highly sialylated recombinant human erythropoietin production in … · glycoprotein can be very...

100 © 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Biotechnol. J. 2014, 9, 100–109 DOI 10.1002/biot.201300301

www.biotechnology-journal.com

BiotechnologyJournal

1 Introduction

Protein N-glycosylation is one of the most common post-translational modifications. Unlike transcription andtranslation, the biosynthesis of glycans is not a template-driven process. As a result, the N-glycans attached to aglycoprotein can be very heterogeneous. Erythropoietin(EPO) is a glycoprotein hormone that regulates the matu-ration of red blood cells [1]. As a recombinant therapeuticdrug, it is utilized to treat anemic patients suffering fromchronic kidney disease or cancer. EPO contains three N-glycans and one O-glycan [2]. In the EPO produced by

Research Article

Highly sialylated recombinant human erythropoietin productionin large-scale perfusion bioreactor utilizing CHO-gmt4 (JW152)with restored GnT I function

John S. Y. Goh1,*, Yingwei Liu2,*, Haifeng Liu2,3, Kah Fai Chan1, Corrine Wan1, Gavin Teo1, Xiangshan Zhou2,Fusheng Xie3,4, Peiqing Zhang1, Yuanxing Zhang2, Zhiwei Song1

1 Bioprocessing Technology Institute, Agency for Science, Technology and Research (A*STAR), Singapore2 State Key Laboratory of Bioreactor Engineering, East China University of Science and Technology, Shanghai, China3 Shandong Dong-e E-Jiao Co., Ltd., Shandong, China4 Shandong Ehua Biopharmaceutical Co., Ltd., Shandong, China

Therapeutic glycoprotein drugs require a high degree of sialylation of their N-glycans for a bettercirculatory half-life that results in greater efficacy. It has been demonstrated that Chinese hamsterovary (CHO) glycosylation mutants lacking N-acetylglucosaminyltransferase I (GnT I), whenrestored by introduction of a functional GnT I, produced highly sialylated erythropoietin (EPO). Wehave now further engineered one of such mutants, JW152, by inactivating the dihydrofolate reduc-tase (DHFR) gene to allow for the amplification of the EPO gene with methotrexate (MTX). Sev-eral MTX-amplified clones maintained the ability to produce highly sialylated EPO and one wasselected for culture in a perfusion bioreactor that is used in an existing industrial EPO-productionbioprocess. Extensive characterization of the EPO produced was performed using total sialic quan-tification, HPAEC-PAD and MALDI-TOF MS analyses. Our results demonstrated that the EPO pro-duced by the mutant line exhibits superior sialylation compared to the commercially used EPO-producing CHO clone cultured under the same conditions. Therefore, this mutant has the indus-trial potential for producing highly sialylated recombinant EPO and potentially other recombinantglycoprotein therapeutics.

Keywords: CHO cells · Erythropoietin (EPO) · Glycosylation · N-Acetylglucosaminyltransferase-I (GnT I) · Sialylation

Correspondence: Dr. Zhiwei Song, Bioprocessing Technology Institute,Agency for Science, Technology and Research (A*STAR), 20 Biopolis Way,#06-01 Centros, Singapore 138668, SingaporeE-mail: [email protected]

Additional Correspondence: Dr. Yuanxing Zhang, State Key Laboratory ofBioreactor Engineering, East China University of Science and Technology,Meilong Road 130, Shanghai 200237, ChinaE-mail: [email protected]

Abbreviations: CHO, Chinese hamster ovary; DEAE, diethylaminoethanol;DHFR, dihydrofolate reductase; EPO, erythropoietin; FBS, fetal bovineserum; GnT I, N-acetylglucosaminyltransferase I; HPAEC-PAD, high pHanionic exchange chromatography-pulsed amperometric detection; IEF,isoelectric focusing; LacNAc, N-acetyllactosamine; MTX, methotrexate;PNGASE F, peptide-N-glycosidase F

Received 16 JUL 2013Revised 04 SEP 2013Accepted 24 OCT 2013Accepted article online 28 OCT 2013

Supporting information available online

* Both the authors contributed equally to this work.

© 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 101

www.biotecvisions.comwww.biotechnology-journal.com

BiotechnologyJournal Biotechnol. J. 2014, 9, 100–109

Chinese hamster ovary (CHO) cells, the glycans make upabout 40% of the total mass of the molecule. High degreeof sialylation of the N-glycans plays an important role inincreasing the circulatory half-life of EPO in vivo [3–5],resulting in higher efficacy of the drug. This is due to thepresence of asialoglycoprotein receptor on hepatocyteswhich bind to and endocytose glycoproteins that haveexposed galactose residues due to reduced sialylation [6,7]. Increased branching of N-glycans on EPO has alsobeen shown to improve the circulatory half-life of the gly-coprotein [8]. Therefore, glycosylation has a significantimpact on therapeutic EPO in terms of its circulatory half-life and consequently its efficacy in vivo.

Producing highly sialylated recombinant glycoproteindrugs has been the goal for numerous cell engineeringapproaches and process optimization efforts [9]. In tryingto improve the sialylation of recombinant glycoproteins inCHO cells, the glycosylation machinery has been manip-ulated in a few reports, via the overexpression of cytidinemonophosphate-sialic acid (CMP-sialic acid) transporteras well as galactosyltransferases and sialyltransferases[10, 11]. The knockdown of sialidase through RNA inter-ference has also demonstrated an increase in sialic acidcontent of human interferon-gamma [12]. Our group hasperformed a comprehensive functional analysis of 31 genesthat are involved in protein N-glycosylation and foundthat none of the overexpressed glycogenes includingCMP-sialic acid transporter, galactosyltransferases andsialyltransferases led to an increase in sialylation of tran-siently expressed human EPO in CHO-K1 cells [13].

