Cloning and expression of human V-genes derived from phage display libraries as fully assembled human anti-TNFα monoclonal antibodies

ELSEVIER

Immunotechnology

3 (1997) 3 l-43

Cloning and expression of human V-genes derived from phage display libraries as fully assembled human anti-TNFa monoclonal antibodies Stephen M. Mahler a,*, Chris P. Marquis a, Gary Brown a, Andy Roberts b, Hennie R. Hoogenboom ’ a Department of Biotechnology, University of New South Wales, Kensington 2052, Sydney, Australia b Cambridge Antibody Technology, The Science Park, Melbourn, Cambridgeshire SG8 6EJ, UK ’ CESAME, Department of Pathologie, Academisch Ziekenhuis, 6202 AZ Maastricht, Netherlands

Received 1 July 1996; received in revised form 19 September 1996; accepted 26 September 1996

Abstract Background: With the advent of phage antibody libraries, access to completely human antibody fragments is feasible, either by direct selection from human antibody libraries, or by guided selection. After selection, Fabs and scFvs may need to be expressed as complete antibodies in mammalian cells for further characterisation, or if effector functions are required. Objectives: To rebuild and express the human anti-TNFu antibody Fab-P3A2 (isolated as a Fab fragment from phage display libraries by guided selection) as a fully assembled, functional human antibody (y-l, A) in Sp2jO myeloma cells, and to perform preliminary characterisation studies of the secreted IgGl molecule. A further objective was to investigate the kinetics of human antibody production and the stability of antibody secretion in transfectomas cultured in various media formulations. Study design: A tripartite strategy was employed for cloning heavy chain gene (V,)-P3 and light chain gene VI-A2-CI into mammalian cell expression vectors pccLys-30 and paLys-17 respectively. The cell line P3A2.B5 was isolated after co-transfection of Sp2/0 mouse myelomas with the constructs, expanded and weaned into a protein free medium. Fully assembled Ig-P3A2 antibody was purified by Protein A affinity chromatography and characterised with respect to size of antibody chains, and affinity for human TNFa. Stability of secretion was investigated by extended serial sub-culture and analysis of P3A2.B5 sub-clones. Strategies of media enrichment were tested for any effect on antibody productivity by selected P3A2.B5 sub-clones. Results: The cell line P3A2.B5 secreted an assembled, human antibody Ig-P3A2, with heavy and light chains of molecular weight 55 and 28 KD respectively. Equilibrium capture studies showed Ig-P3A2 to have a dissociation Abbreviations: ABTS, [2,2’azino-bis-(3-ethyl-benzthiazoline-6 sulphonate)]; BSA, bovine serum albumin; B + F, bound and free; CDR, complementarity determining regions; FCS, fetal calf serum; MAb, monoclonal antibody; MCB, master cell bank; PBS, phosphate-buffered saline; scFv, single-chain Fv; SDS-PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis; TNF, tumor necrosis factor; V, variable region; V,, immunoglobulin heavy chain variable region; V,, immunoglobulin light chain variable region; WCB, working cell bank. * Corresponding author. Tel.: + 61 2 3853899; fax: + 61 2 3136710; e-mail: [email protected]

1380-2933/97/$17.00 0 1997 Elsevier Science B.V. All rights reserved. PIISl380-2933(96)00058-9

32

S.M.

Muhler

et al. / Immunotechnolog~

3 (1997)

31-43

constant of approximately 1.5 x 10 ~ ’ M. The mean specific productivity of the cell line increased from 1.2 pg/cell/day to 7.8 pg/ceil/day by a combination of medium enrichment and serum reduction. Prolonged serial sub-culture of P3A2.BS showed the cell line to be unstable with respect to antibody secretion. Conclusiotzs: We have outlined a method for expression of human V genes as assembled antibodies in Sp2/0 myeloma cells. A cloning strategy for the stable expression of scFv or Fab genes isolated from phage display libraries as assembled human antibodies of the IgGl subclass in Sp2jO myeloma cells has been described. For maximising specific productivity of antibody-producing cell lines, supplementation of culture media with glucose, glutamine and amino acids increases antibody yield significantly compared to that in conventional media, indicating the latter is stoichiometrically limiting for production purposes. 0 1997 Elsevier Science B.V. Keywords: Antibody construction; tion

Characterisation;

