- Email: [email protected]
Recombinant avidin and avidin–fusion proteins
Biomolecular Engineering 16 (1999) 87 – 92 www.elsevier.com/locate/geneanabioeng
Recombinant avidin and avidin–fusion proteins Kari J. Airenne, Varpu S. Marjoma¨ki, Markku S. Kulomaa * Department of Biological and En6ironmental Science, Uni6ersity of Jy6a¨skyla¨, FIN-40351, Jy6a¨skyla¨, Finland
Abstract Both chicken egg-white avidin and its bacterial relative streptavidin are well known for their extraordinary high affinity with biotin (Kd 10 − 15 M). They are widely used as tools in a number of affinity-based separations, in diagnostic assays and in a variety of other applications. These methods have collectively become known as (strept)avidin – biotin technology. Biotin can easily and effectively be attached to different molecules, termed binders and probes, without destroying their biological activity. The exceptional stability of the avidin–biotin complex and the wide range of commercially available reagents explain the popularity of this system. In order by genetic engineering to modify the unwanted properties of avidin and to further expand the existing avidin–biotin technology, production systems for recombinant avidin and avidin – fusion proteins have been established. This review article presents an overview of the current status of these systems. Future trends in the production and applications of recombinant avidin and avidin–fusion proteins are also discussed. © 1999 Elsevier Science B.V. All rights reserved. Keywords: Recombinant avidin; Avidin–biotin; Avidin fusion protein; Baculovirus expression vector system, BEVS; Affinity tag/handle; Semliki Forest virus, SFV; Review
1. Introduction Avidin (Avd) is a widely used tool in numerous applications of (strept)avidin – biotin technology (for a review, see [1]). Avd can readily and economically be purified in high amounts from chicken egg-white [2], and there is thus no direct need to produce natural chicken egg-white Avd in a recombinant system. The high isoelectric point of Avd (pI10.5) together with the presence of a carbohydrate chain have, however, been to the constant detriment of some applications due to its non-specific binding. Streptavidin, a non-glycosylated and neutrally charged bacterial relative of Avd, has therefore displaced Avd in many applications of (strept)avidin–biotin technology, although streptavidin is less abundant and more expensive than chicken Avd. We have developed both bacterial [3,4] and eukaryotic expression systems [5] for the production of recombinant Avd, and for its mutated forms and avidin–fusion proteins. These have granted us a better possibility to understand the structure-function rela* Corresponding author. Tel.: +358-14-2602272; fax: + 358-142602221. E-mail address: [email protected] (M.S. Kulomaa)
tionship of tetrameric avidin and its extraordinary tight interaction with biotin (highest known non-covalent interaction in nature with a Kd of about 10 − 15 M between Avd and biotin). They have also provided a chance to modify by site-directed mutagenesis the above-mentioned unwanted intrinsic properties of Avd and to design Avd mutants more suitable for applications. Avd can be deglycosylated by enzymatic treatment or by incubation with certain microbial culture [6] and the positive charge can be decreased by chemical treatment (acetylation, succinylation, etc.) [1]. In spite of this, recombinant systems offer a more direct and systematic way of achieving all this. In addition, Avd fusion proteins provide novel tools for basic research and for a wide variety of applications due to the ease with which such hybrid proteins can be detected and purified using the biotin-binding activity.
2. Production of recombinant avidin in E. coli Given an easy-to-use and low-cost expression system [7], we decided first to try an E. coli expression system for re-Avd production. Avd is not heavily post-translationally modified and contains only one intramolecular disulfide bridge [8–10]. The carbohydrate chain is not
1389-0344/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved. PII: S 1 0 5 0 - 3 8 6 2 ( 9 9 ) 0 0 0 3 7 - 6
88
K.J. Airenne et al. / Biomolecular Engineering 16 (1999) 87–92
needed for biotin-binding [11] and the size of Avd is relatively small ( 15.7 kDa per identical subunit [12]). Thus, at least in theory, the structural properties of Avd should not restrict its bacterial production [13]. The size per subunit of 14 kDa for re-Avd produced by E. coli [3] agreed well with the computer-predicted size of unglycosylated re-Avd containing the starting methionine (13 960 Da). The accuracy of the N-terminal end of re-Avd was not studied by N-terminal sequencing. The bacterial cells, however, produced biologically active re-Avd, since it could be purified by a single step using biotin-agarose affinity chromatography. Native biotin-binding activity was also demonstrated by immuno-blot analysis using biotinylated alkaline phosphatase as a probe. According to a rough estimation from the immunoblot, the first construct pAVEX8 produced 0.3 – 0.5 mg of re-Avd per liter of culture at +37°C. In order to improve the expression rate, the first ten codons were changed according to the so-called ‘optimal E. coli codon usage’. The resulting plasmid, pAVEX15, yielded at least 2–4 times more of re-Avd than was obtained using pAVEX8. The best record was about 5 mg of re-Avd per liter of culture. Streptavidin has been expressed as an insoluble form (inclusion body) in E. coli, and a successful renaturation/denaturation procedure has been established to make it biologically active after that [14]. In spite of this, we wanted to obtain soluble re-Avd in order to simplify the purification procedure. Indeed, mainly soluble re-Avd was obtained simply by lowering the growth temperature down to 25°C, and the re-Avd was still easily purified in a single step. Expression of re-Avd in soluble form, however, resulted in a significantly lower overall yield, and moreover, recovery was only about 40–50% at best. The amount of purified re-Avd remained as a rule under 1 mg/l of culture. In order to increase both yield and recovery, expression of re-Avd was carried out using a biotin deficient strain of E. coli (SL 133 [15]) and in a biotin-free culture medium (M9) to avoid the saturation of the re-Avd by biotin during expression. Recovery was somewhat better when the SL 133 strain was used, but in practice almost equal amounts of re-Avd were also achieved with the JM109 strain. This was due the fact that the SL 133 cells did not grow well without a rich medium or the addition of biotin to the culture medium to a concentration sufficient to saturate most of the re-Avd biotin-binding sites. The low recovery of re-Avd could partially be explained by the improper folding and assembly of the protein in the reducing environment present in the cytoplasm of E. coli. The C-terminal deletion of the last five amino acids (124 – 128) of re-Avd was not responsible for the low recovery, since there was no major difference between re-Avd and full-length re-Avd in terms of either recovery or yield (unpublished results).
