WO 00/63403
`
`PCT/NL00/00247
`
`32
`
`CGT TTA AAC CTA ACA CTC TCC CCT GTT G—3',
`
`in which the
`
`annealing nucleotides are in italic and an introduced PmeI
`
`restriction enzyme recognition site is underlined.
`
`The PCR was performed on a plasmid (pCMkappa DHFR13 15C5
`
`kappa humanized) carrying the murine signal sequence and
`
`murine variable region of the light chain of 15C5 linked to
`
`the constant domain of the human IgG1 light chain (Vandamme
`
`et al. 1990; Bulens et al. 1991).
`
`The PCR resulted in a product of 340 nucleotides. The SunI
`
`and PmeI sites were introduced for cloning into the
`
`pcDNAZOOl/DHFth polylinker. The SunI site encoded two amino
`
`acids (Arg and Thr) of which the threonine residue is part of
`
`the constant region of human immunoglobulin light chains,
`
`while the arginine residue is part of the variable region of
`
`CAMPATH—lH (Crowe et al. 1992). This enabled subsequent 3'
`
`cloning into the SunI site, which was unique in the plasmid.
`
`The PCR product was digested with SunI and Pmel restriction
`
`enzymes purified over agarose gel,
`
`ligated into a BamHI, PmeI
`
`digested, and agarose gel purified pcDNAZOOl/DHFth, which
`
`was blunted by Klenow enzyme and free nucleotides. Ligation
`
`in the correct orientation resulted in loss of the BamHI site
`
`at the 5' end and preservation of the SunI and PmeI sites.
`
`The resulting plasmid was named pLC2001/DHFth (Fig 8) which
`
`plasmid can be used for linking the human light chain
`
`constant domain to any possible variable region of any
`
`identified immunoglobulin light chain for humanization.
`
`pNUT—C gamma
`
`(Huls et al., 1999) contains the constant
`
`domains,
`
`introns and hinge region of the human IgG1 heavy
`
`chain (Huls et al. 1999) and received upstream of the first
`
`constant domain the variable domain of the gamma chain of
`
`fully humanized monoclonal antibody UBS—54 preceded by the
`
`following leader peptide sequence: MACPGFLWALVISTCLEFSM
`
`(sequence: 5'— ATG GCA TGC CCT GGC TTC CTG TGG GCA CTT GTG
`
`ATC TCC ACC TGT CTT GAA TTT TCC ATG —3'). This resulted in an
`
`insert of approximately 2 kb in length.
`
`The entire gamma
`
`10
`
`15
`
`20
`
`25
`
`3O
`
`35
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`33
`
`chain was amplified by PCR using an upstream primer
`
`(UBS—UP)
`
`and the downstream primer CAME—DOWN. The sequence of UBS—UP
`
`is as follows: 5'—GAT CAC GCG TGC TAG CCA CCA TGG CAT GCC CTG
`
`GCT TC-3'
`
`in which the introduced MluI and NheI sites are
`
`underlined and the perfect Kozak sequence is italicized.
`
`The resulting PCR product was digested with NheI and PmeI
`
`restriction enzymes, purified over agarose gel and ligated to
`
`the pcDNAZOOO/Hygro(-) plasmid that is also digested with
`
`NheI and PmeI, dephosphorylated with tSAP and purified over
`
`gel. The resulting plasmid was named pUBS—Heavy2000/Hyg(—)
`
`(Fig 9). pNUT—C kappa contains the constant domain of the
`
`light chain of human IgG1 kappa (Huls et al. 1999) and
`
`received the variable domain of fully humanized monoclonal
`
`antibody UBS—54 kappa chain preceded by the following leader
`
`peptide: MACPGFLWALVISTCLEFSM (sequence: 5'— ATG GCA TGC CCT
`GGC TTC CTG TGG GCA CTT GTG ATC TCC ACC TGT CTT GAA TTT TCC
`
`ATG ~3', for details on the plasmid see UBiSys). This
`
`resulted in an insert of approximately 1.2 kb in length.
