Balanced expression of single subunits in a multisubunit protein, achieved by cell fusion of individual transfectants Lars Norderhaug 1 , Finn-Eirik Johansen 2 and Inger Sandlie 3 1 Antibody Design AS, Nesoddtangen, Norway; 2 Department of Pathology, Rikshospitalet, Norway; 3 Department of Biology, University of Oslo, Norway To establish stable cell lines that produce recombinant multisubunit proteins, it is usually necessary to cotransfect cells with several independent gene constructs. Here, we show that a stepwise fusion of individually transfected cells, results in a fused cell-line that secretes a complete multi- subunit protein. Functional expression of recombinant multisubunit proteins may require a defined expression ratio between each protein subunit. The cell-fusion technology described allows a predefined expression level of each sub- unit. Using SIgA as a model protein we demonstrate that the majority of the fused cells inherit the molar expression ratio of the parental transfected cells. These results indicate that simplified screening of clones expressing the expected sub- unit ratios may be possible using the cell-fusion technology. This technology may therefore be an alternative to generic transfection methods for the establishment of cells that produce multiprotein complexes such as antibodies, recep- tors, ion channels and other multisubunit proteins. Keywords: antibody; expression; fusion; SIgA; transfection. To establish stable cell lines that produce recombinant multisubunit proteins, it is generally necessary to introduce vectors that direct the expression of each subunit gene. For multisubunit proteins such as secretory antibodies [1,2], receptors [3,4] and ion channels [5–8], a defined expression of each unit may be essential for the specific function of the mature protein product. This is usually achieved by cotrans- fecting cells with several independent constructs or by introducing a single vector harbouring several discrete expression cassettes. Controlled expressions of several poly- peptides within one cell have also been addressed using different vector designs such as; multicistronic vectors where each gene are controlled by internal ribosome entry site (IRES) [9–11], panel of vectors [12] and multiple episomal vectors [13]. We chose secretory IgA (SIgA) as a model protein to investigate the potential of cell-fusion technology for the expression of multisubunit proteins. SIgA is the major immunoglobulin of external secretions and is composed of four different polypeptide chains produced by two distinct cells. Polymeric IgA (pIgA) is produced by B-cells after assembly of light-, a- and joining (J)-chain [14]. SIgA is subsequently generated when a secretory component (SC) is covalently added to pIgA during transport across the epithelial cells lining mucosal surfaces [15]. Expression of the multisubunit SIgA has previously been described using other transfection methods [2,16–18]. Although the principle of cell-fusion is well established by hybridoma technology [19], the development and application of this technology for production of recombinant multisubunit proteins has not been described previously. Here, we achieve multigene expression by utilizing cell-fusion of individually transfected cells, each expressing one or more genes that encode the multisubunit protein. We show that the majority of clones resulting from the fusion inherit the expression levels of the parental cells, thus simplifying screening for clones with stochiometric expression levels of each component that secrete fully functional SIgA. Furthermore, we show that this system enables high-level expression in mammalian cells, which is often a goal in recombinant protein expression. MATERIAL AND METHODS Vectors and cloning of genes Construction of the vector family pLNO and its use for transfection and expression of immunoglobulin genes has been described previously [20–22]. The human Ig a1 gene was subcloned into pLNO/Neo giving the vector pLNOA1/Neo. The Ig heavy-chain variable-gene (VH) SS-269VH [23], specific for the outer membrane protein of the bacteria Neisseria meningitides, was subcloned into pLNOA1/Neo giving pLNOA1/Neo-SS269. The human Ig k gene was