Advances in Engineering Nucleotide Sugar Metabolism for Natural Product Glycosylation in Saccharomyces cerevisiae

Glycosylation is a ubiquitous modification present across all of biology, affecting many things such as physicochemical properties, cellular recognition, subcellular localization, and immunogenicity. Nucleotide sugars are important precursors needed to study glycosylation and produce glycosylated products. Saccharomyces cerevisiae is a potentially powerful platform for producing glycosylated biomolecules, but it lacks nucleotide sugar diversity. Nucleotide sugar metabolism is complex, and understanding how to engineer it will be necessary to both access and study heterologous glycosylations found across biology. This review overviews the potential challenges with engineering nucleotide sugar metabolism in yeast from the salvage pathways that convert free sugars to their associated UDP-sugars to de novo synthesis where nucleotide sugars are interconverted through a complex metabolic network with governing feedback mechanisms. Finally, recent examples of engineering complex glycosylation of small molecules in S. cerevisiae are explored and assessed.


■ INTRODUCTION
Glycosylation is an important modification found across biology present in proteins, lipids, and natural products, and it affects structure and function as well as stability and solubility. 1−3 It is also vital for energy storage as well as cellular communication.Glycosylation is achieved in biology through glycosyltransferases, with the largest family being uridinediphosphate-dependent glycosyltransferases (UGTs). 1 Nucleotide diphosphate (NDP)-sugars serve as the monomer substrates for more than 90% of glycosylation reactions and are essential building blocks for naturally occurring polysaccharides and glycoconjugates. 1 The most common glycosylation precursor is uridine diphosphate α-D-glucose (UDP-D-Glc) (Figure 1).Although there are a wide variety of different nucleotide sugars, glycosylation reactions involving these precursors are understudied due to lack of availability of nucleotide sugars. 4−10 Yeast is also more likely to be easier to engineer to make complex molecules compared with other hosts.Unlike prokaryotic hosts, yeast contains an endoplasmic reticulum (ER) necessary for membrane-bound enzymes involved in plant natural product biosynthesis.Yeast does have an abundant amount of UDP-D-Glc that has been used to glycosylate different molecules such as ginsenosides 11 and steviol. 12−15 This is vastly limiting, considering the plethora of different sugars synthesized across biology that serve as building blocks of a wide variety of different products.In order to increase the diversity of products, the diversity of nucleotide sugars accessible in yeast must increase.
Nucleotide sugars can be made in many different ways (Figure 2). 16Free sugars, such as rhamnose and arabinose, can be salvaged and converted to their activated nucleotide forms. 16Alternatively, nucleotide sugars can be accessed by enzymes that provide a biosynthetic network to interconvert existing nucleotide sugars.This process allows for broad structural variety of NDP-sugar monomers via reactions such as oxidation, decarboxylation, epimerization, etc. 16 Nucleotide sugar metabolism primarily takes place in the cytosol, though it can also occur in the Golgi apparatus, where polysaccharide biosynthesis and protein glycosylation occur.This metabolism is complex and complicated with different feedback mechanisms that make it difficult to tune nucleotide sugar conversion. 16,17This review will focus on the biosynthesis of UDP-sugars (see Table 1) and potential challenges with engineering their metabolism as well as recent engineering efforts toward producing complex glycosylated products in yeast.

IN YEAST
For practicality, it is easiest to make heterologous UDP-sugars in yeast by using either the salvage pathway or the de novo pathway.
