Using Glyco-Engineering to Produce Therapeutic Proteins

Assigned Reading:

Using glyco-engineering to produce therapeutic proteins. Dicker M and Strasser R Expert Opin. Biol. Ther. (2015) 15:1501.

Advances in the production of human therapeutic proteins in yeasts and filamentous fungi Gerngross TU, Nat. Biotech, 22:1409 (2004)

Glycan Engineering for Cell and Developmental Biology Griffin ME and Hsieh-Wilson LC Cell Chem Biol, 23:108 (2016) Expert Opinion on Biological Therapy

ISSN: 1471-2598 (Print) 1744-7682 (Online) Journal homepage: http://www.tandfonline.com/loi/iebt20

Using glyco-engineering to produce therapeutic proteins

Martina Dicker & Richard Strasser

To cite this article: Martina Dicker & Richard Strasser (2015) Using glyco-engineering to produce therapeutic proteins, Expert Opinion on Biological Therapy, 15:10, 1501-1516, DOI: 10.1517/14712598.2015.1069271

To link to this article: http://dx.doi.org/10.1517/14712598.2015.1069271

Published online: 14 Jul 2015.

Submit your article to this journal

Article views: 345

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=iebt20

Download by: [University of California, San Diego] Date: 17 May 2016, At: 09:13 Review Using glyco-engineering to produce therapeutic proteins

† Martina Dicker & Richard Strasser 1. Introduction University of Natural Resources and Life Sciences, Department of Applied Genetics and Cell Biology, Vienna, Austria 2. N-glycosylation of proteins 3. What are the targets for Introduction: Glycans are increasingly important in the development of N-glycan-engineering? new biopharmaceuticals with optimized efficacy, half-life, and antigenicity. 4. O-glycosylation of proteins Current expression platforms for recombinant glycoprotein therapeutics typically do not produce homogeneous glycans and frequently display non- 5. What are the targets for human glycans which may cause unwanted side effects. To circumvent these O-glycan-engineering? issues, glyco-engineering has been applied to different expression systems 6. O-glycan engineering including mammalian cells, insect cells, yeast, and plants. 7. Glyco-engineered therapeutic Areas covered: This review summarizes recent developments in glyco- glycoproteins engineering focusing mainly on in vivo expression systems for recombinant 8. Conclusion proteins. The highlighted strategies aim at producing glycoproteins with 9. Expert opinion homogeneous N- and O-linked glycans of defined composition. Expert opinion: Glyco-engineering of expression platforms is increasingly rec- 10. So what are the challenges for the future? ognized as an important strategy to improve biopharmaceuticals. A better understanding and control of the factors leading to glycan heterogeneity will allow simplified production of recombinant glycoprotein therapeutics with less variation in terms of glycosylation. Further technological advances will have a major impact on manufacturing processes and may provide a completely new class of glycoprotein therapeutics with customized functions.

Keywords: biopharmaceuticals, glycosyltransferases, N-glycosylation, O-glycosylation, recombinant glycoprotein

Expert Opin. Biol. Ther. (2015) 15(10):1501-1516

1. Introduction

Many protein drugs are glycosylated and the attached glycan structures often influence the therapeutic properties. The number and composition of the glycans play an important role for protein folding, solubility, and intracellular trafficking. Glycans can shield the protein backbone to prevent immunogenic reactions and distinct cellular recognition events depend on the presence of specific glycan structures. Downloaded by [University of California, San Diego] at 09:13 17 May 2016 Thus, glycosylation has a huge impact on the biological activity of glycoproteins and should be carefully controlled during manufacturing to achieve optimized therapeutic efficacy. Dependent on the species, cell-type, and physiological status of the production host, the glycosylation pattern on recombinant glycoproteins can differ significantly. Glycans on proteins are structurally quite diverse and consist of a set of monosaccharides that are assembled by different linkages. The glycosylation processes in the ER and Golgi generate the majority of rather heterogeneous glycan structures found on recombinant glycoproteins. Due to the recognized importance in therapy, substantial efforts have been made in recent years to overcome glycan heterogeneity and establish in vivo and in vitro glyco-engineering technologies for efficient production of homogeneous therapeutic glycoproteins. Such recombinant proteins with custom-made glycosylation are essential to understand glycan- mediated functions of glycoprotein therapeutics and represent a new class of next- generation drugs with enhanced biological activity.

3. What are the targets for N-glycan- Article highlights. engineering? . Glyco-engineering offers great potential for the generation of glycoprotein therapeutics with reduced 3.1 Avoidance of macroheterogeneity side effects and enhanced activity. . Analysis of recombinant glycoproteins from different Macroheterogeneity on recombinant glycoproteins arises from mammalian and non-mammalian-based expression variations in glycosylation site occupancy. These differences in systems reveals the presence of unwanted glycan glycosylation efficiency are dependent on the manufacturing modifications that mark targets for glyco-engineering. system (e.g., organism/cell-type-specific) and protein intrinsic . Recent developments in modulation of N-glycosylation and N-glycan processing demonstrate that recombinant features. The presence of the Asn-X-Ser/Thr sequon is neces- glycoproteins with defined homogeneous N-glycans can sary but not sufficient for N-glycosylation of mammalian be efficiently produced in mammalian and non- proteins. While all eukaryotic cells have an overall conserved mammalian expression systems including insect cells, machinery for N-glycosylation and display similar structural yeast, and plants. requirements for efficient N-glycosylation, there are minor dif- . Compared to N-glycosylation, strategies and methods to produce customized O-linked glycans are still in its ferences in utilization of glycosylation sites between species [2,3]. infancy. Nonetheless, future developments in this area Consequently, glycosylation sites may be skipped leading to hold great promise for another class of improved underglycosylation of recombinant proteins or additional glycoprotein biopharmaceuticals. sites may be used leading to aberrant glycosylation. Protein . The first glyco-engineered monoclonal antibodies have intrinsic factors like the surrounding amino-acid sequence already been approved in the US and in Japan. . Further biochemical and molecular understanding of and secondary structure, the positioning of the consensus cellular process will be required to optimize current sequence within the polypeptide as well as the presence of other glyco-engineering approaches and develop strategies for protein modifications like disulfide bond formation so far elusive glycan modifications. contribute to N-glycosylation efficiency [4]. Recognition of these glycoprotein-specific features depends on the presence This box summarizes key points contained in the article. and function of the different OST subunits. Mammalian cells contain two OST complexes that differ in their catalytic sub- 2. N-glycosylation of proteins unit as well as in accessory proteins. These OST complexes display partially overlapping functions, but also preferences The most prominent and best characterized form of protein for certain glycoprotein substrates [3]. Engineering of the glycosylation is the linkage of a glycan to the amide in OST complex is one possible way to overcome differences the side chain of an asparagine (N-glycosylation) on newly related to N-glycosylation site occupancy. Yet, the individual synthetized proteins. N-glycosylation of proteins starts in roles of distinct OST catalytic subunits and their accessory pro- the lumen of the endoplasmic reticulum (ER) by transfer of teins are still not fully understood in eukaryotes making ratio- a conserved preassembled oligosaccharide precursor nal approaches difficult. Moreover, a recent study has indicated that overexpression of individual subunits is not sufficient to (Glc3Man9GlcNAc2)(Figure 1A) to the consensus sequence Asn-X-Ser/Thr (where X is any amino acid except proline) restore N-glycosylation in mutant cells presumably due to the inability to form functional heteromeric OST complexes [5]. exposed on nascent polypeptide chains. This initial glycan However, in protists like Leishmania major, OST is composed transfer reaction is catalyzed by the heteromeric oligosacchar- of a single subunit that can replace the whole Saccharomyces cer- yltransferase (OST) complex and is supposed to precede

Downloaded by [University of California, San Diego] at 09:13 17 May 2016 evisiae OST complex and can be used to overcome underglyco- folding of the protein in the ER [1]. Immediately after the sylation of proteins. Overexpression of the single-subunit OST oligosaccharide transfer, the two terminal glucose residues from L. major in the methylotrophic yeast Pichia pastoris are cleaved off by a-glucosidase I and II and the resulting increased the N-glycosylation site occupancy of a monoclonal polypeptide with mono-glucosylated glycan structures antibody from 75 to 85% to greater than 99% [6]. (Glc1Man9GlcNAc2) can interact with the ER-resident Other engineering attempts to reduce macroheterogeneity membrane-bound lectin calnexin or its soluble homolog are based on mutation of the N-glycosylation consensus calreticulin. These lectins support protein folding in a sequence, for example, from Asn-X-Ser to Asn-X-Thr which glycan-dependent protein quality control cycle. Secretory is more efficiently glycosylated in different eukaryotes [2]. glycoproteins that have acquired their native conformation Besides modifying the consensus sequence itself, adjacent are released from the calnexin/calreticulin cycle and exit the amino acids may also be altered to enhance N-glycosylation ER to the Golgi apparatus. In the Golgi, the ER-derived efficacy. The major drawback of these strategies is an alter- oligomannosidic N-glycans on maturely folded glycoproteins ation of the primary amino acid sequence of the therapeutic are subjected to further N-glycan processing which generates glycoprotein which can have an influence on protein proper- the highly diverse complex N-glycans with different func- ties and may create unwanted regulatory concerns. Nonethe- tional properties (Figure 1B). less, novel N-glycosylation sites have been successfully

1502 Expert Opin. Biol. Ther. (2015) 15(10) Using glyco-engineering to produce therapeutic proteins

M S F G

B. Downloaded by [University of California, San Diego] at 09:13 17 May 2016

Figure 1. Schematic presentation of N-glycan processing pathways in mammals. (A) N-glycosylation is initiated by OST- catalyzed transfer of the lipid-linked preassembled oligosaccharide to Asn with the Asn-X-Ser/Thr consensus sequence. N-glycan processing starts in the endoplasmic reticulum (ER) by the removal of glucose and mannose residues and (B) continues in the different Golgi cisternae by numerous processing reactions. Only a selection of possible complex N-glycan modifications is shown. FUT8: Core a1,6-fucosyltransferase; GALT1: b1,4-galactosyltransferase; GCSI: a-glucosidase I; GCSII: a-glucosidase II; GMI: Golgi-a-mannosidase I; GMII: Golgi-a- mannosidase II; GnTI: N-acetylglucosaminyltransferase I; GnTII: N-acetylglucosaminyltransferase II; GnTIV: N-acetylglucosaminyltransferase IV; GnTV: N-acetylglucosaminyltransferase V; MANI: ER-a-mannosidase I; ST: a2,6-sialyltransferase.

engineered into recombinant proteins. The most prominent acid content which affects the protein half-life and boost the example for this approach is darboetin alfa. This hyperglyco- in vivo potency [8]. In another elegant study, introduction of sylated recombinant erythropoietin (EPO) variant has been a novel N-glycosylation site into the light chain of a monoclo- engineered to carry five instead of three N-glycosylation nal antibody against the HIV-1 receptor CD4 substantially sites [7]. The additional two N-glycans increase the total sialic improved the virus neutralization activity [9]. Collectively,

Expert Opin. Biol. Ther. (2015) 15(10) 1503 M. Dicker & R. Strasser

B. C.

D. E. F.

Figure 2. (A) Illustrations of representative N-glycans from human cells/tissues and characteristic structures from (B) CHO cells, (C) mouse myeloma cells, (D) P. pastoris,(E) insect cells and (F) plants.

these examples illustrate the potential for protein engineering of corresponding fucosyltransferases. On the other hand, for toward incorporation of additional glycosylation sites. complex N-glycan formation or sialylation, this could not be established highlighting a more complicated regulation for these trimming and processing steps [10]. In addition, models have 3.2 Reduction of microheterogeneity been developed to predict the influence of enzyme expression, The commonly used expression systems for recombinant subcellular localization, nucleotide sugar levels, and culture con- glycoproteins typically produce a mixture of different glycans ditions on glycan processing [13]. While the described models on the same protein. This means that different glycans are found may already be useful to predict the rather homogeneous Fc- at the same glycosylation site and that individual sites on the same protein can be furnished with glycans of high structural glycosylation on a recombinant monoclonal antibody [14], important key assumptions of these models have to be validated

Downloaded by [University of California, San Diego] at 09:13 17 May 2016 diversity. The N-glycome of mammalian cells shows a wide range of N-glycan compositions from oligomannosidic to by additional experimental data in the future. Moreover, the branched complex-type N-glycans (Figure 2A) [10,11]. Microhe- robustness of current mathematical models needs to be tested terogeneity arises from variations in N-glycan processing which in host cells on recombinant glycoproteins which display a depends on numerous cellular factors including tissue/cell- more complicated and heterogeneous glycosylation pattern type-specific expression of processing enzymes, their concentra- including branching and sialylation of N-glycans [15]. tions and enzyme kinetic parameters, availability of the nucleo- Other factors that contribute to microheterogeneity are tide sugar donors and the glycoprotein residence, and contact the competition of different modifying enzymes for the time in the reaction compartment (e.g., Golgi apparatus). In same oligosaccharide substrateandthepresenceofendoge- addition, intrinsic protein structural properties can have a strong nous glycoproteins that may interfere with processing or effect on site-specific N-glycan processing. Systems’ glycobiol- contact time in the Golgi. Although it is commonly accepted ogy approaches have been used to correlate the transcript expres- that the Golgi processing enzymes are organized sequentially sion of glycosylation enzymes and the generation of N-glycan in some kind of assembly line across the individual Golgi structures on glycoproteins from mouse tissues or cancer cisternae [16], there is typically an overlapping distribution cells [10,12]. In mouse tissues, a positive correlation was found of multiple enzymes. These enzymes may modify the same between transfer of fucose residues and the transcript expression glycan intermediate structure and block the action of

1504 Expert Opin. Biol. Ther. (2015) 15(10) Using glyco-engineering to produce therapeutic proteins

other glycosyltransferases and glycosidases. The addition of modifications [30]. Despite some progress in this field, funda- the bisecting GlcNAc to complex N-glycans by mental experimental data on Golgi kinetics of different N-acetylglucosaminyltransferase III prevents, for example, recombinant glycoproteins and the impact on glycan process- further modifications by other Golgi-resident glycosyltrans- ing are not available to rationally engineer transport and traf- ferases. This substrate competition and inhibition of further ficking pathways toward more homogeneous glycan processing is the key factor of the GlycoMab technology that structures. In other words, for many cell-based manufacturing was developed by Glycart Biotechnology (acquired by systems, we need a better understanding of cellular transport Roche) and is used to produce glyco-engineered monoclonal processes in order to reduce glycan microheterogeneity related antibodies with reduced amounts of a1,6-linked core to Golgi residence time or intra-Golgi transport. fucose [17]. The glyco-engineered obinutuzumab, an anti- Finally, a major contribution to site-specific modifications CD20 monoclonal antibody produced by this platform, comes from glycoprotein intrinsic features that are deter- was recently approved by the US FDA for the treatment of mined by the protein sequence and structure [31]. Certain pro- patients with previously untreated chronic lymphocytic leu- tein conformations completely prevent or limit the access of kemia [18]. Galactosylation and branching of complex the processing enzymes to the glycans by steric hindrance [32]. N-glycans are other frequent modifications that use the Monoclonal antibodies like cetuximab with an additional same substrates (Figure 1B) [15]. Apart from these competing N-glycan in the variable region of the heavy chain are good reactions that act in the same biosynthetic compartment, a examples for the site-specific processing on the same mole- significant impact of enzymes involved in catabolic processes cule. While the Fc N-glycan of cetuximab is less modified, cannot be excluded. Extracellular a-neuraminidases or the N-glycan in the variable region is more exposed and dis- b-hexosaminidases can act on terminal sugar residues and plays extensive processing [33,34]. Strategies to engineer the remove them from exposed N-glycans after secretion leading protein intrinsic features without altering the protein function tovariationinglycanprofilesonrecombinantglycopro- are challenging, but the increasing availability of protein teins [10,19]. As a consequence, genetic inactivation or structures together with powerful molecular dynamics simula- inhibition of competing glycosyltransferases and (catabolic) tions will open new possibilities for site-specific glycan engi- glycoside hydrolases are beneficial strategies to prevent neering in the future [35]. Alternatively, the rational design unwanted glycan modifications and reduce the complexity of glycan-modifying enzymes toward relaxed or novel of glycosylation. The successful generation of recombinant substrate specificities may allow more efficient processing of glycoprotein therapeutics with homogeneous glycosylation partially accessible sites. in yeast or plants [20,21] whichhavemuchlesscomplicated N-glycan processing machinery than mammalian cells demonstrates the capability of glyco-engineering. The generation 3.3 Elimination of non-human and potentially of defined glycosylation patterns in Chinese Hamster Ovary immunogenic sugar residues (CHO) cells with defined defects in glycosylation shows that The majority of the currently used expression systems for ‘simplifying’ strategies are also feasible in mammalian glycoprotein therapeutics produce N-glycans carrying non- cells [22]. Antibodies produced in Lec8 CHO cells which human structures that can lead to potential side effects of lack UDP-galactose in the Golgi display mainly homoge- the drugs. These non-human epitopes may elicit unwanted neous biantennary N-glycans [23,24]. immune responses that neutralize the applied drug or The residence time of the secreted recombinant glycopro- even worse cause a hypersensitivity reaction. Recombinant tein in the Golgi and the contact time with the membrane- glycoproteins produced in CHO cells carry the non-human

Downloaded by [University of California, San Diego] at 09:13 17 May 2016 anchored glycan-modifying enzymes are other factors that sialic acid N-glycolylneuraminic acid (Neu5Gc) (Figure 2B) may contribute significantly to microheterogeneity. Differen- [36-38]. In addition, CHO cells attach N-acetylneuraminic ces in residence times and transport kinetics through the acid (Neu5Ac) in a2,3-linkage instead of a2,6-linkage that Golgi have been described for several proteins [25,26] and their is mainly found on N-glycans from human glycoproteins influence on glycan modifications is documented [27]. The because CHO cells lack the corresponding a2,6-sialyltransfer- residence time distribution and the interaction with the glyco- ase activity. A number of different approaches have been sylation enzymes are dependent on the mode of intra-Golgi proposed to eliminate the Neu5Gc incorporation [39]. The cargo transport (vesicular transport/tubular connections, most straightforward glyco-engineering strategy is the genetic cisternal maturation, rapid partitioning, or mixed models) knockout of the gene coding for CMP-Neu5Ac hydroxylase and are cell-type and organism-specific [28]. Furthermore, (Figure 3C), the enzyme which converts CMP-Neu5Ac into recent studies suggest that different protein cargos (e.g., large CMP-Neu5Gc. Similarly, the a2,3-sialyltransferase could be vs small cargo) are transported by varying mechanisms and eliminated by genetic modification and ectopically expression therefore very likely interact with different Golgi-resident of the missing a2,6-sialyltransferase will lead to humanized enzymes [29]. For some glycosylation processes, it has been N-glycans. demonstrated that skipping of transit through individual NS0 and SP2/0 mouse myeloma or Baby Hamster Golgi cisternae is an elegant way of regulating glycan Kidney (BHK) cells generate glycoproteins with galactose-a

Expert Opin. Biol. Ther. (2015) 15(10) 1505 M. Dicker & R. Strasser

B. C.

D. E.

F. G.

Figure 3. Strategies for N-glycan engineering. (A) Overexpression of the branching enzymes N-acetylglucosaminyltransferase IV and V (GnTIV/V); (B) knockout of core a1,6-fucosyltransferase (FUT8); (C) knockout of CMP-Neu5Ac hydroxylase (CMAH); (D) knockout of a1,3-galactosyltransferase (a-1,3-GalT); (E) knockout of a1,6-mannosyltransferase (Och1p) in yeast; (F) knockout of core a1,3-fucosyltransferase (FUT) and b-N-acetylglucosaminidase (b-hex) in insect cells; (G) knockout of Downloaded by [University of California, San Diego] at 09:13 17 May 2016 core a1,3-fucosyltransferase (FUT) and b1,2-xylosyltransferase (XylT) in plants.

1,3-galactose (a-Gal epitope) (Figure 2C) that represent anti- presence of the a-Gal epitope [41]. This example highlights genic epitopes for humans. The responsible enzyme, the importance of glyco-engineering for the generation of a1,3-galactosyltransferase, is inactive in humans, but present bio-better therapeutic proteins (Figure 3D). in most non-human mammalian cell lines including Glycosylation of recombinant proteins derived from CHO [40]. BHK-derived recombinant human factor VIII human cell lines such as HEK293 can be highly heteroge- (rhFVIII) carries 3% of the a-Gal epitope, while it is not neous including different biantennary and branched struc- detected on rhFVIII produced in human embryonic kidney tures with variable terminal sugar residues and thus do cells (HEK293) [37]. Significant amounts of the a-Gal epitope not necessarily resemble N-glycosylation of serum-derived are present in the N-glycan from the variable region of cetux- glycoproteins. Despite the fact that human cell lines deficient imab [33,34] which is produced in SP2/0 mouse myeloma cells in distinct N-glycan processing steps (e.g., GnTI-deficient) and worldwide approved for treatment of different types of have been reported [42] and companies like the German-based cancer. A hypersensitive reaction to the drug in a number of Glycotope offer the production of recombinant glycoproteins cetuximab-treated patients in the US has been linked to the in glyco-engineered human cell lines [43], an extensive

1506 Expert Opin. Biol. Ther. (2015) 15(10) Using glyco-engineering to produce therapeutic proteins

re-modelling of the N-glycosylation pathway toward homoge- recombinant IgM variants with defined N-glycosylation [60] neous complex N-glycans has not been described in human and is currently used to manufacture the monoclonal anti- cells. Comparison of several different proteins expressed body cocktail against Ebola virus [61]. in CHO and HEK293 cells indicated significant differences Similar to insect cells, plants also contain in N-glycosylation with reduced sialic acid content in b-hexosaminidases that remove GlcNAc residues from some HEK293 cell produced recombinant glycoproteins [44]. recombinant glycoproteins and expose terminal mannose on Besides reducing heterogeneity, potential targets for glyco- truncated N-glycans. While this post-Golgi processing event engineering of human cells are elimination of core-fucose can be prevented by genetic knockout of the corresponding (Figure 3B) and knockout of N-acetylglucosaminyltransferase b-hexosaminidases [62], N-glycans with terminal mannose III which generates the bisecting GlcNAc and thus prevents residues are beneficial structures for recombinant glycopro- further processing and overexpression of enzymes for branch- teins used for enzyme replacement therapy to treat lysosomal ing or sialylation. storage diseases. Plant-produced recombinant glucocerebrosi- The hypermannosylated N-glycan structures of yeast dase (taliglucerase alfa) has been approved by the FDA in differ significantly from human N-glycans (Figure 2D)and 2013 to treat Gaucher disease. This drug is expressed in genet- may cause immunogenic reactions that affect the biological ically modified carrot cells and contains mainly Man3Xyl- activity of drugs [45].PioneeringworkinP. pastoris demon- FucGlcNAc2 N-glycans [63] which are efficiently internalized strated that humanized complex N-glycans can be generated by mannose receptors in humans. In contrast to other in yeast by elimination of deleterious mannosyltransferases commercially available recombinant glucocerebrosidases, (Figure 3E) and controlled expression of mannosidases and taliglucerase alfa does not require any in vitro exoglycosidase N-acetylglucosaminyltransferases from other species [46]. digestions or a-mannosidase inhibitors during the production The success of this early work from GlycoFi Inc. (now a process to generate N-glycans with terminal mannose residues. division of Merck) was based on high throughput screening of a combinatorial library of glycosylation enzymes designed for differential subcellular localization and efficient 3.4 Optimization of pharmacokinetics and biological production of secreted recombinant proteins with defined activity by incorporation or removal of specific N-glycans. More recently, similar but more rationally monosaccharides designed approaches were used to eliminate hypermannosy- Beneficial properties can be engineered by introduction of lation and engineer P. pastoris [47], Saccharomyces cerevisiae additional sugar residues that form novel recognition sites or [48], Yarrowia lipolytica [49],andHansenula polymorpha [50] mask binding sites for receptors involved in clearance. For toward the production of humanized N-glycans on recombi- instance, optimized pharmacokinetic behavior is achieved by nant glycoproteins. increasing the sialic acid content of recombinant glycopro- Insect cells typically produce truncated (‘paucimannosidic’) teins. Highly sialylated EPO variants have been produced in N-glycans due to the action of a processing b-hexosaminidase mammalian cells (darboetin alfa) [7] as well as in non- [51]. Moreover, a potentially immunogenic core a1,3-fucose mammalian systems including yeast [20], insect cells [64], and residue is present on recombinant glycoproteins expressed in plants [65]. Moreover, sialylation of antibodies is thought to some insect cell lines (Figure 2E) [52]. Different methods play an important role for anti-inflammatory properties [66]. have been developed to overcome these shortcomings and However, the potential use of sialylated IgGs in this context glyco-engineered insect cells that produce galactosylated is still controversial because the used preparations to perform N-glycans on monoclonal antibodies and fucose-deficient studies are often heterogeneous in terms of glycosylation [67].