Expression; Human antibody genes; Mammalian

1. Introduction To date there are few mouse monoclonal antibodies (MAbs) which have been approved for clinical use in humans, including the anti-CD3 OKT3 MAb for prevention of kidney [I] and liver [2] transplant rejection, and the IgG2a/lc 17-1A Mab for the treatment of minimal residual disease in colorectal cancer [3]. This is mainly due to problems associated with the clinical use of mouse MAbs such as the poor recognition of constant regions for the eliciting of effector functions, and more importantly the production of human antimouse antibodies (HAMA), the consequence of which are a shortened in vivo half life, and diversion of MAbs to Fc receptors sites other than those on malignant cells. In the quest to make rodent antibodies more human, the fusion of mouse variable region (V) genes with human constant region genes has allowed the production of chimeric antibodies [4]. Although the plasma half lives of chimeric antibodies are higher than those of murine antibodies, and HAMA response is greatly reduced [5], most chimeric antibodies exhibit significant anti-idiotypic response mainly due to the mouse V regions, rendering repeated dosing ineffective (for review, see [6]). Reshaping mouse antibodies by complementarity determining region (CDR) grafting, whereby mouse antigen binding loops are transplanted onto the p-sheet framework regions of human antibodies, has reduced the rodent component of MAbs even further [7,8], and indeed has led to antibodies with diminished immunogenic-

cells: Produc-

ity. Fully human antibodies, where the CDR regions are also derived from genuine human sequences, are expected to have the lowest possible immunogenicity when used in human therapy. A combination of the cloning of the human immunoglobulin repertoire [9] and the ability to display antibody fragments on the surface of filamentous phage [l&l l] has allowed the isolation of human antibodies of therapeutic significance ([ 121, for review see [ 131). Recently a new technique termed guided selection has been reported [ 141, and is based on human immunoglobulin V gene shuffling with rodent immunoglobulin V genes. Using this technique, a fully human Fab fragment (Fab-P3A2) was isolated by pairing of the V genes of a mouse MAb reactive with human tumor necrosis factor (TNF)a (MAb32) [15] with repertoires of human V genes followed by selection on TNFa [14]. Fab-P3A2 was shown to bind to the same epitope as that of Mab32. For further characterisation of this antibody, it was necessary to rebuild the Fab-P3A2 as a complete y-l human antibody (Ig-P3A2), which has advantages, for example, of an increased serum half life, and increased avidity compared to those of the Fab fragment. Furthermore, we wished to compare antibodies isolated from Escherichiu coli repertoires and expressed in E. coli with the same antibodies expressed in mammalian cells. Expression of such genes in the former may be associated with incorrect or incomplete folding of the antibody fragment, while expression in the latter may be hampered by cryptic splice and O-linked glycosylation sites.

S.M. Mahler et al. / Immunotechnology 3 (1997) 31-43

Recloning and expression of human Fab-P3A2 as a fully assembled human antibody in Sp2/0 myeloma cells, subsequent characterisation of Protein A-purified Ig-P3A2, and stability and productivity studies of the cell lines secreting Ig-P3A2 are described in this report.

2. Materials and methods 2.1. Isolation of V-genes by guided selection The isolation of Fab-P3A2 by guided selection has previously been described [14]. Essentially a repertoire of human 2 light chains was shuffled with the cloned mouse MAb32 immunoglobulin heavy chain variable region (V,) domain. After selection on TNFa, three half human Fabs were identified with light chains designated A2, Dl and C4. The selected 1 genes were then used separately as ‘docking domains’ for a library of human heavy chains in a second shuffle, yielding amongst others, clone Fab-P3A2 with Vu-P3 heavy and V,-A2-C, light chain genes. 2.2. Oligonucleotides used for the amplljication of human V genes by PCR 1. GAG GTG CAG CTG CAG CAG GGG GG 2. AGC AGC GGA TCCAGG ACT CAC AGGAGACGGTGACCGT 3. GCC CGT GAC CAA GCT TAT GAA GCA AA 4. CAG ACT GGG AGT GCA CAC CTG AGA GAA AG 5. CTC CAC AGG TGT GCA CTC CCA TGT GCT GAC GCA G 6. CAT AGC CGG GAG CTC TTA CTA ACA TTC TGT AGG GGT

TCT CTG TAT TGG GTC TGA

2.3. Transfection, selection and production of assembled Ig-P3A2 Sp2/0 myeloma (ATCC:CRL-158) cell lines producing assembled human antibody Ig-P3A2 of the IgGl subclass under the transcriptional control of the immunoglobulin promoter were estab-

33

lished through co-transfection of Sp2/0 cells with pSV-derived mammalian cell expression vectors pcrLys-30 (5 pg) and paLys-17 (5 pg) containing the V,-P3 heavy and VA-A2-CI light chain genes respectively by electroporation. Transfection by electroporation was carried out by delivering a single discharge to the DNA/cell suspension in phosphate buffered saline (PBS) in a sterile plastic cuvette (0.4 cm diameter) from a Gene-Pulser apparatus (Bio-Rad, Richmond, USA) set at a capacitance of 960 ,uF with a field strength across the electrode of 5 kV/cm. After electroporation, the cells were incubated on ice for 1 h, then added to 9 ml of pre-warmed RPM1 1640 medium (Sigma, R-6504) containing 10% fetal calf serum (FCS), 1.8 g/l NaHCO,, and media supplements L-glutamine (Sigma, G-75 13), non essential amino acids (Sigma, M-7145) and sodium pyruvate (Sigma, S-8636) all diluted 1:lOO ml of medium, and cultured for 48 h before passaging into selective medium. After 4 weeks culture in selective medium containing hygromycin (400 pg/ml), mycophenolic acid (1 pg/ ml) and xanthine (250 fig/ml), clones were isolated from transfected cell pools by limiting dilution in non-selective medium, and cell lines with the highest titres of Ig-P3A2 in the supernatant as determined by ELISA isolated. Clones producing assembled Ig-P3A2 at a constant level after five to ten passages were expanded and used for subsequent investigations. After screening expanded clones for productivity, the clone P3A2.B5 was selected. Limiting dilution was performed once again to ensure monoclonality, and the clone expanded to produce a master cell bank (MCB). The recombinant Sp2/0 cell line P3A2.B5 was subsequently weaned into a protein-free medium formulation (Sigma, S-2897). For production of assembled Ig-P3A2 for characterisation studies, P3A2.B5 was cultured in vitro in protein-free medium using spinner flasks (Techne, UK) with a working volume of 250 ml. 2.4. Protein A puriJication The expressed Ig-P3A2 was purified from culture supernatants immediately after harvesting by