`
`The entire insert was amplified by PCR using the upstream
`
`primer UBS—UP and the downstream primer CAML~DOWN, hereby
`
`modifying the translation start site. The resulting PCR
`
`product was digested with NheI and PmeI restriction enzymes,
`
`purified over agarose gel and ligated to pcDNA200l/DHFth
`
`that was also digested with NheI and PmeI, dephosphorylated
`
`by tSAP and purified over gel, resulting in pUBS—
`
`LightZOOl/DHFth (Fig 10). To remove the extra intron which
`is located between the variable domain and the first constant
`
`domain that is present
`
`in pNUT—Cgamma and to link the signal
`
`peptide and the variable domain to the wild type constant
`
`domains of human IgG1 heavy chain,
`
`lacking a number of
`
`polymorphisms present
`
`in the carboxy—terminal constant domain
`
`in pNUT—Cgamma, a PCR product is generated with primer UBS—UP
`
`and primer UBSHV-DOWN that has the following sequence: 5'—
`
`GAT CGC TAG CTG TCG AGA CGG TGA CCA G —3',
`
`in which the
`
`introduced NheI site is underlined and the annealing
`
`nucleotides are italicized. The resulting PCR product is
`
`10
`
`15
`
`2O
`
`25
`
`30
`
`35
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`34
`
`digested with NheI restriction enzyme, purified over gel and
`
`ligated to a NheI digested and SAP—dephosphorylated
`
`pHC2000/Hyg(—) plasmid that was purified over gel. The
`
`plasmid with the insert in the correct orientation and
`
`reading frame is named pUBSZ—Heavy2000/Hyg(—)
`
`(Fig 11).
`
`For removal of an extra intron which is located between the
`
`variable domain and the constant domain that is present
`
`in
`
`pNUT—Ckappa and to link the signal peptide and the variable
`
`domain to the wild type constant domain of human IgG1 light
`
`chain, a PCR product was generated with primer UBS—UP and
`
`primer UBSLV-DOWN that has the following sequence: 5'— GAT
`
`CCG TAC GCT TGA TCT CCA CCT TGG TC -3',
`
`in which the
`
`introduced SunI site is underlined and the annealing
`
`nucleotides are in bold. Then the resulting PCR product was
`
`digested with MluI and SunI restriction enzymes, purified
`
`over gel and ligated to a MluI and SunI digested
`
`pLCZOOl/DHFth plasmid that was purified over gel. The
`
`resulting plasmid was named pUBSZ—LightZOOl/DHFth (Fig 12)
`
`The PCR product of the full—length heavy chain of UBS—54 is
`
`digested with NheI and PmeI restriction enzymes and blunted
`
`with Klenow enzyme. This fragment is ligated to the plasmid
`
`pcDNAsBOOO/DHFth that is digested with BstXI restriction
`
`enzyme, blunted, dephosphorylated by SAP and purified over
`
`gel. The plasmid with the heavy chain insert is named pUBS—
`
`Heavy3000/DHFth. Subsequently,
`
`the PCR of the light chain is
`
`digested with MluI and PmeI restriction enzymes, blunted,
`
`purified over gel and ligated to pUBS—Heavy3000/DHFth that
`
`is digested with HpaI, dephosphorylated by tSAP and purified
`
`over gel. The resulting vector is named pUBS—BOOO/DHFth (Fig
`
`13). The gene that encodes the heavy chain of UBS—54 without
`
`an intron between the variable domain and the first constant
`
`region and with a wild type carboxy terminal constant region
`
`(2031 nucleotides)
`
`is purified over gel after digestion of
`
`pUBSZ—ZOOO/Hyg(—) with EcoRI and PmeI and treatment with
`
`Klenow enzyme and free nucleotides to blunt the EcoRI site.
`
`Subsequently,
`
`the insert is ligated to a pcDNAsBOOO/DHFth
`
`10
`
`15
`
`2O
`
`25
`
`30
`
`35
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`35
`
`plasmid that is digested with BstXI, blunted,
`
`dephosphorylated with SAP and purified over gel. The
`
`resulting plasmid is named pUBSZ—HeavyBOOO/DHFth. pUBSZ—
`
`Light2001/DHFth is then digested with EcoRV and PmeI, and
`
`the 755 nucleotide insert containing the signal peptide
`
`linked to the variable domain of the kappa chain of UBS-54
`
`and the constant domain of human IgG1 kappa chain without an
`
`intron sequence is purified over gel and ligated to pUBSZ—
`
`Heavy3000/DHFth that is digested with HpaI, dephosphorylated
`
`with tSAP and purified over gel. The resulting plasmid is
`
`named pUBSZ—BOOO/DHFth (Fig 14).
`
`Plasmid pRc/CMV (Invitrogen) was digested with BstBI
`
`restriction enzymes, blunted with Klenow enzyme and
`
`subsequently digested with XmaI enzyme. The Neomycin
`
`resistance gene containing fragment was purified over agarose
`
`gel and ligated to pUBS—LightZOOl/DHFth plasmid that was
`
`digested with XmaI and PmlI restriction enzymes,
`
`followed by
`
`dephosphorylation with SAP and purified over gel
`
`to remove
`
`the DHFR CDNA. The resulting plasmid was named pUBS—
`
`LightZOOl/Neo. The fragment was also ligated to a XmaI/PmlI
`
`digested and gelpurified pcDNAZOOl/DHFth plasmid resulting
`
`in pcDNAZOOl/Neo. The PCR product of the UBS—54 variable
`
`domain and the digested and purified constant domain PCR
`
`product were used in a three—point ligation with a MluI/Pmel
`
`digested pcDNAZOOl/Neo. The resulting plasmid was named
`
`pUBSZ—lightZOOl/Neo.