Downloaded by [University of California, San Diego] at 09:13 17 May 2016 N-glycans have been described recently (Figure 3F) [53,54]. In glyco-engineering approaches towards optimizing activ- Plants typically generate rather simple biantennary complex ity, unwanted glycosyltransferases are typically inactivated to N-glycans with attached b1,2-xylose and core a1,3-fucose prevent the transfer of single monosaccharides to glycan struc- residues (Figure 2F) that are not found on N-glycans in tures. Knockout of core a1,6-fucosyltransferases (FUT8) has mammals and therefore represent potentially immunogenic been used to generate CHO cells lacking N-glycans with residues [55]. Genetic knockout as well as knockdown core fucose (Figure 3B) [68]. In another approach, the concen- approaches have successfully been used to eliminate tration of specific nucleotide sugars can be altered, for exam- b1,2-xylose and core a1,3-fucose from recombinant glyco- ple, by inactivation or overexpression of the corresponding protein therapeutics produced in different plants (Figure 3G) nucleotide sugar transporter or metabolic control of nucleo- [56-59]. Compared to conventional CHO-produced recombi- tide sugar flux. It has been shown that branching of nant monoclonal antibodies the N-glycans from plant-derived N-glycans is sensitive toward changes in UDP-GlcNAc con- IgGs were highly homogeneous and displayed mainly the centrations [69] which is used as donor substrate by enzymes GlcNAc2Man3GlcNAc2 glycan which is the optimal substrate that are involved in complex N-glycan formation and initia- for further modifications like branching or galactosylation. tion of N-glycan branching. Since UDP-GlcNAc levels are Notably, the Nicotiana benthamiana-based plant expression dependent on the hexosamine pathway, metabolic control platform also enables the production of functionally active may be used to modify the branching of N-glycans which

Expert Opin. Biol. Ther. (2015) 15(10) 1507 M. Dicker & R. Strasser

can lead to the production of therapeutic proteins that have system displayed interesting features that need to be further increased amounts of sialic acid and consequently an tested to fully assess the potential of this novel glyco- increased half-life (Figure 3A). engineering approach.

3.5 Other recent N-glycan engineering approaches 4. O-glycosylation of proteins Mammalian-type N-glycosylation pathways have been engineered into E. coli and N-glycosylation of recombinant glyco- Different types of O-glycosylation (e.g., O-linked fucose, proteins has been shown [70]. Such glyco-engineering attempts glucose, mannose, xylose, or GalNAc) have been described are very promising, but currently limited by the restricted on secretory proteins in mammals [80]. The formation of the acceptor site specificity of the used bacterial OST [71]. highly abundant mucin-type O-glycans is initiated by the Sequence-guided protein evolution may help to overcome polypeptide GalNAc-transferase family. In this common such bottlenecks [72]. mammalian O-glycosylation pathway, Golgi-resident poly- Synthetic and semi-synthetic approaches have been peptide GalNAc-transferases transfer a single GalNAc residue applied to remodel N-glycans on proteins and generate to Ser/Thr sites of fully folded proteins which do not display a homogeneous glycoproteins with defined glycan structures [73]. clear consensus sequence. Subsequent stepwise elongations are In vitro approaches include total chemical synthesis of glyco- typically carried out by different glycosyltransferases resulting proteins like sialylated EPO [74] or involve cell-based produc- in the formation of highly diverse core O-glycan structures tion followed by chemical or enzymatic post-production (Figure 5A). modifications to obtain homogeneous glycoproteins. Recently, in vitro incubation of b1,4-galactosyltransferase 5. What are the targets for O-glycan- and a2,6-sialyltransferase together with their corresponding engineering? nucleotide sugars (UDP-Gal, CMP-Neu5Ac) was used to increase the disialylated glycan content and homogeneity of Remarkably, the contribution of distinct O-glycans to thera- commercially available intravenous immunoglobulin (IVIg) peutic properties (e.g., for EPO which contains a single [75]. In vitro enzymatic processing steps using a-neuramini- O-glycan) is still not very well investigated. As a consequence, dase, b-galactosidase, and b-N-acetylglucosaminidase have the specific targets for O-glycan engineering are less obvious been applied to generate N-glycans with exposed mannose than for N-glycan engineering. The development of systems residues on recombinant glucocerebrosidase (imiglucerase) capable of producing defined O-glycans is vital to improve [76]. In addition to these stepwise modifications of existing our understanding of O-glycan function for glycoprotein N-glycans, enzymatic removal of heterogeneous glycans and therapeutics. IgA antibodies have, for example, high potential in vitro enzymatic transfer of a preassembled oligosaccharide as a new class of drugs to combat infections or kill tumor has been used to engineer homogeneous disialylated IgG Fc cells [81]. The engineering of O-glycan residues in the hinge fragments in order to investigate the anti-inflammatory role region of IgA1 antibodies toward highly sialylated structures of sialylation [77,78]. The recombinant glycoproteins are pro- represents an interesting target to optimize the stability and duced in eukaryotic cells or even in genetically engineered pharmacokinetic behavior of recombinant IgA1 therapeutics. bacteria that can perform distinct N-glycosylation reac- Mucin-type O-glycosylation in mammalian cells typically tions [70]. The cell-type specific heterogeneous N-glycans are varies in the frequency of site occupancy and display a mixture enzymatically removed by specific endoglycosidases leaving of different structures [82]. In human cells, the generation of glycoproteins with a single N-linked GlcNAc (Figure 4A). homogeneous mucin-type O-glycans is hampered by the large

Downloaded by [University of California, San Diego] at 09:13 17 May 2016 Enzymatic transglycosylation adds a structurally homoge- repertoire of competing glycosyltransferases (Figure 5A and B). neous pre-synthesized oligosaccharide to the N-linked Insect cells can generate some mucin-type modifications, but GlcNAc resulting in a glycoprotein with defined N-glycans. also produce a number of structurally diverse non-human While it is possible to generate homogeneous glycans on gly- O-glycans including the incorporation of fucose and hexur- coproteins by this chemoenzymatic remodeling strategy, the onic acid as well as further substitutions with phosphocholine industrial scale production of semi-synthetic glycoprotein or sulfate [83]. In contrast to animal cells, yeast and plants do therapeutics remains to be shown. not contain the typical mucin-type O-glycosylation machin- The Callewaert group recently developed a similar in vivo ery, which allows the de novo synthesis of these mammalian- remodeling system for N-glycans in human cells (Figure 4B) type O-glycans without any interference from the endogenous [79]. In the GlycoDelete strategy, the endoglycosidase- glycosyltransferases. However, recombinant EPO derived mediated enzymatic removal of N-glycans was engineered from P. pastoris contained two mannose residues linked to into GnTI-deficient HEK293 cells to generate N-GlcNAc- the single O-glycosylation site at Ser126 (Figure 5C) [84]. modified glycoproteins in the Golgi. Endogenous galactosyl- These O-mannose structures differ from mammalian transferase and sialyltransferase can further elongate the a-dystroglycan-type O-mannosylation and may cause N-linked GlcNAc leading to three truncated N-linked gly- unwanted side effects when present on recombinant glycopro- cans. Recombinant antibodies produced in the GlycoDelete teins. Plants can convert exposed proline residues adjacent to

1508 Expert Opin. Biol. Ther. (2015) 15(10) Using glyco-engineering to produce therapeutic proteins

Endo S α-fucosidase Endo S mutant

B. Deglycosylation Transglycosylation

Golgi

Endo T GaIT

ST Deglycosylation

Mannose GlcNAc N-acetylneuraminic acidGalactose Fucose

Figure 4. (A) In vitro chemoenzymatic synthesis of homogeneous glycoproteins [73].(B) In vivo GlycoDelete system [79]:

Recombinant glycoproteins are expressed in mammalian cells lacking N-acetylglucosaminyltransferase I. Man5GlcNAc1 is cleaved off by a specific endo-b-N-acetylglucosaminidase (Endo T) in the Golgi and the resulting N-linked GlcNAc may be further elongated by endogenous galactosyl- (GalT) and sialyltransferases (ST).

O-glycosylation sites of recombinant proteins into hydroxy- generated on N. benthamiana-derived human EPO-Fc [91]. proline residues [85], which can be further modified by arabi- A strategy to avoid the non-desired O-mannosylation of nosyltransferases leading to the presence of small arabinose yeast involves genetic engineering with expression of chains (Figure 5D) [86]. an a-mannosidase. This approach will generate a single O-linked mannose that can either be further elongated with 6. O-glycan engineering mammalian-type modifications such as the generation of a- dystroglycan-type O-glycans [93]. Alternatively, the mannose In mammals, control of mucin-type O-glycosylation initia- residues can be removed by in vitro digestion of P. pastoris- tion and elongation are quite complex as a defined consensus produced glycoproteins with recombinant lysosomal manno- sequence for attachment of the first monosaccharide has not sidases [94] or the generation of O-linked mannose chains been established and the factors that contribute to site-specific may be prevented by inhibition of the involved protein O-glycosylation are not well understood [80]. The human O-mannosyltransferases (Figure 5C). In a semi-synthetic polypeptide GalNAc-transferase family consists of 20 mem- O-glycan engineering approach, a polypeptide GalNAc-

Downloaded by [University of California, San Diego] at 09:13 17 May 2016 bers which are differentially expressed and display distinct as transferase was expressed in E. coli to initiate mucin-type well as overlapping peptide acceptor substrate specificities [87]. O-glycosylation on a recombinant protein [95].TheGalNAc- CHO cells produce mainly core 1 structures that can be engi- containing proteins were further modified in vitro by a neered for the production of defined and elongated struc- glycan-based PEGylation technique to generate proteins tures [88,89]. Strategies to eliminate O-glycan heterogeneity with increased half-life. A similar strategy has been used to include expression of the recombinant glycoprotein in attach PEG-modified sialic acid to the O-glycan of recombi- non-mammalian hosts that cannot perform mucin-type nant blood factor FVIII produced in mammalian cells [96]. O-glycosylation [90-92] or systematic elimination of the enzymes that initiate or elongate O-glycans in mammalian 7. Glyco-engineered therapeutic cells (Figure 5B) [35]. glycoproteins De novo O-glycan engineering has been achieved in yeast and plants. The mucin-type core 1 O-glycan (T-antigen) In recent years, the pharmaceutical industry has put consider- was generated on a peptide derived from human mucin in able effort into the development of novel glycoprotein thera- S. cerevisiae [90]. The machinery for O-linked GalNAc peutics. Table 1 provides an overview of glyco-engineered formation has been successfully expressed in plants [92] and monoclonal antibodies that are approved or in clinical trials di-sialylated core 1 O-glycans (Figure 5A) have been efficiently and have been generated using different technologies.

Expert Opin. Biol. Ther. (2015) 15(10) 1509 M. Dicker & R. Strasser

A. M G A

Golgi

C C E

C O

Figure 5. (A) Human mucin-type O-glycan biosynthesis pathway. Only the enzymes involved in the first biosynthesis steps are shown [80].(B) Strategies for O-glycan engineering in mammalian cells: to produce homogeneous mucin-type O-glycans consisting of O-linked GalNAc without any further extensions (Tn antigen) a knockout of core 1 b1,3-galactosyltransferase (C1GALT1) or its chaperone COSMC can be performed; knockout of individual polypeptide GalNAc-transferases may completely prevent mucin-type O-glycan formation. (C) In yeast transfer and elongation of O-linked mannose can be avoided by knockout or inhibition of protein O-mannosyltransferases (PMTs). (D) In plants, hydroxyproline formation and subsequent transfer of arabinose residues (Hyp) may be abolished by inactivation of the respective prolyl-4-hydroxylases (P4H).

Glyco-engineering of cells for IgG1 expression has focused demonstrate the impact of glycosylation and highlight the Downloaded by [University of California, San Diego] at 09:13 17 May 2016 mainly on the elimination of core-fucose from the N-glycan potential of glyco-engineered mAbs for different applications at Asn297 in the Fc region of the heavy chain. The absence (Table 1) in humans. As a consequence, the number of of core-fucose increases the affinity for FcgRIII receptor bind- glyco-engineered mAbs approved for different treatments is ing leading to improved antibody-dependent cellular cytotox- expected to increase in the near future [43]. Apart from icity on natural killer cells [17,97]. A similar glycosylation- modulating effector functions, future glycosylation remodel- dependent mechanism has an impact on antibody-dependent ing strategies will also focus on the engineering of mAbs cellular phagocytosis by macrophages [98] and influences the for enhanced anti-inflammatory properties [66,67]. IgG receptor-mediated effector function of virus-neutralizing anti- glycoforms with high amounts of terminal sialic acid can bodies [99]. The ZMAPP antibody cocktail for treatment of display increased affinity for lectin-type receptors and may Ebola virus infections and reversion of the disease [61] contains represent another emerging field for the development of three plant-produced antibodies that lack core-fucose resi- next-generation antibodies. Apart from the engineered anti- dues. In mice, the fucose-free monoclonal antibody body therapeutics, recombinant glycoprotein drugs with 13F6 which is one of the ZMAPP components displayed enhanced half-life (similar to darboetin alfa), entirely human- clearly enhanced potency against Ebola virus compared to ized glycosylation (e.g., follicle-stimulating hormone) as well 13F6 variants with core fucose [100]. These examples clearly as different recombinant products for enzyme replacement

1510 Expert Opin. Biol. Ther. (2015) 15(10) Using glyco-engineering to produce therapeutic proteins

Table 1. Examples of glyco-engineered antibodies that are approved or in clinical studies.

Product Company and production Glyco-engineered modification Status platform

Mogamulizumab Kyowa Hakko Kirin Core fucose-deficient CHO cells Approved (anti-CCR4) Potteligent technology FUT8 knockout Obinutuzumab Glycart-Roche Core fucose-reduced CHO cells Approved (anti-CD20) GlycoMab technology GnTIII overexpression ZMAPP MAPP Biopharmaceutical Core fucose and b1,2-xylose-deficient Nicotiana benthamiana In clinical (anti-Ebola RNAi of xylosyl-/fucosyltransferases trials virus) Ublituximab TG Therapeutics Expression in YB2/0 rat hybridoma cells (low amount of FUT8) In clinical (anti-CD20) LFB Biotechnologies trials EMABling technology Roledumab LFB Biotechnologies Expression in YB2/0 rat hybridoma cells (low amount of FUT8) In clinical (anti-RhD) EMABling technology trials SEA-CD40 Seattle Genetics Inhibition of core fucose incorporation by addition of modified In clinical (anti-CD40) SEA technology sugars to the culture medium trials Benralizumab Kyowa Hakko Kirin Core fucose-deficient CHO cells In clinical (anti-IL-5Ra) MedImmune FUT8 knockout trials Potteligent technology MEDI-551 Kyowa Hakko Kirin Core fucose-deficient CHO cells In clinical (anti-CD19) MedImmune FUT8 knockout trials Potteligent technology BIW-8962 Kyowa Hakko Kirin Potteligent Core fucose-deficient CHO cells In clinical (anti-GM2) technology FUT8 knockout trials KHK2804 Kyowa Hakko Kirin Core fucose-deficient CHO cells In clinical (anti-tumor specific Teva FUT8 knockout trials antigen) Potteligent technology KHK2823 Kyowa Hakko Kirin Potteligent Core fucose-deficient CHO cells In clinical (anti-CD123) technology FUT8 knockout trials KHK2898 Kyowa Hakko Kirin Potteligent Core fucose-deficient CHO cells In clinical (anti-CD98) technology FUT8 knockout trials KHK4083 Kyowa Hakko Kirin Potteligent Core fucose-deficient CHO cells In clinical (immunomodulator) technology FUT8 knockout trials Ecromeximab Kyowa Hakko Kirin Core fucose-deficient CHO cells In clinical (anti-GD3) Life Science Pharmaceuticals FUT8 knockout trials Potteligent technology Lumretuzumab Glycart-Roche Core fucose-reduced CHO cells In clinical (anti-HER3) GlycoMab technology GnTIII overexpression trials PankoMab Glycotope Human cell lines In clinical (anti-TA-MUC1) GlycoExpress technology low or no core fucose trials TrasGEX Glycotope Human cell lines In clinical (anti-HER2) GlycoExpress technology low or no core fucose trials CetuGEX Glycotope Human cell lines In clinical (anti-EGFR) GlycoExpress technology Low or no core fucose trials Downloaded by [University of California, San Diego] at 09:13 17 May 2016

CCR4: CC chemokine receptor 4; FUT8: Core a1,6-fucosyltransferase; GnTIII: N-acetylglucosaminyltransferase III; HER: Human epidermal growth factor; IL-5Ra: IL-5 receptor a; MUC1: Mucin 1; RhD: Rhesus D antigen.

therapy (like a-galactosidase A) are in the pipelines of hosts for custom-made expression of recombinant glycopro- companies. tein pharmaceuticals. These glyco-engineering approaches will allow easier production of biosimilar drugs (same homo- 8. Conclusion geneous glycosylation) and lead to the generation of bio-better pharmaceuticals with altered glycosylation and optimized Despite the fact that we still do not understand all factors that efficacy and safety. influence and control glycosylation in eukaryotic cells, major advances in engineering of expression hosts towards homoge- 9. Expert opinion neous glycosylation patterns have been achieved in the past. The use of genome editing tools, extensive integration of ana- The importance of glycosylation for recombinant glycopro- lytical data, and a better understanding of cellular processes tein biopharmaceuticals is more and more recognized and will facilitate the development of next-generation expression enormous efforts are currently undertaken in academic groups

Expert Opin. Biol. Ther. (2015) 15(10) 1511 M. Dicker & R. Strasser

and industry to provide robust solutions for controlled glyco- may allow the development of strategies for site-specific sylation. Recent advances in engineering of cell-based expres- remodeling of N-glycans. Structure-guided protein engineer- sion platforms demonstrate that rather uniform and ing to alter the substrate specificity of glycosylation enzymes customized N-glycan structures can be made in different sys- like the OST catalytic subunit or glycan processing enzymes tems including mammalian cells (human and non-human), may provide additional tools required for site-specific engi- yeast, insect cells, and plants. By employing genome editing neering in cell-based production systems [72]. Alternatively, tools like CRISPR/Cas [35], existing differences and shortcom- site-specific modifications may be accomplished by advanced ings of species and cell-types can be eliminated and numerous chemoenzymatic remodeling or de novo synthesis of glycopro- remodeling tools for the generation of defined N-glycans on teins with precise control of attached glycan structures. In the specific glycoproteins (e.g., monoclonal antibodies, EPO) latter case, it remains to be shown whether such sophisticated are already available in different expression platforms. The and cost-intensive approaches will be of interest for the bio- ultimate goal will be the establishment of manufacturing plat- pharmaceutical industry or remain the field of specialized forms for the production of recombinant glycoprotein thera- laboratories. peutics with homogeneous custom-made N-glycan structures Another challenge for the future will be the coordination of that are designed for desired activities. Currently, such ‘on N- and O-glycan engineering attempts. So far there are only demand’ N-glycans include highly sialylated tetraantennary proof-of-concept studies that have shown the feasibility of complex N-glycans (Figure 3A), terminally sialylated bianten- extensive remodeling of both glycosylation pathways. Due to nary N-glycans (Figure 3C and D), and fucose-deficient bian- the requirement for the simultaneous engineering of two tennary N-glycans (Figure 3B, F and G). In addition, the pathways which act in the same subcellular compartment potential to produce homogeneous oligomannosidic struc- and utilize the same nucleotide sugar donors, the control tures like Man8GlcNAc2 (Figure 3E) and defined O-linked over these modifications under manufacturing conditions glycans (Figure 5A) is of interest. The means to produce glyco- will be a great challenge. A better understanding of cell biol- proteins with defined glycans will drive progress in the under- ogy like cargo transport processes through the Golgi and standing of a glycan’s function leading to novel targets for development of tools to channel nucleotide sugar substrates glyco-engineering of glycoprotein therapeutics. and processing enzymes will be fundamental for more complex approaches that interfere with different cellular processes. 10. So what are the challenges for the Ultimately, the experimentally obtained parameters need to future? be incorporated into robust models to accurately predict the N- and O-glycan profiles of recombinant glycoprotein bio- We still do not understand all pathways leading to glycan pharmaceuticals. Given the great potential that glycoprotein diversity as seen in most eukaryotic cells and our knowledge therapeutics with defined glycans offers, the biopharmaceuti- of essential enzymes involved in initiation of N- and cal industry should extend their efforts in glyco-engineering O-glycosylation (OST complex and polypeptide GalNAc- to spur the development of next-generation drugs with transferases) is limited. In addition, it is still difficult to tailored properties and function. produce N-glycans with long terminal elongations like polysialylation which might significantly contribute to half- life extension. Such extensive modifications might depend Declaration of interest on so far unknown factors that regulate the retention or contact time in the Golgi. A better characterization of cellular M Dicker and glyco-engineering work in the Strasser group

Downloaded by [University of California, San Diego] at 09:13 17 May 2016 factors like Golgi organization and cargo transport processes were supported by a grant from the Austrian Federal Ministry will be essential to overcome some of the current limitations. of Transport, Innovation, and Technology (bmvit) and Moreover, site-specific N-glycosylation and N-glycan process- Austrian Science Fund (FWF): TRP 242-B20. M Dicker ing will be difficult to achieve using available in vivo produc- and R Strasser are employees of University of Natural Resour- tion systems. More research is needed to understand the ces and Life Sciences, Vienna, Austria. The authors have no protein intrinsic factors that influence the relationship of a other relevant affiliations or financial involvement with any particular glycan with the polypeptide sequence and structure organization or entity with a financial interest in or financial as well as the contribution of other still unknown aspects of conflict with the subject matter or materials discussed in the site-specific glycosylation. Together with other experimental manuscript. This includes employment, consultancies, hono- techniques, molecular dynamics simulations will be highly raria, stock ownership or options, expert testimony, grants suitable to examine carbohydrate-- protein interactions and or patents received or pending, or royalties.

