Polypeptide GalNAc-Ts: from redundancy to specificity ScienceDirect

3_Q1

Matilde de las Rivas , Erandi Lira-Navarrete , Thomas A Gerken

and Ramon Hurtado-Guerrero

4 Mucin-typeO-glycosylationisapost-translationalmodification

5_Q2 (PTM)thatispredictedtooccurinmorethanthe80%ofthe

6 proteinsthatpassthroughtheGolgiapparatus.ThisPTMis

7 initiatedbyafamilyofpolypeptideGalNAc-transferases

8 (GalNAc-Ts)thatmodifySerandThrresiduesofproteins

9 throughtheadditionofaGalNAcmoiety.Theseenzymesare

10_Q3 typeIImembraneproteinsthatconsistofaGolgiluminal

11 catalyticdomainconnectedbyaflexiblelinkertoaricintype

12 lectindomain.Together,bothdomainsaccountforthedifferent

13 glycosylationpreferencesobservedamongisoenzymes.

14 Althoughitiswellacceptedthatmostofthefamilymembers

15 sharesomedegreeofredundancytowardtheirproteinand

16 glycoproteinsubstrates,ithasbeenrecentlyfoundthatseveral

17 GalNAc-Tsalsopossessactivitytowardspecifictargets.

18 Despitethehighsimilaritybetweenisoenzymes,structural

19 differenceshaverecentlybeenreportedthatarekeyto

20 understandingthemolecularbasisofboththeirredundancy

21 andspecificity.Thepresentreviewfocusesonthemolecular

22 aspectsoftheproteinsubstraterecognitionandthedifferent

23 glycosylationpreferencesoftheseenzymes,whichinturnwill

24 serveasaroadmaptotherationaldesignofspecific

25 modulatorsofmucin-typeO-glycosylation.

Addresses

26 ¹BIFI,UniversityofZaragoza,BIFI-IQFR(CSIC)JointUnit,Mariano

27 Esquillors/n,Campus,RioEbro,EdificioI+D,Zaragoza,Spain

28 ²CopenhagenCenterforGlycomics,DepartmentofCellularand

29 MolecularMedicine,SchoolofDentistry,UniversityofCopenhagen,

30 Copenhagen,Denmark

31 ³DepartmentsofBiochemistry,ChemistryandPediatricsCaseWestern

32 ReserveUniversity,Cleveland,OH,USA

33 ⁴Fundacio´nARAID,50018,Zaragoza,Spain

Correspondingauthors:Gerken,ThomasA.([email protected]),Hurtado- Guerrero,Ramon([email protected])

34 ⁵Theseauthorshavecontributedequally.

35 CurrentOpinioninStructuralBiology2019,56:xx–yy

36 ThisreviewcomesfromathemedissueonCarbohydrates

37 EditedbyAnnaPanchenkoandMonikaFuxreiter

38 https://doi.org/10.1016/j.sbi.2018.12.007

39 0959-440X/ã2018ElsevierLtd.Allrightsreserved.

Introduction 40

PolypeptideGalNAc-transferases(GalNAc-Ts)areafam- 41

ilyofGolgiresidentenzymes(20inhumans)thattransfera 42

GalNAcmoietyfromUDP-GalNAcontoSerorThrresi- 43

duesoftheirproteinsubstrates.Thisprocessresultsinthe 44

synthesisoftheTnantigen(GalNAca1-O-Ser/Thr),which 45

can be further elongated by the action of subsequent 46

glycosyltransferases(GTs)[1^,2].Historically,thismodifi- 47

cationisknownasmucin-typeO-glycosylation(henceforth 48

O-glycosylation)astheseglycansareabundant(>50%by weight) in mucins. Asthe GalNAc-Tsinitiate and thus 49

