Czachorowski M., Lam-Yuk-Tseung S., Cellier M, Gros P., “Transmembrane topology of the mammalian Slc11a2 iron transporter”, Biochemistry, 48(35): 8422-34, 2009.
TransmembraneTopologyoftheMammalianSlc11a2IronTransporter†
MaciejCzachorowski,‡StevenLam-Yuk-Tseung,‡MathieuCellier,§andPhilippeGros*,‡
‡DepartmentofBiochemistry,McGillUniversity,Montreal,Quebec,CanadaH3G-0B1,and§InstitutNationaldelaRecherche
Scientifique,INRS-InstitutArmand-Frappier,Laval,Quebec,Canada
ReceivedApril8,2009;RevisedManuscriptReceivedJuly17,2009
ABSTRACT:ThemammalianSlc11a1andSlc11a2proteinsdefinealargefamilyofsecondarymetal
transporters.Slc11a1andSlc11a2functionaspH-dependentdivalentcationtransportersthatplayacritical
roleinhostdefensesagainstinfectionsandinFe
2þhomeostasis,respectively.Thepositionandpolarityof
individualtransmembranedomains(TMD)ofSlc11a2werestudiedbyanepitopetaggingmethodbasedon
theinsertionofsmallantigenichemagglutininA(HA)peptides(YPYDVPDYAS)inpredictedintra-or
extracellularloopsoftheprotein.ThetaggedproteinswereexpressedintransfectedLLC-PK1kidneycells
andtestedfortransportactivity,andthepolarityofinsertedtagswithrespecttotheplasmamembranewas
determinedbyimmunofluorescenceinintactandpermeabilizedcells.HAepitopetagswereinsertedat
positions1,98,131,175,201,243,284,344,403,432,468,504,and561.Insertionsatpositions98,131,175,
403,and432abrogatedmetaltransportbySlc11a2,whileinsertionsatpositions1,201,243,284,344,468,504,
and561resultedinfunctionalproteins.TopologymappinginfunctionalHA-taggedSlc11a2proteins
indicatedthattheN-terminus(1),aswellasloopsdelineatedbyTMD4-5(201),TMD6-7(284),and
TMD10-11(468),andC-terminus(561)areintracellular,whileloopsseparatingTMD5-6(243),TMD7-8
(344),andTMD11-12(504)areextracellular.Theseresultsarecompatiblewithatopologyof12
transmembranedomains,withintracellularaminoandcarboxytermini.Structuralmodelsconstructedby
homologythreadingsupportthis12TMDtopologyandshow2-foldstructuralsymmetryinthearrangement
ofmembranehelicesforTM1-5andTM6-10(conservedSlc11hydrophobiccore).
TheSlc11familyofdivalentcationtransportershastwo
membersinmammals,Slc11a1(alsoknownasNramp1)1and
Slc11a2(Nramp2,DMT1),whichareintegralmembranephos-
phoglycoproteinsof90-100kDasharing64%aminoacid
sequenceidentityand78%similarity(1,2).Slc11a1isexpressed
inthelysosomalcompartmentofmacrophagesandintertiary
granulesofneutrophilsandisrapidlyrecruitedtothemembrane
ofmicrobe-containingphagosomesformedinthesecells(3,4).At
thatsite,Slc11a1functionsasapH-dependenteffluxpumpfor
Fe
2þandMn2þ,restrictingtheavailabilityoftheseessential
metalstotheenclosedmicroorganismsandcontributingtothe
antimicrobialdefensesofmacrophages(5-7).Inmice,naturally
occurring(G169D)orexperimentallyinducedmutationsin
Slc11a1causesusceptibilitytoanumberofinfectionsincluding
Mycobacterium,Salmonella,andLeishmania(2).Likewise,poly-
morphicvariantswithinorinproximityofthehumanSLC11A1
genehavebeenfoundtobeassociatedwithincreasedsuscept-
ibilitytotuberculosisandleprosyinseveralpopulationsfrom
areasofendemicdisease(8,9).TheSlc11a2proteinisexpressed
atthebrushborderoftheduodenumwhereitmediatesacquisi-
tionofnon-hemedietaryiron.Itisalsoexpressedinthesyntaxin-
13positiverecyclingendosomecompartmentofmanycelltypes,
whereitisresponsibleforthetransportoftransferrinironfromthelumenofacidifiedendosomesintothecytoplasm(10-12).
Differentialsplicingof3
0-exonsoftheSlc11a2geneyieldstwo
mRNAswithdifferent30-endsequencesdistinguishedbythe
presence(isoformI)orabsence(isoformII)ofanironresponse
element(IRE).ThetwoisoformsencodedbythesemRNAsalso
havedistinctC-terminalaminoacidsequences(1).Although
bothisoformsIandIIarepresentattheplasmamembrane,they
areexpressedingenerallydifferentcelltypesandhavedistinct
organellardistributionsandintracellulartraffickingproperties.
IsoformIIispresentinrecyclingendosomeswhileisoformIis
presentinlateendosomes(13).Slc11a2proteinsplayacentral
roleinironhomeostasis,andaloss-of-functionmutation
(G185R)causesveryseveremicrocyticanemiainthemkmouse
andintheBelgraderat(14,15).Inaddition,anumberofloss-of-
functionmissense(R416C,G212V,delV114)(16-18)andspli-
cingmutations(18,19)havebeendetectedinthehuman
SLC11A2geneinpatientssufferingfromhypochromicmicro-
cyticanemiawithserumandliverironoverload.
Intransfectedmammaliancells,transportstudiesusingiso-
topic
55Fe2þand54Mn2þorthemetal-sensitivefluorescentdyes
(e.g.,calceinandFURA-2)haveestablishedthatSlc11a1and
Slc11a2cantransportbothmetalsinapH-dependentfash-
ion(20).ExperimentsinXenopusoocyteshaveshownthat
Slc11a2isahigh-affinitytransporterforanumberofdivalent
cations(inadditiontoFe
2þandMn2þ).Transportiselectrogenic
andiscausedbyprotonmovementthroughthetransporter
(substrate-dependentandsubstrate-independentH
þleak)(21,
22).SequenceanalysesshowthatSlc11formsafamilyofmetal
transportershighlyconservedfrombacteriatohumans.Func-
tionalstudiesontheyeast(Smf1-3),insect(Malvolio),andplant
†ThisworkwassupportedbygrantstoP.G.fromtheNIH(NIAID,
RO1AI035237).P.G.isaJamesMcGillProfessorofBiochemistry.
*Towhomcorrespondenceshouldbeaddressed.Phone:514-398-
7291.Fax:514-398-2603.E-mail:philippe.gros@mcgill.ca.
1Abbreviations:TMD,transmembranedomains;HA,hemagglutinin
A;OPD,o-phenylenediaminedihydrochloride;DMT1,divalentmetal
transporter-1;Nramp,naturalresistance-associatedmacrophage
protein;CTM,conservedtransportmotif.pubs.acs.org/BiochemistryPublishedonWeb07/21/2009r
2009AmericanChemicalSociety8422Biochemistry2009,48,8422–8434
DOI:10.1021/bi900606y
(AtNramp1-6)relatives,includingyeastcomplementation
experimentsinSmf1-3knockoutstrains,indicatethatSlc11
proteinssharesimilarfunction,substrate,andmechanismof
transport(1,2).Strikingly,eventhesedistantlyrelatedSlc11
proteinsshowanumberofhighlyconservedstructuralfeatures
which,throughfunctionalcharacterizationexperiments,have
beenshowntoplaykeyrolesinthemechanismoftransport.