34

S.M. Mahler et al. /Immunotechnology 3 (1997) 31-43

Protein A affinity chromatography. Cell culture supernatant (10 1) was passed over a 10 ml column of Protein A-Sepharose CL4B (Pharmacia), washed with 20 column volumes of PBS, and eluted with 0.1 M glycine buffer, pH 3.0. The eluate was immediately neutralised with 1 M Tris-HCl, pH 9. 2.5. ELBA 2.5.1. Assembled human IgGl ELISA Quantitation of assembled Ig-P3A2 was performed by a sandwich ELISA using a goat antiFc fragment-specific IgG capture human, antibody (10 pug/ml in carbonate buffer) (Sigma, I-2136), and a biotinylated goat anti-human i. light chain bound and free (B + F) as the secondary antibody (Sigma, B-0900) followed by the complex addition of Extravidin-peroxidase (Sigma, E-2866). The reaction product was detected using [2,2’ azino-bis-(3-ethyl-benzthiazoline-6 sulphonate)] (ABTS) as substrate. Human IgGl 1 (Sigma, I-4014) was used as standard. 2.5.2. Lambda light chain-speciJi‘c ELISA Quantitation of total /z light chain was performed by a sandwich ELISA using a rabbit-antihuman 2 light chain (B + F) capture (Sigma, L-7646) and a biotinylated goat anti-human rZ light chain (B + F) as the secondary antibody (Sigma, B-0900), followed by the addition of Extravidin-peroxidase complex (Sigma, E-2866). The reaction product was detected using ABTS as substrate. Human IgGl rl (Sigma, I-4014) was used as standard. 2.5.3. TNFo-specific ELISA TNFcl (Peptide Technology, Sydney, Australia) was coated at a concentration of 5 pg/ml in PBS. An enzyme amplification system (AmpakTM, Dako, Cambridgeshire, UK) was used for detection of TNFa-specific antibodies, which employs an alkaline phosphatase-conjugated goat anti-human, Fc fragment-specific IgG as the secondary antibody (Jackson, 109-035-008) and NADPH as substrate. Dephosphorylation of NADPH yielding NADH activates a redox cycle culminating in

the production of the coloured dye formazan, which is measured at 492 nm [16]. 2.6. Ajjinity determination by equilibrium capture analysis An indirect competition ELISA 117,181 was used to measure the binding affinity of Ig-P3A2 in solution. In this method, a fixed amount of antibody is incubated with a range of antigen concentrations. After equilibrium is reached, the amount of antibody binding to immobilised antigen is measured. The amount of antibody which binds to the solid phase antigen should be significantly less than the total amount of antibody, so that binding of antibody to solid phase antigen does not appreciably influence the solution equilibrium. 2.6.1. Establishment of’ Ig-P3A2 antibody concentration in equilibrium capture studies Two ELISA plates were coated with TNFa (5 pg/ml in PBS) overnight at 4°C. Ig-P3A2 antibody stock solutions were prepared at concentrations of 7.5, 5, and 2.5 nM in PBS containing 0.02% bovine serum albumin (BSA). (Note: concentration of Ig-P3A2 stock based on A,,, measurement for Protein A-purified Ig-P3A2 from protein-free medium). Plate 1 was washed three times with PBS-Tween, then three times with PBS. A 0.1 ml aliquot of each of the Ig-P3A2 solutions was added to the wells (4 wells/solution). The plate was incubated at 37°C for 1 h, after which Ig-P3A2 solutions were removed from the plate and pooled (400 ~1 each sample). Plate two was washed as previously described. The pre-incubated lg-P3A2 solutions, as well as original stock Ig-P3A2 solutions were assayed for Ig-P3A2 in triplicate by TNFa-specific ELISA as described in Section 2.5.3. 2.62. Equilibrium capture studies TNFa-speciJic competitive ELISA (i) TNFcr solutions of the following concentrations were prepared; 600, 200, 60, 20, 5 and 2 nM (dilutions were performed in PBS containing 0.2% BSA). Ig-P3A2 solution (250 ~1) was added to each of the TNFcl solutions (250 ~1) in Eppendorf tubes to give a final Ig-P3A2 concentration of 5

SM.