define sitesof O-glycosylationin densely O-glycosylated 50

proteins such as mucins, theseenzymes must possess a 51

rangeof propertiesinorderto properlyglycosylatetheir 52

targets[2,3].Itisnow knownthattheGalNAc-Tisoen- 53

zymeshavedifferentglycosylationpreferencesthatallow 54

themtobeclassifiedas:a)glycopeptide/peptide-preferring 55

isoforms (e.g. GalNAc-T1 and GalNAc-T2); b) (glyco) 56

peptide-preferringisoenzymes (e.g.GalNAc-T4); andc) 57

strict glycopeptide-preferring isoenzymes (e.g. GalNAc- 58

T7and GalNAc-T10)[4^].Thisdistinction isbasedon 59

theiractivityagainstsubstrateslackingorcontainingoneor 60

morepriorGalNAc-O-Ser/Thrmoietieswhichallowsthem 61

tobeclassifiedintoearly,intermediateorlateGTs,repre- 62

senting arangein activitiesagainstnakedpeptides/proteins 63

toalreadyhighlyglycosylatedpeptides/proteins(e.g.Gal- 64

NAc-T2,GalNAc-T4andGalNAc-T10areearly,interme- 65

diateandlateGTs,respectively)[1^,4^,5](SeeFigure1). 66

TheglycopeptideactivitiesoftheGalNAc-Tshavebeen 67

further classified into two classes, based ontheir short- 68

range(orneighboring)andlong-range(orremote)glyco- 69

sylating capabilities. The short-range glycosylation pre- 70

ferences account for the glycosylation of glycopeptide 71

substrateswhere thesugarmoietyisbound to thecata- 72

lytic domain (thus glycosylating 1–3 residues from the 73

sugar), while the long-range glycosylation preferences 74

compriseglycopeptidesugarbindingtothelectindomain 75

whichsubsequentlydirectsdistantacceptorsites(6to 76

17residuesaway)ontothecatalyticdomainforglycosyla- 77

tionasdepictedinFigure1[1^,6^].Ithasbeenfoundthat 78

boththelong-rangeandshort-rangeglycopeptideprefer- 79

encescanoperate inanN-terminalorC-terminaldirec- 80

tion depending on isoenzyme and furthermore some 81

isoenzymespossessboththelong-rangeandshort-range 82

glycopeptideactivities(Figure1).Together,theseprop- 83

erties explain how a highly coordinated repertoire of 84

GalNAc-Ts are capable of readily generating multiple 85

closelyspacedTnantigens,asoccursinmucins,aswellas 86

glycosylating proteins containing only a few acceptor 87

Figure1

T2 T3

T12 T13 T1

100

12

88%

56%

77%

58%

72%

47%

44%

67%

72%

50%

43%

48%

48%

62%

37%

83%

100

45

40 98

100

96

17

68

75

100

99

N 13

T13 T3 T6 T4 T12 T15 T2 T14 T16 T5 T7 T10 T17 T11 T20

T4

___

___

___

___

T5

T10

T11 T14 T16

T7 T6

_

_ ___

_

_

_

_ _

_

_

_

_

_

T8,T9,T18,T19

T15 _

_

T17 _

_

_

_

___ ___

_ ___ ___

_

___ ___

_ ___ ___

_ ___ ___

_

___

___

_

_

___ ___

_ ___ ___

_

_

______

_

______

_

______

_

______

_

______

_

_

_

______

_

______

_

______

_

______

___

_

______

____

____

+6 to +15 -6 to -15

____

____

____

____

____

____

___

___

_

_

___

___

_

_

_

_

_

_ ___

_

_

_

_ _

_ _

_

GalNAc-T Isoform

Peptide Motif

Long-Range Glycosylation Neighboring

Glycosylation

la

lc

lla

lg

lb

ld

llb

lf

C N C

C

N C

N

Current Opinion in Structural Biology

SummaryoftheknownpeptideandglycopeptidespecificitiesoftheGalNAc-Tfamily.

PhylogenetictreeoftheGalNAc-Tsshowingtheir1)randompeptidederivedpeptidesubstratemotifsasSequenceLogos[7^,23^,32^,46]2) neighboringpriorglycosylationpreferences(1–3residues)duetocatalyticdomaininteractions[4^],and3)theirlong-rangepriorglycosylation preferences(6to17residues)duetolectindomainbinding[23^,40^].Notethat‘-ND-’standsfornotdeterminedwhile‘---’indicatesnoor weakactivity.AlsonotethatGalNAc-T8,GalNAc-T9,GalNAc-T18andGalNAc-T19exhibitnearlyundetectableactivitiesagainstmostsubstrates [47]andhavenotbeenwellcharacterized.Themodelsatthebottomshowthedifferentsubstratebindingmodesthatleadtotheindicated specificities.Thecatalyticandlectindomainsareshownasoval-shapedfiguresinblueandred,respectively.(Glyco)peptidesareindicatedin purplewhiletheyellowsquaresdenotethepositionofpriorGalNAcmoietiesintheglycopeptides.ArrowsindicatethepositionofGalNActransfer totheacceptorsite.

90 Thefactthatmostisoenzymesofthisfamilyarecapable

91 ofglycosylatingcommonacceptorsubstrates,particularly

92 thosecontainingthe(Thr/Ser)ProXPromotif(where‘X’

93 usuallystandsfor asmallhydrophobicresidue;seeFig-

94 ure 1), suggests that they may also serve redundant

95 functions[4^,7^].Paradoxically,inrecentyearsanincre-

96 mentalnumberofreportshavedemonstratedthatseveral

97 GalNAc-T isoenzymes are highly specific for certain

98 protein substrates, as identified by the Simple Cell

99 (SC) approach developed by theClausen group [8^,9].