subcloned into pLNO/Neo, giving pLNOL/ Neo. The Ig light-chain variable-gene (VL) SS-269VL [23], specific for the outer membrane protein of the bacteria N. meningitides, was subcloned into pLNOL/Neo giving pLNOL/Neo-SS269. The construction of the human J-chain vector pCH (CMV-driven expression, hygromy- cin B resistance) [2] and the human SC vector pcDNA(zeo)- His 6 (CMV-driven expression, Zeocin resistance) has been previously described [2]. Cell culture and transfection CHO-K1 cells were obtained from ATCC (USA) and cultured in HAM F-12 (F-12 derived from hamster) with 1m ML -glutamine supplemented with 10% fetal bovine Correspondence to L. Norderhaug, Antibody Design AS, P.O.Box190, N-1450 Nesoddtangen, Norway. Fax: + 47 66960691, Tel.: + 47 66960690, E-mail: lars.norderhaug@antibodydesign.com Abbreviations: IRES, internal ribosome entry site; SIgA, secretory IgA; pIgA, polymeric IgA; J-chain, joining chain; SC, secretory component; FITC, fluorescein isothiocyanate; HRP, horseradish peroxidase. (Received 14 December 2001, revised 17 April 2002, accepted 14 May 2002) Eur. J. Biochem. 269, 3205–3210 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03000.x serum. All transfections were performed by electroporation with a BTX ECM-600 electroporator (San Diego, CA, USA) with the settings 200 V (500 VÆcm )1 ), 800 lFand 185 Ohm. All transfections were performed in 0.4 mm cuvettes with 20 lg of plasmid DNA and 1 · 10 7 cells per mL in 0.8 mL NaCl/P i at 0 °C giving a 12-ms pulse. Following transfection, the cells were subsequently trans- ferred into 25 mL Dulbecco’s modified Eagle’s medium (DMEM) or HAM F-12 with 10% fetal bovine serum in 25 cm 2 flasks,andallowedtorecoverfor24hbefore addition of appropriate antibiotics; 800 lgÆmL )1 G418 (Invitrogen BV, the Netherlands) or 400 lgÆmL )1 hygromy- cin (Invitrogen BV, the Netherlands) or 400 lgÆmL )1 Zeocin (Invitrogen BV, the Netherlands). Three individual transfec- tions were employed: (a) the vector pCH/J-chain/Hygro (J- chain vector) (b) the vector pcDNA (zeo)his6/SC (SC vector) and (c) cotransfection of the vectors pLNOL/Neo-SS269 (k vector) and pLNOA1/Neo-SS269 (a1vector)inCHO-K1 cells. Cells were allowed to grow for 10 days before protein expression was analysed. Cell fusion Cell fusion was performed in two individual steps (Fig. 1) by mixing equal number of cells (3 · 10 7 cells) of each fusion partner. Cells were centrifuged (5–10 min, 200–400 g) and washed once in serum free medium. The cell mixture was further spun down, and the supernatant was removed completely. The pellet was broken by gently tapping on the bottom of the tube and placed in a 37 °C water-bath. Prewarmed 0.5 mL of 50% poly(ethylene glycol) 1500 (Boehringer Mannheim, Germany) was mixed with the cells over a period of 1 min, under continuous stirring with a pipette tip. The cell/poly(ethylene glycol) solution was stirred for another 1–2 min before further addition of 1 mL prewarmed medium under continuous stirring for 1 min. An additional 3 mL of prewarmed medium was added the same way. The fusion mixture was then slowly mixed with 10 mL of prewarmed medium and incubated for 5 min. The cells were centrifuged, the supernatant discarded and growth medium added to achieve a cell concentration of 5 · 10 5 cellsÆmL )1 . The cells were allowed to recover for 24 h, and then diluted 1 : 100 in 100 mL growth medium supplemen- ted with antibiotics to select for both parental clones. The cells were cultured with fresh medium every 3–5 days until colonies appeared 10–14 days after the fusion. Clones were selected and analysed as described below. Detection of IgA, pIgA, and SIgA expression and bound SC Transfected and fused CHO-K1 cells were analysed for production of IgA, pIgA or SIgA and IgA bound SC by ELISA on supernatant of outgrown cultures. Triplets of 100 lL supernatant of individually fused clones were transferred in dilutions 1 : 1, 1 : 5, 1 : 25 to microtiter plates coated with 4 lgÆmL )1 of N. meningitides OMV (a gift from T. E. Michaelsen, National Institute of Public Health, Norway). Secondary antibodies used for detection were rabbit anti-(human IgA) Ig (DAKO; 1 : 5000 dilution) and rabbit anti-SC Ig (DAKO; 1 : 3000 dilution) and tertiary