Targeting Glycosylation as a Therapeutic Approach

© 2001 Macmillan Magazines Ltd

R E V I E W S

NATURE REVIEWS | DRUG DISCOVERY VOLUME 1 | JANUARY 2002 | 65

Current estimates indicate that genes that regulate N- and O-glycosylation of glycoproteins make up1–2% of the human genome. In the case of N-linkedGLYCANS, more than 30 enzymes, located in the cytosol,the ENDOPLASMIC RETICULUM (ER) and the GOLGI APPARATUS,are required to generate, attach and process theoligosaccharides (FIG. 1). Many functions have beendescribed for protein glycosylation, including promot-ing protein folding in the ER1, stabilizing cell-surfaceglycoproteins2, and providing recognition epitopes thatactivate the innate immune system3. It is therefore notsurprising that genetic mutations that decrease or elimi-nate the activity of GLYCOSYLTRANSFERASES and GLYCOSIDASES

can lead to serious physiological disorders and can belethal in animals as well as in humans4.

Glycoproteins generally exist as populations ofglycosylated variants of a single polypeptide, knownas glycoforms5. The attachment and processing ofsugars is not random, but exquisitely controlled byboth the cell and the three-dimensional structure ofthe protein itself. Changes in protein glycosylation areearly indicators of cellular changes in many diseases, mostnotably cancer6 and rheumatoid arthritis7, providinguseful diagnostic markers and insights into diseaseprogression and pathogenesis.

Important physiological functions have also beenestablished for glycosylated sphingolipids, which formtwo distinct families: the glucosphingolipids and the

galactosphingolipids.Here,we will focus exclusively on theglucosphingolipids (GSLs), which are synthesized fromthe lipid ceramide by several carbohydrate-processingenzymes. GSLs are present on the plasma membranes ofall mammalian cells, and are essential during embryonicdevelopment and differentiation8. They are also exploited as receptors by some bacteria and viruses.Gangliosides — a subclass of GSLs that contain sialicacid — are particularly enriched in neuronal tissues andare important for proper nervous-system function.Maintenance of a balanced population of GSLs on cellmembranes requires stringent control of GSL biosynthe-sis and degradation. However, in GSL storage diseasessuch as Tay–Sachs disease, defects in the degradationpathway result in the accumulation of GSLs in cells, par-ticularly in neurons, which causes neurodegenerationand a shortened lifespan9.

When assessing the potential of drugs that targetprotein and lipid glycosylation, the overriding consider-ation must be the importance of host glycosylation.It is only realistic to use drugs that completely ablatethe activity of carbohydrate-processing enzymes incases in which the target enzyme has no mammaliancounterpart; for example, pathogen-specific enzymesinvolved in the synthesis of the carbohydrate com-ponents of bacterial and fungal cell walls. Otherwise,specific strategies that do not involve totally eliminatingenzyme activity are necessary. Here, we describe two

TARGETING GLYCOSYLATION AS A THERAPEUTIC APPROACHRaymond A. Dwek, Terry D. Butters, Frances M. Platt and Nicole Zitzmann

Increased understanding of the role of protein- and lipid-linked carbohydrates in a wide rangeof biological processes has led to interest in drugs that target the enzymes involved inglycosylation. But given the importance of carbohydrates in fundamental cellular processessuch as protein folding, therapeutic strategies that modulate, rather than ablate, the activity ofenzymes involved in glycosylation are likely to be a necessity. Two such approaches that useimino sugars to affect glycosylation enzymes now show considerable promise in the treatmentof viral infections, such as hepatitis B, and glucosphingolipid storage disorders, such asGaucher disease.

Glycobiology Institute,Department ofBiochemistry, University ofOxford, South Parks Road,Oxford OX1 3QU, UK.Correspondence to R.A.D.e-mail: [email protected]: 10.1038/nrd708

GLYCAN

A polymer consisting ofmonosaccharides linkedtogether by glycosidic bonds.

ENDOPLASMIC RETICULUM

Membrane-boundedcompartment in the cytoplasmof eukaryotic cells, in whichlipids, membrane-bound andsecreted proteins aresynthesized.

GOLGI APPARATUS

Membrane-bounded organellein eukaryotic cells, in whichlipids and proteins made in theendoplasmic reticulum aremodified and sorted.

© 2001 Macmillan Magazines Ltd66 | JANUARY 2002 | VOLUME 1 www.nature.com/reviews/drugdisc

R E V I E W S

Imino sugars in metabolic controlThe potent enzyme-inhibitory activity of imino sugarshas been applied to diseases in which the control ofoligosaccharide metabolism is linked to cellular dysfunction13. The successful treatment of non-insulin-dependent diabetes with the imino sugar N-hydroxy-ethyldeoxynojirimycin (Miglitol) depends on partialinhibition of intestinal disaccharidases to reduce thelevel of postprandial glucose14. Another use of iminosugars in drug development for metabolic control is in themodification of N-linked oligosaccharides on cell-surfaceproteins to reduce tumour-cell metastasis15.

Imino sugars with antiviral activityThe proper folding and controlled assembly of manynascent glycoproteins depends on interactions withchaperones, such as calnexin, in the ER (FIG. 3). Stepwiseremoval by ER α-glucosidases of terminal glucoseresidues from N-glycan chains attached to nascentglycoproteins enables the glycoproteins to interact withthe chaperones calnexin and calreticulin, which bindexclusively to monoglucosylated glycoproteins16,17 (FIGS 1

and 3). Interaction with calnexin is crucial for the correctfolding of some, but not all, glycoproteins, and inhibitorsof the ER α-glucosidases can be used specifically to tar-get proteins that depend on this interaction18.

Mammalian viruses are not known to encode theirown carbohydrate-modifying enzymes; they use the

such approaches that have been used successfully: thefirst interferes with the life cycle of several viruses, suchas hepatitis B virus (HBV), and the second compen-sates for inherited genetic defects in enzymes of theLYSOSOMAL degradation pathway for GSLs. The drugsthat have been evaluated for these diseases are a familyof sugar mimetics termed ‘imino’ sugars and are thefocus of this review.

Imino-sugar inhibitorsThe imino-sugar family of compounds, many of whichoccur naturally in certain plants and microorganisms,have had several important uses for glycobiologists10,and more recently their activities have been exploitedin the potential treatment of human diseases. Iminosugars are monosaccharide mimics, with a nitrogenatom in place of the ring oxygen (TABLE 1). They act asCHARGE-TRANSITION-STATE ANALOGUES and are submicromolar-range inhibitors of many hydrolytic enzymes, includingER α-glucosidases I and II, which are involved in pro-cessing glycoproteins11,12 (FIG. 1). The nitrogen atomprovides a further point for modification, and simplealkylation of imino-sugar glucose or galactose analoguesconfers unexpected activities for other enzyme targets,such as the pathway that is mediated by ceramide-specificglucosyltransferase (CerGlcT; UDP-glucose: N-acyl-sphingosine D-glucosyltransferase), which is crucial forGSL biosynthesis11 (FIG. 2).

GLYCOSYLTRANSFERASE

Glycosyltransferases produceglycosidic bonds by transfer-ring a glycosyl group — anygroup formed by detaching theglycosidic hydroxyl group fromthe cyclic form of a monosacc-haride, oligosaccharide orderivatives.

GLYCOSIDASE

An enzyme that hydrolysesglycosidic bonds.

LYSOSOME

Membrane-bounded organellein eukaryotic cells responsiblefor controlled intracellulardigestion of macromolecules.Lysosomes contain a widerange of hydrolytic enzymes,including glycosidases.

CHARGE-TRANSITION-STATE

ANALOGUES

A structural mimic of thetransition state between reac-tant(s) and product(s) for agiven reaction. Transition-stateanalogues make goodinhibitors because they arebound to the enzyme moretightly than the substrates.

Mannose 1-P DPP-GlcNAc2Man DPP-GlcNAc2Man9Glc3Glucose

GlcNAc

Man

Glc

ER α-glucosidase IER α-glucosidase II

Oligosaccharide transfer

ER α-glucosidase II

Golgi α-mannosidase IIN-acetylglucosaminyl- transferase II

Further glycosyltransferase- catalysed reactions

ER α-mannosidase

Golgi α-mannosidase I

N-acetylglucosaminyl- transferase I

Golgi α-mannosidase I

Golgi endomannosidase

Mature glycoproteins

Figure 1 | Biosynthesis of glycoproteins that possess N-linked glycans. As the nascent glycoprotein enters the endoplasmicreticulum (ER), a preformed oligosaccharide known as the dolicol-phosphate precursor (DPP) is attached co-translationally to someAsn residues that are part of the consensus sequence Asn-Xaa-Ser/Thr. The biosynthesis of this precursor, its attachment to Asnresidues and the subsequent steps of its processing in the ER and the Golgi, are performed by a series of glycosidases andglycosyltransferases. Asn, asparagine; Glc, glucose; GlcNAc, N-acetylglucosamine; Man, mannose; P, phosphate; Ser, serine; Thr, threonine; Xaa, any amino acid.