We have previously shown that treating the CHO-K1cells with the cytotoxic lectin Ricinus communis agglu-tinin I (RCA-I) led to the isolation of mutant CHO cellsspecifically deficient in N-acetylglucosaminyltransferase I(GnT I). Interestingly, the restoration of functional GnT I inthese mutants led to an increase in the sialylation ofrecombinant EPO both in transient expression as well asin stably transfected clones [14]. These CHO mutantswith dysfunctional GnT I are now named CHO-glycosyla-tion mutant (gmt)4 cells. Transiently expressed fusionprotein of erythropoietin and Fc portion of IgG1 molecule(EPO-Fc) in one of the CHO-gmt4 cells, JW152 cells, withrestored GnT I function contained 23% more sialic acidover that expressed in the wild-type CHO-K1 cells [14].Although the recombinant EPO maintained high degreeof sialylation in several stably transfected mutant cloneswhen cultured in the presence of fetal bovine serum (FBS)in tissue culture flasks, it remains unknown whetherthese cells are able to maintain high degree of EPO sialy-lation following methotrexate (MTX) amplification. Inaddition, whether culturing of these cells in chemicallydefined serum-free medium in large scale would affect thedegree of sialylation also remains to be determined.

In this report, we generated a dihydrofolate reductase(DHFR)-deficient CHO-gmt4 cell line using zinc-fingernuclease technology by following a previous report [15].

The resulting cell line, named CHO-gmt4D cells, wastransfected with an EPO expression vector and stablytransfected clones were isolated. MTX-mediated geneamplification was carried out on several stably transfect-ed cells. A series of clones that were able to produce EPOwith superior sialylation were obtained. One of theseclones was cultured in an existing industrial perfusionculture-based bioprocess for producing EPO. It wasobserved that this clone produced EPO with highly sialy-lated N-glycans, demonstrating the industrial potential ofthe CHO-gmt4 mutants for producing well-sialylatedrecombinant EPO and potentially other recombinant gly-coproteins.

2 Materials and methods

2.1 Cell lines and cell culture

One of the CHO-gmt4 lines, JW152, was previously iso-lated from CHO-K1 cells [14]. These cells were cultured inIsocove’s modified Dulbecco’s medium (Life Technolo-gies, USA) supplemented with 10% FBS (Life Technolo-gies). Cells were cultured in a static incubator with 5%carbon dioxide at 37°C. The CHO-gmt4D cell line, theCHO-gmt4 cells with the DHFR gene interrupted with thezinc-finger nucleases, was cultured in the same mediawith the addition of 1 mM sodium hypoxanthine and0.16 mM thymidine. Cells undergoing selection for DHFRand amplification were cultured in IMDM supplementedwith 10% dialyzed FBS and varying amounts of MTX. Theindustrial CHO cell line producing recombinant EPO (E-Hua Biotech Pharmaceutical Co. Ltd.) were main-tained in Dulbecco’s modified Eagle medium: NutrientMixture F-12 (Life Technologies) with 10% fetal calf serum(Life Technologies).

2.2 Interruption of DHFR gene in CHO-gmt4 using zinc-finger nucleases

Constructs that express the zinc finger nucleases to tar-get the DHFR gene in CHO cells were generated based onprevious publication [15]. The zinc-finger motifs wereincluded in a structural scaffold for zinc-finger nucleasesadopted from previous publications [16, 17].

2.3 Vector construction and transfection for the expression of EPO

The pEGD vector (Fig. 1A) was constructed for the tri-cistronic expression of EPO, GnT I, and DHFR driven by asingle cytomegalovirus (CMV) promoter. In the pEGD vec-tor, the open reading frames of EPO, GnT I, and DHFR arelinked together by ECMV internal ribosome entry site (IRESwt) and an attenuated IRES (IRESatt). A modified pcDNA 3.1(+) vector, that was used to create


www.biotechnology-journal.com www.biotecvisions.com


the pEGD vector, was provided by Dr. Yuansheng Yang[18]. CHO-gmt4D cells were transfected with the pEGDvector using Lipofectamine 2000 (Invitrogen, USA).

2.4 Amplification and clone selection

CHO-gmt4D cells transfected with pEGD were subcul-tured 24 h after transfection and the culture medium wasreplaced with selection medium which consists of IMDMsupplemented with 10% dialyzed FBS. When a stablytransfected pool was obtained, single clones were pickedand analyzed by dot blot analyses for EPO expression. Thesialylation patterns of the recombinant EPO from theseclones were analyzed by isoelectric focusing (IEF) fol-lowed by Western blot [19]. Clones with high productivityand superior sialylation patterns were then amplified with50 nM MTX. After amplification, single clones with high-er productivity and superior sialylation patterns were fur-

ther amplified with 250 nM MTX before a final round ofsingle clone selection.

2.5 Isoelectric focusing, immunoblotting, and Coomassie blue staining

The sialylation patterns of recombinant EPO in differentsamples were analyzed by IEF followed by Western blot aspreviously described [19]. For Coomassie blue staining,IEF gels were first fixed in a solution that contains 11.5%w/v tetrachloroacetic acid and 3.5% w/v 5-sulfosalicylicacid in water. The gels were then stabilized in a solutioncontaining 0.5% sodium dodecyl sulfate (SDS), 25%ethanol, and 5% acetic acid in water. The IEF gels werethen stained using 0.1% w/v Coomassie blue G-250 (Bio-Rad, USA) dissolved in the destaining solution with 0.1%w/v copper sulfate. The gels were then destained in thedestaining solution (25% ethanol and 5% acetic acid inwater).

