. and Miller, R.C., Jr. (1992).The binding of Cellulomonas fimi endoglucanase C (CenC) to cellulose and Sephadex is mediatedby the N-terminal repeats. Mol. Microbiol. 6, 1243-1252.Coutinho, J.B., Moser, B., Kilburn, D.G., Warren, R.A.J. and Miller, R.C., Jr. (1991).Nucleotide sequence of the endoglucanase C gene (cenC) of Cellulomonas fimi, its high-levelexpression in Escherichia coli, and characterization of its products. Mol. Microbiol. 5, 1221-1233.Cronan, J.E., Jr. (1990). Biotination of proteins in vivo. J. Biol. Chem. 265, 10327-10333.Csopak, H., Falk, K.-E. and Szajn, H. (1972). Effect of EDTA on Escherichia coli alkalinephosphatase. Biochim. et Biophys. Acta 258, 466-472.Csopak, H., Garellick, G. and Hallberg, B. (1972a). Purification of Escherichia coli alkalinephosphatase. Improved growth conditions for the bacteria, modified methods of preparation ofthe enzyme. Acta Chem. Scand. 26, 2401-2411.Dahlman, K., StrOmstedt, P.-E., Rae, C., JOrnvall, H., Flock, J.-I., Carlstedt-Duke, J. andGustafsson, J.-A. (1989). High level expression in Escherichia coli of the DNA-binding domainof the glucocorticoid receptor in a functional form utilizing domain-specific cleavage of a fusionprotein. J. Biol. Chem. 264, 804-809.Davie, J.R., Wynn, R.M., Cox, R.P. and Chuang, D.T. (1992). Expression and assembly of afunctional El component (alpha 2 beta 2) of mammalian branched-chain alpha-ketoaciddehydrogenase complex in Escherichia coli. J. Biol. Chem. 267, 16601-16606.De Bruijn, F.J. and Lupski, J.R. (1984). The use of transposon Tn5 mutagenesis in the rapidgeneration of correlated physical and genetic maps of DNA segments cloned into multicopyplasmids - a review. Gene 27, 131-149.Derbyshire, K.M. and Grindley, N.D. (1992). Binding of the IS903 transposase to its invertedrepeat in vitro. EMBO J. 11, 3449-3455.DeSerrano, V.S., Menhart, N. and Castellino, F.J. (1992). Expression, purification, andcharacterization of the recombinant kringle 1 domain from tissue-type plasminogen activator.Archives of Biochemistry and Biophysics 294, 282-290.Devos, R., Plaetinck, G., Cheroutre, H., Simons, G., Degrave, W., Tavernier, J., Remaut, E.and Fiers, W. 498.3}. Molecular-cloning-of humaninterlerkin -2- cDNA^ arydit§expres§iaco/i. Nucleic Acids Res. 11, 4307-4323.167di Guan, C., Li, P., Riggs, P.D. and Inouye, H. (1988). Vectors that facilitate the expressionand purification of foreign peptides in Escherichia coli by fusion to maltose-binding protein.Gene 67, 21-30.Din, N., Gilkes, N.R., Tekant, B., Miller, R.C., Jr., Warren, R.A.J. and Kilburn, D.G.(1991). Non-hydrolytic disruption of cellulose fibres by the binding domain of a bacterialcellulase. Bio/Technology 9, 1096-1099.Donaldson, L. and Capone, J.P. (1992). Purification and characterization of the carboxyl-terminal transactivation domain of Vmw65 from herpes simplex virus type 1. J. Biol. Chem.267, 1411-1414.Drickamer, K. (1988). Two distinct classes of carbohydrate-recognition domains in animallectins. J. Biol. Chem. 263, 9557-9560.Dudler, T., Chen, W.-Q., Wang, S., Schneider, T., Annand, R.R., Dempcy, R.O., Crameri,R., Gmachl, M., Suter, M. and Gelb, M.H. (1992). High-level expression in Escherichia coliand rapid purification of enzymatically active honey bee venom phospholipase A2. Biochim. etBiophys. Acta 1165, 201-210.Dykes, C.W., Bookless, A.B., Coomber, B.A., Noble, S.A., Humber, D.C. and Hobden,A.N. (1988). Expression of atrial natriuretic factor as a cleavable fusion protein withchloramphenicol acetyltransferase in Escherichia coli. Eur. J. Biochem. 174, 411-416.Earhart, C.F., Lundrigan, M., Pickett, C.L. and Pierce, J.R. (1979). Escherichia coli K-12mutants that lack major outer membrane protein a. FEMS Microbiol. Lett. 6, 277-280.Ehle, H. and Horn, A. (1990). Immunoaffinity chromatography of enzymes. Bioseparation 1,97-110.Ellinger, S., Glockshuber, R., Jahn, G. and PlUckthun, A. (1989). Cleavage of procaryoticallyexpressed human immunodeficiency virus fusion proteins by Factor Xa and application inwestern blot (immunoblot) assays. J. Clin. Microbiol. 27, 971-976.Evans, D.B., Tarpley, W.G. and Sharma, S.K. (1991). Expression and characterization ofchimeric rDNA proteins engineered for purification and enzymatic cleavage. Protein Expr. Pur.2, 205-213.Ferreira, L.M.A., Durrant, A.J., Hall, J., Hazlewood, G.P. and Gilbert, H.J. (1990). Spatialseparation of protein domains is not necessary for catalytic activity or substrate binding in axylanase. Biochem. J. 269, 261-264.Ferro-Luzzi Ames, G., Prody, C. and Kustu, S. (1984). Simple, rapid, and quantitative releaseof periplasmic proteins by chloroform. J. Bacteriol. 160, 1181-1183.Field, J., Nikawa, J.-I., Broek, D., MacDonald, B., Rodgers, L., Wilson, I.A., Lerner, R.A.and Wigler, M. (1988). Purification of a RAS-responsive adenylyl cyclase complex fromSaccharomyces cerevisiae by use of an epitope addition method. Mol. Cell Biol. 8, 2159-2165.Fikrig, E., Barthold, S.W., Kantor, F.S. and Flavell, R.A. (1990). Protection of mice againstthe Lyme disease agent by immunizing with recombinant OspA. Science 250, 553-556.Fontana, A. and Gross, E. (1986). Fragmentation of polypeptides by chemical methods. InPractical P-roteirt Chemistry-A-Flandbeek,-ed. A. Darbre. Julm ^Wiley and^Sims Ltd., Ne-w-York;pp. 67-120.168Ford, C.F., Suominen, I. and Glatz, C.E. (1991). Fusion tails for the recovery and purificationof recombinant proteins. Protein Expr. Pur. 2, 95-107.Forsberg, G., Baastrup, B., Rondahl, H., Holmgren, E., Pohl, G., Hartmanis, M. and Lake,M. (1992). An evaluation of different enzymatic cleavage methods for recombinant fusionproteins, applied on des(1-3)insulin-like growth factor I. J. Prot. Chem. 11, 201-211.Forsberg, G., Brobjer, M., Holmgren, E., Bergdahl, K., Persson, P., Gautvik, K.M. andHartmanis, M. (1991). Thrombin and H64A subtilisin cleavage of fusion proteins forpreparation of human recombinant parathyroid hormone. J. Prot. Chem. 10, 517-526.Forsberg, G., Palm, G., Ekebacke, A., Josephson, S. and Hartmanis, M. (1990). Separationand characterization of modified variants of recombinant human insulin-like growth factor Iderived from a fusion protein secreted from Escherichia coli. Biochem. J. 271, 357-363.Fronticelli, C., O'Donnell, J.K. and Brinigar, W.S. (1991). Recombinant human hemoglobin:expression and refolding of beta-globin from Escherichia coli. J. Prot. Chem. 10, 495-501.Frorath, B., Abney, C.C., Berthold, H., Scanarini, M. and Northemann, W. (1992).Production of recombinant rat interleukin-6 in Escherichia coli using a novel highly efficientexpression vector pGEX-3T. BioTechniques 12, 558-563.Futai, M. (1974). Orientation of membrane vesicles from Escherichia coli prepared by differentprocedures. J. Membrane Biol. 15, 15-28.Garen, A. and Levinthal, C. (1960). A fine-structure genetic and chemical study of the enzymealkaline phosphatase of E. coli. Biochim. et Biophys. Acta 38, 470-483.Gearing, D.P., Nicola, N.A., Metcalf, D., Foote, S., Willson, T.A., Gough, N.M. andWilliams, R.L. (1989). Production of leukemia inhibitory factor in Escherichia coli by a novelprocedure and its use in maintaining embryonic stem cells in culture. Bio/Technology 7, 1157-1161.Gentz, R., Certa, U., Takacs, B., Matile, H., Diibeli, H., Pink, R., Mackay, M., Bone, N. andScaife, J.G. (1988). Major surface antigen p190 of Plasmodium falciparum: detection ofcommon epitopes present in a variety of plasmodia isolates. EMBO J. 7, 225-230.Gentz, R., Chen, C.-H. and Rosen, C.A. (1989). Bioassay for trans-activation using purifiedhuman immunodeficiency virus tat-encoded protein: trans-activation requires mRNA synthesis.Proc. Natl. Acad. Sci. USA 86, 821-824.Georgiou, G., Palumbo, T., Wilson, D.B. and Shuler, M.L. (1988). Effect of alkaline mediumon the production and excretion of 13-lactamase by Escherichia coli. Biotechnol. Lett. 10, 377-382.Germino, J. and Bastia, D. (1984). Rapid purification of a cloned gene product by genetic fusionand site-specific proteolysis. Proc. Natl. Acad. Sci. USA 81, 4692-4696.Germino, J., Gray, J.G., Charbonneau, H., Vanaman, T. and Bastia, D. (1983). Use of genefusions and protein-protein interaction in the isolation of a biologically active regulatory protein:the replication initiator protein of plasmid R6K. Proc. Natl. Acad. Sci. USA 80, 6848-6852.Ghangas, G.S. and Wilson, D.B. (1988). Cloning of the Thermomonospora fusca-endeglueartme-E2-gene in Streptornycesr ltvidans: affinity pufification and functional domains ofthe cloned gene product. Appl. Environ. Microbiol. 