© 2001 Macmillan Magazines LtdNATURE REVIEWS | DRUG DISCOVERY VOLUME 1 | JANUARY 2002 | 67

R E V I E W S

the folding pathway mediated by calnexin/calreticulin —leads to their misfolding. Perhaps the most importantlong-term advantage of this approach is that, as theenzymes are host-cell- and not virus-encoded,emergenceof drug-resistant viruses is less likely to occur.

No general rules can be made about the effects thechange in N-glycan composition of envelope glycopro-teins or their (partial) misfolding can have on the virallife cycle. Misfolding, especially if restricted to onlysmall areas of the protein, might not necessarily lead todegradation. For many viruses, it results in impairmentof the maturation of envelope precursor proteins24–28,which, in turn, can lead to a decrease in viral assemblyand release26,29 and/or viral infectivity24. However, forother viruses, effects on viral assembly and infectivityare independent of envelope-precursor cleavage22,30.

Three examples that highlight the processing ofN-linked glycans as a target for antiviral interventionare the HIV model, the HBV model, and bovine viraldiarrhoea virus (BVDV), a model organism for thehuman hepatitis C virus (HCV).

HIV. HIV, the causative agent of AIDS, encodes two enve-lope glycoproteins (gp120 and gp41), which are derivedfrom a precursor protein (gp160) by endoproteolytic

host-cell glycosylation machinery to modify their ENVELOPE

proteins. As with some cellular host proteins, the foldingof certain viral glycoproteins has been shown to be calnexin-dependent19–23. So, targeting the ER α-gluco-sidases at a low level could disrupt the folding of theseproteins, and potentially be of therapeutic use in treatingviral infections, without affecting host-cell viability.

Indeed, inhibitors of the N-linked glycosylation path-way (FIG. 1) have been widely tested for antiviral activity(TABLE 2). In general, inhibitors of ER α-glucosidases Iand II might inhibit the replication of some viruses,whereas inhibitors of Golgi α-mannosidases I and IIrarely have any effect. However, it is difficult to generalizeon the effectiveness of ER α-glucosidase inhibitors. Forexample, the replication of certain viruses, such ashuman immunodeficiency virus (HIV), Moloneymurine leukaemia virus, mouse hepatitis virus, HBV,cytomegalovirus and Sindbis virus, is greatly inhibited invitro, whereas the replication of viruses such as RousSarcoma virus, influenza A virus and Semliki forest virusis not affected.

As suggested above, inhibition of viral replication byER α-glucosidase inhibitors is likely to be explained by theretention of glucosylated precursor oligosaccharides on theviral glycoproteins, which — if these proteins depend on

ENVELOPE

A lipoprotein-bilayer outermembrane of many viruses.Envelope proteins are oftenheavily glycosylated.

Table 1 | The imino-sugar family

Imino sugar Compound name Site of action Therapeutic use/ Clinical statuspotential

N-hydroxyethyl-DNJ Intestinal Non-insulin-dependent Approved since 1996 (Glyset, Miglitol) disaccharidases diabetes (Bayer Group)

N-butyl-DNJ CerGlcT GSL lysosomal Submitted for NDA (OGT918) disorders68 for Gaucher

disease (OGS)ER α-glucosidase I Hepatitis B39

N-butyl-DGJ CerGlcT GSL lysosomal Preclinical/(OGT923) disorders46 development (OGS)

Gangliosidoses

N-nonyl-DNJ CerGlcT GSL lysosomal disorders Preclinical/prototypeER α-glucosidase I Hepatitis B47

Hepatitis C surrogate30

N-nonyl-DGJ CerGlcT GSL lysosomal disorders Preclinical/prototypeNot determined Hepatitis B47

Hepatitis C surrogate30

N-7-oxanonyl-6- Not an enzyme Hepatitis C surrogate30 Preclinical Me-DGJ inhibitor development/IND

(Synergy Pharmaceuticals)

CerGIcT, ceramide-specific glucosyltransferase; DGJ, deoxygalactonojirimycin; DNJ, deoxynojirimycin; GSL, glucosphingolipid;IND, investigational new drug application; NDA, new drug application; OGS, Oxford GlycoSciences.

N

OHHO OH

(CH2)2OH

CH2OH

N

OHHO OH

(CH2)3CH3

CH2OH

N

OHHO OH

(CH2)3CH3

CH2OH

N

OHHO OH

(CH2)8CH3

CH2OH

N

OHHO OH

(CH2)8CH3

CH2OH

N

OHHO OH

(CH2)7OCH2CH3

CH3


R E V I E W S

shift of gp120 that results in exposure of gp41 does notoccur, and the process of viral fusion is therefore pre-vented. Consistent with this is the finding that, in thepresence of NB-DNJ, there is a regional misfolding in theV1/V2 LOOP of gp120 (REF. 35). This structural change ingp120 does not prevent the transport of the protein to theplasma membrane or viral budding, but it is sufficient toinhibit viral fusion, a crucial step in the HIV life cycle.

NB-DNJ was evaluated in Phase II clinical trials as ananti-HIV agent, but it was not possible to achieve suffi-ciently high serum concentrations of the drug36,37, andno major impact was observed on VIRAEMIA37. However,important information was gained in the trials. NB-DNJwas well tolerated, with the main side effect being gastrointestinal-tract distress due to intestinal disaccha-ridase inhibition, resulting in osmotic diarrhoea.

HBV. HBV infects over 350 million people worldwideand can cause liver disease and hepatocellularcarcinoma38. The HBV genome encodes three envelopeproteins: large (L), middle (M) and small (S), which arederived from a single open reading frame by use ofalternative translational start sites. In addition to formingthe main component of the viral envelope, these proteinsare also secreted in the form of DNA-free non-infectioussubviral particles, which can outnumber VIRIONS a millionto one. In contrast to the HIV envelope glycoproteins,HBV envelope M proteins contain only two glycosyla-tion sites. However, just as with HIV, HBV is sensitive toinhibitors of the ER α-glucosidases.

The role of the N-glycans of HBV has been probedusing inhibitors of the N-glycosylation pathway (tuni-camycin, NB-DNJ and DMJ)29,39 and site-directedmutagenesis. In an HBV-secreting cell line (HepG2.2.15),neither ER α-glucosidase processing nor interaction withcalnexin/calreticulin is required for the correct folding ofthe S and L proteins40. Even in the presence of NB-DNJ,subviral particles containing S and L proteins are secretedwith a full array of complex glycan structures that havebeen processed through the Golgi endomannosidasepathway (in the presence of the glucosidase blockade, theGolgi endomannosidase removes all three glucoseresidues in one step41; FIG. 1). However, the M proteincrucially depends on calnexin interaction for properfolding. If this interaction is prevented, the M proteindoes not fold correctly and formation of the viral envelopeis prevented.

There is no efficient infectivity assay available forHBV at present. However, the woodchuck chronicallyinfected at birth with woodchuck hepatitis virus(WHV), a close relative of HBV, is recognized as a goodanimal model to test potential antiviral treatments forthe human disease. When chronically infected wood-chucks were treated with N-nonyl-DNJ (NN-DNJ),a nine-carbon alkyl derivative of DNJ (TABLE 1) that hasbeen shown in cell-based assays to be 100–200 timesmore potent at inhibiting HBV secretion than NB-DNJ,virus levels were reduced in a dose-dependent manner42.Interestingly, at NN-DNJ concentrations sufficient toprevent WHV secretion, the glycosylation of mostserum glycoproteins seemed to be unaffected, indicating

cleavage in the cis-Golgi31. Although proteolyticallycleaved, gp120 remains non-covalently attached to thetransmembrane protein gp41, which serves as an anchorfor the complex. During infection, gp120, which is fullyexposed on the outer face of the viral envelope, binds toits cellular receptor (CD4), and undergoes a conforma-tional change that exposes gp41. This, in turn, allowsfusion with the cellular membrane and entry of the virusinto the cell.