2.6 Perfusion culture and EPO ELISA

The perfusion culture followed the standard operationprocedure of Shandong E-Hua Biopharmaceutical Corpo-ration, Ltd. The culture was conducted in a 5-L CelliGenbioreactor (New Brunswick Scientific, Edison, NJ), whichwas equipped with Fibra-Cel® Disks to support cellattachment. The dissolved oxygen concentration wascontrolled at 60% air saturation, and culture pH was keptat 7.0 ± 0.05 by addition of NaOH or CO2 gas. Culture tem-perature was controlled at 37 ± 0.1°C. The perfusion ratewas adjusted to maintain the glucose concentration in theculture medium at 0.5–1.0 g/L throughout the perfusionprocess. The perfusion culture was divided into two phases, the growth phase and the production phase. Inthe growth phase, exponentially growing cells in eightroller bottles were collected and inoculated at about1 × 106 cells/mL into the bioreactor with a working vol-ume of 3.5 L and allowed to attach to the Fibra-Cel® Disks.The cells were cultured in perfusion mode in DMEM/F12with FBS and the concentration of serum was graduallyreduced from 10, 8, to 6.5% before the medium waschanged to serum-free medium. In the production phase,the culture system was washed twice with PBS to removethe serum content. Subsequently, CHO-S-SFM II (LifeTechnologies) with 0.5 mM sodium butyrate was used tomaintain the EPO production. The perfusion rate wasadjusted according to the glucose uptake rate in thebioreactor. The agitation speed of the impeller was main-tained at 120 rpm. The EPO titer was measured using anEPO ELISA kit (R&D Systems, MN, USA).

2.7 Purification of EPO

Every 20 L harvest was filtered with 0.45 (m membranesto remove the cell debris, and then the culture super-

Figure 1. EPO produced by stably transfected CHO-gmt4D cells in thepresence of functional GnT I are highly sialylated. (A) pEGD, a tricistronicvector expressing EPO, GnT I, and DHFR was used to transfect CHO-gmt4D cells. IRESwt, an encephalomycarditis virus (ECMV) internal ribo-some entry site (IRES); IRESatt, an attenuated ECMV IRES. (B) Westernblot analyses of EPO expressed in 11 stably transfected clones. Condi-tioned medium from untransfected CHO cells shows no positive signal(data not shown). (C) IEF analysis of supernatants containing EPOexpressed by nine of the stable clones in which sialylation was restored.From left: Amgen EPO, EPO produced by wild-type CHO-K1 cells, EPOproduced by nine stably transfected clones.




natant containing EPO was loaded onto a column packedwith CM Affi-Gel Blue gel (Bio-Rad) which captures mostof the proteins including EPO. Most of the pigments,nucleic acids and some proteins do not bind to the col-umn, and are removed in this step. The EPO bound to thecolumn was eluted and desalted with a Sephadex G-25column (Amersham Pharmacia Biotech, Sweden) beforebeing loaded onto a diethylaminoethanol (DEAE)Sepharose column (Bio-Rad) to remove most of the otherproteins. As the process was not optimized for the purifi-cation of EPO in the supernatant obtained from the CHO-gmt4D-GnT I culture, the EPO purified after the firstDEAE Sepharose chromatographic step was not as pureas the EPO purified from the biosimilar line cultured at thesame stage. Thus, a C4 reverse phase column (AmershamPharmacia Biotech) and a Q-Sephadex column (Bio-Rad)were used to further remove the miscellaneous proteinswithout losing EPO. To purify the highly sialylated iso-forms of EPO and remove the lesser sialylated isoforms,samples were further purified by another DEAESepharose column.

2.8 Sialic acid quantification

Sialic acid quantification of purified EPO samples wascarried out according to a previously published method[20] with slight modifications. Briefly, without the addi-tion of tetrabutylammonium borohydride, 30 μL of samplewas adjusted to pH 5.2 with the addition of 30 μL acetatebuffer (0.1 M, pH 5.0). Neuraminidase (Roche Diagnostics,Mannheim, Germany) was then added and incubated at37°C for 5 min. Sialic acid released was derivatized by theaddition of 90 μL of borate buffer (0.15 M, pH 9.4) and12 μL of malononitrile (8 g/L) and incubated for 5 min at80°C .The reaction was quenched before the fluorescencewas measured. The concentration of sialic acid was thencalculated as described [20].

2.9 Analysis of the oligosaccharides released from EPO using high pH anion exchangechromatography-pulsed amperometricdetection

One hundred micrograms of EPO, purified after the firstDEAE chromatography step, was used for oligosaccha-ride analysis. The protein samples were treated with pep-tide-N-glycosidase F (PNGase F) to release the attachedN-glycans for subsequent analysis by high pH anionicexchange chromatography-pulsed amperometric detec-tion (HPAEC-PAD) as described previously [19].

2.10 Matrix-assisted laser desorption/ionizationtime-of-flight (MALDI-TOF) mass spectrometry

EPO samples were analyzed by using matrix-assistedlaser desorption/ionization time-of-flight mass spectrom-

etry (MALDI-TOF MS) to further describe the N-glycanstructural information and distribution with reference to apreviously published protocol [21]. The detailed protocolhas been described previously [18]. Samples were treatedwith trypsin (Promega Biosciences, San Luis Obispo, CA,USA) and PNGase F (Prozyme, San Leandro, CA, USA). N-glycan structures were then assigned based on their mass-to-charge ratio (m/z), with the assistance of GlycoWorkBench software.

3 Results

3.1 Generating DHFR-deficient CHO-gmt4 cell lines

We have previously reported that CHO cells that survivedcytotoxic RCA-I treatment are deficient in functional GnTI [14]. These cell lines have now been named CHO-gmt4.One of these cell lines, JW152, was transfected with con-structs encoding zinc-finger nucleases targeting theDHFR gene [15]. Two days after transfection, cells werepassaged and cultured for 2 days followed by incubationwith fluoresceine-MTX, which stains for DHFR positivecells [22]. Cells were then sorted using fluorescence-acti-vated cell sorting (FACs) and negatively stained cellswere pooled and cultured. Two weeks later, this pool wassubjected to a second round of FACs, following which sin-gle cell clones were isolated. Genomic DNA was extract-ed from these clones and the DHFR locus targeted by thezinc-finger nuclease was amplified through PCR andsequenced. Supporting information, Fig. S1A shows thecomparison of the PCR amplicon of the DHFR targetedlocus between CHO-gmt4 and one DHFR-deficient CHO-gmt4 clone (CHO-gmt4D) after DNA gel electrophoresis.Sequencing results showed that CHO-gmt4D had a singleallele of DHFR containing a 130 bp insertion at the zinc-finger target site. Western blot analysis of the CHO-gmt4D cell lysate showed an absence of the DHFR pro-tein, as similarly demonstrated in the DG44 cell lysate(Supporting information, Fig. S1B). DG44 is a CHO cellline that lacks the DHFR gene.

