Strategies for analysing and improving the expression and quality of recombinant proteins made in...

11
Biotechnol. Appl. Biochem. (2009) 53, 73–83 (Printed in Great Britain) doi:10.1042/BA20080258 73 REVIEW Strategies for analysing and improving the expression and quality of recombinant proteins made in mammalian cells Nigel Jenkins* 1 , Paula Meleady, Raymond Tyther* and Lisa Murphy* *National Institute for Bioprocessing Research and Training (NIBRT), Engineering Building, University College Dublin, Belfield, Dublin 4, Republic of Ireland, and National Institute for Cellular Biotechnology (NICB), Dublin City University, Glasnevin, Dublin 9, Republic of Ireland The production of monoclonal antibodies and other recombinant proteins is one of the highest growth areas in the pharmaceutical industry. Mammalian cells are used to manufacture the majority of biotherapeutics, largely due to their ability to perform complex post-translational modifications. Although significant progress has been made recently in improving product yields and protein quality, many challenges still lie ahead to achieve consistently high yields while avoiding potentially damaging protein modifications. The present review first considers the strategies used to analyse and improve recombinant protein expression of industrial cell lines, with an emphasis on proteomic technologies. Next, cellular and environmental influences on protein production and quality are examined, and strategies for improvements in pro- duct yield and quality are reviewed. The analytical techniques required to detect these protein changes are also described, together with prospects for assay improvements. Introduction Recombinant proteins, especially MAbs (monoclonal anti- bodies) and their derivatives, constitute an increasing sector of the biotechnology and pharmaceutical industries. The ability to perform complex PTMs (post-translational modifications) is one of the major reasons that most biotherapeutics are manufactured in animal cells [1]. How- ever, proteins are prone to several modifications that can reduce their efficacy and limit their shelf life [2]. Common protein isoforms and degradation pathways include variable glycosylation, misfolding, aggregation, methionine oxidation, asparagine deamidation and proteolysis. The important field of glycosylation has been covered in several recent reviews [3–5], and the present review will focus on how the expres- sion of biopharmaceuticals can be improved without com- promising the quality attributes of recombinant proteins. Strategies to analyse and improve recombinant protein expression are covered first in this review, with emphasis on current proteomic technologies and how they can provide leads for the genetic manipulation of industrial cell lines. Potential bottlenecks in protein expression, folding and secretion are described, together with strategies to alleviate these constraints. In addition to genetic manipulation of cell lines, many environmental conditions that occur during bioprocessing can contribute to protein secretion and stability, and these are listed in the main text and associated Tables. Accurate and robust assays for PTMs are essential for an understanding of protein stability at all stages of bioprocessing, and recent improvements in protein analysis are described in the present review. Key words: mammalian cell culture, protein expression, protein modification, protein quality, secretion, therapeutics. Abbreviations used: AUC, analytical ultracentrifugation; BiP , heavy-chain (or immunoglobulin) binding protein; CHO, Chinese-hamster ovary; 2D, two-dimensional; DIGE, difference gel electrophoresis; DLS, dynamic light scattering; DSP , downstream processing; eIF, eukaryotic translation initiation factor; EPO, erythropoietin; ER, endoplasmic reticulum; Ero1p, an ER thiol oxidase; ESI, electrospray ionization; FDA, (U.S.) Food and Drug Administration; FFF, field flow fractionation; GAPDH, glycerladehyde-3- phosphate dehydrogenase; GRP , glucose-regulated protein; hGH, human growth hormone; HIC, hydrophobic-interaction chromatography; HMM, high molecular mass; HSP , heat-shock protein; ICAT, isotope-coded affinity tag; IEX, ion-exchange chromatography; iTRAQ TM , isobaric tagging for relative and absolute protein quantitation; LMM, low molecular mass; MAb, monoclonal antibody; MALDI–TOF MS, matrix-assisted laser-desorption ionization–time-of-flight MS; MALLS, multi-angle laser light scattering; MS/MS, tandem MS; MuDPIT, multi-dimensional protein identification technology; NDK, nucleoside diphosphate kinase; PACE-sol, soluble paired basic amino acid converting enzyme; PDI, protein disulfide-isomerase; PMF, peptide mass fingerprint; PTM, post-translational modification; rhBMP-2, recombinant human bone morphogenetic protein-2; SEC, size-exclusion chromatography; SELDI–TOF, surface-enhanced laser-desorption ionization–time-of-flight; SILAC, stable isotope-labelled amino acids in culture; SM, Sec1/Munc18; SNARE, soluble N-ethylmaleimide-sensitive fusion protein-attachment protein receptor; UPR, unfolded protein response; XBP-1, X-box binding protein-1. 1 To whom correspondence should be addressed (email [email protected]). C 2009 Portland Press Ltd

Transcript of Strategies for analysing and improving the expression and quality of recombinant proteins made in...

Page 1: Strategies for analysing and improving the expression and quality of recombinant proteins made in mammalian cells

Biotechnol. Appl. Biochem. (2009) 53, 73–83 (Printed in Great Britain) doi:10.1042/BA20080258 73

REVIEWStrategies for analysing and improving the expression andquality of recombinant proteins made in mammalian cells

Nigel Jenkins*1, Paula Meleady†, Raymond Tyther* and Lisa Murphy*

*National Institute for Bioprocessing Research and Training (NIBRT), Engineering Building, University College Dublin,Belfield, Dublin 4, Republic of Ireland, and †National Institute for Cellular Biotechnology (NICB), Dublin City University,Glasnevin, Dublin 9, Republic of Ireland

The production of monoclonal antibodies and otherrecombinant proteins is one of the highest growth areasin the pharmaceutical industry. Mammalian cells areused to manufacture the majority of biotherapeutics,largely due to their ability to perform complexpost-translational modifications. Although significantprogress has been made recently in improving productyields and protein quality, many challenges still lieahead to achieve consistently high yields while avoidingpotentially damaging protein modifications. Thepresent review first considers the strategies used toanalyse and improve recombinant protein expressionof industrial cell lines, with an emphasis on proteomictechnologies. Next, cellular and environmentalinfluences on protein production and quality areexamined, and strategies for improvements in pro-duct yield and quality are reviewed. The analyticaltechniques required to detect these protein changesare also described, together with prospects for assayimprovements.

Introduction

Recombinant proteins, especially MAbs (monoclonal anti-bodies) and their derivatives, constitute an increasingsector of the biotechnology and pharmaceutical industries.The ability to perform complex PTMs (post-translationalmodifications) is one of the major reasons that mostbiotherapeutics are manufactured in animal cells [1]. How-ever, proteins are prone to several modifications that canreduce their efficacy and limit their shelf life [2]. Commonprotein isoforms and degradation pathways include variableglycosylation, misfolding, aggregation, methionine oxidation,asparagine deamidation and proteolysis. The important fieldof glycosylation has been covered in several recent reviews[3–5], and the present review will focus on how the expres-sion of biopharmaceuticals can be improved without com-promising the quality attributes of recombinant proteins.

Strategies to analyse and improve recombinant proteinexpression are covered first in this review, with emphasis oncurrent proteomic technologies and how they can provideleads for the genetic manipulation of industrial cell lines.Potential bottlenecks in protein expression, folding andsecretion are described, together with strategies to alleviatethese constraints. In addition to genetic manipulationof cell lines, many environmental conditions that occurduring bioprocessing can contribute to protein secretionand stability, and these are listed in the main text andassociated Tables. Accurate and robust assays for PTMsare essential for an understanding of protein stabilityat all stages of bioprocessing, and recent improvementsin protein analysis are described in the presentreview.

