Metabolic flux control in glycosylation

Available online at www.sciencedirect.com

ScienceDirect Metabolic flux control in glycosylation Andrew G McDonald, Jerrard M Hayes and Gavin P Davey Glycosylation is a common post-translational protein modification, in which glycans are built onto proteins through the sequential addition of monosaccharide units, in reactions catalysed by glycosyltransferases. Glycosylation influences the physicochemical and biological properties of proteins, with subsequent effects on subcellular and extracellular protein trafficking, cell–cell recognition, and ligand–receptor interactions. Glycan structures can be complex, as is the regulation of their biosynthesis, and it is only recently that the systems biology of metabolic flux control and glycosyltransferase networks has become a study in its own right. We review various models of glycosylation that have been proposed to date, based on current knowledge of Golgi structure and function, and consider how metabolic flux through glycosyltransferase networks regulates glycosylation events in the cell. Address School of Biochemistry and Immunology, Trinity College Dublin, Dublin 2, Ireland Corresponding author: Davey, Gavin P ([email protected])

Current Opinion in Structural Biology 2016, 40:97–103 This review comes from a themed issue on Carbohydrate–protein interactions and glycosylation Edited by Nagasuma Chandra and Harry Gilbert

http://dx.doi.org/10.1016/j.sbi.2016.08.007 0959-440/# 2016 Elsevier Ltd. All rights reserved.

Introduction Glycosylation is a major cellular activity in which oligosaccharides are covalently attached to proteins and lipids to form glycoconjugates. The type of glycosylation can vary, from a single monosaccharide to the formation of complex, highly branched structures known as glycans. The biosynthesis of glycans involves a large family of enzymes known as glycosyltransferases, which transfer the monosaccharide portion of a nucleotide sugar to its glycoprotein or glycolipid acceptor. In the case of protein, glycans typically fall into two major classes: N-linked, in which the carbohydrate is linked via the terminal (amide) nitrogen of the side chain of an asparagine residue; and Olinked, in which the attachment point is the side-chain oxygen of serine or threonine. The carbohydrate content of a glycoprotein can contribute significantly to its overall www.sciencedirect.com

mass, and alter its biophysical properties. Glycosylation is a key determinant of many biological processes, such as intracellular and extracellular trafficking, cellular differentiation and development. In disease states, such as cancer, where alterations to the cellular glycosylation profile are present, a deeper knowledge of metabolic flux and how it regulates glycosyltransferase activity is needed. Improved control of glycosylation networks during bioprocessing is also fundamental to the development of biopharmaceuticals. Of increasing relevance is the use of mathematical modelling for understanding and prediction of glycoform heterogeneity, leading to new insights and to new directions for experimental work. We review some of the systems-biological models that have been developed, focusing in particular on the concept of fluxbased competition amongst the glycosyltransferases, as well as among the substrates common to a given enzyme.

Enzymes of glycosylation and their substrates This article focuses on the most commonly encountered N-linked and O-linked glycans and the enzymes responsible for their synthesis. N-linked glycosylation starts on the cytoplasmic side of the endoplasmic reticulum through the transfer of a dolichol phosphate-linked glycan consisting of two N-acetylglucosamine, nine mannose and three glucose residues (Glc3Man9GlcNAc2) to a vacant glycosylation site on a protein catalysed by oligosaccharide transferase (OST). As the protein translocates through the ER/Golgi network, a combination of membrane-bound glycosyltransferases and glycosidases act thereafter to remodel the N-glycan structure to form a wide range of possible glycoforms. The different types of glycosylation are shown in Figure 1. The consensus sequence (sequon) for the initiation of asparagine-linked glycosylation is the Asn-X-Ser/Thr motif, where X is not proline, whereas for O-linked glycosylation the sequon is an unoccupied side chain of a Ser/ Thr residue. O-linked glycans of mucin type are formed initially by the addition of a N-acetylgalactosamine (GalNAc), catalysed by a family of 20 GalNAc-transferases distributed throughout the Golgi [1]. Whereas all N-glycans have in common a single, trimannosyl pentasaccharide core structure, O-glycans are known to form as many as eight different core structures [2]. With the exception of glucosidases and mannosidases, which hydrolyse single glucose and mannose residues, respectively, from the non-reducing end of a glycan, the other enzymes of glycosylation are transferases, catalysing the general reaction: Ax þ B ¼ A þ Bx Current Opinion in Structural Biology 2016, 40:97–103