With 30 potential N-glycan sites between them, gp41and gp120 are among the most heavily N-glycosylatedviral proteins known. They normally carry a mixture ofoligomannose and complex glycans. Treatment of HIV-1with N-butyldeoxynojirimycin (NB-DNJ), an ER α-glu-cosidase inhibitor (TABLE 1), suppresses viral infectivityand SYNCYTIUM formation in vitro32. By contrast, treatmentwith deoxymannojirimycin (DMJ), a Golgi mannosi-dase inhibitor (FIG. 1), has no effect on the secretion ofinfectious virus33, emphasizing the need of the virus forER α-glucosidase processing and interaction with cal-nexin and/or calreticulin.

The reduction in viral infectivity caused by NB-DNJ isthe result of impairments in post-CD4 binding steps32,34.Although binding to CD4 occurs, the conformational

Lactosylceramide

Lacto (neo)-series Globo-seriesGanglio-series

UDP-glucose: N-acylsphingosine D-glucosyltransferase (CerGlcT)

UDP

UDP-glucose

Glucosylceramide

HN

O

OH

OO

OH

HOHO OH

NB-DNJ NB-DGJNN-DNJ NN-DGJ

L-serine, palmitoyl-coenzyme A

Ceramide

HN

O

OH

HO

Figure 2 | Biosynthesis of glucosphingolipids. The first committed step in glucosphingolipidbiosynthesis is the transfer of glucose to ceramide by the ceramide-specific glucosyl-transferase (CerGlcT; UDP-glucose: N-acylsphingosine D-glucosyltransferase). This step isinhibited by the imino sugars N-butyldeoxynojirimycin (NB-DNJ) and N-nonyldeoxynojirimycin(NN-DNJ), and their galactose analogues N-butyldeoxygalactonojirimycin (NB-DGJ) and N-nonyldeoxygalactonojirimycin (NN-DGJ).

SYNCYTIUM

A mass of cytoplasm containingseveral separate nuclei enclosedin a continuous membraneresulting from the fusion ofindividual cells.

V1/V2 LOOP

The gp120 protein has elevendefined loop segments, five ofwhich are termed variable(designated V1–V5). The tip of one of the six non-variableloops forms a β-hairpin thathydrogen bonds with twoparallel strands from the V1/V2loop to form a β-sheet thateffectively connects the innerand outer domains.

VIRAEMIA

The presence of viruses in theblood.

VIRION

A mature infectious virusparticle.


R E V I E W S

glucosphingolipid biosynthesis (FIG. 2). In addition,the alkyl side chains might influence their antiviralbehaviour (TABLE 1).

For BVDV, these different activities have been investi-gated in more detail than for the other viral systems outlined above.When N-butyldeoxygalactonojirimycin(NB-DGJ; the galactose analogue of NB-DNJ), aninhibitor that targets only CerGlcT (not ER α-glucosidas-es I and II), was used in PLAQUE-REDUCTION ASSAYS, no effectwas observed on BVDV plaque formation, even at con-centrations high enough to completely inhibit theenzyme45. As NB-DNJ and NB-DGJ carry the same alkylside chain, this implies that the antiviral effectobserved using NB-DNJ can be attributed to the inhi-bition of the ER α-glucosidases involved in N-glycanprocessing, not the inhibition of GSL synthesis. Thisresult was confirmed by experiments showing that, in thepresence of NB-DNJ, the interaction of BVDV E1 and E2with calnexin is prevented, which leads to the misfoldingof the envelope glycoproteins and inefficient formationof the E1–E2 heterodimers. For NB-DNJ, the degree ofthis effect correlates with the dose-dependent antiviraleffect observed22.

However, for the long alkyl-chain compound NN-DNJ, the situation is not as straightforward. Thereis a lack of correlation between the ability of long alkyl-chain DNJ derivatives to inhibit ER α-glucosidases andtheir antiviral effect against BVDV, ruling out ER α-glu-cosidase inhibition as the sole antiviral mechanismresponsible. For example, the long alkyl-chain com-pound NN-DGJ, which is not an ER α-glucosidaseinhibitor, is just as effective against BVDV as NN-DNJwhen a low MULTIPLICITY OF INFECTION (MOI) is used in theplaque-reduction assays, with NN-DNJ showing superi-ority only at higher MOIs30.

These results hint at an entirely new mechanism bywhich imino-sugar derivatives carrying longer alkyl sidechains might exert their antiviral effect. Using short andlong alkyl-chain DNJ- and DGJ-derivatives, the possibil-ity that this mechanism could be acting at the level ofreplication, protein synthesis or protein processing wasruled out30. Long alkyl-chain derivatives induce anincrease in the accumulation of E2–E2 dimers in the ER,and these homodimers are subsequently also enriched insecreted virus particles; further investigations will showwhether this causes or reflects the antiviral mechanism.NN-DNJ caused a reduction in viral secretion (probablydue to misfolding of viral envelope glycoproteinscaused by ER α-glucosidase inhibition), as well as areduction in the infectivity of newly released virions.NN-DGJ exerted its antiviral effect solely through theproduction of particles with reduced infectivity30. Thefact that NN-DNJ combines both activities could explainits superiority at higher MOIs. However, these inhibitorsare being investigated as potential antiviral drugs againsthepatitis C, a chronic disease that is not usually charac-terized by high viral titres in the blood (although it isnot possible to estimate local virus concentrations; forexample, in the liver). In addition, patients might requirelong-term treatment, in which case a DGJ-based com-pound, which avoids some of the known side effects

that this class of therapeutics might be selective againstHBV. Alternatively, a mechanism other than, and inaddition to, ER α-glucosidase inhibition might beresponsible in part for the antiviral effect observed. Thisadditional mechanism, which seems to be associated withthe length of the alkyl side chain attached to the imino-sugar headgroup of the inhibitor, was described in detailfor another viral system, BVDV.

BVDV. Because of its similarity to HCV and the fact thatthere is no efficient cell-culture system available to supportHCV replication, BVDV43 has been used as a modelorganism in infectivity assays for the screening of potentialanti-HCV drugs. Most BVDV and HCV proteins arefunctionally homologous. The envelope glycoproteinsE1 and E2 interact either non-covalently (HCV)44 orthrough disulphide bonds (BVDV)22 to form a dimer,which has been proposed to be the functional complexpresent on the surface of mature virions. The envelopeglycoproteins of both HCV and BVDV interact withcalnexin during productive folding22,23. NB-DNJ andNN-DNJ have an antiviral effect against BVDV in vitro,with the long alkyl-chain derivative NN-DNJ being atleast tenfold more potent than NB-DNJ30,45.

NB-DNJ and NN-DNJ have at least three featuresthat might be involved in their antiviral activity: they areinhibitors not only of the ER α-glucosidases, but also ofCerGlcT, which catalyses the first committed step in

Golgi

Glc

Man

Glucosyl- transferase

Other chaperones

Protein folding on/off calnexin

Endoplasmic reticulum

α-glucosidase I & II

α-glucosidase II

NB-DNJNN-DNJ

NB-DNJNN-DNJ

Cx

G II

P

P

P

P

P

Figure 3 | The role of glycosylation in protein folding. The folding and assembly of manynewly synthesized glycoproteins depends on interactions with chaperones. Processing of theattached common oligosaccharide precursor GlcNAc2Man9Glc3 (which is N-linked; see FIG. 1)

in the endoplasmic reticulum (ER) by α-glucosidases I and II gives GlcNAc2Man9Glc, which canbind to chaperones such as calnexin (Cx). Chaperones act as quality-control factors byretaining glycoproteins in the ER until they are correctly folded. Inhibition of the α-glucosidasesby the imino sugars N-butyldeoxynojirimycin (NB-DNJ) and N-nonyldeoxynojirimycin (NN-DNJ)can interfere with this process, leading to misfolded proteins. G II, ER α-glucosidase II; Glc,glucose; GlcNAc, N-acetylglucosamine; Man, mannose; P, protein.

PLAQUE-REDUCTION ASSAY

Virus-induced cell death causesa plaque in a cell monolayer.Plaques can be counted toindicate how much virus ispresent.

MULTIPLICITY OF INFECTION

(MOI). The (average) numberof virus particles that infecteach cell in an experiment.