Key words: mammalian cell culture, protein expression, protein modification,protein quality, secretion, therapeutics.

Abbreviations used: AUC, analytical ultracentrifugation; BiP, heavy-chain (orimmunoglobulin) binding protein; CHO, Chinese-hamster ovary; 2D,two-dimensional; DIGE, difference gel electrophoresis; DLS, dynamic lightscattering; DSP, downstream processing; eIF, eukaryotic translation initiationfactor; EPO, erythropoietin; ER, endoplasmic reticulum; Ero1p, an ER thioloxidase; ESI, electrospray ionization; FDA, (U.S.) Food and DrugAdministration; FFF, field flow fractionation; GAPDH, glycerladehyde-3-phosphate dehydrogenase; GRP, glucose-regulated protein; hGH, humangrowth hormone; HIC, hydrophobic-interaction chromatography; HMM,high molecular mass; HSP, heat-shock protein; ICAT, isotope-coded affinitytag; IEX, ion-exchange chromatography; iTRAQTM, isobaric tagging forrelative and absolute protein quantitation; LMM, low molecular mass; MAb,monoclonal antibody; MALDI–TOF MS, matrix-assisted laser-desorptionionization–time-of-flight MS; MALLS, multi-angle laser light scattering;MS/MS, tandem MS; MuDPIT, multi-dimensional protein identificationtechnology; NDK, nucleoside diphosphate kinase; PACE-sol, soluble pairedbasic amino acid converting enzyme; PDI, protein disulfide-isomerase; PMF,peptide mass fingerprint; PTM, post-translational modification; rhBMP-2,recombinant human bone morphogenetic protein-2; SEC, size-exclusionchromatography; SELDI–TOF, surface-enhanced laser-desorptionionization–time-of-flight; SILAC, stable isotope-labelled amino acids inculture; SM, Sec1/Munc18; SNARE, soluble N-ethylmaleimide-sensitivefusion protein-attachment protein receptor; UPR, unfolded proteinresponse; XBP-1, X-box binding protein-1.

1 To whom correspondence should be addressed ([email protected]).

C© 2009 Portland Press Ltd

Page 2: Strategies for analysing and improving the expression and quality of recombinant proteins made in mammalian cells

74 N. Jenkins and others

Table 1 Proteomic studies on industrial cell lines

Condition Method Cell type Reference(s)

Comparing high 2D PAGE/MS NS0 [22–24,102]and low producers 2D DIGE/MS CHO [21]

iTRAQ /MS CHO [20]2D PAGE/iTRAQ/MS NS0 [25]

Temperature shift 2D PAGE CHO [36,38]

Sodium butyrate 2D PAGE/MS CHO [27,29]

DMSO 2D PAGE/MS CHO [31]

Hyperosmotic pressure 2D PAGE/MS CHO [33]

Proteomic technologies used toidentify proteins that affectproductivity

The bottlenecks in the cellular machinery for the efficientproduction of recombinant proteins are poorly understood.There have been a number of studies published usingexpression microarray [6–8] and proteomic technologies(Table 1) to gain insights into the biology of mammaliancell lines used for biopharmaceutical production. Somestudies have compared cell lines producing different levelsof recombinant protein (from low to high producers),whereas others have used profiling tools to deduce whysome media supplements (e.g. butyrate or DMSO) orenvironmental conditions (e.g. hyperosmotic pressure ortemperature shift) results in an increase in productivity.These studies have revealed many genes and proteins thatare altered under such conditions and are related to diversebiological functions such as protein folding and secretion,cell metabolism, cytoskeletal architecture, cell growth orapoptosis. The challenge is to translate these findings intofunctional cell responses to bioprocess conditions, in orderto improve productivity and product quality attributes.

There are several proteomic technologies used forprotein expression profiling, including 2D (two-dimensional)PAGE, SELDI-TOF (surface-enhanced laser-desorptionionization–time-of-flight) MS, protein arrays and stableisotope labelling technologies such as ICAT (isotope-codedaffinity tag), iTRAQTM (isobaric tagging for relative andabsolute protein quantitation) and SILAC (stable-isotope-labelled amino acids in culture).

2D PAGE and MS techniques2D PAGE is the most widely used proteomic techniquewith which to study the proteome [9], although thetechnique can be challenging, owing to its low sensitivity andpoor reproducibility. There are other limitations, includingdifficulties in separating low-abundance proteins, membrane

proteins, HMM (high-molecular mass) and LMM (lowmolecular mass) proteins and also proteins with extremepI values. DIGE (difference gel electrophoresis, a modifiedversion of 2D PAGE) enables separation of two or threefluorescently labelled protein samples using Cy2, Cy3 andCy5 dyes on the same gel. This technique offers increasedthroughput, ease of use, good reproducibility and accuratequantification of protein-expression differences [10].

MS techniques such as MALDI–TOF MS (matrix-assisted laser-desorption ionization–time-of-flight MS)generate PMFs (peptide mass fingerprints), allowing proteinsto be identified through subsequent database searching. Fur-thermore, ESI-MS/MS (electrospray-ionization tandem MS)techniques are capable of providing amino-acid-sequenceinformation from peptide fragments of the parent protein[11] and are often used together with MALDI-TOF MS todetect protein PTMs.

Stable-isotope labellingICAT uses stable-isotope labelling to perform quantitativeanalysis of paired protein samples [12]. Two differentisotope tags are generated using linkers that contain eithereight hydrogen (1H) atoms (d0, light reagent) or eight2H atoms (d8, heavy reagent) which bind covalently tocysteine residues within proteins. Both samples are mixed,digested with trypsin and fractionated by avidin-affinitychromatography. Spectral analysis in single-MS mode ofthe isotopically resolved peptides enables quantificationof the relative amounts of the peptide and, hence, theprotein levels. Differentially expressed proteins can then beidentified by MS/MS sequencing.

iTRAQ is an isobaric labelling technology similar toICAT, the main difference being that the amine residuesin peptides are differentially labelled [13]. After trypsindigestion up to eight different samples can be labelled witheight independent iTRAQ reagents. The reporter groups ofthe iTRAQ reagents split from the peptide and generatesmall fragments for each sample, with m/z (mass-to-charge)ratios of 113, 114, 115, 116, 117, 118, 119 and 121. Theintensity of each peak represents the quantity of the smallreporter fragment, and the quantity of each peptide canbe derived. Peaks in the spectrum can be used to identifypeptide sequences and thus the proteins of interest. Quant-itative differences can be readily measured by comparing theamounts of each peptide labelled with the iTRAQ reagent.

SILAC is a metabolic labelling technique that can beused for quantitative proteomic profiling of cell lysates.SILAC experiments are carried out by pairing heavy and lightisotopic forms of amino acids within the protein samples.Cells are grown in culture medium where the natural formof the amino acid is replaced with a ‘heavy’ version of theamino acid, usually a modified lysine or arginine residue

C© 2009 Portland Press Ltd

Page 3: Strategies for analysing and improving the expression and quality of recombinant proteins made in mammalian cells

Protein expression and quality in mammalian cells 75

bearing an extra six or eight 13C atoms. The SILAC methodis particularly useful for analysis of membrane proteins [14]and the phosphoproteome [15].

Other proteomic technologiesThere are a number of additional proteomic profiling tech-nologies such as SELDI–TOF MS, protein arrays and MuDPIT(multi-dimensional protein identification technology) thathave been extensively used in biomedical applications [16–18]. However, to date their use in profiling recombinantmammalian cell lines for understanding phenotypic charac-teristics has been limited. However, SELDI–TOF MS hasbeen used to profile secreted proteins in conditioned mediafrom CHO (Chinese-hamster ovary) cells [19].