98 Carbohydrate–protein interactions and glycosylation

Figure 1

Golgi

1

O-linked

N-linked

2 GalT

3 UDP-Gal

Factors affecting regulation of glycosylation: 1. Polypeptide folded state 2. Glycosyltransferase enzyme concentration 3. Donor concentration 4. Nucleotide-sugar transporter (NST) 5. Nucleoside diphosphatase (NDPase)

UDP

UMP 4

NDPase

5

NST

UMP UDP-Gal Nucleotide-sugar metabolism Current Opinion in Structural Biology

Types of glycosylation and some key points of regulation. Examples are shown of N-linked and O-linked glycan, the latter including O-GalNAc, Omannose and O-fucose variants. The major regulation points include availability of substrate, enzyme activity, and transport. Additional regulation points include localisation of enzymes within the Golgi, and their co-association.

where Ax is a nucleotide sugar donor (NSD), B is a proteoglycan acceptor, and x is the monosaccharyl unit transferred. Donor substrates are transported into the Golgi by specific transport proteins, as illustrated in Figure 1 for UDP-galactose (UDP-Gal). In addition to the glycosidases involved in N-glycan processing, GlcNAc-transferases (GnTs), GalNAc-transferases (GalNAcTs), galactosyltransferases (GalTs), sialytransferases (SiaTs) and fucosyltransferases (FucTs) are localised to separate regions of the Golgi. Glycans can be modified further by phosphorylation and sulfation through the action of kinases, sulfotransferases and O-acetyltransferases. There is also evidence of homomeric and heteromeric associations of the glycan processing enzymes [3], which may influence their kinetic properties and their localisation.

Models of glycosylation A comparative study of the major systems glycobiology models in relation to glycoengineeering has recently appeared [4]. Here we give a brief account of the models Current Opinion in Structural Biology 2016, 40:97–103

of glycosylation that have been proposed. The earliest theoretical study, by Shelikoff and co-workers [5] studied the initiation of N-linked glycosylation and modelled the co-translational attachment of oligosaccharides to glycosylation sites. An important aspect of this initiation process, that of sequon occupancy, has been receiving increased attention of late due to the difficulty in predicting the initiation of protein glycosylation from a knowledge of the primary sequence alone, although some patterns are discernable in the case of N-glycans [6]. The location and decision to glycosylate a protein at a particular site is likely to be influenced by many factors, including the presence of neighbouring glycosylation sites. Murray et al. have shown that any aromatic amino acid at position n-2 relative to the Asn-X-Thr/Ser sequon results in glycoproteins possessing N-glycans of lower complexity [7]. A kinetic model of mucin Core 1 formation was developed by Gerken in 2004 to explain the influence of neighbouring glycosylation sites on O-glycan formation [8]. Currently, machine-learning methods are most commonly used to predict the location of glycan www.sciencedirect.com