antibody used for detection was horseradish peroxidase (HRP)-conjugated goat anti-(rabbit IgG) Ig (DAKO; 1 : 3000 dilution). The absorbance was read by TitertekÒ Multiskan (ICN Flow, USA). The amount of IgA present in each supernatant was calculated relative to a standard preparation with known concentration. Verification of J-chain expression To examine production of J-chain, transfected cells were screened by immunofluorescence staining. CHO-K1 cells transfected with J-chain were cultured on micro slides. Cells were fixed and permeabilized in methanol ()20 °Cfor 4 min). The cells were then washed twice with NaCl/P i and incubated at room temperature for 20 min with 1 : 3000 diluted rabbit anti-(J-chain) Ig (P. Brandtzaeg, LIIPAT, National Hospital, Norway). Cells were then washed three times with NaCl/P i and incubated for 20 min with (1 : 200) fluorescein isothiocyanate (FITC)-conjugated goat anti- (rabbit IgG) Ig (DAKO). Cells were washed as above and analysed by fluorescence microscopy. Verification of SC production Transfected and fused CHO-K1 cells were analysed for production of total SC, by a dot blot approach. Triplets of Fig. 1. A schematic diagram of the fusion of cells producing individual protein units of a multisubunit protein. Three transfected CHO-K1 cell lines producing either IgA, J-chain or SC, respectively, were fused in two steps; fusion I and fusion II. Fusion I between cells producing IgA and J-chain resulted in clones producing pIgA. Fusion II between cells producing pIgA and SC resulted in clones producing SIgA. 3206 L. Norderhaug et al. (Eur. J. Biochem. 269) Ó FEBS 2002 supernatants from individual clones were applied in 1 : 1, 1 : 5, 1 : 25, 1 : 125 dilutions onto poly(vinylidene difluo- ride) paper (Millipore; Sweden), preincubated for 3–5 s in methanol, followed by 5 min incubation in dH 2 Oand subsequently 10 min incubation in blotting buffer. The membranes were then dried at 37 °C for 1 h, before incubation in 25 mL NaCl/P i /0.1% Tween-20 (NaCl/P i / Tween) with 10% skimmed milk and 1 : 1000 diluted rabbit anti-SC Ig. The membranes were then washed three times in NaCl/P i /Tween and incubated with HRP-conjugated goat anti-(rabbit IgG) Ig (Bio-Rad; 1 : 3000 dilution) for another 1 h. Finally, the membranes were washed three times in NaCl/P i /Tween before addition of substrate (Bio-Rad) for 5 min. The membranes were covered with plastic film and exposed to Kodak X-OMAT film for 15–60 s. Dot blot density was analysed by TOTALLAB gel software (Phonetix, UK). The amount of SC present in each supernatant was calculated relative to a standard preparation with known concentration. Western blot of IgA, pIgA and SIgA Aliquots of 10 lL supernatant from selected clones were analysed under nonreducing conditions on a 4–15% Tris/ HCl SDS/PAGE ReadyGel (Bio-Rad) run at 200 V for 1 h. The gel was blotted onto PVDF paper (Millipore; Sweden) in a Bio-Rad Miniblotter for 1 h at 100 V. Following the transfer, the membranes were washed in NaCl/P i for 5–10 min with gentle agitation, and blocked for 45 min in NaCl/P i /Tween with 10% skimmed milk. The membrane was washed once in NaCl/P i before incubation with either 1 : 5000 dilution rabbit anti-(human IgA) or 1 : 3000 dilution rabbit anti-(human SC) Ig for 1 h. The membranes were washed twice followed by incubation with 1 : 3000 dilution HRP-conjugated goat anti-(rabbit IgG) Ig (Immun-Star TM Chemiluminescent Protein Detection Systems, Bio-Rad) for 1 h. The blot was developed as described above and exposed to X-ray film for 1–10 min. RESULTS To obtain SIgA-producing cells we first established cells that stably expressed the individual protein subunits by conven- tional transfection. Transfection of the SC gene, the J-chain gene and cotransfection of the a1 heavy-chain and k light- chain genes resulted in clones that produced SC, J-chain or IgA, respectively. Selected clones were then subjected to two fusions: the first between IgA and J-chain producing cells, and the second between pIgA-producing cells from the first fusion and SC-producing cells (Fig. 1). Generation of clones expressing IgA, J-chain or SC To establish cells expressing IgA, CHO-K1 cells were cotransfected with a1andk genes using