3.2 Isolating the initial stable EPO-producing cell lines

The pEGD vector was constructed for the tricistronicexpression of EPO, GnT I, and DHFR driven by a singleCMV promoter. In the vector, the open reading frames ofGnT I and DHFR were linked to that of EPO by IRESwt andIRESatt (Fig. 1A). CHO-gmt4D cells were transfected withpEGD and then subcultured in IMDM supplemented with10% dialyzed FBS without hypoxanthine and thymidinesupplementation. After 2 weeks, single clones were iso-lated and 11 of them that expressed EPO were chosenafter dot blot analysis. Western blot analysis of EPO




showed that 9 out of 11 clones expressed EPO with a high-er molecular weight (Fig. 1B), suggesting the presence ofrestored GnT I function [14]. EPO expressed in the othertwo clones showed the same molecular weight as thatexpressed in CHO-gmt4 cells (data not shown), suggest-ing the absence of GnT I activity. The sialylation patternof EPO expressed in the nine clones was analyzed usingIEF/Western blot and the results are shown in Fig. 1C.Clinical EPO produced by Amgen was used as a controland shown on the far left. In this sample, the EPO bandsare located in the acidic region of the IEF gel, suggestingthat the EPO is highly sialylated. EPO produced by wild-type CHO-K1 cells is more heterogeneous with bands alsofound in the more basic region of the IEF gel (second lanein Fig. 1C). The sialylation patterns of EPO produced bynine stably transfected clones were all superior comparedto that of EPO produced by wild-type CHO-K1 cells andthey were very similar to that of clinical EPO (Fig. 1C).These results are similar to what we reported earlier [14].

3.3 Gene amplification with methotrexate in EPO-producing CHO-gmt4D lines

In order to investigate whether CHO-gmt4 cells could stillproduce highly sialylated EPO after gene amplificationwith MTX, 4 stably transfected clones (clones 9, 13, 14,and 15 shown in Fig. 1C) were cultured in amplificationmedia containing 50 nM MTX. Two weeks later, singleclones that survived 50 nM MTX were isolated from eachof the 4 original clones. Supernatants from single cellclones producing the highest levels of EPO were analyzedusing IEF. Clones 13-3 and 13-4 that were isolated fromoriginal clone 13 expressed highly sialylated EPO. Theother amplified clones had either lost functional GnT I or

had decreased sialylation (data not shown). Clones 13-3and 13-4 were chosen for further amplification with250 nM MTX and single clones were isolated 2 weeks lat-er. The EPO expressed by the top 10 clones isolated fromeach stable population were analyzed using IEF. All 10clones isolated from 13-4 cultured with 250 nM MTX dis-played highly sialylated profiles (Fig. 2). The specific pro-ductivity of clone 13-4-1, shown in Fig. 2, was 2.6 timeshigher than the parent clone 13. In comparison, the sialy-lation patterns of EPO produced by the clones isolatedfrom clone 13-3 were less superior compared to that iso-lated from clone 13-4 (data not shown). Clone 13-4-1 wasnamed CHO-gmt4D-GnT I and was cultured and analyzedin an industrial bioprocess with a perfusion-based biore-actor.

3.4 Production of EPO in a perfusion bioreactor

Next we investigated whether the clone, CHO-gmt4D-GnT I, was able to retain the superior sialylation of EPOwhen cultured in an industrial setting using a 5 L perfu-sion bioreactor. The bioreactor run was divided into agrowth phase and a production phase. In the growthphase, CHO cells were introduced into the bioreactorwhere they were allowed to adhere to the Fibra-Cel®

Disks by culturing them in DMEM/F12 media supple-mented with 10% FBS. The amount of serum in the medi-um was then gradually reduced to 8 and then 6.5%. OnDay 10, the cells were washed with phosphate bufferedsaline and the culture media was switched to a proteinfree CHO-S-SFM II media containing 0.5 mM of sodiumbutyrate for the production phase. A total of 94 L of mediawas used in the perfusion culture. The rate of media per-fusion was regulated according to the glucose level thatremained in the bioreactor. Figure 3A shows the changesin pH value, perfusion rate, and the glucose concentrationduring the bioreactor culture with the clone CHO-gmt4D-GnT I. The highest perfusion rate was 10 L/day and theproduction phase lasted for 19 days. The industrial clonewas cultured using the same bioreactor and the same cul-ture process (data not shown).

3.5 Sialylation patterns of EPO produced by CHO-gmt4D-GnT I and an industrial line in a perfusion bioreactor

During the culture, supernatant was sampled daily fromthe bioreactor and the sialylation pattern of EPO was ana-lyzed by IEF. As shown in Fig. 3B, clone CHO-gmt4D-GnTI produced better sialylated EPO compared to that pro-duced by the industrial (biosimilar) line from E-HuaBiotech Pharmaceutical Co. Ltd (Fig. 3C). In both cases,the sialylation pattern remained consistent throughoutthe production phase.