Mahler

et al. / Immunotechnology

3 (1997) 31-43

35

Fig. I. Mammalian cell expression vectors paLys-17 and palys-30 respectively in Sp2/0 mouse myeloma cells.

for the expression of immunglobulin light and heavy chains

nM. The solutions were incubated for 18 h at 30°C. Samples were assayed for Ig-P3A2 as per ELISA method in Section 2.4. Plotting of data and statistical analysis were performed using the statistics and graphics program Statview 512TM (Abacus Concepts).

pressures of firstly mycophenolic acid with xanthine, and secondly hygromycin. Surviving cells were expanded and frozen down to produce a new WCB. Enriched batch and fed-batch cultures were performed in T-75 flasks (Corning) using DF12 and Ultradoma PF medium (a protein free medium formulation for hybridoma growth) (BioWhittaker, Maryland, USA), supplemented with various levels of FCS, concentrates of amino acids (Sigma, M-7020) and RPM1 1640 vitamins (Sigma, R-7256).

2.7. Stability and productivity studies To investigate the stability of antibody secretion and productivity of P3A2.B5, a frozen vial containing P3A2.B5 was thawed from a working cell bank (WCB) and the cell line re-cloned by serial dilution. The supernatant of 15 sub-clones was assayed for Ig-P3A2 expression by the assembled human IgGl ELISA. Of the sub-clones expressing Ig-P3A2, the two sub-clones producing the highest levels of antibody (P3A2.B5.11D and P3A2.B5.6H) were expanded. They were then maintained in culture for a period of 170 days by serial sub-culture in DF12 media (a 1:l mixture of DMEM and Coons F12) (CSL, Melbourne, Australia) with 10, 2 and 0.2% FCS, and assessed for growth and stability of antibody production in periodic batch culture at days 0, 46, 109, 148 and 170. To study the productivity of P3A2.B5, a vial containing the cell line was thawed and cultured for three to four passages under the selective

3. Results 3.1. Cloning of phage library-derived human V genes into mammalian cell expression vectors The strategy for the cloning of the Vn-P3 heavy and VA-AZC, light chain genes comprising the human anti-TNFa Fab-P3A2 into mammalian cell expression vectors paLys-30 and pccLys-17 (Fig. 1) [19] respectively is shown in Fig. 2. For the heavy chain, the three DNA segments were; paLys-30 previously digested with Hind111 and Barn H 1, a Hin dIII- Pst 1 fragment obtained from the construct M13-V,PCRl (Fig. 3) [20], and a Pst 1-BamHl V, fragment obtained by PCR with primers 1 and 2 using fd-tet-DOG1 contain-

SM.

Mahler et al. / Inwwwtechnology

3 (1997) 31-43

A Hind III

Pstl

Pst I

Barn HI

LI i pal Lys-30

B Hind III

Apa LJ L 5\

\

Sac I vu2

‘i 4

Ch x

template (b)

6

t-

template (c)

pa Lys-17 Fig. 2. Cloning of the V,-P3 (A) and V,-A2-C, (B) antibody genes by tripartite ligations into pSV-derived expression vectors paLys-30 and pctlys-17 respectively. Oligonucleotides used for generation of PCR fragments are indicated by numbered arrows. Templates: (a) fd-tet-DOGI-V,-P3-C,l, (b) M13-VKPCRl, (c) pUCl9-Vi-A2-C,; L, leader sequence.

ing V,-P3-C&l heavy chain gene as template 1141. The HindIII-Pstl fragment contains a 19 amino acid leader sequence, and the first three amino acids of framework 1 of the Vu-P3 heavy chain PCR fragment contains gene. The Pst I-BamHl the remaining cDNA sequence encoding V,-P3. For the cloning of the V,-A2-C, light chain into paLys-17, the three DNA segments were; pcr Lys-17 previously digested with Hind111 and Sacl, a HindIII-ApaLl fragment obtained by PCR with primers 3 and 4 using the construct M13-VKPCRl (Fig. 3) as template [20], and an ApaLl-Sac2 light chain fragment obtained by PCR with primers 5 and 6 using pUC19 containing the V,-A2-C, light chain gene as template [14]. The HindIII-ApaLl PCR fragment contains

the first 16 amino acid residues of the leader sequence, while the ApaLl -Sue1 fragment contains the last three amino acid residues of the leader, and the V,-A2-C1 light chain gene. Upon tripartite ligation and transformation of E. coli, selected transformants were cultured, and DNA constructs prepared for electroporation. Restriction digest analysis and sequencing of the coding regions showed successful ligation of the DNA fragments (results not shown). 3.2. Expression of assembled Ig-P3A2 After transfection of Sp2/0 myeloma cells by initial expression studies on electroporation, transfected pools for total light chain V,-A2-C;

S.M. Mahler et al. /Immunotechnology 3 (1997) 31-43

31

w HindIll

TCTATACCA

AAGCITAXNATATGCAAATCCTGAATCTACATGGTAAATATAGGTTI-G

CAAACAGAAAAACATGAGATCACAGTTCKU?l-ACAG’ITACIGAGCACACAGGACCXAC LFLVAT AT MGWSCII CA’IGGGATGGAGCK3TATCATCCRXTCITGGTAGCAACAGCI’ACA~TAA-AC