`
`Example 4: Construction of CAMPATH—lH expression vectors
`
`Cambridge Bioscience Ltd.
`
`(UK) generates a 396
`
`nucleotide fragment containing a perfect Kozak sequence
`
`followed by the signal sequence and the variable region of
`
`the published CAMPATH—lH light chain (Crowe et al. 1992).
`
`This fragment contains, on the 5' end, an introduced and
`
`10
`
`15
`
`20
`
`25
`
`3O
`
`35
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`36
`
`unique HindIII site and, on the 3' end, an introduced and
`
`unique SunI site and is cloned into an appropriate shuttle
`
`vector. This plasmid is digested with HindIII and SunI and
`
`the resulting CAMPATH—lH light chain fragment is purified
`
`over gel and ligated into a HindIII/Sunl digested and agarose
`
`gel purified pLCZOOl/DHFth. The resulting plasmid is named
`
`pCAMPATH-LightZOO1/DHFth. Cambridge Bioscience Ltd.
`
`(UK)
`
`generated a 438 nucleotide fragment containing a perfect
`
`Kozak sequence followed by the signal sequence and the
`
`published variable region of the CAMPATH—lH heavy chain
`
`(Crowe et al. 1992), cloned into an appropriate cloning
`
`vector. This product contains a unique HindIII restriction
`
`enzyme recognition site on the 5' end and a unique NheI
`
`restriction enzyme recognition site on the 3' end. This
`
`plasmid was digested with HindIII and NheI and the resulting
`
`CAMPATH—lH heavy chain fragment was purified over gel and
`
`ligated into a purified and HindIII/NheI digested
`
`pHCZOOO/Hyg(—). The resulting plasmid was named pCAMPATH—
`
`Heavy2000/Hyg(-) .
`
`Example 5
`
`Construction of 15C5 expression vectors
`
`The heavy chain of the humanized version of the
`
`monoclonal antibody 15C5 directed against human fibrin
`
`fragment D—dimer
`
`(Bulens et al. 1991; Vandamme et al. 1990)
`
`consisting of human constant domains including intron
`
`sequences, hinge region and variable regions preceded by the
`
`signal peptide from the 15GB kappa light chain is amplified
`
`by PCR on plasmid ”pCMgamma NEO Skappa Vgamma Cgamma hu" as a
`
`template using CAMH—DOWN as a downstream primer and 15C5—UP
`
`as upstream primer.
`
`15C5—UP has the following sequence:
`
`5'—
`
`GA TCA CGC GTG CTA GCC ACC ATG GGT ACT CCT GCT CAG TTT CTT
`
`GGA ATC —3',
`
`in which the introduced MluI and NheI
`
`restriction recognition sites are underlined and the perfect
`
`Kozak sequence is italicized. To properly introduce an
`
`10
`
`15
`
`2O
`
`25
`
`30
`
`35
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`37
`
`adequate Kozak context,
`
`the adenine at position +4
`
`(the
`
`adenine in the ATG start codon is +1)
`
`is replaced by a
`
`guanine resulting in a mutation from an arginine into a
`
`glycine amino acid. To prevent primer dimerization, position
`
`+6 of the guanine is replaced by a thymine and the position
`
`+9 of the cytosine is replaced by thymine. This latter
`
`mutation leaves the threonine residue intact. The resulting
`
`PCR was digested with NheI and PmeI restriction enzymes,
`
`purified over gel and ligated to a NheI and PmeI digested
`
`pcDNA2000/Hygro(—),
`
`that is dephosphorylated by SAP and
`
`purified over agarose gel. The resulting plasmid is named
`
`p15C5—Heavy2000/Hyg(-). The light chain of the humanized
`
`version of the monoclonal antibody 15C5 directed against
`
`human fibrin fragment D—dimer
`
`(Bulens et al. 1991; Vandamme
`
`et al. 1990) consisting of the human constant domain and
`
`variable regions preceded by a 20 amino acid signal peptide
`
`is amplified by PCR on plasmid pCMkappa DHFR13 15C5kappa hu
`
`as a template, using CAML-DOWN as a downstream primer and
`
`15C5—UP as the upstream primer. The resulting PCR is digested
`
`with NheI and PmeI restriction enzymes, purified over gel and
`
`ligated to a NheI and PmeI digested pcDNAZOOl/DHFth that is
`
`dephosphorylated by SAP and purified over agarose gel. The
`
`resulting plasmid is named p15C5—Light2001/DHFth.