54, 2521-2526.169Gilbert, H.J., Hall, J., Hazlewood, G.P. and Ferreira, L.M.A. (1990). The N-terminal regionof an endoglucanase from Pseudomonas fluorescens subspecies cellulose constitutes a cellulose-binding domain that is distinct from the catalytic centre. Mol. Microbiol. 4, 759-767.Gilkes, N.R., Claeyssens, M., Aebersold, R., Henrissat, B., Meinke, A., Morrison, H.D.,Kilburn, D.G., Warren, R.A.J. and Miller, R.C., Jr. (1991a). Structural and functionalrelationships in two families of 0-1,4-glycanases. Eur. J. Biochem. 202, 367-377.Gilkes, N.R., Henrissat, B., Kilburn, D.G., Miller, R.C., Jr. and Warren, R.A.J. (1991).Domains in microbial f3-1,4-glycanases: sequence conservation, function, and enzyme families.Microbiol. Rev. 55, 303-315.Gilkes, N.R., Jervis, E., Henrissat, B., Tekant, B., Miller, R.C., Jr., Warren, R.A.J. andKilburn, D.G. (1992). The adsorption of a bacterial cellulase and its two isolated domains tocrystalline cellulose. J. Biol. Chem. 267, 6743-6749.Gilkes, N.R., Kilburn, D.G., Langsford, M.L., Miller, R.C., Jr., Wakarchuk, W.W., Warren,R.A.J., Whittle, D.J. and Wong, W.K.R. (1984). Isolation and characterization of Escherichiacoli clones expressing cellulase genes from Cellulomonas fimi. J. Gen. Microbiol. 130, 1377-1384.Gilkes, N.R., Kilburn, D.G., Miller, R.C., Jr. and Warren, R.A.J. (1989). Structural andfunctional analysis of a bacterial cellulase by proteolysis. J. Biol. Chem. 264, 17802-17808.Gilkes, N.R., Warren, R.A.J., Miller, R.C., Jr. and Kilburn, D.G. (1988). Precise excision ofthe cellulose binding domains from two Cellulomonas fimi cellulases by a homologous proteaseand the effect on catalysis. J. Biol. Chem. 263, 10401-10407.Gillis, S. and Smith, K.A. (1977). Long term culture of tumour-specific cytotoxic T cells.Nature 268, 154-156.Gillis, S., Ferm, M.M., Ou, W. and Smith, K. (1978). T cell growth factor: parameters ofproduction and a quantitative microassay for activity. J. Immunol. 120, 2027-2033.Goeddel, S.V., Kleid, D.G., Bolivar, F., Heyneker, H.L., Yansura, D.G., Crea, R., Hirose,T., Kraszewski, A., Itakura, K. and Riggs, A.D. (1979). Expression in Escherichia coli ofchemically synthesized genes for human insulin. Proc. Natl. Acad. Sci. USA 76, 106-110.Gottschalk, N. and Jaenicke, R. (1991). Authenticity and reconstitution of immobilizedenzymes: characterization and denaturation/renaturation of glucoamylase II. Biotechnol. Appl.Biochem. 14, 324-335.Greenwood, J.M., Gilkes, N.R., Kilburn, D.G., Miller, R.C., Jr. and Warren, R.A.J. (1989).Fusion to an endoglucanase allows alkaline phosphatase to bind to cellulose. FEBS Lett. 244,127-131.Greenwood, J.M., Ong, E., Gilkes, N.R., Warren, R.A.J., Miller, R.C., Jr. and Kilburn, D.G.(1992). Cellulose-binding domains: potential for purification of complex proteins. Protein Eng.5, 361-365.Grewal, T.S., Lowry, P.J. and Savva, D. (1989). Expression and partial purification of humanpro-opiomelanacortin in Escherichia_coli. -J-Mol-Endocrinol,  3, 105 112.170Grodberg, J. and Dunn, J.J. (1988). ompT encodes the Escherichia coli outer membraneprotease that cleaves T7 RNA polymerase during purification. J. Bacteriol. 170, 1245-1253.Guan, K.L. and Dixon, J.E. (1991). Eukaryotic proteins expressed in Escherichia coli: animproved thrombin cleavage and purification procedure of fusion proteins with glutathione S-transferase. Anal. Biochem. 192, 262-267.Guo, Z., Adman, N., Ong, E., Gilkes, N.R., Kilburn, D.G., Warren, R.A.J. and Miller, R.C.,Jr. (1988). Leakage of Cellulomonas find cellulases from Escherichia coli. FEMS Microbiol.Lett. 49, 279-283.Gutierrez, C., Barondess, J., Manoil, C. and Beckwith, J. (1987). The use of transposonTnphoA to detect genes for cell envelope proteins subject to a common regulatory stimulus.Analysis of osmotically regulated genes in Escherichia coli. J. Mol. Biol. 195, 289-297.Haffey, M.L., Lehman, D. and Boger, J. (1987). Site-specific cleavage of a fusion protein byrenin. DNA 6, 565-571.Hakes, D.J. and Dixon, J.E. (1992). New vectors for high level expression of recombinantproteins in bacteria. Anal. Biochem. 202, 293-298.Halliwell, G. and Griffin, M. (1978). Affinity chromatography of the cellulase system ofTrichoderma koningii. Biochem. J. 169, 713-715.Hammarberg, B., Nygren, P.-A., Holmgren, E., Elmblad, A., Tally, M., Hellman, U., Moks,T. and Uhler', M. (1989). Dual affinity fusion approach and its use to express recombinanthuman insulin-like growth factor II. Proc. Natl. Acad. Sci. USA 86, 4367-4371.Hammarberg, B., Tally, M., Samuelsson, E., Wadensten, H., Holmgren, E., Hartmanis, M.,Hall, K., Uhlen, M. and Moks, T. (1991). Characterization of an extended form of recombinanthuman insulin-like growth factor II. J. Biol. Chem. 266, 11058-11062.Hanada, K., Yamato, I. and Anraku, Y. (1988). Purification and reconstitution of Escherichiacoli proline carrier using a site specifically cleavable fusion protein. J. Biol. Chem. 263, 7181-7185.Harlow, E. and Lane, D. (1988). Antibodies: A Laboratory Manual. Cold Spring HarborLaboratory Press, Cold Spring Harbor.Harris, M.I. and Coleman, J.E. (1968). The biosynthesis of apo- and metalloalkalinephosphatases of Escherichia coli. J. Biol. Chem. 243, 5063-5073.Hartman, J., Daram, P., Frizzell, R.A., Rado, T., Benos, D.J. and Sorscher, E.J. (1992).Affinity purification of insoluble recombinant fusion proteins containing glutathione-S-transferase. Biotechnol. Bioeng. 39, 828-832.Haun, R.S. and Moss, J. (1992). Ligation-independent cloning of glutathione S-transferasefusion genes for expression in Escherichia coli. Gene 112, 37-43.Helfman, D.M. and Hughes, S.H. (1987). Use of antibodies to screen cDNA expressionlibraries prepared in plasmid vectors. Methods Enzymol. 152, 451-457.Hellman, J. and Mantsala, P. (1992). Construction of an Escherichia coli export-affinity vectorfor  expre ......^t • t^. •^ by ftrsiorrto -tyclometodextringlucanotransferase. J. Biotechnol. 23, 19-34.171Henrissat, B., Vigny, B., Buleon, A. and Peres, S. (1988). Possible adsorption sites ofcellulases on crystalline cellulose. FEBS Lett. 231, 177-182.Hickman, R.K., McMurry, L.M. and Levy, S.B. (1990). Overproduction and purification of theTn/O-specified inner membrane tetracycline resistance protein Tet using fusions to p-galactosidase. Mol. Microbiol. 4, 1241-1251.Hirabayashi, J. and Kasai, K. (1992). Arginine-tail method, an affinity tag procedure utilizinganhydrotrypsin agarose. J. Chromatog. 597, 181-187.Hirel, P.-H., Leveque, F., Mellot, P., Dardel, F., Panvert, M., Mechulam, Y. and Fayat, G.(1988). Genetic engineering of methionyl-tRNA synthetase: in vitro regeneration of an activesynthetase by proteolytic cleavage of a methionyl-tRNA synthetase-13-galactosidase chimericprotein. Biochimie 70, 773-782.Hochuli, E., Bannwarth, W., DObeli, H., Gentz, R. and Stiiber, D. (1988). Genetic approach tofacilitate purification of recombinant proteins with a novel metal chelate adsorbent.Bio/Technology 6, 1321-1325.Hochuli, F., Dtibeli, H. and Schacher, A. (1987). New metal chelate adsorbent selective forproteins and peptides containing neighbouring histidine residues. J. Chromatog. 