1512 Expert Opin. Biol. Ther. (2015) 15(10) Using glyco-engineering to produce therapeutic proteins

Bibliography Papers of special note have been highlighted as 11. North SJ, Huang HH, Sundaram S, produced recombinant proteins. either of interest () or of considerable interest et al. Glycomics profiling of Chinese Curr Opin Biotechnol 2014;30C:95-100 () to readers. hamster ovary cell glycosylation mutants 22. Stanley P. Chinese hamster ovary cell reveals N-glycans of a novel size and 1. Aebi M. N-linked protein glycosylation mutants with multiple glycosylation complexity. J Biol Chem in the ER. Biochim Biophys Acta defects for production of glycoproteins 2010;285:5759-75 2013;1833:2430-7 with minimal carbohydrate heterogeneity. 12. Bennun SV, Yarema KJ, Betenbaugh MJ, Mol Cell Biol 1989;9:377-83 2. Zielinska DF, Gnad F, Schropp K, et al. et al. Integration of the transcriptome Mapping N-glycosylation sites across 23. Wright A, Sato Y, Okada T, et al. In and glycome for identification of glycan seven evolutionarily distant species reveals vivo trafficking and catabolism of cell signatures. PLoS Comput Biol a divergent substrate proteome despite a IgG1 antibodies with Fc associated 2013;9:e1002813 common core machinery. Mol Cell carbohydrates of differing structure. 2012;46:542-8 13. Spahn PN, Lewis NE. Systems Glycobiology 2000;10:1347-55 glycobiology for glycoengineering. 3. Shrimal S, Gilmore R. Glycosylation of 24. Yoo EM, Yu LJ, Wims LA, et al. Curr Opin Biotechnol 2014;30:218-24 closely spaced acceptor sites in human Differences in N-glycan structures found . Comprehensive summary of systems glycoproteins. J Cell Sci on recombinant IgA1 and IgA2 produced glycobiology approaches aiming at 2013;126:5513-23 in murine myeloma and CHO cell lines. prediction of product glycosylation. MAbs 2010;2:320-34 4. Shrimal S, Cherepanova NA, Gilmore R. 14. Jedrzejewski PM, del Val IJ, . Comparison of N-glycan profiles found Cotranslational and posttranslocational Constantinou A, et al. Towards on IgA subtypes expressed in murine N-glycosylation of proteins in the controlling the glycoform: a model myeloma and CHO cells. endoplasmic reticulum. Semin Cell framework linking extracellular Dev Biol 2015;41:71-8 25. Patterson G, Hirschberg K, metabolites to antibody glycosylation. Polishchuk R, et al. Transport through 5. Shrimal S, Gilmore R. Reduced Int J Mol Sci 2014;15:4492-522 the Golgi apparatus by rapid partitioning expression of the oligosaccharyltransferase 15. Hossler P, Mulukutla BC, Hu WS. within a two-phase membrane system. exacerbates protein hypoglycosylation in Systems analysis of N-glycan processing Cell 2008;133:1055-67 cells lacking the fully assembled in mammalian cells. PLoS ONE oligosaccharide donor. Glycobiology 26. Beznoussenko GV, Parashuraman S, 2007;2:e713 2015;25(7):774-83 Rizzo R, et al. Transport of soluble 16. Moremen KW, Tiemeyer M, Nairn AV. proteins through the Golgi occurs by 6. Choi BK, Warburton S, Lin H, et al. Vertebrate protein glycosylation: diffusion via continuities across cisternae. Improvement of N-glycan site occupancy diversity, synthesis and function. Nat Rev Elife 2014;3:doi: 10.7554/eLife.02009 of therapeutic glycoproteins produced in Mol Cell Biol 2012;13:448-62 Pichia pastoris. 27. Nabi IR, Dennis JW. The extent of Appl Microbiol Biotechnol 17. Uman˜a P, Jean-Mairet J, Moudry R, polylactosamine glycosylation of MDCK 2012;95:671-82 et al. Engineered glycoforms of an LAMP-2 is determined by its Golgi antineuroblastoma IgG1 with optimized residence time. Glycobiology 7. Elliott S, Lorenzini T, Asher S, et al. antibody-dependent cellular cytotoxic 1998;8:947-53 Enhancement of therapeutic protein activity. Nat Biotechnol 1999;17:176-80 in vivo activities through 28. Glick BS, Luini A. Models for Golgi . Describes the GlycArt technology for glycoengineering. Nat Biotechnol traffic: a critical assessment. Cold Spring the reduction of core fucose 2003;21:414-21 Harb Perspect Biol 2011;3:a005215 on antibodies. 8. Egrie JC, Browne JK. Development and 29. Lavieu G, Dunlop MH, Lerich A, et al. 18. Goede V, Fischer K, Busch R, et al. characterization of novel erythropoiesis The Golgi ribbon structure facilitates Downloaded by [University of California, San Diego] at 09:13 17 May 2016 Obinutuzumab plus chlorambucil in stimulating protein (NESP). Br J Cancer anterograde transport of large cargoes. patients with CLL and coexisting 2001;84:3-10 Mol Biol Cell 2014;25:3028-36 conditions. N Engl J Med 30. D’Angelo G, Uemura T, Chuang CC, 9. Song R, Oren DA, Franco D, et al. 2014;370:1101-10 Strategic addition of an N-linked glycan et al. Vesicular and non-vesicular 19. Zhang M, Koskie K, Ross JS, et al. to a monoclonal antibody improves its transport feed distinct glycosylation Enhancing glycoprotein sialylation by HIV-1-neutralizing activity. pathways in the Golgi. Nature targeted gene silencing in mammalian Nat Biotechnol 2013;31:1047-52 2013;501:116-20 cells. Biotechnol Bioeng . This elegant study demonstrates the 31. Thaysen-Andersen M, Packer NH. 2010;105:1094-105 power of protein glyco-engineering. Site-specific glycoproteomics confirms 20. Hamilton SR, Davidson RC, 10. Nairn AV, York WS, Harris K, et al. that protein structure dictates formation Sethuraman N, et al. Humanization of Regulation of glycan structures in animal of N-glycan type, core fucosylation and yeast to produce complex terminally tissues: transcript profiling of glycan- branching. Glycobiology sialylated glycoproteins. Science related genes. J Biol Chem 2012;22:1440-52 2006;313:1441-3 .. 2008;283:17298-313 This study analyzed N-glycan profiles 21. Strasser R, Altmann F, Steinkellner H. from 117 puplished papers to Controlled glycosylation of plant- investigate how the individual protein

Expert Opin. Biol. Ther. (2015) 15(10) 1513 M. Dicker & R. Strasser

structures affect the N-glycan type, specific for galactose-alpha-1,3-galactose. 51. Altmann F, Schwihla H, Staudacher E, core fucosylation, and branching. N Engl J Med 2008;358:1109-17 et al. Insect cells contain an unusual, 32. Wormald MR, Dwek RA. Glycoproteins: 42. Reeves PJ, Callewaert N, Contreras R, membrane-bound glycan presentation and protein-fold et al. Structure and function in beta-N-acetylglucosaminidase probably stability. Structure 1999;7:R155-60 rhodopsin: high-level expression of involved in the processing of protein N-glycans. J Biol Chem 33. Stadlmann J, Pabst M, Kolarich D, et al. rhodopsin with restricted and 1995;270:17344-9 Analysis of immunoglobulin glycosylation homogeneous N-glycosylation by a by LC-ESI-MS of glycopeptides and tetracycline-inducible 52. Shi X, Jarvis DL. Protein N-glycosylation oligosaccharides. Proteomics N-acetylglucosaminyltransferase in the baculovirus-insect cell system. 2008;8:2858-71 I-negative HEK293S stable mammalian Curr Drug Targets 2007;8:1116-25 cell line. Proc Natl Acad Sci USA 34. Ayoub D, Jabs W, Resemann A, et al. 53. Palmberger D, Wilson IB, Berger I, et al. 2002;99:13419-24 Correct primary structure assessment and SweetBac: a new approach for the extensive glyco-profiling of cetuximab by 43. Beck A, Reichert JM. Marketing production of mammalianised a combination of intact, middle-up, approval of mogamulizumab: a triumph glycoproteins in insect cells. PLoS One middle-down and bottom-up ESI and for glyco-engineering. MAbs 2012;7:e34226 MALDI mass spectrometry techniques. 2012;4:419-25 54. Mabashi-Asazuma H, Kuo CW, MAbs 2013;5:699-710 44. Croset A, Delafosse L, Gaudry JP, et al. Khoo KH, et al. A novel baculovirus 35. Steentoft C, Bennett EP, Differences in the glycosylation of vector for the production of Schjoldager KT, et al. Precision genome recombinant proteins expressed in HEK nonfucosylated recombinant editing: a small revolution for and CHO cells. J Biotechnol glycoproteins in insect cells. Glycobiology glycobiology. Glycobiology 2012;161:336-48 2014;24:325-40 2014;24:663-80 45. Chiba Y, Suzuki M, Yoshida S, et al. 55. Altmann F. The role of protein 36. Ghaderi D, Taylor RE, Production of human compatible high glycosylation in allergy. Int Arch Padler-Karavani V, et al. Implications of mannose-type (Man5GlcNAc2) sugar Allergy Immunol 2007;142:99-115 the presence of N-glycolylneuraminic chains in Saccharomyces cerevisiae. 56. Cox K, Sterling J, Regan J, et al. Glycan acid in recombinant therapeutic J Biol Chem 1998;273:26298-304 optimization of a human monoclonal glycoproteins. Nat Biotechnol 46. Hamilton SR, Bobrowicz P, antibody in the aquatic plant Lemna 2010;28:863-7 Bobrowicz B, et al. Production of minor. Nat Biotechnol 2006;24:1591-7 .. Demonstration of the presence of complex human glycoproteins in yeast. 57. Scha¨hs M, Strasser R, Stadlmann J, et al. covalently bound Neu5Gc in Science 2003;301:1244-6 Production of a monoclonal antibody in .. cetuximab and generation of immune Using a combinatorial library plants with a humanized N-glycosylation in vitro complexes . approach, the first production of pattern. Plant Biotechnol J 37. Kannicht C, Ramstrom€ M, Kohla G, complex N-glycans in yeast is shown. 2007;5:657-63 et al. Characterisation of the 47. Jacobs PP, Geysens S, Vervecken W, 58. Strasser R, Stadlmann J, Scha¨hs M, et al. post-translational modifications of a et al. Engineering complex-type Generation of glyco-engineered Nicotiana novel, human cell line-derived N-glycosylation in Pichia pastoris using benthamiana for the production of recombinant human factor VIII. GlycoSwitch technology. Nat Protoc monoclonal antibodies with a Thromb Res 2013;131:78-88 2009;4:58-70 homogeneous human-like N-glycan 38. Shahrokh Z, Royle L, Saldova R, et al. 48. Parsaie Nasab F, Aebi M, Bernhard G, structure. Plant Biotechnol J Erythropoietin produced in a human cell et al. A combined system for engineering 2008;6:392-402 line (Dynepo) has significant differences glycosylation efficiency and glycan 59. Koprivova A, Stemmer C, Altmann F, in glycosylation compared with structure in Saccharomyces cerevisiae. Downloaded by [University of California, San Diego] at 09:13 17 May 2016 et al. Targeted knockouts of erythropoietins produced in CHO cell Appl Environ Microbiol Physcomitrella lacking plant-specific lines. Mol Pharm 2011;8:286-96 2013;79:997-1007 immunogenic N-glycans. 39. Ghaderi D, Zhang M, Hurtado-Ziola N, 49. De Pourcq K, Tiels P, Van Hecke A, Plant Biotechnol J 2004;2:517-23 et al. Production platforms for et al. Engineering Yarrowia lipolytica to 60. Loos A, Gruber C, Altmann F, et al. biotherapeutic glycoproteins. Occurrence, produce glycoproteins homogeneously Expression and glycoengineering of impact, and challenges of non-human modified with the universal functionally active heteromultimeric IgM sialylation. Biotechnol Genet Eng Rev Man3GlcNAc2 N-glycan core. in plants. Proc Natl Acad Sci USA 2012;28:147-75 PLoS One 2012;7:e39976 2014;111:6263-8 40. Bosques CJ, Collins BE, Meador JW, 50. Wang H, Song HL, Wang Q, et al. .. The assembly of pentameric and et al. Chinese hamster ovary cells can Expression of glycoproteins bearing hexameric glyco-engineered forms of produce galactose-a-1,3-galactose complex human-like glycans with IgM is demonstrated using a Nicotiana antigens on proteins. Nat Biotechnol galactose terminal in Hansenula benthamiana plant system. 2010;28:1153-6 polymorpha. World J 61. Qiu X, Wong G, Audet J, et al. 41. Chung CH, Mirakhur B, Chan E, et al. Microbiol Biotechnol 2013;29:447-58 Reversion of advanced Ebola virus disease Cetuximab-induced anaphylaxis and IgE

1514 Expert Opin. Biol. Ther. (2015) 15(10) Using glyco-engineering to produce therapeutic proteins

in nonhuman primates with ZMapp. eukaryotic protein glycosylation pathway 83. Gaunitz S, Jin C, Nilsson A, et al. Nature 2014;514:47-53 in Escherichia coli. Nat Chem Biol Mucin-type proteins produced in the . ZMappTM is composed of three 2012;8:434-6 Trichoplusia ni and Spodoptera “humanized” monoclonal antibodies 72. Ollis AA, Zhang S, Fisher AC, et al. frugiperda insect cell lines carry novel manufactured in glyco-engineered Engineered oligosaccharyltransferases with O-glycans with phosphocholine and Nicotiana benthamiana plants. greatly relaxed acceptor-site specificity. sulfate substitutions. Glycobiology 62. Castilho A, Windwarder M, Gattinger P, Nat Chem Biol 2014;10:816-22 2013;23:778-96 et al. Proteolytic and N-Glycan 73. Wang LX, Amin MN. Chemical and 84. Gong B, Burnina I, Stadheim TA, et al. Processing of Human a1-Antitrypsin chemoenzymatic synthesis of Glycosylation characterization of Expressed in Nicotiana benthamiana. glycoproteins for deciphering functions. recombinant human erythropoietin Plant Physiol 2014;166:1839-51 Chem Biol 2014;21:51-66 produced in glycoengineered Pichia pastoris by mass spectrometry. 63. Tekoah Y, Tzaban S, Kizhner T, et al. 74. Wang P, Dong S, Shieh JH, et al. J Mass Spectrom 2013;48:1308-17 Glycosylation and functionality of Erythropoietin derived by chemical recombinant b-glucocerebrosidase from synthesis. Science 2013;342:1357-60 85. Parsons J, Altmann F, Graf M, et al. various production systems. Biosci Rep A gene responsible for prolyl- 75. Washburn N, Schwab I, Ortiz D, et al. 2013;33:doi: 10.1042/BSR20130081 hydroxylation of moss-produced Controlled tetra-Fc sialylation of IVIg recombinant human erythropoietin. 64. Mabashi-Asazuma H, Shi X, Geisler C, results in a drug candidate with Sci Rep 2013;3:3019 et al. Impact of a human CMP-sialic consistent enhanced anti-inflammatory acid transporter on recombinant activity. Proc Natl Acad Sci USA 86. Karnoup AS, Turkelson V, glycoprotein sialylation in 2015;112:E1297-306 Anderson WH. O-linked glycosylation in glycoengineered insect cells. Glycobiology maize-expressed human IgA1. 76. Zimran A, Elstein D, Levy-Lahad E, 2013;23:199-210 Glycobiology 2005;15:965-81 et al. Replacement therapy with 65. Castilho A, Neumann L, Gattinger P, imiglucerase for type 1 Gaucher’s disease. 87. Kong Y, Joshi HJ, Schjoldager KT, et al. et al. Generation of biologically active Lancet 1995;345:1479-80 Probing polypeptide GalNAc-transferase multi-sialylated recombinant human isoform substrate specificities by in vitro 77. Huang W, Giddens J, Fan SQ, et al. EPOFc in plants. PLoS One analysis. Glycobiology 2015;25:55-65 Chemoenzymatic glycoengineering of 2013;8:e54836 intact IgG antibodies for gain of 88. Lindberg L, Liu J, Gaunitz S, et al. 66. Kaneko Y, Nimmerjahn F, Ravetch JV. functions. J Am Chem Soc Mucin-type fusion proteins with blood Anti-inflammatory activity of 2012;134:12308-18 group A or B determinants on defined immunoglobulin G resulting from Fc O-glycan core chains produced in 78. Ahmed AA, Giddens J, Pincetic A, et al. sialylation. Science 2006;313:670-3 glycoengineered Chinese hamster ovary Structural characterization of anti- cells and their use as immunoaffinity 67. Raju TS, Lang SE. Diversity in structure inflammatory immunoglobulin G Fc matrices. Glycobiology 2013;23:720-35 and functions of antibody sialylation in proteins. J Mol Biol 2014;426:3166-79 the Fc. Curr Opin Biotechnol 89. Liu J, Jin C, Cherian RM, et al. 79. Meuris L, Santens F, Elson G, et al. 2014;30:147-52 O-glycan repertoires on a mucin-type GlycoDelete engineering of mammalian reporter protein expressed in CHO cell 68. Yamane-Ohnuki N, Kinoshita S, cells simplifies N-glycosylation of pools transiently transfected with Inoue-Urakubo M, et al. Establishment recombinant proteins. Nat Biotechnol O-glycan core enzyme cDNAs. of FUT8 knockout Chinese hamster 2014;32:485-9 J Biotechnol 2015;199:77-89 ovary cells: an ideal host cell line for .. This study describes a novel approach . Describes the effect of transiently producing completely defucosylated to reduce the glycan heterogeneity of expressed O-glycan core chain antibodies with enhanced antibody- recombinant proteins by simplification dependent cellular cytotoxicity. glycosyltransferases on O-glycan Downloaded by [University of California, San Diego] at 09:13 17 May 2016 of N-glycan processing steps. Biotechnol Bioeng 2004;87:614-22 biosynthesis pathways in CHO cells. 80. Bennett EP, Mandel U, Clausen H, et al. . Provides a description of the 90. Amano K, Chiba Y, Kasahara Y, et al. Control of mucin-type O-glycosylation: Potteligent technology. Engineering of mucin-type human A classification of the polypeptide glycoproteins in yeast cells. Proc Natl 69. Lau KS, Partridge EA, Grigorian A, et al. GalNAc-transferase gene family. Acad Sci USA 2008;105:3232-7 Complex N-glycan number and degree Glycobiology 2012;22:736-56 of branching cooperate to regulate cell 91. Castilho A, Neumann L, Daskalova S, 81. Bakema JE, van Egmond M. proliferation and differentiation. Cell et al. Engineering of Sialylated Mucin- Immunoglobulin A: A next generation of 2007;129:123-34 type O-Glycosylation in Plants. therapeutic antibodies? MAbs 70. Baker JL, C¸ elik E, DeLisa MP. J Biol Chem 2012;287:36518-26 2011;3:352-61 . Expanding the glycoengineering toolbox: Demonstrates the de novo generation 82. Taschwer M, Hackl M, the rise of bacterial N-linked protein of sialylated core 1 O-glycans on Herna´ndez Bort JA, et al. Growth, glycosylation. Trends Biotechnol recombinant EPO-Fc produced productivity and protein glycosylation in 2013;31:313-23 in plants. a CHO EpoFc producer cell line adapted 92. Yang Z, Drew DP, Jørgensen B, et al. 71. Valderrama-Rincon JD, Fisher AC, to glutamine-free growth. J Biotechnol Engineering mammalian mucin-type Merritt JH, et al. An engineered 2012;157:295-303

Expert Opin. Biol. Ther. (2015) 15(10) 1515 M. Dicker & R. Strasser

O-glycosylation in plants. J Biol Chem demonstrates full efficacy and prolonged activity of monoclonal antibody 2G12. 2012;287:11911-23 effect in hemophilic mice models. Blood J Immunol 2010;185:6876-82 93. Hamilton SR, Cook WJ, 2013;121:2108-16 100. Zeitlin L, Pettitt J, Scully C, et al. Gomathinayagam S, et al. Production of 97. Ferrara C, Grau S, Ja¨ger C, et al. Unique Enhanced potency of a fucose-free sialylated O-linked glycans in Pichia carbohydrate-carbohydrate interactions monoclonal antibody being developed as pastoris. Glycobiology 2013;23:1192-203 are required for high affinity binding an Ebola virus immunoprotectant. . Generation of a-dystroglycan-type between FcgammaRIII and antibodies Proc Natl Acad Sci 2011;108:20690-4 O-glycans in yeast. lacking core fucose. Proc Natl Acad 94. Hopkins D, Gomathinayagam S, Sci USA 2011;108:12669-74 Affiliation .. † Hamilton SR. A practical approach for Insight into the role of specific glycans Martina Dicker1 & Richard Strasser 2 † O-linked mannose removal: the use of in receptor-mediated antibody- Author for correspondence recombinant lysosomal mannosidase. dependent cytotoxocity using 1PhD candidate, Appl Microbiol Biotechnol X-ray crystallography. University of Natural Resources and Life 2015;99:3913-27 98. Golay J, Da Roit F, Bologna L, et al. Sciences, Department of Applied Genetics and 95. Henderson GE, Isett KD, Gerngross TU. Glycoengineered CD20 antibody Cell Biology, Muthgasse 18, Vienna, Austria 2 Site-specific modification of recombinant obinutuzumab activates neutrophils and Associate Professor, Group Leader, proteins: a novel platform for modifying mediates phagocytosis through CD16B University of Natural Resources and Life Sciences, Department of Applied Genetics and glycoproteins expressed in E. more efficiently than rituximab. Blood coli. Bioconjug Chem 2011;22:903-12 2013;122:3482-9 Cell Biology, Muthgasse 18, Vienna, Austria 99. Forthal DN, Gach JS, Landucci G, et al. Tel: +43 1 47654 6705; 96. Stennicke HR, Kjalke M, Karpf DM, Fax: +43 1 47654 6392; et al. A novel B-domain Fc-glycosylation influences Fcg receptor E-mail: [email protected] O-glycoPEGylated FVIII (N8-GP) binding and cell-mediated anti-HIV Downloaded by [University of California, San Diego] at 09:13 17 May 2016

1516 Expert Opin. Biol. Ther. (2015) 15(10)

Advances in the production of human therapeutic proteins in yeasts and filamentous fungi Gerngross TU, Nat. Biotech, 22:1409 (2004) REVIEW

Advances in the production of human therapeutic proteins in yeasts and filamentous fungi

Tillman U Gerngross

Yeast and fungal protein expression systems are used for the production of many industrially relevant enzymes, and are widely used by the research community to produce proteins that cannot be actively expressed in Escherichia coli or require glycosylation for proper folding and biological activity. However, for the production of therapeutic glycoproteins intended for use in humans, yeasts have been less useful because of their inability to modify proteins with human glycosylation structures. Yeast glycosylation http://www.nature.com/naturebiotechnology is of the high-mannose type, which confers a short in vivo half-life to the protein and may render it less efficacious or even immunogenic. Several ways of humanizing yeast-derived glycoproteins have been tried, including enzymatically modifying proteins in vitro and modulating host glycosylation pathways in vivo. Recent advances in the glycoengineering of yeasts and the expression of therapeutic glycoproteins in humanized yeasts have shown significant promise, and are challenging the current dominance of therapeutic protein production based on mammalian cell culture.