100 Usingthisstrategy,ApoC-IIIwasidentifiedasaspecific

101 substrate of GalNAc-T2 [9] and GalNAc-T11 was

102 reported to be specifically involved in glycosylation of

103 thepeptidelinkersbetweenclassArepeatsoftheLDL

104 receptor family [10,11]. One common theme found for

105 isoenzyme-specificglycosylationisthatsuchaglycosyla-

106 tion can interfere with the proprotein processing of

107 neighboring sites thus controlling such a processing

108 [12]. Two of the most studied examples are GalNAc-

109 T3 and its substrate, the fibroblast growth factor 23

110 (FGF23) [13],as well as GalNAc-T2 and angiopoietin-

111 likeProtein3(ANGPTL3)[14],wheremiss-regulationof O-glycosylationcanleadtofamilialtumoralcalcinosisand

112 dyslipidemia,respectively[13,14].Inaddition,theaber-

113 rantexpression ormutationof severalGalNAc-T isoen-

114 zymes and the overexpression of the Tn-antigen is

115 directly associated with many cancers [15,16^,17], the

116 mechanisms of whichare still notunderstood,although

117 Tn–Tnself-associationmayplayarole[16^].Finally,the

118 GalNAc-Ts are involved or implicated in many other

119 biologicalfunctionsincludingdevelopment,receptortraf-

120 fickingand modulationandproteinsecretion,which are

121 beyond thescopeof thisreview[18].

122 HowthisfamilyofGTscanglycosylatemultiplecommon

123 sites in some proteins and at the same time be highly

124 isoenzyme-specific for sites in other proteins remains

125 unanswered.Herein,wereviewtherecentadvancesthat

126 havebeguntounravel thesubstrate-recognitionmecha-

127 nismofseveralofthemostrepresentativeisoenzymes,as

128 wellaspresentingthestructuralandkineticbasisforboth

129 theiroverlappingandselectiveactivities.