As an essential cofactor in biotin-dependent carboxylases, biotin plays an important role in all cells as an absolute requirement for higher organisms [16]. The detrimental effect of biologically active re-Avd on E. coli cells may thus be difficult or impossible to resolve in terms of increasing the yield and recovery of re-Avd when produced in the cytoplasmic space in soluble (active) form. One way to overcome or reduce this lethal effect could be the production of re-Avd into the periplasmic space of E. coli. As a secretory protein Avd should also be recognized by the bacterial secretory machinery. The periplasmic space should also favor the correct folding of re-Avd by providing an oxidizing environment in which the intramolecular disulfide bond of re-Avd subunit could form more easily. Our preliminary, still unpublished, results suggest that the natural signal sequence of Avd is capable of leading re-Avd to a secretory pathway in E. coli. The overall yield of re-Avd in E. coli might also be improved by producing it as a tandem repeat or as another kind of fusion protein. Such a strategy has successfully been used with some recombinant proteins, especially with small polypeptides [17–20], which are recognized as abnormal by the proteolytic system of the cell [21]. Indeed, the results of the Avd fusion protein experiments [4] showed that the overall yield of Avd fusion proteins was considerably better than that of re-Avd. However, a production strategy of this kind suffers from the fact that an additional protease digestion step is needed to cleave the re-Avd from its fusion partner(s). The study of other bacteria than E. coli as a host might also lead to improved recovery and yield, although preliminary trials with Bacillus have so far been unsuccessful (Dr. Yuval Shoham and Dr. Edward A. Bayer, personal communication). In spite of all this, biologically active recombinant chicken Avd can be produced in E. coli in a soluble form which can be easily purified in a single step using affinity chromatography. However, without further improvements, the relatively low productivity of re-Avd in E. coli reduces the value of bacterial cells as an attractive host for Avd-related studies.
3. Production of recombinant avidin in insect cells After experimenting with a prokaryotic expression system, it was a natural next step to try an eukaryotic one. Of all the available eukaryotic expression systems, the baculovirus expression vector system (BEVS, [22]) was the most attractive due to the recent advances in this system. The latest virus construction techniques have speeded up the process of recombinant virus construction so that the primary results of protein production can be achieved in about two weeks (Bac-to-Bac Baculovirus expression system, Gibco BRL) [23,24].
K.J. Airenne et al. / Biomolecular Engineering 16 (1999) 87–92
This together with all its other promising properties led us finally to choose the BEVS as the next re-Avd production system. According to quantitative analysis by ELISA, the insect cells produced about 12 mg of re-Avd per liter of suspension culture (106 cells/ml or 109 cells/l) [5]. Most of this re-Avd was secreted into the culture medium, suggesting that the natural signal sequence of the Avd was efficiently recognized and processed by the insect cell secretory machinery. A portion of the re-Avd detected in the culture medium was probably due to lysis of the cells late in the infection. The electron microscopy results, however, supported the notion that the natural signal sequence of Avd in the insect cells was functional. As in the bacterial system, the initial attempts to purify intracellular re-Avd by 2-iminobiotin-agarose resulted in low recovery (12%) of highly purified re-Avd. This recovery was, however, notably ( \ 90%) enhanced simply by excluding biotin from the culture medium, indicating that the recovery (activity) problems were connected to the existence of biotin in the culture medium. Indeed, the original medium contained enough biotin to saturate all the free biotin-binding sites of the re-Avd produced. Unlike the bacterial system, biotin did not seem to be needed for the efficient production of re-Avd or the successive stages of the baculovirus infection. In order also to allow for the efficient recovery of re-Avd from the culture medium, we even omitted fluronic-F68 from it. The major intracellular form of re-Avd produced in the baculovirus-infected cells was a polypeptide of about 18 kDa. In addition, two other minor bands corresponding in size to about 17 and 15.5 kDa were detected by immuno-blot analysis [5]. These different forms were shown to represent glycosylated (18 and 17 kDa) and unglycolsylated (15.5 kDa) re-Avd. It would be interesting to study the composition of the carbohydrate residues of re-Avd in more detail to find out if they correspond to those of natural Avd. Natural Avd has been shown to contain at least three distinct oligosaccharide structures of similar composition and size. In addition to oligomannosidic and bisected hybrid components, the Avd carbohydrate also contains nonbisected hybrid structures [25]. The purified re-Avd was assembled mainly into tetramers, which exhibited a high level of thermostability, and was further stabilized following biotin binding. These results together with the glycosylation data strongly suggest that the biotin-binding and structural properties of the re-Avd were similar or identical to those of natural egg-white Avd. Hence the insect system is worth considering in future site-directed mutagenesis studies of re-Avd.
89
4. Production of avidin–fusion proteins in E. coli Chicken egg-white Avd meets most of the criteria required from an ideal polypeptide tail to be used in fusion (chimeric) proteins. As a protein it is relatively small ( 15.7 kDa per identical subunit), stable, heat resistant, soluble, and needs no co-factors to be functional [12]. Although many peptide or protein tails have been introduced during the past decade [26,27], none of them have so far been perfect. Avd could thus offer an efficient additional tool for fusion protein applications due to the ease with which such hybrid proteins could be detected and purified using the biotin-binding activity. A large number of reagents for (strept)avidin–biotin technology are already commercially available [1], and Avd–fusion proteins could further expand this already widely used technology. In order to study whether it is possible to use Avd as an affinity tail in fast and simple detection and purification of fusion proteins, we produced Avd as C- and N-terminally linked with Schistosoma japonicum glutathione S-transferase (GST) in E. coli [4]. A commercial pGEX-3X expression vector (Pharmacia Biotech) was used to produce a fusion protein in which Avd was linked to the C-terminus of GST. In order to enable the production of GST or any desired protein as N-terminally linked with the C-terminus of re-Avd, a new expression vector, pAVEX16C, was constructed. The pAVEX16C was based on the pAVEX15 [3], into which the IgG3 hinge region [28] and a multiple cloning site were added behind the re-Avd region. The hinge region was included into the pAVEX16C vector to facilitate the autonomous folding and activity of both parts of the fusion protein. Approximately 5–10 mg of both fusion proteins per liter of culture was produced at room temperature (21–22°C). From this we were able to recover 1–2 mg of highly purified fusion proteins per liter of culture. In accordance with the previous re-Avd results, both fusion proteins were synthesized in considerably (5–10 times) higher amounts at 37°C, but most of the proteins accumulated were in inclusion bodies. As with re-Avd, it proved possible to purify only a relatively small fraction (10–20%, estimation from immuno-blot) of the soluble Avd-GST and GST-Avd with the affinity chromatography protocol used. The reasons were most probably largely the same as those discussed in conjunction with re-Avd. To avoid the toxic effect to the cells of the re-Avd part of the fusion protein, biotin had to be present in the culture medium at a concentration high enough to saturate most of the four biotinbinding sites. According to purification results, the activity of the Avd part of the fusion protein did not seem to be affected by the orientation of the Avd in the fusion protein. We were able to purify both the Avd-GST and
90
K.J. Airenne et al. / Biomolecular Engineering 16 (1999) 87–92
GST-Avd by affinity chromatography on a biotinagarose column. In another series of our experiments, b-galactosidase activity was neither affected when it was fused to the C-terminus of re-Avd (unpublished results), suggesting that the desired proteins produced by the pAVEX16C vector are fully functional. In addition to Avd-GST and GST-Avd, we have produced several other prokaryotic and eukaryotic proteins fused to the C-terminal end of re-Avd. These include, for instance, b-galactosidase (130.9 kDa), EPA (barley endoproteinase A, 41.2 kDa), chicken PR A and B (progesterone receptor A and B; 72 and 86 kDa, respectively), Parvo (part of the nuclear localization signal of canine parvovirus capsid protein, 23.2 kDa) and IgG3 (hinge region of mouse IgG3, 14.9 kDa) (unpublished results). All of these were easily detected by the Avd-antibody after production in E. coli cells, but only in the case of Avd-IgG3 and Avd-Parvo was the purification efficiency at the same level as in the GST fusions when purification was done by biotinagarose affinity chromatography. In accordance with published results [29], the level of purification efficiency thus seemed to be better the smaller the fusion partner. In the case of Avd-b-galactosidase, the tetrameric nature of both re-Avd and b-galactosidase might lead to problems such as unnatural or intermolecular oligomerization, resulting in aggregation or reduced activity of the Avd tail. The problems with Avd-PRA and B might be related to the extremely unstable nature of these receptors. Biologically active re-Avd can thus be produced as C- or N-terminally linked to the desired protein in E. coli. The production system could, however, be improved in several ways. As with re-Avd, the production of the fusion proteins into the periplasmic space of E. coli could lead to better recovery and higher overall yields of pure proteins. By developing a general denaturation/renaturation protocol for Avd – fusion proteins, the overall yield of fusion proteins might be considerably increased simply by synthesizing them as inclusion bodies in a higher growth temperature (37°C).