R E V I E W S

neurodegeneration and are fatal in early infancy. GSLstorage diseases occur at a collective frequency of 1 in18,000 live births and are the most common cause ofneurodegenerative disease in infants and children48.They result from the inheritance of mutations in genesthat encode acid glycosidases or their protein cofactors,which participate in the sequential removal of mono-saccharide units from GSLs in the lysosome. Most areautosomal recessive diseases, with the exception ofFabry disease, which is X-linked. The specific diseasesare: Gaucher types 1, 2 and 3, Fabry, Tay–Sachs,Sandhoff and GM2 gangliosidosis. With the exceptionof type 1 Gaucher disease, all are associated with GSLstorage in the nervous system9, reflecting the particularabundance of GSLs in neural tissue.

Individual mutations have different consequences onthe residual activity of the specific enzyme in question,which provides a guide as to the severity of clinical mani-festations. The infantile-onset disease variants have low orundetectable residual enzyme activity, the juvenile-onsetpatients have detectable but low enzymatic activities,whereas the adult-onset group have moderate residualenzyme activity. The more residual enzyme activity anindividual has, the longer it takes for storage of GSLs tobuild up to pathological levels. The molecular cell pathol-ogy of the GSL storage diseases is not clearly understoodand, until recently, no appropriate small animal modelsof these diseases were available for experimental study.

The options for treating the GSL storage disordersare limited at present and, for most affected patients, nospecific therapy is available. This difficulty is furthercompounded by the inaccessibility of the central nervoussystem (CNS). To date, most research has focused onmethods to augment the level of enzymatic activitywithin the lysosome by direct enzyme replacement49,bone-marrow transplantation (BMT) or gene therapy.Enzyme-replacement therapy is an established treatmentfor non-neuropathic Gaucher disease, but delivery ofthe enzyme is intravenous and therefore invasive, andlack of uptake of enzymes across the blood–brain barrierlimits the potential of this approach in neuropathic dis-eases. BMT has had mixed success in this group of dis-orders; it requires matched donors and is a procedureassociated with high mortality rates. Gene-therapystrategies, although promising, pose many unansweredquestions about efficacy and safety, and major technicaldifficulties mean that they are many years from clinicalimplementation.

So, is there any place for more conventional drug-based strategies for the management of these diseases?One attractive approach proposed by Radin and col-leagues was the concept of partially inhibiting GSL syn-thesis using a pharmacological agent50,51. Slowing therate of synthesis of GSLs will lead to fewer GSLs enteringthe lysosome for catabolism, reducing the rate of storage.In principle, complete balance would be achieved if suf-ficient residual enzyme activity were present. However,even if enzyme levels were low or undetectable, it isanticipated that severe disease could be converted into amilder, slower-progressing form. Many terms have beengiven to this approach, including ‘substrate deprivation’,

associated with DNJ-based compounds, might bepreferable46. Treatment of an MDBK cell line chronicallyinfected with non-cytopathic BVDV using long alkyl-chain imino-sugar derivatives that were chemically mod-ified to reduce in vitro toxicity, showed that both DNJ andDGJ derivatives can ‘cure’ the infection (D. Duranteland N. Z., unpublished observations). Promising resultsfrom in vivo preclinical toxicology and pharmacokineticstudies with one of these compounds could lead to itbeing evaluated in clinical studies for treatment againstHCV in the near future, subject to a successful investiga-tional new drug filing (IND).

The results obtained using ER α-glucosidaseinhibitors and non-inhibitors in the BVDV systemmight prompt investigators to revisit the HIV, HBV47

and WHV systems described above, using long alkyl-chain derivatives of DGJ as well as DNJ. The mechanismof action in each case is not a trivial task to evaluate, anda mixture of different mechanisms could well beresponsible for the antiviral effects achieved.

The fact that imino-sugar derivatives can disrupt thegeneral cellular function of glycoprotein processing gaverise to both the hope that they could be used as thera-peutics for various diseases and the fear that the significantmammalian toxicity of the compounds would be ahindrance to their usefulness. However, this is frequentlytrue of many drug candidates and it is not unreasonableto expect that an appropriate therapeutic window can befound. Improvement of their pharmacokinetic propertiesshould result in lower dose rates being necessary so thatundesirable side effects are limited.

Imino-sugar inhibitors and GSL storage diseasesThe GSL storage diseases are a family of progressivedisorders in which GSL species are stored in the lysosomeas a result of defects in the GSL degradation pathway(FIG. 4). In their most severe forms, they cause progressive

Table 2 | N-glycosylation inhibitors as potential antivirals

Virus family Examples of viruses tested References

Retroviruses Human immunodeficiency virus 71Moloney murine leukaemia virus 71,72

Hepadnaviruses Hepatitis B virus 28,38

Coronaviruses Murine hepatitis virus 23

Herpesviruses Herpes simplex virus type 1 73Herpes simplex virus type 2 74Cytomegalovirus 75,76

Alphaviruses Sindbis virus 24,77,78Semliki forest virus 79,80

Rhabdoviruses Vesicular stomatitis virus 81

Orthomyxoviruses Influenza A virus 77

Paramyxoviruses Measles virus 82

Flaviviruses Dengue virus 83Japanese encephalitis 84

Pestiviruses Bovine viral diarrhoea virus 22,45

Arenaviruses Lymphocytic choriomeningitis virus 85Junin virus 86

Baculoviruses Autographa californica multicapsid 87nuclear polyhedrosis virus


R E V I E W S

that a single drug could treat a family of GSL storagediseases,without the need for disease-specific intervention.

Imino-sugar inhibitors of GSL biosynthesisN-alkylated imino sugars with glucose and galactosestereochemistries inhibit GSL biosynthesis by inhibitingCerGlcT, which catalyses the transfer of glucose toceramide11 (FIG. 2). This is the first committed step ofthe GSL biosynthetic pathway, and glucosylceramideis the precursor for all glucosphingolipids, includingneutral GSLs and gangliosides. These analogues do notinhibit the ceramide galactosyltransferase that syn-thesizes galactosyl ceramide, which — along with itssulphated derivative, sulphatide — is an importantlipid in myelin36.

The prototypic compound, NB-DNJ (TABLE 1)52,53,was developed as an antiviral agent (see above). Iminosugars were not identified as inhibitors of CerGlcT until1994 (REF. 54), despite the fact they were in common useas inhibitors of enzymes that process N-glycans10,55.Inhibition of GSL biosynthesis was observed only incells treated with compounds that had a minimal alkyl-chain length of at least three carbons54,56. Molecularmodelling studies indicated that ceramide mimicrymight contribute, at least in part, to the mechanism ofinhibition of CerGlcT by NB-DNJ11,36; the nature of thering moiety is also important. So, in addition to the glu-cose analogue NB-DNJ, the galactose analogue NB-DGJ(TABLE 1) inhibits CerGlcT, but the mannose, fucose andN-acetylglucosamine derivatives do not56.

Whereas α-glucosidases I and II are resident in the ER,CerGlcT is located on an early Golgi compartment withits catalytic domain exposed to the cytosol. AlthoughNB-DNJ is a more potent inhibitor of ER α-glucosidases Iand II than it is of CerGlcT, both in vitro and in vivo, thedominant activity of the compound is against CerGlcTas, owing to its cytosolic orientation, this enzyme ismuch more accessible to the drug36.

Evaluation of SRT in mouse modelsSeveral knockout mouse models of GSL storage diseaseshave been generated57, allowing in vivo studies of SRT tobe performed. Before evaluating SRT in disease models,it was shown that healthy adult mice treated orally withNB-DNJ could tolerate partial GSL depletion58, eventhough mice null for GSLs do not develop8.