411, 177-184.Hoffman, A. and Roeder, R.G. (1991). Purification of his-tagged proteins in non-denaturingconditions suggests a convenient method for protein interaction studies. Nucleic Acids Res. 19,6337-6338.Hoffman, C.S. and Wright, A. (1985). Fusions of secreted proteins to alkaline phosphatase: anapproach for studying protein secretion. Proc. Natl. Acad. Sci. USA 82, 5107-5111.Hopp, T.P., Prickett, K.S., Price, V.L., Cerretti, D.P., March, C.J. and Conlon, P.J. (1988a).Flag fusions: an efficient system for identification and purification of recombinant proteins. InProceedings: Advances in Gene Technology: Protein Engineering and Production, ed. K. Brewet al. Miami. ICSU Press; p. 138.Hopp, T.P., Prickett, K.S., Price, V.L., Libby, R.T., March, C.J., Cerretti, D.P., Urdal, D.L.and Conlon, P.J. (1988). A short polypeptide marker sequence useful for recombinant proteinidentification and purification. Bio/Technology 6, 1204-1210.Howard-Flanders, P., Simson, E. and Theriot, L. (1964). A locus that controls filamentformation and sensitivity to radiation in Escherichia coli K-12. Genetics 49, 237-246.Hsiung, H.M. and MacKellar, W.C. (1987). Expression of bovine growth hormone derivativesin Escherichia coli and the use of the derivatives to produce natural sequence growth hormone bycathepsin C cleavage. Methods Enzymol. 153, 390-401.Hsiung, H.M., Mayne, N.G. and Becker, G.W. (1986). High-level expression, efficientsecretion and folding of human growth hormone in Escherichia coli. Bio/Technology 4, 991-995.Hummel, M., Herbst, H. and Stein, H. (1989). Gene synthesis, expression in Escherichia coliand purification of immunoreactive human insulin-like growth factors I and II. Eur. J. Biochem.180, 555-561.Inouye, H. and Beckwith, J. (1977). Synthesis and processing of an Escherichia coli alkalinephosphatase precursor in vitro. Proc. Natl. Acad. Sci. USA 74, 1440-1444.172Itakura, K., Hirose, T., Crea, R. and Riggs, A.D. (1977). Expression in Escherichia coli of achemically synthesized gene for the hormone somatostatin. Science 198, 1056-1063.Iwakura, M., Furusawa, K., Kokubu, T., Ohashi, S., Tanaka, Y., Shimura, Y. and Tsuda, K.(1992). Dihydrofolate reductase as a new "affinity handle". J. Biochem. 111, 37-45.Iwakura, M., Obara, K., Kokubu, T., Ohashi, S. and Izutsu, H. (1992a). Expression andpurification of growth hormone-releasing factor with the aid of dihydrofolate reductase handle. J.Biochem. 112, 57-62.Janknecht, R. and Nordheim, A. (1992). Affinity purification of histidine-tagged proteinstransiently produced in HeLa cells. Gene 121, 321-324.Janknecht, R., de Martynoff, G., Lou, J., Hipskind, R.A., Nordheim, A. and Stunnenberg,H.G. (1991). Rapid and efficient purification of native histidine-tagged protein expressed byrecombinant vaccinia virus. Proc. Natl. Acad. Sci. USA 88, 8972-8976.Jansson, B., Palmcrantz, C., Uhlen, M. and Nilsson, B. (1989). A dual-affinity gene fusionsystem to express small recombinant proteins in a soluble form: expression and characterizationof protein A deletion mutants. Protein Eng. 2, 555-561.Johnson, K.S.J., Harrison, G.B.L., Lightowlers, M.W., O'Hoy, K.L., Cougle, W.G.,Dempster, R.P., Lawrence, S.B., Vinton, J.G., Heath, D.D. and Rickard, M.D. (1989).Vaccination against ovine cysticercosis using a defined recombinant antigen. Nature 338, 585-587.Johnson, L.N., Cheetham, J., McLaughlin, P.J., Acharya, K.R., Barford, D. and Phillips,D.C. (1988). Protein-oligosaccharide interactions: lysozyme, phospharylase, amylases. Cuff.Top. Microbiol. Immunol. 139, 81-134.Kaplan, G., Britton, W.J., Hancock, G.E., Theuvenet, W.J., Smith, K.A., Job, C.K., Roche,P.W., Molloy, A., Burkhardt, R., Barker, J., Pradhan, H.M. and Cohn, Z.A. (1991). Thesystemic influence of recombinant interleukin 2 on the manifestations of lepromatous leprosy. J.Exp. Med. 173, 993-1006.Kaplan, G., Cohn, Z.A. and Smith, K.A. (1992). Rational immunotherapy with interleukin 2.Bio/Technology 10, 157-162.Kasahara, M. and Anraku, Y. (1974). Succinate dehydrogenase of Escherichia coli membranevesicles. J. Biochem. 76, 959-966.Kato, K., Yamada, T., Kawahara, K., Onda, H., Asano, T., Sugino, H. and Kakinuma, A.(1985). Purification and characterization of recombinant human interleukin-2 produced inEscherichia coli. Biochem. Biophys. Res. Comm. 130, 692-699.Kennedy, R.M. (1990). Hydrophobic chromatography. Methods Enzymol. 182, 339-343.Kilburn, D.G., Miller, R.C., Jr., Warren, R.A.J. and Gilkes, N.R. (1992a). Cellulose bindingfusion proteins for immobilization and purification of polypeptides. U.S. Pat. 5137819Kilburn, D.G., Turner, R.F.B., Coutinho, J.B., Din, N., Gilkes, N.R., Greenwood, J.M.,Hobbs, J.B., Miller, R.C., Jr., Ong, E., Phelps, M.R., Ramirez, C. and Warren, R.A.J.(1992). Cellulose hinding_domains: applicationin Noteehrtology. In Pi oceedings: C-ettucon'92, Wrexham, Wales. In press.173Kilgus, J., Romagnoli, P., Guttinger, M., Stuber, D., Adorini, L. and Sinigaglia, F. (1989).Vaccine T-cell epitope selection by a peptide competition assay. Proc. Natl. Acad. Sci. USA 86,1629-1633.Klyosov, A.A. (1990). Trends in biochemistry and enzymology of cellulose degradation.Biochemistry 29, 10577-10589.Knott, J.A., Sullivan, C.A. and Weston, A. (1988). The isolation and characterisation of humanatrial natriuretic factor produced as a fusion protein in Escherichia coli. Eur. J. Biochem. 174,405-410.Knudsen, C.R., Clark, B.F., Degn, B. and Wiborg, 0. (1992). One-step purification of E. colielongation factor Tu. Biochemistry International 28, 353-362.KOhler, K., Ljungquist, C., Kondo, A., Veide, A. and Nilsson, B. (1991). Engineeringproteins to enhance their partition coefficients in aqueous two-phase systems. Bio/Technology 9,642-646.Koland, J.G., Maddison O'Brien, K. and Cerione, R.A. (1990). Expression of epidermalgrowth factor receptor sequences as E. coli fusion proteins: applications in the study of tyrosinekinase function. Biochem. Biophys. Res. Comm. 166, 90-100.Kolmar, H., Ferrando, E., Hennecke, F., Wippler, J. and Fritz, H.-J. (1992). Generalmutagenesis/gene expression procedure for the construction of variant immunoglobulin domainsin Escherichia coli. Production of the Bence-Jones protein REI V via fusion to f3-lactamase. J.Mol. Biol. 228, 359-365.Koths, K., Thomson, J., Kunitani, M., Wilson, K. and Hanisch, W. (1986). Process forrecovering microbially produced interleukin-2 and purified recombinant interleukin-2compositions. U.S. Pat. 4569790Kraft, R., Tardiff, J., Krauter, K.S. and Leinwand, L.A. (1988). Using mini-prep plasmidDNA for sequencing double stranded templates with SequenaseTM. BioTechniques 6, 544-546.Kraulis, P.J., Clore, G.M., Nilges, M., Jones, T.A., Pettersson, G., Knowles, J. andGronenborn, A.M. (1989). Determination of the three-dimensional solution structure of the C-terminal domain of cellobiohydrolase I from Trichoderma reesei. A study using nuclear magneticresonance and hybrid distance geometry-dynamical simulated annealing. Biochemistry 28,7241-7257.Kremer, R.D. and Tabb, D. (1990). Paper: the beneficially interactive support medium fordiagnostic test development. American Laboratory 2/90, 136-143.Krivi, G.G., Bittner, M.L., Rowold, E., Jr., Wong, E.Y., Glenn, K.C., Rose, K.S. andTiemeier, D.C. (1985). Purification of recA-based fusion proteins by immunoadsorbentchromatography. J. Biol. Chem. 260, 10263-10267.Kulshreshtha, A.K. and Dweltz, N.E. (1973). Paracrystalline lattice disorder in cellulose. I.Reappraisal of the application of the two-phase hypothesis to the analysis of powder x-raydiffractograms of native and hydrolysed cellulosic materials. J. Polym. Sci. 11, 487-497.Kunkel, T.A., Roberts, J.D. and Zakour, R.D. (1987). Rapid and efficient site-specificmutagenesis without phenotypic selection,- Me-thods Enzymol-. 1-54-, 367-382.174Laemmli, U.K. (1970). Cleavage of structural proteins during the assembly of the head ofbacteriophage T4. Nature 227, 680-685.Landon, M. (1977). Cleavage at aspartyl-prolyl bonds. Methods Enzymol. 