We also analyzed the purified EPO from both cell linesusing IEF. Cell culture supernatant containing EPO was

Figure 2. EPO produced by the mutant cells remained highly sialylatedafter MTX gene amplification. After selection with 50 nM MTX, EPO pro-duced by two clones isolated from clone 13 (Fig. 1), 13-3, and 13-4, wasanalyzed by IEF (left part). After selection with 250 nM MTX, EPO producedby 10 clones isolated from clone 13-4 was analyzed by IEF (right part).




first applied onto a column packed with CM Affi-Gel Bluegel. Under our experimental conditions, EPO was retainedin the column and unwanted pigments, nucleic acids andsome contaminating proteins were removed by the wash-ing buffer. EPO eluted from the CM Affi-Gel Blue gel col-umn was purified with a DEAE Sepharose column. ThisDEAE Sepharose chromatography removed most of theimpurities and EPO represents approximately 95% of thetotal protein in the eluted fraction. This partially purifiedEPO was further purified by a second DEAE Sepharosecolumn, which removes remaining impurities and lesssialylated EPO isoforms. This purified EPO contained onlyhighly sialylated EPO isoforms, as observed in theIEF/Western blot analysis of the clinical EPO produced byAmgen shown in Fig. 3D (lane 1). We analyzed EPO elut-ed after the first DEAE Sepharose chromatography usingIEF/Western blot for both cell lines. As shown in Fig. 3D,EPO produced by CHO-gmt4D-GnT I (lane 3) was bettersialylated than that produced by the industrial clone (lane 2).

The purification process was optimized for EPO pro-duced by the industrial line and not for CHO-gmt4D-GnTI cells. As a result, the EPO produced by CHO-gmt4D-GnTI was not as pure as the EPO produced by the industrialline following the first DEAE Sepharose chromatography(data not shown). Hence, additional purification stepswere performed to further purify the EPO produced byCHO-gmt4D-GnT I. However, no EPO was removed dur-ing these extra purification steps as shown by comparinglane 3 and 4 in Fig. 3D. For ease of description in subse-quent analyses, EPO produced by CHO-gmt4D-GnT I thatwas additionally purified after the first DEAE Sepharosechromatography is called CHO-gmt4D-GnT I EPO.

Purified EPO samples were analyzed using IEF andthe gel was subsequently stained with Coomassie blue tovisualize the protein bands. The results are shown inFig. 3E. Lane 1 shows EPO produced by the industrial linethat has been purified after the second DEAE Sepharosechromatography by E-Hua Pharmaceutical Co., Ltd. EPOproduced by the industrial line, purified after the firstDEAE Sepharose chromatography and CHO-gmt4D-GnTI EPO, are shown in lanes 2 and 3, respectively. Theseresults show that CHO-gmt4D-GnT I EPO is better sialy-lated than EPO produced by the industrial line after thefirst DEAE Sepharose chromatography.

3.6 Sialic acid quantification and glycan structureanalyses

To compare the sialic acid contents in the EPO producedby the two cell lines, sialic acid was removed from equalamounts of the respective purified EPO samples (Fig. 3D,lane 2 and lane 4) using neuraminidase and quantifiedusing a method described previously [20]. The resultsshowed that the amount of sialic acid released from EPOproduced by the industrial line was 0.25 nM/μg of protein

Figure 3. The mutant line produces better sialylated EPO than an industrialEPO-producing line in an established industrial process utilizing a perfusionbioreactor. (A) Variation of glucose concentration, pH, and perfusion rateduring the bioreactor run for the mutant, clone CHO-gmt4D-GnT I. (B) IEFanalysis of EPO produced by CHO-gmt4D-GnT I during the productionphase as shown in (A). (C) IEF analysis of EPO produced by the industrialEPO-producing line (from E-Hua Pharmaceutical Co., Ltd.) during the pro-duction phase. (D) IEF analysis of purified EPO produced by the industrialline and CHO-gmt4D-GnT I cell line. Lane 1: Amgen EPO. Lane 2: EPO pro-duced by the industrial line, purified after the first DEAE Sepharose chro-matography. Lane 3: EPO produced by CHO-gmt4D-GnT I, purified after thefirst DEAE Sepharose chromatography. Lane 4: EPO produced by CHO-gmt4D-GnT I was purified with DEAE Sepharose and two extra steps,reverse phase and anion exchange chromatographies. (E) IEF/Coomassieblue staining of purified EPO samples. Lane 1: EPO produced by the indus-trial line (Shandong E-Hua Biotech Pharmaceutical Co. Ltd), purified afterthe second DEAE Sepharose chromatography. Lane 2: EPO produced by theindustrial line, purified after first DEAE Sepharose chromatography. Lane 3:EPO produced by CHO-gmt4D-GnT I, purified with DEAE Sepharose andtwo extra steps, reverse phase and anion exchange chromatographies.




and the sialic acid content in CHO-gmt4D-GnT I EPO was0.4 nM/μg of protein, 60% more than the sialic acidreleased from EPO produced by the industrial line(Fig. 4A).

N-glycans released from EPO samples containing dif-ferent number of sialic acid were analyzed using HPAEC-PAD as previously described [19]. Figure 4B shows theelution profile of N-linked glycans released from equalamounts of EPO samples using PNGase F. Higher peakswere observed for N-glycans from CHO-gmt4-GnT I EPOin the tri-sialylated and tetra-sialylated glycans (Fig. 4B).The percentage composition of different sialylated gly-cans was calculated from the area under the curve in the

different elution regions for both EPO samples and sum-marized in the table in Fig. 4C. This indicated that CHO-gmt4-GnT I EPO contained higher proportions of tri-and tetra-sialylated glycans.