`
`Establishment of methotrexate hygromycim and
`Example 6:
`G418 selection levels.
`
`PER.C6 and PER.C6/E2A were seeded in different
`
`densities. The starting concentration of methotrexate (MTX)
`
`in these sensitivity studies ranged between 0 nM and 2500nM.
`
`The concentration which was just lethal for both cell lines
`
`was determined; when cells were seeded in densities of
`
`100.000 cells per well
`
`in a 6—wells dish, wells were still
`
`100% confluent at 10nM, approximamtely 90—100% confluent at
`
`25nM, while most cells were killed at concentration at SOnM
`
`10
`
`15
`
`20
`
`25
`
`3O
`
`35
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`38
`
`and above after 6 days to 15 days of incubation. These
`
`results are summarized in table 1. PER.C6 cells were tested
`
`for their resistance to a combination of Hygromycin and G418
`
`to select outgrowing stable colonies that expressed both
`
`heavy and light chains for the respective recombinant
`
`monoclonal antibodies encoded by plasmids carrying either a
`
`hygromycin or a neomycin resistance gene. When cells were
`
`grown on normal medium containing lOOug/ml hygromycin and 250
`
`ug/ml G418 non—transfected cells were killed and stable
`
`colonies could appear
`
`(see example 7)
`
`CHO—dhfr cells ATCC:CRL9096 are seeded in different
`
`densities in their respective culture medium. The starting
`
`concentration of methotrexate in these sensitivity studies
`
`ranges from approximately 0.5 nM to 500 nM. The
`
`concentration, which is just lethal for the cell line,
`
`is
`
`determined, and subsequently used directly after growth
`
`selection on hygromycin in the case of IgG heavy chain
`
`selection (hyg) and light chain selection (dhfr).
`
`Example 7:
`
`Transfection of EPO expression vectors to
`
`obtain stable cell lines.
`
`Cells of cell lines PER.C6 and PER.C6/E2A were seeded in
`
`40 tissue culture dishes (10 cm diameter) with approximately
`
`2—3 million cells/dish and were kept overnight under their
`
`respective conditions (10% CO2 concentration and temperature,
`
`which is 39°C for PER.C6/E2A and 37°C for PER.C6). On the
`
`next day,
`
`transfections were all performed at 37°C using
`
`Lipofectamine (Gibco). After replacement with fresh (DMEM)
`
`medium after 4 hours, PER.C6/E2A cells were transferred to
`
`39°C again, while PER.C6 cells were kept at 37°C. Twenty
`
`dishes of each cell line were transfected with 5 ug Scal
`
`digested pEPOZOOO/DHFth and twenty dishes were transfected
`
`with 5 ug ScaI digested pEPO2000/DHFRm all according to
`
`standard protocols. Another 13 dishes served as negative
`
`10
`
`15
`
`20
`
`25
`
`3O
`
`35
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`39
`
`controls for methotrexate killing and transfection efficiency
`
`which was approximately 50%. On the next day, MTX was added
`
`to the dishes in concentrations ranging between 100 and 1000
`
`nM for DHFth and 50.000 and 500.000nM for DHFRm dissolved in
`
`medium containing dialyzed FBS. Cells were incubated over a
`
`period of 4—5 weeks. Tissue medium (including MTX) was
`
`refreshed every two—three days. Cells were monitored daily
`
`for death, comparing between positive and negative controls.
`
`Outgrowing colonies were picked and subcultured. No positive
`
`clones could be subcultured from the transfectants that
`
`received the mutant DHFR gene most likely due to toxic
`
`effects of the high concentrations of MTX that were applied.
`
`From the PER.C6 and PER.C6/E2A cells that were transfected
`
`with the wild type DHFR gene, only cell lines could be
`
`established in the first passages when cells were grown on
`
`lOOnM MTX, although colonies appeared on dishes with 250 and
`
`500 nM MTX. These clones were not viable during subculturing,
`
`and were discarded.
`
`Example 8:
`
`Subculturing of transfected cells.