5. Production of avidin – fusion proteins in insect cells We have recently constructed a novel baculovirus transfer vector (pbacAVs+C) which also enables Avd fusion protein production, detection and purification in insect cells [30]. The pbacAVs+ C encodes for a secretion-compatible form of recombinant Avd (signal peptide and amino acids 1 – 123 of Avd+Gln at the C-terminal end) connected C-terminally to the IgG hinge region and multiple cloning site. It thus encodes an otherwise similar re-Avd as does the pAVEX16C [4],
except it also encodes the signal sequence in front of re-Avd. Hevein (Hev) is a small (43 kDa) cysteine-rich protein in the latex of the rubber tree (He6ea brasiliensis [31]). It is reported to be a major allergen in allergies against natural rubber latex products [32]. The results of using the pbacAVs+ C for Hev production in insect cells has clearly shown that the pbacAVs+ C provides an efficient tool for combining the power of the baculovirus system with the multipurpose Avd-tag [30]. Milligrams of highly purified Av-hev were obtained from a 25 ml culture (5 × 107 cells) of re-baculovirus-infected cells in a single step. The purified Av-hev formed tetramers in solution and had the tight biotin-binding activity characteristic of tetrameric Avd. The Av-hev was purified efficiently on a 2-iminobiotin agarose column both from the cells and also from the medium when biotin and Pluronic F-68 were omitted from culture medium. Furthermore, the Hev component could be released from Avd-tag by an enterokinase cleavage. The antigenic properties of the released reHev were indistinguishable from the native molecule, which makes it (and its mutated forms) an interesting object of study in research into the development and maintenance of allergies at the molecular level. These results also support the general view favoring the of independent folding of both ends of Avd fusion proteins. Altogether, our results indicate that Avd works well in the baculovirus expression system as a multipurpose secretable tag. The Avd-tag is very stable and retains its characteristically high biotin-binding activity. It is also appropriate for use with the large range of commercially available reagents in avidin–biotin technology. The reliability with which Avd fusion proteins can be detected, purified and immobilized, is the basis for the use of our system as an elegant alternative in eukaryotic fusion protein production.
6. Membrane traffic research using avidin fusion proteins Since Avd is normally expressed only in avian and amphibian tissues and is absent from mammalian cells (for a review, see [12] or [33]), it therefore provides a practical marker for a wide range of morphological studies in several animal cell lines. We have expressed chimeric proteins containing Avd fused to for example cation-independent mannose 6-phosphate receptor (CIMPR; [34]). The constructed fusion proteins contained Avd in the N-terminus followed by the transmembrane and selected cytoplasmic sequences from the CI-MPR containing various signals for endocytosis and intracellular sorting.
K.J. Airenne et al. / Biomolecular Engineering 16 (1999) 87–92
The Semliki Forest virus vector was used to express the fusion proteins in BHK cells [35,36]. The Semliki Forest virus expression systems have recently successfully been used to express several membrane and cytoplasmic proteins in various cell lines (for a review, see [35,36]). This system is advantageous for several reasons. Semliki Forest virus contains a capped and polyadenylated single-stranded RNA genome of positive polarity which when introduced into an animal cell is effectively read as a message. After translation of its own RNA replicase, the recombinant genome is effectively replicated and transcribed. Co-transfection of a helper RNA and a recombinant RNA into BHK cells by electroporation produces infectious viruses which can be used to infect several cell lines to produce a recombinant protein. The infection is carried out in an infection medium (culture medium supplemented with 0.2% BSA without serum) for 1 h, which is long enough for efficient replication to take place. The cells are then chased in a complete medium (with serum) until the required amount of the protein is obtained. For our purposes, infection for 1 h followed by a 4 – 5-h chase produced : 1 mg of Avd fusion protein per 1 mg of total protein in BHK cells. For our studies on membrane trafficking, it is essential not to overproduce the protein, since it may cause mistargeting and disturbances in the various steps involved. However, if needed, protein expression may be extended up to at least 48 h after infection, since apparent cytopathogenic effects appear very late, in contrast to many other viral expression systems. It was found possible to label Avd chimeras in vivo during their expression with a membrane permeable dye, e.g. with biotinylated fluorescein [34]. Biotin-binding suggested that the Avd was indeed correctly folded. Double-labeling with biotinylated fluorescein and antiAvd suggested that the entire Avd pool was able to bind biotin in various intracellular locations. This is an advantage, since various functional tests can be created with biotinylated conjugates. We have developed an in vivo meeting assay which detects the formation of the Avd –biotin complex after the internalization of biotinylated horse radish peroxidase (HRP; [34]). Using different internalization time periods for biotinylated HRP, the cellular location of Avd chimeras can be studied in the endosomal pathway. The labeling of Avd with biotinylated fluorescein also make it possible to label the chimeras during expression and then follow the fate of the chimera during subsequent experimental conditions. We have also prepared green fluorescent protein (GFP)-chimeras of Avd fusion proteins which contain GFP in the C-terminus. Because GFP-fluoresescence is very stable we have been able to follow the chimeras for long time periods under a fluoresence microscope.