Tay–Sachs disease and Sandhoff disease are caused by the accumulation of the GM2 ganglioside due to adeficiency in the degradation enzyme, β-hexosaminidase(FIG. 4). There are two main isoenzymes of β-hexos-aminidase — hexosaminidase A, a heterodimer formedfrom an α-chain and a β-chain, and hexosaminidase B,a β–β homodimer. Tay–Sachs disease and Sandhoff dis-ease are caused by mutations in the genes coding for theα-chain (HEXA) and β-chain (HEXB), respectively. Inthe mouse model of Tay–Sachs disease (Hexa knockout),the mice store GM2 ganglioside in a progressive fashion,but the levels never exceed the threshold required to elicitneurodegeneration59,60.This is because in mice (but not inhumans), a lysosomal sialidase is sufficiently abundantor active to convert GM2 into GA2, which can then be

‘substrate inhibition’ and ‘substrate balance’. For thepurpose of this discussion, it will be termed ‘substratereduction therapy’ (SRT). There are three main goals forthis approach: first, to use an oral drug; second, the drugshould penetrate the CNS; and third, an early step in theGSL biosynthetic pathway should be targeted, such

GM2 ganglioside

GalNAcβ4NeuAcα3

Galβ4GlcCer

β-hexosaminidase Aβ-hexosaminidase A

Tay–Sachs disease Sandhoff disease

β-hexosaminidase B

Fabry disease

α-galactosidase

β-glucocerebrosidase

Gaucher disease

Sphingomyelinase

Niemann–Pick disease

GalNAcβ3Galα4Galβ4GlcCer

Galα4Galβ4GlcCer

Galβ4GlcCer

GlcCer

Sphingomyelin

Ceramide

NeuAcα3Galβ4GlcCer

b

a

Efflux

Nucleus

Synthesis

Membrane GSLs

Golgi

Lysosome

Impaired enzyme degradation

Diseased cell (lysosomal storage of GSLs)

Influx

Figure 4 | The glucosphingolipid cycle and glucosphingolipid storage diseases.a | Glucosphingolipids (GSLs) are synthesized from ceramide (FIG. 2) by the sequential additionof monosaccharides in the Golgi, and are then transported to the cell plasma membrane.As part of the normal turnover of components of the plasma membrane, GSLs are transportedto the lysosome for degradation. GSL storage diseases, such as Tay–Sachs disease, arecaused by defects in enzymes in the degradation pathway, which result in accumulation ofGSL species in the lysosome. b | Selected steps in the GSL degradation pathway in humans,showing the defective enzyme in several GSL storage diseases. Gal, galactose; GalNAc, N-acetylgalactosamine; GlcCer, glucosylceramide; NeuAc, N-acetylneuraminic acid.


R E V I E W S

Proof of concept in a genetic model of SRT. The efficacyof SRT was further shown in the Sandhoff mousemodel by Proia and colleagues, who crossed theSandhoff mouse with a mouse engineered to block syn-thesis of the storage GSL65. The resulting mice, whichhad defects in both GM2 ganglioside synthesis andcatabolism, no longer stored GSL, lived much longerand had greatly improved neurological function. It isinteresting to note that these mice, which were defi-cient in both the hexosaminidase A and B isoenzymes,eventually developed storage of oligosaccharidesderived from the catabolism of N-linked glycans.These are additional substrates for β-hexosaminidasesand are known to accumulate in Sandhoff-diseasepatients. This was the first clear indication that storedN-glycans might also contribute to neuropathology inthis disease. This study elegantly highlighted the limita-tion of SRT in diseases in which the defective enzymehas other non-GSL substrates65.

SRT in a mouse model of NPC disease. GSL accumula-tion is not restricted to the ‘classical’ GSL storage diseasesin which the inherited defect is in a lysosomal hydrolase.Other lysosomal disorders, including Niemann–Picktype C (NPC) disease, result in the accumulation ofGSLs in the brain, secondary to the primary defect. InNPC disease, the primary defect is in the NPC1 gene(and, less commonly, the NPC2 gene). NPC1 is localizedto vesicles that are thought to recycle unesterified choles-terol from late endosomes or lysosomes to the ER andGolgi. For this reason, NPC has historically been consid-ered to be a disorder of cholesterol transport. However,the fact that certain GSLs are stored in the brain in NPCdisease indicates that NPC1 might also be involved inGSL homeostasis. A crucial question in NPC disease isthe cause of the neurodegenerative phenotype. As GSLaccumulation in the classical GSL storage diseases isknown to lead to neuronal dysfunction and death(although precisely how remains to be determined),Walkley and colleagues proposed that the stored GSLsmight be major contributors to the neuropathologyassociated with this devastating disease66.

Two classes of GSL accumulate in the brain in NPCdisease: neutral GSL species and gangliosides. Thedominant species is GM2, which is known to causeprogressive neurodegeneration when it is stored inTay–Sachs and Sandhoff disease. Proia and colleaguestherefore investigated the contribution of the ganglio-sides to the pathology of NPC disease by crossing the NPCmouse (spontaneous model with a lesion in the Npc1gene) with a mouse genetically engineered to preventit synthesizing GM2 and some other higher GSLspecies67. Neuronal GM2 storage was prevented, but itdid not alter the disease course. So, if GSLs areinvolved in pathogenesis, it must be those that remainin the double-mutant mice, namely GM3, lactosylce-ramide, glucosylceramide and free sphingosine.Walkley and colleagues used NB-DNJ treatment in theNPC mouse model to determine whether or not anyglucosylceramide-based GSLs contributed to thepathology66, and found that life expectancy of the NPC

catabolised by the unaffected hexosaminidase-Bisoenzyme61. Disruption of Hexb to produce a mousemodel of Sandhoff disease knocks out both hexos-aminidase A and B, resulting in the storage of GM2 andGA2 gangliosides in the CNS and periphery61. TheSandhoff-disease mouse has very low levels of residualenzyme activity, conferred by the minor hexos-aminidase S (α–α) isoenzyme. The mice cannot bypassthe block in catabolism, however, and so undergorapid, progressive neurodegeneration and die at 4–5 months of age61. The Tay–Sachs mouse modeltherefore showed whether or not SRT could preventstorage in the brain and the Sandhoff mouse provideda model in which to study whether SRT would affectclinical pathology and extend survival by slowing therate of storage.

SRT in a mouse model of Tay–Sachs disease. Tay–Sachsmice were reared on food containing NB-DNJ andmonitored for 12 weeks. A reduction in stored GM2ganglioside was observed in all animals from the NB-DNJ-treated group (50% reduction in GM2 gan-glioside in the brains of treated mice relative to theuntreated controls)62. NB-DNJ was therefore able tocross the blood–brain barrier to an extent that pre-vented storage62.

SRT in a mouse model of Sandhoff disease. WhenSandhoff mice were treated with NB-DNJ, their lifeexpectancy was increased by 40% and GSL storagewas reduced in peripheral tissues and in the CNS63.Following the onset of symptoms, the rate of declinewas significantly slower in NB-DNJ-treated mice, andthe age at which deterioration could first be detectedwas delayed (approximately 100 days for untreated miceand approximately 135 days for NB-DNJ-treatedmice). However, the terminal stage of the disease(when the mice are moribund) was prolonged in NB-DNJ-treated mice. When GSL storage levels weremeasured in the untreated and NB-DNJ-treatedSandhoff mice at their end points (at 125 days and170 days, respectively), the levels of GM2 and GA2were comparable, indicating that death correlatedwith the same levels of GSL storage in the brains ofthe two groups of mice. Histological examination ofthe mice at 120 days showed reduced storage in thebrain of NB-DNJ-treated mice, consistent with theclinical improvement.

Combination therapy in the Sandhoff mouse.Sandhoff-disease mice treated with both BMT andNB-DNJ survived significantly longer than those treatedwith BMT or NB-DNJ alone. When the mice weresubdivided into two groups on the basis of the enzymelevels in their CNS, which were derived from donorbone marrow, the high enzyme group had a greaterdegree of synergy (25%) than the group as a whole(13%). Combination therapy might therefore be thestrategy of choice for treating the infantile-onset diseasevariants in which lack of enzyme limits the potential of SRT64.


R E V I E W S

four centres — Cambridge, Amsterdam, Prague andJerusalem — into a 1-year open-label clinical trial ofNB-DNJ. Data from this trial68 (BOX 1), and subsequentstudies in an extended-use protocol, strongly indicatethat GSL depletion improves all key clinical features ofGaucher disease. The data from the preclinical, clinicaland toxicology studies have recently been submitted tothe regulatory authorities in Europe and the UnitedStates for consideration for approval.

Future prospectsIn principle, it would be predicted that, in the same waythat BMT and substrate deprivation are synergistic intheir action in the Sandhoff mouse model64, combiningintravenous enzyme replacement and SRT in Gaucherpatients would be a rational treatment option69. Severalpermutations could be envisaged, including monother-apy, sequential therapy — that is, enzyme followed byNB-DNJ maintenance — or co-administration; that is,continuous NB-DNJ administration with periodicenzyme administration.