Metabolic flux control in glycosylation McDonald, Hayes and Davey 99

attachments [9–12]. A model by Monica et al., focusing on sialyltransferase activity in the trans-Golgi network was developed to describe the compartmentalized sialylation of N-linked glycans and predict the outcome of sialylation reactions [13]. The first kinetic model of the core pathways of N-glycosylation followed and was proposed by Uman˜a and Bailey [14]. The Uman˜a–Bailey model was subsequently extended by Krambeck and Betenbaugh to account for the complex glycans expressed by wild-type Chinese hamster ovary (CHO) cells [15]. This model was able to predict the levels of the more commonly occurring glycans, although it did not account for the high-mass oligosaccharides in CHO-cell mutants that were reported by North et al. [16]. More recently, further extensions and developments to the N-glycosylation model have appeared, which are able to predict N-glycan structures and enzyme activities from mass-spectrometric data [17,18] and mRNA-expression arrays [18], a significant achievement that demonstrates the utility of a systems-biological approach. Hossler et al. [19] used a mathematical model of glycan biosynthesis based on the effect of reaction variables on glycan distribution and heterogeneity to compare the two competing views of Golgi processing, the cisternal maturation view versus that of vesicular transport. The modelling results and comparison to experimentally generated glycan profiles support the view that the Golgi maturation model is likely to be a valid mechanism. The original Krambeck and Betenbaugh model was further used as the theoretical basis for a study of N-glycan antennarity in CHO cells, through a combination of in silico and in vivo variation of enzyme expression levels. In this study a galactosyltransferase enzyme (b-1,4-galactosyltransferase 4) was found to be an important regulator of tri-antennary and tetra-antennary glycoforms [20,21]. Whereas the majority of kinetic studies have treated the ER/Golgi as a four-compartment system, a detailed study by del Val and co-workers [22], treated the Golgi network as a continuum, and in addition incorporated the kinetics of nucleotide-sugar donor transporters (NSTs). A perennial problem of quantitative modelling based on enzymekinetic rate laws is the absence of reliable estimates of the kinetic parameters, especially in glycosylation where there are a large number of potential acceptors. Spahn et al. employed a Markov-chain model of an N-glycosylation reaction network to minimise the number of parameters required, and made predictions of cellular behaviour that were subsequently validated against glycoanalysis of secreted glycoproteins in CHO cell lines [23]. The majority of studies to date have concentrated on N-linked rather than O-linked glycosylation, however, models specific to O-linked glycosylation have now begun to appear. The Glycosylation Network Analysis Toolbox [24] provides a computational framework for modelling reaction networks of both N-linked and www.sciencedirect.com

O-linked glycosylation, as well as the handling and analysis of glycomics generated mass spectrometry (MS) data. Before this, an object-oriented programming method was described by Liu et al. and together with biochemical data was used to predict and define the reaction networks leading to the formation of sialyl Lewis-X glycans linked to P-selectin Glycoprotein Ligand-1 [25]. The first in silico meta-glycome of human milk glycans has recently appeared [26]. A formal grammar for modelling the enzymes of O-glycosylation has been used as the basis of a study into the effects of enzyme knockouts on the appearance of several glycan epitopes [27].

Control of flux in glycosylation Glycosylation is an extensive, heterogeneous and complex process, involving many enzymes, transporters, substrate donors and acceptors. Varying metabolic flux in cells regulates the synthesis and antennarity of glycan structures [28]. The extent of the glycosylation process, and the heterogeneity of glycans, is also dependent on a number of factors, including, but not limited to: enzyme expression levels, their association, location and activity within Golgi; the biosynthesis of nucleotide-sugar donors in cytoplasm and nucleus and their entry into the Golgi by specific transporters; the removal of donor products via nucleoside diphosphatase (NDPase) activity; the nature and availability of the acceptor(s); and the rate of maturation of the Golgi. Furthermore, the levels of particular types of glycan, at steady state, will be dependent on two types of competition: that which exists between multiple substrates for a given enzyme, and flux-based competition that exists between different enzymes acting on the same substrate (see Figure 2). Uman˜a and Bailey [14], and others, have given consideration to the first of these, in which access to the active site of an enzyme is blocked when another acceptor is currently bound. This can be illustrated by the galactosyltransferase (GalT) enzymes (EC 2.4.1.38), which can act on either a biantennary glycan with two terminal GlcNAc residues: GlcNAc b1-3(GlcNAc b1-6)-R, to form either Gal b1-4 GlcNAc b1-3(GlcNAc b1-6)-R or GlcNAc b1-3(Galb1-4 GlcNAcb1-6)-R as the product. With each of these being a substrate of the same enzyme, the net effect is an increase in the apparent Km (Michaelis constant) for the GalT reaction in the network. Accordingly, several modelling studies have incorporated competitive inhibition terms in their enzyme-kinetic rate laws to account for this behaviour [14,15,22,29]. Monica et al. [13] reasoned that, since the acceptor-Km values of the sialyltransferases are generally tenfold higher than the total protein concentration in the Golgi stacks, the effects of substrate competition would be negligible, and consequently they excluded it from their model. By contrast, the flux-based competition between different enzymes in the network, will be a function of the Vmax, In this case, as Hossler et al. pointed out, varying one Current Opinion in Structural Biology 2016, 40:97–103