Salvage Pathway.The salvage pathway involves converting a simple sugar to its UDP-sugar form by a specific kinase and a UDP-sugar pyrophosphorylase that can be either specific or promiscuous.The kinase phosphorylates the sugar at the C1 residue to make the sugar-1-phosphate, and then the UDPsugar pyrophosphorylase converts it to the UDP-sugar (see Figure 3).Kinases are typically highly specific for their sugar substrates. 16For some sugars, such as D-xylose and D-fucose, the kinases have not been described, though there have been efforts to engineer promiscuous kinases. 18UDP-sugar pyrophosphorylases are either specific, in the case of UDP-D-Glc pyrophosphorylases (UGPs) and UDP-D-N-acetylglucosamine pyrophosphorylases (UAGPs), or quite promiscuous, such as in the case of "sloppy" UDP-sugar pyrophosphorylases (USPs) that can show different levels of promiscuity toward different sugar-1-phosphates. 19he salvage pathway may be advantageous for the synthesis of specific UDP-sugars that have readily available substrates and pathway enzymes.One such pathway is the myo-inositol pathway for UDP-D-GlcA production, where myo-inositol is oxidatively cleaved by myo-inositol oxidase to form D- glucuronic acid, which can then be converted to UDP-D-GlcA via the salvage pathway with D-glucuronic acid kinase Enzymes in the cytosol that interconvert or salvage UDP-sugars can be found either purely cytosolic or anchored to the cell membrane, Golgi apparatus, etc. S. cerevisiae's native metabolism is depicted with black arrows, and non-native pathways to access other UDP-sugars are depicted with red arrows.A full list of enzymes needed for each part of this metabolic network can be found in Table S1.followed by UDP-sugar pyrophosphorylase. 20UDP-L-Arap can also be synthesized from free arabinose utilizing the salvage pathway with arabinokinase and UDP-sugar pyrophosphorylase. 21The salvage pathway is also orthogonal to yeast's native metabolism and does not rely on nucleotide sugar interconversion, which may suffer from low yield.This pathway may be the best way to synthesize nonstandard sugars such as fluorinated sugars. 22It is unfortunately limited in other ways, as not every sugar has a known kinase (e.g., xylose) and individual sugars can be prohibitively expensive to feed in or are not commercially available.For the production of a wide variety of UDP-sugars, the salvage pathway is not very optimal.
De Novo Pathway.The de novo pathway interconverts UDP-sugars into different UDP-sugars directly, typically starting from UDP-D-Glc, and can access a wide variety of UDP-sugars.A broad class of nucleotide sugar interconversion enzymes carry out these reactions by modifying the sugar attached to UDP through dehydration, reduction, decarboxylation, epimerization, ring restructuring, etc. 16 As many sugars are either prohibitively expensive or unavailable commercially and some do not have a known kinase (e.g., xylose), the de novo pathway using nucleotide sugar interconversion enzymes is often advantageous.Starting from UDP-D-Gal and UDP-D-Glc (as yeast makes both), many other UDP-sugars can be synthesized by the de novo pathway (Figure 4).A small number of these enzymes have been previously tested in yeast, 23,24 but there are still many challenges to address in engineering these pathways.
Yeast natively makes UDP-D-Glc and UDP-D-Gal from either glucose or galactose. 25-Glucose can be phosphorylated to D-Glc-6-P and isomerized by phosphoglucomutase (PGM) to D-Glc-1-P, which is then converted to UDP-D-Glc by UGP.D-Galactose can be phosphorylated by galactokinase (GALK)  to make D-Gal-1-P, which can then be converted by galactose-1-phosphate uridylyltransferase (GALT) with UDP-D-Glc to form UDP-D-Gal and D-Glc-1-P (this is also known as the Leloir pathway).17 S. cerevisiae also contains the enzyme UDPglucose 4-epimerase (UGE), which interconverts UDP-D-Glc and UDP-D-Gal by C4 epimerization.25 Most of these UDP-sugars originate from UDP-D-Glc.UDP-L-Rha is made from UDP-D-Glc via the UDP-L-Rha synthase (RHM), which is a large three-domain enzyme that performs a 4,6-dehydration (N-terminus), 3,5-epimerization, and 4-keto reduction (C-terminus).24 UDP-D-Fuc is made in a similar manner from UDP-D-Glc with a 4,6-dehydration performed by UDP-D-Glc 4,6-dehydratase (UG46DH) followed by a 4-keto reduction by a homologue of neomenthol dehydrogenase (NMD) without a 3,5-epimerization.26 UDP-D-GlcA is made from