Protein-based therapeutics are emerging as the largest class of new recombinant protein, together with the ease of scale-up, have made chemical entities being developed by the drug industry1.Unlike these hosts the system of choice for producing many industrial small molecules, which typically are synthesized by chemical means, enzymes, where cost of goods is a primary concern2.The research most proteins are sufficiently complex to necessitate their produc- community has relied on yeasts for the production of recombinant tion in living systems, mostly by recombinant DNA technology. proteins that cannot be obtained from E. coli because of folding As such, the choice of recombinant expression hosts has been the problems or the requirement for glycosylation. Yeasts have been © 2004 Nature Publishing Group subject of ongoing evaluation and much effort has been directed at used to express proteins at very high protein titers, including developing novel protein expression systems with improved char- mammalian proteins, as exemplified by the production of 14.8 g/l of acteristics. The five primary metrics for evaluating such novel gelatin in Pichia pastoris3.Expression kits based on yeasts are avail- protein expression hosts are: (i) the cost of manufacturing and puri- able from almost all major research tool companies and their ease fication (ii) the ability to control the final product including its post- of use has made them an attractive solution for many protein translational processing (iii) the time required from gene to purified expression needs. protein (iv), the regulatory path to approve a drug produced on a However, the choice of protein expression system is most often dic- given expression platform and (v) the overall royalties associated tated by the ultimate use of the product and not necessarily the ease of with the production of a recombinant product in a given host. initial production. Proteins intended for use in humans impose the Yeasts and filamentous fungi have been extensively used in the most significant challenges in this context, and the aberrant nature or industrial enzyme industry for the production of recombinant pro- absence of post-translational processing associated with a given host, teins—and most companies (DSM (Heerlen, The Netherlands), often becomes an overriding consideration. In this article, we review Genencor (Palo Alto, CA, USA), Novozymes (Bagsveard, Denmark) the challenges involved in the production of therapeutic glycopro- and others) have developed large-scale fermentation capacity teins and the current state of development of engineered yeast strains around yeasts and/or fungi. The ability of these organisms to grow that are able to produce humanized glycoproteins. in chemically defined medium in the absence of animal-derived growth factors (e.g., calf serum) and to secrete large amounts of Glycosylation of therapeutic proteins There are approximately 140 therapeutic proteins approved in the United States and Europe, and an additional 500 in clinical trials1; Thayer School of Engineering, the Department of Biological Sciences and the with an even larger number in preclinical development. The thera- Department of Chemistry, Dartmouth College, Hanover, New Hampshire 03755, USA and GlycoFi Inc., 21 Lafayette Street, Lebanon, New Hampshire 03766, peutic protein market can be roughly divided into two segments— USA. Correspondence should be addressed to T.U.G. proteins that are not post-translationally modified and those that ([email protected]). require post-translational processing (mostly N-glycosylation) to Published online 4 November 2004; doi:10.1038/nbt1028 attain full biological function. Nonglycosylated proteins are typi-

NATURE BIOTECHNOLOGY VOLUME 22 NUMBER 11 NOVEMBER 2004 1409 REVIEW

a It has to be recognized that glycoproteins produced by mammalian cell culture are inherently heterogenous mixtures of glycoforms, and that scale-up and regulatory approval have been a matter of maintaining a reproducible composition and ratio of glycoforms. Because it is well understood that certain glycoforms are more active than others9, the relative distribution of individual glycoforms has to be kept constant during preclinical manufacturing and post- approval production—often representing a major bioprocessing challenge. A purified protein preparation derived from a mammalian cell culture process is essentially a mixture of individual drugs, each with its own pharmacokinetic, pharmacodynamic prop- b erties and efficacy profile10.For example, human erythropoietin (EPO), a glycoprotein that plays a major role in regulating the level of circulating erythrocytes, has found wide use in the treatment of anemia. Natural EPO contains three N-glycans (glycans linked to asparagine residues of the glycoprotein), which are critical to its therapeutic activity, and one O-glycosylation site which has no known function; removal of the N-glycans from EPO results in a protein with a very short half-life and virtually no erythropoietic function in vivo10.Whereas glycosylation of EPO appears to be essential, it is not sufficient to ensure full biological activity. Recent work has shown that the production of recombinant EPO is highly Figure 1 Protein structure of interferon-β illustrates the impact of sensitive to the production process. Different process conditions glycosylation on hydrodynamic volume; glycosylated interferon-β (top) and have been shown to yield material that has a fivefold difference in http://www.nature.com/naturebiotechnology glycosylated interferon-β with glycosylation highlighted in blue (bottom). erythropoietic function in vivo, and a detailed analysis of this effect Source: http://www.dkfz.de/spec/glycosciences.de/modeling/glyprot/php/ demonstrated that variations in the glycosylation are responsible for main.php these differences11.Recent work by Erbayraktar and colleagues has shown that by modifying the glycosylation of EPO, one can create a molecule that entirely lacks erythropoietic function while retaining cally expressed in either E. coli or the yeast Saccharomyces cerevisiae the neuroprotective function of the protein12. and currently constitute about 40% of the therapeutic protein mar- Some therapeutic proteins require specific glycosylation to ensure ket with an annual growth rate of about 12%4.Some aglycosylated proper cellular targeting. For example, glucocerebrosidase, the proteins are expressed in both systems, for example, recombinant enzyme used for replacement therapy in patients with Gaucher dis- insulin is produced in E. coli (Humulin, Eli Lilly, Indianapolis, IN, ease, is based on a recombinant enzyme produced in mammalian USA) and yeast (Novolog, Novo Nordisk, Bagsvaerd, Denmark). cells7.However, the therapeutically active form of glucocerebrosi- The production of recombinant proteins that are N-glycosylated dase requires terminal mannose sugars, which are responsible for in their native state has in most cases required mammalian expres- delivery of the protein to macrophages and subsequent incorpora- © 2004 Nature Publishing Group sion hosts that have the ability to mimic human glycosylation. tion into lysozomes13.As discussed, glycoproteins from CHO cells Prokaryotic hosts such as E. coli do not glycosylate proteins, and contain complex N-glycans which contain: N-acetyl glucosamine lower eukaryotic expression systems such as yeast and insect cells (GlcNAc), galactose (Gal) and terminal sialic acid (NANA) (Fig. 2). are typically unable to provide mammalian glycosylation. Aglyco- It is therefore necessary to remove all terminal nonmannose sugars sylated forms of glycoproteins tend to be misfolded, biologically after isolation of the glycoprotein. To do so, in vitro treatment with inactive or rapidly cleared from circulation5.Yeasts are able to glyco- neuraminidase (to remove sialic acid), galactosidase (to remove sylate proteins, yet the nonhuman nature of yeast glycans negatively galactose) and hexosaminidase (to remove GlcNAc) is required. The impacts protein half-life because of the affinity of these glycans to process thus involves a mammalian cell culture step to produce the high-mannose receptors present on macrophages and endothelial protein of interest and a purification step followed by three enzy- cells6.For these reasons, mammalian cells, in particular Chinese matic processing steps to yield a glycan structure that is appropriate hamster ovary (CHO) cells, have emerged as the most widely used for therapeutic use. expression system for the production of complex human glycopro- Collectively, these examples are intended to support the view that teins—currently about 60% of the therapeutic protein market are glycosylation plays an essential role in the function of many thera- glycoproteins with an annual growth rate of approximately 26%4. peutically relevant proteins and that controlling glycosylation can CHO cells are typically able to express glycoproteins with human- be used to modulate and improve therapeutic function, or even like glycosylation patterns, although it is commonly recognized that eliminate certain undesirable functions. glycan structures produced from CHO cell lines differ from those Because mammalian cells typically express a heterogenous mix- produced in human cells and sometimes have to be modified to ture of glycoforms, several approaches have been developed to over- meet therapeutic efficacy7,8.A few natively glycosylated proteins are come this shortcoming including modulating glycosylation produced in nonglycosylating hosts (e.g. Betaferon (interferon-β- pathways in mammalian host cells or using in vitro enzymatic meth- 1b); Schering-Plough, Kenilworth, NJ, USA) yet the lack of glycosy- ods to modify glycosylation after purification of the protein. For lation can compromise half-life and other clinically relevant example, Umana and coworkers at the Institute of Biotechnology properties of the molecule such as renal clearance, which is affected (ETH, Zurich) showed that CHO cells that overexpress UDP- by the hydrodynamic volume of the protein (Fig. 1). glucosaminyl transferase III, produce IgGs with increased bisecting

1410 VOLUME 22 NUMBER 11 NOVEMBER 2004 NATURE BIOTECHNOLOGY REVIEW

GlcNAc and elevated antibody-dependent cell-mediated cytotoxic- ER ity (ADCC) activity14.Methods to engineer glycosylation in mam- (human, P. pastoris) malian cells have been reviewed elsewhere15,16.

Yeast and fungal expression systems The growing demand for improved biopharmaceutical expression hosts has led several companies to return to yeast and fungal expres- Golgi (human) Man8B Golgi (P. pastoris) sion systems, which provide high protein titers (>1 g/l) in fermentation processes that last a few days, are scalable to the 100 m3 scale, Mns IA, IB and IC 1,6 MnT (Och1p) allow for rapid turnaround from gene to protein and offer an array High-mannose of genetic tools to manipulate the host organism. Decades of glycans research and the production of many industrial enzymes in both yeast and fungi have resulted in robust, scalable processes that allow Man5 Man9 the low-cost production of many recombinant enzymes. In addi- GnTI 1,2 MnTs tion, about a sixth of all currently approved therapeutics are made 1,6 MnTs in yeasts1,and thus fewer regulatory hurdles are expected when Hybrid comparing yeast-based production to entirely new expression plat- glycans GlcNAcMan5 forms, such as transgenic animals or plants. MnsII All yeast-based therapeutic proteins are currently produced in the Man9-14 Baker’s yeast S. cerevisiae (Table 1), but other yeasts have been developed to make therapeutic proteins. The yeast P. pastoris was origi- GlcNAcMan nally developed as a single-cell protein-production system by 3 β-1,N-GlcNAc GnTII Philips Petroleum (Bartlesville, OK, USA) but was subsequently β-1,4-GlcNAc adapted for heterologous protein expression. More than 120 recom- β-1,2-GlcNAc http://www.nature.com/naturebiotechnology binant proteins have been expressed in this host—many of which β-1,4-Man are of human or mammalian origin17.More recently, P. pastoris α-1,6-Man α-1,2-Man has been used to express therapeutic proteins that have entered clin- GlcNAc2Man3 GalT α-1,3-Man ical trials (Table 1), although some proteins expressed in this host Complex β glycans -1,4-Gal have been shown to contain O-mannosylation not present on the α-2,3-NANA/α-2,6-NANA native protein18.

Genencor (Palo Alto, CA, USA) has used the filamentous fungi Gal2GlcNAc2Man3 Aspergillus niger and Trichoderma reesei for the large-scale produc- ST tion of recombinant industrial enzymes. Recent efforts to express therapeutically relevant proteins have focused on the expression of full-length IgGs in A. niger19.This antibody is produced at titers just under 1 g/l, is correctly assembled and binds antigen. Ogunjimi NANA2Gal2GlcNAc2Man3 et al.20 have had similar success with the expression of intact anti- © 2004 Nature Publishing Group body in P. pastoris;however, titers were below 40 mg/l. The structure of the glycans added to the antibodies by A. niger and P. pastoris Figure 2 Glycosylation pathways in humans and yeast. Mns, α-1,2- are of the high-mannose type, as expected from fungi and yeasts. mannosidase; MnsII, mannosidase II; GnT1, β-1,2-N-acetylglucosaminyl- β β Ward and colleagues19 demonstrated that the antibody produced transferase I; GnTII, -1,2-N-acetylglucosaminyltransferase II; GalT, -1, 4-galactosyltransferase; ST, sialyltransferase; MnT, mannosyltransferase. in A. niger had similar pharmacokinetic behavior and ADCC activity when compared with a mammalian cell–derived antibody, although the glycan structures differed from mammalian glycosylation and not all glycosylation sites were occupied with glycans. glycosylation pathways in these hosts to produce human-like glyco- Although this is promising, concern remains that these nonhuman proteins. Early attempts to replace mammalian cell lines with fungal aberrant sugar structures may be immunogenic in humans upon expression hosts were based on the observation that the filamentous long-term administration. fungus T. reesei was able to secrete glycoproteins containing inter- Berna Biotech, formerly Rhein Biotech (Bern, Switzerland), has mediates of human glycosylation that were amenable to in vitro pro- developed a proprietary protein expression platform based on the cessing by mammalian glycosyltransferases21.In this approach, a methylotrophic yeast Hansenula polymorpha that is currently being fungal precursor protein serves as a substrate for one or several further developed for the production of recombinant vaccines. chemoenzymatic glycosylation steps that result in hybrid glycopro- Although yeast systems (e.g., H. polymorpha, S. cerevisiae and P. pas- teins, albeit it at low yield22.Although encouraging, this approach toris) are often ideal for producing high titers of recombinant pro- requires several in vitro glycosylation reactions, which are costly and teins, their nonhuman glycosylation patterns have in the past cumbersome, and it became apparent that a more attractive solu- precluded them from consideration for the production of glycopro- tion for high-titer production of human glycoproteins in yeast and teins intended for therapeutic use in humans. filamentous fungi would be to genetically engineer the human glycosylation pathway into the host organism itself. Humanizing glycosylation pathways in yeasts and fungi Although conceptually straightforward, the technical complexity Given the otherwise attractive attributes of yeasts and filamentous of replicating human glycosylation pathways in fungal hosts is fungi, several groups have investigated the possibility of humanizing daunting. First, any attempt to reengineer a protein expression host

NATURE BIOTECHNOLOGY VOLUME 22 NUMBER 11 NOVEMBER 2004 1411 REVIEW

to perform human glycosylation requires a Table 1 Therapeutic proteins produced in the yeasts S. cerevisiae and P. pastoris complete inventory of all glycosylation reactions known to occur in a given host, includ- Products on the market ing O-glycosylation (i.e., the glycome). By the early 1990s, much was known about the Commercial name Recombinant protein Company Expression system principal glycosylation pathways in humans Actrapid Insulin NovoNordisk S. cerevisiae and several yeasts23,24.Whereas human Ambirix Hepatitis B surface antigen GlaxoSmithKline S. cerevisiae N-glycosylation is of the complex type, built on a tri-mannose core extended with Comvax Hepatitis B surface antigen Merck S. cerevisiae GlcNAc, galactose, and sialic acid (Fig. 2), Elitex Urate oxidase Sanofi-Synthelabo S. cerevisiae

yeast N-glycosylation is of the high- Glucagen Glucagon Novo Nordisk S. cerevisiae mannose type containing up to 100 or more mannose sugars (hypermannosylation)25,26. HBVAXPRO Hepatitis B surface antigen Aventis Pharma S. cerevisiae Early N-glycan processing, involving (i) Hexavac Hepatitis B surface antigen Aventis Pasteur S. cerevisiae the assembly of the core oligosaccharide, (ii) Infanrix-Penta Hepatitis B surface antigen GlaxoSmithKline S. cerevisiae its site-specific transfer onto the protein, Leukine Granulocyte-macrophage colony Berlex S. cerevisiae and (iii) its trimming by glucosidase I, II stimulating factor and an endoplasmic reticulum (ER) resident α-1,2-mannosidase, is highly con- Novolog Insulin Novo Nordisk S. cerevisiae served in mammals and yeast. In fact, up to Pediarix Hepatitis B surface antigen GlaxoSmithKline S. cerevisiae the formation of the Man8 glycan interme- Procomvax Hepatitis B surface antigen Aventis Pasteur S. cerevisiae diate (Man B) in the ER both pathways are 8 Refuldan Hirudin/lepirudin Hoechst S. cerevisiae identical (Fig. 2). However, following transport of the protein to the Golgi apparatus, Regranex rh Platelet-derived growth factor Ortho-McNeil Phama S. cerevisiae

http://www.nature.com/naturebiotechnology (US), Janssen-Cilag (EU) the two pathways diverge significantly. It is at this juncture that mammalian cells rely Revasc Hirudin/desirudin Aventis S. cerevisiae on additional α-1,2-mannosidases to trim Twinrix Hepatitis B surface antigen GlaxoSmithKline S. cerevisiae 27 mannose residues ,whereas yeasts initiate Products under development a series of mannosyltransferase reactions to yield the above-mentioned hypermannosy- Protein Indication Company Expression system lated glycan structures. To date, all success- Angiostatin Antiangiogenic factor EntreMed P. pastoris ful efforts to humanize yeast glycosylation pathways have focused on the deletion of Elastase inhibitor Cystic fibrosis Dyax P. pastoris specific yeast genes involved in hyperman- Endostatin Antiangiogenic factor EntreMed P. pastoris nosylation, and the introduction of genes Epidermal growth Diabetes Transition Therapeutics P. pastoris catalyzing the synthesis, transport, and factor analog addition of human sugars. Insulin-like growth Insulin-like growth factor-1 Cephalon P. pastoris © 2004 Nature Publishing Group 28 Schachter’s group at the University of factor-1 deficiency Toronto was the first to report an attempt Human serum Stabilizing blood volume in Mitsubishi Pharma P. pastoris to introduce a mammalian glycosylation albumin burns/shock (formerly Welfide) step in the filamentous fungus A. nidulans Kallikrein inhibitor Hereditary angiodema Dyax P. pastoris by expressing β-1,4-N-acetylglucosaminyltransferase I (GnTI), the key enzyme Includes approved products and those in clinical development. Sources: (G. Walsh, 2003)1 or personal responsible for the formation of hybrid gly- communication. cans in mammals. However, although high transcription levels of the enzyme were observed, and active enzyme could be assayed in cell-free extracts, the ER, some Man5 N-glycans could be generated on an intracellular 28 30 no impact on the N-glycans was observed . reporter protein .However, the paucity of Man5 obtained (less One of the critical steps in obtaining complex glycoproteins in a than 30% of total glycans), despite the expression of an α-1,2- µ fungal host requires the generation of Man5 in high yields, because mannosidase from a multicopy plasmid (yeast 2 ) and a strong con- this substrate is the precursor for all subsequent downstream stitutive promotor (glyceraldehyde-3-phosphate, GAPDH), was of processing steps (Fig. 2). Early attempts to engineer active α-1,2- concern. These initial attempts in fungi and yeasts showed some mannosidases into yeasts were rather discouraging, mostly because promise; however, they also revealed that genetically humanizing P. pastoris appeared to compensate for mannose trimming by upreg- glycosylation pathways in these hosts was not trivial and that merely ulating mannosyltransferase activity leading to even higher molecu- expressing enzymes involved in mammalian glycosylation pathways lar weight glycan structures29.Jigami’s group30 at Japan’s National was unlikely to be successful. Institute of Bioscience and Human Technology (Tsukuba, Ibaraki, Japan), in collaboration with researchers at Kirin (Yokohama, The secretory pathway is a cellular assembly line Japan), showed that after systematically eliminating yeast-specific In yeasts and humans, host-specific glycosyltransferases and gly- glycosylation reactions by deleting och1, mnn1 and mnn4 in the cosidases line the luminal surface of the ER and Golgi apparatus and yeast S. cerevisiae, and expressing an α-1,2-mannosidase targeted to thereby provide catalytic surfaces that allow the sequential process-