100 Carbohydrate–protein interactions and glycosylation

Figure 2

(a)

(b)

GalT

GalT

GnTIII GalT

GnTIV

GalT

GalT

Current Opinion in Structural Biology

Competitive aspects of flux regulation during glycosylation. (a) Multi-substrate competition for a single enzyme; (b) enzymes in competition for a single substrate.

enzyme concentration to alter the flux through its subset of the reaction network could activate, or suppress, the activities of other enzymes downstream, with unexpected results [19]. An adaptation of the Krambeck and Betenbaugh (2005) model [15] can be used to illustrate this (Figure 3). Predicted changes to overall N-glycan antennarity during GalT knockdown, in conjunction with the overexpression of GnT-IV and GnT-V, have been validated in a CHO cell line [20]. In another study, an a2,6-sialyltransferase (ST6GAL1) activity was introduced into CHO cell lines by means of plasmid expression vectors, and overexpression of ST6GAL1 was able to produce more human-like immunoglobulin-G (IgG) glycoprofiles, possessing both a2,3linked and a2,6-linked terminal sialic acids [30]. Investigations into the parameters controlling flux through glycosylation have application to the biotechnology and biopharmaceutical industries, especially to monoclonal antibody (mAb) production [31–36], where systems-biological approaches continue to provide new insights [4]. Control of nucleotide-sugar donor biosynthesis was the focus of a model by Jedrzejewski et al. [29], who used the output of their model as input to the earlier model of IgG glycosylation [22], and were able to predict changes to the Fc region of antibody in a murine cell line. This detailed mechanistic model has recently been adapted for the prediction of N-glycosylation patterns of fed-batch cell cultures [37]. Altering cell culture conditions to obtain a desired glycoprofile has been a common consideration to several studies (e.g. [28,29,38,39]). One such study, in CHO cells, used a small shift to lower Current Opinion in Structural Biology 2016, 40:97–103

temperatures, during the transition to stationary phase, to observing an increase in both cell viability and specific antibody productivity. The results of flux balance (FBA) analysis suggested decreased flux through the nucleotide and NSD biosynthetic pathways, partly as a result of reduced glucose consumption during mild hypothermia. In addition to reduced glycosyltransferase expression, the lower availability of donor substrates UDP-Glc, UDP-Gal and UDP-GlcNAc, essential for glycan initiation, branching and elongation, led to glycans of lower complexity [40]. CHO cells grown with mannose as the major carbon source had higher concentrations of intracellular mannose, which through product inhibition of a-mannosidase in ER and Golgi led to increased high-mannose glycans on expressed IgG [36]. The effect of oxidative stress on glycosylation has been the subject of several recent studies [41–46]. Using 13C-metabolic flux analysis, Nargund et al. showed that CHO cells deficient in copper have a significantly altered carbon flux distribution, which was increased through glycolysis but decreased through the TCA and pentose-phosphate pathways [44]. They proposed that copper deficiency results in lower cytochrome-c oxidase activity, consequently destabilising Complex I, with concomitantly increased production of reactive-oxygen species (ROS) and lactate via a switch to aerobic glycolysis. Since the nature of the acceptor can be an important influence on the observed glycoforms [8,47], an open question is whether it is possible to deduce, from structural knowledge of the interacting species, the enzymekinetic parameters of the reactions. A combination of the site-occupancy predictive tools referred to earlier, with www.sciencedirect.com

Metabolic flux control in glycosylation McDonald, Hayes and Davey 101

100

100

80

80

80

60 40

60 40 20

20 3 2 4 5 [GnT-II]total (µM)