UDP-D-Glc by UDP-D-Glc 6dehydrogenase (UGD), which oxidizes the C6 hydroxy residue on D-glucose to the carboxylic acid.23 UDP-D-GlcA can be reversibly interconverted to UDP-D-Gal via C4 epimerization by UDP-D-GlcA 4-epimerase (UGlcAE).27 UDP-D-Xyl synthase (UXS) performs a C6 decarboxylation on UDP-D-GlcA to make UDP-D-Xyl.28 A similar C6 decarboxylation performed by UDP-GalA decarboxylase (UGC) that converts UDP-D-GalA to UDP-L-Arap has been reported in the pathogenic fungus Ampullariella digitata.29 However, UGC activity has not been reported from other species, and the putative enzyme has not been identified.UDP-D-GlcA can also be converted to UDP-D-Api by UDP-D-Api/Xyl synthase (AXS), which performs a C6 decarboxylation and ring contraction.30 AXS also makes UDP-D-Xyl by performing only the decarboxylation reduction.UDP-Xyl 4-epimerase (UXE) interconverts UDP-D-Xyl and UDP-L-Arap to about equal molar concentrations.31 UDP-L-Araf, the furanose form of UDP-L-Arap and the primary form of arabinose found in plant saccharides, 21,32 is made from UDP-L-Arap by UDP-L-Ara mutase (UAM).32 This enzyme has also been named "reversible glycosylated polypeptide" (RGP).32 Similarly, UDP-D-Galf is made from UDP-D-Gal by UDP-D-Gal mutase (UGM).33 Both UAM and UGM are ruled by thermodynamic equilibria that favor the pyranose form of the sugar over the furanose form: for UAM this equilibrium is 90:10 favoring UDP-L-Arap, 32 and for UGM this equilibrium is 11:1 favoring UDP-D-Gal.33 There are many common reactions involved in the interconversion of UDP-sugars including C6 oxidation, C6 decarboxylation, C4 epimerization, C4,6 dehydration, ring restructuring, etc. C4 epimerization is commonly used to interconvert UDP-D-Glc/UDP-D-Gal, UDP-D-GlcNAc/UDP-GalNAc, UDP-D-GlcA/UDP-D-GalA, and UDP-D-Xyl/UDP-L-Arap. Some epimerases have been discovered to work on nucleotide sugars other than UDP-D-Glc/UDP-D-Gal, namely, UDP-D-GlcA/UDP-D-GalA and UDP-D-Xyl/UDP-L-Arap, which is advantageous because UGlcAE and UXE are usually membrane-bound and active in the Golgi lumen. 34C4,6 dehydration followed by 4-keto reduction is common to both UDP-L-Rha and UDP-D-Fuc biosynthesis, and a similar 4,6dehydration, 3,5-epimerization, and 4-keto reduction is used to convert GDP-D-mannose to GDP-L-fucose.16 It is widely observed across biology that UDP-D-Xyl is a strong inhibitor of its upstream enzyme UGD even in organisms that do not make UDP-D-Xyl.23,35 UDP-D-Xyl allosterically binds to UGD with a very high affinity.Previously, UGD1 from Arabidopsis thaliana was measured to have a K i of 4.9 μM for UDP-D-Xyl and 99 μM for UDP-D-GlcA.23 It is also a known inhibitor of other enzymes involved in UDP-sugar metabolism such as UGP, 36 UXS, 28 and RHM.24 There have been efforts to alleviate this inhibition of UGD by implementing point mutations on the Homo sapiens UGD that increased the apparent K i of UDP-D-Xyl 10-fold.37 UDP-D-Xyl is hypothesized to be a very important nucleotide sugar in Figure 6. of glycosylated terpene products made in S. cerevisiae and their biosynthetic pathways. Nave yeast metabolism is shaded in orange, steviol glycoside biosynthesis in blue, and saponin biosynthesis in green.
regulating nucleotide sugar metabolism, mostly in maintaining a high UDP-D-Glc concentration.Its strong inhibition of UGD likely helps control the irreversible conversion of UDP-D-Glc and the synthesis of UDP-pentose sugars.
Another common feedback mechanism is the inhibition of the N-terminus of RHM (the 4,6-dehydratase domain) by the end product of UDP-L-Rha. 24This inhibition again is likely to maintain high UDP-D-Glc levels in vivo, as the synthesis of UDP-L-Rha from UDP-D-Glc is irreversible.As UDP-D-Xyl is also a known inhibitor of the N-terminal domain of RHM, 24 it seems that UDP-D-Xyl is an important overall regulator of UDP-sugar metabolism.In terms of de novo synthesis, there seems to be either reversible interconversion or high inhibition that serves to prevent UDP-D-Glc depletion.−39

GLYCOSYLATION IN S. cerevisiae
Engineering S. cerevisiae to produce glycosylated molecules requires a variety of tools and strategies.Examples of different glycosylated natural products include flavonoids, polyketides, and terpenoids, which can be found in nature to be decorated by a variety of different sugars (Figure 5).Many of the molecules of interest for production in yeast are derived from terpene and flavonoid metabolism.