91
7. Drug targeting Two recent papers have described the production of recombinant Avd for affinity targeting purposes. Shin and co-workers [37] produced Avd joined to various positions within the mouse/human chimeric IgG3 heavy chain in the TWS2 cell line. The purified antibodyavidin (Ab-Avd) fusion proteins retained their specificity for antigen and bound biotin with sufficient affinity for the retention of biotinylated therapeutics in vivo. Such Ab-Avd chimeras might prove to be valuable in delivering biotinylated ligands such as nuclides, drugs, and cytotoxic agents to a specific target expressing an antigen recognized by the associated antibody. Another potential approach for targeting tumors or tissues were described by Walker and his collaborators [38]. A functional re-Avd was introduced into two human tumor cell lines (HeLa and Hep G2) and into a murine breast carcinoma cell line (16/C) using either a transient or stable gene expression, respectively. The results showed that transfected human or murine carcinoma cells could sequester and concentrate a measurable amount of biotin, suggesting that a similar approach could be attempted using the Avd gene in vivo. Indeed, the targeted expression of the Avd gene in different tissues in vivo would provide great potential for therapeutic and imaging purposes.
8. Future perspectives The established production systems provide a basis for the efficient production of the desired re-Avds as well as Avd–fusion proteins for different purposes. They should further expand the existing widely used and versatile (strept)avidin–biotin technology. Improvements in the expression systems as well as in the mutated versions of Avd (which are under construction in our laboratory) may lead to an even more efficient expression of re-Avds and avidin fusion proteins with improved and desired properties. Such proteins could well find their way to a wide range of novel applications.
Acknowledgements We wish to express our deepest appreciation to the following people for fruitful and indispensable cooperation over the past years; Dr. Edward A Bayer and Professor Meir Wilchek (Department of Biological Chemistry, The Weizmann Institute of Science, Rehovot, Israel), Dr. Christian Oker-Blom (VTT, Biotechnology and Food Research, Espoo, Finland), Professor Jean Gruenberg, University of Geneva, Geneva, Switzerland), Dr. Vesa Olkkonen (Finnish Institute of
92
K.J. Airenne et al. / Biomolecular Engineering 16 (1999) 87–92
Occupational Health, Helsinki, Finland), Professor Timo Palosuo and Mr. Jari Mikkola (National Public Health Institute, Helsinki, Finland), Dr. Harri Alenius (Finnish Institute of Occupational Health, Helsinki, Finland), Dr. Nisse Kalkkinen (Institute of Biotechnology, University of Helsinki, Helsinki, Finland), Dr. M. Siti Arija and H.Y. Yeang (Rubber Research Institute, Kuala Lumbur, Malaysia) and Kati Juuti, Anna Laukkanen, Olli Laitinen, Ari Marttila, Eeva-Liisa Punnonen, and Paula Sarkkinen from our department.
References [1] Wilchek M, Bayer EA, editors. Methods in Entzymol 184, 1990. [2] Heney G, Orr GA. Anal Biochem 1981;114:92–6. [3] Airenne KJ, Sarkkinen P, Punnonen E-L, Kulomaa MS. Gene 1994;144:75 – 80. [4] Airenne KJ, Kulomaa MS. Gene 1995;167:63–8. [5] Airenne KJ, Oker-Blom C, Marjoma¨ki VS, Bayer EA, Wilchek M, Kulomaa MS. Prot Expr Purif 1997;9:100–8. [6] Bayer EA, Meester FD, Kulik T, Wilchek M. Appl Biochem Biotech 1995;53:1 – 9. [7] Gold L. Methods Entzymol 1990;185:11–4. [8] Livnah O, Bayer EA, Wilchek M, Sussman JL. Proc Natl Acad Sci USA 1993;90:5076–80. [9] Pugliese L, Coda A, Malcovati M, Bolognesi M. J Mol Biol 1993;231:698 – 710. [10] Pugliese L, Malcovati M, Coda A, Bolognesi M. J Mol Biol 1994;235:42 – 6. [11] Hiller Y, Gershoni JM, Bayer EA, Wilchek M. Biochem J 1987;248:167 – 71. [12] Green NM. Avd Protein Chem 1975;29:85–133. [13] Goeddel DV. Methods Enzymol 1990;185:3–7. [14] Sano T, Cantor CR. Proc Natl Acad Sci USA 1990;87:142 – 6. [15] Lesley SA, Patten PA, Schultz PG. Proc Natl Acad Sci USA 1993;90:1160 – 5.
.
[16] Dakshinamurti K, Chauhan J. Methods Enzymol 1990;184:93– 102. [17] Shen S-H. Proc Natl Acad Sci USA 1984;81:4627 – 31. [18] Schulz M-F, Buell G, Schmid E, Movva R, Selzer G. J Bacteriol 1987;169:5385– 92. [19] Butt TR, Jonnalagadda S, Monia BP, Sternberg EJ, Marsh JA, Stadel JM, Ecker DJ, Crooke ST. Proc Natl Acad Sci USA 1989;86:2540 – 4. [20] Hammarberg B, Nygren P-A, , Holmgren E, Elmblad A, Tally M, Hellman U, Moks T, Uhle´n M. Proc Natl Acad Sci USA 1989;86:4367 – 71. [21] Itakura K, Hirose T, Crea R, Riggs AD, Heyneker HL, Bolivar F, Boyer HW. Science 1977;198:1056– 63. [22] O’Reilly DR, Miller LK, Luckov VA. Baculovirus Expression Vectors: A Laboratory Manual. New York: Oxford University Press, 1994. [23] Davies AH. Biotechnology 1994;12:47 – 50. [24] Jones I, Morikawa Y. Curr Opin Biotech 1996;7:512 – 6. [25] Bruch RC, White HB. Biochemistry 1982;21:5334 – 41. [26] Uhle´n M, Moks T. Methods Enzymol 1990;185:129 – 43. [27] Nilsson J, Sta˚hl S, Lundeberg J, Uhle´n M, Nygren P-A, . Prot Expr Purif 1997;11:1 – 16. [28] Pack P, Plu¨ckthun A. Biochemistry 1992;31:1579 – 84. [29] Frangioni JV, Neel BG. Anal Biochem 1993;210:179 – 87. [30] Airenne KJ, Laitinen OH, Alenius H, Mikkola J, Palosuo T, Kalkkinen N, Arija MS, Yeang HY, Kulomaa MS. Prot Expr Purif 1999;17:139 – 45. [31] Acher BL. Biochem J 1960;75:236 – 40. [32] Alenius H, Kalkkinen N, Reunala T, Turjanmaa K, Palosuo T. J Immunol 1996;156:1618– 25. [33] Korpela JK. Med Biol 1984;62:5 – 26. [34] Marjoma¨ki V, Juuti K, Airenne KJ, Laukkanen A, Punnonen E-L, Olkkonen VM, Gruenberg J, Kulomaa M. Resubmitted, 1999. [35] Liljestro¨m P, Garoff H. Biotechnology 1991;9:1356 – 61. [36] Olkkonen VM, Liljestro¨m P, Dupree P, Garoff H, Simons K. Methods Cell Biol 1994;43:43 – 53. [37] Shin S-U, Wu D, Ramanathan R, Pardridge WM, Morrison SL. J Immunol 1997;158:4797– 804. [38] Walker L, Kulomaa MS, Bebok Z, Parker WB, Allan P, Logan J, Huang Z, Reynolds RC, King S, Sorscher EJ. J Drug Targ 1996;4:41 – 9.