Thesestructuralfeaturesinclude(a)anumberofpredicted
membranespanningdomains(TMD),someofwhichhave
amphipathicproperties(23),(b)severalhighlyconservedcharged
aminoacidresidueswithinthesepredictedTMDs,withmuta-
tionsatseveralnegativelychargedpositions(D86,E154,D192,
E299)abrogatingtransport(24),(c)apairofinvarianthistidines
(H267,H272)inpredictedTMD6thatplayakeyroleinpH
regulationoftransportand/orH
þmovementthroughthe
transporter(23,24),(d)amutation-sensitivesegmentattheC-
terminalendofpredictedTMD1(25-27),(e)ahighlyconserved
sequencemotif(residues384-403),inwhichalterationsabrogate
transport(28),and(f)aclusterofpredictedAsn-linkedglyco-
sylationsitesinanotherwisepoorlyconservedsegmentdeli-
neatedbypredictedTMD7and8.Finally,afunctionalfish
Slc11γtransporterlackingpredictedTMD11and12hasbeen
described,suggestingthatthefirsttenTMDsconstitutethemain
functionalunitofthisfamilyoftransporters(29).
High-resolutionstructuralinformationisultimatelyrequired
notonlytounderstandthemechanismofmetaltransportby
Slc11proteinsbutalsotodeterminehowconserved,mutation-
sensitiveregionsofthetransporterscoordinatesimultaneous
metalandH
þtransport.Determiningthesecondarystructureof
Slc11proteinsisanecessaryfirststeptowardthisgoal.A
combinationofmultiplesequencealignments,hydropathypro-
filing,hydrophobicmoments,andotherpredictivemethodshas
beenusedtoproposethatSlc11proteinsfromanimalscontain12
TMD,whileyeastandmicrobialproteinswouldcontain11(23).
Immunofluorescencestudieshaveshownthattheaminoand
carboxylterminiofmouseSlc11a2areintracellular(30,31)while
theglycosylatedloopbetweenpredictedTMDs7and8is
extracytoplasmicinSlc11a1andSlc11a2(20).Thesestudies
suggestamodelinwhichtheN-andC-terminiareintracellular
withanevennumberof12TMdomains.Althoughparallel
topologystudiesoftheEscherichiacolitransporter(MntH)
supportthismodel(32),onestudyusingananti-Slc11a1antibody
toinhibitFe
2þtransportsuggeststhattheN-andC-terminiof
Slc11a1areextracytoplasmic(33).
Inthepresentstudy,weinvestigatedthetopologyofthe
mammalianSlc11a2protein,includingthenumber,position,
andorientationofindividualtransmembranedomains.Forthis,
weusedahemagglutininA(HA)epitope-taggingapproachin
whichthemembranetopologyoftheinsertedexofacialepitopes
isestablishedintransport-competentproteinsexpressedatthe
plasmamembraneofLLC-PK1kidneycells.Thetransmem-
branetopologyobtainedexperimentallywasusedtotestthree-
dimensionalmodelsofthe10TMDstructuralfoldthatis
conservedamongseveralfamiliesofcation-driventransporters.
EXPERIMENTALPROCEDURES
Materials.
Reagentgradechemicalswerepurchasedfrom
SigmaChemicals(St.Louis,MO).Geneticin(G418)wasob-
tainedfromInvitrogen.Monoclonalmouseantibody(HA.11)
directedagainsttheinfluenzahemagglutinepitope(HA)was
purchasedfromCovance(Berkeley,CA).Calceinacetoxymethylester(calcein-AM)waspurchasedfromMolecularProbes
(Eugene,OR).Cy3-conjugatedgoatanti-mousesecondaryanti-
bodiesandperoxidase-coupleddonkeyanti-mouseandgoatanti-
rabbitantiserawerepurchasedfromJacksonImmunoResearch
Laboratories(WestGrove,PA).AllrestrictionenzymesandVent
DNApolymerasewereobtainedfromNewEnglandBiolabs
(Ipswich,MA).
PlasmidsandSite-DirectedMutagenesis.Theconstruc-
tionofthefull-lengthcDNAformurineSlc11a2isoformII
(-IRE)clonedintotheHindIIIsiteofpCB6(pN2-2myc)was
describedelsewhere(24).TheconstructionofSlc11a2isoformI
(þIRE)full-lengthcDNAinpCB6andcontainingahemagglu-
tinin(HA)taginthefourthpredictedextracellularloop(pN2-
HAIRE)hasbeendescribed(13).TheSlc11a2isoformIIcDNA
wasexcisedfrompN2-2mycbyHindIIIdigestionandclonedinto
pBluescriptKSþ(Stratagene,LaJolla,CA)modifiedbythe
eliminationoftheSacIandXbaIsitesfromthepolylinkerto
generatepN2myc-B/S-KO.Toimproveproteinexpressionin
transfectedcells,thefull-lengthSlc11a2cDNAwasmodifiedat
its5
0-endbytheadditionofaconsensusGCCACCKozak
sequenceandadditionalEcoRI,MluI,andHindIIIsitestoenable
subsequentcloningintothemammalianexpressionvector
pCB6(31).ThiswasdonebyPCRmutagenesis,usingolig-
onucleotideprimersEMHK-NTF(5
0-CAGAATTCACGCG-
TAAGCTTGCCACCATGGTGTTGGATCCTAAAGAAAA-
GATG-3
0)andN2501R(50-GCCAATGACTTCCTGCAT-
GTC-30)andpN2-HAIREasatemplate.The30-endofSlc11a2
inpN2myc-B/S-KOwasmodifiedtoreplaceisoformII-specific
3
0-endsequenceswithisoformIsequences,usingPCRmutagen-
esiswitholigonucleotidesN21051F(50-ACTCTGGCTGTG-
GACATCTAC-30)andN2-I-CT-HXR(50-CAGTCTCGA-
GAAGCTTTTACTTAATGTTGCCACCGCTGG-30)with
pN2-HAIREasthetemplate.Thesemodified50(N-terminal)
and30(C-terminal)fragmentsofSlc11a2werethensubclonedas
EcoRI/BstEIIandSacI/XhoIfragments,respectively,into
pN2myc-B/S-KOtogeneratethefinalconstructpN2IRE-B/
S-KO.
HAepitopes(YPYDVPDYAS)wereinsertedinthemouse
Slc11a2(þIRE)protein(13)byrecombinantPCR(34)andusing
mutagenicprimerslistedinTable1.Tofacilitatethepotential
insertionofadditionalHAepitopes,aNheIrestrictionsitewas
insertedatthe3
0-endofeachHAsequence,accountingforthe
additionalSresiduefoundattheC-terminusofeachHAtag.
pN2-HAIREwasusedasatemplateforallmutagenesisreac-
tions,exceptforconstruct6forwhichpN2-2mycwasusedasa
template.ThemodifiedSlc11a2productswerethendigestedand
subclonedaseitherEcoRI/BstEII(constructsNTand1),BstEII/
XbaI(constructs2,3,4,and5),XbaI/SacI(construct6),orSacI/
XhoI(constructs8,9,10,11,andCT)fragmentsintopN2IRE-B/
S-KO.Allconstructsweresequencedtoverifytheabsenceof
undesiredmutations,priortosubcloningofeachHA-taggedfull-
lengthSlc11a2fragmentintotheHindIIIsiteofpCB6for
transfection.Inthecaseofconstructs5b,8b,and11b,asecond
HAepitopewasinsertedbycloningashortdouble-stranded
fragmentcomposedoftwocomplementaryoligonucleotidesinto
theengineeredNheIsitesofthepreviouslyinsertedHAepitope
sequences.