Terpenoids are a diverse class of molecules, comprising more than 50,000 known molecules, mostly coming from plants. 40Examples of terpenes include saponins, which are derived from triterpene (C30) 2,3-oxidosqualene.Saponins are a structurally diverse set of natural products found in many types of plants, consisting of a triterpenoid or steroid core that is glycosylated at one or more residues. 41,42They are widely desired for a variety of purposes, ranging from the adjuvant and anticancer activity of QS-21 43 to the anti-inflammatory aescin 44 and the low-calorie sweetener glycyrrhizin, which may also be a COVID-19 therapeutic. 45,46The biosynthesis of saponins requires 2,3-oxidosqualene cyclases (OSCs) that cyclize 2,3-oxidosqualene (derived from the mevalonate pathway) to a four-or five-ring structure core, typically followed by cytochrome P450-dependent monooxygenases (P450s) that are localized to the ER and oxidize selected carbon residues on the 30-carbon core.Further modification involves UDP-glycosyltransferases (UGTs) that glycosylate selected oxygen residues with specific sugars and are typically in the cytoplasm (Figure 6).Various other enzymes, such as methyltransferases, acyltransferases, etc., can also be a part of saponin synthesis. 41The synthesis of other terpene glycosides such as rebaudioside M, a diterpene glycoside, similarly involves P450s and UGTs. 40,47S. cerevisiae natively harbors the mevalonate-based terpene synthesis pathways, so biosynthesis typically requires tailoring enzymes from plants such as P450s, UGTs, and enzymes to increase nucleotide sugar supply or diversity.
Flavonoids are synthesized from phenylalanine via the shikimate pathway and are similarly oxidized by P450s and glycosylated by UGTs (Figure 7).The biosynthesis of flavonoids in yeast has been achieved. 48However, most examples of this type of glycosylation employ the feed-in of the associated flavonoid aglycone due to challenges in engineering the full pathway.
Increasing UDP-D-Glc Pools and Knocking out Glucosidases in Improving Flavonoid-7-O-Glc Production.For the production of flavonoid-7-O-glycosides by S. cerevisiae, efforts have been made to engineer UDP-D-Glc pools and knock out known endogenous glucosidases.To increase production of glycosylated products, the deletion of endogenous glucosidases will be crucial, as these enzymes can noticeably degrade production. 49In increasing the titer of naringenin-7-O-Glc production in S. cerevisiae, Li et al. knocked out both known flavonoid-7-O glycoside glucosidases, EXG1 and SPR1, in addition to increasing UDP-D-Glc pools by overexpressing enzymes involved in UTP biosynthesis, UDP-D-Glc biosynthesis, and D-glucose uptake. 50This resulted in a 7.9-fold increase in the titer of their desired product (2S)naringenin.A similar approach was taken in the production of scutellarein 7-O-glucoside, where UGP1 and PGM2 were overexpressed and the genes encoding the native βglucosidases, EXG1, SPR1, and YIR007W, were knocked out, resulting in 1.2 g/L of scutellarein 7-O-glucoside produced in a 4 L reactor supplemented with ∼3.5 g of scutellarein. 51These classic metabolic engineering strategies of knocking out degradation enzymes and overexpressing upstream enzymes to increase precursor pools are important to improve glycoside biosynthesis.Specifically increasing UDP-D-Glc pools will be important for glycosylating with nucleotide sugars that are derived from UDP-D-Glc, such as UDP-D-GlcA, UDP-L-Rha, etc., due to complicated feedback mechanisms that preserve UDP-D-Glc levels, as highlighted in the previous section of this review.