1

6

2

4

6

8

10

1

80

40

60 40

20

20

0

0 4

6

8

mole %

100

80 mole %

100

80 60

10

40

0 4

2

6

8

10

2

80 mole %

100

80 mole %

100

80

60 40 20

0 2.0

[GnT-IV]total (µM)

3.0

6

8

10

25

30

60 40 20

0 1.0

4

[SiaT]total (µM)

100

20

4

60

[Man-II]total (µM)

40

3

20

[Man-I]total (µM)

60

2

[GalT]total (µM)

[GnT-V]total (µM)

100

2

mole %

40

0 0

7

monoantennary biantennary triantennary tetraantennary sialylated

60

20

0

0

mole %

mole %

100

mole %

mole %

Figure 3

0 4

2

6

8

[GnT-III]total (µM)

10

5

10

15

20

[GnT-I]total (µM) Current Opinion in Structural Biology

Regulation of flux by varying enzyme concentration. Shown are the predicted changes in the steady-state levels of N-glycans of different antennarity, and the degree of sialylation, while varying enzyme concentrations ten-fold above and below their presumed wild-type values, as indicated by arrows on the abscissa of each graph. Predictions are based on the model of CHO-cell N-linked glycosylation by Krambeck and Betenbaugh (2005) [15], in which donor concentrations are assumed to be saturating.

quantitative methods based on interaction fields [48], may be one way to achieve this goal, and provide additional information for systems-biological models of glycosylation.

Conclusions In conclusion, systems glycobiology is a valuable addition of ever-increasing importance to the glycomics toolbox, which complements current technologies that generate vast amounts of glycomics data through high content screens and GWAS studies, glycoarrays and MS based glycomics systems. Current models have evolved from the modelling of initiation sites and site occupancy on glycoproteins to N-linked pathways and computational frameworks for the analysis of both N-linked and O-linked networks and the handling of experimentally generated www.sciencedirect.com

glycomics data [49,50]. Predictions based on modelling studies of glycosylation networks have been experimentally verified and have led to novel findings and identification of control points in these networks, leading to vindication of the model systems and simultaneously showing the utility of these approaches. Flux control in glycosylation pathways and networks resulting from different enzymes acting on the same substrates, the levels of the enzymes and multiple substrates for a given enzyme has implications for glycan heterogeneity and complexity and the control of glycosylation and the final glycoforms of a given glycoprotein. This in turn has important implications for the biopharmaceutical industry and therapeutic glycoproteins such as monoclonal antibodies where activity and interaction with biological systems is modulated and controlled by the final glycoform. Current Opinion in Structural Biology 2016, 40:97–103

102 Carbohydrate–protein interactions and glycosylation

Conflict of interest statement Nothing declared.

Acknowledgments

16. North SJ, Huang HH, Sundaram S, Jang-Lee J, Etienne AT, Trollope A, Chalabi S, Dell A, Stanley P, Haslam SM: Glycomics profiling of Chinese hamster ovary cell glycosylation mutants reveals N-glycans of a novel size and complexity. J Biol Chem 2010, 285:5759-5775.

This work was part supported by an EU Marie Curie Initial Training Network, Project No. 608381 and Science Foundation Ireland Grant No. SFI-13/SP SSPC/I2893.

17. Krambeck FJ, Bennun SV, Narang S, Choi S, Yarema KJ, Betenbaugh MJ: A mathematical model to derive N-glycan structures and cellular enzyme activities from mass spectrometric data. Glycobiology 2009, 19:1163-1175.

References and recommended reading

18. Bennun SV, Yarema KJ, Betenbaugh MJ, Krambeck FJ: Integration of the transcriptome and glycome for identification of glycan cell signatures. PLoS Comput Biol 2013, 9:e1002813.

Papers of particular interest, published within the annual period of review, have been highlighted as:  of special interest  of outstanding interest 1.