`
`From each cell line, approximately 50 selected colonies
`
`that were resistant to the threshold MTX concentration were
`
`grown subsequently in 96—wells, 24—wells, 6—wells plates and
`
`T25 flask in their respective medium plus MTX. When cells
`
`reached growth in T25 tissue culture flasks at least one vial
`
`of each clone was frozen and stored, and was subsequently
`
`tested for human recombinant EPO production. For this,
`
`the
`
`commercial ELISA kit from R&D Systems was used (Quantikine
`
`IVD human EPO, Quantitative Colorimetric Sandwich ELISA,
`
`cat.# DEPOO). Since the different clones appeared to have
`
`different growth characteristics and growth curves,
`
`a
`
`standard for EPO production was set as fellows: At day 0
`
`cells were seeded in T25 tissue culture flasks in
`
`10
`
`15
`
`2O
`
`25
`
`3O
`
`35
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`40
`
`concentrations ranging between 0.5 to 1.5 million per flask.
`
`At day 4. Supernatant was taken and used in the EPO ELISA.
`
`From this the production level was set as ELISA units per
`
`million seeded cells per day.
`
`(U/lE6/day) A number of these
`
`clones are given in table 2.
`
`The following selection of good producer clones was
`
`based on high expression, culturing behaviour and viability.
`
`To allow checks for long term viability, suspension growth in
`
`roller bottles and bioreactor during extended time periods,
`
`more vials of the best producer clones were frozen, and the
`
`following best producers of each cell line were selected for
`
`further investigations P8, P9, E17 and E55 in which "P"
`
`stands for PER.C6 and "E" stands for PER.C6/E2A. These clones
`
`are subcultured and subjected to increasing doses of
`
`methotrexate in a time span of two months. The concentration
`starts at the threshold concentration and increases to
`
`approximately 0.2 mM. During these two months, EPO ELISA
`
`experiments are performed on a regular basis to detect an
`
`increase in EPO production. At the highest methotrexate
`
`concentration,
`
`the best stable producer is selected and
`
`compared to the amounts from the best CHO clone and used for
`
`cell banking (RL). From every other clone 5 vials are frozen.
`
`The number of amplified EPO cDNA copies is detected by
`
`Southern blotting.
`
`Example 9:
`
`EP0 production in bioreactors
`
`The best performing EPO producing transfected stable
`
`cell line of PER.C6, P9, was brought into suspension and
`
`scaled up to l to 2 liter fermentors. To get P9 into
`
`suspension, attached cells were washed with PBS and
`
`subsequently incubated with JRH EXCell 525 medium for PER.C6
`
`(JRH), after which the cells losen from the flask and form
`
`the suspension culture. Cells were kept at two concentrations
`
`of MTX:
`
`0 nM and 100 nM. General production levels of EPO
`
`10
`
`15
`
`2O
`
`25
`
`3O
`
`35
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`41
`
`that were reached at these concentrations (in rollerbottles)
`
`were respectively 1500 and 5700 units per million seeded
`
`cells per day. Although the lower yields in the absence of
`
`MTX can be explained by removal of the integrated DNA, it
`
`seems as if there is a shut—down effect of the integrated DNA
`
`since cells that are kept at lower concentrations of MTX for
`
`longer periods of time are able to reach their former yields
`
`when they are transferred to 100 nM MTX concentrations again
`
`(see example 11).
`
`Suspension P9 cells were grown normally with 100 nM MTX and
`
`used for inoculation of bioreactors. Two bioreactor settings
`
`were tested: Perfusion and repeated batch cultures.
`
`A. Perfusion in a 2 liter bioreactor.
`
`Cells were seeded at a concentration of 0.5 x 106 cells per
`
`ml and perfusion was started at day 3 after cells reached a
`
`density of approximately 2.3 x 106 cells per ml. The
`
`perfusion rate was 1 volume per 24 hours with a bleed of
`
`approximately 250 ml per 24 hours.
`
`In this setting, P9 cells
`
`stayed at a constant density of approximately 5 x 106 cells
`
`per ml and a viability of almost 95% for over a month. The
`
`EPO concentration was determined on a regular basis and shown
`
`in Figure 15.
`
`In the 2 liter perfused bioreactor the P9 cells
`
`were able to maintain a production level of approximately
`
`6000 ELISA units per ml. With a perfusion rate of 1 working
`
`volume per day (1.5 to 1.6 liter),
`
`this means that in this 2
`
`liter setting the P9 cells produced approximately 1 x 107
`
`units per day per 2 liter bioreactor in the absence of MTX.
`
`B. Repeated batch in a 2 liter bioreactor.