AGTAGCAGGCITGAGGTCTGG

TCKCACA

ACATATATATGGGTGACAATGACATCCA-C

Pvu II GVHSDIQLTQSPS GGTGTCCAC!KCGACATCCAGCX’GACCCAGAGCCCAAGCAGC

S

f!2 HindIll AAGCITATGAATATGCAAATCCT~GAATCTACATGGTAAATATAGGTTTG’KXATACCA

CAAACAGAAAAACATGAGATCACAGlTCKXTACAGTTACTGAGCACACAGGACCTCAC MGWSCII LF L V AT A T CATGGGATGGAGCXTATCATCCI-CITClTGGTAGCAACA~ACAGGTAA-CAC

AGTAGCAGGCRGAGGTCI-GGACATATATATGGGTGACAATGACATCCACl-ITGCCI.ITC Pstl

GVHSQVQLQESGPG TCKCACAGGTGTCCACICCCAGGTCCAACIUCAGGAGAGCGGTCCAGGT

Fig. 3. Sequences of relevant segments contained in constructs M13-VKPCRl (A) and M13-VHPCRl (B). The constructs harbor human antibody_ V -genes, and the first ten residues of human framework regions are shown. The first 15 and last four residues of the leader sequence are separated by an intron.

and Ig-P3A2 by ELISA showed the presence of a significant excess of light chain V,-A2-CI (Fig. 4). Subsequent isolation of the clone P3A2.B5 by limiting dilution, and ELISA analysis of P3A2.B5 cell culture supernatant showed the ratio of V,A2-C, light chain to Ig-P3A2 to be approximately 4:l (results not shown), while no free heavy chain was detected. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis of Protein A-purified Ig-P3A2 showed the heavy and light chains to have molecular weights of approximately 55 and 28 KD respectively (Fig. 5). 3.3. Determination of afJinity of Ig-P3A2 for TNFa by equilibrium capture analysis The affinity of Protein A-purified Ig-P3A2 for TNFa was estimated by equilibrium capture anal-

ysis. The chosen concentration of Ig-P3A2 used in equilibrium capture analysis was 5 nm, established by ELISA (Section 2.6.1). A plot of the absorbance versus the Ig-P3A2 antibody concentration for both Ig-P3A2 stock solutions and preincubated Ig-P3A2 solutions is shown in Fig. 6. The plots were linear over the concentration range (fitted by linear regression), with the slopes of the fitted lines differing by approximately 11%. At the concentration of Ig-P3A2 used in equilibrium capture studies (5 nM), the amount of Ig-P3A2 which binds to solid phase TNFa was approximately 10% of the total amount of antibody in solution, indicating binding of the antibody to solid phase antigen would have a minimal effect on the solution equilibrium in equilibrium capture studies. Equilibrium capture studies were performed over a concentration range of O-300 nm TNFa.

SM. Mahler et al. /Immunotechnology

38

3 (1997) 31-43

The results are presented as the TNFcl binding curve and Klotz plot (Fig. 7). The dissociation constant (&) determined by Klotz plot was 15 nM similar to the affinity determined for Fab-P3A2 [ 141. 3.4. Cell growth and antibody production In vitro studies of the growth and expression of Ig-P3A2 by P3A2.B5 in protein free media showed the antibody titre to fall over time (results not shown). Subsequent sub-cloning of P3A2.B5 by limiting dilution and clonal expansion showed that of 15 sub-clones tested for antibody production, six were found to express Ig-P3A2 at levels up to 1.6 pg/ml in 1 ml well cultures, while the remaining nine sub-clones were found to be nonproducers. Instability of Ig-P3A2 production in this cell line over time was subsequently investigated by maintaining two of the six producing sub-clones (P3A2.B5.11D and P3A2.B5.6H) in serial subculture for a period of 170 days. At various intervals, batches were inoculated and antibody production determined. The decline in Ig-P3A2 produced in batch cultures for both sub-clones where the in600

0

1

2

3

4

5

6

7

8

9

10 11 12

Pool Fig. 4. Level of fully assembled Ig-P3A2 (black) and total V,-A2-C1 light chain, bound and free (striped), in 12 transfected pools of Sp2/0 myeloma cells

94 67 * 43 30 20 14 -

Fig. 5. SDS-PAGE analysis of Protein A-purified Ig-P3A2 produced in protein-free media. Lane 1, low molecular weight standards (Pharmacia); Lanes 2, Protein A eluate fraction 1; Lane 3, fraction 2; Lane 4, fraction 3.

oculum had been in serial sub-culture for the indicated number of days post-thawing is shown in Fig. 8. Not surprisingly, the data also highlights the inability of media supplemented with low levels of FCS to achieve a high viable cell density, with maximum viable cell numbers being up to ten fold higher in media supplemented with 10% FCS compared to those in media supplemented with 0.2% FCS. Although the maximum number of viable cells was lower in media with low levels of FCS, antibody titres in the latter stages of continuous sub-culture were similar. Preliminary findings on the effect of nutrient addition to conventional media are shown in Fig. 9. The inocula for these experiments (cultures 2-8) were prepared from a cell bank that was produced by re-subjecting P3A2.B5 to the selective pressures of mycophenolic acid and hygromycin, whereas the inoculum for culture 1 was derived from cells prior to the re-subjecting to selection pressure. Culture 1 had an almost five fold lower final production level compared to culture 2. The highest yields were obtained in

SM.