Rottger S, White J, Wandall HH, Olivo JC, Stark A, Bennett EP, Whitehouse C, Berger EG, Clausen H, Nilsson T: Localization of three human polypeptide GalNAc-transferases in HeLa cells suggests initiation of O-linked glycosylation throughout the Golgi apparatus. J Cell Sci 1998, 111(Pt 1):45-60.

2.

Brockhausen I, Schachter H, Stanley P: O-GalNAc glycans. In Essentials of Glycobiology, edn 2nd. Edited by Varki A, Cummings RD, Esko JD, Freeze HH, Stanley P, Bertozzi CR, Hart GW, Etzler ME.2009.

3.

Hassinen A, Kellokumpu S: Organizational interplay of Golgi Nglycosyltransferases involves organelle microenvironmentdependent transitions between enzyme homo- and heteromers. J Biol Chem 2014, 289:26937-26948.

4.

Spahn PN, Lewis NE: Systems glycobiology for glycoengineering. Curr Opin Biotechnol 2014, 30:218-224.

5.

Shelikoff M, Sinskey AJ, Stephanopoulos G: A modeling framework for the study of protein glycosylation. Biotechnol Bioeng 1996, 50:73-90.

6.

Ben-Dor S, Esterman N, Rubin E, Sharon N: Biases and complex patterns in the residues flanking protein N-glycosylation sites. Glycobiology 2004, 14:95-101.

7. 

Murray AN, Chen W, Antonopoulos A, Hanson SR, Wiseman RL, Dell A, Haslam SM, Powers DL, Powers ET, Kelly JW: Enhanced aromatic sequons increase oligosaccharyltransferase glycosylation efficiency and glycan homogeneity. Chem Biol 2015, 22:1052-1062. Provides further evidence that N-glycosylation is strongly influenced by the environment of the sequon.

8.

Gerken TA: Kinetic modeling confirms the biosynthesis of mucin core 1 (beta-Gal(1-3) alpha-GalNAc-O-Ser/Thr) Oglycan structures are modulated by neighboring glycosylation effects. Biochemistry 2004, 43:4137-4142.

9.

Hansen JE, Lund O, Tolstrup N, Gooley AA, Williams KL, Brunak S: NetOglyc: prediction of mucin type O-glycosylation sites based on sequence context and surface accessibility. Glycoconj J 1998, 15:115-130.

10. Li S, Liu B, Zeng R, Cai Y, Li Y: Predicting O-glycosylation sites in mammalian proteins by using SVMs. Comput Biol Chem 2006, 30:203-208. 11. Nishikawa I, Nakajima Y, Ito M, Fukuchi S, Homma K, Nishikawa K: Computational prediction of O-linked glycosylation sites that preferentially map on intrinsically disordered regions of extracellular proteins. Int J Mol Sci 2010, 11:4991-5008. 12. Zhou K, Ai C, Dong P, Fan X, Yang L: A novel model to predict Oglycosylation sites using a highly unbalanced dataset. Glycoconj J 2012, 29:551-564. 13. Monica TJ, Andersen DC, Goochee CF: A mathematical model of sialylation of N-linked oligosaccharides in the trans-Golgi network. Glycobiology 1997, 7:515-521. 14. Umana P, Bailey JE: A mathematical model of N-linked glycoform biosynthesis. Biotechnol Bioeng 1997, 55:890-908. 15. Krambeck FJ, Betenbaugh MJ: A mathematical model of Nlinked glycosylation. Biotechnol Bioeng 2005, 92:711-728. Current Opinion in Structural Biology 2016, 40:97–103