47, 145-149.Langsford, M.L. (1988). The purification and characterization of two cellulose-binding,glycosylated cellulases from the bacterium Cellulomonas fimi. Ph.D. thesis, University ofBritish Columbia.Langsford, M.L., Gilkes, N.R., Singh, B., Moser, B., Miller, R.C., Jr., Warren, R.A.J. andKilburn, D.G. (1987). Glycosylation of bacterial cellulases prevents proteolytic cleavagebetween functional domains. FEBS Lett. 225, 163-167.Lauritzen, C., Tuchsen, E., Hansen, P.E. and Skovgaard, 0. (1991). BPTI and N-terminalextended analogues generated by factor Xa cleavage and cathepsin C trimming of a fusion proteinexpressed in Escherichia coli. Protein Expr. Pur. 2, 372-378.Le, D.K. (1992). Production and preliminary characterization of a fusion protein comprisingstreptavidin and a cellulose binding domain. M.Sc. thesis, University of British Columbia.Le, H.V., Syto, R., Mays, C., Reichert, P., Narula, S., Gewain, K., Greenberg, R., Kastelein,R., van Kimmenade, A., Nagabhushan, T.L. and Trotta, P.P. (1987). Isolation of E. coliderived murine interleukin-2 from intracellular and secretory expression systems. In ProteinPurification: Micro to Macro, ed. R. Burgess. Alan Liss, Inc., New York; pp. 383-391.Leemans, R., Remaut, E. and Fiers, W. (1989). Correlation between temperature-dependentcytoplasmic solubility and periplasmic export of a heterologous protein in Escherichia coli. Gene85, 99-108.LeGrice, S.F.J. and Grilninger-Leitch, F. (1990). Rapid purification of homodimer andheterodimer HIV-1 reverse transcriptase by metal chelate affinity chromatography. Eur. J.Biochem. 187, 307-314.Levens, D. and Howley, P.M. (1985). Novel method for identifying sequence-specific DNA-binding proteins. Mol. Cell Biol. 5, 2307-2315.Levinthal, C., Signer, E.R. and Fetherolf, K. (1962). Reactivation and hybridization of reducedalkaline phosphatase. Proc. Natl. Acad. Sci. USA 48, 1230-1237.Li, Z., Gambino, R., Fless, G.M., Copeland, R.A., Halfpenny, A.J. and Scanu, A.M. (1992).Expression and purification of kringle 4-type 2 of human apolipoprotein (a) in Escherichia coli.Protein Expr. Pur. 3, 212-222.Liang, S.-M., Allet, B., Rose, K., Hirschi, M., Liang, C.-M. and Thatcher, D.R. (1985).Characterization of human interleukin 2 derived from Escherichia coli. Biochem. J. 229, 429-439.Lilius, G., Persson, M., Billow, L. and Mosbach, K. (1991). Metal affinity precipitation ofproteins carrying genetically attached polyhistidine affinity tails. Eur. J. Biochem. 198, 499-504.Lin, Y., Robb, R.J., Gray, J.E. and Simon, P. (1988). The construction and characterization ofa biologically active recombinant IL-2 containing a lysine-rich C-terminal extension. J. Immunol.1-4-l y 3847 3-8-5-1.175Lin, Y.-S. and Green, M.R. (1991). Mechanism of action of an acidic transcriptional activator invitro. Cell 64, 971-981.Liu, D.T.-Y. (1992). Deamidation: a source of microheterogeneity in pharmaceutical proteins.Trends Biotechnol. 10, 364-369.Ljungquist, C., Breitholtz, A., Brink-Nilsson, H., Moks, T., Uhlën, M. and Nilsson, B.(1989). Immobilization and affinity purification of recombinant proteins using histidine peptidefusions. Eur. J. Biochem. 186, 563-569.Lorberboum-Galski, H., FitzGerald, D., Chaudhary, V., Sankar, A. and Pastan, I. (1988).Cytotoxic activity of an interleukin 2-Pseudomonas exotoxin chimeric protein produced inEscherichia coli. Proc. Natl. Acad. Sci. USA 85, 1922-1926.Lottenberg, R., Hall, J.A., Pautler, E., Zupan, A., Christensen, U. and Jackson, C.M. (1986).The action of Factor Xa on peptide p-nitroanilide substrates: substrate selectivity and examinationof hydrolysis with different reaction conditions. Biochim. et  Biophys. Acta 874, 326-336.Lotter, H., Mannweiler, E., Schreiber, M. and Tannich, E. (1992). Sensitive and specificserodiagnosis of invasive amebiasis by using a recombinant surface protein of pathogenicEntamoeba histolytica. J. Clin. Microbiol. 30, 3163-3167.Louis, J.M., McDonald, R.A., Nashed, N.T., Wondrak, E.M., Jerina, D.M., Oroszlan, S. andMora, P.T. (1991). Autoprocessing of the HIV-1 protease using purified wild-type and mutatedfusion proteins expressed at high levels in Escherichia coli. Eur. J. Biochem. 199, 361-369.Lowe, P.A., Rhind, S.K., Sugrue, R. and Marston, F.A.O. (1987). Solubilisation, refoldingand purification of eukaryotic proteins expressed in E. coli. In Protein Purification: Micro toMacro; UCLA symposia on molecular and cellular biology, ed. R. Burgess. Alan R. Liss, Inc.,New York; pp. 429-442.LOwenadler, B., Nilsson, B., Abrahms6n, L., Moks, T., Ljungqvist, L., Holmgren, E., Paleus,S., Josephson, S., Philipson, L. and Uhler', M. (1986). Production of specific antibodiesagainst protein A fusion proteins. EMBO J. 5, 2393-2398.Lundeberg, J., Wahlberg, J. and Uhlen, M. (1990). Affinity purification of specific DNAfragments using a lac repressor fusion protein. Genet. Anal. Techn. Appl. 7, 47-52.Lydon, N.B., Adams, B., Poschet, J.F., Gutzwiller, A. and Matter, A. (1990). An E. coliexpression system for the rapid purification and characterization of a v-abl tyrosine proteinkinase. Oncogene Research 5, 161-173.Mach, H., Middaugh, C.R. and Lewis, R.V. (1992). Statistical determination of the averagevalues of the extinction coefficients of tryptophan and tyrosine in native proteins. Anal.Biochem. 200, 74-80.Maglione, G., Russell, J.B. and Wilson, D.B. (1992). Properties of a genetically reconstructedPrevotella ruminicola endoglucanase. Appl. Environ. Microbiol. 58, 3593-3597.Maina, C.V., Riggs, P.D., Grandea III, A.G., Slatko, B.E., Moran, L.S., Tagliamonte, J.A.,McReynolds, L.A. and di Guan, C. (1988). An Escherichia coli vector to express and purifyforeign proteins by fusion to and separation from maltose-binding protein. Gene 74, 365-373.Malarny i. MIL artel-Horecker, -R.L. (1964). Purification and crysfallization of the alkalinephosphatase of Escherichia coli. Biochemistry 3, 1893-1897.176Malamy, M.H. and Horecker, B.L. (1964a). Release of alkaline phosphatase from cells ofEscherichia coli upon lysozyme spheroplast formation. Biochemistry 3, 1889-1893.Manoil, C. and Beckwith, J. (1985). TnphoA: A transposon probe for protein export signals.Proc. Natl. Acad. Sci. USA 82, 8129-8133.Manoil, C., Mekalanos, J.J. and Beckwith, J. (1990). Alkaline phosphatase fusions: sensors ofsubcellular location. J. Bacteriol. 172, 515-518.Markmeyer, P., Riihlmann, A., Englisch, U. and Cramer, F. (1990). The pAX plasmids: newgene-fusion vectors for sequencing, mutagenesis and expression of proteins in Escherichia coli.Gene 93, 129-134.Marshall, K. and Sixsmith, D. (1974). Some physical characteristics of microcrystallinecellulose. 1. Powders for pharmaceutical use. Drug Devel. Comm. 1, 51-71.Marston, F.A.O., Lowe, P.A., Doel, M.T., Schoemaker, J.M., White, S. and Angal, S. (1984).Purification of calf prochymosin (prorennin) synthesized in Escherichia coli. Bio/Technology 2,800-804.Mart'yanov, V.A., Rabinovich, M.L., Klesov, A.A., Myagkikh, I V and Gernet, M.V. (1984).Affinity chromatography of cellulases on cellulose carriers. App. Biochem. Microbiol. 19, 405-411.Martineau, P., Szmelcman, S., Spurlino, J.C., Quiocho, F.A. and Hofnung, M. (1990).Genetic approach to the role of tryptophan residues in the activities and fluorescence of a bacterialperiplasmic maltose-binding protein. J. Mol. Biol. 214, 337-352.Masuta, C., Furuno, M., Tanaka, H., Yamada, M. and Koiwai, A. (1992). Molecular cloningof a cDNA clone for tobacco lipid transfer protein and expression of the functional protein inEscherichia coli. FEBS Lett. 311, 119-123.Matsudaira, P. (1990). Limited N-terminal sequence analysis. Methods Enzymol. 182, 602-613McComb, R.B., Bowers, G.N., Jr. and Posen, S. (1979). Alkaline phosphatase. PlenumPress, New York.McCracken, S. and Meighen, E. (1980). Functional and structural properties of immobilizedsubunits of Escherichia coli alkaline phosphatase. J. Biol. Chem. 255, 2396-2404.McGhee, J.D. and von Hippel, P.H. (1974). Theoretical aspects of DNA-protein interactions:co-operative and non-co-operative binding of large ligands to a one-dimensional homogeneouslattice. J. Mol. Biol. 86, 469-489.Meinke, A., Braun, C., Gilkes, N.R., Kilburn, D.G., Miller, R.C., Jr. and Warren, R.A.J.(1991). Unusual sequence organization in CenB, an inverting endoglucanase from Cellulomonasfimi. J. Bacteriol. 173, 308-314.Meinke, A., Gilkes, N.R., Kilburn, D.G., Miller, R.C., Jr. and Warren, R.A.J. (1993).Cellulose-binding polypeptides from Cellulomonas fimi: endoglucanase D (CenD), a family A 13-1,4-glucanase. J. Bacteriol. in press.177Meinke, A., Schmuck, M., Gilkes, N.R., Kilburn, D.G., Miller, R.C., Jr. and Warren, R.A.J.