the two vectors pLNOA1/Neo-SS269 and pLNOL/Neo-SS269. Superna- tants from 20 clones were analysed by ELISA and seven of these were positive for IgA production. One of these clones, IgA-29, was chosen for further expansion and fusion with J-chain producing cells. The J-chain gene was transfected into CHO-K1 cells using the vector pCH/J-chain/Hygro. Because J-chain is retained within the endoplasmic reticu- lum and only secreted when joined to IgA or IgM, cells were screened for J-chain-expression by immunofluorescence. Cells from 6 clones were fixed and stained with an anti- (human J-chain) Ig to verify the presence of intracellular J-chain. A further attempt to directly quantify the amount of intracellular J-chain was avoided, as retention is closely linked to degradation [24,25]. One clone, J-1, with high fluorescence intensity was selected for further expansion and fusion to IgA-29. The SC gene was transfected into CHO- K1 cells on the vector pcDNA(zeo)his6/SC. Twenty-four clones were analysed by dot blot as described, and six of these were positive for SC production. One clone, SC-4, was expanded for further fusion. The clones IgA-29 (4.5 lgÆmL )1 ) and SC-4 (2.3 lgÆmL )1 ) were chosen for cell fusions because the amount of expressed protein on a molar basis is almost equal in these cells, as the M r of IgA and SC are 160 and 80 kDa, respectively. Fusion of single transfectants to achieve SIgA-producing clones The first fusion of IgA-producing cells (IgA-29) and J chain- producing cells (J-1) resulted in numerous G418 and hygromycin B resistant colonies. The overall fusion effi- ciency was as high as 1 · 10 )3 . Five colonies were analysed for production of IgA and J-chain by ELISA. All five colonies were shown to produce both polypeptide chains. One clone (pIgA-D) producing polymeric IgA was expan- ded for fusion with SC-producing cells (SC-4). This fusion was as effective as the first and resulted in numerous G418, hygromycin B and Zeocin resistant colonies. Expanded clones from the second fusion expressed all four genes: k-chain and a1-chain, J-chain and SC. Expression levels and SIgA quality To investigate the expression ratio between the introduced protein components of the fused cell, an IgA and SC expression level analysis was done. Whereas SC is readily secreted without assembly with the other components, this is not the case for a1 heavy-chain and J-chain. Both are retained and degraded intracellularly unless in complex with their appropriate partners. Uncomplexed k light-chain shows some retention, but is also secreted as light-chain dimers [26]. The parental cell IgA-29, the pIgA clone pIgA- D, and eight SIgA clones (SIgA-1–8) were analysed by ELISA for expression levels of IgA and total SC (Fig. 2A) and by dot blot for total SC (Fig. 2A). The IgA expression levels varied from 0.5 to 5.2 lgÆmL )1 and total SC expression levels varied from 0.5 to 2.3 lgÆmL )1 (Fig. 2A). Importantly, more than 50% of the fused clones showed expression levels almost equal to the parental cells which was 4.5 lgÆmL )1 for IgA and 2.3 lgÆmL )1 for SC (Fig. 2A). The molar expression ratio between IgA and total SC was calculated for each fused clone, and compared with the molar expression ratio of the parental cells SC-4/pIgA-D. The molar ratio varied from 0.9 to 2.2, while 50% of the clones maintained the molar ratio of  1 (Fig. 2B). This shows that the selection and isolation of clones expressing a stochiometric or predefined ratio of different protein subunits is well within reach of a simple screening proce- dure. SC bound to IgA also correlated with IgA expression levels in all fused clones shown by ELISA (data not shown). Because SC only interacts with J-chain-positive pIgA, the Ó FEBS 2002 Multi-subunit protein production by cell fusion (Eur. J. Biochem. 