130 Structural similarities between GalNAc-Ts

131 isoenzymes

132 To obtain acompleteunderstanding of themechanism

133 underlying the substrate specificity of this family of

134 enzymes, crystal structures of GalNAc-T isoenzymes

135 have beensolved, either intheapo formor in complex

136 with(glyco)peptidesubstratesandproducts.Todate,the

137 following structures havebeen reported (seeFigure 2a

138 andb):a)MusmusculusGalNAc-T1withMn²⁺(MmGal-

139 NAc-T1) (PDBentry 1XHB)[19^];b) humanGalNAc-

140 T2 (HsGalNAc-T2) complexed with UDP-Mn²⁺ (PDB

5AJN)[21^];c)HsGalNAc-T10 complexedwithUDP- 143

Mn²⁺ and Ser-GalNAc (PDB entries 2D7I and 2D7R) 144

[22^];d)HsGalNAc-T4complexedwithUDP,Mn²⁺and 145

a glycopeptide (PDB entry 5NQA) [6^]; and e) the 146

recentlyreportedstructuresofHsGalNAc-T4complexed 147

withUDP,Mn²⁺andadiglycopeptide(PDBentry6H0B) 148

[23^]andtwosplicevariantsoftheflyPGANT9A/Bwith 149

UDP-Mn²⁺andapeptidesubstrate(PDBentries6E4Q 150

and6E4R; Figure2b)[24^]. 151

ThesestructuresallshowanN-terminalcatalyticdomain 152

adopting the typical GT-A fold, characterized by two 153

abuttingRossmann-likefoldswhich islinkedbyashort 154

flexible linker to a C-terminal ricin-like lectin domain 155

[6^,19^,20^,21^,22^] (Figure 2a). The lectin domain, a 156

unique structural feature only present in this family of 157

eukaryotic GTs, has a b-trefoil fold built from three 158

repeatunits (a,b andg) thatarepotentially capableto 159

bindaGalNAcmoiety[25,26^].Itshouldbenoted that 160

theserepeatsarenotnecessarilyallactivebindersbased 161

ontheirsequencemotif[4^]andbyexperiment[7^,27]. 162

GalNAc-T catalyticdomain: UDP-GalNAc 163

binding site and flexible loop 164

ThefirstcrystalstructureofaGalNAc-TwasforMmGal- 165

NAc-T1[19^].Itprovidedtheinitialpictureoftheactive 166

site for which subsequent GalNAc-T structures highly 167

resemble, particularly in the architecture of the critical 168

and conserved Mn⁺² binding site (formed by Asp209, 169

His211(theDXHmotif)andHis344).Thisstructurealso 170

showedthatthecatalyticandlectindomainswereclosely 171

associated [19^] (Figure 2b). Subsequent crystal struc- 172

tures ofGalNAc-T2complexedtobothUDP-Mn²⁺and 173

toUDP-Mn²⁺andtheEA2peptide(AspSerThrThrProA- 174

laProThrThrLys) [20^], together with theGalNAc-T10 175

crystal structure complexedwith hydrolyzed UDP-Gal- 176

NAcandMn²⁺[22^],furtherdefinedtheGT-Afoldactive 177

siteresiduesthattetheredtheuridinediphosphateofthe 178

UDP-GalNAcdonorsubstrate[20^,22^].Inaddition,the 179

structure of GalNAc-T10 catalyticdomain revealed the 180

GalNAc moiety bound in the UDP-GalNAc-binding 181

pocket [22^,28]. Thesestructures,as well as thecrystal 182

structures ofthefirst pre-Michaelis andMichaeliscom- 183

plexes of HsGalNAc-T2 [29^], were of fundamental 184

importance fordefining modelsof thedynamicsofGal- 185

NAc-T2 during its catalyticcycle, which consistsof an 186

ordered bi–bi kinetic mechanism [29^,30,31]. In addi- 187

tion, the Michaelis complex revealed that these GTs 188

followafront-faceSNi-type reactionmechanism[29^]. 189

StructuresofGalNAc-Ts, bothwithandwithoutbound 190

peptidesubstrate,haverevealedadynamicflexibleloop 191

atthe surfaceof thecatalyticdomain substrate binding 192

siteasanimportantstructuralfeatureoftheGalNAc-Ts 193

[20^,21^,29^]. This flexible loop, formed by residues 194

Figure2

Flexible linker

Flexible loop

Flexible loop open conformation

Flexible loop closed conformation

(2FFV) (2FFU)

HsGalNAc-T2

MmGalNAc-T1 (1XHB)

HsGalNAc-T4 (5NQA)

HsGalNAc-T2 (5AJP/5AJN)

HsGalNAc-T10 (2D7R) DmPGANT9A

(6E4Q)

C-terminalN-terminal

(a)

(b)

(c)

Current Opinion in Structural Biology

CartoonandsurfacerepresentationofGalNAc-Tsstructures.

Catalyticandlectindomainsareshowninpurpleandsalmonrespectively,whileflexibleloopandlinkeraredisplayedinyellow.(a)Extended(left

197 theenzymeinanactiveform,oranopenconformationin

198 which the loop folds back to expose UDP to the bulk

199 solvent(inactiveform;Figure2c)[20^,29^].Thisinter-

200 conversionhasrecentlybeenshowntobedependentof

201 the presence ofUDP-GalNAc in HsGalNAc-T2, where

202 this sugar nucleotide stabilizes theclosed conformation

203 and consequently allows the binding of the substrate

204 peptide [32^]. The importance of this flexible loop in

205 catalysis is exemplified by the molecular basis of the

206 inactivation of the HsGalNAc-T2-Phe104Ser mutant,

207 whichislinkedtolowlevelsof highdensitylipoprotein

208 cholesterol [32^,33].It wasfound that Phe104 controls

209 theinactive-to-activetransitionoftheflexibleloopdueto

210 itshydrophobicinteractionwithAla151/Ile256/Val360,as

211 wellasaCH-pinteractionwiththeside-chainofArg362

212 located in the flexible loop [32^]. The hydrophilic

213 Phe104Ser mutationfailsto locktheflexible loopin its

214 activeform,thusimpedingpeptidesubstratebindingand

215 thefailuretoglycosylateitstargetedpeptidesubstrates(i.

216 e.ApoC-IIIandANGPTL3).Thisresultsinlowlevelsof

217 HDL[32^,33,34].

218 GalNAc-T catalyticdomain: peptide-binding

219 site

220 ItisnoteworthythatseveralGalNAc-Tcrystalssoakedor

221 cocrystallized with(glyco)peptides show indeterminate/

222 disordered structures for the substrate bound to the

223 catalyticdomain[6^,24^].Nevertheless,structuralinfor-

224 mation of the GalNAc-T-peptide acceptor recognition

225 couldbeinferredfromtheseriesofstructuresof(glyco)

226 peptides bound to the HsGalNAc-T2isoenzyme [21^]

227 and very recently the diglycopeptide bound to HsGal-

228 NAc-T4[23^].Intheselatterstructurestheinteractions

229 between the transferase and its acceptor substrates are

230 dissimilar,suggestingdifferencesbetweenisoenzymesat

231 the peptide binding groove level. However, this could

232 alsobeduetothedifferent(glyco)peptidesusedforboth

233 isoenzymes.TheHsGalNAc-T2structuresrevealedthat

234 theEA2peptideboundinashallowcleftonthesurfaceof HsGalNAc-T2, beingrecognized byhydrophobic inter-