CellCulture,Transfection,andWesternBlotting.LLC-
PK1cells(35)weregrowninDulbecco’smodifiedEagle’s
medium(Invitrogen)supplementedwith10%fetalbovineserum
and100units/mLpenicillin/streptomycin(37C,5%CO
2).Cells
weretransfectedwithdifferentSlc11a-HA/pCB6constructsArticleBiochemistry,Vol.48,No.35,20098423
usinglipofectamine2000(Invitrogen)asperthemanufacturer’s
instructions.Selectionforstablytransfectedcloneswascarried
outinmediumcontainingG418(geneticin,1.4mg/mL)for14
days.Individualcolonieswerethenisolatedandexpanded.LLC-
PK1cellclonesstablytransfectedwiththevariousSlc11a-HA/
pCB6constructswerelysedinbuffer(50mMTris-HCl,pH8.0,
137mMNaCl,1%NP40,10%glycerol)supplementedwith
proteaseinhibitors(1mMphenylmethanesulfonylfluoride
[PMSF],1μMpepstatin,0.3μMaprotinin,1μMleupeptin).
ThesewholecellextractswereseparatedbySDS-PAGE,and
clonesexhibitinghighlevelsofSlc11a2-HAproteinexpression
wereidentifiedbyimmunoblotting,aspreviouslydescribed(31),
andusingaffinity-purifiedrabbitpolyclonalanti-mouse
Nramp2/DMT1(1:1000)oramonoclonalmouseanti-HAanti-
body(1:2000).Secondaryanti-rabbitandanti-mouseantibodies
conjugatedtohoseradishperoxidasewereusedat1:20000.
DivalentMetalTransportAssay.Metaltransportby
individualtaggedSlc11a2-HAproteinsexpressedintransfected
LLC-PK1cellcloneswascarriedoutusingafluorescence
quenchingassay,exactlyaswehavepreviouslydescribed(31).
Calcein-AMwaspreparedasa500μMstocksolutionindimethyl
sulfoxide(DMSO),whileferrousammoniumsulfate(Fe
2þ)and
cobaltchloride(Co2þ)werepreparedasa2mMstockin
deionizedwater.MetaltransportinindividualLLC-PK1trans-
fectedcloneswasassessed(initialquenchingrates)fromfluores-
cencequenchingcurves(31).
Immunofluorescence.Avariationofacellsurfacelabeling
protocolwasusedtodeterminethemembranetopologyofHA
tagspresentinindividualrecombinantSlc11a2proteins(13).
Stabletransfectantsweregrowntoconfluencyonglasscover-
slips,withemptyvectortransfectedcellsservingasanegativecontrol.TolocalizetheHAepitopesexpressedonthecellsurface
(nonpermeabilizedconditions),cellswereincubatedwithanti-
HAantibody(1:200dilution)in2%nonfatmilk/DMEMfor2h
at37Cpriortofixationwith4%paraformaldehydein
phosphate-bufferedsaline(PBS)supplementedwith1mM
MgCl
2and0.1mMCaCl2(PBSþþ),followedbyincubation
withgoatanti-mouserhodamine-conjugatedsecondaryantibody
(1:1000dilution)in5%nonfatmilk/PBSfor1hatroom
temperature.TolocalizetheHAepitopesexpressedintracellu-
larly(permeabilizedconditions),cellswerefixedwith4%paraf-
ormaldehyde,followedbypermeabilizationwith0.1%Triton
X-100for30minatroomtemperature.Thecellswerethen
incubatedinprimaryantibodyin5%nonfatmilk/PBSfor1hat
roomtemperaturebeforeexposuretosecondaryantibodyas
describedabove.Allcells,intactandpermeabilized,werein-
cubatedinDAPI(4
0,6-diamidino-2-phenylindole)for5minat
roomtemperature(tostainnuclei)priortobeingmountedon
glassmicroscopeslides.CellswerevisualizedwithanAxiovert
200Mepifluorescencemicroscopeusinga63oilimmersion
objective.DigitalimageswereacquiredwithaZeissAxioCam
HRmcameraoperatedwithAxioVision4.3.Imageswere
cropped,assembled,andlabeledusingAdobePhotoshop7.0
andIllustrator10.0software.
QuantificationofSlc11a2-HASurfaceExpression.To
quantifysurfaceexpressionofextracellularHAtags(detectedby
immunofluorescenceundernonpermeabilizedconditions),we
usedacolorimetricassay(36).Forthis,2.010
4cells/well(24-
wellplasticplates)wereallowedtoreachconfluency,followedby
incubationwithanti-HAantibodyexactlyasdescribedfor
immunofluorescence(omittingtheDAPIstainingstep).The
secondaryantibodywasahorseradishperoxidase(HRP)
Table1:OligonucleotidesUsedforEpitopeInsertionbySite-DirectedMutagenesis
M1
bV2
NTCAGAATTCACGCGTAAGCTTGCCACCATG(HA)cGTGTTGGATCCTAAAGAAAAGATG(27)
A98V99
1(271)
aGAATCTGATTTGCAGTCTGGAGCA(HA)GTGGCTGGATTTAAGCTGCTCTG(317)
L131H132
2(393)TTGGAGTGGTCACCGGCTTG(HA)CATCTTGCTGAAGTATGTCACCG(416)
L175S176
3(524)CAGCCATCGCCATCAATCTGCTG(HA)TCTGCAGGAAGGGTCCCCCTGTG(548)
K201Y202
4(603)GTGTTTCTTTTTTTGGACAAA(HA)TATGGCTTGCGGAAGCTGGAAGC(626)
P243S244
5(729)CTCAGGGGCATGTTCGTGCCG(HA)TCCTGTCCAGGGTGCCGCACC(750)
5b(HA)x2
A248N249
6(852)CAAGTCTAGACAGGTGAATCGGGCC(HA)AATAAGCAGGAAGTGCGGGAAGC(875)
L344F345
7(HA)
N403L404
8(1209)GTCATGGAGGGATTCCTGAAC(HA)CTAAAATGGTCGCGCTTTGCC(1230)
8b(HA)x2
V432E433
9(1296)CGTCGCTGTCTTCCAGGATGTG(HA)GAGCACCTAACGGGGATGAATG(1318)
S468E469
10(1404)CAAGCCTGCGGCCAGTGATGAGT(HA)GAGTTTTCCAATGGAATAGGCTGG(1421)
H504V505
11(1512)GTTTATGTCCAGGAGCTAGGGCAT(HA)GTGGCACTCTATGTGGTGGCTGC(1532)
11b(HA)x2
K561
CT(1660)GATACCAGCGGTGGCAACATTAAG(HA)TAAAAGCTTCTCGAGACTG
(HA)
cYPYDVPDYAS
aNucleotidepositioninsequenceofSlc11a2isoformI.bAminoacidresidueimmediatelyprecedingthesiteofinsertionoftheHAepitope.cHAepitope
sequence.8424Biochemistry,Vol.48,No.35,2009Czachorowskietal.
conjugatedanti-mouseantiserum(1:4000dilution),andincuba-
tionwasfor1hatroomtemperaturein5%nonfatmilk/PBS.