Proteomic analysis of high- andlow-producing cell lines

Most studies aimed at understanding productivity in large-scale mammalian cell culture have involved direct proteomiccomparisons between low- and high-producing cells. Ina study comparing recombinant CHO cells expressingdifferent amounts of green fluorescent protein, 20 proteinsinvolved in protein biosynthesis and folding, includingeIF2S3 (eukaryotic translation initiation factor 2, subunit3 gamma, 52 kDa) and Hspd1 (heat-shock 60 kDa protein1 or chaperonin) [20], were identified as being differentiallyregulated. A number of proteins involved in chromatinmodification [Hmgb1 (high-mobility group box 1) andhistone H1.2], cell growth [S100a11 (S100 calcium-bindingprotein A11)], carbohydrate metabolism [Mdh2 (malatedehydrogenase 2)] and signal transduction (annexin A1)were also found to be altered in high-producing CHOcells.

In another study using 2D DIGE and MS, differencesin the proteome of CHO-DUKX (CHO dehydrofolatereductase-negative) cells expressing rhBMP-2 (recombinanthuman bone morphogenetic protein-2) were comparedwith cells co-expressing the cleavage enzyme PACE-sol(soluble paired basic amino acid converting enzyme), whichimproves post-translational processing of the rhBMP-2dimer [21]. PACE-sol co-expression was associated witha significant increase in cellular productivity of rhBMP-2.A total of 60 proteins were also found to be differentiallyexpressed, and a substantial number of these proteinswere found to have chaperone or folding activity [e.g.the chaperone BiP (heavy-chain or immunoglobulin bindingprotein) and PDI (protein disulfide-isomerase)]. Otherdifferentially expressed proteins were involved in proteinsynthesis and translation (eIF4B and eEF1D) or cytoskeletalarchitecture (vimentin and tropomyosin 3). Four mousemyeloma (NS0) cells that had approximately equal growth

rates but various MAb specific production rates werecompared in another study [22]. Proteins that were foundto be up-regulated with increasing levels of productivityincluded ER (endoplasmic reticulum) luminal chaperonesinvolved in folding [endoplasmin and ENPL (endoplasminprecursor)] and BiP. Other up-regulated proteins includedcytosolic and mitochondrial chaperones [HSC70 (heat-shock cognate 70 stress protein) and HSP60 (heat-shockprotein of 60 kDa)], proteins involved in calcium bindingand microtubule stabilization [TCTP (translationally con-trolled tumour protein)], nucleoside metabolism [NDKA(nucleoside diphosphate kinase A)] and oxidative stress[PDX1 (pancreatic and duodenal homeobox 1)]. Down-regulated proteins included two glycolytic enzymes [fructosebisphosphate aldolase and KPY2c (pyruvate kinase, cytosolicisozyme)]. In a related study, the microsomal componentwas isolated from the same set of NS0 cells, and PDI andBiP were again found to be increased with increased levels ofMAb productivity [23]. In a more recent study on the sameset of samples, a set of 79 proteins were identified in NS0cells and analysed across all four NS0 lines with various levelsof specific MAb productivity [24]. The relative abundance ofseveral ER chaperones, non-ER chaperones, cytoskeletal,cell signalling, metabolic and mitochondrial proteins weresignificantly increased with increased productivity of MAb.The authors suggested that individual cells within parentalpopulations are more equipped for high-level recombinantprotein production than others, and that this differencecould be used to select cells that are likely to achievehigh productivity. A study profiling NS0 cells with highproductivity levels using both cDNA microarray andproteomic technologies (i.e. 2D PAGE and iTRAQ labelling)found that a large number of genes and proteins related toprotein synthesis, cell growth and cell death pathways werealtered in high-producing cell lines [25]. A total of 22 proteinchanges were identified using 2D PAGE and MS, and a further30 proteins were identified using iTRAQ labelling and MS. Ina follow-on study from this group, the transcriptional profilesof nine NS0 cell lines with various productivity levels wereanalysed using software tools (e.g. MAPPFinder) to identifypathways and biological functions that are altered statisticallyin the high producers rather than at an individual gene level[6]. Functional classes identified as being regulated includeprotein modification, vesicle trafficking, protein turnover,cell-cycle regulation and cytoskeleton-related elements [6].

Proteomic analysis of media andenvironmental influences onproductivity

Sodium butyrate is well known to enhance the productivityof recombinant proteins [26] and appears to cause a

C© 2009 Portland Press Ltd

Page 4: Strategies for analysing and improving the expression and quality of recombinant proteins made in mammalian cells

76 N. Jenkins and others

complex series of changes in CHO cells related to longevityof culture and specific productivity. A study investigatedproteins altered following elevated levels of hGH (humangrowth hormone) expression in recombinant CHO cells[27]. Cells were exposed to zinc and butyrate, and theexpression of a number of cellular proteins besides hGHwere increased in response, including GRP75 (glucose-regulated 75 kDa protein), enolase and thioredoxin. Twoproteomic studies analysing the effect of butyrate onproductivity of CHO [28,29] demonstrated changes inexpression of a number of proteins involved in either proteinfolding (GRP78), cell metabolism [GAPDH (glyceraldehyde-3-phosphate dehydrogenase)] or cytoskeletal architecture(tropomyosin 4 and β-tubulin) among others. DMSO haspreviously been shown to increase specific productivityof a fusion protein in CHO cells [30], but it is not clearhow DMSO exerts this effect. In addition to recombinanthepatitis B surface antigen, seven proteins were alteredfollowing exposure to DMSO, including four glycolyticenzymes (triose-phosphate isomerase, GAPDH, aldolaseand phosphoglycerate kinase [31]).

Hyperosmotic pressure (induced by adding salts orsugars to the culture medium) has been suggested as aneconomical solution to increase protein productivity inrecombinant CHO cells [32]. Recombinant CHO cellsexpressing a MAb were subjected to hyperosmosis in orderto understand intracellular responses of these cells topressure. Proteome profiling following CHO cell exposureto abnormal osmotic pressure (450 mOsM/l) revealedproteins that were differentially regulated compared withcontrol medium (300 mOsM/l), namely the glycolyticenzymes GAPDH and pyruvate kinase, both of which couldalter cell energy availability [33].

A switch to low-temperature cultivation has beenshown to significantly improve the specific productivity ofrecombinant cells [34,35], but the molecular mechanismsunderlying this response are poorly understood. The effectof lowering cultivation temperature from 37 to 30 ◦C on theproductivity of CHO cells producing SEAP (secreted alkalinephosphatase) was monitored using 2D PAGE and Westernblotting [36]. Ten 2D PAGE spots had significantly alteredintensities from cells incubated at 30 ◦C compared with the37 ◦C controls, and Western blots showed altered levels oftwo tyrosine-phosphorylated proteins. The transcriptomeand proteome of CHO cells producing recombinant EPO(erythropoietin) grown at different temperatures wasinvestigated [37]. The expression levels of several proteins[PDI, vimentin, NDKB, ERp57 (ER protein of 57 kDa),phosphoglycerate kinase and 71 kDa HSP] were increasedover 2-fold at 33 ◦C and two proteins [HSP90-β (β-subunitof HSP of 90 kDa) and EF2 (elongation factor 2)] weredecreased over twofold compared with the 37 ◦C controlculture. Recently, a cDNA microarray study to evaluate

the effects of temperature shift on mouse hybridoma andCHO cells found that many of the transcriptional changeswere specific to only one cell line. For example, ribosomalgenes and genes involved in oxidative phosphorylationwere altered in MAK but not in CHO cells, whereas someprotein trafficking and cytoskeletal genes were up-regulatedin CHO but not in MAK cells [7].