`
`P9 suspension cells that were grown on rollerbottles were
`
`used to inoculate a 2 liter bioreactor in the absence of MTX
`
`and were left to grow till a density of approximately 1.5
`
`million cells per ml after which a third of the population
`
`was removed (i 1 liter per 2
`
`to 3 days) and the remaining
`
`culture was diluted with fresh medium to reach again a
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`42
`
`density of 0.5 million cells per ml. This procedure was
`
`repeated for 3 weeks and the working volume was kept at 1.6
`
`liter. EPO concentrations in the removed medium was
`
`determined and shown in Figure 16. The average concentration
`
`was approximately 3000 ELISA units per ml. With an average
`
`period of 2 days after which the population was diluted,
`
`this
`
`means that in this 2 liter setting the P9 cells produced
`
`approximately 1.5 x 106 units per day in the absence of MTX.
`
`10
`
`15
`
`20
`
`25
`
`30
`
`C. Repeated batch in a 1 liter bioreactor with different
`
`concentrations dissolved oxygen,
`
`temperatures and pH
`
`settings.
`
`Fresh P9 suspension cells were grown in the presence of 100
`
`nM MTX in rollerbottles and used for inoculation of 4 x 1
`
`liter bioreactors to a density of 0.3 million cells per ml
`
`in
`
`JRH ExCell 525 medium. EPO yields were determined after 3,
`
`5
`
`and 7 days. The first settings that were tested were: 0.5%,
`
`10%,
`
`150% and as a positive control 50% Dissolved Oxygen
`
`(%DO).
`
`50% D0 is the condition in which PER.C6 and P9 cells
`
`are normally kept.
`
`In another run, P9 cells were inoculated
`
`and tested for EPO production at different temperatures
`
`(32°C, 34°C, 37°C and 39°C)
`
`in which 37°C is the normal setting
`
`for PER.C6 and P9, and in the third run fresh P9 cells were
`
`inoculated and tested for EPO production at different pH
`
`settings (pH 6.5, pH 6.8, pH 7.0 and pH 7.3). PER.C6 cells
`
`are normally kept at pH 7.3. An overview of the EPO yields
`
`(three days after seeding)
`
`is shown in Figure 17. Apparently,
`
`EPO concentrations increase when the temperature is rising
`
`from 32 to 39°C as was also seen with PER.C6/E2A cells grown
`
`at 39°C (Table 4), and 50% D0 is optimal for P9 in the range
`
`that was tested here. At pH 6.5, cells can not survive since
`
`the viability in this bioreactor dropped beneath 80% after 7
`
`days. EPO samples produced in these settings are checked for
`
`glycosylation and charge in 2D electrophoresis (see also
`
`35
`
`example 17).
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`43
`
`Example 10:
`
`Amplification of the DHFR gene
`
`A number of cell lines described in example 8 were used
`
`5
`
`in an amplification experiment to determine the possibility
`
`of increasing the number of DHFR genes by increasing the
`
`concentration of MTX in a time span of more than two months.
`
`The concentration started at the threshold concentration (100
`
`nM) and increased to 1800 nM with in—between steps of 200 nM,
`
`10
`
`400 nM, 800 nM and 1200 nM. During this period, EPO ELISA
`
`experiments were performed on a regular basis to detect the
`
`units per million seeded cells per day (Fig. 18). At the
`
`highest MTX concentration (1800 nM),
`
`some vials were frozen.
`
`Cell pellets were obtained and DNA was extracted and
`
`15
`
`subsequently digested with BglII, since this enzyme cuts
`
`around the wild type DHFR gene in pEPOZOOO/DHFth (Fig. 5),
`
`so a distinct DHFR band of that size would be distinguishable
`
`from the endogenous DHFR bands in a Southern blot. This DNA
`
`was run and blotted and the blot was hybridized with a
`
`20
`
`radioactive DHFR probe and subsequently with an adenovirus E1
`
`probe as a background control
`
`(Fig. 19). The intensities of
`
`the hybridizing bands were measured in a phosphorimager and
`
`corrected for background levels. These results are shown in
`
`table 3. Apparently it is possible to obtain amplification of
`
`25
`
`the DHFR gene in PER.C6 cells albeit in this case only with
`
`the endogenous DHFR and not with the integrated vector.
`
`Example 11:
`
`Stability of EPO expression in stable cell
`
`30
`
`lines
`
`A number of cell lines mentioned in example 8 were
`
`subject
`
`to long term culturing in the presence and absence of
`
`MTX. EPO concentrations were measured regularly in which 1.0
`
`35
`
`to 1.5 x 106 cells per T25 flask were seeded and left for 4
`
`days to calculate the production levels of EPO per million
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`44
`
`seeded cells per day. The results are shown in Figure 20.