The N-glycans released from the EPO samples werealso analyzed by MALDI-TOF mass spectrometry to pro-vide detailed description and quantification of glycansfound in the EPO produced by both cell lines. EPO pro-duced by the industrial line and CHO-gmt4D-GnT I thatwas purified using the first step of DEAE Sepharose chro-matography were treated with PNGase F to release N-gly-cans. O-Glycans were released using reductive elimina-tion. Glycans were then permethylated and analyzed byMALDI-TOF in positive reflectron mode. The results areshown in Fig. 5A and B. In the mass spectrum for EPOproduced by the industrial line (Fig. 5A), the peaks corre-sponding to di-sialylated, tri-sialylated, and tetra-sialylat-ed peaks are highlighted in red. The glycans from EPOproduced by the industrial line have almost comparableabundance of each species of di-sialylated, tri-sialylated,and tetra-sialylated glycans. In contrast, the major peaksfound in CHO-gmt4D-GnT I EPO are tetra-sialylated gly-can species with very low abundance in di-sialylated andtri-sialylated glycans (Fig. 5B). It was also observed that inCHO-gmt4D-GnT I EPO, there was a greater abundanceof tetra-sialylated glycans with 1–2 additional N-acetyl-lactosamine (LacNAc) units (m/z 5035.9, 5485.1 in Fig. 5B).The O-linked glycosylation for both samples were largelysimilar, with the major species being mono- and di-sialy-lated O-glycans (Fig. 5C and D). This suggests that theimprovement in sialylation in CHO-gmt4D-GnT I EPO, asseen in various IEF analyses, is attributed to the increasedsialylation of N-glycans.

4 Discussion

Sialylation of therapeutic glycoproteins has been shownto affect their circulatory half-life and this has beendemonstrated in EPO. An in vivo study in mice showedthat fully sialylated EPO had a circulatory half-life of30 min whilst that of asialylated EPO was reduced to6 min [23]. It has also been shown in mice that the effica-cy of EPO varied with the number of sialic acids on theEPO molecule. The isoform that contained 14 sialic acidsshowed the greatest efficacy [24]. Consequently, com-mercially available recombinant human EPO containsonly isoforms with 9–14 sialic acid residues [24]. EPO iso-forms with less sialic acid have to be removed during thepurification process, which can amount to as much as80% of purified EPO (see review [25]). A very good exam-ple is the EPO produced by the industrial cell line used inthis study, which is provided by Shandong E-Hua BiotechPharmaceutical Co. Ltd. The EPO produced by these cellsare not highly sialylated as shown in Fig. 3. However, aftertwo rounds DEAE purification, the final product contains

Figure 4. EPO produced by the mutant line contains more sialic acidcompared to that produced by the industrial line. (A) The amounts of sialic acid in purified EPO samples produced by two different cell lines.(B) HPAEC-PAD elution profile of N-glycans released from EPO producedby the industrial line (as shown in Fig. 3D, lane 2) and CHO-gmt4D-GnT I(as shown in Fig. 3D, lane 4). The elution regions denoted by 0S, 1S, 2S,3S, and 4S correspond to peaks with asialylated, mono-, di-, tri-, and tetra-sialylated glycans. (C) The percentages of glycans with different numbersof sialic acid were tabulated from the elution profile.




Figure 5. MALDI-TOF analyses of N- and O-glycans released from purified EPO samples. (A) MALDI-TOF spectrum of N-glycans released from EPO pro-duced by the industrial line (as shown in Fig. 3D, lane 2). (B) MALDI-TOF spectrum of N-glycans released from EPO produced by CHO-gmt4D-GnT I (asshown in Fig. 3D, lane 4). The corresponding peaks in (A) and (B) are highlighted in red for comparison reasons. (C) MALDI-TOF spectrum of O-glycansreleased from EPO produced by the industrial line (as shown in Fig. 3D, lane 2) (D) MALDI-TOF spectrum of O-glycans released from EPO produced byCHO-gmt4D-GnT I (as shown in Fig. 3D, lane 4).




only highly sialylated EPO isomers (Fig. 3E, lane 1). Thesialylation pattern of this product is comparable to thatproduced by Amgen (compare Fig. 3D and 3E). To reachthis quality, however, the company has to remove 80% ofits product during the second DEAE step. We had previ-ously shown that CHO-gmt4 (JW152) cells that lack func-tional GnT I produced highly sialylated EPO if the GnT IcDNA is reintroduced into the cells [14]. In this study, wewanted to determine whether the superior sialylationwould be maintained after gene amplification with MTX.In addition, we wanted to determine whether the sialyla-tion remained superior in an industrial bioreactor settingusing serum-free medium.

The DHFR gene in JW152 was interrupted using zinc-finger nucleases, creating the cell line CHO-gmt4D. Sta-bly transfected clones expressing EPO and GnT I wereisolated. The sialylation of EPO produced by these cloneswere shown to be superior to that produced by wild-typeCHO-K1 cells. The CHO-gmt4D-GnT I cell line was cul-tured in an industrial perfusion-based bioprocess and weobserved that EPO produced by this cell line maintainedits superior sialylation and the EPO was better sialylatedthan that produced by the industrial EPO-producing line.The productivity of the current EPO-producing line pre-sented in this paper is still low for manufacturing ofrecombinant EPO. The productivity can be improved byselecting higher producing clones using increased con-centrations of MTX. In addition, the culture media andprocess parameters used in this perfusion bioreactor maybe optimized for our mutant line to achieve higher titers.Similarly, the sialylation patterns shown in Figs. 1 and 2are better than that shown in Fig. 3, suggesting that themutant cells did not achieve the highest sialylation poten-tial when cultured in the perfusion bioreactor. The medi-um and process will be optimized with the aim to improvethe sialylation pattern.

In addition to the IEF/Western blot analyses showingthe superior sialylation of EPO produced by CHO-gmt4D-GnT I cells, the improvement in sialylation compared toEPO produced by the industrial line was shown using sial-ic acid quantification, HPAEC-PAD and MALDI-TOFmass spectrometry. EPO produced by CHO-gmt4D-GnT Ihad a greater amount of sialic acid content and more tri-and tetra-antennary glycans. It also contained a greaterabundance in tetra-sialylated structures and containedone or two additional LacNAc units compared to EPO pro-duced by the industrial line. Extra LacNAc units havebeen found in EPO produced by a human cell line usinggene activation technology by Shire Pharmaceuticals.LacNAc units have also been found in CHO-producedEPO, NeoRecorman and Eprex [26]. Whether the presenceof LacNAc units on EPO would affect the efficacy of thedrug has yet to be established.