The preclinical studies in mouse models of Tay–Sachsand Sandhoff disease62,63 offer the prospect that SRTmight be of benefit to patients with CNS involvement, atleast those with the juvenile- and adult-onset variants ofthese disorders.With the advent of more effective meansof delivering enzymes to the CNS (BMT, gene therapyand neuronal stem-cell therapy), several strategies mightbecome available for improving the lives of patientssuffering from these devastating neurological diseases.Combining substrate-reducing drugs with enzyme-augmenting therapies could even make therapy in thevery severe infantile-onset patients a reality.

mouse was significantly extended and the neu-ropathology significantly delayed. They obtained similardata in the NPC cat model66. These observations indicatethat GSLs are involved in the neuropathology of NPC,although precisely which stored GSL species contribute isunknown. The other possibility is that NB-DNJ mediatesthe clinical improvement owing to an as yet unidentifiedactivity of this drug, independent of GSL depletion.However, the imino sugar NB-DGJ, which also inhibitsGSL biosynthesis, but does not cause any of the sideeffects attributable to NB-DNJ, has the same effect in theNPC mouse. Life expectancy was extended to the sameextent, making a mechanism based on GSL depletionhighly likely66. The study of NB-DNJ in the NPC mousehas led to the planning of clinical trials of NB-DNJ inNPC patients, due to start in 2002.

Clinical trial of SRT in type 1 Gaucher diseaseGaucher disease results from mutations in the geneencoding glucocerebrosidase, the enzyme thatremoves glucose from ceramide (FIG. 4). The symp-toms of Gaucher disease are primarily due to Gauchercells (MACROPHAGES engorged with glucosylceramide),which commonly collect in the spleen and liver,resulting in enlargement of these organs, as well asblood abnormalities, such as anaemia. Type I Gaucherdisease was chosen for the first clinical trial of SRTbecause it has no CNS involvement, has well-definedclinical endpoints and there is an existing effectivetherapy to make comparisons with (intravenousenzyme-replacement therapy).

Therefore, from 1998 to 1999, patients with type IGaucher (non-neuropathic) disease were recruited at

MACROPHAGE

A type of white blood cell that isspecialized for the uptake ofmaterial by phagocytosis.

Box 1 | Clinical trial of substrate reduction therapy in type I Gaucher disease

The trial was coordinated by Oxford GlycoSciences and N-butyldeoxynojirimycin (NB-DNJ) was referred to asOGT918. The trial enrolled 28 adult patients (14 females and 14 males); 7 of whom had had previous splenectomies.All patients were unable or unwilling to take enzyme-replacement therapy.

Side effects. The main known side effect of OGT918 is diarrhoea, which was noted in the previous trial in whichthis compound was used as an agent against human immunodeficiency virus (HIV)37. In the Gaucher study,a tenfold lower dose of drug was used relative to the HIV trial. Although most patients reported gastrointestinal(GI)-tract symptoms as soon as they started taking OGT918, the diarrhoea spontaneously resolved in mostpatients within several weeks and did not, in general, pose a significant problem68.

Of the 28 patients enrolled in the trial, 6 withdrew (2 owing to the GI-tract side effects, 2 owing to pre-existingmedical conditions and 2 for personal reasons). The remaining 22 patients were monitored at 6 and 12 months forsigns of clinical improvement. Two further patients withdrew because of symptoms of peripheral neuropathy.All other patients on OGT918 have been investigated by electromyography, and to date no other cases of peripheralneuropathy have been identified. Beyond the 12-month study, 18 patients have continued to receive OGT918 in anextended-treatment protocol, with some patients having so far taken therapy for 2.5 years.

Clinical efficacy. Both spleen and liver volumes showed a statistically significant reduction (15% and 7%,respectively) after 6 months of therapy. At 12 months, the decreases from baseline were 19% and 12%,respectively66. This was comparable to the response observed in patients of similar baseline disease severityreceiving enzyme-replacement therapy70. Chitotriosidase activity, a marker of disease activity, was reduced in atime-dependent manner, indicating a reduction in the total pool of Gaucher cells within the patients treated withOGT918 (REF. 68). Haemoglobin and platelet counts showed trends towards improvement, with a greaterimprovement in haemoglobin noted in patients who were anaemic at baseline. A statistically significantimprovement in platelet counts was achieved following 12 months of treatment68. Assessment of 18 patients in theextended-use protocol has shown continued reduction of organ volume, further improvements in platelet andhaemoglobin counts (all values now statistically significant) and continued decline in chitotriosidase activity (A. Zimran, presented at the Fourth European Working Group of Gaucher Disease (EWGED) Workshop, 2000).


R E V I E W S

1. Zapun, A. et al. Conformation-independent binding ofmonoglucosylated ribonuclease B to calnexin. Cell 88,29–38 (1997).

2. Rudd, P. M. et al. Roles for glycosylation of cell surfacereceptors involved in cellular immune recognition. J. Mol.Biol. 293, 351–366 (1999).

3. Weis, W. I., Taylor, M. E. & Drickamer, K. The C-type lectinsuperfamily in the immune system. Immunol. Rev. 163,19–34 (1998).

4. Marquardt, T. & Freeze, H. Congenital disorders ofglycosylation: glycosylation defects in man and biologicalmodels for their study. Biol. Chem. 382, 161–177 (2001).

5. Dwek, R. A. Glycobiology: ‘towards understanding thefunction of sugars’. Biochem. Soc. Trans. 23, 1–25(1995).A comprehensive review of concepts in glycobiology.

6. Dennis, J. W., Granovsky, M. & Warren, C. E. Glycoproteinglycosylation and cancer progression. Biochim. Biophys.Acta 1473, 21–34 (1999).

7. Parekh, R. B. et al. Association of rheumatoid arthritisand primary osteoarthritis with changes in theglycosylation pattern of total serum IgG. Nature 316,452–457 (1985).

8. Yamashita, T. et al. A vital role for glycosphingolipidsynthesis during development and differentiation. Proc.Natl Acad. Sci. USA 96, 9142–9147 (1999).

9. Walkley, S. U. Cellular pathology of lysosomal storagedisorders. Brain Pathol. 8, 175–193 (1998).

10. Winchester, B. & Fleet, G. W. Amino-sugar glycosidaseinhibitors: versatile tools for glycobiologists. Glycobiology2, 199–210 (1992).

11. Butters, T., Dwek, R. & Platt, F. Inhibition ofglycosphingolipid biosynthesis: application to lysosomalstorage disorders. Chem. Rev. 100, 4683–4696 (2000).

12. Butters, T. D. et al. Molecular requirements of imino sugarsfor the selective control of N-linked glycosylation andglycospingolipid biosynthesis. Tetrahedron Asymmetry11, 113–124 (2000).

13. Jacob, G. S. Glycosylation inhibitors in biology andmedicine. Curr. Opin. Struct. Biol. 5, 605–611 (1995).

14. Mitrakou, A. et al. Long-term effectiveness of a new α-glucosidase inhibitor (BAY m1099–miglitol) in insulin-treated type 2 diabetes mellitus. Diabet. Med. 15,657–660 (1998).

15. Goss, P. E., Reid, C. L., Bailey, D. & Dennis, J. W. Phase1B clinical trial of the oligosaccharide processing inhibitorswainsonine in patients with advanced malignancies. Clin. Cancer Res. 3, 1077–1086 (1997).

16. Bergeron, J. J., Brenner, M. B., Thomas, D. Y. &Williams, D. B. Calnexin: a membrane-bound chaperoneof the endoplasmic reticulum. Trends Biochem. Sci. 19,124–128 (1994).

17. Peterson, J. R., Ora, A., Van, P. N. & Helenius, A.Transient, lectin-like association of calreticulin with foldingintermediates of cellular and viral glycoproteins. Mol. Biol.Cell 6, 1173–1184 (1995).

18. Mehta, A., Zitzmann, N., Rudd, P. M., Block, T. M. &Dwek, R. A. α-glucosidase inhibitors as potential broadbased anti-viral agents. FEBS Lett. 430, 17–22 (1998).

19. McGinnes, L. W. & Morrison, T. G. Role of carbohydrateprocessing and calnexin binding in the folding and activityof the HN protein of Newcastle disease virus. Virus Res.53, 175–185 (1998).

20. Mirazimi, A., Nilsson, M. & Svensson, L. The molecularchaperone calnexin interacts with the NSP4 enterotoxin of rotavirus in vivo and in vitro. J. Virol. 72, 8705–8709(1998).

21. Werr, M. & Prange, R. Role for calnexin and N-linkedglycosylation in the assembly and secretion of hepatitis Bvirus middle envelope protein particles. J. Virol. 72,778–782 (1998).