1412 VOLUME 22 NUMBER 11 NOVEMBER 2004 NATURE BIOTECHNOLOGY REVIEW

ing of glycoproteins as they proceed from the ER through the Golgi network. As a gly- β-1,N-GlcNAc β-1,4-Man α-1,6-Man α-1,3-Man β-1,4-Gal coprotein proceeds from the ER through the ab secretory pathway, it is sequentially exposed 100 100 to different mannosidases and glycosyl- 75 transferases. An essential step in recreating 75 human glycosylation pathways in lower 50 50 eukaryotes is recreating the sequential 25 Man 25 Intensity (%) 5 nature of complex N-glycan processing. In Intensity (%) 2000, I initiated a research program with 0 0 three main focal points: (i) overcoming the 850 1,350 1,850 2,350 2,850 3,350 850 1,350 1,850 2,350 2,850 Mass (m/z) Mass (m/z) uncertainty related to the targeting of het- cd erologous glycosylation enzymes across dif- 100 100 ferent species and developing methods to target enzymes to predetermined locations 75 75

in the secretory pathway of yeast and fila- 50 50 mentous fungi, (ii) developing methods to GlcNAcMan3 GlcNAc2Man3 Intensity (%) identify enzymes that are in fact active at the Intensity (%) 25 25 site of localization (since the different 0 0 organelles of the yeast secretory pathway 850 1,350 1,850 2,350 2,850 850 1,350 1,850 2,350 2,850 provide different processing environments Mass (m/z) Mass (m/z) with different pHs and sugar nucleotide pools, one has to consider how a recombi- Figure 3 MALDI-TOF mass spectrometry of oligosaccharides released from purified kringle 3 protein nant enzyme targeted to a particular produced in wild-type P. pastoris and glycoengineered strains of P. pastoris. (a–d) MALDI-TOF mass http://www.nature.com/naturebiotechnology organelle will behave in that nonnative envi- spectra positive mode: Wild type (a), and increasingly humanized glycan structures (see ref. 31, (b); ref. 34, (c); and ref. 35, (d)). ronment), and (iii) develop efficient screens to identify cell lines that have acquired the ability to process recombinant glycoproteins in a human-like fashion. by the elimination of two genes associated with the transfer of this By 2003, my team had shown that many of the problems previ- sugar32.We have expressed several recombinant proteins in these ously encountered by other researchers could be overcome by gener- strains and have found expression levels to be similar to wild-type ating large-scale combinatorial genetic libraries of glycosylation yeast. Recent work by Contreras’ group33 at Ghent University enzymes fused to libraries of yeast-targeting peptides31.By screen- (Belgium) has also shown the in vivo production of nonsialylated ing a library of over 600 fusion constructs consisting of catalytic hybrid N-linked glycans in a glycoengineered strain of P. pastoris. domains of α-1,2-mannosidases, and yeast type II leader peptides, After the in vivo production of hybrid N-glycans, we have gone on we were able to find several specific combinations of catalytic to further humanize the glycosylation of P. pastoris to perform domains and leader peptides that displayed efficient mannosidase glycosylation of the complex type34.By successfully introducing © 2004 Nature Publishing Group activity in vivo in P. pastoris. Screening was done by purifying a α-mannosidase II and GlcNAc-transferase II (GnTII) activities, fol- recombinant reporter protein secreted by each strain, and analyzing lowing the combinatorial library approach discussed above, we were its glycans by digestion with protein N-glycanase followed by able to generate yeast strains that perform essentially homogenous matrix-assisted laser desorption ionization time-of-flight (MALDI- glycosylation containing GlcNAc2Man3 (Fig. 3c). TOF) mass spectrometry. The entire procedure of purification and More recently, we have demonstrated the proper localization of analysis was carried out by an automated liquid handler allowing active β-1,4-galactosyltransferase (GalT) and UDP-galactose-4- the processing of several thousand samples per week. epimerase in P. pastoris conferring the transfer of galactose onto Glycoproteins expressed in these engineered strains contain over both terminal GlcNAc residues of a GlcNAc2Man3 glycan, resulting 35 90% Man5 and thus provided the basis for the further humanization in the production of Gal2GlcNAc2Man3 in vivo (Fig. 3d). This has of glycosylation pathways in P. pastoris. After the successful genera- rendered the humanization of P. pastoris one step away from com- tion of Man5, we showed that by introducing an active GnTI gene pletion. Unfortunately, the final transfer of sialic acid will be most and a UDP-GlcNAc transporter, we were able to add a terminal challenging, as unlike GlcNAc and galactose, a source of endoge- GlcNAc residue to the 1,3 arm of the Man5 glycan, thereby generat- nous sialic acid is not known to exist in yeasts or other fungal organ- ing the first genetically engineered yeast that efficiently produces isms. Thus, in addition to the requirement for a sialyltransferase hybrid N-glycans. As in the mannosidase case, active GnTI was enzyme, several other genes involved in the synthesis of CMP-sialic identified by introducing a GnTI catalytic domain/Golgi leader acid must also be genetically engineered into the host. We have library into the yeast and screening for transformants that acquired recently shown that the precursor CMP-sialic acid, can be produced the ability to produce GlcNAcMan5 at high efficiency. After knock- in P. pastoris and used to transfer sialic acid onto a secreted reporter ing out the initiating α-1,6-mannosyltransferase activity, and intro- protein (Gerngross, T.U., unpublished data). ducing α-1,2-mannosidase, GnT1 and the UDP-GlcNAc transporter Unlike proteins obtained from mammalian cell culture, which are into P. pastoris we were able to show the secretion of a reporter heterogenous, we have found that yeast cells can be engineered to protein with over 90% of the N-glycans have the desired secrete glycoproteins with exceptional glycan uniformity31,34,35. GlcNAcMan5 structure (Fig. 3b)31.The remaining glycans contain Although yeast-based manufacturing is expected to reduce the mostly mannosylphosphate, which can be eliminated in P. pastoris production cost of protein-based therapeutics, we believe the main

NATURE BIOTECHNOLOGY VOLUME 22 NUMBER 11 NOVEMBER 2004 1413 REVIEW

benefit of using glycoengineered yeast lies in the ability to rigorously cells. Glyconj. J. 16, 81–97 (1999). 9. Suzuki, T., Kitajima, K., Inoue, S. & Inoue, Y. N-glycosylation/deglycosylation as a control glycosylation of the final product and thereby improve clin- mechanism for the post-translational modification/remodification of proteins. ically relevant characteristics such as half-life or tissue distribution. Glycoconj J. 12, 189–193 (1995). Using different yeast strains, a protein can be expressed in its vari- 10. Takeuchi, M. et al. Relationship between sugar chain structure and biological activity of recombinant human erythropoietin produced in Chinese hamster ovary cells. Proc. ous glycoforms (Fig. 3) and glycosylation-mediated functions can Natl. Acad. Sci. USA. 86, 7819–7822 (1989). be elucidated, provided an appropriate assay can be developed. Most 11. Yuen, C.T. et al. Relationships between the N-glycan structures and biological activities of recombinant human erythropoietins produced using different culture condi- importantly, once a given glycoform has been identified, scaling-up tions and purification procedures. Br. J. Haematol. 121, 511–526 (2003). and manufacturing are done on the same platform allowing the 12. Erbayraktar, S. et al. Asialoerythropoietin is a nonerythropoietic cytokine with broad large-scale production of specific glycoforms. neuroprotective activity in vivo. Proc. Natl. Acad. Sci. USA 100, 6741–6746 (2003). Erratum in: Proc. Natl. Acad. Sci. USA 100, 9102 (2003). 13. Friedman, B. et al. A comparison of the pharmacological properties of carbohydrate The future of humanized yeast remodeled recombinant and placental-derived beta-glucocerebrosidase: implications With the advent of humanized yeast strains and their ability to exer- for clinical efficacy in treatment of Gaucher disease. Blood 93, 2807–2816 (1999). 14. Umana, P., Jean-Mairet, J., Moudry, R., Amstutz, H. & Bailey, J.E. Engineered glyco- cise direct control over glycosylation, we see the advantages of forms of an antineuroblastoma IgG1 with optimized antibody-dependent cellular mammalian cell culture fading, and expect a significant number of cytotoxic activity. Nat. Biotechnol. 17, 176–180 (1999). 15. Warner, T.G. Enhancing therapeutic glycoprotein production in Chinese hamster ovary biopharmaceuticals to be switched to yeast-based expression plat- cells by metabolic engineering endogenous gene control with antisense DNA and forms. The short turn-around time to obtain gram quantities of a gene targeting. Glycobiology 9, 841–850 (1999). therapeutic protein (typically 4–6 months for a purified protein), 16. Warner, T.G. Metabolic engineering glycosylation: biotechnology’s challenge to the glycobiologist in the next millennium. in Part IV Biology of Saccharides. (eds. Ernst, combined with the demonstrated scalability of yeast offer an attrac- B., Hart, G. & Sinay, P.) 1043–1064 (VCH-Wiley, New York, 2001). tive alternative to current mammalian cell culture processes. 17. Cregg, J.M., Cereghino, J.L., Shi, J. & Higgins, D.R. Recombinant protein expression Although this is expected to reduce the cost of biomanufacturing, in Pichia pastoris. Mol. Biotechnol. 16, 23–52 (2000). 18. Mochizuki, S. et al. Expression and characterization of recombinant human we believe that the main benefit of using glycoengineered yeast antithrombin III in Pichia pastoris. Protein Expr. Purif. 23, 55–65 (2001). strains lies in the ability to rigorously control glycosylation of the 19. Ward, M. et al. Characterization of humanized antibodies secreted by Aspergillus niger. Appl. Environ. Microbiol. 70, 2567–2576 (2004). final product and thereby improve clinically relevant properties of 20. Ogunjimi, A.A., Chandler, J.M., Gooding, C.M. Recinos, A. & Choudary, P.V. High- the molecule. Instead of glycoform mixtures produced by mam- level secretory expression of immunologically active intact antibody from the yeast http://www.nature.com/naturebiotechnology malian cell culture processes, it is now possible to use glycoengi- P. pastoris. Biotechnol. Letters, 21, 561–567 (1999). 21. Maras, M. & Contreras, R. Methods for modifying carbohydrate moieties. US patent neered yeast strains to produce a glycoprotein with a single type of 5,834,251 (Dec. 30, 1994). N-glycosylation (Fig. 3b–d). Most cost-of-goods analyses focus on 22. Maras, M. et al. In vitro conversion of the carbohydrate moiety of fungal glycoproteins the cost required to produce a given amount of recombinant pro- to mammalian-type oligosaccharides. Eur. J. Biochem. 249, 701–707 (1997). 23. Kornfeld, R. & Kornfeld, S. Assembly of asparagine-linked oligosaccharides. Annu. tein. What is often absent in this analysis is the quality of the final Rev. Biochem. 54, 631–664 (1985). product, which can have an impact on dosage requirements and 24. Brockhausen, I., Carver, J.P. & Schachter, H. Control of glycoprotein synthesis. The use of oligosaccharide substrates and HPLC to study the sequential pathway for therapeutic effect and ultimately market share. Considering the N-acetylglucosaminyltransferases I, II, III, IV, V and VI in the biosynthesis of highly 20 ability of yeast systems to express full antibodies ,cytokines, serum branched N-glycans by hen oviduct membranes. Biochem. Cell Biol. 66, 1134–1151 proteins and other therapeutically relevant proteins, we expect (1988). 25. Silberstein, S. & Gilmore, R. Biochemistry, molecular biology and genetics of yeast-based therapeutic protein expression to become a serious con- oligosaccharyltransferase. FASEB J. 10, 849–858 (1996). tender in any manufacturing strategy. 26. Gemmill, T.R. & Trimble, R.B. Overview of N- and O-linked oligosaccharide structures found in various yeast species. Biochim. Biophys. Acta 1426, 227–237 (1999). COMPETING INTERESTS STATEMENT 27. Tulsiani, D.R., Hubbard, S.C., Robbins, P.W. & Touster, O. Alpha-D-mannosidases of The authors declare competeing financial interests (see the Nature Biotechnology rat liver Golgi membranes. Mannosidase II is the GlcNAcMan5-cleaving enzyme in © 2004 Nature Publishing Group website for details). glycoprotein biosynthesis and mannosidases IA and IB are the enzymes converting Man9 precursors to Man5 intermediates. J. Biol. Chem. 257, 3660–3668 (1982). 28. Kalsner, I., Hintz, W., Reid, L.S. & Schachter, H. Insertion into Aspergillus nidulans Published online at http://www.nature.com/naturebiotechnology/ of functional UDP-GlcNAc: α3-D-mannoside β-1,2-N-acetyl-glucosaminyltransferase I, the enzyme catalyzing the first committed step from oligomannose to hybrid and 1. Walsh, G. Biopharmaceutical Benchmarks—2003. Nat. Biotechnol. 21, 865–870 complex N-glycans. Glycoconj. J. 12, 360–370 (1995). (2003). 29. Martinet, W., Maras, M., Saelens, X., Jou, W.M. & Contreras, R. Modification of the 2. Punt, P.J., van Biezen, N. Conesa, A., Albers, A., Mangnus, J. & van den Hondel, C. protein glycosylation pathway in the methylotrophic yeast, Pichia pastoris. Filamentous fungi as cell factories for heterologous protein production. Trends Biotechnol. Letters 20, 1171–1177 (1998). Biotechnol. 20, 200–206 (2002) 30. Chiba, Y. et al. Production of human compatible high mannose-type (Man5GlcNAc2) 3. Werten, M.W.T., van den Bosch, T. J., Wind, R.D., Mooibroek, H. & De Wolf, F.A. sugar chains in Saccharomyces cerevisiae. J. Biol. Chem. 273, 26298–26304 High-yield secretion of recombinant gelatins by Pichia pastoris. Yeast 15, (1998). 1087–1096 (1999). 31. Choi, B.K. et al. Use of combinatorial genetic libraries to humanize N-linked glycosy- 4. Humphreys, A. & Boersig, C. Cholesterol drugs dominate. MedAdNews 22, 42–57 lation in the yeast Pichia pastoris. Proc. Natl. Acad. Sci. USA 100, 5022–5027 (2003). (2003). 5. Coloma, M.J., Clift, A., Wims, L. & Morrison, S.L. The role of carbohydrate in the 32. Stadheim, T.A., Bobrowicz, P., & Wildt, S. Methods for eliminating mannosylphos- assembly and function of polymeric IgG. Mol Immunol. 37, 1081–1090 (2000). phorylation of glycans in the production of glycoproteins. US Patent Appl. No. 6. Mistry, P.K., Wraight, E.P. & Cox, T.M. Therapeutic delivery of proteins to 60/532,461 (December 23, 2003). macrophages: implications for treatment of Gaucher’s disease. The Lancet, 348, 33. Vervecken, W. et al. In vivo synthesis of mammalian-like, hybrid-type N-glycans in 1555–1559 (1996). Pichia pastoris. Appl Environ Microbiol. 70, 2639–2646 (2004). 7. Grabowski, G.A. et al. Enzyme therapy in type 1 Gaucher disease: comparative effi- 34. Hamilton, S.R. et al. Production of complex human glycoproteins in yeast. Science, cacy of mannose-terminated glycocerebrosidase from natural and recombinant 301, 1244–1246 (2003). sources. Ann. Intern. Med. 122, 33–39 (1995). 35. Bobrowicz, P. et al. Engineering of an artificial glycosylation pathway blocked in core 8. Grabenhorst, E., Schlenke, P., Pohl, S., Nimtz, M. & Conradt, H.S. Genetic engineer- oligosaccharide assembly in the yeast Pichia pastoris: production of complex human- ing of recombinant glycoproteins and the glycosylation pathway in mammalian host ized glycoproteins with terminal galactose. Glycobiology 14, 757–766 (2004).

1414 VOLUME 22 NUMBER 11 NOVEMBER 2004 NATURE BIOTECHNOLOGY

Glycan Engineering for Cell and Developmental Biology Griffin ME and Hsieh-Wilson LC Cell Chem Biol, 23:108 (2016) Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

Glycan Engineering for Cell and Developmental Biology

Matthew E. Grifﬁn1 and Linda C. Hsieh-Wilson1,* 1Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA 91125, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.chembiol.2015.12.007

Cell-surface glycans are a diverse class of macromolecules that participate in many key biological processes, including cell-cell communication, development, and disease progression. Thus, the ability to modulate the structures of glycans on cell surfaces provides a powerful means not only to understand fundamental processes but also to direct activity and elicit desired cellular responses. Here, we describe methods to sculpt glycans on cell surfaces and highlight recent successes in which artiﬁcially engineered glycans have been employed to control biological outcomes such as the immune response and stem cell fate.

Introduction cise manipulation of specific genes in a cell-specific and induc- Cell-surface glycans participate in many important processes ible manner. However, as GTs typically operate on multiple pro- throughout the lifespan of an organism, ranging from cell migra- tein substrates and assemble a variety of glycan structures, it is tion and tissue patterning to the immune response, disease pro- difficult to study the impact of a single glycan on an individual gression, and cell death (Fuster and Esko, 2005; Ha¨ cker et al., protein of interest. Moreover, genetic approaches usually sub- 2005; Lichtenstein and Rabinovich, 2013; van Kooyk and Rabi- tract from existing glycan structures rather than add new novich, 2008). At a molecular level, glycans are often the first chemical functionalities. Finally, knocking out GTs can lead to points of contact between cells, and they function by facilitating developmental defects or embryonic lethality, which can hinder a variety of interactions both in cis (on the same cell) and in trans the identification of functions in the adult organism. Nonethe- (on different cells). The glycan covering that surrounds the cell less, genetic approaches have provided invaluable information surface, termed the glycocalyx, can both promote and hinder on the functional importance of GTs and protein glycosylation the binding of canonical protein ligands to their cell-surface re- in vivo. ceptors, as well as mediate ligand-independent receptor clus- The power of genetic approaches is exemplified by elegant tering and activation (Bishop et al., 2007; Coles et al., 2011; studies on the Notch signaling pathway. Notch signaling is Haines and Irvine, 2003; Rogers et al., 2011). Indeed, the integral essential for proper development, and dysregulation of the roles of cell-surface glycans in regulating cellular signaling pathway leads to various human diseases, including congenital events are only beginning to be understood. disorders and cancer (Andersson et al., 2011; Kopan and Ila- The ability to modulate and re-engineer the diverse structures gan, 2009). Glycosylation of the extracellular domain of the of glycans at the cell surface provides a powerful means to eluci- Notch receptor has emerged as an important mechanism for date the molecular mechanisms that underlie glycan-mediated the regulation of Notch activity (Figure 2A). Early studies iden- signaling events and their downstream cellular consequences. tified a b-1,3-N-acetylglucosaminyltransferase called Fringe From a mechanistic standpoint, systematically altering glycan that modifies O-linked Fuc residues on Notch (Bru¨ ckner structures provides insights into structure-function relationships et al., 2000; Moloney et al., 2000). Ectopic expression of a cata- and the importance of individual structures in glycan-mediated lytically inactive, mutant form of Fringe in Drosophila demon- processes. From an engineering standpoint, the ability to strated that the GT activity of Fringe was required for certain remodel glycan architectures on cell surfaces offers a novel Notch ligand-receptor interactions and proper wing formation approach to manipulate cellular physiology and phenotypic out- in vivo (Moloney et al., 2000). Other GTs have also been shown comes. Here, we will describe the methods available to tailor the to be critical for Notch signaling. For example, protein O-fuco- structures of glycans on cells and provide notable examples of syltransferase 1 (Pofut1 in mammals or Ofut1 in Drosophila) how remodeling of cell-surface glycans have led to both new catalyzes the addition of O-Fuc to Notch, generating the biological insights and novel cellular functions. acceptor substrate for Fringe (Wang et al., 2001). RNAi-mediated knockdown of Ofut1 showed that the enzyme is required Genetic Approaches for both Fringe-dependent and Fringe-independent Notch func- Genetic manipulation of glycosyltransferases (GTs) or other tion in Drosophila (Okajima and Irvine, 2002). In mice, genetic genes involved in glycan biosynthesis provides a powerful deletion of Pofut1 results in embryonic lethality with phenotypic method to perturb specific glycan subpopulations in cells and defects in cardiovascular and neurologic development, consis- organisms. A variety of approaches have been developed, tent with loss of all Notch paralog signaling (Shi and Stanley, including gene deletion or knockout, gene knockdown by 2003). In addition, inducible inactivation of Pofut1 in the im- RNAi, and gene overexpression (Figure 1). Genetic methods mune-responsive cells and hematopoietic tissues of adult offer excellent spatial and temporal control, enabling the pre- mice revealed a key role for O-Fuc glycans on Notch in the

108 Cell Chemical Biology 23, January 21, 2016 ª2016 Elsevier Ltd All rights reserved Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

A Figure 1. Genetic Approaches GT GT (A) Transient transfection of cells in vitro with either RNAi x shRNA or siRNA (RNAi) targeting a gene causes a GT GT decrease in expression of the protein. Transfection protein with the cDNA of a protein leads to its overexpression expression. (B) Mouse knockout models are traditionally pro- GT GT cDNA GT duced by ﬁrst stably disrupting gene expression in GT mouse embryonic stem cells by homologous GT recombination. Mutated stem cells are injected into blastocysts, which are then implanted into mice to B produce mosaic offspring. Mice are further bred to produce a stable knockout line. offspring (C) To generate stable CHO knockout cell lines, breeding parental cells were treated with a mutagenic agent (e.g., ionizing radiation) and then selected for resistance to cytotoxic lectins, which bind to spe- blastocyst ciﬁc carbohydrate epitopes now missing on the mutated pregnant stable injection mutated cells. stem cells mouse knockout line

C Another example of the power of genetic approaches to reveal important develop- mutagenesis selection mental functions of glycans involves studies on heparan sulfate (HS) glycosami- mutated noglycans (GAGs). HS binds to diffusible cell line chemical signals known as morphogens and aids in establishing gradients of parental cells these secreted proteins to relay positional information in developing organisms (Figure 3A) (Ha¨ cker et al., 2005). Pheno- regulation of hematopoietic homeostasis (Yao et al., 2011). typic screens in Drosophila showed that mutations in tout-velu Together, these genetic studies demonstrate the essential na- (ttv), a homolog of the mammalian HS polymerase Ext1, caused ture of glycosylation for Notch receptor function. pattern defects in the wing imaginal disc similar to the loss of Mass spectrometry analysis indicated that the O-Fuc glycans Hedgehog (Hh) signaling (Bellaiche et al., 1998; The et al., 1999). on mammalian Notch exist as a mixture of di-, tri-, and tetra- The morphogen Hh specifies anterior-posterior orientation during saccharides of NeuAca(2–3)Galb(1–4)GlcNAcb(1–3)Fuc (Molo- wing development by diffusing from the posterior compartment of ney et al., 2000). To decipher the minimum structure necessary the imaginal disc to anterior cells in a graded fashion (Tabata and for activity, a library of mutant Chinese hamster ovary (CHO) Kornberg, 1994). To elucidate the importance of ttv, mosaic cell lines deficient in GTs and activated nucleotide (NXP)-sugar knockout clones were produced in which defined regions of the transporters was employed (Figures 1C and 2B) (Chen et al., Drosophila embryo lacked ttv (Bellaiche et al., 1998). Loss of ttv 2001). Expression of Fringe in parental CHO cells led to a in the posterior compartment where Hh is secreted showed no ef- decrease in Jagged1 stimulation of Notch, presumably due to fect on the distance of Hh-mediated activation into the anterior the loss of Jagged1-Notch complexation mediated by the compartment. However, anterior ttvÀ/À clones showed a severe O-Fuc glycan. However, the Lec8 cell line, which contains mu- spatial attenuation of active Hh signaling in the anterior compart- tations in the uridine diphosphate (UDP)-Gal Golgi transporter ment. Defective distribution of Hh signaling was also observed in that result in defective protein galactosylation, showed no mutant clones of the Ext2 homolog sister-of-tout-velu (sotv) change in Notch activation upon Fringe expression, suggesting and the Extl3 homolog brother-of-tout-velu (botv)(Han et al., that the Gal moiety was required to disrupt the Jagged1-Notch 2004), suggesting that the HS produced by these enzymes is interaction. Conversely, the Lec2 cell line containing an inactive necessary for Hh binding, gradient formation, and signal trans- cytidine monophosphate (CMP)-NeuAc Golgi transporter that duction. Furthermore, wingless (Wg) and decapentaplegic (dpp) prevents protein sialylation showed the same response to Jag- morphogen signaling were shown to be disrupted by altered HS ged1 stimulation as the parental cell line, indicating that the biosynthesis and display (Baeg et al., 2004; Belenkaya et al., NeuAc residue was unnecessary. Of the six mammalian 2004; Kirkpatrick et al., 2004), providing evidence for a general b-1,4-galactosyltransferase (b4GalT) enzymes that transfer mechanism by which HS mediates morphogen gradients. Gal to GlcNAc in CHO cells, only b4GalT-1 was required for Interestingly, ostensibly conflicting data were obtained from Notch stimulation, providing the identity of a key enzyme mouse models (Koziel et al., 2004). Here, hypomorphic mutation involved in synthesizing the minimum trisaccharide. Thus, ge- of Ext1, which severely limits HS production, led to an extended netic approaches not only illustrate the necessity of specific tissue distribution of Indian hedgehog (Ihh), the mouse homolog glycans for activity in physiological systems but also can be of Hh, during endochondral ossification (Figure 3B). Using hap- systematically used to determine their structure and the en- loinsufficiency studies in which one or both alleles of Ext1 were zymes required for their biosynthesis. replaced with the hypomorphic variant, lower production of HS

Cell Chemical Biology 23, January 21, 2016 ª2016 Elsevier Ltd All rights reserved 109 Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

Figure 2. Genetic Studies of Notch A (A) Structure of the O-fucose glycan found on mammalian Notch. (B) The O-fucose glycan of Notch causes a decrease in Jagged1-mediated stimulation. By eliminating speciﬁc monosaccharides on the Notch glycan using genetic methods, it was found that a trisaccharide was the minimum structure necessary to decrease Jagged1 activity.