235 actionsandtoalesserextenthydrogenbondinteractions

236 [20^](seeFigure3a).Itwasalsoobservedthatthemethyl

237 groupoftheacceptorThrresiduewasembeddedwithina

238 hydrophobicpocket,providingaplausibleexplanationof

239 why most GalNAc-T isoenzymes prefer to glycosylate

240 ThroverSeracceptorresidues[20^,21^,35](Figure3a).

241 SeveralothercrystalstructuresofHsGalNAc-T2incom-

242 plexwithUDP-Mn²⁺andglycopeptidesalsoshowedthat

243 theglycopeptidesactedasbridgesbetweenthecatalytic

glycopeptideswereboundtoanadaptablesugar-nucleo- 246

tide bindingsite, withtheflexible loop adoptingeither 247

open or closedconformations(Figure2c). Interestingly, 248

the binding of a mono-glycopeptide to GalNAc-T4 249

revealed peptide GalNAc bindingat the lectindomain 250

butnoobservablepeptideelectrondensityinitscatalytic 251

domain[6^]whilerecentlyahomologousdiglycopeptide 252

showed a well-resolved peptide bound to the catalytic 253

domain in a closed conformation due to GalNAc-T4’s 254

neighboring glycopeptide binding activity (discussed 255

below) [23^] (Figure 3b). Interestingly, in the Gal- 256

NAc-T4 structure, theportion of the peptidespanning 257

the catalytic and lectin domains was found disordered 258

[21^,23^]. 259

Atthelevelofthepeptide-bindinggroove,itwasfurther 260

observed that three highly conservedaromatic residues 261

(namely Phe361, Phe280 and Trp282 in GalNAc-T2), 262

interactwiththe(Thr/Ser)-Pro-X-Prosubstratesequence 263

[20^].Thusfarthe(Thr/Ser)ProXProsequenceistheonly 264

substrate consensus motif remotely conserved among 265

mostGalNAc-Ts(Figures1,3aandb)[1^,21^].Indeed, 266

allisoenzymesthatexperimentallydisplaythis(Thr/Ser) 267

ProXPropreferencepossessthehomologusPheandTrp 268

residues [4^,7^,23^] including GalNAc-T4 and Gal- 269

NAc-T12. GalNAc-T7 and GalNAc-T10, which lack 270

these conserved residues and do not exhibit the (Thr/ 271

Ser)ProXPro preference, instead display strong neigh- 272

bouringglycosylationpreferencesatthe+1positionrela- 273

tive to the acceptor (i.e. (Thr/Ser)(Thr*/Ser*), where 274

*=-O-GalNAc) [4^,28]. These latter two isoenzymes 275

are, therefore, expected to contain a GalNAc binding 276

site in place of the ProXPro binding site found in the 277

otherisoenzymes.Presently,thestructuralandmolecular 278

bases for the neighboring glycosylation preferences of _Q4279

GalNAc-T7andGalNAc-T10remainto bedetermined. 280

Thus, thenear lackof a conservedsubstrate consensus 281

motif togetherwith theiractive site flexibilitypointsto 282

theversatilityoftheseenzymes,allowingthemtosculpt 283

their binding sites to accommodate a wide range of 284

acceptor substrates. 285

GalNAc-T catalyticdomain: glycopeptide- 286

binding site 287

Untilveryrecentlytherewerenostructuresdescribinghow 288

the so-called neighbouring glycosylation activity of the 289

GalNAc-Ts could beaccommodated.The recent report 290

of a diglycopeptide (GlyAlaThr*3GlyAlaGlyAlaGlyAla- 291

GlyThr*11Thr12ProGlyProGly, where Thr*=Thr-O- 292

(Figure2LegendContinued)panel)andcompact(rightpanel)formsofmonomericHsGalNAc-T2.(b)Cartoonandsurfacerepresentationof MmGalNAc-T1,HsGalNAc-T4,DmPGANT9AandHsGalNAc-T10.(c)SurfacerepresentationoftheHsGalNAc-T2-UDP-MUC5AC-13complex.The overallstructureisshowninsalmon,monoglycopeptideMUC5AC-13andtheflexiblelooparedepictedincyanandyellow,respectively.The flexibleloopoftheenzymeisshowninitsclosedandopenconformations.

Figure3

(a)

(b)

(c)

A7 A9 G8

Y536 P6

P4

F361 F361

T12

T11 P8

Current Opinion in Structural Biology

InteractionsbetweenGalNAc-Tsandtheirsubstrates.