To summarize, a number of studies have been per-formed to date profiling mammalian cells used forbiopharmaceutical production to gain a better understandingof the biology of these cells, with the ultimate aim togenerate a cellular phenotype capable of high productivity ofrecombinant protein. The data generated so far from thesestudies suggests that the expression of a wide range of genesand proteins are altered, especially from functional classessuch as protein folding and secretion, protein synthesis,cellular architecture (cytoskeleton), cellular metabolism (e.g.glucose, protein, lipid, etc.), cellular growth/death and signaltransduction. When the data is taken all together, it suggeststhat the simultaneous modulation of several physiologicalfunctions may be a potential approach to achieving highproductivity.

Cellular interventions in the proteinfolding and secretory pathways

Protein folding and disulfide-bond formationThe lumen of the ER provides an oxidizing environmentand access to several enzymes and chaperones that assist inthe folding and secretion of proteins. PDI is a 58 kDa ER-resident enzyme that catalyses the formation and breakageof disulfide bonds between thiol groups of cysteine residuesusing the substrate glutathione. It allows proteins to find thecorrect alignment of disulfide bonds [38] and operates as achaperone to inhibit the aggregation of misfolded proteins[39]. This reaction normally allows proteins to fold quickly[40], but a mismatch of thiol pairing can also be the sourceof covalent aggregate formation and protein misfolding [41].Results from transfecting extra copies of the gene codingfor PDI alone has yielded mixed results. For example, a2.5-fold increase in the level of PDI resulted in a 15–27%rise in antibody productivity in one CHO line [38], but failedto increase thrombopoietin secretion in another CHO line.

Besides the overall level of PDI, glutathione availabilityand the activity of Ero1p (the ER thiol oxidase that oxidizesPDI) may control PDI activity. Increased expression of Ero1results in the acceleration of disulfide-bond formation andcorrect protein folding. However, reducing the levels ofglutathione in the cell can, by compromising the reductivepathway, lead to an increase in the rate of disulfide-bondformation without leading to correct protein folding [42].It is likely that multiple intervention points in the reductive

C© 2009 Portland Press Ltd

Page 5: Strategies for analysing and improving the expression and quality of recombinant proteins made in mammalian cells

Protein expression and quality in mammalian cells 77

pathway and formation of disulfide bonds will be requiredfor maximum secretion and product quality.

Role of chaperonesChaperones such as BiP, a member of the hsp70 family,have been linked to many beneficial ER functions suchas the protein translocation, folding and oligomerization[43,44]. However, insufficient ATP levels or the lack of co-chaperones, such as Lhs1p, may become rate-limiting to BiPfunctions, and increased BiP activity may stall other chaper-one functions, e.g. GRP94, calnexin or calreticulin [44].

Engineering chaperone systems by overexpressinga single component of the ER secretion machinery hasyielded mixed results as regards improving productivity.In one case the overexpression of BiP actually decreasedthe secretion rate of recombinant antibody in one CHOline [45]. The calreticulin and calnexin chaperones alsoprovide a quality-control mechanism to ensure properprotein folding. They transiently bind to newly synthesizedglycoprotein intermediates and ensure that only correctlyfolded proteins are transported to the Golgi apparatus.Calnexin and calreticulin overexpression were found tonearly double the specific productivity of thrombopoietinin recombinant CHO cultures [46]. As seen with PDI,multiple chaperone points may need to be engineered toconsistently improve secretion rates.

Cell clearance of misfolded proteinsIn theory misfolded proteins undergo proteolysis in theendosome and the resultant amino acids are recycledin the UPR (unfolded protein response) and the ER-overload response [47]. However, in practice, cells usedfor bioprocessing can become overloaded with misfoldedrecombinant protein, leading to release of misfolded andaggregated proteins, particularly at high levels of proteinexpression [48]. This can be a particular problem withmultimeric proteins such as recombinant IgG [49] andblood-clotting proteins such as Factor VIII [50], whereprotein aggregates trigger an immune response in thepatient, resulting in inhibitory antibodies to the therapeuticprotein [51]. Various strategies are being evaluated tocombat aggregation and misfolding, and these includealtering expression levels of chaperone proteins [45] ormodulating the redox potential of the cell [40]. However,more work needs to be done. Genetic engineering of thesecretion pathway in cells used for bioprocessing (of, forexample, CHO cells) holds the promise of increasing proteinyields without compromising product quality, (for example,causing the formation of aggregates or misfolded proteins).

Multiple gene activatorsXBP-1S [the truncated form of XBP-1 (X-box bindingprotein-1)] may be a more effective target for improving

productivity, since it regulates multiple genes in thesecretory pathway. XBP-1S is essential for the generation ofplasma cells, a cell type optimized for high-level productionand secretion of antibodies [52]. Activation of the precursor,XBP-1, is triggered by the accumulation of unfolded ormisfolded proteins in the ER, and the UPR activatesthe formation of its active form, XBP-1S [53,54]. XBP-1S induces the expression of many ER, Golgi andmitochondrial chaperones [55]. Overexpression of XBP-1Shas overall yielded mixed results in improving productivity.Heterologous expression of XBP-1S led to an increasein ER content and specific MAb productivity of CHO-DG44 (dihydrofolate reductase-deficient CHO) cells [56],resulting in a 40% increase in antibody titres. However,Ku et al. [54] showed that overexpressing XBP-1S didnot improve recombinant protein production in stable celllines that did not show secretory bottlenecks. By contrast,overexpression of XBP-1S in transient transfection systemsimproved protein titres of EPO in NS0 mouse cells (2-fold)and in proline-auxotrophic CHO-K1 cells (2.5-fold). Themost common interventions in the secretory pathway forrecombinant proteins are shown in Table 2.

The vesicle trafficking system manages and regulatesthe distribution of proteins within cells as well as regulatingprotein secretion. The SM (Sec1/Munc18) proteins areintegral to this system, as they regulate membrane fusion.How they contribute to exocytosis is unknown. A recentstudy has shown that the SNARE (soluble N-ethylmalei-mide-sensitive fusion protein-attachment protein receptor)-interacting SM proteins Sly1 and Mun18c have a positiveeffect on SNARE-based fusion of ER-to-Golgi- and Golgi-to-plasma-membrane-addressed exocytic vesicles. An increasein secretion of up to 40 pg/day per cell was observedin mammalian cells. This approach also co-operated withXBP-1-mediated ER/Golgi organelle engineering, whichshould dramatically increase the secretory capacity ofmammalian production cell lines [57].

If cells are selected on the basis of high expression,there is a risk that protein quality may be compromised,since post-translational machinery of the cells may becomesaturated.