22. Branza-Nichita, N., Durantel, D., Carrouee-Durantel, S.,Dwek, R. A. & Zitzmann, N. Antiviral effect of N-butyldeoxynojirimycin against bovine viral diarrhea viruscorrelates with misfolding of E2 envelope proteins andimpairment of their association into E1–E2 heterodimers.J. Virol. 75, 3527–3536 (2001).

23. Choukhi, A., Ung, S., Wychowski, C. & Dubuisson, J.Involvement of endoplasmic reticulum chaperones in thefolding of hepatitis C virus glycoproteins. J. Virol. 72,3851–3858 (1998).

24. Repp, R. et al. The effects of processing inhibitors of N-linked oligosaccharides on the intracellular migration ofglycoprotein E2 of mouse hepatitis virus and thematuration of coronavirus particles. J. Biol. Chem. 260,15873–15879 (1985).

25. Schlesinger, S., Koyama, A. H., Malfer, C., Gee, S. L. & Schlesinger, M. J. The effects of inhibitors of glucosidase Ion the formation of Sindbis virus. Virus Res. 2, 139–149(1985).

26. Pinter, A., Honnen, W. J. & Li, J. S. Studies with inhibitorsof oligosaccharide processing indicate a functional role forcomplex sugars in the transport and proteolysis of Friendmink cell focus-inducing murine leukemia virus envelopeproteins. Virology 136, 196–210 (1984).

27. Poss, M. L., Mullins, J. I. & Hoover, E. A. Posttranslationalmodifications distinguish the envelope glycoprotein of theimmunodeficiency disease-inducing feline leukemia virusretrovirus. J. Virol. 63, 189–195 (1989).

28. Ratner, L., Vander Heyden, N. & Dedera, D. Inhibition ofHIV and SIV infectivity by blockade of α-glucosidaseactivity. Virology 181, 180–192 (1991).

29. Lu, X., Mehta, A., Dwek, R., Butters, T. & Block, T.Evidence that N-linked glycosylation is necessary forhepatitis B virus secretion. Virology 213, 660–665 (1995).

30. Durantel, D. et al. Study of the mechanism of antiviralaction of iminosugar derivatives against bovine viraldiarrhea virus. J. Virol. 75, 8987–8998 (2001).Shows the antiviral properties of long alkyl-chainimino sugars, including those with a galactose-analogue headgroup, against a surrogate model ofHCV.

31. Freed, E. O. & Martin, M. A. in Fields Virology 4th edn Vol. 2(eds Knipe, D. M. et al.) 1971–2041 (Lippincott, Williamsand Willkins, Philadelphia, 2001).

32. Fischer, P. B. et al. The α-glucosidase inhibitor N-butyldeoxynojirimycin inhibits human immunodeficiencyvirus entry at the level of post-CD4 binding. J. Virol. 69, 5791–5797 (1995).

33. Elbein, A. D. Glycosidase inhibitors as antiviral and/orantitumor agents. Semin. Cell Biol. 2, 309–317 (1991).

34. Fischer, P. B., Karlsson, G. B., Dwek, R. A. & Platt, F. M. N-butyldeoxynojirimycin-mediated inhibition of humanimmunodeficiency virus entry correlates with impairedgp120 shedding and gp41 exposure. J. Virol. 70,7153–7160 (1996).

35. Fischer, P. B., Karlsson, G. B., Butters, T. D., Dwek, R. A.& Platt, F. M. N-butyldeoxynojirimycin-mediated inhibitionof human immunodeficiency virus entry correlates withchanges in antibody recognition of the V1/V2 region ofgp120. J. Virol. 70, 7143–7152 (1996).Elucidates the mechanism of action of NB-DNJagainst HIV.

36. Platt, F. M. & Butters, T. D. Substrate deprivation: A newtherapeutic approach for the glycosphingolipid lysosomalstorage diseases. Exp. Rev. Mol. Med. (2000).[http://www-ermm.cbcu.cam.ac.uk/00001484h.htm].

37. Fischl, M. A. et al. The safety and efficacy of combinationN-butyl-deoxynojirimycin (SC-48334) and zidovudine inpatients with HIV-1 infection and 200–500 CD4 cells/mm3.J. Acquir. Immune Defic. Syndr. 7, 139–147 (1994).

38. Hollinger, F. B. & Liang, T. J. in Fields Virology 4th edn Vol.2 (eds Knipe, D. M. et al.) 2971–3036 (Lippincott, Williamsand Willkins, Philadelphia, 2001).

39. Block, T. M. et al. Secretion of human hepatitis B virus isinhibited by the imino sugar N-butyldeoxynojirimycin.Proc. Natl Acad. Sci. USA 91, 2235–2239 (1994).

40. Mehta, A., Lu, X., Block, T. M., Blumberg, B. S. & Dwek,R. A. Hepatitis B virus (HBV) envelope glycoproteins varydrastically in their sensitivity to glycan processing:evidence that alteration of a single N-linked glycosylationsite can regulate HBV secretion. Proc. Natl Acad. Sci. USA94, 1822–1827 (1997).

41. Moore, S. E. & Spiro, R. G. Demonstration that Golgiendo-α-D-mannosidase provides a glucosidase-independent pathway for the formation of complex N-linked oligosaccharides of glycoproteins. J. Biol. Chem.265, 13104–13112 (1990).

42. Block, T. M. et al. Treatment of chronic hepadnavirus infectionin a woodchuck animal model with an inhibitor of proteinfolding and trafficking. Nature Med. 4, 610–614 (1998).First demonstration of antiviral action of N-nonyl-DNJin an animal model of HBV.

43. Lindenbach, B. D. & Rice, C. M. Flaviviridae: in FieldsVirology 4th edn Vol. 1 (eds Knipe, D. M. et al.) 991–1041(Lippincott, Williams and Willkins, Philadelphia, 2001).

44. Deleersnyder, V. et al. Formation of native hepatitis C virusglycoprotein complexes. J. Virol. 71, 697–704 (1997).

45. Zitzmann, N. et al. Imino sugars inhibit the formation andsecretion of bovine viral diarrhea virus, a pestivirus modelof hepatitis C virus: implications for the development ofbroad spectrum anti-hepatitis virus agents. Proc. NatlAcad. Sci. USA 96, 11878–11882 (1999).

46. Andersson, U., Butters, T. D., Dwek, R. A. & Platt, F. M. N-butyldeoxygalactonojirimycin: a more selective inhibitorof glycosphingolipid biosynthesis than N-butyldeoxynojirimycin, in vitro and in vivo. Biochem.Pharmacol. 59, 821–829 (2000).Evidence of greater tolerability of NB-DGJ relative toNB-DNJ.

47. Mehta, A. et al. Inhibition of hepatitis B virus DNAreplication by imino sugars without the inhibition of theDNA polymerase: therapeutic implications. Hepatology33, 1488–1495 (2001).

48. Meikle, P. J., Hopwood, J. J., Clague, A. E. & Carey, W. F.Prevalence of lysosomal storage disorders. J. Am. Med.Assoc. 281, 249–254 (1999).

49. Beutler, E. & Grabowski, G. in Gaucher Disease (edsScriver, C. R., Beadet, A. L., Valle, D. & Sly, W. S.)3635–3668 (McGraw Hill, New York, 2001).

50. Vunnam, R. R. & Radin, N. S. Analogs of ceramide thatinhibit glucocerebroside synthetase in mouse brain.Chem. Phys. Lipids 26, 265–278 (1980).

51. Radin, N. S. Treatment of Gaucher disease with anenzyme inhibitor. Glycoconj. J. 13, 153–157 (1996).

52. Platt, F. M. & Butters, T. D. Inhibitors of glycosphingolipidbiosynthesis. Trends Glycosci. Glycotechnol. 7, 495–511(1995).

53. Platt, F. M. & Butters, T. D. New therapeutic prospects forthe glycosphingolipid lysosomal storage diseases.Biochem. Pharmacol. 56, 421–430 (1998).

54. Platt, F. M., Neises, G. R., Dwek, R. A. & Butters, T. D. N-butyldeoxynojirimycin is a novel inhibitor of glycolipidbiosynthesis. J. Biol. Chem. 269, 8362–8365 (1994).Identification of NB-DNJ as an inhibitor of GSLbiosynthesis.

55. Platt, F. M., Karlsson, G. B. & Jacob, G. S. Modulation of cell-surface transferrin receptor by the imino sugar N-butyldeoxynojirimycin. Eur. J. Biochem. 208, 187–193(1992).