(1992). The tertiary structure of endo-f3-1,4-glucanase B (CenB), a multidomain cellulase fromthe bacterium Cellulomonas fimi. Glycobiology 2, 321-326.Men6ndez-Arias, L., Young, M. and Oroszlan, S. (1992). Purification and characterization ofthe mouse mammary tumor virus protease expressed in Escherichia coli. J. Biol. Chem. 267,24134-24139.Michaelis, S., Inouye, H., Oliver, D. and Beckwith, J. (1983). Mutations that alter the signalsequence of alkaline phosphatase in Escherichia coli. J. Bacteriol. 154, 366-374.Miller, H. (1987). Practical aspects of preparing phage and plasmid DNA: growth, maintenance,and storage of bacteria and bacteriophage. Methods Enzymol. 152, 145-170.Miller, J.H. (1972). Experiments in Molecular Genetics. Cold Spring Harbor Laboratory Press,Cold Spring Harbor.Miller, P.B., Shen, H., Gilkes, N.R., Kilburn, D.G., Miller, R.C., Jr., Plaut, A.G. andWarren, R.A.J. (1992). Endoglucanase A from Cellulomonas fimi in which the hinge sequenceof human IgA1 is substituted for the linker connecting its two domains is hydrolyzed by IgAproteases from Neisseria gonorrhoeae. FEMS Microbiol. Lett. 92, 199-204.Moks, T., Abrahmsen, L., Holmgren, E., Bilich, M., Olsson, A., Pohl, G., Sterky, C.,Hultberg, H., Josephson, S., Holmgren, A., Thmvall, H., Uhlen, M. and Nilsson, B. (1987).Expression of human insulin-like growth factor I in bacteria: use of optimized gene fusionvectors to facilitate protein purification. Biochemistry 26, 5239-5244.Moks, T., Abrahms6n, L., Nilsson, B., Hellman, U., Sjoquist, J. and Uhlen, M. (1986).Staphylococcal protein A consists of five IgG-binding domains. Eur. J. Biochem. 156, 637-643.Moks, T., Abrahmsen, L., OsterlOf, B., Josephson, S., Ostling, M., Enfors, S.-O., Persson, I.-L., Nilsson, B. and Uhler', M. (1987a). Large-scale affinity purification of human insulin-likegrowth factor I from culture medium of Escherichia coli. Bio/Technology 5, 379-382.Monaco, L., Bond, H.M., Howell, K.E. and Cortese, R. (1987). A recombinant apoA-1—protein A hybrid reproduces the binding parameters of HDL to its receptor. EMBO J. 6, 3253-3260.Moser, B., Gilkes, N.R., Kilburn, D.G., Warren, R.A.J. and Miller, R.C., Jr. (1989).Purification and characterization of endoglucanase C of Cellulomonas fimi, cloning of the gene,and analysis of in vivo transcripts of the gene. Appl. Environ. Microbiol. 55, 2480-2487.Mossman, T. (1983). Rapid colorimetric assay for cellular growth and survival: application toproliferation and cytotoxicity assays. J. Immunol. Methods 65, 55-63.Murby, M., Cedergren, L., Nilsson, J., Nygren, P.-A, Hammarberg, B., Nilsson, B., Enfors,S.-O. and Uhler', M. (1991). Stabilization of recombinant proteins from proteolytic degradationin Escherichia coli using a dual affinity fusion strategy. Biotechnol. Appl. Biochem. 14, 336-346.Murby, M., Nygren, P.-A., Rondahl, H., Hellman, U., Enfors, S.-O. and Uhlen, M. (1991a).Differential degradation of a recombinant albumin-binding receptor in Escherichia coli. Eur. IBiochent. 199, 44-46.178Nagai, K. and ThOgersen, H.C. (1984). Generation of (3-globin by sequence-specificproteolysis of a hybrid protein produced in Escherichia coli. Nature 309, 810-812.Nagai, K. and Thogersen, H.C. (1987). Synthesis and sequence-specific proteolysis of hybridproteins produced in Escherichia coli. Methods Enzymol. 153, 461-481.Nakagawa, S., Yamada, T., Kato, K. and Nishimura, 0. (1987). Enzymatic cleavage of aminoterminal methionine from recombinant human interleukin 2 and growth hormone byaminopeptidase M. BiolTechnology 5, 824-827.Nakata, A., Shinagawa, H. and Rothman, F.G. (1987). Molecular mechanism of isozymeformation of alkaline phosphatase in Escherichia coli. In Phosphate metabolism and cellularregulation in microorganisms, ed. A. Torriani-Gorini et al. American Society for Microbiology,Washington, D.C.; pp. 139-141.Nardin, E.H., Herrington, D.A., Davis, J., Levine, M., Stuber, D., Takacs, B., Caspers, P.,Barr, P., Altszuler, R., Clavijo, P. and Nussenzweig, R.S. (1989). Conserved repetitive epitoperecognized by CD4+ clones from a malaria-immunized volunteer. Science 246, 1603-1606.Neu, H.C. and Heppel, L.A. (1965). The release of enzymes from Escherichia coli by osmoticshock and during the formation of spheroplasts. J. Biol. Chem. 240, 3685-3692.Neuberger, M.S., Williams, G.T. and Fox, R.O. (1984). Recombinant antibodies possessingnovel effector functions. Nature 312, 604-608.Nilsson, B., Abrahms6n, L. and Uhlen, M. (1985). Immobilization and purification of enzymeswith staphylococcal protein A gene fusion vectors. EMBO J. 4, 1075-1080.Nilsson, B., Forsberg, G. and Hartmanis, M. (1991). Expression and purification ofrecombinant insulin-like growth factors from Escherichia coli. Methods Enzymol. 198, 3-16.Nilsson, B., Forsberg, G., Moks, T., Hartmanis, M. and Uhlen, M. (1992). Fusion proteins inbiotechnology and structural biology. Curr. Opin. Struct. Biol. 2, 569-575.Nilsson, B., Holmgren, E., Josephson, S., Philipson, L. and Uhlen, M. (1985a). Efficientsecretion and purification of human insulin-like growth factor I with a gene fusion vector inStaphylococci. Nucleic Acids Res. 13, 1151-1162.Nilsson, B., Moks, T., Jansson, B., Abrahmsen, L., Elmblad, A., Holmgren, E., Henrichson,C., Jones, T.A. and Uhler', M. (1987). A synthetic IgG-binding domain based onstaphylococcal protein A. Protein Eng. 1, 107-113.Nordquist, D. (1992). Characterization of double binding domain derivatives of CenA fromCellulomonas fimi. M.Sc. thesis, University of British Columbia.Nossal, N.G. and Heppel, L.A. (1966). The release of enzymes by osmotic shock fromEscherichia coli in exponential phase. J. Biol. Chem. 241, 3055-3062.Nummi, M., Niku-Paavola, M.-L., Enari, T.-M. and Raunio, V. (1981). Isolation of cellulasesby means of biospecific sorption of amorphous cellulose. Anal. Biochem. 116, 137-141.Nunberg, J.H., Doyle, M.V., York, S.M. and York, C.J. (1989). Interleukin 2 acts as anadjuvant to increase the potency of inactivated rabies virus vaccine. Pr_oc_ NatL  Mad  Sci. USA 8t, 424O-4243.179Nurse, F., Dagan, S., Tackney, C. and Goldstein, N.I. (1991). Isolation of a monoclonalantibody to the TrpE protein and its use for the purification of recombinant fusion proteins.Hybridoma 10, 753-760.Nygren, P.-A., Ljungquist, C., Tromborg, H., Nustad, K. and Uhler', M. (1990). Species-dependent binding of serum albumins to the streptococcal receptor protein G. Eur. J. Biochem.193, 143-148.O'Neill, G., Goh, S.H., Warren, R.A.J., Kilburn, D.G. and Miller, R.C., Jr. (1986). Structureof the gene encoding the exoglucanase of Cellulomonas fimi. Gene 44, 325-330.O'Neill, G.P., Warren, R.A.J., Kilburn, D.G. and Miller, R.C., Jr. (1986a). Secretion ofCellulomonas fimi exoglucanase by Escherichia coli. Gene 44, 331-336.Offensperger, W., Wahl, S., Neurath, A.R., Price, P., Strick, N., Kent, S.B.H., Christman,J.K. and Acs, G. (1985). Expression in Escherichia coli of a cloned DNA sequence encoding thepre-S2 region of hepatitis B virus. Proc. Natl. Acad. Sci. USA 82, 7540-7544.Olsen, E. and Mohapatra, S.S. (1992). Expression and thrombin cleavage of Poa p IXrecombinant allergens fused to glutathione S-transferase. Int. Arch. Allergy 98, 343-348.Ong, E., Gilkes, N.R., Miller, R.C., Jr., Warren, R.A.J. and Kilburn, D.G. (1991). Enzymeimmobilization using a cellulose-binding domain: properties of a P-glucosidase fusion protein.Enzyme Microb. Technol. 13, 59-65.Ong, E., Gilkes, N.R., Miller, R.C., Jr., Warren, R.A.J. and Kilburn, D.G. (1993a). Thecellulose-binding domain (CBDcex) of an exoglucanase from Cellulomonas fimi: production inEscherichia coli and characterisation of the polypeptide. Biotechnol. Bioeng. in press.Ong, E., Gilkes, N.R., Warren, R.A.J., Miller, R.C., Jr. and Kilburn, D.G. (1989). Enzymeimmobilization using the cellulose-binding domain of a Cellulomonas fimi exoglucanase.Bio/Technology 7, 604-607.Ong, E., Greenwood, J.M., Gilkes, N.R., Kilburn, D.G., Miller, R.C., Jr. and Warren, R.A.J.