269) 3207 complexing of IgA and J-chain is a prerequisite for SC binding. Therefore, the correlation between the IgA and SC levels in all the clones demonstrated that a sufficient amount of J-chain was available for SIgA complex formation. One SIgA-producing clone (SIgA-3) along with pIgA-D and IgA-29 were analysed by SDS/PAGE gel and Western blot (Fig. 3) to characterize the molecular size and composition of the secreted products. This Western blot clearly demon- strated that cells fused to produce all the four polypeptides of SIgA, assembled and secreted SIgA of the expected molecular size with reactivity against antibodies towards both IgA and SC. The fused cells expressed both monomeric and polymeric IgA as described for hybridoma cells expressing monoclonal IgA [27], and was also seen when the J-chain gene was transfected into an IgA-producing CHO cell line [2]. The production level of the fused cells grown in culture, without any selective pressure, was measured by ELISA for two months. During this time period, no change in expression levels was observed. The fused cells were also frozen and thawed without any change in growth or expression rate. In conclusion the fused cells behaved as their transfected parental cells with respect to morphology, growth rate and expression rate. DISCUSSION Stable transfection of mammalian cells requires that the gene of interest is integrated into the chromosome by a nonhomologous recombination event [28,29]. This is a rare event and results in a very low frequency of integration and variable expression levels due to the so-called Ôposition effectÕ [30] and the copy number of integrated genes [31–33]. It has been observed that the variation in expression levels among individual clones varies with > 30-fold in a single gene transfection assay [11,34–36] (L. Norderhaug & I. Sandlie, unpublished data). Thus, to establish a single cell line expressing defined ratios of multiple protein units, excessive screening is necessary, as each gene introduced into the cell multiplies the complexity of the screening. Others have addressed the problem of stochiometric expression by construction of special vectors with IRES elements that guide bicistronic or multicistronic expression [9–11,37]. However, the use of IRES elements may require complex plasmid construction or be limited to the use of short cDNAs due to the size of the complete gene. Stochiometric expressions by multiple episomal vectors [13,38–40] have been addressed. The advantage of episomal vectors is the property of the vector to replicate extra- chromosomally, and thereby eliminate the positional effect [30] of chromosomal integrating vectors. However, for stable expression of multiple genes over time, the usefulness of such episomal vector is limited, mainly because of the slow loss of the vector over time when unselected [39]. A defined expression ratio between each protein subunit may be essential for the specific function of the mature protein product. Study of ion channels [8] and receptors [3,41] have shown that their structure and functions actually depend on the level of expression of the different subunits. Thus, cell- fusion technology will allow generation of cells with a Fig. 3. Western blot of assembled IgA, pIgA and SIgA. SDS/PAGE and Western blot of untransfected CHO-K1 cells (lane 1), IgA-29 (lane 2), pIgA-D (lane 3 and 5) and SIgA-3 (lane 4 and 6) detected by antihuman IgA or strippedandredetectedwithantihumanSC,as indicated. The blot shows assembly of all protein subunits in the fused cells. Fig. 2. IgA and total SC production levels (A) and molar expression ratios (B). (A) IgA and SC expression levels of outgrown supernatants of the clones: SC-4, IgA-29, pIgA-D and SIgA-1–8 measured in triplets by ELISA or dot blot. (B) Calculation of the molar expression ratio between the subcomponents SC and IgA in each fusion clone SIgA1-8 and also between their parental cells SC-4 and IgA-29. Calculations were based on measured expression levels of both SC and IgA (A) and the M r of SC (80 kDa) and IgA (160 kDa). 3208 L. Norderhaug et al. (Eur. J. Biochem. 269) Ó FEBS 2002 predefined expression level of each protein subunit and hence predefined molar ratios. The result presented here show that the fusion event typically does not alter the expression level of the recombinant protein (Fig. 2B). Furthermore, in the study of structure–function relation- ships of protein complexes, individual components may be altered by site directed mutagenesis. A cell line expressing the altered gene product at a given level may then be fused to constitute an altered complex. Cell-fusion is possible using CHO-K1, one of the cell-lines most widely used for recombinant protein production, and the technology is also applicable to NS0 cells (L. Norderhaug & I. Sandlie, unpublished data). ACKNOWLEDGEMENTS This research was supported by Oslo Research Park AS and The Research Council of Norway. REFERENCES 1. Chintalacharuvu, K.R. & Morrison, S.L. (1997) Production of secretory immunoglobulin A by a single mammalian cell. Proc. Natl Acad. Sci. USA 94, 6364–6368. 