0

Mahler et al. /Immunotechnology

1

2

4

3 Ig-BA2

5

4. Discussion The V genes of the human monoclonal FabP3A2 reactive with human TNFa, isolated by guided selection [14], were re-cloned and expressed as a complete IgGl, 2 antibody molecule in the Sp2/0 mouse myeloma. About a four-fold excess of free V,-A2-CA light chain, and fully assembled Ig-P3A2 were secreted in protein-free culture of P3A2.B5, however there was no free heavy chain secreted. This is not surprising as the mechanism of intracellular antibody assembly is known to involve the association of the immunoglobulin heavy chain with the heavy chain binding protein BiP (grp 78) attached to the luminal membrane of the endoplasmic reticulum [21,22]. Furthermore, expression of the heavy and light chains are independent of one another, so that although both heavy and light chains are under the transcriptional control of the im-

6

7

8

(nm)

Fig. 6. Determination of the amount of Ig-P3A2 which binds to solid phase TNFcl solutions (7.5, 5, 2.5 nm) were incubated on TNFu coated plates (5 pg/ml) for 1 h, Ig-P3A2). Stock Ig-P3A2 (circle) and pre-incubated Ig-P3A2 (square) were then assayed results are the mean of triplicate determinations with SD. Where error bars are absent, difference in absorbance between the lines indicates the amount of Ig-P3A2 which binds of Ig-P3A2

batch and fed-batch cultures where supplementation resulted in final titres of 22-25 mg/l. Cultures 6-8 demonstrate that elevation of antibody production can also be achieved in low-serum media, despite the poorer growth performance in this formulation.

39

3 (1997) 31-43

in TNFa-specific ELISA. Stock lg-P3A2 then removed (now termed pre-incubated for Ig-P3A2 by TNFcc-specific ELISA. The the S.D. is contained within the point. The to solid phase TNFr at each concentration

munoglobulin promoter, different levels of expression may be due to copy number and position of insertion in the chromosomes. To ensure the heavy and light chains were of the correct size, and also to check for any anomalous translation products such as truncated antibody chains, SDS-PAGE analysis of Ig-P3A2 was performed. The heavy and light chains were observed to be in the size range (55 and 28 KD respectively) for IgG molecules, while no anomalous translation products were observed. Equilibrium capture analysis showed Ig-P3A2 to have a K,, of around 1.5 x lo-’ M, which is in the same order as those determined for single chain Fv (scFv)-P3A2 and Fab-P3A2 fragments binding to TNFa [14]. Clonal screening of the cell-line P3A2.B5 in the absence of selection pressure indicated a significant number of non-producing clones. The results of stability studies illustrate the decline in maximum titres achieved for the sub-clones P3A2.B5.11D and P3A2.B5.6H in batch culture when the inoculum was prepared from a flask maintained in serial sub-culture for extended periods. The maximum antibody levels obtained are shown to decline with time, indicating that during serial passage, an increasing proportion of the cells are non-producers.

S.M. Mahler et al. / Immunotechnology 3 (1997) 31-43

5 4 3. .2. .I 0

1

.5

1.5

2

2.5

log TNFa (nM)

I

t

0

5

10

IS

20

25 x 10.‘M-1

30

35

IA&+F)I

Fig. 7. Binding of Ig-P3A2 human antibody (5 nm) to solid phase TNFx in the presence of TNFr over a concentration range of O-300 nM TNFa (A). Each point represents the average of triplicate assays, and the S.D. is shown. The data are also represented as Klotz plot (B). Ag, antigen (TNFn); Ab, antibody (IgP3A2); B, bound; F. free.

Instability of antibody production has been previously reported for murine hybridoma cells [23,24] and for human-mouse heterohybridomas [25]. More recently, instability of antibody production has been reported for a transfectoma producing a chimeric antibody, in the absence of selection pressure [26]. Possible reasons for instability of antibody secretion include loss of chrothe gene(s) for the mosomes containing production of antibody chains as in the case of hybridomas [23,24]. The Sp2/0 myeloma contains 73 chromosomes, while normal mouse cells contain 40 chromosomes [27]. It is well established that transfection of Sp2/0 myeloma cells by electroporation results in low copy number, thus it is not inconceivable that random loss of chromosomes may account for loss of recombinant anti-

body genes, resulting in a decrease in antibody productivity over time. Another possibility for loss of productivity in the case of transfectomas is homologous recombination events leading to immunoglobulin gene deletion due to identical sequences in the co-transfected constructs. It has been reported that in cultured mouse cells, vector DNAs with overlapping homologous sequences undergo both extrachromosomal and chromosoma1 recombination events [28]. Instability of antibody production by transfectomas may be managed through adopting similar strategies as those adopted for hybridomas, i.e. regular recloning (after 30-35 passages) to select for antibody-producing cells. This demonstration of diminished specific productivity of a transfectoma as a function of generation number in culture

ml

148

Time poet-t?rewkrg(days)