(1989a). The cellulose-binding domains of cellulases: tools for biotechnology. TrendsBiotechnol. 7, 239-243.Ong, E., Greenwood, J.M., Gilkes, N.R., Miller, R.C., Jr. and Warren, R.A.J. (1993).Proteins designed for adherence to cellulose. In Biocatalyst Design for Stability and Specificity;ACS Symposium Series 516, ed. M.E. Himmel and G. Georgiou. American Chemical Society,Washington DC; pp. 185-194.Osborn, M.J. and Munson, R. (1974). Separation of the inner (cytoplasmic) and outermembranes of gram-negative bacteria. Methods Enzymol. 31, 642-653.Ostrove, S. (1990). Affinity chromatography: general methods. Methods Enzymol. 182, 357-371Otter, D.E., Munro, P.A., Scott, G.K. and Geddes, R. (1989). Desorption of Trichodermareesei cellulase from cellulose by a range of desorbents. Biotechnol. Bioeng. 34, 291-298.Owolabi, J.B., Beguin, P., Kilburn, D.G., Miller, R.C., Jr. and Warren, R.A.J. (1988).Expression in Escherichia coli of the Cellulomonas fimi structural gene for endoglucanase B.Appl. Environ. Mi tiol -5181521180Parker, D.E., Glatz, C.E., Ford, C.F., Gendel, S.M., Suominen, I. and Rougvie, M.A. (1990).Recovery of a charged-fusion protein from cell extracts by polyelectrolyte precipitation.Biotechnol. Bioeng. 36, 467-475.Perrin, D.D. and Dempsey, B. (1974). Buffers for pH and Metal Ion Control. Chapman andHall, New York.Persson, M., Bergstrand, M.G., Billow, L. and Mosbach, K. (1988). Enzyme purification bygenetically attached polycysteine and polyphenylalanine affinity tails. Anal. Biochem. 172, 330-337.Persson, M., Billow, L. and Mosbach, K. (1990). Purification and site-specific immobilizationof genetically engineered glucose dehydrogenase on thiopropyl-Sepharose. FEBS Lett. 270, 41-44.Phalipon, A. and Kaczorek, M. (1987). Genetically engineered diphtheria toxin fusion proteinscarrying the hepatitis B surface antigen. Gene 55, 255-263.Pilz, I., Schwarz, E., Kilburn, D.G., Miller, R.C., Jr., Warren, R.A.J. and Gilkes, N.R.(1990). The tertiary structure of a bacterial cellulase determined by small-angle X-ray-scatteringanalysis. Biochem. J. 271, 277-280.Plocke, D.J., Levinthal, C. and Vallee, B.L. (1962). Alkaline phosphatase of Escherichia coli: azinc metalloenzyme. Biochemistry 1, 373-378.Pliickthun, A. (1991). Antibody engineering: advances from the use of Escherichia coliexpression systems. Bio/Technology 9, 545-551.Pohiner, J., Klauser, T., Kuttler, E. and Halter, R. (1992). Sequence-specific cleavage ofprotein fusions using a recombinant Neisseria type 2 IgA protease. Bio/Technology 10, 799-804.Poole, D.M., Dun-ant, A.J., Hazlewood, G.P. and Gilbert, H.J. (1991). Characterization ofhybrid proteins consisting of the catalytic domains of Clostridium and Ruminococcusendoglucanases, fused to Pseudomonas non-catalytic cellulose-binding domains. Biochem. J.279, 787-792.Poole, D.M., Hazlewood, G.P., Huskisson, N.S., Virden, R. and Gilbert, H.J. (1993). Therole of conserved trytophan residues in the interaction of a bacterial cellulose binding domain withits ligand. FEMS Microbiol. Lett. 106, 77-84.Porath, J., Carlsson, J., Olsson, I. and Belfrage, G. (1975). Metal chelate affinitychromatography, a new approach to protein fractionation. Nature 258, 598-599.Power, B.E., Ivancic, N., Harley, V.R., Webster, R.G., Kortt, A.A., Irving, R.A. andHudson, P.J. (1992). High-level temperature-induced synthesis of an antibody VH-domain inEscherichia coli using the PelB secretion signal. Gene 113, 95-99.Preston, R.D. (1986). Natural Celluloses. In Cellulose. Structure, Modification andHydrolysis, ed. R.A. Young and R.M. Rowell. John Wiley and Sons, New York; pp. 3-27.Prickett, K.S., Amberg, D.C. and Hopp, T.P. (1989). A calcium-dependent antibody foridentification and purification of recombinant proteins. BioTechniques 7, 580-589.Quiocho, F.A. (1986). Carbohydrate-binding proteins: tertiary structures and protein-sugarinteractions. Ann. Rev. Biochem. 55, 287-315.181Quiocho, F.A. (1988). Molecular features and basic understanding of protein-carbohydrateinteractions: the arabinose-binding protein-sugar complex. Cum Top. Microbiol. Immunol.139, 135-148.Rahman, M.A., Brot, N. and Weissbach, H. (1992). High level expression and purification ofpeptide methionine sulfoxide reductase in Escherichia coli. Cellular and Molecular Biology 38,529-542.Ramshaw, I.A., Andrew, M.E., Phillips, S.M., Boyle, D.B. and Coupar, B.E.H. (1987).Recovery of immunodeficient mice from a vaccinia virus/IL-2 recombinant infection. Nature329, 545-546.Reese, E.T. (1982). Elution of cellulase from cellulose. Process Biochem. 17, 2-6.Reid, T.W. and Wilson, I.B. (1971). E. coli alkaline phosphatase. In The Enzymes, Vol. IV,ed. P.D. Boyer. Academic Press, New York; pp. 373-415.Reinikainen, T., Ruohonen, L., Nevanen, T., Laaksonen, L., Kraulis, P., Jones, T.A.,Knowles, J.K.C. and Teeri, T.T. (1992). Investigation of the function of mutated cellulose-binding domains of Trichoderma reesei cellobiohydrolase I. Proteins 14, 475-482.Roberts, C.H. and Chlebowski, J.F. (1984). Trypsin modification of Escherichia coli alkalinephosphatase. J. Biol. Chem. 259, 729-733.Roehrborn, C.G., Zoppi, S., Gruber, J.A., Wilson, C.M. and McPhaul, M.J. (1992).Expression and characterization of full-length and partial human androgen receptor fusionproteins. Implications for the production and applications of soluble steroid receptors inEscherichia coli. Mol. Cell. Endocrinol. 84, 1-14.Rosenberg, A.H., Lade, B.N., Chui, D.-S., Lin, S.-W., Dunn, J.J. and Studier, F.W. (1987).Vectors for selective expression of cloned DNAs by T7 RNA polymerase. Gene 56, 125-135.Rothman, F. and Byrne, R. (1963). Fingerprint analysis of alkaline phosphatase of Escherichiacoli K12. J. Mol. Biol. 6, 330-340.Sabin, E.A., Lee-Ng, C.T., Shuster, J.R. and Barr, P.J. (1989). High-level expression and invivo processing of chimeric ubiquitin fusion proteins in Saccharomyces cerevisiae.Bio/Technology 7, 705-709.Sambrook, J., Fritsch, E.F. and Maniatis, T. (1989). Molecular Cloning. A LaboratoryManual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor.Samuelsson, E., Wadensten, H., Hartmanis, M., Moks, T. and Uhlen, M. (1991). Facilatatedin vitro refolding of human recombinant insulin-like growth factor I using a solubilizing fusionpartner. Bio/Technology 9, 363-366.Sanchez-Puelles, J.M., Sanz, J.M., Garcia, J.L. and Garcia, E. (1992). Immobilization andsingle-step purification of fusion proteins using DEAE-cellulose. Eur. J. Biochem. 203, 153-159.Sandermann, H., Jr. and Strominger, J.L. (1972). Purification and properties of C 55-isoprenoidalcohol phosphokinase from Staphylococcus aureus. J. Biol. Chem. 2475 123-5131. Sankar, S. and Porter, A.G. (1991). Expression, purification, and properties of recombinantencephalomyocarditis virus RNA-dependent RNA polymerase. J. Virol. 65, 2993-3000.182Sano, T. and Cantor, C.R. (1991). A streptavidin-protein A chimera that allows one-stepproduction of a variety of specific antibody conjugates. Bio/Technology 9, 1378-1381.Sarthy, A., Michaelis, S. and Beckwith, J. (1981). Deletion map of the Escherichia colistructural gene for alkaline phosphatase, phoA. J. Bacteriol. 145, 288-292.Sassenfeld, H.M. (1984). A polypeptide fusion designed for the purification of recombinantproteins. BiolTechnology 2, 76-81.Sassenfeld, H.M. (1990). Engineering proteins for purification. Trends Biotechnol. 8, 88-93.Sassenfeld, H.M., Deeley, M., Rubero, J., Shriner, J.C. and Madani, H. (1991). Geneticalterations which facilitate protein purification: applications in the biopharmaceutical industry. InApplications of Enzyme Biotechnology, ed. J.W. Kelly and T.O. Baldwin. Plenum Press, NewYork; pp. 251-259.Scherf, A., Mattei, D. and Schreiber, M. (1990). Parasite antigens expressed in Escherichia coli.A refined approach for epidemiological analysis. J. Immunol. Methods 128, 81-87.Schlesinger, M.J. (1965). The reversible dissociation of the alkaline phosphatase of Escherichiacoli. II. Properties of the subunit. J. Biol. Chem. 240, 4293-4298.Schlesinger, M.J. and Barrett, K. (1965). The reversible dissociation of the alkaline phosphataseof Escherichia coli. I. Formation and reactivation of subunits. J. Biol. Chem. 240, 4284-4292.Schlesinger, M.J. and Olsen, R. (1970). A new, simple, rapid procedure for purification ofEscherichia coli alkaline phosphatase. Anal. Biochem. 