(a)Close-upviewofthecatalyticdomainofHsGalNAc-T2withtwodifferentpeptides,EA2(cyansticks;leftpanel)andglycopeptideMUC5AC-13 (GlyThrThrProSerProValProThrThrSerThrThr*SerAlaPro)(yellowsticks;rightpanel),whicharesimilarlyrecognizedbyHsGalNAc-T2througha hydrophobicpatch.UDPisdepictedasstickswithmagentacarbonatomsandMn²⁺isshownasapurplesphere.(b)Ontheleftpanel,surface representationoftheHsGalNAc-T4-UDP-Diglycopeptide6(GlyAlaThr*3GlyAlaGlyAlaGlyAlaGlyThr*11Thr12ProGlyProGly)complex.Peptide backboneisdepictedincyanwiththetwoGalNAcgroupsasblueandredsticks;theenzymeflexibleloopisshowninitsclosedconformationin yellow.Ontherightpanel,close-upviewofthemaininteractionsbetweenHsGalNAc-T4catalyticdomainglycopeptidebinding-siteandthe GalNAcgrouponT*11ofdiglycopeptide6.TheGalNAc-T4residuesformingthepeptide-bindingsitearedepictedinsalmonandyellowandthe glycopeptideisdepictedincyan,withtheGalNAcgroupsshownasblueandredsticks.Mn²⁺andwatermoleculesaredepictedasgreenandred spheres,respectively,andhydrogenbondsappearsasdottedyellowlines.Pleasenotethatweonlyshowwater-mediatedinteractionsinwhich

294 isoenzyme[23^](Figure3b).InthisstructuretheGalNAc

295 ofThr*11isshowntetheredbyhydrogenbondandhydro-

296 phobicinteractionstothesidechainsofThr283and Gln285

297 andthebackboneofLys366atthesurfaceofthecatalytic

298 domain. Importantly, these residues are not conserved

299 among otherGalNAc-Tisoenzymesandthereisnodis-

300 cernible cleft or pocket for the binding of theGalNAc

301 (Figure3b).SuchGalNAcbindingpresentstheadjacent

302 Thr12intothecorrectorientationtoacceptGalNAcfrom

303 the UDP-GalNAc donor. Kinetic studies on a series of

304 glycopeptidesubstratesfurtherconfirmedthattheneigh-

305 bouring GalNAc binding at the catalytic domain was

306 weakerthantheremoteGalNAcbindingofThr*3tothe

307 lectin domain and further revealed substrate inhibition

308 kinetics onthediglycopeptide,presumably duetocom-

309 petitivebindingofthetwoThr*’softhesubstrateatthe

310 lectindomain[23^].Thisworkisofadditionalsignificance

311 as the individual GalNAc-T4 remote and neighboring

312 glycopeptideactivities,andbothtogether,couldbeelimi-

313 natedorgreatlyreducedbyselectivemutagenesis.

314 GalNAc-Ts lectindomain

315 The GalNAc-T-glycopeptide recognition at the lectin

316 domain ismore easilycompared among isoenzymes,as

317 therearecrystalstructuresofHsGalNAc-T2,HsGalNAc-

318 T4 and HsGalNAc-T10complexed withSer-O-GalNAc

319 aswellaslongerThr-O-GalNAccontainingglycopeptides

320 [6^,21^,22^,23^] (Figure 3c). The GalNAc-T1 lectin

321 domain containstwoknownfunctionalGalNAc-binding

322 sites out of the possible three (i.e. the a and b subdo-

323 mains) [36],whereas the GalNAc-T2,GalNAc-T4, and

324 GalNAc-T10 lectin domains contain only one known

325 active site(i.e.a-,a-andb-respectively)[4^].Thefirst

326 structureof aglycopeptideboundtothelectindomain,

327 thatis,Ser-O-GalNAcboundtoHsGalNAc-T10,revealed

328 thesugarmoietyboundtotheb-siteinteractingthrough

329 several hydrogen bonds (including residues Asp525,

330 Asn544, Tyr536) and one CH-p interaction (His539)

331 (Figure 3c). Subsequentstructures of GalNAc-T2com-

332 plexedwithlongerglycopeptidesshowednodiscernible

333 interactions with the peptide backbone of the lectin

334 domainlevel,whiletheGalNAcmoietyinteractedexclu-

335 sivelywithresiduesinthea-subdomainbindingsiteby

336 similarinteractionsasdescribedforGalNAc-T10(i.e.via

337 residuesAsp458/Asn479/Tyr471andHis474).Theseresi-

338 duesare conservedinnearly allisoenzymes(Figure 3c)