19. Hossler P, Mulukutla BC, Hu WS: Systems analysis of N-glycan processing in mammalian cells. PLoS One 2007, 2:e713. 20. McDonald AG, Hayes JM, Bezak T, Gluchowska SA, Cosgrave EF, Struwe WB, Stroop CJ, Kok H, van de Laar T, Rudd PM et al.: Galactosyltransferase 4 is a major control point for glycan branching in N-linked glycosylation. J Cell Sci 2014, 127:5014-5026. 21. McDonald AG, Tipton KF, Stroop CJ, Davey GP: GlycoForm and Glycologue: two software applications for the rapid construction and display of N-glycans from mammalian sources. BMC Res Notes 2010, 3:173. 22. del Val IJ, Nagy JM, Kontoravdi C: A dynamic mathematical model for monoclonal antibody N-linked glycosylation and nucleotide sugar donor transport within a maturing Golgi apparatus. Biotechnol Prog 2011, 27:1730-1743. 23. Spahn PN, Hansen AH, Hansen HG, Arnsdorf J, Kildegaard HF,  Lewis NE: A Markov chain model for N-linked protein glycosylation — towards a low-parameter tool for modeldriven glycoengineering. Metab Eng 2016, 33:52-66. A novel approach to modelling glycosylation that requires knowledge of fewer parameters to predict flux through reaction networks than the more commonly encountered models based on ordinary differential equations. 24. Liu G, Neelamegham S: A computational framework for the  automated construction of glycosylation reaction networks. PLoS One 2014, 9:e100939. A computational framework to deduce glycosylation networks from user supplied data, including substrate/product information and the biosynthetic enzymes involved. 25. Liu G, Marathe DD, Matta KL, Neelamegham S: Systems-level modeling of cellular glycosylation reaction networks: O-linked glycan formation on natural selectin ligands. Bioinformatics 2008, 24:2740-2747. 26. Agravat SB, Song X, Rojsajjakul T, Cummings RD, Smith DF: Computational approaches to define a human milk metaglycome. Bioinformatics 2016, 32:1471-1478. 27. McDonald AG, Tipton KF, Davey GP: A knowledge-based system for display and prediction of O-glycosylation network behaviour in response to enzyme knockouts. PLoS Comput Biol 2016, 12:e1004844. 28. Burleigh SC, van de Laar T, Stroop CJ, van Grunsven WM, O’Donoghue N, Rudd PM, Davey GP: Synergizing metabolic flux analysis and nucleotide sugar metabolism to understand the control of glycosylation of recombinant protein in CHO cells. BMC Biotechnol 2011, 11:95. 29. Jedrzejewski PM, del Val IJ, Constantinou A, Dell A, Haslam SM, Polizzi KM, Kontoravdi C: Towards controlling the glycoform: a model framework linking extracellular metabolites to antibody glycosylation. Int J Mol Sci 2014, 15:4492-4522. 30. Lin N, Mascarenhas J, Sealover NR, George HJ, Brooks J, Kayser KJ, Gau B, Yasa I, Azadi P, Archer-Hartmann S: Chinese hamster ovary (CHO) host cell engineering to increase sialylation of recombinant therapeutic proteins by modulating sialyltransferase expression. Biotechnol Prog 2015, 31:334-346. 31. Aghamohseni H, Ohadi K, Spearman M, Krahn N, Moo-Young M, Scharer JM, Butler M, Budman HM: Effects of nutrient levels and average culture pH on the glycosylation pattern of camelidhumanized monoclonal antibody. J Biotechnol 2014, 186:98-109. www.sciencedirect.com