46

170

0

0

0

46

109

148 TihIl@oat-thawing(days)

170

Fig. 8. Maximum antibody levels obtained in batch cultures of the sub-clones P3A2.B5,1lD (A) and P3A2.B5.6H (B) in media supplemented with 10% FCS (left column), 2% FCS (middle column) and 0.2% FCS (right column), commenced from an inoculum prepared after 46, 109, 148 and 170 days in culture. Maximum viable cell numbers in media supplemented with 10% FCS (diamond), 2% FCS (circle), 0.2% FCS (triangle) are also indicated.

0

a,:

g

0.2

II

1.2

_ 1.6 I 1.4

1.8

0.2

0.6

0.8

*

1.2 I

1.4 Qo _ Y

1.6 -$

1.8

f

42

SM.

1

Mahler

2

et al. / Immunotechnology

3

4

3 (1997)

5

6

31-43

7

8

Culture Fig. 9. Maximum antibody titres obtained using the indicated media and culture method. Inoculum was prepared from a WCB prepared by re-submission of P3A2.B5 to MPA and hygromycin selective pressure, except for culture 1 where inoculum was from original vial, prior to antibiotic selection): culture 1, culture in DF12 with 10% FCS; culture 2, culture in DF12 with 10% FCS; culture 3, culture in DF12 with 10% FCS, medium supplemented with glucose and glutamine; culture 4, culture in DF12 with 10% FCS, medium supplemented with glucose, glutamine and MEM amino acids; culture 5, fed-batch culture in DF12 with 10% FCS, with medium supplemented daily with glucose, glutamine and MEM amino acids; culture 6, cultured in ultradoma-PF with 0.5% FCS; culture 7, cultured in ultradoma-PF with 0.5% FCS and supplemented with additional glucose and glutamine; culture 8, cultured in ultradoma-PF with 0.5% FCS and supplemented with additional glucose, glutamine and MEM amino acids.

implies that continuous or semi-continuous fermentation strategies may not be optimal with respect to overall productivity. For both sub-clones of P3A2.B5, maximum final cell density was a function of initial serum levels. However final antibody titres were similar, suggesting that higher specific rates of productivity were being achieved in low serum-supplemented media. This was confirmed by comparing the calculated mean specific productivity of cells cultured in different levels of serum as shown in Fig. 9, cultures 2 and 6. Cells cultured in DF12 media supplemented with 10% FCS, and Ultradoma-PF supplemented with 0.5% FCS had mean specific productivities of 1.2 + 0.4 pg/cell/day and 7.8 &-2.3 pg/cell/day respectively. Final antibody levels did not increase markedly due to lower viable cell densities when the serum was removed. Higher specific Mab secretion rates have also been reported in serum-free, lipid-lean hybridoma cul-

ture, however the reasons for increased productivity are not well understood [29]. The results suggest that culture of cell lines in media with low FCS supplementation, or protein free media formulations, may be beneficial with respect to both specific antibody productivity as well as facilitating purification. As well as being able to express V genes isolated from phage display libraries as fully assembled human antibodies of a specific sub-class, it is also desirable to maximise the productivity of the host cell line used for expression. There are a number of mammalian expression systems available for the production of recombinant antibodies, however the two mammalian hosts of choice are the Chinese hamster ovary (CHO) cells, and mouse myeloma cells. Recently a whole human anti-tetanus toxoid antibody has been expressed in CHO cells, where the V genes were isolated from combinatorial phage libraries [30]. The CHO

S.M. Mahler et al. /Immunotechnology

cells constitutively express a viral transactivator protein to enhance transcription, and thus increase antibody productivity. Generally maximum productivity is achieved using amplification systems such as the glutamine synthetase (GS) system (Celltech) which utilises the NSO myeloma host, and methotrexate amplification using CHO cells. Often however, improvements in productivity of cell lines can be obtained by nutrient supplementation of production media alone [31]. Our results show that supplementation of rich media (10% FCS) with concentrates of amino acids and vitamins can increase productivity up to five fold. This supports data obtained from the culture of hybridoma lines which shows that conventional media formulations are limiting with respect to growth and antibody production [3 1,321. We have presented here a method for the production of assembled, human antibodies, including a cloning strategy useful for the expression of scFv and Fab genes as whole, assembled antibodies, and a method for improving antibody yields from in vitro cell cultures through fed batch strategies. The techniques allow production of significant amounts of human antibody for characterisation studies.

Acknowledgements We thank Jeff Foote for providing the paLys expression vectors. S.M.M. is grateful for support by Peptide Technology Limited, Australia.