Recombinant avidin and avidin–fusion proteins Kari J. Airenne, Varpu S. Marjoma¨ki, Markku S. Kulomaa * Department of Biological and En6ironmental Science, Uni6ersity of Jy6a¨skyla¨, FIN-40351, Jy6a¨skyla¨, Finland
Abstract Both chicken egg-white avidin and its bacterial relative streptavidin are well known for their extraordinary high affinity with biotin (Kd 10 − 15 M). They are widely used as tools in a number of affinity-based separations, in diagnostic assays and in a variety of other applications. These methods have collectively become known as (strept)avidin – biotin technology. Biotin can easily and effectively be attached to different molecules, termed binders and probes, without destroying their biological activity. The exceptional stability of the avidin–biotin complex and the wide range of commercially available reagents explain the popularity of this system. In order by genetic engineering to modify the unwanted properties of avidin and to further expand the existing avidin–biotin technology, production systems for recombinant avidin and avidin – fusion proteins have been established. This review article presents an overview of the current status of these systems. Future trends in the production and applications of recombinant avidin and avidin–fusion proteins are also discussed. © 1999 Elsevier Science B.V. All rights reserved. Keywords: Recombinant avidin; Avidin–biotin; Avidin fusion protein; Baculovirus expression vector system, BEVS; Affinity tag/handle; Semliki Forest virus, SFV; Review
1. Introduction Avidin (Avd) is a widely used tool in numerous applications of (strept)avidin – biotin technology (for a review, see [1]). Avd can readily and economically be purified in high amounts from chicken egg-white [2], and there is thus no direct need to produce natural chicken egg-white Avd in a recombinant system. The high isoelectric point of Avd (pI10.5) together with the presence of a carbohydrate chain have, however, been to the constant detriment of some applications due to its non-specific binding. Streptavidin, a non-glycosylated and neutrally charged bacterial relative of Avd, has therefore displaced Avd in many applications of (strept)avidin–biotin technology, although streptavidin is less abundant and more expensive than chicken Avd. We have developed both bacterial [3,4] and eukaryotic expression systems [5] for the production of recombinant Avd, and for its mutated forms and avidin–fusion proteins. These have granted us a better possibility to understand the structure-function rela* Corresponding author. Tel.: +358-14-2602272; fax: + 358-142602221. E-mail address: [email protected] (M.S. Kulomaa)
tionship of tetrameric avidin and its extraordinary tight interaction with biotin (highest known non-covalent interaction in nature with a Kd of about 10 − 15 M between Avd and biotin). They have also provided a chance to modify by site-directed mutagenesis the above-mentioned unwanted intrinsic properties of Avd and to design Avd mutants more suitable for applications. Avd can be deglycosylated by enzymatic treatment or by incubation with certain microbial culture [6] and the positive charge can be decreased by chemical treatment (acetylation, succinylation, etc.) [1]. In spite of this, recombinant systems offer a more direct and systematic way of achieving all this. In addition, Avd fusion proteins provide novel tools for basic research and for a wide variety of applications due to the ease with which such hybrid proteins can be detected and purified using the biotin-binding activity.
2. Production of recombinant avidin in E. coli Given an easy-to-use and low-cost expression system [7], we decided first to try an E. coli expression system for re-Avd production. Avd is not heavily post-translationally modified and contains only one intramolecular disulfide bridge [8–10]. The carbohydrate chain is not
1389-0344/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved. PII: S 1 0 5 0 - 3 8 6 2 ( 9 9 ) 0 0 0 3 7 - 6
88
K.J. Airenne et al. / Biomolecular Engineering 16 (1999) 87–92
needed for biotin-binding [11] and the size of Avd is relatively small ( 15.7 kDa per identical subunit [12]). Thus, at least in theory, the structural properties of Avd should not restrict its bacterial production [13]. The size per subunit of 14 kDa for re-Avd produced by E. coli [3] agreed well with the computer-predicted size of unglycosylated re-Avd containing the starting methionine (13 960 Da). The accuracy of the N-terminal end of re-Avd was not studied by N-terminal sequencing. The bacterial cells, however, produced biologically active re-Avd, since it could be purified by a single step using biotin-agarose affinity chromatography. Native biotin-binding activity was also demonstrated by immuno-blot analysis using biotinylated alkaline phosphatase as a probe. According to a rough estimation from the immunoblot, the first construct pAVEX8 produced 0.3 – 0.5 mg of re-Avd per liter of culture at +37°C. In order to improve the expression rate, the first ten codons were changed according to the so-called ‘optimal E. coli codon usage’. The resulting plasmid, pAVEX15, yielded at least 2–4 times more of re-Avd than was obtained using pAVEX8. The best record was about 5 mg of re-Avd per liter of culture. Streptavidin has been expressed as an insoluble form (inclusion body) in E. coli, and a successful renaturation/denaturation procedure has been established to make it biologically active after that [14]. In spite of this, we wanted to obtain soluble re-Avd in order to simplify the purification procedure. Indeed, mainly soluble re-Avd was obtained simply by lowering the growth temperature down to 25°C, and the re-Avd was still easily purified in a single step. Expression of re-Avd in soluble form, however, resulted in a significantly lower overall yield, and moreover, recovery was only about 40–50% at best. The amount of purified re-Avd remained as a rule under 1 mg/l of culture. In order to increase both yield and recovery, expression of re-Avd was carried out using a biotin deficient strain of E. coli (SL 133 [15]) and in a biotin-free culture medium (M9) to avoid the saturation of the re-Avd by biotin during expression. Recovery was somewhat better when the SL 133 strain was used, but in practice almost equal amounts of re-Avd were also achieved with the JM109 strain. This was due the fact that the SL 133 cells did not grow well without a rich medium or the addition of biotin to the culture medium to a concentration sufficient to saturate most of the re-Avd biotin-binding sites. The low recovery of re-Avd could partially be explained by the improper folding and assembly of the protein in the reducing environment present in the cytoplasm of E. coli. The C-terminal deletion of the last five amino acids (124 – 128) of re-Avd was not responsible for the low recovery, since there was no major difference between re-Avd and full-length re-Avd in terms of either recovery or yield (unpublished results).