Cellular and environmental influenceson protein modifications

Protein aggregationAggregation during the manufacture of recombinantproteins can occur at all stages of bioprocessing, namelyduring cell culture, protein purification, formulation andfilling. Protein aggregates can arise by different mechanisms,including reversible and irreversible reactions, non-covalentinteractions between hydrophobic domains or the formation

C© 2009 Portland Press Ltd

Page 6: Strategies for analysing and improving the expression and quality of recombinant proteins made in mammalian cells

78 N. Jenkins and others

Table 2 Genetic manipulation studies on industrial cell lines

Protein Mechanism of action Effect on productivity Notes Reference(s)

PDI Formation and breakage ofdisulfide bonds

Increased MAb production inCHO cells

Did not increase thrombopoietinproduction

[38,45]

BiP ER-based protein chaperone Overexpression delayed thyroglobulinsecretion in CHO cells; increasedMAb production in insect cells

Not useful for production [104,105]

Calnexin andcalreticulin

Chaperones that detect glycosylationstatus

Increased thrombopoietin production inCHO cells

– [46]

Activating transcriptionfactor 4

Transcription factor involved inUPR function

Increased antithrombin III production inCHO cells

– [106]

Human XBP-1S Gene activator of multiple ER proteins Increased MAb production in CHO cells Comparable glycosylation patterns [56]

Mouse XBP-1S – Increased EPO production in CHO andNS0 cells

– [54]

of intermolecular disulfide bonds [58]. Some aggregatesare insoluble, whereas others remain in solution. Fornon-vaccine biotherapeutics, all types of aggregates areconsidered undesirable, since small soluble aggregates maybecome immunogenic [59], resulting in inhibitory antibodiesto the therapeutic protein [51]. This can be a particularproblem with multimeric proteins such as recombinantIgG, Fc-fusion proteins and blood-clotting proteins such asFactor VIII [50]. Larger particulates may cause problemsat the site of administration [60]. Although there are USPharmacopeia guidelines for the number of particles of size�10 μm and � 25μm that are acceptable in pharmaceuticalpreparations, the permissible levels of soluble aggregates,e.g. dimers and HMM soluble aggregates are ill-defined.

The first steps in protein aggregation typically arisefrom weak non-covalent protein interactions. Exposureof hydrophobic surfaces in partially denatured proteinscan lead to non-covalent aggregates, and these can beprecursors to the formation of covalent aggregates. Thereis an equilibrium between the monomers and higher-orderforms (e.g. dimers and tetramers) that may shift as a resultof a change in conditions, such as protein concentration orpH [60]. A further step in aggregate formation is disulfidebonding between unpaired thiols. These covalently bondedaggregates are very difficult to disrupt and, once formed,are often discarded during the protein purification processthrough chromatography or filtration.

One way of reducing the free thiol content in recom-binant CHO cells is to add low amounts (up to 100 μM) ofthe oxidizing agent copper sulfate. This results in a 10-foldreduction of percentage of free thiols in IgG, and significant(3-fold) reductions were observed with as little as 5 μMcopper sulfate [61]. Other cell-culture-media componentscan also affect the level of total aggregates and the distribu-

tion between non-covalent and disulfide-bonded aggregates[62]. Evidence has been gained that protein aggregates cancontinue to form in the supernatant after cells producingrecombinant IgG have been harvested [60], and thisnecessitates cooling or other controlled hold steps betweenbioreactor harvest and the first chromatographic capturestep.

The aggregation problem is not confined to upstreambioprocessing. For example, the techniques used to inactiv-ate viruses during downstream processing, such as exposureto detergents or extremes of pH, can inadvertently damagethe protein product and cause it to aggregate [63].Low pH conditions (pH 2–4) are also commonly used toelute antibodies from Protein A capture columns duringpurification. Downstream intermediate and polishing stepstypically include ion-exchange chromatography, which elutesthe protein with high-ionic-strength solutions or under high-pH conditions which can damage the product. Multiplefiltration steps are also used in protein purification forconcentration, buffer exchange and virus removal. Largeprotein aggregates can cause membrane fouling, and the highpressures employed may also increase aggregation duringthese process steps. Stresses to proteins such as freezing,thawing, freeze-drying (lyophilization), prolonged exposureto air, light, or interactions with metal surfaces, may resultin surface denaturation, which then leads to the formationof aggregates [1].

Excipients such as sugars [64] and arginine [65] areoften used to suppress aggregate formation during proteinpurification and formulation [66]. However, there areexamples where the drug product forms aggregates withits excipients. For example, a formulation of interferon-αbecame oxidized at room temperature and formedaggregates with the excipient human serum albumin;

C© 2009 Portland Press Ltd

Page 7: Strategies for analysing and improving the expression and quality of recombinant proteins made in mammalian cells

Protein expression and quality in mammalian cells 79

Table 3 Effects of environmental conditions on protein modifications

Environmental conditions Main effect Context and other effects Reference

Temperature, protein concentration,ionic strength and pH

All influence aggregate levels Can occur at all phases of theproduction cycle

[107]

Electrostatic interactions Increase viscosity and aggregation Can occur between monomers and withsurfaces of containers

[108]

Mechanical stress Increases aggregation Impellar speed in bioreactor, rate ofpumping and filtration

[109]

Oxidative intracellular environment Decreases aggregate formation Bioreactor conditions [40]

Medium components Influence distribution betweenmonomers, dimers and HMM species

Bioreactor conditions [62]

Copper sulfate Decreases aggregation Toxic to cells in millimolar doses [61]

Time in bioreactor Increases aggregation Affects duration of bioreactor run [60]

Freeze–thaw cycles and freeze-drying(lyophilization)

Increases aggregation Can occur at process hold stepsand in formulation

[58]

Viral inactivation and low pH elutionfrom Protein A

Increases aggregation Minimize time the protein is exposed [110]

Excipients, e.g. arginine and sugars Decrease aggregation Used in formulation and somepurification steps

[64]

this induced an immune response to interferon-α, andchanging to a liquid albumin-free formulation stored at 4 ◦Creduced the immune reaction [67]. Table 3 summarizes theenvironmental effects on protein modifications.

Several analytical techniques exist to detect proteinaggregates (see Table 4), but all have both their strengthsand their weaknesses. The standard [and FDA (U.S.Food and Drug Administration-)approved] method hasbeen SEC (size-exclusion chromatography) performed byHPLC [68], where aggregated species are separated by sizefrom monomers in a porous gel matrix. SEC–HPLC permitsrelatively rapid and cost-effective analysis of aggregatecontent, but is hampered by non-specific interactions withthe column matrix [69]. Adding excipients such as arginineto the SEC elution buffer reduces non-specific interactionsand permits a more accurate measurement of HMMspecies. Arginine suppresses protein aggregation throughinteractions between the guanidinium group of arginine andtryptophan side chains [70]. Including arginine in the elutionbuffer in SEC can provide a more accurate estimation oftotal aggregate content [71]. However, the largest solubleHMM species may not interact with the column materialat all and be of sufficiently low concentration to escapedetection. A tandem combination of SEC–HPLC and MALLS(multi-angle laser light scattering) facilitates the detectionof larger HMM species [72].

Other modes of chromatography such as HIC(hydrophobic-interaction chromatography) [73] and IEX(ion-exchange chromatography) [60] may be more useful in

DSP (downstream processing) at separating aggregates frommonomers. However, both HIC and IEX typically use highsalt concentrations in the mobile phase, and this may alterthe nature or concentration of aggregates. Other, column-free, methods have been used to detect aggregates suchas AUC (analytical ultracentrifugation) and FFF (field flowfractionation). AUC separates molecules on the basis oftheir their different sedimentation coefficients, which them-selves derive from the molecule’s size and molecular mass[74]. AUC is arguably the optimal means of analysing HMMdistribution [68] and is an excellent choice for orthogonalconfirmation of results obtained using SEC. Unfortunately,AUC analysis requires a specialized centrifuge, complex dataanalysis and is low-throughput in nature. An emerging altern-ative is FFF, which offers similar benefits to AUC with fasterrun times and simpler post-run analysis, and it can be coupledto MALLS. However, FFF needs considerable optimization toensure reproducible results [75] and involves sampledilution. PAGE performed under native conditions canalso demonstrate aggregation [76,77]. This is inexpensiveand can deliver nanogram detection limits with sensitivemethods involving staining with, for example, silver (0.2 ngof protein/lane) [78] or Coomassie Brilliant Blue G-250 [79],but they are seriously constrained by size limits, a factorthat restricts their usefulness for many biopharmaceuticals.Other problems are the need to exclude SDS and reducingagents, which increase gel running times and prevent analysisof proteins without a net negative charge under nativeconditions.