TheHRPsubstrate[0.4mg/mLo-phenylenediaminedihy-
drochloride(OPD);SigmaFASTOPD;550μL/well]wasused
accordingtothemanufacturer’sinstructions.Absorbanceread-
ings(492nm)weretakeninanELISAplatereader(Bio-Rad
Model450),andbackgroundabsorbancereadingsfrom(i)
nonspecificbindingofsecondaryantibodyand(ii)nonspecific
bindingofprimaryantibodytovector-transfectedcellswere
subtractedforeachsample.
SequenceAnalyses.BlastanalysesusingthemurineSlc11a2
proteinsequenceasquerywereperformedtoselectsequences
encodingSlc11homologues(Blastscore>400,full-length%
identity>45,expect<2e
-110)fromspeciesrepresentativeof
Unikonta(motileeukaryoticcellswithasingleflagellum),suchas
AmoebozoaandOpisthokonta,comprisingChoanozoa
(unicellulareukaryotewithaflagellum),Fungi,andMetazoa
(multicellularmotileorganisms).Thenonmetazoansequences
wereusedasthephylogeneticoutgroup(37).Thesequencesfrom
selectedmodelspecieswereretrievedusingNCBI’sEntrezlife
sciencessearchengine.Unikonta:Amebozoa(Dictyostelium
discoideum,AAO52395,DdNR2),Opisthokonta:Choanoflagel-
late(Monosigabrevicollis,XP_001750777,MbreNR),Fungi
(Zygomycota,Phycomycesblakesleeanus,PblNR2),andMeta-
zoa:Porifera(Reniera,tropicalmarinedemosponge,ReniNR);
Eumetazoa:Cnidaria(Nematostellavectensis,seaanemone,
XP_001635621,NvecNR)andBilateria:Protostomia:Cuticulata:
arthropoda(Drosophilamelanogaster,Fruitfly,NP_524425
DmeNR,Malvolio)andLophotrozoa,Platyhelminthe
(Schistosomamansoni,AAS99648,SmNRa)andMollusca
(Mizuhopectenyessoensis,Giantscallop,BAD99106,MyesNR),
Deuterostomia:Echinodermata(Strongylocentrotuspurpuratus,
seaurchin,XP_781841,SpurNRa),Chordata(Cionaintestinalis,
Seasquirt,XP_002127954,CinNR),Vertebrata:Actinopterygii,
Teleostfishes(Pagrusmajor,Redseabream,AAR83912,
PmaNR,andDaniorerio,Zebrafish,NP_001035460,DreNR)
Sarcopterygii,Tetrapoda,Amphibia(Frogs)Xenopus(Silurana)
tropicalis(NP_001116938,XtrNR2)andXenopuslaevis
(NP_001088008,XlaeNR1),Amniota,Archosauria(Gallusgal-
lus,chicken,ABI24020,GgNR2,andNP_990295.1,GgNR1),
Mammalia(Musmusculus,laboratorymouse,P49282,MmNR2,
P41251,MmNR1).The18sequenceswerealignedusingClus-
talX(38);themultiplesequencealignmentwaseditedmanually
usingSeaview(39)anddisplayedwithGenedoc.
HomologyModeling.ThemouseSlc11a2sequencewasused
asquerytogeneratethebeststructuralmodelsresembling
templatespresentinthePDBusingathreadingapproach,which
combinestridimensionalfoldrecognitionbysequencealignment
withtemplatecrystalstructuresandmodelstructurerefining(40).
Threecomplementarysuitesofprogramswereused:(a)the
LOcalMEta-Threading-Server(41)thatreturnsthebeststruc-
turepredictionscollectedfromnineindependentthreading
algorithmsandrankedusingaconsensusmethod;i.e.,best
modelsaremorefrequentandstructurallyrefinedusingMode-
ler(42);(b)theI-TASSERserver(43),whichusesvariationsof
thethreadingapproachbasedonsequenceprofile-profilealign-
mentscombinedtoabinitiomodelingbyiterativeimplementa-
tionoftheThreadingASSEmblyRefinement(TASSER);(c)the
MUSTERserver(44)thatusestheMUlti-SourceThreadER
(MUSTER)toincorporateweightedinformationfromsix
sourcestoimprovefoldrecognitionandthatusesalsoModeler
forstructuralrefinement.Thebestmodelswereselectedamongthosepredictedbythethreeapproaches,andcandidatemodels
werecheckedforacceptableCRroot-mean-squareddeviation
(rmsd,<6A
fortemplatewithlessthan20%sequenceidentity
withSlc11a2)andconsistencybetweenlocationsmodeledfor
(Slc11familyspecific)functionalsitesandpriorexperimental
resultsofsolventaccessibilityinsitu:MntHTMD1Asp
34and
TMD6His211apparentlyaccessible,andAsn37,250,401inTMD1,
7,and11,notaccessible(27).
RESULTS
ConstructionofEpitope-TaggedSlc11a2Mutants.
The
multiplesequencealignmentpresentedinFigure1showsthe
positionoftheTMdomainsfortheanimalmembersoftheSlc11
transporterfamilyaspredictedbyhomologythreading(27).To
verifythismodelaswellasestablishthepositionandpolarityof
individualTMDs,weusedanepitopeinsertiontechniquepre-
viouslyutilizedtoestablishthemembranetopologyofseveral
independentABCtransporters(34,45).Thismethodconsistsof
insertinghemmaglutininA(HA)epitopesinvariousextracyto-
plasmic(EC)andintracytoplasmic(IC)loopsdelineatedbypairs
ofpredictedTMDs,expressingtherecombinantproteinsin
transfectedcellsandmonitoringthepolarityoftheengineered
HAtagsbyimmunofluorescenceinintactorpermeabilizedcells
torevealextra-andintracytoplasmicloops.Thisapproachallows
onetoobtaintopologicaldataintransport-competentproteins
properlyexpressedattheplasmamembrane.Additionally,
certainHAtaginsertionscausinglossoftransportcanprovide
additionalcluesonstructure-functionrelationships.
HAepitopes(YPYDVPDYAS)wereinsertedatbothtermini
andineachpredictedloop(fromhydropathyplot(23))between
pairsofconsecutiveTMDs(Figures1and2).Theinsertion
positionsoftheHAepitopesweredeliberatelychosenastheleast
conservedregionsamongSlc11familymemberstominimize
possibledisruptionsonproteinfunction(Figure1).HAepitopes
wereinsertedatpositions1(Nterminus;NTconstruct),98
(construct1),131(construct2),175(construct3),201(construct4),
243(construct5a),284(construct6),344(construct7),403
(construct8a),432(construct9),468(construct10),504
(construct11a),and561(C-terminus,CTconstruct)(Figure2).
Inaddition,twoadjacentHAtagswereinsertedatpositions243
(5b),403(8b),and504(11b)toincreasetheimmunofluorescence
signal.WeusedSlc11a2isoformIcDNAasthebackboneforHA
taginsertionasitshowshighersurfaceexpressionandslower
internalizationkineticsthanisoformIIinporcineLLC-PK1
cells(13,36),facilitatingdetectionofcellsurfaceproteinexpres-
sionbyimmunofluorescence.
ExpressionofEpitope-TaggedSlc11a2Mutantsin
TransfectedLLC-PK1KidneyCells.