2. Johansen, F., Norderhaug, I., Roe, M., Sandlie, I. & Brandtzaeg, P. (1999) Recombinant expression of polymeric IgA: incorpora- tion of J chain and secretory component of human origin. Eur. J. Immunol. 29, 1701–1708. 3. Tretter, V., Hauer, B., Nusser, Z., Mihalek, R., Hoger, H., Homanics, G., Somogyi, P. & Sieghart, W. (2001) Targeted disruption of the GABA (subA) receptor (delta) subunit gene leads to an upregulation of (gamma) (sub2) subunit-containing receptors in cerebellar granule cells. J. Biol. Chem. 276, 10532– 10538. 4. Sieghart, W., Fuchs, K., Tretter, V., Ebert, V., Jechlinger, M., Hoger, H. & Adamiker, D. (1999) Structure and subunit com- position of GABA (A) receptors. Neurochem. Int. 34, 379–385. 5. Xu,J.,Yu,W.,Wright,J.,Raab,R.&Li,M.(1998)Distinct functional stoichiometry of potassium channel beta subunits. Proc. Natl Acad. Sci. USA 95, 1846–1851. 6. Tucker, S., Pessia, M. & Adelman, J. (1996) Muscarine-gated K+ channel: subunit stoichiometry and structural domains essential for G protein stimulation. Am. J Physiol. 271, H379–H385. 7. Inagaki, N., Gonoi, T. & Seino, S. (1997) Subunit stoichiometry of thepancreaticbeta-cellATP-sensitiveK+channel.FEBS Lett. 409, 232–236. 8. Manganas, L. & Trimmer, J. (2000) Subunit composition determines Kv1 potassium channel surface expression. J. Biol. Chem. 275, 29685–29693. 9. Zhu, J., Musco, M. & Grace, M. (1999) Three-color flow cyto- metry analysis of tricistronic expression of eBFP, eGFP, and eYFP using EMCV-IRES linkages. Cytometry 37, 51–59. 10. Gurtu, V., Yan, G. & Zhang, G. (1996) IRES bicistronic expres- sion vectors for efficient creation of stable mammalian cell lines. Biochem. Biophys. Res. Commun. 229, 295–298. 11. Liu, X., Constantinescu, S., Sun, Y., Bogan, J., Hirsch, D., Weinberg, R. & Lodish, H. (2000) Generation of mammalian cells stably expressing multiple genes at predetermined levels. Anal. Biochem. 280, 20–28. 12. Cockett, M., Ochalski, R., Benwell, K., Franco, R. & Wardwell- Swanson, J. (1997) Simultaneous expression of multi-subunit proteins in mammalian cells using a convenient set of mammalian cell expression vectors. Biotechniques 23, 406–407. 13. Horlick, R., Schilling, A., Samama, P., Swanson, R., Fitzpatrick, V., Robbins, A. & Damaj, B. (2000) Combinatorial gene expres- sion using multiple episomal vectors. Gene 243, 187–194. 14. Johansen, F., Braathen, R. & Brandtzaeg, P. (2000) Role of J chain in secretory immunoglobulin formation. Scand. J. Immunol. 52, 240–248. 15. Norderhaug, I., Johansen, F., Schjerven, H. & Brandtzaeg, P. (1999) Regulation of the formation and external transport of secretory immunoglobulins. Crit.Rev.Immunol.19, 481–508. 16. Chintalacharuvu, K.R. & Morrison, S.L. (1999) Production and characterization of recombinant IgA. Immunotechnology 4, 165– 174. 17. Frigerio,L.,Vine,N.,Pedrazzini,E.,Hein,M.,Wang,F.,Ma,J. & Vitale, A. (2000) Assembly, secretion, and vacuolar delivery of a hybrid immunoglobulin in plants. Plant Physiol. 123, 1483–1494. 18. Ma, J., Hiatt, A., Hein, M., Vine, N., Wang, F., van Stabila, P.D.C., Mostov, K. & Lehner, T. (1995) Generation and assembly of secretory antibodies in plants. Science 268, 716–719. 19. Kohler, G. & Milstein, C. (1975) Continuous cultures of fused cells secreting antibody of predefined specificity. Nature 256, 495–497. 20. Sorensen, V., Rasmussen, I., Norderhaug, L., Natvig, I., Michaelsen, T. & Sandlie, I. (1996) Effect of the IgM and IgA secretory tailpieces on polymerization and secretion of IgM and IgG. J. Immunol. 156, 2858–2865. 21. Norderhaug, L., Olafsen, T., Michaelsen, T.E. & Sandlie, I. (1997) Versatile vectors for transient and stable expression of recombinant antibody molecules in mammalian cells. J. Immunol. Methods 204, 77–87. 