In conclusion, we have shown that CHO-gmt4 cellsare able to produce highly sialylated EPO after MTXamplification. We have further shown that the superior

sialylation of EPO is maintained in an industrial bioreac-tor culture setting. The results suggest that CHO-gmt4cells can be used for the large-scale industrial productionof EPO and possibly other therapeutic glycoproteins.

We would like to thank Dr. Natasha Pereira for criticalreview of the manuscript. This work was jointly funded bythe Agency for Science, Technology and Research(A*STAR), Singapore, and the Open Funding Project ofthe State Key Laboratory of Bioreactor Engineering(SKLBE), East China University of Science and Technolo-gy, Shanghai, China.

The authors declare no conflict of interest.

5 References

[1] Graber, S. E., Krantz, S. B., Erythropoietin and the control of red cellproduction. Annu. Rev. Med. 1978, 29, 51–66.

[2] Lin, F. K., Suggs, S., Lin, C. H., Browne, J. K. et al., Cloning andexpression of the human erythropoietin gene. Proc. Nat. Acad. Sci.USA 1985, 82, 7580–7584.

[3] Fukuda, M. N., Sasaki, H., Lopez, L., Fukuda, M., Survival of recom-binant erythropoietin in the circulation: The role of carbohydrates.Blood 1989, 73, 84–89.

[4] Takeuchi, M., Inoue, N., Strickland, T. W., Kubota, M. et al., Rela-tionship between sugar chain structure and biological activity ofrecombinant human erythropoietin produced in Chinese hamsterovary cells. Proc. Nat. Acad. Sci. USA 1989, 86, 7819–7822.

[5] Wasley, L. C., Timony, G., Murtha, P., Stoudemire, J. et al., The impor-tance of N- and O-linked oligosaccharides for the biosynthesis andin vitro and in vivo biologic activities of erythropoietin. Blood 1991,77, 2624–2632.

[6] Hudgin, R. L., Pricer, W. E., Ashwell, G., Stockert, R. J. et al., The iso-lation and properties of a rabbit liver binding protein specific forasialoglycoproteins. J. Biol. Chem. 1974, 249, 5536–5543.

[7] Morell, A. G., Gregoriadis, G., Scheinberg, I. H., Hickman, J. et al.,The role of sialic acid in determining the survival of glycoproteins inthe circulation. J. Biol. Chem. 1971, 246, 1461–1467.

[8] Misaizu, T., Matsuki, S., Strickland, T. W., Takeuchi, M. et al., Role ofantennary structure of N-linked sugar chains in renal handling ofrecombinant human erythropoietin. Blood 1995, 86, 4097–4104.

[9] Hossler, P., Khattak, S. F., Li, Z. J., Optimal and consistent proteinglycosylation in mammalian cell culture. Glycobiology 2009, 19, 936–949.

[10] Wong, N. S., Yap, M. G., Wang, D. I., Enhancing recombinant glyco-protein sialylation through CMP-sialic acid transporter over expres-sion in Chinese hamster ovary cells. Biotechnol. Bioeng. 2006, 93,1005–1016.

[11] Weikert, S., Papac, D., Briggs, J., Cowfer, D. et al., Engineering Chi-nese hamster ovary cells to maximize sialic acid content of recom-binant glycoproteins. Nat. Biotechnol. 1999, 17, 1116–1121.

[12] Zhang, M., Koskie, K., Ross, J. S., Kayser, K. J. et al., Enhancing gly-coprotein sialylation by targeted gene silencing in mammalian cells.Biotechnol. Bioeng. 2010, 105, 1094–1105.

[13] Zhang, P., Tan, D. L., Heng, D., Wang, T. et al., A functional analysisof N-glycosylation-related genes on sialylation of recombinant ery-thropoietin in six commonly used mammalian cell lines. Metab. Eng.2010, 12, 526–536.




[14] Goh, J. S., Zhang, P., Chan, K. F., Lee, M. M. et al., RCA-I-resistantCHO mutant cells have dysfunctional GnT I and expression of nor-mal GnT I in these mutants enhances sialylation of recombinant ery-thropoietin. Metab. Eng. 2010, 12, 360–368.

[15] Santiago, Y., Chan, E., Liu, P. Q., Orlando, S. et al., Targeted geneknockout in mammalian cells by using engineered zinc-finger nucle-ases. Proc. Nat. Acad. Sci. USA 2008, 105, 5809–5814.

[16] Doyon, Y., McCammon, J. M., Miller, J. C., Faraji, F. et al., Heritabletargeted gene disruption in zebrafish using designed zinc-fingernucleases. Nat. Biotechnol. 2008, 26, 702–708.

[17] Urnov, F. D., Miller, J. C., Lee, Y. L., Beausejour, C. M. et al., Highlyefficient endogenous human gene correction using designed zinc-finger nucleases. Nature 2005, 435, 646–651.

[18] Ho, S. C., Bardor, M., Feng, H., Mariati et al., IRES-mediated Tri-cistronic vectors for enhancing generation of high monoclonal anti-body expressing CHO cell lines. J. Biotechnol. 2012, 157, 130–139.

[19] Lim, S. F., Lee, M. M., Zhang, P., Song, Z., The Golgi CMP-sialic acidtransporter: A new CHO mutant provides functional insights. Gly-cobiology 2008, 18, 851–860.

[20] Markely, L. R., Ong, B. T., Hoi, K. M., Teo, G. et al., A high-through-put method for quantification of glycoprotein sialylation. Anal. Bio -chem. 2010, 407, 128–133.

[21] Morelle, W., Michalski, J. C., Analysis of protein glycosylation bymass spectrometry. Nat. Protoc. 2007, 2, 1585–1602.