Similar to classical forward genetics, genetic mapping of disease-related muta- B tions has led to the discovery of novel genes responsible for a single set of path- ological symptoms. Hereditary multiple exostoses (HME), a condition characterized by abnormal growths at the ends of long bones, has been associated with loss-of-function mutations in the Ext1, Ext2, and Ext3 genes by chromosomal linkage analysis (Cook et al., 1993; Le Merrer et al., 1994; Wu et al., 1994). These genes were later determined to encode GTs responsible for HS biosynthesis (Mc- Cormick et al., 1998). Missense mutations that were mapped for Ext1 were shown to cause a deficiency in HS production. As correlated with higher rates of proliferation in a region of chon- described above, decreased levels of HS results in increased drocytes distal from Ihh secretion, implying increased Ihh activation of Ihh (Koziel et al., 2004). This signaling dysregulation diffusion. In vitro growth of embryonic mouse forelimbs with triggers uncontrolled cellular growth and leads to cartilaginous exogenous GAGs showed that Ihh diffusion was inversely pro- abscesses and benign tumors in endochondrial bone. Thus, portional to HS concentration, suggesting that HS retarded the this particular example highlights how the mutational profile of diffusion of Ihh, unlike the results seen for ttv in Drosophila.It a disease enhanced an understanding of the molecular basis has been proposed that the disparity in these results is due to dif- of exostoses formation. ferences in the genetic approaches used (Ha¨ cker et al., 2005). As Genetic mapping of distinct but related disease states identi- HS is required for ligand-receptor binding, the use of a null allele fied multiple important genes that are components of the same for ttv prohibits all HS biosynthesis and Hh-mediated signaling. glycosylation pathway. Mutations in at least ten enzymes have In contrast, the hypomorphic Ext1 allele produces enough HS been linked to congenital muscular dystrophies (Wells, 2013), to elicit Ihh-protein complexation, and the lower overall amount which are commonly associated with defective O-glycosylation of HS may facilitate further morphogen diffusion. Thus, complete of a-dystroglycan, an integral component of muscle fibers ablation of HS in Drosophila decreases Hh signaling due to a loss (Figure 4). The O-glycans of a-dystroglycan mediate interactions of ligand-receptor binding, but the decrease of HS in mice in- with extracellular matrix proteins such as laminin (Yoshida-Mor- creases Ihh signaling by allowing greater diffusion of Ihh. The iguchi et al., 2010) and facilitate the formation of compact base- interpretation of these results highlights that careful consider- ment membranes and mature neuromuscular junctions (God- ation is necessary in selecting genetic models due to the multi- deeris et al., 2013). Loss of these glycans disrupts extracellular faceted nature of glycan-protein interactions in vivo. matrix organization and predisposes muscular tissue to dystro- The examination of naturally occurring mutations in human phy. Walker-Warburg syndrome and muscle-eye-brain disease, genes provides a complementary approach to establish the mo- two similar types of muscular dystrophy typified by anomalies of lecular determinants that underlie development and disease- the brain and eye, were linked to a number of proteins that related phenotypes. Human genetic mutations that lead to produce O-mannosyl glycans, including protein O-mannosyl- abnormal glycosylation have been well documented and span transferase 1 and 2 (POMT1/2) (Beltran-Valero de Bernabe´ a wide clinical spectrum, impacting nearly every organ system et al., 2002; van Reeuwijk et al., 2005), protein O-mannosyl N-ace- in the body. Advances in next-generation DNA sequencing tech- tylglucosaminyltransferase 1 and 2 (POMGnT1/2) (Manzini et al., nologies have helped to propel the discovery of >100 such 2012; Yoshida et al., 2001), b-1,3-N-acetylgalactosaminyltrans- congenital disorders of glycosylation (CDGs), providing unique ferase 2 (b3GalNT2) (Stevens et al., 2013), and protein O-manno- opportunities to understand the significance of glycosylation in syl kinase (POMK) (Yoshida-Moriguchi et al., 2013). Merosin-defi- human physiology. As CDGs have been well reviewed (Freeze, cient congenital muscular dystrophy 1D (MDC1D) was associated 2013; Hennet and Cabalzar, 2015), we highlight only a few repre- with mutations in like-acetylglucosaminyltransferase (LARGE) sentative examples. (Longman et al., 2003), which transfers GlcAb(1–3)Xyla(1–3)

110 Cell Chemical Biology 23, January 21, 2016 ª2016 Elsevier Ltd All rights reserved Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

A wild-type

HSPGs

posterior anterior

ttv-/-

Figure 4. O-Mannosyl Glycan of a-Dystroglycan posterior anterior The laminin-binding glycan of a-dystroglycan contains a core trisaccharide produced by POMT1/2, POMGnT2, and b3GalNT-2. The 6-OH position of mannose is phosphorylated by POMK, on which an unknown structure is added. b4GAT-1 makes an acceptor substrate for LARGE, which adds repeating GlcAb(1–3)Xyla(1–3) units. The required roles of other associated B enzymes such as FKTN are still unknown. Ext1Gt/Gt

can O-mannosyl glycan, linking the biosynthetic pathway of a complex glycan to a suite of developmental disorders. Next-generation sequencing has allowed for both targeted and untargeted screens to identify genes involved in the etiology of glycosylation-related diseases. Targeted sequencing of 25 known CDG-related genes uncovered no observable mutations prehypertrophic hypertrophic in a patient with widespread mental and physical developmental delays (Jones et al., 2012). Subsequent untargeted whole- Figure 3. Genetic Studies Provide Insights into How HS Glycans exome sequencing against roughly 22,000 genes revealed two Control Morphogen Gradients mutations that caused premature truncation of DDOST, a (A) In wild-type Drosophila, Hedgehog ligands (Hh, orange) are secreted from subunit of the oligosaccharyltransferase (OST) complex. The posterior cells of the imaginal wing disc and diffuse to anterior cells. HS chains (light green) on HS proteoglycans (HSPGs, dark green) displayed on the cell DDOST mutations impaired cell-surface expression of intracel- surface trap these ligands for cellular signaling and prevent further diffusion. lular adhesion molecule 1 (ICAM-1), whose trafficking to the À/À However, the lack of HS in ttv Drosophila prevents Hh recruitment to the cell surface depends on proper N-glycosylation. Complementa- cell surface and downstream signaling. (B) Mice with a hypomorphic mutation in Ext1 (Ext1Gt/Gt) produced less HS on tion with wild-type DDOST partially restored ICAM-1 trafficking, the cell surface. When Indian hedgehog ligands (Ihh, blue) were secreted by indicating that the DDOST mutations were indeed pathogenic prehypertrophic chondrocytes, the reduced amount of cell-surface HS al- and that their clinical manifestations were likely due to wide- lowed Ihh to diffuse further into the hypertrophic region. Unlike full knockout as À À spread defects in N-glycosylation. Untargeted genetic ap- seen in the ttv / Drosophila, chondrocytes with a hypomorphic mutation in Ext1 maintained enough cell-surface HS to transduce signaling, but not proaches have also revealed unanticipated genes important for enough to impede diffusion. proper glycosylation. Microarray and sequencing analysis of genes within autozygous regions of two glycosylation-deficient siblings with multiple forms of dysplasia uncovered mutations repeating units onto the O-mannosyl glycan (Goddeeris et al., in TMEM165 (Foulquier et al., 2012). This protein has no known 2013). Finally, patients with Fukuyama-type congenital muscular function but is implicated in maintaining calcium levels within dystrophy, a frequent disorder in Japan that presents as brain the Golgi (Demaegd et al., 2013), suggesting that proper glyco- micro-polygria due to improper neuronal migration, contained sylation relies on not only GTs and transporters but also mutations in a novel gene called fukutin (FKTN)(Kobayashi homeostatic regulation of Golgi pH and ion concentrations. et al., 1998). Although its exact function remains unknown, The identification of DDOST and TMEM165 mutations by these FKTN was postulated to be a glycosyltransferase and was methods illustrates the ability of next-generation sequencing to found to be required for LARGE activity (Ohtsuka et al., 2015). reveal pathologies caused by glycosylation defects and poten- Thus, an examination of these and other a-dystroglycanopathies tially expose novel regulatory mechanisms of glycosylation as converged on genes involved in the synthesis of the a-dystrogly- targets for therapeutic intervention.

Cell Chemical Biology 23, January 21, 2016 ª2016 Elsevier Ltd All rights reserved 111 Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

Figure 5. Metabolic Oligosaccharide Engineering MOE requires the passage of peracetylated glycans across the cell membrane and into the cell, where they are deacetylated by endogenous esterases. The monosaccharides are phosphorylated and activated with a nucleotide phosphate group. Glycosyltransferases then incorporate the non-natural sugars into nascent glycans. The functional handle (X) installed by the biosynthetic machinery can be reacted with other groups (Y) via bioorthogonal chemistry (X + Y produce Z) for detection, imaging, or introduction of novel biological activity.

Thus, genetic methods provide a powerful means to under- pathways, converted into activated nucleotide phosphate do- stand the importance of glycosylation in development and nors, and incorporated by GTs into specific glycan structures disease. However, knockout, knockdown, and overexpression (Figure 5). A range of non-natural sugar analogs based on approaches result in loss of glycan structures or enhanced Fuc, NeuAc, GalNAc, and GlcNAc have been successfully expression of only natural structures. A complementary installed in a variety of cells and organisms such as zebrafish approach is the use of chemical biology methods that enable and Caenorhabditis elegans. In principle, the analog is inserted expansion past the confines of naturally occurring glycans into all glycans containing the monosaccharide of interest, through incorporation of non-natural sugars. These non-natural which allows for the simultaneous detection of multiple glycan labeling approaches have facilitated the imaging, tracking, and structures. However, this also precludes the selective detec- identification of glycan structures. Beyond advancing the study tion of specific structures, such as motifs that differ by neigh- of glycans, such approaches provide a novel way to control boring glycans or by glycosidic linkage. As the non-natural an- cellular physiology and function by altering the glycan structures alogs compete with natural substrates, their incorporation into presented on the cell surface. Below, we highlight a few notable cellular glycans is sub-stoichiometric, which minimizes pertur- examples whereby cell-surface glycan engineering has been ex- bation to the system but also reduces detection sensitivity. The ploited to interrogate and control glycan-mediated biological analogs must also contain relatively small non-natural function- processes. alities to allow efficient recognition by enzymes and incorporation into glycans. Nonetheless, a number of chemical groups, Metabolic Oligosaccharide Engineering including inert alkyl chains (Keppler et al., 1995), as well as Metabolic oligosaccharide engineering (MOE) is a widely used reactive groups such as ketones, thiols, azides, alkynes, al- method for modifying cellular glycans with non-natural sugars kenes, isonitriles, and cyclopropenes have been installed (Dube and Bertozzi, 2003; Laughlin and Bertozzi, 2009a; Rou- (Cole et al., 2013; Hang et al., 2003; Hsu et al., 2007; Laughlin hanifard et al., 2013). In this method, non-natural monosaccha- et al., 2008; Mahal et al., 1997; Sampathkumar et al., 2006; ride analogs are taken up by cells into biosynthetic salvage Spa¨ te et al., 2014; Stairs et al., 2013). Reactive groups have

112 Cell Chemical Biology 23, January 21, 2016 ª2016 Elsevier Ltd All rights reserved Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

enabled the covalent attachment of a variety of small-molecule and Jork, 1987). At the cellular level, 2-dGal decreased neurite reporters (e.g., biotin, fluorescent dyes, crosslinking agents) outgrowth of embryonic cortical neurons and induced collapse via bioorthogonal chemistry (McKay and Finn, 2014; Sletten of synapses, while 3-deoxygalactose (3-dGal) and 6-deoxyga- and Bertozzi, 2009). Consequently, MOE provides a facile plat- lactose (6-dGal) had no effect (Kalovidouris et al., 2005; Murrey form for endowing glycan structures with new chemical func- et al., 2006). Treatment with Gal reversed the inhibitory effects tionalities. of 2-dGal on neurite outgrowth, presumably by enabling re-syn- A beautiful example is the application of MOE to image cell- thesis of Fuca(1–2)Gal-containing glycans (Kalovidouris et al., surface glycans in developing zebrafish (Laughlin et al., 2008). 2005). Interestingly, Fuca(1–2)Gal glycans were enriched on syn- By incubation of embryos with peracetylated N-azidoacetylga- aptic proteins such as synapsin Ia and Ib (Murrey et al., 2006), lactosamine (Ac4GalNAz), followed by two different strained which regulate neurotransmitter release and synapse formation. cyclooctyne-conjugated fluorophores, newly synthesized gly- 2-dGal disrupted the fucosylation of synapsin and promoted its cans were distinguished from older glycans and tracked in vivo. degradation. Thus, the alteration of Fuca(1–2)Gal glycans on Interestingly, distinct regions of the embryo containing only cell surfaces using non-natural sugars has provided insights newly synthesized glycans were observed, suggesting that the into the functions of these glycans in the nervous system, sug- production of glycoproteins is highly choreographed during gesting that they play roles in the regulation of important synaptic differentiation. To image glycosylation at earlier stages of devel- proteins and morphological changes that underlie synaptic opment, N-azidoacetylsialic acid (NeuAz/SiaNAz) was microin- plasticity. jected into zebrafish embryos to bypass diffusion into the cell An alternative to inhibiting protein glycosylation with deoxy and label sialic acid-containing glycans (Dehnert et al., 2012). sugars is the use of fluorinated sugar analogs, which can serve Dual-labeling experiments were also conducted in C. elegans as inhibitors of GTs. Peracetylated and fluorinated Fuc and with Ac4GalNAz to distinguish regions of active glycan produc- NeuAc analogs were used to block protein fucosylation and tion during both larval and adult stages of development (Laughlin sialylation, respectively, through a two-pronged mechanism and Bertozzi, 2009b). Thus, MOE provides a powerful approach (Figure 6B) (Rillahan et al., 2012). First, the fluorinated mono- to visualize the spatiotemporal dynamics of glycan biosynthesis saccharide inhibitors were converted by salvage pathways to in live organisms. the corresponding activated nucleotide sugar donors. Once In addition to imaging applications, MOE has been used to activated, the inhibitors bound to GTs but were not transferred identify the binding partners of lectins such as the sialic acid- to nascent glycan structures, preventing protein fucosylation or binding immunoglobulin-like lectin (Siglec) CD22 (Han et al., sialylation. The resulting accumulation of sugar donors also 2005). CD22 functions as a negative regulator of B cell receptor stimulated feedback inhibition loops that suppressed de novo activation by establishing a minimum signaling threshold to pre- biosynthesis of Fuc or NeuAc. This approach was utilized to vent aberrant immune reactions (Nitschke et al., 1997). Although inhibit the production of Lewis X (LeX) and sialyl LeX antigens, it was known that cis interactions of CD22 with glycoproteins on resulting in reduced binding to E- and P-selectins and the same cell modulated CD22 activity, the identity of the glyco- enhanced leukocyte rolling, a process important for fighting protein(s) involved was unknown. Cells were incubated with infections. NeuAc containing an aryl-azide moiety (9-AAz-NeuAc) to photo- The remodeling of cell-surface glycans with reactive chemical chemically crosslink NeuAc-bearing glycoproteins to their cis- handles provides another powerful means to modulate cellular binding partners. Surprisingly, the only proteins pulled down responses. For example, the incorporation of a thiol group into with CD22 were other CD22 receptors, which formed self-regu- NeuAc-containing glycoproteins promoted spontaneous cell- lating, homomultimeric complexes. To find the trans-binding cell clustering of Jurkat T cell lymphoma cells, as well as line- partners of CD22, the soluble ectodomain of CD22 was photo- age-specific differentiation of human embryoid body-derived chemically crosslinked to cells displaying 9-AAz-NeuAc-con- (hEBD) stem cells (Figure 6C) (Sampathkumar et al., 2006). The taining glycans (Ramya et al., 2010). The B cell receptor IgM hEBD cells showed a decrease in proliferation, along with an in- was discovered to be the preferred trans-binding partner. crease in b-catenin expression and morphological changes Notably, IgM did not appear to crosslink to CD22 when pre- consistent with differentiation to a neural cell type. Moreover, Ju- sented in cis, underscoring the ability of this MOE approach to rkat cells decorated with thiol-containing NeuAc could be immo- dissect components of different CD22 binding complexes. bilized onto gold or maleimide-functionalized glass surfaces The ability of MOE to remodel glycan structures at the cell sur- without compromising cell viability. Thus, the remodeling of face provides a powerful means to modulate cellular responses. cell surfaces with non-natural sugars can both change their For example, Fuca(1–2)Gal-containing glycans have been impli- adhesive properties and artificially induce cellular signaling cated in cognitive processes such as learning and memory. To processes. study these glycans, 2-deoxygalactose (2-dGal) was used to In another notable example, MOE was used to trigger an arti- prevent the attachment of Fuc to the 2-position of Gal in the ficial immune response against Helicobacter pylori, a bacterium Fuca(1–2)Gal structure (Figure 6A). Intracerebral injection of responsible for stomach ulcers (Kaewsapsak et al., 2013).

2-dGal into day-old chicks induced reversible amnesia and inter- Although tetraacetyl N-azidoacetylglucosamine (Ac4GlcNAz) is fered with the maintenance of long-term potentiation (Bullock incorporated primarily into cytosolic glycoproteins in mammalian et al., 1990; Krug et al., 1991; Rose and Jork, 1987), an electro- cells, H. pylori robustly processed the sugar for cell-surface physiological model of learning and memory. The amnesic effect display (Figure 6D). The azido-functionalized glycoproteins on was not observed with other monosaccharides, and memory for- the cell surface were subsequently conjugated to an immunosti- mation was rescued by co-injection with Gal but not Fuc (Rose mulant, 2,4-dinitrophenol (DNP), via Staudinger ligation. After

Cell Chemical Biology 23, January 21, 2016 ª2016 Elsevier Ltd All rights reserved 113 Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

A Figure 6. Non-Natural Monosaccharides HH Used for MOE HO (A) 2-dGal was incorporated into cell-surface gly- OH GDP cans to prevent attachment of Fuc in neurons. A O representative glycan containing terminal Fuca(1– HO OH X 2)Gal is shown. H (B) Fluorinated analogs of Fuc and NeuAc were metabolized to their respective nucleotide phos- H phate donors and used to inhibit native GTs and affect leukocyte rolling velocity. 2-dGal (C) Ac5ManNTGc was converted to a thiol-containing analog of NeuAc and incorporated into cell- surface glycans to alter cell-cell adhesion and stem cell differentiation. B (D) Ac4GlcNAz was selectively incorporated into cell-surface glycans of Helicobacter pylori, allowing AcO OAc OAc the installation of a functional handle into unknown glycan structures to elicit an immune response AcO O CO Me against the bacteria. AcHN 2 F CMP X OAc F F Chemoenzymatic Labeling

Ac5-3Fax-NeuAc With MOE, the non-natural analog in principle can be incorporated into all glycans containing the monosaccharide of interest. However, determining the precise C HS SH O glycan structures labeled by the non-natural sugar and demonstrating labeling SAc AcO HN HS CMP specificity can be challenging in practice AcO O and can complicate efforts to study glycan AcO OAc function (Boyce et al., 2011; Chuh et al., AcS 2014). The chemoenzymatic labeling (CL) of glycans represents a complementary Ac4ManNTGc strategy to remodel glycans with non-natural functionalities. In this approach, an exogenous GT transfers a non-natural D sugar that contains a reactive group onto OAc the glycan structure of interest (Figure 7). AcO O AcO OAc Subsequent reaction allows for the attach- N NH N3 UDP 3 ment of a variety of chemical reporters as O readouts for the glycan of interest. The specificity of the exogenous GT can be N3 determined using glycan microarrays, N3 and this approach can be used to distinguish and selectively label disaccharide Ac4GlcNAz or higher-order motifs of specific sugar composition and glycosidic linkage, such as Fuca(1–2)Gal and N-acetyllactosamine labeling, the bacteria were treated with anti-DNP antibodies, (LacNAc). Another advantage of this approach is that the chemo- leading to a nearly two-fold increase in bacterial cell death, as enzymatic reactions often proceed quantitatively, which leads to measured by an antibody-dependent cell-mediated cytotoxicity highly sensitive detection. Finally, this system can be employed assay. This method highlights how the differential metabolic in biological contexts where feeding cells non-natural sugar an- utilization of glycans can potentially be exploited to selectively alogs is not possible, such as the ex vivo detection of disease target and elicit an immune response against pathogenic biomarkers. Although identifying a suitable GT that recognizes bacteria. the motif of interest is no small task and may limit the overall Overall, MOE provides a versatile approach to control the scope of the approach, CL has been used both to provide novel glycan architecture on cell surfaces through the incorporation insights into glycan function and to control cellular processes. of non-natural monosaccharides. In addition to facilitating the A powerful example is the application of CL to study the func- detection and study of glycoconjugates, MOE can be exploited tion of O-linked GlcNAc glycosylation on cAMP response to modulate cellular processes and possibly engineer new activ- element-binding protein (CREB), a crucial transcription factor ities through chain termination, installation of reactive groups, involved in neuronal development, plasticity, and long-term and functionalization with bioactive compounds. memory formation (Lonze and Ginty, 2002). Whereas traditional