AlltaggedSlc11a2
cDNAsweretransfectedintoLLC-PK1cells,andtheexpression
oftheresultingHA-taggedSlc11a2proteinswasverifiedby
immunoblotting.Representativeimmunoblotsusingapolyclo-
nalantibodyrecognizingtheN-terminusofSlc11a2(Figure3A)
andananti-HAmonoclonalantibody(Figure3B)areshownfor
singleclones.Previousstudiesinthesecellshaveshownthat
Slc11a2isdetectedastwoimmunoreactivespeciesof∼60and
∼90kDa,correspondingtocoreglycosylatedandcomplex
glycosylatedforms,respectively(46).Briefly,wecouldobtain
stabletransfectantsexpressingrobustlevelsofSlc11a2for
constructs4-11aswellasforconstructsbearingN-andC-
terminiepitopetags.Forconstructs1-3,itwasdifficulttoobtain
stablytransfectedclonesexpressingrobustlevelsofSlc11a2,
andincellclonespositiveforexpression,onlylowlevelsoftheArticleBiochemistry,Vol.48,No.35,20098425
∼60kDaimmunoreactivebandcouldbedetected(Figure3A).In
addition,expressionofconstructs1-3couldonlybedetectedwiththeanti-Slc11a2antiserumandnotwiththeanti-HA
antibody,suggestingthattheHAepitopemaybeunstableor
FIGURE1:EvolutionaryconservationofthehydrophobiccoreofSlc11a2(TMD1-10).MultiplesequencealignmentofSlc11homologues
fromrepresentativeUnikontaspecies(phylogeneticrelationshipsschematizedandinsertedatthebeginningofthealignment,A,animals,
andV,vertebrates):Amebozoa(DdNR2),Choanoflagellate(MbreNR),Zygomycota(PblNR2)andMetazoa:Porifera(ReniNR)and
Eumetazoa:Cnidaria(NvecNR)andBilateria:Protostomia,Cuticulata(DmeNR)andLophotrozoa,(SmNRaandMyesNR),and
Deuterostomia,Echinodermata,(SpurNRa),Chordata,(CinNR),Vertebrata,Teleostfishes(PmaNRandDreNR),Tetrapoda,Amphibia
(XtrNR2andXlaeNR1),Amniota(GgNR2andGgNR1),andMammalia(MmNR2andMmNR1).Residueconservationisindicatedwith
threecutofflevelsofidentity(40%,80%,and100%).ThepositionofSlc11a2transmembranedomains(TMD)wasdeducedfromhomology
modelingusingthreedifferenttemplatestructures(2a65,LeuT-Slc6(53),3dh4,vSGLT-Slc5,(50)andMhp1-Slc23(51),andhighlighted
withrainbowcolorcode(TMD1,blue,TMD12,red).ThepositionofthemouseNramp2/Slc11a2(MmNR2)proteinusedforallstudiesin
thisworkisthefourthfromthebottomofthemultiplesequencealignment.TheinsertionsitesforHAepitopetags(triangles)resultingin
transportactive(orange)ortransportinactive(black)Slc11a2proteinsareindicated.Thepositionofcriticalresiduesatwhichmutationsare
knowntocauselossoffunction(D86,G88,E299,andR416)ortoimpairseverelyMmNR2transportactivity(E154,G185,D192,Q384,
G394)areindicatedinthesequenceofMmNR2.Theyarecoloredintheprimarysequencerespectivelyinredandmagenta,andtheirposition
isindicatedbyatickofthecorrespondingcolorunderthemultiplesequencealignment.8426Biochemistry,Vol.48,No.35,2009Czachorowskietal.
inaccessiblefordetectionintheseconstructs.Theseresultsalso
suggestthattheHAtaginsertioninconstructs1-3mayhavea
detrimentaleffectonproteinstability,maturation,and/orpro-
cessingtothe∼90kDafullyglycosylatedandmembrane-
expressedisoform.
FunctionalAnalysisofEpitope-TaggedSlc11a2Mut-
antsinTransfectedLLC-PK1KidneyCells.
ToevaluatepossibledetrimentaleffectsoftheHAepitopetaginsertionson
proteinfunction,transportpropertiesofindividualSlc11a2
taggedproteinswereassessedincorrespondingLLC-PK1trans-
fectedcells(Figure4).Indeed,properplasmamembraneexpres-
sionoftheproteinisrequiredtodetectSlc11a2-mediatedmetal
importinintactcells(24,47).Divalentmetaltransportactivity
wasmeasuredusingafluorescencequenchingassayincalcein-
loadedcells(31).ThepH-dependenttransport(pH5.5)oftwo
divalentmetals(Fe
2þandCo2þ)wastestedtoassesspossible
effectsoftaginsertiononionselectivityofSlc11a2(24).Trans-
portrateswereestimatedfromtheslopeofinitialquenching
curvesfollowingadditionofmetaltothecalcein-loadedcells.In
thesestudies,LLC-PK1cellstransfectedwithemptyvectorswere
usedasanegativecontrol,whilecellsexpressingwild-type
Slc11a2wereusedasapositivecontrol.Wenotedthreepheno-
typicconsequencesoftaginsertiononSlc11a2function.For
constructs1-3,HAtaginsertioncompletelyabrogatedprotein
function,withtransportratessimilartothosedetectedinthe
pCB6-transfectednegativecontrol(nosignificantdifference,p>
0.05).Theseresultsindicatethatthissegmentoftheproteinis
highlymutationsensitive(affectingeithertransportperseor
targetingoftheproteintothecellsurface),inagreementwith
resultspreviouslyobtainedwithpointmutantsinthesere-
gions(25-27).Inasecondgroupofconstructs,namely,4,
5(b),6,7,10,and11(b)aswellasNTandCT,theepitope
insertionhadnomajorfunctionalconsequence,withtransport
ratessimilarorsuperiortothosedetectedinthepositivecontrol
(7-12-foldstimulationabovebackground).Finally,constructs
8bor9repeatedlyshowedseverelyreducedtransportrates,which
werenotstatisticallydifferent(p>0.05)fromthosemeasuredin
thenegativecontrol.NomajoreffectofHAtaginsertiononion
selectivitywasdetectedinanyoftheconstructsthatretained
detectablemetaltransportactivity.Wenextproceededtodeter-
minethesubcellularlocation(intra-orextracytoplasmic)ofHA
epitopesinSlc11a2variantsthatretainedtransportactivity.Due
FIGURE2:ConstructionofSlc11a2proteinscontaininghemagglutinin(HA)epitopetags.ThesitesforinsertionofYPYDVPDYAhemagglu-
tinin(HA)epitopesinindividualSlc11a2proteinsareindicated,alongwiththenumericaldesignationoftheconstruct(NT,1-11,CT).Thesehave
beensuperimposedonthehydropathyplotoftheSlc11a2proteinproducedbytheTOPPREDsoftwarepackageusinga20-residuelongsliding
window(corewindow,12residues;wedgewindows,4residues).ThedefaultparameterssuggestedforpredictingthepresenceofTMDsin
eukaryoticproteinswereused,suchastheKyte-Doolittlecutoffvalues(lower,0.6;putative;upper,1.0;certain).Orangearrowheadsindicate
insertionsitesthatproducedfunctionalproteinsinwhichthepolarityofthetagcouldbeestablishedbyimmunofluorescence.Blackarrowheads
indicateinsertionsitesthatcausedpartialorcompletelossoftransport.Singlearrowheadsrepresentsingleepitopetagswhiledoublearrowheads
(5b,8b,11b)representconstructswithtwoepitopetagsinsertedsidebyside.Boxesatthetopofthegraphindicatethelocationoftheproposed2-
foldstructuralsymmetryinSlc11a2.