`
`From this it is concluded that there is a relatively stable
`
`expression of EPO in P9 cells when cells are cultured in the
`
`presence of MTX and that there is a decrease in EPO
`
`production in the absence of MTX. However, when P9 cells were
`
`placed on 100 nM MTX again after being cultured for a longer
`
`period of time without MTX,
`
`the expressed EPO reached it
`
`original level
`
`(iBOOO ELISA units per million seeded cells
`
`per day), suggesting that the integrated plasmids are shut
`
`off but are stably integrated and can be switched back on
`
`again. It seems as if there are differences between the cell
`
`lines P8 and P9 because the production level of P8 in the
`
`presence of MTX is decreasing in time over a high number of
`
`passages (Fig. 20A), while P9 production is stable for at
`
`least 62 passages (Fig. 20B).
`
`Example 12:
`
`Transient expression of recombinant EPO on
`
`attached and suspension cells after plasmid DNA transfections
`
`pEPOZOOO/DHFth pEPOZOOO/DHFRm and pAdApt.EPO plasmids from
`
`example 2 are purified from E. coli over columns, and are
`
`transfected using Lipofectamine, electroporation, PEI or
`
`other methods. PER.C6 or PER C6/E2A cells are counted and
`
`seeded in DMEM plus serum or JRH ExCell 525 medium or the
`
`appropriate medium for transfection in suspension.
`
`Transfection is performed at 37°C up to 16 hours depending on
`
`the transfection method used, according to procedures known
`
`by a person skilled in the art. Subsequently the cells are
`
`placed at different temperatures and the medium is replaced
`
`by fresh medium with or without serum.
`
`In the case when it is
`
`necessary to obtain medium that completely lacks serum
`
`components,
`
`the fresh medium lacking serum is removed again
`
`after 3 hours and replaced again by medium lacking serum
`
`components. For determination of recombinant EPO production,
`
`samples are taken at different time points. Yields of
`
`10
`
`15
`
`2O
`
`25
`
`30
`
`35
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`45
`
`recombinant protein is determined using an ELISA kit
`
`(R&D
`
`Systems)
`
`in which 1 Unit equals approximately 10 ng of
`
`recombinant CHO- produced EPO protein (100,000 Units/mg). The
`
`cells used in these experiments grow at different rates, due
`
`to their origin, characteristics and temperature. Therefore,
`
`the amount of recombinant EPO produced is generally
`
`calculated in ELISA units/106 seeded cells/day,
`
`taking into
`
`account that the antisera used in the ELISA kit do not
`
`discriminate between non— and highly glycosylated recombinant
`
`EPO. Generally, samples for these calculations are taken at
`
`day 4 after replacing the medium upon transfection.
`
`PER.C6/E2A cells,
`
`transfected at 37°C using Lipofectamine and
`
`subsequently grown at 39°C in the presence of serum produced
`
`typically 3100 units/10°cells/day.
`
`In the absence of serum
`
`components without any refreshment of medium lacking serum
`
`these Lipofectamine—transfected cells produced typically 2600
`
`units/10°cells/day. PER.C6 cells,
`
`transfected at 37°C using
`
`Lipofectamine and subsequently grown at 37°C in the presence
`
`of serum produced typically 750 units/106 cells/day, and in
`
`the absence of serum 590 units/IO6 cells/day. For comparison,
`
`the same expression plasmids pEPOZOOO/DHFth and
`
`pEPOZOOO/DHFRm, were also applied to transfect Chinese
`
`Hamster Ovary cells (CHO, ECACC nr.85050302) using
`
`Lipofectamine, PEI, Calcium Phosphate procedures and other
`
`methods. When CHO cells were transfected using lipofectamine
`
`and subsequently cultured in Hams F12 medium in the presence
`
`of serum, a yield of 190 units/lo6 cells/day was obtained.
`
`In
`
`the absence of serum 90 units/IO6 cells/day was produced,
`
`although higher yields can be obtained when transfections are
`
`being performed in DMEM.
`
`Different plates containing attached PER.C6/E2A cells were
`
`also transfected at 37°C with pEPOZOOO/DHFth plasmid and
`
`subsequently placed at 32°C, 34°C, 37°C or 39°C to determine
`
`the influence of temperature on recombinant EPO production. A
`
`temperature dependent production level was observed ranging
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`

`

`WO 00/63403
`
`PCTINL00]00247
`
`46
`
`from 250 to 610 units/106 seeded cells/day, calculated from a
`
`day 4 sample, suggesting that the difference between
`
`production levels observed in PER.C6 and PER.C6/E2A is partly
`
`due to incubation temperatures (see also Fig 17). Since
`
`PER.C6/E2A grows well at 37°C, further studies were performed
`at 37°C.