114 Cell Chemical Biology 23, January 21, 2016 ª2016 Elsevier Ltd All rights reserved Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

HO OH X X Z Z O HO NH OUDP Y X O Exogenous Glycosyltransferase

Figure 7. Chemoenzymatic Labeling In CL, cells are treated with a purified, exogenous GT and an appropriate nucleotide phosphate donor to stoichiometrically attach non-natural sugars onto glycans of defined structure. The functional handle (X) can be further elaborated using bioorthogonal chemistry to detect and study glycans or to endow cells with new biological properties. methods such as wheat germ agglutinin lectin and O-GlcNAc an- in Fuca(1–2)Gal expression on prostate cancer cells compared tibodies failed to detect O-GlcNAc glycosylated CREB, CL using with normal prostate epithelial cells. Thus, this CL method may an engineered b-1,4-galactosyltransferase enzyme and UDP-2- provide a novel tool to discriminate cancerous cells from normal acetonyl-2-deoxygalactose (UDP-ketoGal) allowed for its bio- cells and to study glycans associated with tumor pathogenesis. tinylation and subsequent rapid, sensitive detection (Figure 8A) CL has also been applied to detect and image LacNAc, a uni- (Khidekel et al., 2003; Rexach et al., 2012). To determine the stoi- versal component of the antennae of hybrid- and complex-type chiometry of glycosylation, O-GlcNAc glycosylated proteins N-glycans (Zheng et al., 2011). Here, a-1,3-fucosyltransferase from neuronal lysates were labeled with a 2,000-Da polyethylene (FucT) from H. pylori was combined with GDP-Fuc derivatives glycol mass tag (Rexach et al., 2010), which shifted the molecu- to label LacNAc on the surface of cell lines, primary spleeno- lar weight of the glycosylated proteins. Immunoblotting for cytes, and zebrafish embryos (Figure 8C). To demonstrate its ef- CREB allowed for quantification of the glycosylated species, ficacy at labeling cells ex vivo, T and B cells isolated from mice demonstrating that nearly 50% of the CREB population was were treated in situ with exogenous FucT and GDP-6-azidofu- O-GlcNAc-modified in neurons (Rexach et al., 2012). Further- cose (GDP-FucAz), followed by reaction with strained cyclooc- more, expression of CREB mutants in which glycosylation sites tyne-conjugated biotin and incubation with fluorescently labeled were mutated to alanine, followed by measurement of their streptavidin. Flow cytometry analysis of the labeled cell popula- glycosylation stoichiometry, enabled determination of the major tion revealed that activated immune cells produced higher fluo- glycosylation site in neurons, which was also found to be the site rescence signal than their naı¨ve counterparts, suggesting that induced by neuronal depolarization. Identification of this key site immune cell activation increased cell-surface LacNAc. When facilitated in-depth functional studies and revealed that glycosyl- applied to zebrafish, only a short incubation with the non-natural ation at this site repressed both basal and activity-induced sugar analog and exogenous enzyme were needed to produce CREB-mediated transcription, regulating both neuronal growth robust labeling, thereby obviating the need for microinjection and long-term memory. and minimizing perturbation to the developing animal. Moreover, In another example, a CL approach was developed to detect FucT tolerated fluorescein-labeled GDP-Fuc as a substrate, the potential biomarker Fuca(1–2)Gal, which is upregulated in which avoided the subsequent dye functionalization step. inflammatory diseases and cancer (Figure 8B) (Chaubard et al., In addition to glycan detection, the modification of cell-surface 2012). Terminal Fuca(1–2)Gal residues on complex cell-surface glycans by CL has been exploited to control biological activity. glycans were labeled using a bacterial homolog of the human Although the use of mesenchymal stem cells (MSCs) to regrow blood group A antigen GT (BgtA) and UDP-GalNAz or UDP-keto- tissue and cure skeletal diseases holds great potential (Horwitz Gal. Notably, no mutagenesis of the BgtA active site was et al., 2002), injected MSCs often do not migrate efficiently and required to transfer the non-natural sugars, highlighting the sub- selectively to the tissues of interest. To provide a homing target strate promiscuity of prokaryotic GTs. Glycan microarray data for circulating MSCs, the CD44 glycan on MSCs was modified demonstrated that BgtA specifically labeled glycans containing ex vivo by human-derived a-1,3-fucosyltransferase (FTVI) to pro- a terminal Fuca(1–2)Gal disaccharide motif, in contrast to duce the SLeX glycan motif (Figure 8D) (Sackstein et al., 2008). MOE, which would have labeled all fucosylated glycans. More- This motif binds to E-selectin receptors that are specifically ex- over, BgtA tolerated a variety of distal architectures appended pressed in bone marrow tissue. Engineered HCELL+ MSCs to the reducing end of Gal, labeling diverse N- and O-glycan showed much stronger sheer-resistant adhesive interactions structures with a terminal Fuca(1–2)Gal moiety, including Globo with E-selectin-expressing endothelial cells compared with un- H and Fuc-GM1. Fluorescent labeling of Fuca(1–2)Gal glycans modified MSCs, suggesting that the Fuc modification of CD44 and flow cytometry analysis demonstrated a significant increase glycans endowed cells with new adherent properties.

Cell Chemical Biology 23, January 21, 2016 ª2016 Elsevier Ltd All rights reserved 115 Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

Figure 8. Examples of CL (A) The O-GlcNAc modification was detected and quantified by labeling with ketoGal, followed by installation of a biotin or PEG mass tag. (B) The Fuca(1–2)Gal epitope was selectively labeled using BgtA to install GalNAc derivatives with functional handles for downstream detection. (C) a1,3FucT from H. pylori modified LacNAc moieties with Fuc derivatives for flow cytometry and imaging analysis. (D) Human FTVI was used to functionalize the N-glycans of CD44 on MSCs, thereby producing SLeX moieties to help target MSCs to bone marrow tissue.

Furthermore, HCELL+ MSCs that were injected intravenously scoring the opportunity to expand the collection of GTs and into mice exhibited much greater levels of homing to bone chemical diversity of glycan architectures remodeled by CL. marrow tissue, highlighting the ability of CL to engineer cellular functions in vivo. De novo Glycan Display Overall, CL provides a complementary alternative to MOE for The methods described thus far involve the modiﬁcation of nat- labeling and modifying intracellular and cell-surface glycans. ural glycans on cell surfaces by manipulating the biosynthetic Both methods have strengths in particular biological systems machinery used for glycan production. An alternative approach, and can alter the properties of cells. In the future, mining of pro- which we term de novo glycan display, focuses on inserting karyotic genomes may reveal many novel GT candidates, under- deﬁned glycan or glycomimetic structures directly into plasma

116 Cell Chemical Biology 23, January 21, 2016 ª2016 Elsevier Ltd All rights reserved Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

AB C

Figure 9. De novo Display of Cell-Surface Glycans (A) NeuAc-containing glycans and HS GAGs were chemically appended to a ketone-functionalized polymer and inserted into the plasma membrane using a phospholipid tail. (B) CS GAGs were displayed on the cell surface by functionalization onto ketone-containing liposomes. (C) Genetic incorporation of the HaloTag protein allowed for the immobilization of HS GAGs modified with a chlorohexyl linker. membranes using approaches such as lipid insertion, liposomal erythrocytes (which lack A- and B-type antigens) to allow for lipid fusion, or protein conjugation (Frame et al., 2007; Huang et al., insertion of the glycomimetics in the membrane. The newly 2014; Hudak et al., 2013; Paszek et al., 2014; Pulsipher et al., remodeled blood cells displaying the artificial A- and B-type 2014, 2015). De novo display methods provide excellent control glycolipids acquired a strong immune response to anti-A and over the glycan structure of interest and enable the presentation anti-B antibodies, respectively. This proof-of-concept approach of both small and large glycans at the cell surface. Moreover, this demonstrated that de novo glycan display through the passive approach is not constrained by the substrate specificity of natu- insertion of lipid-anchored glycans into membranes could be ral enzymes, thus expanding the scope of glycan engineering to used to control biological responses. complex structures unattainable by other methods. However, More recently, the immunoevasive properties of tumorigenic exogenous sugars are typically displayed alongside the native cells were examined through membrane insertion of synthetic glycan population, which could obscure the biological effects glycopolymers. Many cancer cells develop mechanisms such of the newly added carbohydrates. To address this complica- as increased surface sialylation to avoid detection by natural tion, de novo glycan display methods can be used in combina- killer (NK) cells in the innate immune response (Bu¨ ll et al., tion with genetic depletion approaches to minimize the contribu- 2014; Fuster and Esko, 2005). Using glycan-functionalized poly- tions of interfering endogenous glycans (Huang et al., 2014). The mers containing a terminal phospholipid, different carbohy- versatility of the technique also allows for the display of a wide drates were anchored to the surface of cancer cells (Figure 9A) range of carbohydrate-based structures, including glycomimet- (Hudak et al., 2013). Cells were then co-incubated with NK cells ics such as synthetic glycopolymers, glycans appended to to determine if the exogenous glycans recapitulated the naturally simplified proteins, or even the glycan component of glycopro- observed avoidance of NK-mediated cell death. Only sialylated teins alone (Huang et al., 2014; Hudak et al., 2013; Paszek NeuAc-containing polymers significantly attenuated cytotox- et al., 2014; Pulsipher et al., 2014, 2015). Despite the use of mini- icity, and the polymer-displaying cells also were found to prevent malist structures and possible interference from endogenous NK degranulation, a key process in cell killing. Notably, the carbohydrates, de novo glycan display has provided remarkable NeuAc polymers protected bone marrow-derived hematopoietic insights into a variety of biological systems. stem cells against NK-mediated cytotoxicity, suggesting that de One of the first examples of de novo glycan display examined novo glycan display of such polymers could help prevent cellular the effects of altering blood group antigens on erythrocyte cell rejection during transplantation. surfaces (Frame et al., 2007). Blood type is determined by the The remodeling of cell surfaces with lipid-anchored glycopol- presence of specific glycolipid structures, the O-, A- and B- ymers has also provided key insights into the mechanobiology of type antigens, on red blood cells. Synthetic glycolipid mimics integrin signaling. Integrins are cell-surface receptors that regu- that contained each antigen type were incubated with O-type late cancer cell migration, invasion, and proliferation (Guo and

Cell Chemical Biology 23, January 21, 2016 ª2016 Elsevier Ltd All rights reserved 117 Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

Giancotti, 2004). Computational modeling studies predicted that stem cell fate by regulating growth factor-mediated signaling the glycocalyx promotes integrin receptor clustering and en- events that determine both pluripotent self-renewal and differen- hances integrin signaling through mechanical interactions tiation (Forsberg et al., 2012; Kraushaar et al., 2012). To explore between bulky carbohydrates and transmembrane proteins whether artificial cell-surface glycans could be used to direct this (Paszek et al., 2009). To test this hypothesis, synthetic mucin- process, synthetic phospholipid-anchored ketone polymers like glycopolymers of different lengths were conjugated to a were functionalized with different aminooxy-tagged HS disac- phospholipid tail and inserted into the membranes of mammary charides produced by enzymatic digestion (Huang et al., 2014). epithelial cells (Paszek et al., 2014). Cells displaying longer Passive insertion of glycomimetics containing both 2-O- and 6- glycopolymers (80 nm) showed segregated regions of clustered O-sulfation, but not other sulfation motifs, activated fibroblast integrins, whereas cells decorated with shorter glycopolymers growth factor 2 (FGF2) signaling pathways involved in stem cell (3 and 30 nm) showed no effect. The larger glycopolymers differentiation. Furthermore, stem cells decorated with the active were found to expand the distance between the cell surface HS glycopolymers formed neural rosettes, suggesting the for- and the extracellular matrix by 19 nm, as determined by scanning mation of intermediate neuroprogenitor cells. angle interference microscopy. This larger spacing enhanced A major challenge of de novo glycan display methods is that the clustering of integrins into fewer, larger focal adhesions, the presentation of glycans at the cell surface is often short- promoting the formation of active signaling nodes on the lived. Lipid-anchored glycans typically exhibit a membrane plasma membrane by physically restructuring the cell surface. half-life of about 6–8 hr (Huang et al., 2014; Hudak et al., Corroborating these data, overexpression of mucin glycoprotein 2013; Pulsipher et al., 2014). To achieve long-lived display, a MUC1 recapitulated the overall heightened membrane and genetically encoded, transmembrane-anchored HaloTag pro- isolated regions of focal adhesion formation. Thus, de novo tein was exploited (Figure 9C) (Pulsipher et al., 2015). The glycan display provided a facile method to modulate systemati- HaloTag protein is a bacterial alkyl dehalogenase that forms a cally the chemical composition and physical parameters of the covalent bond with chloroalkyl ligands (Los et al., 2008), which glycocalyx. allows for attachment of HS GAGs and other molecules. De novo glycan display has also provided a physiologically Notably, fluorescent probes anchored to HaloTag proteins per- relevant model for studying developing neurons in the hippo- sisted on the cell surface for over 8 days. Moreover, stem cells campus. Neuronal growth and synapse formation are highly decorated with highly sulfated HS on HaloTag proteins under- regulated processes in the hippocampus, a brain region critical went an accelerated loss of pluripotency and differentiation for the consolidation of short- and long-term memory. Studies into mature neuronal cells. Homogeneous neuronal cell popula- suggested that chondroitin sulfate glycosaminoglycans (CS tions could be valuable as cell replacement therapies for neuro- GAGs) containing certain sulfation motifs can stimulate the degenerative disorders such as Parkinson’s disease (Freed outgrowth of embryonic hippocampal neurons by recruiting et al., 2001; Mendez et al., 2008). This method constitutes the growth factors to the cell surface and facilitating activation of first example of long-term glycan display, significantly broad- their cognate cell-surface receptors (Clement et al., 1999; ening the potential application of de novo display to a wide Gama et al., 2006; Rogers et al., 2011). However, evidence for range of biological processes. this mechanism was primarily based on the growth-promoting De novo glycan display provides a powerful, versatile means effects of CS GAGs adhered to cell culture plates or other artifi- to direct key signaling events and biological outcomes such as cial substrata. A liposome-based approach was developed to the immune response, cancer proliferation, and stem cell differ- recapitulate the natural display of CS GAGs on neuronal cell sur- entiation. The technique is modular by design, providing a faces and probe the role of sulfation (Pulsipher et al., 2014). Lipo- general platform to test numerous glycan structures and presen- somes were formed from commercially available lipids in the tation architectures that best suit the biological system of presence of 2-dodecanone, which provided a reactive ketone interest. Moreover, the method allows for the controlled display handle for covalent attachment of aminooxy-functionalized CS of large polysaccharides, which are inaccessible to other forms polysaccharides (Figure 9B). Membrane fusion led to presenta- of glycan engineering. De novo glycan display approaches tion of CS GAGs enriched in specific sulfation motifs on the sur- have provided novel insights into the roles of specific carbohy- face of embryonic hippocampal neurons. Notably, only neurons drates in biology and have the potential to direct biomedically displaying the disulfated CS-E motif showed significantly greater important processes such as tissue transplantations or stem neurite outgrowth in response to nerve growth factor (NGF) cell fate. compared with neurons lacking lipid-anchored CS GAGs. The extent of neurite outgrowth could be finely tuned by controlling Outlook the amount of CS-E delivered to the cell surface. As further sup- Considerable progress has been made toward developing novel port for the mechanism, the presence of CS-E led to enhanced methods to remodel the structures of glycans at cell surfaces. activation of NGF-mediated signaling pathways, suggesting These studies have broadly demonstrated the ability to modulate that CS-E facilitates surface recruitment of NGF. Thus, these signaling pathways and important downstream biological events studies demonstrated the proof-of-principle that de novo glycan by controlling glycan-protein interactions. To date, the majority display of CS GAGs can be used to control neuronal signaling of glycan engineering techniques have been applied in vitro. pathways and downstream cellular responses. The dearth of in vivo experiments underscores the need to GAG-mediated cell signaling plays a critical role not only in expand these approaches to complex living systems. The ability neurons but also in stem cell development. For example, previ- to perturb glycans in a tissue-specific manner or at an organism- ous reports have suggested that HS sulfation levels influence wide level will present new opportunities to unravel the

118 Cell Chemical Biology 23, January 21, 2016 ª2016 Elsevier Ltd All rights reserved Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

intricacies of glycan-mediated processes that cannot be fully REFERENCES recapitulated in vitro. Excitingly, recent advances in other fields promise to fuel the Andersson, E.R., Sandberg, R., and Lendahl, U. (2011). Notch signaling: simplicity in design, versatility in function. Development 138, 3593–3612. future development and expansion of glycan engineering technologies. With the advent of genome editing methods such as Baeg, G.H., Selva, E.M., Goodman, R.M., Dasgupta, R., and Perrimon, N. (2004). The wingless morphogen gradient is established by the cooperative action of CRISPR/Cas9 (Hsu et al., 2014), scientists now can more rapidly Frizzled and heparan sulfate proteoglycan receptors. Dev. Biol. 276, 89–100. knock out genes and screen for enzymatic function, and perhaps Belenkaya, T.Y., Han, C., Yan, D., Opoka, R.J., Khodoun, M., Liu, H.Z., and Lin, even more impressively, knock in exogenous or evolved enzymes X.H. (2004). Drosophila Dpp morphogen movement is independent of dyna- with new activities. These advances should greatly facilitate the min-mediated endocytosis but regulated by the glypican members of heparan identification of still unknown GTs and glycosylation-associated sulfate proteoglycans. Cell 119, 231–244. proteins critically involved in development and disease. Further- Bellaiche, Y., The, I., and Perrimon, N. (1998). Tout-velu is a Drosophila homo- more, the integration of facile genetic editing with tissue-specific logue of the putative tumour suppressor EXT-1 and is needed for Hh diffusion. or temporally regulated expression should enable the production Nature 394, 85–88. of animal models that possess modified glycan structures during Beltran-Valero de Bernabe´ , D., Currier, S., Steinbrecher, A., Celli, J., van Beu- specific stages of development, growth, and disease. sekom, E., van der Zwaag, B., Kayserili, H., Merlini, L., Chitayat, D., Dobyns, W.B., et al. (2002). Mutations in the O-mannosyltransferase gene POMT1 Such technologies should also accelerate the expansion of give rise to the severe neuronal migration disorder Walker-Warburg syndrome. MOE and CL to a wider range of non-natural glycan structures. Am. J. Hum. Genet. 71, 1033–1043. Genetic knockins or targeted delivery of agents could produce Bishop, J.R., Schuksz, M., and Esko, J.D. (2007). Heparan sulphate proteogly- specific cell populations and even entire organisms capable of cans fine-tune mammalian physiology. Nature 446, 1030–1037. incorporating a diverse assortment of non-natural sugars inac- Boyce, M., Carrico, I.S., Ganguli, A.S., Yu, S.-H., Hangauer, M.J., Hubbard, cessible to the natural glycosylation machinery, possibly allow- S.C., Kohler, J.J., and Bertozzi, C.R. (2011). Metabolic cross-talk allows labeling multiple glycan populations to be monitored in real time using ing of O-linked beta-N-acetylglucosamine-modified proteins via the N-acetyl- galactosamine salvage pathway. Proc. Natl. Acad. Sci. USA 108, 3141–3146. orthogonal non-natural functionalities. Furthermore, tagged glycans could be purified from living tissues to ascertain the differ- Bru¨ ckner, K., Perez, L., Clausen, H., and Cohen, S. (2000). Glycosyltransferase activity of Fringe modulates Notch-Delta interactions. Nature 406, 411–415. ences in individual glycoprotein structures over time. With the increase in non-natural sugars that could be simultaneously Bu¨ ll, C., den Brok, M.H., and Adema, G.J. (2014). Sweet escape: sialic acids in incorporated, a complementary wider set of functional handles tumor immune evasion. Biochim. Biophys. Acta 1846, 238–246. would provide further power to control cellular function and Bullock, S., Potter, J., and Rose, S.P.R. (1990). Effects of the amnesic agent endow cells with novel properties. 2-deoxygalactose on incorporation of fucose into chick brain glycoproteins. J. Neurochem. 54, 135–142. Ongoing developments in DNA sequencing and bioinformatics have dramatically increased the number of sequenced and anno- Chaubard, J.-L., Krishnamurthy, C., Yi, W., Smith, D.F., and Hsieh-Wilson, L.C. (2012). Chemoenzymatic probes for detecting and imaging fucose-a(1-2)- tated microbial genomes (Markowitz et al., 2012), which could galactose glycan biomarkers. J. Am. Chem. Soc. 134, 4489–4492. provide novel enzymes involved in glycosylation and expand Chen, J., Moloney, D.J., and Stanley, P. (2001). Fringe modulation of Jagged1- the chemical diversity of glycan architectures remodeled by CL. induced Notch signaling requires the action of b4galactosyltransferase-1. As CL is uniquely suited for biomarker detection, these new GTs Proc. Natl. Acad. Sci. USA 98, 13716–13721. could also provide a suite of tools to monitor human pathology Chuh, K.N., Zaro, B.W., Piller, F., Piller, V., and Pratt, M.R. (2014). Changes in and perhaps predict the onset and progression of disease states. metabolic chemical reporter structure yield a selective probe of O-GlcNAc Finally, recent advances in de novo glycan display, particularly modification. J. Am. Chem. Soc. 136, 12283–12295. the expansion of its temporal limits, have opened an exciting new Clement, A.M., Sugahara, K., and Faissner, A. (1999). Chondroitin sulfate E frontier to explore the structure-function relationships of carbo- promotes neurite outgrowth of rat embryonic day 18 hippocampal neurons. hydrates. These methods now present unparalleled opportu- Neurosci. Lett. 269, 125–128. nities to engineer novel signaling pathways and functions into Cole, C.M., Yang, J., Seckute, J., and Devaraj, N.K. (2013). Fluorescent live- cells. Liposomes decorated with tissue-specific ligands could cell imaging of metabolically incorporated unnatural cyclopropene-mannos- amine derivatives. Chembiochem 14, 205–208. be used to deliver exogenous glycans to specific cell populations in vivo. Similarly, protein anchors like the HaloTag protein could Coles, C.H., Shen, Y., Tenney, A.P., Siebold, C., Sutton, G.C., Lu, W., Gal- lagher, J.T., Jones, E.Y., Flanagan, J.G., and Aricescu, A.R. (2011). Proteogly- be incorporated through genetic engineering to direct glycan can-specific molecular switch for RPTPs clustering and neuronal extension. display in model organisms. The attachment of glycans to Science 332, 484–488. various protein architectures could help mimic the natural pre- Cook, A., Raskind, W., Blanton, S.H., Pauli, R.M., Gregg, R.G., Francomano, sentation of sugars as glycoproteins and dissect the importance C.A., Puffenberger, E., Conrad, E.U., Schmale, G., Schellenberg, G., et al. of different protein-carbohydrate combinations. Together, the (1993). Genetic heterogeneity in families with hereditary multiple exostoses. Am. J. Hum. Genet. 53, 71–79. successes in glycan engineering underscore the powerful integration of chemistry and biology and its ability to drive techno- Dehnert, K.W., Baskin, J.M., Laughlin, S.T., Beahm, B.J., Naidu, N.N., logical innovations and new discoveries. Amacher, S.L., and Bertozzi, C.R. (2012). Imaging the sialome during zebrafish development with copper-free click chemistry. Chembiochem 13, 353–357.