FIGURE3:ExpressionofHAepitope-taggedSlc11a2constructsin
LLC-PK1cells.LLC-PK1cellswerestablytransfectedwithSlc11a2
isoformIcDNAconstructs(listed)modifiedbytheadditionof
individualordouble(b)HAepitopesandinsertedinthemammalian
expressionvectorpCB6.Totalcelllysates(5μgperlane)were
prepared,resolvedbyelectrophoresisona10%SDS-polyacryla-
midegel,andanalyzedbyWesternblottingwithapolyclonalrabbit
anti-Slc11a2antibody(panelA)orwiththemousemonoclonalanti-
HAepitopeantibody16B12(panelB).Theasteriskindicatesthat
construct7wascreatedbyourgroupinanotherstudy(20).The
positionsofthemolecularmassmarkersareindicatedontheleftof
thefigureinkDa.ArticleBiochemistry,Vol.48,No.35,20098427
tothedetrimentaleffectofthetagsonSlc11a2proteinexpression
andmetaltransportforconstructs1-3,wedidnotpursuefurther
theircharacterization.
MembranePolarityoftheInsertedHA-EpitopeTags.
Theintra-orextracytoplasmicnatureofHAtagswasdetermined
foreachconstructbyimmunofluorescencewiththeircorrespond-
ingLLC-PK1transfectantsusingtheanti-HAmonoclonalanti-
body(Figure5).Immunofluorescencewasperformedinparallel
onintactcellstodetectextracytoplasmictagsandinpermeabi-
lizedcellstreatedwithdetergenttodetectintracytoplasmictags.
Inallcases,cellnucleiwerestainedwithDAPItoidentifythe
positionofthecorrespondingcells(Figure5).LLC-PK1cells
transfectedwithemptypCB6vectorwereusedasnegative
controls,andnosignificantcell-associatedfluorescencecould
bedetectedinthesecellswiththeanti-HAantibodyundereither
nonpermeabilizingorpermeabilizingconditions(Figure5,panels
A-D).Incontrast,strongfluorescentsignalswereobservedin
bothintactandpermeabilizedcellsforconstructs5b(TMD5-6
segment,twoepitopes),7(TMD7-8segment),8b(TMD8-9
segment),and11b(TMD11-12segment,twoepitopes)
(Figure5:N,P;V,X;Z,AB;AL,AN),suggestingthatthe
correspondingsegmentsofSlc11a2areextracytoplasmic.Con-
versely,strongfluorescencesignalswereobtainedonlyunder
permeabilizingconditionsforcellstransfectedwithconstructs
NT(N-terminus),4(HAbetweenTMD4-5segment),6(TMD
6-7segment),9(TMD9-10segment),10(TMD10-11
segment),andCT(C-terminus)(Figure5:F,H;J,L;R,T;
AH,AJ;AP,AR),suggestingthatthecorrespondingtagged
segmentsofSlc11a2areintracytoplasmic.Forcertainconstructs
(5,8,and11),theinsertionofasingleHAtagdidnotproducea
clearfluorescencesignalintransfectedcellsinanyconditions
(datanotshown).However,theinsertionofasecondadjacentHAtagproducedunambiguousimmunofluorescenceresults(5b,
8b,and11b,twoepitopes;Figure5).Foralloftheconstructs,the
polarityoftheHAtagswasvalidatedbyimmunofluorescencein
multipleindependentlytransfectedcellclones.
ValidationofHA-EpitopeLocalizationinSLC11A2.
ThetopologyofinsertedHAtagsestablishedbyimmunofluor-
escencewasvalidatedbyanenzymaticmethod,whichaddition-
allyprovidesaquantitativemeasureofexofacialtagsurface
expression.Forthis,LLC-PK1transfectantsexpressingindivi-
dualHA-taggedSlc11a2constructswerefixedandincubated
withprimaryanti-HAantibodywithorwithoutpriorpermea-
bilizationwithdetergent.Thecellswerethenincubatedwitha
secondaryHRP-conjugatedantibody,andtheamountofbound
secondaryantibodywasquantifiedcolorimetricallyusingthe
HRPsubstrateo-phenylenediaminedihydrochloride.Resultsin
Figure6showtheamountofsurfaceexpression(detectedunder
nonpermeabilizedconditions)ofHA-taggedconstructsex-
pressedasafractionoftotalprotein(detectedunderpermeabi-
lizedconditions).LLC-PK1cellsstablytransfectedwith
constructs5b,7,8b,and11bwerefoundtohaveasignificantly
greaterproportionofaccessibleHA-taggedSlc11a2attheir
surfacethanconstructsNT,4,6,9,10,andCT(Figure6).These
resultsareinagreementwithimmunofluorescenceexperiments
andindicatethatthetagsinconstructs5b,7,8b,and11bare
extracellularwhilethetagsinconstructsNT,4,6,9,10,andCT
areintracellular.
AThree-DimensionalFrameworkforSlc11a2.Homol-
ogymodelinghasrecentlybecomeanimportanttooltopredict
secondaryandtertiarystructuresofmembraneproteins.For
instance,structuralmodelsofthehumandopaminetransporter
DATweregeneratedusingtheLeuTAacrystalstructureasa
template,andthesewerethenusedtopredictprobableprimary
andsecondarypharmacologicalbindingsitesonthetransporter.
Suchbindingsitesweresubsequentlyvalidatedindetaileddock-
ingstudieswithknownDATligands(56).Inanindependent
study,itwasnotedthatthe12TMhelicesinthecrystalstructure
ofE.coliNa
þ/HþexchangerA(EcNhaA)superimposedstrik-
inglywiththepatternofevolutionaryconservationofboththe
bacterialandhumanNa
þ/Hþexchangers.Homologymodeling
wasthenusedtoassigntheposition,boundaries,andangulation
ofthe12TMdomainsinthehumanhomologueNHE1(∼10%
sequenceidentity;14TMDsintotal).Thisexercisepositioned
somekeyresiduesandfunctionalsiteswithinapredictedhighly
conservedmembranefoldinthehumanproteinwhichwerethen
validatedexperimentallyinsite-directedmutagenesisstudies(57).
Topologydataobtainedexperimentallywereevaluatedusinga
tridimensionalframeworkderivedbysequencethreadingusing
availablestructuresofdistantlyrelatedproteins.Currentnum-
bersofproteinfamilies(>30%sequenceidentity)representa3-
foldexcesscomparedtoproteinfolds(<20%sequenceidentity),
andasimilartrendwasreportedformembraneproteinswith
highR-helicalcontent(49).Accordingly,despitelittleaminoacid
sequencesimilarity,threecation-drivenmembranetransporters
(Slc23Mhp1,Slc6LeuT,andSlc5vSGLT)exhibitsimilar
structures,includinganovelfoldforR-helicalintegralmembrane
proteins(50-53).Inaddition,thesestructuresrepresentdiscrete
possiblestepsincation-driventransportcycle(opentoout,open
toout-occludedbysubstrate,opentoin).Homologymodelingof
Slc11a2showedexcellentsuperimpositionontheseapparently
unrelatedtemplates(Figure7),implyingcommonarchitectural
featuresunderlyingsharedmechanisticaspectsofmembrane
transport.ThemodelsthatarepresentedinFigure7havebeen
FIGURE4:MetaltransportactivityofHAepitopetaggedSlc11a2
proteinsexpressedinLLC-PK1cells.Metal-transportactivityof
HA-taggedSlc11a2variantswastestedusingafluorescence-quench-
ingassay.Briefly,110
6cellswereloadedwiththemetal-sensitive
fluorescentdyecalcein-AM,followedbyincubationwitheither
20μMCo
2þor20μMFe2þinacidic(pH5.0)bufferandcontinuous
monitoringoffluorescencequenchingovertime(2min)usinga
spectrofluorometer(excitationwavelength,488nm;emissionwave-
length,517nm).Theslopesoftheinitialfluorescencequenching
curveswerecalculatedandreflectcationtransportactivityofindivi-
dualHAconstructs.Resultsareshownastheinitialratesof
fluorescencequenching,asdescribedpreviously(24),witherrorbars
representingthestandarderroronthemeansofthreetofive
independentexperiments.*,one-wayANOVAfollowedbyaDun-
nett’sposthoctestindicatednosignificantdifference(P>0.05)in
quenchratebetweentheemptyvectorcontrol(pCB6)andconstructs
1-3,8,and9.Furthermore,nosignificantdifference(P=0.95;two-
wayANOVA)wasobservedinquenchrateforthetwometalstested
foranyoftheclonesanalyzed.8428Biochemistry,Vol.48,No.35,2009Czachorowskietal.