36, 86-90.Schlesinger, M.L. and Levinthal, C. (1963). Hybrid formation of E. coli alkaline phosphataseleading to in vitro complementation. J. Mol. Biol. 7, 1-12.Schnaitman, C.A. (1981). Cell fractionation. In Manual of Methods for General Bacteriology,ed. P. Gerhardt. American Society for Microbiology, Washington, D.C.; pp. 52-61.Scholtissek, S. and Grosse, F. (1988). A plasmid vector system for the expression of atriprotein consisting of D-galactosidase, a collagenase recognition site and a foreign gene product.Gene 62, 55-64.Scopes, R.K. (1974). Measurement of protein by spectrophotometry at 205 nm. Anal.Biochem. 59, 277-282.Sharma, S.K., Evans, D.B., Vosters, A.F., McQuade, T.J. and Tarpley, W.G. (1991). Metalaffinity chromatography of recombinant HIV-1 reverse transcriptase containing a human renincleavable metal binding domain. Biotechnol. Appl. Biochem. 14, 69-81.Shen, H., Schmuck, M., Pilz, I., Gilkes, N.R., Kilburn, D.G., Miller, R.C., Jr. and Warren,R.A.J. (1991). Deletion of the linker connecting the catalytic and cellulose-binding domains ofendoglucanase A (CenA) of Cellulomonas fimi. J. Biol. Chem. 266, 11335-11340.Sherwood, R. (1991). Protein fusions: bioseparation and application. Trends Biotechnol. 9, 1-3.Shim;^ettes,-L-Man,^T.Z. and Baxter, J.D. (1980). Expression of cloned13-endorphin gene sequences by Escherichia coli. Nature 285, 456-461.183Sieg, K., Kun, J., Pohl, I., Scherf, A. and Muller-Hill, B. (1989). A versatile phage lambdaexpression vector system for cloning in Escherichia coli. Gene 75, 261-270.Simpson, I.A. and Sonne, 0. (1982). A simple, rapid, and sensitive method for measuringprotein concentration in subcellular membrane fractions prepared by sucrose densityultracentrifugation. Anal. Biochem. 119, 424-427.Skerra, A., Pfitzinger, I. and Pliickthun, A. (1991). The functional expression of antibody F vfragments in Escherichia coli: improved vectors and a generally applicable purification technique.Bio/Technology 9, 273-278.Smith, D.B. and Johnson, K.S. (1988). Single-step purification of polypeptides expressed inEscherichia coli as fusions with glutathione S-transferase. Gene 67, 31-40.Smith, J.C., Derbyshire, R.B., Cook, E., Dunthorne, L., Viney, J., Brewer, S .J., Sassenfeld,H.M. and Bell, L.D. (1984). Chemical synthesis and cloning of a poly(arginine)-coding genefragment designed to aid polypeptide purification. Gene 32, 321-327.Smith, K.A. (1988). Interleukin-2: inception, impact, and implications. Science 240, 1169-1176.Smith, K.A. (1992). Interleukin-2. Curr. Op. Immunol. 4, 271-276.Smith, M.C., Furman, T.C. and Pidgeon, C. (1987). Immobilized iminodiacetic acid metalpeptide complexes. Identification of chelating peptide purification handles for recombinantproteins. Inorg. Chem. 26, 1965-1969.Smith, M.C., Furman, T.C., Ingolia, T.D. and Pidgeon, C. (1988). Chelating peptide-immobilized metal ion affinity chromatography. J. Biol. Chem. 263, 7211-7215.Sowadski, J.M., Handschumacher, M.D., Murthy, H.M.K., Foster, B.A. and Wyckoff, H.W.(1985). Refined structure of alkaline phosphatase from Escherichia coli at 2.8 A resolution. J.Mol. Biol. 186, 417-433.Spurlino, J.C., Lu, G.-Y. and Quiocho, F.A. (1991). The 2.3-A resolution structure of themaltose- or maltodextrin-binding protein, a primary receptor of bacterial active transport andchemotaxis. J. Biol. Chem. 266, 5202-5219.Stahl, S., Sjolander, A., Hansson, M., Nygren, P.A. and Uhlen, M. (1990). A general strategyfor polymerization, assembly and expression of epitope-carrying peptides applied to thePlasmodium fakiparum antigen Pf155/RESA. Gene 89, 187-193.Stahl, S., Sjolander, A., Nygren, P.-A., Berzins, K., Perlmann, P. and Uhlen, M. (1989). Adual expression system for the generation, analysis and purification of antibodies to a repeatedsequence of the Plasmodium falciparum antigen Pf155/RESA. J. Immunol. Methods 124, 43-52.Stahlberg, J., Johansson, G. and Pettersson, G. (1991). A new model for enzymatic hydrolysisof cellulose based on the two-domain structure of cellobiohydrolase I. Bio/Technology 9, 286-290.Steers, E., Jr., Cuatrecasas, P. and Pollard, H.B. (1971. The purification  of 41-zalactosida,se,from Escherichia coli by affinity chromatography. J. Biol. Chem. 246, 196-200.184Stein, R.C., Malkovska, V., Morgan, S., Galazka, A., Aniszewski, C., Roy, S.E., Shearer,R.J., Marsden, R.A., Bevan, D., Gordon-Smith, E.C. and Coombes, R.C. (1991). The clinicaleffects of prolonged treatment of patients with advanced cancer with low-dose subcutaneousinterleukin 2. Brit. J. Cancer 63, 275-278.Stirling, D.A., Petrie, A., Pulford, D.J., Paterson, D.T. and Stark, M.J. (1992). Protein A-calmodulin fusions: a novel approach for investigating calmodulin function in yeast. Mol.Microbiol. 6, 703-713.Stofko-Hahn, R.E., Carr, D.W. and Scott, J.D. (1992). A single step purification forrecombinant proteins. Characterization of a microtubule associated protein (MAP 2) fragmentwhich associates with the type II cAMP-dependent protein kinase. FEBS Lett. 302, 274-278.Strauch, K.L., Johnson, K. and Beckwith, J. (1989). Characterization of degP, a gene requiredfor proteolysis in the cell envelope and essential for growth of Escherichia coli at hightemperature. J. Bacteriol. 171, 2689-2696.Stringer, K.F., Ingles, C.J. and Greenblatt, J. (1990). Direct and selective binding of an acidictranscriptional activation domain to the TATA-box factor TFIID. Nature 345, 783-786.Stilber, D., Bannwarth, W., Pink, J.R.L., Meloen, R.H. and Matile, H. (1990). New B cellepitopes in the Plasmodium falciparum malaria circumsporozoite protein. Eur. J. Immunol. 20,819-824.Sttiber, D., Matile, H. and Garotta, G. (1990a). System for high-level production in Escherichiacoli and rapid purification of recombinant proteins: application to epitope mapping, preparation ofantibodies, and structure-function analysis. In Immunological Methods, Vol. IV, ed. I. Lefkovitsand B. Pernis. Academic Press, Inc., New York; pp. 121-152.Studier, F.W. and Moffatt, B.A. (1986). Use of bacteriophage T7 RNA polymerase to directselective high-level expression of cloned genes. J. Mol. Biol. 189, 113-130.Suh, S.-S., Haymore, B.L. and Arnold, F.H. (1991). Characterization of His-X3-His sites inoc-helices of synthetic metal-binding bovine somatotropin. Protein Eng. 4, 301-305.Sulkowski, E. (1989). The saga of IMAC and MIT. BioEssays 10, 170-175.Svensson, B., Jespersen, H., Sierks, M.R. and Macgregor, E.A. (1989). Sequence homologybetween putative raw-starch binding domains from different starch-degrading enzymes.Biochem. J. 264, 309-311.Szoka, P.R., Schrieber, A.B., Chan, H. and Murthy, J. (1986). A general method for retrievingthe components of a genetically engineered fusion protein. DNA 5, 11-20.Takahara, M., Sagai, H., Inouye, S. and Inouye, M. (1988). Secretion of human superoxidedismutase in Escherichia coli. Bio/Technology 6, 195-198.Taniguchi, T., Matsui, H., Fujita, T., Takaoka, C., Kashima, N., Yoshimoto, R. and Hamuro,J. (1983). Structure and expression of a cloned cDNA for human interleukin-2. Nature 302,305-310.Taylor, M.E. and Drickamer, K. (1991). Carbohydrate-recognition domains as tools for rapidpurification of recombinant eukaryotic proteins. Biochem. J. 274, 575-580.185Taylor, M.E. and Drickamer, K. (1992). Expression and purification of the cytoplasmic tail ofan endocytic receptor by fusion to a carbohydrate-recognition domain. Protein Expr. Pur. 3,308-312.Teather, R.M. and Wood, P.J. (1982). Use of Congo Red-polysaccharide interactions inenumeration and characterization of cellulolytic bacteria from the bovine rumen. Appl. Environ.Microbiol. 