`
`Different plates containing attached PER.C6 and PER.C6/E2A
`
`cells were transfected with pEPOZOOO/DHFth, pEPOZOOO/DHFRm
`
`and pAdApt.EPO using Lipofectamine. Four hours after
`
`transfection the DMEM was replaced with either DMEM plus
`
`serum or JRH medium lacking serum and EPO was allowed to
`
`accumulate in the supernatant for several days to determine
`
`the concentrations that are produced in the different
`
`mediums. PER.C6 cells were incubated at 37°C, while
`
`PER.C6/E2A cells were kept at 39°C. Data from the different
`
`plasmids were averaged since they contain a similar
`
`expression cassette. Calculated from a day 6 sample the
`
`following data was obtained: PER.C6 grown in DMEM produced
`
`400 units/106 seeded cells/day, and when they were kept
`
`in
`
`JRH medium they produced 300 units/lo6 seeded cells/day.
`
`PER.C6/E2a grown in DMEM produced 1800 units/106 seeded
`
`cells/day, and when they were kept
`
`in JRH they produced
`
`1100 units/106 seeded cells/day. Again a clear difference was
`
`observed in production levels between PER.C6 and PER.C6/E2A,
`
`although this might partly be due to temperature differences
`
`(see above). There was however a significant difference with
`
`PER.C6/E2A cells between the concentration in DMEM vs the
`
`concentration in JRH medium, although this effect was almost
`
`completely lost in PER.C6 cells.
`
`10
`
`15
`
`20
`
`25
`
`30
`
`EPO expression data obtained in this system are summarized in
`
`table 4. PER.C6 and derivatives thereof can be used for
`
`scaling up the DNA transfections system. According to Wurm
`
`and Bernard (1999)
`
`transfections on suspension cells can be
`
`performed at 1—10 liter set—ups in which yields of 1-10 mg/l
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`47
`
`(0 1—1 pg/cell/day) of recombinant protein have been obtained
`
`using electroporation. There is a need for a system in which
`
`this can be well controlled and yields might be higher,
`
`especially for screening of large numbers of proteins and
`
`toxic proteins that cannot be produced in a stable setting.
`
`With the Lipofectamine transfections on the best PER.C6 cells
`
`in the absence of serum, we reached 590 units/million
`
`cells/day (+/—5.9 pg/cell/day when 1 ELISA unit is
`
`approximately 10 ng EPO), while PER.C6/E2A reached 31
`
`pg/cell/day (in the presence of serum). The medium used for
`suspension cultures of PER.C6 and PER.C6/E2A (JRH ExCell 525)
`
`does not support efficient transient DNA transfections using
`
`components like PEI. Therefore the medium is adjusted to
`
`enable production of recombinant EPO after transfection of
`
`pEPO2000/DHFth and pEPOZOOO/DHFRm containing a recombinant
`
`human EPO cDNA, and pcDNAZOOO/DHFth containing other cDNA's
`
`encoding recombinant proteins.
`l to 10 liter suspension cultures of PER.C6 and PER.C6/E2A
`
`growing in adjusted medium to support transient DNA
`
`transfections using purified plasmid DNA, are used for
`
`electroporation or other methods, performing transfection
`
`with the same expression plasmids. After several hours the
`
`transfection medium is removed and replaced by fresh medium
`
`without serum. The recombinant protein is allowed to
`
`accumulate in the supernatant for several days after which
`
`the supernatant is harvested and all the cells are removed.
`
`The supernatant is used for down stream processing to purify
`
`the recombinant protein.
`
`Example 13:
`viruses.
`
`Generation of AdApt.EPO.recombinant Adeno
`
`pAdApt.EPO was co—transfected with the pWE/Ad.AflII—
`
`rITR.tetO—E4, pWE/Ad.AflII—rITR.AE2A, pWE/Ad.AflII—rITR.A
`E2A.tetO—E4 cosmids in the appropriate cell lines using
`
`10
`
`15
`
`2O
`
`25
`
`3O
`
`35
`
`

`

`WO 00/63403
`
`PCT/NL00/00247
`
`48
`
`procedures known to persons skilled in the art. Subsequently,
`
`cells were left at their appropriate temperatures for several
`
`days until full cpe was observed. Then cells were applied to
`
`several freeze/thaw steps to free all viruses from the cells,
`
`after which the cell debris was spun down. For IGAdS/AdApt
`
`EPO.dE2A. The supernatant was used to infect cells followed
`
`by an agarose overlay for plaque purification using several
`
`dilutions. After a number of days, when single plaques were
`
`clearly visible in the highest dilutions, nine plaques and
`
`one negative control
`
`(picked cells between clear plaques, so
`
`most likely not containing virus) were picked and checked for
`
`EPO production on A549 cells. All plaque picked viruses we