339 [21^,22^].Similarly,bindinginteractionsof thepeptide

340 GalNAc residue to thelectin a-domain of GalNAc-T4

341 wererecentlyreported[6^,23^](Figure3c);however,a

342 largedifferenceintheorientationofthelectindomainsof

GalNAc-T4 and GalNAc-T2 relative their catalytic 343

domains was observed. As discussed in the sections 344

below, these differences readily explain the origins of 345

theirdifferentlongrangeN-priororC-priorglycosylation 346

preferences(seeFigures1 and4). 347

Earlier work had suggested that the lectin domain of 348

GalNAc-Tscouldlikelyinfluencesubstratespecificityby 349

steric hindrance that would depend on the size of the 350

aminoacidsidechainsoftheglycopeptidesubstrate[37], 351

whileithasalsobeensuggestedthatthelectindomainsof 352

someGalNAc-Tscouldformhetero-dimersand/orhomo- 353

dimers that could also alter their specificity [38]. The 354

recent crystal structures of the fly PGANT9-A and 355

PGANT9-B lectin domain splice variants now offers 356

intriguing evidence for something likethe former[24^ 357

].Inthiscase,alooponthelectindomainthatprotrudes 358

towardthecatalyticdomainpeptidebindingsitediffersin 359

charge between the splice variants. These charge 360 CATALYTIC

DOMAIN

LECTIN DOMAIN

N ^T*

T*

C

Like T4

Like T2

Current Opinion in Structural Biology

SuperpositionofHsGalNAc-T2andHsGalNAc-T4.

SuperimposedcartoonrepresentationsofHsGalNAc-T2-UDP- MUC5AC-13glycopeptide

(GlyThrThrProSerProValProThrThrSerThrThr*SerAlaPro)complex depictedinredandHsGalNAc-T4-UDP-diglycopeptide6(GlyAlaThr

*3GlyAlaGlyAlaGlyAlaGlyThr*11Thr12ProGlyProGly)complexdepicted inblue.TheMUC5AC-13,diglycopeptide6andGalNAcmoietiesare showninred,blue,andorangeatoms,respectively.Thearrows indicatethedirectionofthelong-rangeglycosylationpreferenceof eachenzyme,basedontheorientationoftheirrespectivelectin domainswithrespecttheircatalyticdomains.NotethatthecriticalAsp residuesofthelectindomainGalNAc-bindingsitesareindicatedfor clarificationpurposes.

(Figure3LegendContinued)onlythewatermoleculeactasabridgebetweentheresidues.(c)MaininteractionsbetweenGalNAc-T2,GalNAc- T4andGalNAc-T10isoenzymeslectindomain(shownassalmon,purpleandslate,respectively)andtheGalNAcmoiety(shownasstickswith orangecarbonatoms).

361 differencescorrelate with their activities toward highly

362 chargedsubstrates,thus suggestingthatatleastelectro-

363 staticinteractions,ifnotdirectpeptidesubstratebinding,

364 ofthelectindomaincansignificantlyinfluencetransfer-

365 aseactivity[24^].Biologically,thesesplicevariancesare

366 usedto properlyglycosylate differentsecretory mucins,

367 whoseincompleteglycosylation isshownto alter secre-

368 torygranulemorphology[24^].Concurrently,structural

369 andmoleculardynamicsstudiesonGalNAc-T4boundto

370 adiglycopeptidehaverevealedaflexiblelooponitslectin

371 domainthatcanapproachtheGalNAcresidueofcatalytic

372 domain bound glycopeptide [23^]. Mutagenesisof this

373 loopwasshowntoalterthekineticpropertiesofGalNAc-

374 T4againstbothpeptideandglycopeptidesubstratesthus

375 again confirming that additional features of the lectin

376 domain beyond glycan binding will likely play roles in

377 substrateselection ofthesetransferases.

378 Theflexible linker and itsrole in theremote

379 glycosylation preferencesofthe GalNAc-Ts

380 ThecatalyticandlectindomainsofalloftheGalNAc-Ts

381 (exceptforT20thatlacksthelectindomain)areconnected

382 byalinker sequencewhose length andsequence varies

383 amongisoenzymes[6^,19^,20^,21^,22^].Comparinglin-

384 kers,theN-terminalregionsaremoreconservedwhilethe

385 C-terminalregionsarelessconserved[6^].Previousstud-

386 ieshaveattributedtherelativepositioningofthecatalytic

387 andlectin domainstothenatureof thelinkersequence

388 [21^,39],thusthemorestretched-outlinkerofGalNAc-

389 T10 [22^] results in fewer interactions between both

390 domains comparedto the more closely spaceddomains

391 inGalNAc-T1[19^](Figure2b).Thissuggestedthatlinker

392 flexibilitycouldfunctiontocontroltherelativeorientation

393 oflectinandcatalyticdomains,therefore,modulatingthe

394 selectionofnewGalNAc-modificationsitesinpreviously

395 glycosylatedsubstrates[4^,39,40^].

396 One of the largestquestions in the field hasbeen how

397 theseenzymesdifferentiallyrecognizeremotepriorgly-

398 cosylationsitesinanN-terminalorC-terminaldirection.