C© 2009 Portland Press Ltd

Page 8: Strategies for analysing and improving the expression and quality of recombinant proteins made in mammalian cells

80 N. Jenkins and others

Table 4 Analytical methods used to detect protein quality changes

Method Principle Advantages Limitations Reference(s)

SEC Separations based onhydrodynamic radius

FDA-approved; simple and robust Can underestimate HMMaggregates; sample is diluted

[68,69,111]

HIC Adsorption of hydrophobicprotein residues underhigh-salt conditions

Concentrates sample morethan SEC

High-salt conditions;can be corrosive

[73]

AUC Separates molecules based onsedimentation coefficients

Minimal dilution effects; goodorthogonal method to SEC

Time-consuming; complexanalysis

[68,74,112]

FFF Separates species based ondiffusion coefficients

Broad dynamic range,column-free

Requires extensive optimizationand low throughput

[75]

DLS and MALLS Uses scattered light to measurethe rate of diffusion orgeometric size of proteinparticles

Avoids dilution effects; gooddetection of HMM aggregates

Cannot resolve LMW species [72,111,113]

Liquid chromatography–MS MS used on proteolytic digests Comprehensive characterizationof protein primary structureand PTMs

Low throughput; expensiveequipment

[99]

An asset of optics-based detection technologies such asDLS (dynamic light scattering) is the ability to resolve largerHMM species that may escape detection by conventionalSEC [80]. Because the signal increases with increasingmolecular size, DLS is ideal for detection of HMM species atthe upper scale of >1000 kDa. The unfortunate converse ofthis is that the resolution of DLS suffers below this threshold,and it is unable to resolve monomers from dimers. Thereis a requirement for more rapid assessment methods foraggregation that could be used in process development (forexample in clone selection, setting bioreactor parameters,chromatography column and filter selection). Dye-basedassays are already used in process analytics, e.g. Ellman’s re-agent [5,5′-dithiobis-(2-nitrobenzoic acid)] [58] is routinelyused to determine total free thiol content in productbatches. Dye-based analysis is also compatible with the 96-or 384-well microtitre platforms that lend themselves toautomated liquid handling and spectrophotometric read-outs [82]. However, their use in routinely detectingbiopharmaceutical aggregates has yet to be established.

Asparagine deamidationDeamidation is reported to be the most common PTMfound in proteins [83] and involves the non-enzymic conver-sion of asparagine residues into cyclic imide intermediates,which rapidly hydrolyse to form a mixture of isoasparticand aspartic acid at a ratio of 3:1 [84]. This PTM can ariseduring every stage of biopharmaceutical production from cellculture though to formulation [85] and can even occur post-administration in vivo [86]. Asparagine residues at the proteinsurface appear more vulnerable, and the micro-environment

is known to influence the rate and extent of deamidation[87]. Also, the presence of a carboxy glycine greatlyenhances the rate of deamidation. Computer modellinghas been used to predict deamidation [88]. In one casedeamidation was found to cause a 70% loss of potency of amarketed MAb [89], and it is known to promote aggregationin certain proteins [90] and enhance the autoimmuneresponse [91].

Deamidated proteins have been identified by means ofa variety of chromatographic techniques, including HIC [92]and IEX [93], but are most comprehensively characterizedthrough peptide mapping of proteolytic digests, wheredeamidation can be tracked to a specific asparagine residue[94]. Promega’s ISOQUANTTM kit exploits the ability of theenzyme PIMT (protein-L-isoaspartyl O-methyltransferase)to transfer the active methyl group of SAM (S-adenosyl-L-methionine) on to the free α-carboxy group of isoaspartateto form an O-methyl ester. The isoaspartyl methyl esterrapidly breaks down at neutral pH to form the cyclic imide,with concomitant release of methanol.

Methionine oxidationMethionine oxidation is another common PTM that isminimized within the cell by the methionine sulfoxidereductase pathway [95]. Once secreted, vulnerablemethionine residues can be oxidized (particularlyin protein-free or low protein culture media), butformulation excipients can help protect the residues[66]. Methionine oxidation has been reported in MAbsduring bioprocessing [96] and after long-term storage.Methionine oxidation in the Fc portion of IgG1 adversely

C© 2009 Portland Press Ltd

Page 9: Strategies for analysing and improving the expression and quality of recombinant proteins made in mammalian cells

Protein expression and quality in mammalian cells 81

affects both MAb structure and stability [97] andcan decrease its binding to Protein A- and ProteinG-based resins [98] that are commonly used in capture DSP.No rapid method for methionine oxidation detection exists,but it can be detected by peptide mapping with MS detection[99], reversed-phase HPLC, HIC and weak cation-exchangechromatography [100]. Currently, only MS can determinewhich methionine residue within the protein is oxidized.

Conclusions

Much progress has been made in recent years, througha combination of analytical and cellular techniques, inunderstanding the bottlenecks to protein secretion andmodification. However, many challenges still exist to achieveconsistently high yields in biopharmaceutical productionwhile avoiding post-translational modifications that maydamage protein stability, efficacy or shelf-life.

References

1 Jenkins, N. (2007) Cytotechnology 53, 121–1252 Chirino, A. J. and Mire-Sluis, A. (2004) Nat. Biotechnol. 22,

1383–13913 Marth, J. D. and Grewal, P. K. (2008) Nat. Rev. Immunol. 8,

874–8874 Sola, R. J. and Griebenow, K. (2009) J. Pharm. Sci. 98,

1223–12455 Lairson, L. L., Henrissat, B., Davies, G. J. and Withers, S. G.

(2008) Annu. Rev. Biochem. 77, 521–5556 Charaniya, S., Karypis, G. and Hu, W. S. (2009) Biotechnol.

Bioeng. 102, 1654–16697 Yee, J. C., Gerdtzen, Z. P. and Hu, W. S. (2009) Biotechnol.

Bioeng. 102, 246–2638 Doolan, P., Melville, M., Gammell, P., Sinacore, M., Meleady, P.,

McCarthy, K., Francullo, L., Leonard, M., Charlebois, T. andClynes, M. (2008) Mol. Biotechnol. 39, 187–199

9 Gorg, A., Weiss, W. and Dunn, W. J. (2004) Proteomics 4,3665–3685

10 Unlu, M., Morgan, M. E. and Minden, J. S. (1997)Electrophoresis 18, 2071–2077

11 Mann, M., Hendrickson, R. C. and Pandey, A. (2001) Ann. Rev.Biochem. 70, 437–473

12 Gygi, S. P., Rist, B., Gerber, S. A., Turecek, F., Gelb, M. H. andAebersold, R. (1999) Nat. Biotechnol. 17, 994–999

13 Ross, P. L., Huang, Y. L. N., Marchese, J. N., Williamson,B., Parker, K., Hattan, S., Khainovski, N., Pillai, S., Dey, S.,Daniels, S., Purkayastha, S. et al. (2004) Mol. Cell. Proteomics3, 1154–1169

14 Dreisbach, A., Otto, A., Becher, D., Hammer, E., Teumer, A.,Gouw, J. W., Hecker, M. and Volker, U. (2008) Proteomics 8,2062–2076

15 Kruger, M., Kratchmarova, I., Blagoev, B., Tseng, Y. H., Kahn,C. R. and Mann, M. (2008) Proc. Natl. Acad. Sci. U.S.A. 105,2451–2456

16 Kislinger, T., Gramolini, A. O., MacLennan, D. H. and Emili, A.(2005) J. Am. Soc. Mass Spectrom. 16, 1207–1220

17 Poon, T. C. W. (2007) Expert Rev. Proteomics 4, 51–6518 Reid, J. D., Parker, C. E. and Borchers, C. H. (2007) Curr. Opin.