Non-glucose Glycosylation in S. cerevisiae.As mentioned before, most reports of natural product glycosylation in yeast have been restricted to glucosylations, which can take advantage of the endogenous pool of UDP-D-Glc in yeast.However, there are several examples of successful glycosylations.Flavonoids that serve as precursors to breviscapine were synthesized in yeast, including the production of apigenin 7-O-GlcA and scutellarin (which contains a D-GlcA residue) by introducing EbUGD and its respective flavonoid 7-O-glucuronosyltransferase EbF7GAT from Erigeron breviscapus; titers were increased by increasing the flux of the precursor malonyl-CoA to improve flavonoid production (Figure 6). 13Recently, the biosynthesis of flavonoid 7-O-disaccharides was reported in S. cerevisiae, specifically the production of eriocitrin, naringin, hesperidin, neohesperidin, diosmin, and neoeriocitrin, each of which contains either a D-glucose-(1,2)-L-rhamnose or D-glucose-(1,6)-L-rhamnose disaccharide. 14The biosynthesis of precursor UDP-L-Rha was achieved by introducing RHM from A. thaliana into S. cerevisiae, and further optimization of glycosylation was achieved by both creating a chimeric UDP-L-Rha-producing enzyme that is able to regenerate NADH and overexpressing upstream enzymes in UDP-D-Glc biosynthesis, resulting in titers of hundreds of mg/L of flavonoid 7-Odisaccharides with a feed-in of the respective aglycones. 14. cerevisiae has been important in the discovery of pathways in saponin synthesis and the characterization of UGTs.The production of glycyrrhizin, a desired sweetener that is 150 times sweeter than sucrose, was achieved at 225.3 μg/L in S. cerevisiae with the discovery of a cellulose synthase-like enzyme that glucuronidates the C3 residue of glycyrrhetinic acid and expression of UGD from A. thaliana to produce UDP-D-GlcA. 9his was reported at the same time as the discovery that cellulose synthase-like enzymes are involved in the pathway for producing yossoside V, an abundant saponin found in spinach, where the production of medicagenic acid 3-O-GlcA was achieved in S. cerevisiae with UGD from Spinacia oleracea, producing UDP-D-GlcA. 8However, significant amounts of unglycosylated medicagenic acid remained, showing that there is an opportunity for optimization to either improve glycosyltransferase activity or UDP-sugar availability. 8Yossosides I−V also contain a D-fucose residue, and since there was no known UDP-D-Fuc biosynthetic pathway, the fucosylation was tested and confirmed in Nicotiana benthamiana.The recent discovery of the biosynthesis of UDP-D-Fuc via UDP-D-Glc 4,6-dehydratase and a homologue of neomenthol dehydrogenase is exciting for the potential to produce many glycosylated products that contain this elusive sugar. 26The addition of D-xylose to make both notoginsenoside R1 and R2 was achieved in yeast with the integration of AtUGD and AtUXS to produce UDP-D-Xyl as well as the discovery of UGT94Q13 that performs the xylosylation to produce both saponins. 52here have been some efforts in engineering heterologous nucleotide sugar pathways from plants and microbes in yeast to study the metabolism of nucleotide sugar interconversion. 53articularly, delaying the production of UDP-D-Xyl can allow for higher accumulation of the upstream product UDP-D-GlcA. 53Also, it was observed that the production of UDP-L-Rha can inhibit the production of UDP-D-Fuc. 53Studies such as these may become more important to understand bottlenecks caused by UDP-sugar interconversion limits.
Engineering UGTs and Their Expression in S. cerevisiae.There are a few examples of increasing glycosylation efficiency either by traditional metabolic engineering methods of increasing enzyme expression through promoter engineering and testing homologues 54 or by using site-directed mutagenesis to increase enzyme activity. 47roduction of ginsenoside Rh2 at 2.25 g/L in a 10 L bioreactor (79.3 mg/L in shake flasks) was achieved in S. cerevisiae both through engineering higher titers of the aglycone protopanaxadiol by balancing P450 levels and by improving the C3 glycosylation conversion through increasing the copy number, improving protein expression through promoter engineering, and increasing enzyme activity through directed evolution. 54There have been some efforts to engineer UGTs that can perform a wide variety of glycosylation reactions.The production of artificial sweeteners rebaudioside D (Reb D) and rebaudioside M (Reb M) was increased in S. cerevisiae relative to alternative products 1,2-stevioside and rebaudioside A by mutating UGT76G1, an enzyme that is able to glucosylate both steviol C13 and C19 residues and perform up to eight different glucosylation reactions. 47These mutations were achieved by homology modeling to target 1,3glycosylation and site-directed mutagenesis that was then screened to find enzyme mutants that increased accumulation of Reb D and Reb M. 47 In order to screen large libraries of mutants and homologues of UGTs, coupled in vitro fluorescence assays that could be used to screen UGT variants in a 384-well format though the correlation between in vitro and in vivo in S. cerevisiae would need to be tested. 55OUTLOOK It is a very exciting time to engineer the production of heterologous products in yeast with the discovery of new biosynthetic pathways and the development of new strategies for metabolic engineering.While this review has focused on small-molecule glycosylation, yeast is also a potential host for production of other glycosylated products such as proteins. 56It is important to note that multiplexed approaches will be necessary to balance UDP-sugar concentration, concentration of aglycones, and enzyme expression.