highlighted(bold)inSupportingInformationTable1.Models
wererankedaccordingtotheirpredictionscores,wherehigher
valueincreasesconfidenceinthemodelpredicted(takingintoaccounttherespectivecutoffscoresofeachprogram).Acompi-
lationoftheseresultsisprovidedinSupportingInformation
Table1togetherwithdetailsregardingtheLOMETSpredictions
FIGURE5:DetectionofHAepitope-taggedSlc11a2proteinsbyimmunofluorescence.LLC-PK1cellsstablytransfectedwitheitheremptypCB6
vectororindividualHA-taggedSlc11a2variantswereanalyzedbyimmunofluorescence.AschematicrepresentationofthesiteofHAtaginsertion
inSlc11a2variantsisshownintheleftpanel.Immunofluorescencewascarriedoutwiththemousemonoclonalanti-HAepitopeantibody16B12
oncellseitheruntreated(intactcells;lefttwocolumns)orpretreatedwith0.1%TritonX-100(permeabilized;righttwocolumns).Cellswerethen
incubatedwithsecondarygoatanti-mouseantibodyconjugatedtorhodamine(Cy3),andimageswereacquiredbyepifluorescencemicroscopy.
TheHAepitopesinconstructs5b(M-P),7(U-X),8b(Y-AB),and11b(AK-AN)weredetectedinintactandpermeabilizedcells
(extracytoplasmic),whiletheHAepitopesinconstructsNT(E-H),4(I-L),6(Q-T),9(AC-AF),10(AG-AJ),andCT(AO-AR)weredetectable
onlyinpermeabilizedcells(intracytoplasmic).Nuclearstaining(DAPI,columns1and3)wasusedasacontroltoascertainthepresenceofcellsin
themicroscopefield.ArticleBiochemistry,Vol.48,No.35,20098429
whicharegiveninSupportingInformationTable2.Similar
resultswereobtainedusingdifferentapproaches(mGenTHRE-
ADER,GenTHREADER,CBSWWWServer)(datanot
shown).Modelsfortheopen-to-outform(resemblingLeuT
andMhp1),andfortheopen-to-inform(vSGLT)werefre-
quentlyobtainedwithsignificantscores.Althoughthemodels
obtainedwiththetemplateLeuThadlowerconfidencevalues,
theirarchitecturewashighlyconsistentwiththosederivedfrom
bothMhp1andvSGLTtemplates.Finally,themodelspresented
wereselectedforconsistencywithadditionalfunctionaldata
publishedbyourgroup.Toevaluateaccuracyinmodelingthe
Slc11a2fold,weusedarepresentativesetofcandidatemodels
thatcomprisedthetopfourcandidatestructuresproducedbythe
programMUSTER,whichrepresenteithertheLeuT/vSGLT/
Mhp1foldortheGlpT/MFSfold(SupportingInformation
Table1).ThesemodelswerecheckedforacceptableCRroot-
mean-squareddeviation(rmsd)usingthreedifferentprograms
thatperformpairwisestructurealignmentandprovideaglobal
rmsd.Eachmodelwasalignedbothwithitsrespectivetemplate
andwithafourthtemplatethatalsofitstheLeuT/vSGLT/Mhp1
fold(BetP).TheresultspresentedinSupportingInformation
Table3demonstrateexcellentsuperpositionofeachmodel
structurewithitsrespectivetemplateandshowthatmodels
representingtheLeuTfoldsuperimposebetterontheBetP
templatethanaMFSmodel.Notably,themostconserved
elementsoftheSlc11hydrophobiccore(TMD1,3,6,and8;
Figure1)correspondtoTMDsthatplaykeyrolesinthecation-
drivenmembranetransportmechanism(50-53)(Figure7).
TheSlc11a2tridimensionalmodelsderivedfromMhp1,LeuT,
andSGLTwereusedtolocateresiduesatwhichnaturally
occurringandexperimentallyinducedpointmutationsabrogate
SLC11A2transportactivityinmiceandhumans.Thesemuta-
tionswerefoundtoaffectsymmetricelementssituatedinthe
sameregionofthepredictedfold:threemaptolongTMD3(E154)andTMD8(Q384,G394),andthemutationscausing
anemiainmammalsaffecttheadjacentTMD4and9(G185,
R416,SupportingInformationFigure1).Suchstructure-func-
tionrelationshipssupportthestructuralmodelweproposefor
Slc11a2,includingalongandobliquearrangementforTMD3
and8(Figure7andSupportingInformationFigure2).Addi-
tionalmutationsdetrimentalforSlc11a2affectotherpartsofthe
moleculethatarepredictedtobemobileduringthetransport
cycle(respectivelyTMD1and7andTMD2and5,Supporting
InformationFigures1and2).Thetridimensionallocalizationof
criticalSlc11a2mutationsprovidescircumstantialsupportforthe
proposedstructuralmodel.Thus,ourexperimentalresultswith
insertedHAtagsinintactcells,includingtheireffectson
transport,canbediscussed,takingintoconsiderationthis
structuralandfunctionalframework.
DISCUSSION
Wehaveusedepitopetaggingtomapthepositionandpolarity
ofindividualTMDoftheSlc11a2metaltransporter.This
approachconsistsofinsertingHAtagsatsitesshowingrelatively
lowsequenceconservationanddelineatingTMDsegments
predictedbyhydropathyprofilingandothercomputationaltools
(Figures1and2).Therecombinant,HA-taggedproteinsare
FIGURE6:DetectionofHAepitope-taggedproteinsbysurface
labeling.StablytransfectedLLC-PK1cellswereincubatedwith
anti-HAantibodywithorwithoutpriordetergent(TritonX-100)
permeabilization,followedbyincubationwithahorseradishperox-
idase(HRP)-coupledsecondaryantibody.Theamountofbound
secondaryantibodywasdeterminedbyacolorimetricassayusing
o-phenylenediaminedihydrochloride(measuredat492nm).The
presenceofSlc11a2-HAexpressedatthecellsurface(inintactcells)
isshownasafractionoftotalproteinexpression(inpermeabilized
cells)normalizedfornonspecificbindingofprimaryandsecondary
antibodiesinnontransfectedLLC-PK1cells.Errorbarsrepresent
standarderrorsofthemeansoftwoindependentexperimentscarried
outinduplicate.*,one-wayANOVAfollowedbyaDunnett’s
posthoctestindicatedasignificantdifference(P<0.01)incell
surfacefluorescencebetweencellstransfectedwiththeemptyvector
control(pCB6)andconstructs5b,7,8b,and11b.