399 A recent work on HsGalNAc-T2 and HsGalNAc-T4

400 showsthattheirflexiblelinkersdisplaybothinterdomain

401 rotationandinterdomaintranslational-likemotionwhich

402 couldberesponsibleoftheirdifferentlongrangeglyco-

403 peptidepreferences[6^].ThecrystalstructureofHsGAl-

404 NAc-T4 with glycopeptide boundto thelectin domain

405 [6^]revealedthatitsGalNAc-bindingsitewaslocatedon

406 theoppositesideofthelectindomainwhencomparedto

407 thehomologoussiteinHsGalNAc-T2(Figure4).These

408 differentpositionsofthelectindomain(Figure4),readily

409 account how GalNAc-T4 promotes the opposite long-

410 rangeglycosylationpreferencecomparedtoGalNAc-T2

411 and other isoenzymes [1^,4^] (seeFigure 1).That this

412 rotationiscausedbythenatureoftheflexiblelinkerwas

413 supported by molecular dynamics simulations, site-

414 directed mutagenesis, and kinetics experiments [6^].

415 Indeed, the glycopeptide kinetics of GalNAc-T2

chimerascontainingaGalNAc-T3orGalNAc-T4flexible 416

linker and a series of flexible linker mutants, demon- 417

stratedthatitslong-rangeglycosylationpreferencecould 418

bemodulatedandevenreversedsimplybymodifyingits 419

linker[6^].Thissuggeststhattheflexiblelinkerplaysa 420

majorrolein dictatingeachisoenzyme’s long-rangegly- 421

cosylation preference by altering the lectin domain’s 422

orientation relative to its catalytic domain [6^]. All 423

together,thesefindings showedforthefirsttime howa 424

structuralfeaturethatisneitherin theactive sitenorin 425

the lectin domain GalNAc-binding site is capable of 426

modifyingtheactivityandtheglycosylationpreferences 427

oftheseisoenzymes. 428

Final remarks 429

ThattheGalNAc-Tsareassociatedwithnumeroushuman 430

diseasesincludingcancer[14,15,16^,41]clearlyjustifiesthe 431

importance of unravelling the molecular basis that lie 432

beneaththeirsubstraterecognition,rangingfromredun- 433

dantoverlappingsites[42]tohighlyspecifictargets.Here, 434

wehavebrieflysummarizedthemostimportantadvances 435

atstructurallevelofthisfamilyofenzymesthatbeginto 436

revealthemolecularoriginsoftheiruniquepeptide and 437

glycopeptidespecificities.However,additionalstructures 438

oftheseisoenzymesincomplexwithboththeirredundant 439

andspecific(glyco)peptidesubstrateswillbenecessaryfor 440

athoroughmechanisticunderstandingoftheirpromiscuity, 441

specificity,anddistinctglycosylationpreferences.Inpar- 442

ticular,muchmoreneedstobeunderstoodregardingtheir 443

short-rangeglycosylationpreferencesaswecurrentlyhave 444

onlyoneexampledescribingsuchGalNAc-T-(glyco)pep- 445

tide recognition. Hence, it is of utmost importance to 446

continuestudyingthiscomplexfamilyofenzymestofully 447

understandhowtheyselectivelyrecognizetheirtargetsin 448

multiple signaling pathways. Such studies will in turn 449

facilitatethedevelopmentof GalNAc-Tmodulatorsand 450

inhibitors that would certainly be useful forthetreatment of 451

manydiseases[11,13,16^,43,44].Finally, onecannotdis- 452

cardthepotentialforNatureorganizingtheGalNAc-T’sin 453

acellaccordingtotheirisoenzymeclass(e.g.early,inter- 454

mediateandlateGTs)utilizingtheirdifferentglycosyla- 455

tionpreferencestoproducethevastrepertoireofglycosyl- 456

ationsitesobservedinvitro.Such organizationisclearly 457

presentastheretrogradeintroductionofGalNAc-Tsinto 458

theER(thesocalledGALApathway)hasbeenshownto 459

manifestlyalterthepatternsof O-glycosylationandmay 460

play arole in cancer [26^,41].However, this pathwayis 461

currently under an intense debate in the Glycobiology 462

communityhenceitsimportancehasyettobefullyunder- 463

stood[45]. 464

Acknowledgements 465

466

TheauthorswouldliketoacknowledgethesupportoftheNational _Q5

467

InstitutesofHealth(grantR01GM113534toT.A.Gerken),ARAID,MEC

468

(grantsCTQ2013-44367-C2-2-PandBFU2016-75633-PtoR.Hurtado-

469

Guerrero),andtheDGA(E34_R17).Theresearchleadingtotheseresults

470

hasalsoreceivedfundingfromtheFP7(2007–2013)underBioStruct-X

471

(Grantagreement283570andBIOSTRUCTX_5186).