Metabolic flux control in glycosylation McDonald, Hayes and Davey 103

32. Agarabi CD, Schiel JE, Lute SC, Chavez BK, Boyne MT 2nd, Brorson KA, Khan MA, Read EK: Bioreactor process parameter screening utilizing a Plackett-Burman design for a model monoclonal antibody. J Pharm Sci 2015, 104:1919-1928. 33. Blondeel EJ, Braasch K, McGill T, Chang D, Engel C, Spearman M, Butler M, Aucoin MG: Tuning a MAb glycan profile in cell culture: supplementing N-acetylglucosamine to favour G0 glycans without compromising productivity and cell growth. J Biotechnol 2015, 214:105-112. 34. Zupke C, Brady LJ, Slade PG, Clark P, Caspary RG, Livingston B, Taylor L, Bigham K, Morris AE, Bailey RW: Real-time product attribute control to manufacture antibodies with defined Nlinked glycan levels. Biotechnol Prog 2015, 31:1433-1441. 35. Sha S, Agarabi C, Brorson K, Lee DY, Yoon S: N-glycosylation design and control of therapeutic monoclonal antibodies. Trends Biotechnol 2016 http://dx.doi.org/10.1016/ j.tibtech.2016.02.013. 36. Slade PG, Caspary RG, Nargund S, Huang CJ: Mannose metabolism in recombinant CHO cells and its effect on IgG glycosylation. Biotechnol Bioeng 2016, 113:1468-1480. 37. Villiger TK, Scibona E, Stettler M, Broly H, Morbidelli M, Soos M: Controlling the time evolution of mAb N-linked glycosylation — Part II: model-based predictions. Biotechnol Prog 2016 http://dx.doi.org/10.1002/btpr.2315. 38. Kyriakopoulos S, Kontoravdi C: A framework for the systematic design of fed-batch strategies in mammalian cell culture. Biotechnol Bioeng 2014, 111:2466-2476. 39. Fan Y, del Val IJ, Muller C, Wagtberg Sen J, Rasmussen SK, Kontoravdi C, Weilguny D, Andersen MR: Amino acid and glucose metabolism in fed-batch CHO cell culture affects antibody production and glycosylation. Biotechnol Bioeng 2015, 112:521-535. 40. Sou SN, Sellick C, Lee K, Mason A, Kyriakopoulos S, Polizzi KM, Kontoravdi C: How does mild hypothermia affect monoclonal antibody glycosylation? Biotechnol Bioeng 2015, 112:1165-1176. 41. Brodsky AN, Caldwell M, Bae S, Harcum SW: Glycosylationrelated genes in NS0 cells are insensitive to moderately

www.sciencedirect.com

elevated ammonium concentrations. J Biotechnol 2014, 187:78-86. 42. Nicolae A, Wahrheit J, Bahnemann J, Zeng AP, Heinzle E: Nonstationary 13C metabolic flux analysis of Chinese hamster ovary cells in batch culture using extracellular labeling highlights metabolic reversibility and compartmentation. BMC Syst Biol 2014, 8:50. 43. Wahrheit J, Nicolae A, Heinzle E: Dynamics of growth and metabolism controlled by glutamine availability in Chinese hamster ovary cells. Appl Microbiol Biotechnol 2014, 98:1771-1783. 44. Nargund S, Qiu J, Goudar CT: Elucidating the role of copper in CHO cell energy metabolism using (13)C metabolic flux analysis. Biotechnol Prog 2015, 31:1179-1186. 45. Kizuka Y, Nakano M, Kitazume S, Saito T, Saido TC, Taniguchi N: Bisecting GlcNAc modification stabilizes BACE1 protein under oxidative stress conditions. Biochem J 2016, 473:21-30. 46. Taniguchi N, Kizuka Y, Takamatsu S, Miyoshi E, Gao C, Suzuki K, Kitazume S, Ohtsubo K: Glyco-redox, a link between oxidative stress and changes of glycans: lessons from research on glutathione, reactive oxygen and nitrogen species to glycobiology. Arch Biochem Biophys 2016, 595:72-80. 47. Hang I, Lin CW, Grant OC, Fleurkens S, Villiger TK, Soos M, Morbidelli M, Woods RJ, Gauss R, Aebi M: Analysis of sitespecific N-glycan remodeling in the endoplasmic reticulum and the Golgi. Glycobiology 2015, 25:1335-1349. 48. Gabdoulline RR, Stein M, Wade RC: qPIPSA: relating enzymatic kinetic parameters and interaction fields. BMC Bioinformatics 2007, 8:373. 49. Liu G, Neelamegham S: Integration of systems glycobiology with bioinformatics toolboxes, glycoinformatics resources, and glycoproteomics data. Wiley Interdiscip Rev Syst Biol Med 2015, 7:163-181. 50. Del Val IJ, Polizzi KM, Kontoravdi C: A theoretical estimate for nucleotide sugar demand towards Chinese Hamster Ovary cellular glycosylation. Sci Rep 2016, 6:28547.

Current Opinion in Structural Biology 2016, 40:97–103