As an essential cofactor in biotin-dependent carboxylases, biotin plays an important role in all cells as an absolute requirement for higher organisms [16]. The detrimental effect of biologically active re-Avd on E. coli cells may thus be difficult or impossible to resolve in terms of increasing the yield and recovery of re-Avd when produced in the cytoplasmic space in soluble (active) form. One way to overcome or reduce this lethal effect could be the production of re-Avd into the periplasmic space of E. coli. As a secretory protein Avd should also be recognized by the bacterial secretory machinery. The periplasmic space should also favor the correct folding of re-Avd by providing an oxidizing environment in which the intramolecular disulfide bond of re-Avd subunit could form more easily. Our preliminary, still unpublished, results suggest that the natural signal sequence of Avd is capable of leading re-Avd to a secretory pathway in E. coli. The overall yield of re-Avd in E. coli might also be improved by producing it as a tandem repeat or as another kind of fusion protein. Such a strategy has successfully been used with some recombinant proteins, especially with small polypeptides [17–20], which are recognized as abnormal by the proteolytic system of the cell [21]. Indeed, the results of the Avd fusion protein experiments [4] showed that the overall yield of Avd fusion proteins was considerably better than that of re-Avd. However, a production strategy of this kind suffers from the fact that an additional protease digestion step is needed to cleave the re-Avd from its fusion partner(s). The study of other bacteria than E. coli as a host might also lead to improved recovery and yield, although preliminary trials with Bacillus have so far been unsuccessful (Dr. Yuval Shoham and Dr. Edward A. Bayer, personal communication). In spite of all this, biologically active recombinant chicken Avd can be produced in E. coli in a soluble form which can be easily purified in a single step using affinity chromatography. However, without further improvements, the relatively low productivity of re-Avd in E. coli reduces the value of bacterial cells as an attractive host for Avd-related studies.
3. Production of recombinant avidin in insect cells After experimenting with a prokaryotic expression system, it was a natural next step to try an eukaryotic one. Of all the available eukaryotic expression systems, the baculovirus expression vector system (BEVS, [22]) was the most attractive due to the recent advances in this system. The latest virus construction techniques have speeded up the process of recombinant virus construction so that the primary results of protein production can be achieved in about two weeks (Bac-to-Bac Baculovirus expression system, Gibco BRL) [23,24].
K.J. Airenne et al. / Biomolecular Engineering 16 (1999) 87–92
This together with all its other promising properties led us finally to choose the BEVS as the next re-Avd production system. According to quantitative analysis by ELISA, the insect cells produced about 12 mg of re-Avd per liter of suspension culture (106 cells/ml or 109 cells/l) [5]. Most of this re-Avd was secreted into the culture medium, suggesting that the natural signal sequence of the Avd was efficiently recognized and processed by the insect cell secretory machinery. A portion of the re-Avd detected in the culture medium was probably due to lysis of the cells late in the infection. The electron microscopy results, however, supported the notion that the natural signal sequence of Avd in the insect cells was functional. As in the bacterial system, the initial attempts to purify intracellular re-Avd by 2-iminobiotin-agarose resulted in low recovery (12%) of highly purified re-Avd. This recovery was, however, notably ( \ 90%) enhanced simply by excluding biotin from the culture medium, indicating that the recovery (activity) problems were connected to the existence of biotin in the culture medium. Indeed, the original medium contained enough biotin to saturate all the free biotin-binding sites of the re-Avd produced. Unlike the bacterial system, biotin did not seem to be needed for the efficient production of re-Avd or the successive stages of the baculovirus infection. In order also to allow for the efficient recovery of re-Avd from the culture medium, we even omitted fluronic-F68 from it. The major intracellular form of re-Avd produced in the baculovirus-infected cells was a polypeptide of about 18 kDa. In addition, two other minor bands corresponding in size to about 17 and 15.5 kDa were detected by immuno-blot analysis [5]. These different forms were shown to represent glycosylated (18 and 17 kDa) and unglycolsylated (15.5 kDa) re-Avd. It would be interesting to study the composition of the carbohydrate residues of re-Avd in more detail to find out if they correspond to those of natural Avd. Natural Avd has been shown to contain at least three distinct oligosaccharide structures of similar composition and size. In addition to oligomannosidic and bisected hybrid components, the Avd carbohydrate also contains nonbisected hybrid structures [25]. The purified re-Avd was assembled mainly into tetramers, which exhibited a high level of thermostability, and was further stabilized following biotin binding. These results together with the glycosylation data strongly suggest that the biotin-binding and structural properties of the re-Avd were similar or identical to those of natural egg-white Avd. Hence the insect system is worth considering in future site-directed mutagenesis studies of re-Avd.
89
4. Production of avidin–fusion proteins in E. coli Chicken egg-white Avd meets most of the criteria required from an ideal polypeptide tail to be used in fusion (chimeric) proteins. As a protein it is relatively small ( 15.7 kDa per identical subunit), stable, heat resistant, soluble, and needs no co-factors to be functional [12]. Although many peptide or protein tails have been introduced during the past decade [26,27], none of them have so far been perfect. Avd could thus offer an efficient additional tool for fusion protein applications due to the ease with which such hybrid proteins could be detected and purified using the biotin-binding activity. A large number of reagents for (strept)avidin–biotin technology are already commercially available [1], and Avd–fusion proteins could further expand this already widely used technology. In order to study whether it is possible to use Avd as an affinity tail in fast and simple detection and purification of fusion proteins, we produced Avd as C- and N-terminally linked with Schistosoma japonicum glutathione S-transferase (GST) in E. coli [4]. A commercial pGEX-3X expression vector (Pharmacia Biotech) was used to produce a fusion protein in which Avd was linked to the C-terminus of GST. In order to enable the production of GST or any desired protein as N-terminally linked with the C-terminus of re-Avd, a new expression vector, pAVEX16C, was constructed. The pAVEX16C was based on the pAVEX15 [3], into which the IgG3 hinge region [28] and a multiple cloning site were added behind the re-Avd region. The hinge region was included into the pAVEX16C vector to facilitate the autonomous folding and activity of both parts of the fusion protein. Approximately 5–10 mg of both fusion proteins per liter of culture was produced at room temperature (21–22°C). From this we were able to recover 1–2 mg of highly purified fusion proteins per liter of culture. In accordance with the previous re-Avd results, both fusion proteins were synthesized in considerably (5–10 times) higher amounts at 37°C, but most of the proteins accumulated were in inclusion bodies. As with re-Avd, it proved possible to purify only a relatively small fraction (10–20%, estimation from immuno-blot) of the soluble Avd-GST and GST-Avd with the affinity chromatography protocol used. The reasons were most probably largely the same as those discussed in conjunction with re-Avd. To avoid the toxic effect to the cells of the re-Avd part of the fusion protein, biotin had to be present in the culture medium at a concentration high enough to saturate most of the four biotinbinding sites. According to purification results, the activity of the Avd part of the fusion protein did not seem to be affected by the orientation of the Avd in the fusion protein. We were able to purify both the Avd-GST and
90
K.J. Airenne et al. / Biomolecular Engineering 16 (1999) 87–92
GST-Avd by affinity chromatography on a biotinagarose column. In another series of our experiments, b-galactosidase activity was neither affected when it was fused to the C-terminus of re-Avd (unpublished results), suggesting that the desired proteins produced by the pAVEX16C vector are fully functional. In addition to Avd-GST and GST-Avd, we have produced several other prokaryotic and eukaryotic proteins fused to the C-terminal end of re-Avd. These include, for instance, b-galactosidase (130.9 kDa), EPA (barley endoproteinase A, 41.2 kDa), chicken PR A and B (progesterone receptor A and B; 72 and 86 kDa, respectively), Parvo (part of the nuclear localization signal of canine parvovirus capsid protein, 23.2 kDa) and IgG3 (hinge region of mouse IgG3, 14.9 kDa) (unpublished results). All of these were easily detected by the Avd-antibody after production in E. coli cells, but only in the case of Avd-IgG3 and Avd-Parvo was the purification efficiency at the same level as in the GST fusions when purification was done by biotinagarose affinity chromatography. In accordance with published results [29], the level of purification efficiency thus seemed to be better the smaller the fusion partner. In the case of Avd-b-galactosidase, the tetrameric nature of both re-Avd and b-galactosidase might lead to problems such as unnatural or intermolecular oligomerization, resulting in aggregation or reduced activity of the Avd tail. The problems with Avd-PRA and B might be related to the extremely unstable nature of these receptors. Biologically active re-Avd can thus be produced as C- or N-terminally linked to the desired protein in E. coli. The production system could, however, be improved in several ways. As with re-Avd, the production of the fusion proteins into the periplasmic space of E. coli could lead to better recovery and higher overall yields of pure proteins. By developing a general denaturation/renaturation protocol for Avd – fusion proteins, the overall yield of fusion proteins might be considerably increased simply by synthesizing them as inclusion bodies in a higher growth temperature (37°C).