43, 777-780.Thomas, G., Thorne, B., Thomas, L., Allen, R., Hruby, D., Fuller, R. and Thorner, J. (1988).Yeast KEX2 endopeptidase correctly cleaves a neuroendocrine prohormone in mammalian cells.Science 241, 226-231.Thomas, L.M., Huntington, P.J., Mead, L.J., Wingate, D.L., Rogerson, B.A. and Lew, A.M.(1992). A soluble recombinant fusion protein of the transmembrane envelope protein of equineinfectious anaemia virus for ELISA. Veterinary Microbiology 31, 127-137.Tomme, P., van Tilbeurgh, H., Pettersson, G., van Damme, J., Vandekerckhove, J., Knowles,J., Teeri, T. and Claeyssens, M. (1988). Studies of the cellulolytic system of Trichoderma reeseiQM 9414. Analysis of domain function in two cellobiohydrolases by limited proteolysis. Eur. J.Biochem. 170, 575-581.Tonouchi, N., Oouchi, N., Kashima, N., Kawai, M., Nagase, K., Okano, A., Matsui, H.,Yamada, K., Hirano, T. and Kishimoto, T. (1988). High-level expression of human BSF-2/IL6cDNA in Escherichia coli using a new type of expression-preparation system. J. Biochem. 104,30-34.Torriani, A. (1966). Alkaline phosphatase from Escherichia coli. In Procedures in Nucleic AcidsResearch, ed. G.L. Cantoni and D.R. Davies. Harper and Row, New York; pp. 224-235.Towbin, H., Staehelin, T. and Gordon, J. (1979). Electrophoretic transfer of proteins frompolyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc. Natl. Acad.Sci. USA 76, 4350-4354.Traunecker, A., Oliveri, F. and Karjalainen, K. (1991). Myeloma based expression system forproduction of large mammalian proteins. Trends Biotechnol. 9, 109-113.Tsuji, T., Nakagawa, R., Sugimoto, N. and Fukuhara, K. (1987). Characterization of disulfidebonds in recombinant proteins: reduced human interleukin 2 in inclusion bodies and its oxidativerefolding. Biochemistry 26, 3129-3134.Uhlen, M. and Moks, T. (1990). Gene fusions for purpose of expression: an introduction.Methods Enzymol. 185, 129-143.Uhlen, M., Nilsson, B., Guss, B., Lindberg, M., Gatenbeck, S. and Philipson, L. (1983).Gene fusion vectors based on the gene for staphylococcal protein A. Gene 23, 369-378.Ullman, A. (1984). One-step purification of hybrid proteins which have P-galactosidase activity.Gene 29, 27-31.van Dyke, M.W., Sirito, M. and Sawadogo, M. (1992). Single-step purification of bacteriallyexpressed polypeptides containing an oligo-histidine domain. Gene 111, 99-104.van Kimmenade, A., Dang, W. and Kastelein, R.A. (1989). Secretion of rnurine and humaninterleukin-2 by Escherichia coli. J. Biotechnol. 11, 11-24.186van Tilbeurgh, - H., Tomme, P., Claeyssens, M., Bhikhabhai, R. and Pettersson, G. (1986).Limited proteolysis of the cellobiohydrolase I from Trichoderma reesei. Separation of functionaldomains. FEBS Lett. 204, 223-227.Varadarajan, R., Szabo, A. and Boxer, S.G. (1985). Cloning, expression in Escherichia coli,and reconstitution of human myoglobin. Proc. Natl. Acad. Sci. USA 82, 5681-5684.Vieira, J. and Messing, J. (1987). Production of single-stranded plasmid DNA. MethodsEnzymol. 160, 3-11.Villa, R., de Fazio, G., Donini, S., Tarchi, G. and Canosi, U. (1988). Expression inEscherichia coli and characterization of human growth-hormone-releasing factor. Eur. J.Biochem. 171, 137-141.Volkin, D.B. and Klibanov, A.M. (1989). Minimizing protein inactivation. In Protein function.A practical approach, ed. T.E. Creighton. IRL Press, Oxford; pp. 1-24.Vosters, A.F., Evans, D.B., Tarpley, W.G. and Sharma, S.K. (1992). On the engineering ofrDNA proteins for purification by immobilized metal affinity chromatography: applications toalternating histidine-containing chimeric proteins from recombinant Escherichia coli. ProteinExpr. Pur. 3, 18-26.Vyas, N.K., Vyas, M.N. and Quiocho, F.A. (1988). Sugar and signal-transducer binding sitesof the Escherichia coli galactose chemoreceptor protein. Science 242, 1290-1295.Wadensten, H., Ekebacke, A., Hammarberg, B., Holmgren, E., Kalderen, C., Tally, M.,Moks, T., Uhlen, M., Josephson, S. and Hartmanis, M. (1991). Purification andcharacterization of recombinant human insulin-like growth factor II (IGF-II) expressed as asecreted fusion protein in Escherichia coli. Biotechnol. Appl. Biochem. 13, 412-421.Waldenstrom, M., Holmgren, E., Attersand, A., Kalderen, C., Lowenadler, B., Raden, B.,Hansson, L. and Pohl, G. (1991). Synthesis and secretion of a fibrinolytically active tissue-typeplasminogen activator variant in Escherichia coli. Gene 99, 243-248.Wang, A., Lu, S.-D. and Mark, D.F. (1984). Site-specific mutagenesis of the humaninterleukin-2 gene: structure-function analysis of the cysteine residues. Science 224, 1431-1433.Warren, R.A.J., Beck, C.F., Gilkes, N.R., Kilburn, D.G., Langsford, M.L., Miller, R.C., Jr.,O'Neill, G.P., Scheufens, M. and Wong, W.K.R. (1986). Sequence conservation and regionshuffling in an endoglucanase and an exoglucanase from Cellulomonas fimi. Proteins 1, 335-341Weigle, U., Meyer, M. and Sebald, W. (1989). Mutant proteins of human interleukin 2.Renaturation yield, proliferative activity and receptor binding. Eur. J. Biochem. 180, 295-300.Weir, M.P. and Sparks, J. (1987). Purification and renaturation of recombinant humaninterleukin-2. Biochem. J. 245, 85-91.Weissbach, A. and Hurwitz, J. (1959). The formation of 2-keto-3-deoxyheptonic acid in extractsof Escherichia coli B. J. Biol. Chem. 234, 705-709.Wels, W., Harwerth, I.-M., Zwickl, M., Hardman, N., Groner, B. and Hynes, N.E. (1992).Construction, bacterial expression and characterization- of a hifunetional- singlc-ehairt -antibody-phosphatase fusion protein targeted to the human erbB-2 receptor. Bio/Technology 10, 1128-1132.187Whittle, D.J., Kilburn, D.G., Warren, R.A.J. and Miller, R.C., Jr. (1982). Molecular cloningof a Cellulomonas fimi cellulase gene in Escherichia coli. Gene 17, 139-145.Williams, D.P., Parker, K., Bacha, P., Bishai, W., Borowski, M., Genbauffe, F., Strom, T.B.and Murphy, J.R. (1987). Diphtheria toxin receptor binding domain substitution withinterleukin-2: genetic construction and properties of a diphtheria toxin-related interleukin-2 fusionprotein. Prot. Eng. 1, 493-498.Williams, G.T. and Neuberger, M.S. (1986). Production of antibody-tagged enzymes bymyeloma cells: application to DNA polymerase I Klenow fragment. Gene 43, 319-324.Wolber, V., Maeda, K., Schumann, R., Brandmeier, B., Wiesmiiller, L. and Wittinghofer, A.(1992). A universal expression-purification system based on the coiled-coil interaction of myosinheavy chain. Bio/Technology 10, 900-904.Wong, W.K.R., Gerhard, B., Guo, Z.M., Kilburn, D.G., Warren, R.A.J. and Miller, R.C., Jr.(1986). Characterization and structure of an endoglucanase gene cenA of Cellulomonas fimi.Gene 44, 315-324.Wood, T.M. (1988). Preparation of crystalline, amorphous, and dyed cellulase substrates.Methods Enzymol. 160, 19-25.Woodward, J., Hayes, M.K. and Lee, N.E. (1988). Hydrolysis of cellulose by saturating andnon-saturating concentrations of cellulase: implications for synergism. Bio/Technology 6, 301-304.Wynn, R.M., Davie, J.R., Cox, R.P. and Chuang, D.T. (1992). Chaperonins groEL and groESpromote assembly of heterotetramers (alpha 2 beta 2) of mammalian mitochondrial branched-chain alpha-keto acid decarboxylase in Escherichia coli. J. Biol. Chem. 267, 12400-12403.Yanisch-Perron, C., Vieira, J. and Messing, J. (1985). Improved M13 phage cloning vectorsand host strains: nucleotide sequences of the Ml3mp and pUC vectors. Gene 33, 103-119.Yansura, D.G. (1990). Expression as trpE fusion. Methods Enzymol. 185, 161-166.Yasukawa, K. and Saito, T. (1990). A new system for the expression of human interleukin-6 inEscherichia coli. Biotechnol. Lett. 12, 419-424.Young, R.A. (1986). Structure, swelling and bonding of cellulose fibres. In Cellulose.Structure, Modification and Hydrolysis, ed. R.A. Young and R.M. Rowell. John Wiley andSons, New York; pp. 91-128.Zhou, C., Abaigar, L. and Jong, A.Y. (1990). A protocol for using 17 DNA polymerase inoligonucleotide site-directed mutagenesis. BioTechniques 8, 503.Zweig, M., Showalter, S.D., Du Bois, G.C., Sisk, W.P. and Court, D.L. (1987). Detection ofheterologous fusion proteins in Escherichia colt with a monoclonal antibody. Gene 55, 47-53.188


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items