Mol. Ther. 9, 216–22119 Kumar, N., Maurya, P., Ganunell, P., Dowling, P., Clynes, M. and

Meleady, P. (2008) Biotechnol. Prog. 24, 273–27820 Nissom, P. M., Sanny, A., Kok, Y. J., Hiang, Y. T., Chuah, S. H.,

Shing, T. K., Lee, Y. Y., Wong, K. T. K., Hu, W. S., Sim, M. Y. G.and Philp, R. (2006) Mol. Biotechnol. 34, 125–140

21 Meleady, P., Henry, M., Gammell, P., Doolan, P., Sinacore, M.,Melville, M., Francullo, L., Leonard, M., Charlebois, T. andClynes, M. (2008) Proteomics 8, 2611–2624

22 Smales, C. M., Dinnis, D. M., Stansfield, S. H., Alete, D., Sage,E. A., Birch, J. R., Racher, A. J., Marshall, C. T. and James, D. C.(2004) Biotechnol. Bioeng. 88, 474–488

23 Alete, D. E., Racher, A. J., Birch, J. R., Stansfield, S. H., James,D. C. and Smales, C. M. (2005) Proteomics 5, 4689–4704

24 Dinnis, D. M., Stansfield, S. H., Schlatter, S., Smales, C. M.,Alete, D., Birch, J. R., Racher, A. J., Marshall, C. T., Nielsen, L. K.and James, D. C. (2006) Biotechnol. Bioeng. 94, 830–841

25 Seth, G., Philp, R. J., Lau, A., Jiun, K. Y., Yap, M. and Hu, W. S.(2007) Biotechnol. Bioeng. 97, 933–951

26 Palermo, D. P., Degraaf, M. E., Marotti, K. R., Rehberg, E. andPost, L. E. (1991) J. Biotechnol. 19, 35–47

27 Van Dyk, D. D., Misztal, D. R., Wilkins, M. R., Mackintosh, J. A.,Poljak, A., Varnail, J. C., Teber, E., Walsh, B. J. and Gray, P. P.(2003) Proteomics 3, 147–156

28 Baik, J. Y., Joo, E. J., Kim, Y. H. and Lee, G. M. (2008) J. Biotechnol.133, 461–468

29 Yee, J. C., Gatti, M. D., Philp, R. J., Yap, M. and Hu, W. S. (2008)Biotechnol. Bioengin. 99, 1186–1204

30 Liu, C. H., Chu, I. M. and Hwang, S. M. (2001) Biotechnol. Lett.23, 1641–1645

31 Li, J. H., Huang, Z., Sun, X. M., Yang, P. Y. and Zhang, Y. X.(2006) Enzyme Microb. Technol. 38, 372–380

32 Kim, T. K., Ryu, J. S., Chung, J. Y., Kim, M. S. and Lee, G. M.(2000) Biotechnol. Prog. 16, 775–781

33 Lee, M. S., Kim, K. W., Kim, Y. H. and Lee, G. M. (2003)Biotechnol. Prog. 19, 1734–1741

34 Yoon, S. K., Song, J. Y. and Lee, G. M. (2003) Biotechnol. Bioeng.82, 289–298

35 Fox, S. R., Patel, U. A., Yap, M. G. S. and Wang, D. I. C. (2004)Biotechnol. Bioeng. 85, 177–184

36 Kaufmann, H., Mazur, X., Fussenegger, M. and Bailey, J. E. (1999)Biotechnol. Bioengin. 63, 573–582

37 Baik, J. Y., Lee, M. S., An, S. R., Yoon, S. K., Joo, E. J., Kim, Y. H.,Park, H. W. and Lee, G. M. (2006) Biotechnol. Bioeng 93,361–371

C© 2009 Portland Press Ltd

Page 10: Strategies for analysing and improving the expression and quality of recombinant proteins made in mammalian cells

82 N. Jenkins and others

38 Mohan, C., Park, S. H., Chung, J. Y. and Lee, G. M. (2007)Biotechnol. Bioeng. 98, 611–615

39 Wilkinson, B. and Gilbert, H. F. (2004) Biochim. Biophys. Acta1699, 35–44

40 Chakravarthi, S., Jessop, C. E. and Bulleid, N. J. (2006) EMBORep. 7, 271–275

41 Harris, R. J. (2005) Dev. Biol. (Basel) 122, 117–12742 Chakravarthi, S. and Bulleid, N. J. (2004) J. Biol. Chem. 279,

39872–3987943 Ailor, E. and Betenbaugh, M. J. (1998) Biotechnol. Bioeng. 58,

196–20344 Schroder, M. (2008) Biotechnol Lett. 30, 187–19645 Borth, N., Mattanovich, D., Kunert, R. and Katinger, H. (2005)

Biotechnol. Prog. 21, 106–11146 Chung, J. Y., Lim, S. W., Hong, Y. J., Hwang, S. O. and Lee, G. M.

(2004) Biotechnol. Bioeng. 85, 539–54647 Cudna, R. E. and Dickson, A. J. (2003) Biotechnol. Bioeng. 81,

56–6548 Schroder, M., Schafer, R. and Friedl, P. (2002) Biotechnol.

Bioeng. 78, 131–14049 Demeule, B., Lawrence, M. J., Drake, A. F., Gurny, R. and

Arvinte, T. (2007) Biochim. Biophys. Acta 1774, 146–15350 Purohit, V. S., Middaugh, C. R. and Balasubramanian, S. V.

(2006) J. Pharm. Sci. 95, 358–37151 Hermeling, S., Crommelin, D. J., Schellekens, H. and Jiskoot, W.

(2004) Pharm. Res. 21, 897–90352 Brewer, J. W. and Hendershot, L. M. (2005) Nat. Immunol. 6,

23–2953 Mori, K. (2003) Traffic 4, 519–52854 Ku, S. C., Ng, D. T., Yap, M. G. and Chao, S. H. (2008)

Biotechnol. Bioeng. 99, 155–16455 Lee, A. H., Iwakoshi, N. N. and Glimcher, L. H. (2003)

Mol. Cell. Biol. 23, 7448–745956 Becker, E., Florin, L., Pfizenmaier, K. and Kaufmann, H. (2008)

J. Biotechnol. 135, 217–22357 Peng, R. W. and Fussenegger, M. (2009) Biotechnol. Bioeng.

102, 1170–118158 Andya, J. D., Hsu, C. C. and Shire, S. J. (2003) AAPS PharmSci.

5, E1059 Rosenberg, A. S. (2006) AAPS J. 8, E501–E50760 Cromwell, M. E. and E. Hilario, F. Jacobson (2006) AAPS J. 8,

E572–E57961 Chaderjian, W. B., Chin, E. T., Harris, R. J. and Etcheverry, T. M.

(2005) Biotechnol. Prog. 21, 550–55362 Jenkins, N., Murphy, L. and Tyther, R. (2008) Mol. Biotechnol.