56. Platt, F. M., Neises, G. R., Karlsson, G. B., Dwek, R. A. & Butters, T. D. N-butyldeoxygalactonojirimycin inhibitsglycolipid biosynthesis but does not affect N-linkedoligosaccharide processing. J. Biol. Chem. 269,27108–27114 (1994).First description of NB-DGJ as an inhibitor of GSLbiosynthesis.

57. Suzuki, K., Proia, R. L. & Suzuki, K. Mouse models of humanlysosomal diseases. Brain Pathol. 8, 195–215 (1998).

58. Platt, F. M., Reinkensmeier, G., Dwek, R. A. & Butters, T. D.Extensive glycosphingolipid depletion in the liver andlymphoid organs of mice treated with N-butyldeoxynojirimycin. J. Biol. Chem. 272,19365–19372 (1997).

59. Yamanaka, S. et al. Targeted disruption of the Hexa generesults in mice with biochemical and pathologic features ofTay–Sachs disease. Proc. Natl Acad. Sci. USA 91,9975–9979 (1994).

60. Taniike, M. et al. Neuropathology of mice with targeteddisruption of Hexa gene, a model of Tay–Sachs disease.Acta NeuroPathol. (Berl.) 89, 296–304 (1995).

61. Sango, K. et al. Mouse models of Tay–Sachs and Sandhoffdiseases differ in neurologic phenotype and gangliosidemetabolism. Nature Genet. 11, 170–176 (1995).

62. Platt, F. M. et al. Prevention of lysosomal storage inTay–Sachs mice treated with N-butyldeoxynojirimycin.Science 276, 428–431 (1997).Demonstrates the in vivo efficacy of SRT in a mousemodel of a GSL storage disease.

63. Jeyakumar, M. et al. Delayed symptom onset andincreased life expectancy in Sandhoff disease mice treatedwith N-butyldeoxynojirimycin. Proc. Natl Acad. Sci. USA96, 6388–6393 (1999).

64. Jeyakumar, M. et al. Enhanced survival in Sandhoffdisease mice receiving a combination of substratedeprivation therapy and bone marrow transplantation.Blood 97, 327–329 (2001).

65. Liu, Y. et al. A genetic model of substrate deprivationtherapy for a glycosphingolipid storage disorder. J. Clin.Invest. 103, 497–505 (1999).

66. Zervas, M., Somers, K. L., Thrall, M. A. & Walkley, S. U.Critical role for glycosphingolipids in Niemann–Pickdisease type C. Curr. Biol. 11, 1283–1287 (2001).

67. Liu, Y. et al. Alleviation of neuronal ganglioside storagedoes not improve the clinical course of the Niemann–Pick Cdisease mouse. Hum. Mol. Genet. 9, 1087–1092 (2000).

68. Cox, T. et al. Novel oral treatment of Gaucher’s diseasewith N-butyldeoxynojirimycin (OGT 918) to decreasesubstrate biosynthesis. Lancet 355, 1481–1485 (2000).First clinical study reporting efficacy of SRT in type 1Gaucher disease.

69. Priestman, D. A., Platt, F. M., Dwek, R. A. & Butters, T. D.Imino sugar therapy for type 1 Gaucher disease.Glycobiology 10, 4–5 (2000).

70. Lachmann, R. H. & Platt, F. M. Substrate reductiontherapy for glycosphingolipid storage disorders. Exp. Opin. Invest. Drugs 10, 455–466 (2001).

71. Taylor, D. L. et al. 6-0-butanoylcastanospermine (MDL28,574) inhibits glycoprotein processing and the growth of HIV. AIDS 5, 693–698 (1991).


R E V I E W S

72. Sunkara, P. S., Bowlin, T. L., Liu, P. S. & Sjoerdsma, A.Antiretroviral activity of castanospermine anddeoxynojirimycin, specific inhibitors of glycoproteinprocessing. Biochem. Biophys. Res. Commun. 148,206–210 (1987).

73. Bridges, C. G. et al. The effect of oral treatment with 6-O-butanoyl castanospermine (MDL 28,574) in themurine zosteriform model of HSV-1 infection. Glycobiology5, 249–253 (1995).

74. Ahmed, S. P. et al. Antiviral activity and metabolism of thecastanospermine derivative MDL 28,574, in cells infectedwith herpes simplex virus type 2. Biochem. Biophys. Res.Commun. 208, 267–273 (1995).

75. Gretch, D. R., Gehrz, R. C. & Stinski, M. F.Characterization of a human cytomegalovirusglycoprotein complex (gcI). J. Gen. Virol. 69, 1205–1215(1988).

76. Taylor, D. L. et al. Loss of cytomegalovirus infectivity aftertreatment with castanospermine or related plant alkaloidscorrelates with aberrant glycoprotein synthesis. AntiviralRes. 10, 11–26 (1988).

77. Datema, R., Romero, P. A., Rott, R. & Schwarz, R. T. On the role of oligosaccharide trimming in the maturationof Sindbis and influenza virus. Arch. Virol. 81, 25–39(1984).

78. McDowell, W., Romero, P. A., Datema, R. & Schwarz, R. T.Glucose trimming and mannose trimming affect differentphases of the maturation of Sindbis virus in infected BHKcells. Virology 161, 37–44 (1987).

79. Kaluza, G., Repges, S. & McDowell, W. The significance ofcarbohydrate trimming for the antigenicity of the Semliki

Forest virus glycoprotein E2. Virology 176, 369–378(1990).

80. Repges-Illguth, S. & Kaluza, G. Differences in antigenicityof E2 in Semliki Forest virus particles and in infected cells.Arch. Virol. 135, 433–435 (1994).

81. Schlesinger, S., Malfer, C. & Schlesinger, M. J. Theformation of vesicular stomatitis virus (San Juan strain)becomes temperature-sensitive when glucose residuesare retained on the oligosaccharides of the glycoprotein. J. Biol. Chem. 259, 7597–7601 (1984).

82. Malvoisin, E. & Wild, F. The role of N-glycosylation in cellfusion induced by a vaccinia recombinant virus expressingboth measles virus glycoproteins. Virology 200, 11–20 (1994).

83. Courageot, M. P., Frenkiel, M. P., Dos Santos, C. D.,Deubel, V. & Despres, P. α-glucosidase inhibitors reducedengue virus production by affecting the initial steps ofvirion morphogenesis in the endoplasmic reticulum.J. Virol. 74, 564–572 (2000).

84. Lad, V. J., Shende, V. R., Gupta, A. K., Koshy, A. A. &Roy, A. Effect of tunicamycin on expression of epitopeson Japanese encephalitis virus glycoprotein E in porcinekidney cells. Acta Virol. 44, 359–364 (2000).

85. Wright, K. E., Spiro, R. C., Burns, J. W. & Buchmeier, M. J.Post-translational processing of the glycoproteins oflymphocytic choriomeningitis virus. Virology 177, 175–183(1990).

86. Silber, A. M., Candurra, N. A. & Damonte, E. B. The effectsof oligosaccharide trimming inhibitors on glycoproteinexpression and infectivity of Junin virus. FEMS Microbiol.Lett. 109, 39–43 (1993).

87. Jarvis, D. L. & Garcia, A. Jr. Biosynthesis and

processing of the Autographa californica nuclearpolyhedrosis virus gp64 protein. Virology 205, 300–313(1994).

AcknowledgementsF.M.P. is a Lister Institute Research Fellow. N.Z. is a DorothyHodgkin Royal Society Research Fellow.

Online links

DATABASESThe following terms in this article are linked online to:LocusLink: http://www.ncbi.nlm.nih.gov/LocusLink/calnexin | calreticulin | CD4 | ceramide galactosyltransferase |CerGlcT | chitotriosidase | ER α-glucosidase I | ER α-glucosidase II | glucocerebrosidase | HEXA | Hexa | HEXB |Hexb | β-hexosaminidase | α-mannosidase I | α-mannosidase II |NPC1 | Npc1 | NPC2OMIM: http://www.ncbi.nlm.nih.gov/Omim/Fabry disease | Gaucher type 1 | Gaucher type 2 | Gaucher type 3 |GM2 gangliosidosis | non-insulin-dependent diabetes | NPCdisease | Sandhoff disease | Tay–Sachs diseaseMedscape DrugInfo:http://promini.medscape.com/drugdb/search.aspMiglitol

FURTHER INFORMATIONEncyclopedia of Life Sciences: http://www.els.netGlycosylation and diseaseAccess to this interactive links box is free online.

Documents

Targeting Glycosylation as a Therapeutic Approach