[22] Gaudray, P., Trotter, J., Wahl, G. M., Fluorescent methotrexate label-ing and flow cytometric analysis of cells containing low levels ofdihydrofolate reductase. J. Biol. Chem. 1986, 261, 6285–6292.

[23] Fukuda, M. N., Sasaki, H., Lopez, L., Fukuda, M., Survival of recom-binant erythropoietin in the circulation: the role of carbohydrates.Blood 1989, 73, 84–89.

[24] Egrie, J. C., Browne, J. K., Development and characterization of nov-el erythropoiesis stimulating protein (NESP). Nephrol. Dial. Trans-plant. 2001, 16, 3–13.

[25] Grabenhorst, E., Schlenke, P., Pohl, S., Nimtz, M. et al., Genetic engi-neering of recombinant glycoproteins and the glycosylation path-way in mammalian host cells. Glycoconj. J. 1999, 16, 81–97.

[26] Shahrokh, Z., Royle, L., Saldova, R., Bones, J. et al., Erythropoietinproduced in a human cell line (Dynepo) has significant differencesin glycosylation compared with erythropoietins produced in CHOcell lines. Mol. Pharm. 2011, 8, 286–296.

© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.biotechnology-journal.com

Editorial: Latest methods and advances in biotechnologySang Yup Lee and Alois Jungbauer

http://dx.doi.org/10.1002/biot.201300522

Editorial: Biotechnology Journal – a review of 2013 and a preview of 2014Judy Peng


ReviewPhysiologically relevant organs on chipsKyungsuk Yum, Soon Gweon Hong, Kevin E. Healy and Luke P. Lee


ReviewLarge-scale production of red blood cells from stem cells: What are the technical challenges ahead?Guillaume F. Rousseau, Marie-Catherine Giarratana and Luc Douay


ReviewMolecular farming of human cytokines and blood products from plants: Challenges in biosynthesis and detection of plant-produced recombinant proteinsNicolau B. da Cunha, Giovanni R. Vianna, Thaina da Almeida Limaand Elíbio Rech


ReviewBiomaterial and cellular properties as examined through atomic forcemicroscopy, fluorescence optical microscopies and spectroscopictechniquesBirgit Kainz, Ewa A. Oprzeska-Zingrebe and José L. Toca-Herrera


ReviewMicrobial heterogeneity affects bioprocess robustness: Dynamic single-cell analysis contributes to understanding of microbial populationsFrank Delvigne and Philippe Goffin


ReviewAlgal biomass conversion to bioethanol – a step-by-step assessmentRazif Harun, Jason W. S. Yip, Selvakumar Thiruvenkadam, Wan A.W. A. K. Ghani, Tamara Cherrington and Michael K. Danquah


Research ArticleRecovery of Chinese hamster ovary host cell proteins for proteomicanalysisKristin N. Valente, Amy K. Schaefer, Hannah R. Kempton,Abraham M. Lenhoff and Kelvin H. Lee


Research ArticleHighly sialylated recombinant human erythropoietin production inlarge-scale perfusion bioreactor utilizing CHO-gmt4 (JW152) withrestored GnT I functionJohn S. Y. Goh, Yingwei Liu, Haifeng Liu, Kah Fai Chan, Corrine Wan, Gavin Teo, Xiangshan Zhou, Fusheng Xie, Peiqing Zhang, Yuanxing Zhang, Zhiwei Song


Research ArticleSecretory ranalexin produced in recombinant Pichia pastoris exhibitsadditive or synergistic bactericidal activity when used in combinationwith polymyxin B or linezolid against multi-drug resistant bacteriaRasha Abou Aleinein, Holger Schäfer and Michael Wink


Research ArticleEngineering stress tolerance of Escherichia coli by stress-inducedmutagenesis (SIM)-based adaptive evolutionLinjiang Zhu, Zhen Cai, Yanping Zhang and Yin Li


Research ArticleMini-scale cultivation method enables expeditious plasmidproduction in Escherichia coliPetra Grunzel, Maciej Pilarek, Dörte Steinbrück, Antje Neubauer,Eva Brand, Michael U. Kumke, Peter Neubauer and Mirja Krause


Research ArticleA magnetic nanobead-based bioassay provides sensitive detection ofsingle- and biplex bacterial DNA using a portable AC susceptometerMattias Strömberg, Teresa Zardán Gómez de la Torre, MatsNilsson, Peter Svedlindh and Maria Strømme


Research ArticleHydrostatic pressure and shear stress affect endothelin-1 and nitricoxide release by endothelial cells in bioreactorsFederico Vozzi, Francesca Bianchi, Arti Ahluwalia and Claudio Domenici


Technical reportA protease substrate profiling method that links site-specificproteolysis with antibiotic resistanceLisa Sandersjöö, George Kostallas, John Löfblom and Patrik Samuelson


Rapid CommunicationAlbumin-based nanocomposite spheres for advanced drug deliverysystemsHeath E. Misak, Ramazan Asmatulu, Janani S. Gopu, Ka-PohMan, Nora M. Zacharias, Paul H. Wooley and Shang-You Yang


Biotechnology Journal – list of articles published in the January 2014 issue.

Our latest Biotech Methods & Advances special issue is edited by our Editors-in-Chief Prof. Alois Jungbauerand Prof. Sang Yup Lee. As always, the special issue is a collection of the latest breakthroughs in biotech-nology. The cover is a graphical representation of some of the tools in biotechnology research. Image: © Bank-Bank – Fotolia.com.

Systems & Synthetic Biology ·Nanobiotech · Medicine

ISSN 1860-6768 · BJIOAM 9 (1) 1–170 (2014) · Vol. 9 · January 2014

1/2014Stem cellsBioreactorPlant biotech

www.biotechnology-journal.com

Biotech Methods & Advances


















highly sialylated recombinant human erythropoietin production in … · glycoprotein can be very...

Documents