39, 113–11863 Lin, J. J., Meyer, J. D., Carpenter, J. F. and Manning, M. C. (2000)

Pharm. Res. 17, 391–39664 Cleland, J. L., Lam, X., Kendrick, B., Yang, J., Yang, T. H.,

Overcashier, D., Brooks, D., Hsu, C. and Carpenter, J. F. (2001)J. Pharm. Sci. 90, 310–321

65 Tsumoto, K., Ejima, D., Kita, Y. and Arakawa, T. (2005) ProteinPept. Lett. 12, 613–619

66 Soenderkaer, S., Carpenter, J. F., van de Weert, M., Hansen,L. L., Flink, J. and Frokjaer, S. (2004) Eur. J. Pharm. Sci. 21,597–606

67 Ryff, J. C. (1997) J. Interferon Cytokine Res. 17 (Suppl. 1),S29–S33

68 Gabrielson, J. P., Brader, M. L., Pekar, A. H., Mathis, K. B.,Winter, G., Carpenter, J. F. and Randolph, T. W. (2007)J. Pharm. Sci. 96, 268–279

69 Stulik, K., Pacakova, V. and Ticha, M. (2003) J. Biochem. Biophys.Methods 56, 1–13

70 Tsumoto, K., Umetsu, M., Kumagai, I., Ejima, D., Philo, J. S. andArakawa, T. (2004) Biotechnol. Prog. 20, 1301–1308

71 Ejima, D., Yumioka, R., Arakawa, T. and Tsumoto, K. (2005)J. Chromatogr. A 1094, 49–55

72 Ahrer, K., Buchacher, A., Iberer, G. and Jungbauer, A. (2004)J. Chromatogr. A 1043, 41–46

73 McCue, J. T., Engel, P., Ng, A., Macniven, R. and Thommes, J.(2008) Bioprocess Biosyst. Eng. 31, 261–275

74 Gabrielson, J. P., Arthur, K. K., Kendrick, B. S., Randolph, T. W.and Stoner, M. R. (2008) J. Pharm. Sci. 18, 18

75 Silveira, J. R., Hughson, A. G. and Caughey, B. (2006) MethodsEnzymol. 412, 21–33

76 Arakawa, T., Philo, J. S. and Kita, Y. (2001) Biosci. Biotechnol.Biochem. 65, 1321–1327

77 Narhi, L. O., Arakawa, T., Aoki, K. H., Elmore, R., Rohde, M. F.,Boone, T. and Strickland, T. W. (1991) J. Biol. Chem. 266,23022–23026

78 Jin, L. T., Li, X. K., Cong, W. T., Hwang, S. Y. and Choi, J. K.(2008) Anal. Biochem. 383, 137–143

79 Neuhoff, V., Arold, N., Taube, D. and Ehrhardt, W. (1988)Electrophoresis 9, 255–262

80 Mahler, H. C., Muller, R., Friess, W., Delille, A. and Matheus, S.(2005) Eur. J. Pharm. Biopharm. 59, 407–417

81 Reference deleted82 Jordan, M. and Jenkins, N. (2007) In Animal Cell Biotechnology,

Methods and Protocols (Portner, R., ed.), pp. 193–202,Humana Press, Totowa

83 Cleland, J. L., Powell, M. F. and Shire, S. J. (1993) Crit. Rev. Ther.Drug Carrier Syst. 10, 307–377

84 Gestwicki, J. E., Crabtree, G. R. and Graef, I. A. (2004) Science306, 865–869

85 Liu, H., Gaza-Bulseco, G., Faldu, D., Chumsae, C. and Sun, J.(2008) J. Pharm. Sci. 97, 2426–2447

86 Huang, L., Lu, J., Wroblewski, V. J., Beals, J. M. and Riggin, R. M.(2005) Anal. Chem. 77, 1432–1439

87 Weintraub, S. J. and Deverman, B. E. (2007) Sci. STKE 2007,re7

88 Catak, S., Monard, G., Aviyente, V. and Ruiz-Lopez, M. F. (2008)J. Phys. Chem. A 112, 8752–8761

89 Harris, R. J., Kabakoff, B., Macchi, F. D., Shen, F. J., Kwong, M.,Andya, J. D., Shire, S. J., Bjork, N., Totpal, K. and Chen, A. B.(2001) J. Chromatogr. B Biomed. Sci. Appl. 752, 233–245

C© 2009 Portland Press Ltd

Page 11: Strategies for analysing and improving the expression and quality of recombinant proteins made in mammalian cells

Protein expression and quality in mammalian cells 83

90 Takata, T., Oxford, J. T., Demeler, B. and Lampi, K. J. (2008)Protein Sci. 17, 1565–1575

91 Falini, M. L., Elli, L., Caramanico, R., Bardella, M. T., Terrani, C.,Roncoroni, L., Doneda, L. and Forlani, F. (2008) Dig. Dis. Sci.53, 2697–2701

92 Zhang, Y., Martinez, T., Woodruff, B., Goetze, A., Bailey, R.,Pettit, D. and Balland, A. (2008) Anal. Chem. 80, 7022–7028

93 Teshima, G., Porter, J., Yim, K., Ling, V. and Guzzetta, A. (1991)Biochemistry 30, 3916–3922

94 Chelius, D., Rehder, D. S. and Bondarenko, P. V. (2005) Anal.Chem. 77, 6004–6011

95 Moskovitz, J. (2005) Biochim. Biophys. Acta 1703,213–219

96 Matamoros Fernandez, L. E., Kalume, D. E., Calvo, L., FernandezMallo, M., Vallin, A. and Roepstorff, P. (2001) J. Chromatogr. BBiomed. Sci. Appl. 752, 247–261

97 Liu, D., Ren, D., Huang, H., Dankberg, J., Rosenfeld, R., Cocco,M. J., Li, L., Brems, D. N. and Remmele, Jr, R. L. (2008)Biochemistry 47, 5088–5100

98 Gaza-Bulseco, G., Faldu, S., Hurkmans, K., Chumsae, C. andand Liu, H. (2008) J. Chromatogr. B Anal. Technol. Biomed.Life Sci. 870, 55–62

99 Houde, D., Kauppinen, P., Mhatre, R. and Lyubarskaya, Y. (2006)J. Chromatogr. A 1123, 189–198

100 Chumsae, C., Gaza-Bulseco, G., Sun, J. and Liu, H. (2007)J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 850,285–294

101 Reference deleted

102 Stansfield, S. H., Allen, E. E., Dinnis, D. M., Racher, A. J.,Birch, J. R. and James, D. C. (2007) Biotechnol. Bioeng. 97,410–424

103 Reference deleted104 Muresan, Z. and Arvan, P. (1998) Mol. Endocrinol. 12, 458–467105 Hsu, T. A. and Betenbaugh, M. J. (1997) Biotechnol. Prog. 13,

96–104106 Ohya, T., Hayashi, T., Kiyama, E., Nishii, H., Miki, H.,

Kobayashi, K., Honda, K., Omasa, T. and Ohtake, H. (2008)Biotechnol. Bioeng. 100, 317–324

107 Chi, E. Y., Krishnan, S., Randolph, T. W. and Carpenter, J. F.(2003) Pharm. Res. 20, 1325–1336

108 Zhang, Y. B., Howitt, J., McCorkle, S., Lawrence, P.,Springer, K. and Freimuth, P. (2004) Protein Expression Purif.36, 207–216

109 Mahler, H. C., Muller, R., Friess, W., Delille, A. and Matheus, S.(2005) Eur. J. Pharm. Biopharm. 59, 407–417

110 Smales, C. M., Pepper, D. S. and James, D. C. (2001) Biotechnol.Prog. 17, 974–978

111 Ahrer, K., Buchacher, A., Iberer, G., Josic, D. and Jungbauer, A.(2003) J. Chromatogr. A 1009, 89–96

112 Gabrielson, J. P., Randolph, T. W., Kendrick, B. S. and Stoner,M. R. (2007) Anal. Biochem. 361, 24–30

113 van Dijk, J. A. and Smit, J. A. (2000) J. Chromatogr. A 867,105–112

Received 17 November 2008/13 February 2009; accepted 19 February 2009Published on the Internet 6 May 2009, doi:10.1042/BA20080258

C© 2009 Portland Press Ltd