References 111Ponticelli, C., Rivolta, E., Tarantino, A., Egidi, F., Ban& G., DeVecchi, A., Montagnino, G. and Vegeto, A. (1986) Transplant. Proc. 18, 9422948. PI Cosimi, A.B., Jenkins, R.L., Rohrer, R.J., Delmonico, F.L., Hoffman, N. and Monaco, A.P. (1990) Arch. Surg. 125, 781-784. [31 Sears, H.F., Herlyn, D., Steplewski, Z. and Koprowski, H. (1985) Cancer Res. 45, 5910-5913. [41 Oi, V.T., Morrison, S.L., Herzenberg, L.A. and Berg, P. (1983) Proc. Natl. Acad. Sci. USA 80, 825-829. 151LoBugho, A.F., Wheeler, R.H, Trang, J., Haynes, J., Rogers, K., Harvey, E.B., Sun, L., Ghrayeb, J and Khazaeli, M.B. (1989) Proc. Natl. Acad. Sci. USA 86, 42204224.

3 (1997) 31-43

43

[6] Mountain, A. and Adair, J. R. Biotechnol. Genet. Eng. Rev. 10, 1-142. [7] Jones, P.T., Dear, P.H., Foote, J., Neuberger, M.S. and Winter, G. (1986) Nature 321, 522-525. [8] Reichmann, L., Clark, M., Waldmann, H. and Winter, G. (1988) Nature 332, 3233327. t91 Tomlinson, I.M., Walter, G., Marks, J.D., Llewelyn, M.B. and Winter, G. (1992) J. Mol. Biol. 227, 7766798. UOI McCafferty, J., Griffiths, A.D., Winter, G. and Chiswell, D.J. (1990) Nature 348, 5522554. 1111 Hoogenboom, H.R., Griffiths, Johnson, K.S., Chiswell, D.J., Hudson, P. and Winter, G. (1991) Nucleic Acids Res. 19, 4133-4137. u21 Griffiths, A.D., Malmqvist, M., Marks, J.D., Bye, J.M., Embleton, M.J., McCafferty, J., Baier, M., Holliger, K.P., Gorick, B.D., Hughes-Jones, N.C., Hoogenboom, H.R. and Winter, G. (1993) EMBO J. 12, 7255734. u31 Winter, G., Griffiths, A.D., Hawkins, R.E. and Hoogenboom, H.R. (1994) Making antibodies by phage display technology. Ann. Rev. Immunol. 12, 4333455. v41 Jespers, LX, Roberts, A., Mahler, S.M., Winter, G. and Hoogenboom, H.R. (1994) Bio/Technology 12, 899-903. [I51 Rathjen, D. A., Furphy, L. J. and Aston, R. (1992) Br. J. Cancer 65, 852-856. 1161Stanley, C.J., Johannsson, A. and Self, C.H. (1986) J. Immunol. Methods 83, 89-95. [I71 Friguet, B., Chaffottee, A.F., Djavadi, O.L. and Goldberg, M.E. (1985) J. Immunol. Methods 77, 305-319. 1181Goldberg, M.E. and Djavadi, O.L. (1993) Curr. Opin. Immunol. 5, 278-281. 1191Foote, J. and Winter, G. (1992) J. Mol. Biol. 224, 487499. PO1 Orlandi, R., Gussow, D.H., Jones, P.T. and Winter, G. (1989) Proc. Natl. Acad. Sci. USA 86, 383333837. PU Bole, D.G., Hendershot, L.M. and Kearney, J.F. (1986) J. Cell Biol. 102, 1558-1568. LQ.1Hendershot, L.M., Bole, D.G., Kohler, G. and Kearney, J.F. (1986) J. Cell Biol. 104, 761-769. v31 Frame, K.K. and Hu, W.S. (1990) Biotechnol Bioeng. 35, 469-476. v41 Chuck, AS. and Palsson, B.O. (1992) Biotechnol. Bioeng. 39, 3544360. WI Koropatnick, J., Pearson, J. and Harris, J.F. (1988) Mol. Biol. Med. 5, 69-83. WI Bae, SW., Hong, H.Y. and Lee, G.M. (1995) Biotechnol. Bioeng. 47, 243-251. ~271Shulman, M., Wilde, C.D. and Kohler, G. (1978) Nature 276, 2699270. 1281Subramani, S. and Rubnitz, J. (1985) Molec. Cell. Biol. 5, 659-666. 1291Glassy, M.C, Tharakan, J.P. and Chau, P.C. (1988) Biotechnol. Bioeng. 32, 101551028. [301 Bender, E., Woof, J.M., Atkin, J.D., Barker, M.D., Bebbington, C.R and Burton, D.R. (1993) Human Antibodies and Hybridomas. 4, 74-79. [311 Bibila, T.A. and Robinson, D.K (1995) Biotechnol. Prog. 11, 1-13. 1321Jo, E.C., Park, H.J., Park, J.M. and Kim, K.H. (1993) Biotechnol. Bioeng. 36, 717-722.