5. Production of avidin – fusion proteins in insect cells We have recently constructed a novel baculovirus transfer vector (pbacAVs+C) which also enables Avd fusion protein production, detection and purification in insect cells [30]. The pbacAVs+ C encodes for a secretion-compatible form of recombinant Avd (signal peptide and amino acids 1 – 123 of Avd+Gln at the C-terminal end) connected C-terminally to the IgG hinge region and multiple cloning site. It thus encodes an otherwise similar re-Avd as does the pAVEX16C [4],
except it also encodes the signal sequence in front of re-Avd. Hevein (Hev) is a small (43 kDa) cysteine-rich protein in the latex of the rubber tree (He6ea brasiliensis [31]). It is reported to be a major allergen in allergies against natural rubber latex products [32]. The results of using the pbacAVs+ C for Hev production in insect cells has clearly shown that the pbacAVs+ C provides an efficient tool for combining the power of the baculovirus system with the multipurpose Avd-tag [30]. Milligrams of highly purified Av-hev were obtained from a 25 ml culture (5 × 107 cells) of re-baculovirus-infected cells in a single step. The purified Av-hev formed tetramers in solution and had the tight biotin-binding activity characteristic of tetrameric Avd. The Av-hev was purified efficiently on a 2-iminobiotin agarose column both from the cells and also from the medium when biotin and Pluronic F-68 were omitted from culture medium. Furthermore, the Hev component could be released from Avd-tag by an enterokinase cleavage. The antigenic properties of the released reHev were indistinguishable from the native molecule, which makes it (and its mutated forms) an interesting object of study in research into the development and maintenance of allergies at the molecular level. These results also support the general view favoring the of independent folding of both ends of Avd fusion proteins. Altogether, our results indicate that Avd works well in the baculovirus expression system as a multipurpose secretable tag. The Avd-tag is very stable and retains its characteristically high biotin-binding activity. It is also appropriate for use with the large range of commercially available reagents in avidin–biotin technology. The reliability with which Avd fusion proteins can be detected, purified and immobilized, is the basis for the use of our system as an elegant alternative in eukaryotic fusion protein production.
6. Membrane traffic research using avidin fusion proteins Since Avd is normally expressed only in avian and amphibian tissues and is absent from mammalian cells (for a review, see [12] or [33]), it therefore provides a practical marker for a wide range of morphological studies in several animal cell lines. We have expressed chimeric proteins containing Avd fused to for example cation-independent mannose 6-phosphate receptor (CIMPR; [34]). The constructed fusion proteins contained Avd in the N-terminus followed by the transmembrane and selected cytoplasmic sequences from the CI-MPR containing various signals for endocytosis and intracellular sorting.
K.J. Airenne et al. / Biomolecular Engineering 16 (1999) 87–92
The Semliki Forest virus vector was used to express the fusion proteins in BHK cells [35,36]. The Semliki Forest virus expression systems have recently successfully been used to express several membrane and cytoplasmic proteins in various cell lines (for a review, see [35,36]). This system is advantageous for several reasons. Semliki Forest virus contains a capped and polyadenylated single-stranded RNA genome of positive polarity which when introduced into an animal cell is effectively read as a message. After translation of its own RNA replicase, the recombinant genome is effectively replicated and transcribed. Co-transfection of a helper RNA and a recombinant RNA into BHK cells by electroporation produces infectious viruses which can be used to infect several cell lines to produce a recombinant protein. The infection is carried out in an infection medium (culture medium supplemented with 0.2% BSA without serum) for 1 h, which is long enough for efficient replication to take place. The cells are then chased in a complete medium (with serum) until the required amount of the protein is obtained. For our purposes, infection for 1 h followed by a 4 – 5-h chase produced : 1 mg of Avd fusion protein per 1 mg of total protein in BHK cells. For our studies on membrane trafficking, it is essential not to overproduce the protein, since it may cause mistargeting and disturbances in the various steps involved. However, if needed, protein expression may be extended up to at least 48 h after infection, since apparent cytopathogenic effects appear very late, in contrast to many other viral expression systems. It was found possible to label Avd chimeras in vivo during their expression with a membrane permeable dye, e.g. with biotinylated fluorescein [34]. Biotin-binding suggested that the Avd was indeed correctly folded. Double-labeling with biotinylated fluorescein and antiAvd suggested that the entire Avd pool was able to bind biotin in various intracellular locations. This is an advantage, since various functional tests can be created with biotinylated conjugates. We have developed an in vivo meeting assay which detects the formation of the Avd –biotin complex after the internalization of biotinylated horse radish peroxidase (HRP; [34]). Using different internalization time periods for biotinylated HRP, the cellular location of Avd chimeras can be studied in the endosomal pathway. The labeling of Avd with biotinylated fluorescein also make it possible to label the chimeras during expression and then follow the fate of the chimera during subsequent experimental conditions. We have also prepared green fluorescent protein (GFP)-chimeras of Avd fusion proteins which contain GFP in the C-terminus. Because GFP-fluoresescence is very stable we have been able to follow the chimeras for long time periods under a fluoresence microscope.