22. Olafsen, T., Rasmussen, I.B., Norderhaug, L., Bruland, O.S. & Sandlie, I. (1998) IgM secretory tailpiece drives multimerisation of bivalent scFv fragments in eukaryotic cells. Immunotechnology 4, 141–153. 23. Wang, J., Jarvis, G.A., Achtman, M., Rosenqvist, E., Michaelsen, T.E., Aase, A. & Griffiss, J.M. (2000) Functional activities and immunoglobulin variable regions of human and. Infect. Immun. 68, 1871–1878. 24. Klausner, R. & Sitia, R. (1990) Protein degradation in the endoplasmic reticulum. Cell 62, 611–614. 25. Mancini, R., Fagioli, C., Fra, A., Maggioni, C. & Sitia, R. (2000) Degradation of unassembled soluble Ig subunits by cytosolic proteasomes: evidence that retrotranslocation and degradation are coupled events. FASEB J. 14, 769–778. 26. Reddy, P., Sparvoli, A., Fagioli, C., Fassina, G. & Sitia, R. (1996) Formation of reversible disulfide bonds with the protein matrix of the endoplasmic reticulum correlates with the retention of unassembled Ig light chains. EMBO J. 15, 2077–2085. 27. Stoll, T., Chappaz, A., von Stockar, U. & Marison, I. (1997) Effects of culture conditions on the production and quality of monoclonal IgA. In Enzyme and Microbial Technology,pp. 203–211. 28. Roth, D.B., Porter, T.N. & Wilson, J.H. (1985) Mechanisms of nonhomologous recombination in mammalian cells. Mol. Cell. Biol. 5, 2599–2607. 29. Roth, D.B. & Wilson, J.H. (1986) Nonhomologous recombina- tion in mammalian cells: role for short sequence homologies in the joining reaction. Mol. Cell. Biol. 6, 4295–4304. 30. Wilson, C., Bellen, H. & Gehring, W. (1990) Position effects on eukaryotic gene expression. Annu.Rev.CellBiol.6, 679–714. 31. Zastrow, G., Koehler, U., Muller, F., Klavinius, A., Wegner, M., Wienberg, J., Weidle, U.H. & Grummt, F. (1989) Distinct mouse DNA sequences enable establishment and persistence of plasmid DNA polymers in mouse cells. Nucleic Acids Res. 17, 1867–1879. 32. Pulm, W. & Knippers, R. (1985) Transfection of mouse fibroblast cells with a promoterless herpes simplex virus thymidine kinase gene: number of integrated gene copies and structure of single and amplified gene sequences. Mol. Cell. Biol. 5, 295–304. 33. Hemann, C., Gartner, E., Weidle, U.H. & Grummt, F. (1994) High-copy expression vector based on amplification-promoting sequences. DNA Cell Biol. 13, 437–445. Ó FEBS 2002 Multi-subunit protein production by cell fusion (Eur. J. Biochem. 269) 3209 34. Bebbington, C.R., Renner, G., Thomson, S., King, D., Abrams, D. & Yarranton, G.T. (1992) High-level expression of a recombinant antibody from myeloma cells using a glutamine synthetase gene as an amplifiable selectable marker Biotechnology 10, 169–175. 35. Hendricks, M.B., Banker, M.J. & McLaughlin, M. (1988) A high- efficiency vector for expression of foreign genes in myeloma cells. Gene 64, 43–51. 36. Wurm, F.M., Johnson, A., Ryll, T., Ko ¨ hne, C., Scherthan, H., Glaab, F., Lie, Y.S., Petropoulos, C.J. & Arathoon, W.R. (1996) Gene transfer and amplification in CHO cells. Efficient methods for maximizing specific productivity and assessment of genetic consequences. Ann. NY Acad. Sci. 782, 70–78. 37. Martinez-Salas, E. (1999) Internal ribosome entry site biology and its use in expression vectors. Curr. Opin. Biotechnol. 10, 458–464. 38. Van, C.K., Vanhoenacker, P., Duchau, H. & Haegeman, G. (2000) Molecular integrity and usefulness of episomal expression vectors derived from BK and Epstein–Barr virus. Gene 253, 293–301. 39. Vanhoenacker, P. & Haegeman, G. (2000) Episomal vectors for gene expression in mammalian cells. Eur. J. Biochem. 267, 5665– 5678. 40. Mizuguchi, H., Hosono, T. & Hayakawa, T. (2000) Long-term replication of Epstein–Barr virus-derived episomal vectors in the rodent cells. FEBS Lett. 472, 173–178. 41. Eertmoed, A., Vallejo, Y. & Green, W. (1998) Transient expression of heteromeric ion channels. Methods Enzymol. 293, 564–585. 3210 L. Norderhaug et al. (Eur. J. Biochem. 269) Ó FEBS 2002 . Balanced expression of single subunits in a multisubunit protein, achieved by cell fusion of individual transfectants Lars Norderhaug 1 , Finn-Eirik. blot density was analysed by TOTALLAB gel software (Phonetix, UK). The amount of SC present in each supernatant was calculated relative to a standard preparation