Beyond UGTs and precursors, other efforts also involve the engineering of yeast's native morphology to increase glycosylated products, for example, the expansion of the endoplasmic reticulum to allow for a larger area for P450 docking to increase product titers.The knockout of phosphatidic acid phosphatase (PAH1), an enzyme that generates neutral triglycerides from phosphatidic acid in S. cerevisiae, increased the production of medicagenic acid C28-O-Glc 16-fold. 57The expansion of the ER can be achieved by overexpressing INO2, an important ER size regulatory factor, that increased production of the ginsenoside aglycone protopanaxadiol 8-fold in S. cerevisiae. 58The engineering of yeast to produce such complexly decorated molecules will continue to require a careful balance of relative gene expression of a variety of enzymes, production balancing of various precursors, and metabolic burden.
With the synthesis of large molecules that come from heterologous species, transporters and accessory proteins may become increasingly important for the glycosylation of these natural products.The plant vacuole is often used to store secondary metabolites like saponins, and ABC-type transporters are likely used to allow these metabolites to enter the vacuole. 59,60For example, avenacin A1 is a saponin found in Avena strigosa, where the last two biosynthetic steps occur in the vacuole, including the final glycosylation step, and this molecule could be synthesized in N. benthamiana without the addition of any transporters. 15,61These transporters are largely uncharacterized, though it is likely that many plants share this machinery.Recent achievements in the production of medicinal tropane alkaloids in S. cerevisiae show that it is possible to engineer up to six subcellular compartments in yeast, including the vacuole and the peroxisome, for the production of heterologous small products. 6t may also be advantageous to utilize yeast native compartments, such as the peroxisome, for heterologous natural product biosynthesis.The peroxisome has already been repurposed from its native function to easily import heterologously expressed enzymes to produce natural products, showing its potential as a compartment of engineered specialized metabolism. 62It has also been used for the storage of protopanaxadiol in the peroxisomal membrane to increase production by 78%. 63s stated in the Introduction, S. cerevisiae is an advantageous platform compared with others for a variety of reasons.Yeast may also express heterologous enzymes like nucleotide sugar enzymes found in plants, whereas other microbes like Escherichia coli have a hard time expressing activated forms of the enzyme UGD that produces UDP-D-GlcA. 64,65Yeast also grows much faster compared to native producers of these products, such as plants, which can take decades to mature.Alternative plant platforms such as N. benthamiana offers the advantage of containing more metabolic precursors, such as nucleotide sugar diversity, and possibly contain other helper enzymes not yet identified. 2 However, the most sufficient of these methods requires transient expression using agrobacterium infiltration that produces gram-scale quantities on the scale of months.We hope that the expansion of nucleotide sugar metabolism in yeast will lead to the establishment of stable strains that can easily be used to test different glycosylations in a matter of weeks and eventually establish stable strains for the large-scale production of glycosylated molecules.
There are significant challenges with balancing the production of UDP-sugar metabolism with internal feedback mechanisms.With the elucidation and engineering of UDPsugar metabolism and UGTs, not only will there be more production of glycosylated natural and new-to-nature products and proteins but also a greater understanding of glycosylation in general.

■ ASSOCIATED CONTENT
* sı Supporting Information

Figure 1 .
Figure 1.Structure of uridine diphosphate α-D-glucose (UDP-D-Glc) consisting of uracil, ribose, and diphosphate (which compose UDP) and D-glucose.This is a common activated form of D-glucose which is essential for recognition by UDP-dependent glycosyltransferases and its use in glycosylation.

Figure 2 .
Figure 2. Major pathways to produce different UDP-sugars from simple sugars in vivo.Pathways that start from individual sugars with known enzymes are included.Gray dashed arrows indicate transporters from the cytosol into the lumen of the Golgi apparatus.Many UDP-sugars are transported into the Golgi apparatus by specific UDP-sugar transporters and can be interconverted in the Golgi lumen, but only a few are shown as examples.Enzymes in the cytosol that interconvert or salvage UDP-sugars can be found either purely cytosolic or anchored to the cell membrane, Golgi apparatus, etc. S. cerevisiae's native metabolism is depicted with black arrows, and non-native pathways to access other UDP-sugars are depicted with red arrows.A full list of enzymes needed for each part of this metabolic network can be found in TableS1.

Figure 3 .
Figure 3. Salvage pathway.The salvage pathway consists of the phosphorylation of a sugar by an associated kinase and then the transfer of a UMP moiety from UTP by a UDP-sugar pyrophosphorylase.

Figure 5 .
Figure 5. Examples of glycosylated natural products from flavonoids to polyketides and terpenoids.Sugars are labeled and highlighted in red.

Table 1 .
Sugars Discussed in This Review and Their Activated Forms