91
7. Drug targeting Two recent papers have described the production of recombinant Avd for affinity targeting purposes. Shin and co-workers [37] produced Avd joined to various positions within the mouse/human chimeric IgG3 heavy chain in the TWS2 cell line. The purified antibodyavidin (Ab-Avd) fusion proteins retained their specificity for antigen and bound biotin with sufficient affinity for the retention of biotinylated therapeutics in vivo. Such Ab-Avd chimeras might prove to be valuable in delivering biotinylated ligands such as nuclides, drugs, and cytotoxic agents to a specific target expressing an antigen recognized by the associated antibody. Another potential approach for targeting tumors or tissues were described by Walker and his collaborators [38]. A functional re-Avd was introduced into two human tumor cell lines (HeLa and Hep G2) and into a murine breast carcinoma cell line (16/C) using either a transient or stable gene expression, respectively. The results showed that transfected human or murine carcinoma cells could sequester and concentrate a measurable amount of biotin, suggesting that a similar approach could be attempted using the Avd gene in vivo. Indeed, the targeted expression of the Avd gene in different tissues in vivo would provide great potential for therapeutic and imaging purposes.
8. Future perspectives The established production systems provide a basis for the efficient production of the desired re-Avds as well as Avd–fusion proteins for different purposes. They should further expand the existing widely used and versatile (strept)avidin–biotin technology. Improvements in the expression systems as well as in the mutated versions of Avd (which are under construction in our laboratory) may lead to an even more efficient expression of re-Avds and avidin fusion proteins with improved and desired properties. Such proteins could well find their way to a wide range of novel applications.
Acknowledgements We wish to express our deepest appreciation to the following people for fruitful and indispensable cooperation over the past years; Dr. Edward A Bayer and Professor Meir Wilchek (Department of Biological Chemistry, The Weizmann Institute of Science, Rehovot, Israel), Dr. Christian Oker-Blom (VTT, Biotechnology and Food Research, Espoo, Finland), Professor Jean Gruenberg, University of Geneva, Geneva, Switzerland), Dr. Vesa Olkkonen (Finnish Institute of
92
K.J. Airenne et al. / Biomolecular Engineering 16 (1999) 87–92
Occupational Health, Helsinki, Finland), Professor Timo Palosuo and Mr. Jari Mikkola (National Public Health Institute, Helsinki, Finland), Dr. Harri Alenius (Finnish Institute of Occupational Health, Helsinki, Finland), Dr. Nisse Kalkkinen (Institute of Biotechnology, University of Helsinki, Helsinki, Finland), Dr. M. Siti Arija and H.Y. Yeang (Rubber Research Institute, Kuala Lumbur, Malaysia) and Kati Juuti, Anna Laukkanen, Olli Laitinen, Ari Marttila, Eeva-Liisa Punnonen, and Paula Sarkkinen from our department.
References [1] Wilchek M, Bayer EA, editors. Methods in Entzymol 184, 1990. [2] Heney G, Orr GA. Anal Biochem 1981;114:92–6. [3] Airenne KJ, Sarkkinen P, Punnonen E-L, Kulomaa MS. Gene 1994;144:75 – 80. [4] Airenne KJ, Kulomaa MS. Gene 1995;167:63–8. [5] Airenne KJ, Oker-Blom C, Marjoma¨ki VS, Bayer EA, Wilchek M, Kulomaa MS. Prot Expr Purif 1997;9:100–8. [6] Bayer EA, Meester FD, Kulik T, Wilchek M. Appl Biochem Biotech 1995;53:1 – 9. [7] Gold L. Methods Entzymol 1990;185:11–4. [8] Livnah O, Bayer EA, Wilchek M, Sussman JL. Proc Natl Acad Sci USA 1993;90:5076–80. [9] Pugliese L, Coda A, Malcovati M, Bolognesi M. J Mol Biol 1993;231:698 – 710. [10] Pugliese L, Malcovati M, Coda A, Bolognesi M. J Mol Biol 1994;235:42 – 6. [11] Hiller Y, Gershoni JM, Bayer EA, Wilchek M. Biochem J 1987;248:167 – 71. [12] Green NM. Avd Protein Chem 1975;29:85–133. [13] Goeddel DV. Methods Enzymol 1990;185:3–7. [14] Sano T, Cantor CR. Proc Natl Acad Sci USA 1990;87:142 – 6. [15] Lesley SA, Patten PA, Schultz PG. Proc Natl Acad Sci USA 1993;90:1160 – 5.
.
[16] Dakshinamurti K, Chauhan J. Methods Enzymol 1990;184:93– 102. [17] Shen S-H. Proc Natl Acad Sci USA 1984;81:4627 – 31. [18] Schulz M-F, Buell G, Schmid E, Movva R, Selzer G. J Bacteriol 1987;169:5385– 92. [19] Butt TR, Jonnalagadda S, Monia BP, Sternberg EJ, Marsh JA, Stadel JM, Ecker DJ, Crooke ST. Proc Natl Acad Sci USA 1989;86:2540 – 4. [20] Hammarberg B, Nygren P-A, , Holmgren E, Elmblad A, Tally M, Hellman U, Moks T, Uhle´n M. Proc Natl Acad Sci USA 1989;86:4367 – 71. [21] Itakura K, Hirose T, Crea R, Riggs AD, Heyneker HL, Bolivar F, Boyer HW. Science 1977;198:1056– 63. [22] O’Reilly DR, Miller LK, Luckov VA. Baculovirus Expression Vectors: A Laboratory Manual. New York: Oxford University Press, 1994. [23] Davies AH. Biotechnology 1994;12:47 – 50. [24] Jones I, Morikawa Y. Curr Opin Biotech 1996;7:512 – 6. [25] Bruch RC, White HB. Biochemistry 1982;21:5334 – 41. [26] Uhle´n M, Moks T. Methods Enzymol 1990;185:129 – 43. [27] Nilsson J, Sta˚hl S, Lundeberg J, Uhle´n M, Nygren P-A, . Prot Expr Purif 1997;11:1 – 16. [28] Pack P, Plu¨ckthun A. Biochemistry 1992;31:1579 – 84. [29] Frangioni JV, Neel BG. Anal Biochem 1993;210:179 – 87. [30] Airenne KJ, Laitinen OH, Alenius H, Mikkola J, Palosuo T, Kalkkinen N, Arija MS, Yeang HY, Kulomaa MS. Prot Expr Purif 1999;17:139 – 45. [31] Acher BL. Biochem J 1960;75:236 – 40. [32] Alenius H, Kalkkinen N, Reunala T, Turjanmaa K, Palosuo T. J Immunol 1996;156:1618– 25. [33] Korpela JK. Med Biol 1984;62:5 – 26. [34] Marjoma¨ki V, Juuti K, Airenne KJ, Laukkanen A, Punnonen E-L, Olkkonen VM, Gruenberg J, Kulomaa M. Resubmitted, 1999. [35] Liljestro¨m P, Garoff H. Biotechnology 1991;9:1356 – 61. [36] Olkkonen VM, Liljestro¨m P, Dupree P, Garoff H, Simons K. Methods Cell Biol 1994;43:43 – 53. [37] Shin S-U, Wu D, Ramanathan R, Pardridge WM, Morrison SL. J Immunol 1997;158:4797– 804. [38] Walker L, Kulomaa MS, Bebok Z, Parker WB, Allan P, Logan J, Huang Z, Reynolds RC, King S, Sorscher EJ. J Drug Targ 1996;4:41 – 9.