Demaegd, D., Foulquier, F., Colinet, A.-S., Gremillon, L., Legrand, D., Mariot, P., ACKNOWLEDGMENTS Peiter, E., Van Schaftingen, E., Matthijs, G., and Morsomme, P. (2013). Newly characterized Golgi-localized family of proteins is involved in calcium and pH homeostasis in yeast and human cells. Proc. Natl. Acad. Sci. USA 110, 6859–6864. This research was supported by the NIH (R01-GM093627 and R01- GM084724) and a National Science Foundation Graduate Research Fellow- Dube, D.H., and Bertozzi, C.R. (2003). Metabolic oligosaccharide engineering ship (DGE-1144469). as a tool for glycobiology. Curr. Opin. Chem. Biol. 7, 616–625.

Cell Chemical Biology 23, January 21, 2016 ª2016 Elsevier Ltd All rights reserved 119 Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

Forsberg, M., Holmborn, K., Kundu, S., Dagalv, A., Kjelle´ n, L., and Forsberg- Kaewsapsak, P., Esonu, O., and Dube, D.H. (2013). Recruiting the host’s im- Nilsson, K. (2012). Undersulfation of heparan sulfate restricts differentiation mune system to target Helicobacter pylori’s surface glycans. Chembiochem potential of mouse embryonic stem cells. J. Biol. Chem. 287, 10853–10862. 14, 721–726.

Foulquier, F., Amyere, M., Jaeken, J., Zeevaert, R., Schollen, E., Race, V., Kalovidouris, S.A., Gama, C.I., Lee, L.W., and Hsieh-Wilson, L.C. (2005). A role Bammens, R., Morelle, W., Rosnoblet, C., Legrand, D., et al. (2012). for fucose-a(1À2)-galactose carbohydrates in neuronal growth. J. Am. Chem. TMEM165 deficiency causes a congenital disorder of glycosylation. Am. J. Soc. 127, 1340–1341. Hum. Genet. 91, 15–26. Keppler, O.T., Stehling, P., Herrmann, M., Kayser, H., Grunow, D., Reutter, W., Frame, T., Carroll, T., Korchagina, E., Bovin, N., and Henry, S. (2007). Synthetic and Pawlita, M. (1995). Biosynthetic modulation of sialic acid-dependent virus- glycolipid modification of red blood cell membranes. Transfusion 47, 876–882. receptor interactions of two primate polyoma viruses. J. Biol. Chem. 270, 1308–1314. Freed, C.R., Greene, P.E., Breeze, R.E., Tsai, W.-Y., DuMouchel, W., Kao, R., Dillon, S., Winfield, H., Culver, S., Trojanowski, J.Q., et al. (2001). Transplanta- Khidekel, N., Arndt, S., Lamarre-Vincent, N., Lippert, A., Poulin-Kerstien, K.G., tion of embryonic dopamine neurons for severe Parkinson’s Disease. N. Engl. Ramakrishnan, B., Qasba, P.K., and Hsieh-Wilson, L.C. (2003). A chemoenzy- J. Med. 344, 710–719. matic approach toward the rapid and sensitive detection of O-GlcNAc post- translational modifications. J. Am. Chem. Soc. 125, 16162–16163. Freeze, H.H. (2013). Understanding human glycosylation disorders: biochemistry leads the charge. J. Biol. Chem. 288, 6936–6945. Kirkpatrick, C.A., Dimitroff, B.D., Rawson, J.M., and Selleck, S.B. (2004). Spatial regulation of wingless morphogen distribution and signaling by dally- Fuster, M.M., and Esko, J.D. (2005). The sweet and sour of cancer: glycans as like protein. Dev. Cell 7, 513–523. novel therapeutic targets. Nat. Rev. Cancer 5, 526–542. Kobayashi, K., Nakahori, Y., Miyake, M., Matsumura, K., Kondo-Iida, E., No- Gama, C.I., Tully, S.E., Sotogaku, N., Clark, P.M., Rawat, M., Vaidehi, N., God- mura, Y., Segawa, M., Yoshioka, M., Saito, K., Osawa, K., et al. (1998). An dard, W.A., III, Nishi, A., and Hsieh-Wilson, L.C. (2006). Sulfation patterns of ancient retrotransposal insertion causes Fukuyama-type congenital muscular glycosaminoglycans encode molecular recognition and activity. Nat. Chem. dystrophy. Nature 394, 388–392. Biol. 2, 467–473. Kopan, R., and Ilagan, M.X.G. (2009). The canonical Notch signaling pathway: Goddeeris, M.M., Wu, B., Venzke, D., Yoshida-Moriguchi, T., Saito, F., Matsu- unfolding the activation mechanism. Cell 137, 216–233. mura, K., Moore, S.A., and Campbell, K.P. (2013). LARGE glycans on dystroglycan function as a tunable matrix scaffold to prevent dystrophy. Nature 503, Koziel, L., Kunath, M., Kelly, O.G., and Vortkamp, A. (2004). Ext1-dependent 136–140. heparan sulfate regulates the range of Ihh signaling during endochondral ossification. Dev. Cell 6, 801–813. Guo, W., and Giancotti, F.G. (2004). Integrin signalling during tumour progres- Kraushaar, D.C., Rai, S., Condac, E., Nairn, A., Zhang, S., Yamaguchi, Y., sion. Nat. Rev. Mol. Cell Biol. 5, 816–826. Moremen, K., Dalton, S., and Wang, L. (2012). Heparan sulfate facilitates Ha¨ cker, U., Nybakken, K., and Perrimon, N. (2005). Heparan sulphate proteo- FGF and BMP signaling to drive mesoderm differentiation of mouse embryonic glycans: the sweet side of development. Nat. Rev. Mol. Cell Biol. 6, 530–541. stem cells. J. Biol. Chem. 287, 22691–22700. Krug, M., Jork, R., Reymann, K., Wagner, M., and Matthies, H. (1991). The Haines, N., and Irvine, K.D. (2003). Glycosylation regulates Notch signalling. amnesic substance 2-deoxy-D-galactose suppresses the maintenance of hip- Nat. Rev. Mol. Cell Biol. 4, 786–797. pocampal LTP. Brain Res. 540, 237–242. Han, C., Belenkaya, T.Y., Khodoun, M., Tauchi, M., Lin, X., and Lin, X. (2004). Laughlin, S.T., and Bertozzi, C.R. (2009a). Imaging the glycome. Proc. Natl. Distinct and collaborative roles of Drosophila EXT family proteins in Acad. Sci. USA 106, 12–17. morphogen signalling and gradient formation. Development 131, 1563–1575. Laughlin, S.T., and Bertozzi, C.R. (2009b). In vivo imaging of Caenorhabditis Han, S., Collins, B.E., Bengtson, P., and Paulson, J.C. (2005). Homomultimeric elegans glycans. ACS Chem. Biol. 4, 1068–1072. complexes of CD22 in B cells revealed by protein-glycan cross-linking. Nat. Chem. Biol. 1, 93–97. Laughlin, S.T., Baskin, J.M., Amacher, S.L., and Bertozzi, C.R. (2008). In vivo imaging of membrane-associated glycans in developing zebrafish. Science Hang, H.C., Yu, C., Kato, D.L., and Bertozzi, C.R. (2003). A metabolic labeling 320, 664–667. approach toward proteomic analysis of mucin-type O-linked glycosylation. Proc. Natl. Acad. Sci. USA 100, 14846–14851. Le Merrer, M., Legeai-Mallet, L., Jeannin, P.M., Horsthemke, B., Schinzel, A., Plauchu, H., Toutain, A., Archard, F., Munnich, A., and Maroteaux, P. (1994). A Hennet, T., and Cabalzar, J. (2015). Congenital disorders of glycosylation: a gene for hereditary multiple exostoses maps to chromosome 19p. Hum. Mol. concise chart of glycocalyx dysfunction. Trends Biochem. Sci. 40, 377–384. Genet. 3, 717–722.

Horwitz, E.M., Gordon, P.L., Koo, W.K.K., Marx, J.C., Neel, M.D., McNall, R.Y., Lichtenstein, R.G., and Rabinovich, G.A. (2013). Glycobiology of cell death: Muul, L., and Hofmann, T. (2002). Isolated allogeneic bone marrow-derived when glycans and lectins govern cell fate. Cell Death Differ. 20, 976–986. mesenchymal cells engraft and stimulate growth in children with osteogenesis imperfecta: implications for cell therapy of bone. Proc. Natl. Acad. Sci. USA 99, Longman, C., Brockington, M., Torelli, S., Jimenez-Mallebrera, C., Kennedy, 8932–8937. C., Khalil, N., Feng, L., Saran, R.K., Voit, T., Merlini, L., et al. (2003). Mutations in the human LARGE gene cause MDC1D, a novel form of congenital muscular Hsu, T.-L., Hanson, S.R., Kishikawa, K., Wang, S.-K., Sawa, M., and Wong, C.- dystrophy with severe mental retardation and abnormal glycosylation of H. (2007). Alkynyl sugar analogs for the labeling and visualization of glycocon- a-dystroglycan. Hum. Mol. Genet. 12, 2853–2861. jugates in cells. Proc. Natl. Acad. Sci. USA 104, 2614–2619. Lonze, B.E., and Ginty, D.D. (2002). Function and regulation of CREB family Hsu, P.D., Lander, E.S., and Zhang, F. (2014). Development and applications transcription factors in the nervous system. Neuron 35, 605–623. of CRISPR-Cas9 for genome engineering. Cell 157, 1262–1278. Los, G.V., Encell, L.P., McDougall, M.G., Hartzell, D.D., Karassina, N., Zim- Huang, M.L., Smith, R.A.A., Trieger, G.W., and Godula, K. (2014). Glycocalyx prich, C., Wood, M.G., Learish, R., Ohana, R.F., Urh, M., et al. (2008). HaloTag: remodeling with proteoglycan mimetics promotes neural speciﬁcation in em- a novel protein labeling technology for cell imaging and protein analysis. ACS bryonic stem cells. J. Am. Chem. Soc. 136, 10565–10568. Chem. Biol. 3, 373–382.

Hudak, J.E., Canham, S.M., and Bertozzi, C.R. (2013). Glycocalyx engineering Mahal, L.K., Yarema, K.J., and Bertozzi, C.R. (1997). Engineering chemical reveals a Siglec-based mechanism for NK cell immunoevasion. Nat. Chem. reactivity on cell surfaces through oligosaccharide biosynthesis. Science Biol. 10, 69–75. 276, 1125–1128.

Jones, M.A., Ng, B.G., Bhide, S., Chin, E., Rhodenizer, D., He, P., Losfeld, M.- Manzini, M.C., Tambunan, D.E., Hill, R.S., Yu, T.W., Maynard, T.M., Heinzen, E., He, M., Raymond, K., Berry, G., et al. (2012). DDOST mutations identiﬁed by E.L., Shianna, K.V., Stevens, C.R., Partlow, J.N., Barry, B.J., et al. (2012). Exome whole-exome sequencing are implicated in congenital disorders of glycosyla- sequencing and functional validation in zebraﬁsh identify GTDC2 mutations as a tion. Am. J. Hum. Genet. 90, 363–368. cause of Walker-Warburg syndrome. Am. J. Hum. Genet. 91, 541–547.

120 Cell Chemical Biology 23, January 21, 2016 ª2016 Elsevier Ltd All rights reserved Please cite this article as: Grifﬁn and Hsieh-Wilson, Glycan Engineering for Cell and Developmental Biology, Cell Chemical Biology (2016), http:// dx.doi.org/10.1016/j.chembiol.2015.12.007 Cell Chemical Biology Review

Markowitz, V.M., Chen, I.-M.A., Palaniappan, K., Chu, K., Szeto, E., Grechkin, Sackstein, R., Merzaban, J.S., Cain, D.W., Dagia, N.M., Spencer, J.A., Lin, Y., Ratner, A., Jacob, B., Huang, J., Williams, P., et al. (2012). IMG: the Inte- C.P., and Wohlgemuth, R. (2008). Ex vivo glycan engineering of CD44 pro- grated Microbial Genomes database and comparative analysis system. Nu- grams human multipotent mesenchymal stromal cell trafﬁcking to bone. Nat. cleic Acids Res. 40, D115–D122. Med. 14, 181–187.

McCormick, C., Leduc, Y., Martindale, D., Mattison, K., Esford, L.E., Dyer, Sampathkumar, S.-G., Li, A.V., Jones, M.B., Sun, Z., and Yarema, K.J. (2006). A.P., and Tufaro, F. (1998). The putative tumour suppressor EXT1 alters the Metabolic installation of thiols into sialic acid modulates adhesion and stem expression of cell-surface heparan sulfate. Nat. Genet. 19, 158–161. cell biology. Nat. Chem. Biol. 2, 149–152.

McKay, C.S., and Finn, M.G. (2014). Click chemistry in complex mixtures: bio- Shi, S.L., and Stanley, P. (2003). Protein O-fucosyltransferase 1 is an essential orthogonal bioconjugation. Chem. Biol. 21, 1075–1101. component of Notch signaling pathways. Proc. Natl. Acad. Sci. USA 100, 5234–5239. Mendez, I., Vin˜ uela, A., Astradsson, A., Mukhida, K., Hallett, P., Robertson, H., Tierney, T., Holness, R., Dagher, A., Trojanowski, J.Q., et al. (2008). Dopamine Sletten, E.M., and Bertozzi, C.R. (2009). Bioorthogonal chemistry: ﬁshing for neurons implanted into people with Parkinson’s disease survive without pa- selectivity in a sea of functionality. Angew. Chem. Int. Ed. Engl. 48, 6974–6998. thology for 14 years. Nat. Med. 14, 507–509. Spa¨ te, A.K., Schart, V.F., Scho¨ llkopf, S., Niederwieser, A., and Wittmann, V. Moloney, D.J., Panin, V.M., Johnston, S.H., Chen, J.H., Shao, L., Wilson, R., (2014). Terminal alkenes as versatile chemical reporter groups for metabolic Wang, Y., Stanley, P., Irvine, K.D., Haltiwanger, R.S., et al. (2000). Fringe is a oligosaccharide engineering. Chemistry 20, 16502–16508. glycosyltransferase that modiﬁes Notch. Nature 406, 369–375. Stairs, S., Neves, A.A., Sto¨ ckman, H., Wainman, Y.A., Ireland-Zecchini, H., Murrey, H.E., Gama, C.I., Kalovidouris, S.A., Luo, W.I., Driggers, E.M., Porton, Brindle, K.M., and Leeper, F.J. (2013). Metabolic glycan imaging by isoni- B., and Hsieh-Wilson, L.C. (2006). Protein fucosylation regulates synapsin la/lb trile-tetrazine click chemistry. Chembiochem 14, 1063–1067. expression and neuronal morphology in primary hippocampal neurons. Proc. Natl. Acad. Sci. USA 103, 21–26. Stevens, E., Carss, K.J., Cirak, S., Foley, A.R., Torelli, S., Willer, T., Tambunan, D.E., Yau, S., Brodd, L., Sewry, C.A., et al. (2013). Mutations in B3GALNT2 Nitschke, L., Carsetti, R., Ocker, B., Ko¨ hler, G., and Lamers, M.C. (1997). CD22 cause congenital muscular dystrophy and hypoglycosylation of a-dystrogly- is a negative regulator of B-cell receptor signalling. Curr. Biol. 7, 133–143. can. Am. J. Hum. Genet. 92, 354–365.

Ohtsuka, Y., Kanagawa, M., Yu, C.-C., Ito, C., Chiyo, T., Kobayashi, K., Okada, Tabata, T., and Kornberg, T.B. (1994). Hedgehog is a signaling protein with a T., Takeda, S., and Toda, T. (2015). Fukutin is prerequisite to ameliorate key role in patterning Drosophila imaginal discs. Cell 76, 89–102. muscular dystrophic phenotype by myofiber-selective LARGE expression. Sci. Rep. 5, 8316. The, I., Bellaiche, Y., and Perrimon, N. (1999). Hedgehog movement is regulated through tout velu-dependent synthesis of a heparan sulfate proteogly- Okajima, T., and Irvine, K.D. (2002). Regulation of Notch signaling by O-linked can. Mol. Cell 4, 633–639. fucose. Cell 111, 893–904. van Kooyk, Y., and Rabinovich, G.A. (2008). Protein-glycan interactions in the Paszek, M.J., Boettiger, D., Weaver, V.M., and Hammer, D.A. (2009). Integrin control of innate and adaptive immune responses. Nat. Immunol. 9, 593–601. clustering is driven by mechanical resistance from the glycocalyx and the substrate. PLoS Comput. Biol. 5, e1000604. van Reeuwijk, J., Janssen, M., van den Elzen, C., Beltran-Valero de Bernabe´ , D., Sabatelli, P., Merlini, L., Boon, M., Scheffer, H., Brockington, M., Mutoni, F., Paszek, M.J., DuFort, C.C., Rossier, O., Bainer, R., Mouw, J.K., Godula, K., et al. (2005). POMT2 mutations cause a-dystroglycan hypoglycosylation and Hudak, J.E., Lakins, J.N., Wijekoon, A.C., Cassereau, L., et al. (2014). The can- Walker-Warburg syndrome. J. Med. Genet. 42, 907–912. cer glycocalyx mechanically primes integrin-mediated growth and survival. Nature 511, 319–325. Wang, Y., Shao, L., Shi, S., Harris, R.J., Spellman, M.W., Stanley, P., and Hal- tiwanger, R.S. (2001). Modification of epidermal growth factor-like repeats Pulsipher, A., Griffin, M.E., Stone, S.E., Brown, J.M., and Hsieh-Wilson, L.C. with O-fucose: molecular cloning and expression of a novel GDP-fucose pro- (2014). Directing neuronal signaling through cell-surface glycan engineering. tein O-fucosyltransferase. J. Biol. Chem. 276, 40338–40345. J. Am. Chem. Soc. 136, 6794–6797. Wells, L. (2013). The O-mannosylation pathway: glycosyltransferases and pro- Pulsipher, A., Griffin, M.E., Stone, S.E., and Hsieh-Wilson, L.C. (2015). Long- teins implicated in congenital muscular dystrophy. J. Biol. Chem. 288, 6930– lived engineering of glycans to direct stem cell fate. Angew. Chem. Int. Ed. 6935. Engl. 54, 1466–1470. Wu, Y.Q., Heutink, P., de Vries, B.B.A., Sandkuijl, L.A., van den Ouweland, Ramya, T.N.C., Weerapana, E., Liao, L., Zeng, Y., Tateno, H., Liao, L., Yates, A.M.W., Neirmeijer, M.F., Galjaard, H., Reyniers, E., Willems, P.J., and Halley, J.R., III, Cravatt, B.F., and Paulson, J.C. (2010). In situ trans ligands of CD22 D.J.J. (1994). Assignment of a second locus for multiple exostoses to the peri- identified by glycan-protein photocross-linking-enabled proteomics. Mol. centromeric region of chromosome 11. Hum. Mol. Genet. 3, 167–171. Cell Proteomics 9, 1339–1351. Yao, D., Huang, Y., Huang, X., Wang, W., Yan, Q., Wei, L., Xin, W., Gerson, S., Rexach, J.E., Rogers, C.J., Yu, S.-H., Tao, J., Sun, Y.E., and Hsieh-Wilson, Stanley, P., Lowe, J.B., et al. (2011). Protein O-fucosyltransferase 1 (Pofut1) L.C. (2010). Quantification of O-glycosylation stoichiometry and dynamics us- regulates lymphoid and myeloid homeostasis through modulation of Notch re- ing resolvable mass tags. Nat. Chem. Biol. 6, 645–651. ceptor ligand interactions. Blood 117, 5652–5662.

Rexach, J.E., Clark, P.M., Mason, D.E., Neve, R.L., Peters, E.C., and Hsieh- Yoshida, A., Kobayashi, K., Manya, H., Taniguchi, K., Kano, H., Mizuno, M., In- Wilson, L.C. (2012). Dynamic O-GlcNAc modiﬁcation regulates CREB-medi- azu, T., Mitsuhashi, H., Takahashi, S., Takeuchi, M., et al. (2001). Muscular ated gene expression and memory formation. Nat. Chem. Biol. 8, 253–261. dystrophy and neuronal migration disorder caused by mutations in a glycosyltransferase, POMGnT1. Dev. Cell 1, 717–724. Rillahan, C.D., Antonopoulos, A., Lefort, C.T., Sonon, R., Azadi, P., Ley, K., Dell, A., Haslam, S.M., and Paulson, J.C. (2012). Global metabolic inhibitors of sialyl- Yoshida-Moriguchi, T., Yu, L., Stalnaker, S.H., Davis, S., Kunz, S., Madson, M., and fucosyltransferases remodel the glycome. Nat. Chem. Biol. 8, 661–668. Oldstone, M.B.A., Schachter, H., Wells, L., and Campbell, K.P. (2010). O-Man- nosyl phosphorylation of alpha-dystroglycan is required for laminin binding. Rogers, C.J., Clark, P.M., Tully, S.E., Abrol, R., Garcia, K.C., Goddard, W.A., Science 327, 88–92. III, and Hsieh-Wilson, L.C. (2011). Elucidating glycosaminoglycan-protein- protein interactions using carbohydrate microarray and computational ap- Yoshida-Moriguchi, T., Willer, T., Anderson, M.E., Venzke, D., Whyte, T., Mun- proaches. Proc. Natl. Acad. Sci. USA 108, 9747–9752. toni, F., Lee, H., Nelson, S.F., Yu, L., and Campbell, K.P. (2013). SGK196 is a glycosylation-speciﬁc O-mannose kinase required for dystroglycan function. Rose, S.P.R., and Jork, R. (1987). Long-term memory formation in chicks is Science 341, 896–899. blocked by 2-deoxygalactose, a fucose analog. Behav. Neural Biol. 48, 246–258. Zheng, T., Jiang, H., Gros, M., Soriano del Amo, D., Sundaram, S., Lauvau, G., Rouhanifard, S.H., Nordstrøm, L.U., Zheng, T., and Wu, P. (2013). Chemical Marlow, F., Liu, Y., Stanley, P., and Wu, P. (2011). Tracking N-acetyllactosamine probing of glycans in cells and organisms. Chem. Soc. Rev. 42, 4284–4296. on cell-surface glycans in vivo. Angew. Chem. Int. Ed. Engl. 50, 4113–4118.