FIGURE7:PositionofHAepitopeinsertionsitesonthepredicted
structuralmodelofSlc11a2.Slc11a2modelswerededucedfrom
homologythreadingandstructurerefiningbasedonthecrystal
structuresofLeuT(top)andvSGLT(bottom).Thesemayrepresent
discretestepsinthetranslocationcycle(opentooutside/occludedand
opentoinside,respectively)ofastructuralfoldsharedbydifferent
families(e.g.,Slc5,6,and23).Theviewispresentedfromtheexternal
surface(outside).Forclarity,thepredicted10TMDformingthe
hydrophobiccoreofSlc11a2hasbeensplitatsite5topresentthe
N-andC-terminalhalvesjuxtaposedhorizontallyandcomprising
respectivelyTMD1-5(bluetogreen)andTMD6-10(greento
orange).TheSlc11-specificfunctionalmotifsAsp-Pro-Glyand
Met-Pro-His(27)arehighlightedinredandmagenta,andtheir
respectivelocationsbetweenTMD1a/bandTM6a/bareindicated
intheLeuTmodel(top).Thesidechainsoftheaminoacidsflanking
eachinsertionsite(numberedaccordingtotheprecedingTMD)are
highlightedinblack.Comparisonofthetopandbottompanelsshows
possiblemovementsthatmayfacilitatesubstrateuptake(N-domain,
TMDs1,2,and5,andC-domain,TMDs6,7,and10).Detailsonthe
modelspresentedinFigure7areprovidedinSupportingInformation
Tables1and3.8430Biochemistry,Vol.48,No.35,2009Czachorowskietal.
stablyexpressedandtestedfunctionallyinmammaliancells,
wherethepolarityofthetagisestablishedbyimmunofluores-
cence(Figure5)inintact(extracytoplasmic)vspermeabilized
cells(intracytoplasmic).Theexperimentalresultsaretheneval-
uatedusingcandidatetridimensionalmodelsobtainedbyse-
quencethreadingusingavailablestructuresinthePDBdatabase.
Thisapproachprovidesseveraladvantagesovermappingtech-
niquesusingtruncatedproteinsfusedtoindicatorgenesand
proteins(32,48).Indeed,topologydataobtainedbyepitope
mappingarewithfunctionalproteinsthatareproperlyprocessed
andinsertedinthemembraneofintactcellsinatransport-
competentstate.Bythesametoken,topologydataobtainedfrom
HA-taggedproteinsshowingalteredtransportfunctionmustbe
interpretedwithgreatcaution.Analysesofsequencethreading
andmotifconservationareusefulforthispurpose;forinstance,
taginsertionsimpairingSlc11a2activitywerefoundmostlyin(a)
shortloopsconnectingconservedTMDs(constructs1,3,8b,and9),
(b)thegroupoffivepositivechargesthatareconservedinthe15
residuestretchspanningtheloopbetweenTM8-9,and(c)theN-
endofTMD9whichmaycontainatopogenic(positive-inside)
signal.Thus,epitopeinsertionsthatimpairtransportactivity
maypointtoproteinsegmentsthatareessentialforproper
structure,membranearrangement,andfunctionoftheproteins
andthatmaybestudiedbyfurthermutagenesis.
ConstructsNT,4,5b,6,7,10,11b,andCTweretransport
activeandshowedunambiguouspolarityinimmunofluorores-
cencestudiesbeingeitherintracytoplasmic(NT,4,6,10,CT),or
extracytoplasmic(5b,7,11b).Thisdemonstratesthatthe
N-terminus,theC-terminus,andtheproteinsegmentsseparating
TM4-5,TM6-7,andTM10-11areintracytoplasmic.Onthe
otherhand,thesegmentsseparatingTM5-6,TM7-8,and
TM11-12canbeunambiguouslyassignedasextracellular.These
resultsareinagreementwithstructuralmodelsdeducedfrom
sequenceprofilingandmodeling(Figures1and2)butalsowith
topologydataobtainedforthedistantbacterialSlc11relative
MntH(32).Geneticfusionsofcytoplasmicorperiplasmic
reporterproteinsatdifferentlocationsinMntHsuggestedthat
loops1/2,5/6,7/8,and9/10andtheC-terminusareperiplasmic
whileloops2/3,6/7,8/9,and10/11arecytoplasmic.Also,both
theC-terminusandthesegmentseparatingTMD7and8were
showntobeperiplasmicintransport-activeMntH.Theseand
additionaldatasuggestedan11TMDtopologymodel,wherethe
MntHN-terminusiscytoplasmicandtheC-terminusperiplas-
mic,whichwasconsistentwithcomputer-basedpredictionsbased
onhydrophobicity,secondarystructurepropensity,conserva-
tion,andchargedensity(32).
Ontheotherhand,notopologydatacouldbeobtainedfor
Slc11a2proteinsegmentsdelineatedbytheTM1-2,TM2-3,
andTM3-4intervalsandrepresentedbyHAtaginsertionsin
constructs1-3.Fortheseconstructs,wewereunabletoisolate
stableLLC-PK1transfectantsexpressinghighlevelsofthe
correspondingproteinsdespiteseveralattempts.Positiveclones
onlyshowedlowlevelexpressionofaproteinisoform(∼60kDa)
thatcorrespondstotheimmatureformofSlc11a2.Weinterpret
thisresultasepitopeinsertioninthisportionoftheprotein
causingimpropermaturationandprocessingoftheprotein
(retentionintheendoplasmicreticulum),preventingfunctional
insertioninthemembraneasdeterminedbytheabsenceofan
immunofluorescencesignalintransfectedcells(datanotshown)
andultimatelyleadingtoabsenceofdetectabletransport.The
mutation-sensitivenatureofthisregionofSlc11a2hasbeen
previouslynotedinparallelmutagenesisexperiments(25).ComplementationstudiesinyeastSmfmutantsandelectrophy-
siologicalmeasurementsinXenopusoocytesshowedthatthe
highlyconservedC-terminalportionofpredictedTMD1andthe
N-terminalportionofpredictedTMD2arecriticalfortransport.
Thesestudiesestablishedtheimportanceofseveralresiduesin
substratespecificity(includingG119,D124,andQ126)and
establishedacriticalrolefortheG88residuefoundinthe
invariantDPGNmotifinthisregion(25,28).Finally,theparallel
mutationalstudiesofpositionsD93andQ95andoftheDPGN
motiffromtheMntHbacterialrelativehaveledtotheproposal
thatthisregionofTM1-TM2mayformpartofametalbinding
siteintheprotein(26,27,54).TheLeuTstructuralmodel(52,53)
suggestsapseudo-2-foldsymmetry,inwhichTMDs1and6
containmembranespanningdiscontinuoushelicesinvolvedin
substraterecognitionandionbinding(Figure7).Suchacritical
roleforTMD1couldhelptoexplainthelossoffunction
mutationofSlc11a2construct1,whichbearsanHAtaginsertion
immediatelyC-terminaltoTMD1.Likewise,thecrystalstructure
ofLeuTrevealssubstantialsecondarystructureinthehydro-
philicloopsbetweenTMDs2-4and7-8,loopsimplicatedin
conformationalchangesduringsubstratetransport(52,53).
Therefore,itistemptingtospeculatethatdisruptionofsuch
shortR-helices(possiblyconservedinSlc11a2)mayalsoexplain
thelossofproteinstability,maturation,and/ortransportnoted
forconstructs2and3bearingHAepitopesinthesesegments.
Constructs8b(insertioninTM8-9segment,position403)and
9(insertioninTM9-10segment,position432)gaveambiguous
resultsthatwarrantadditionaldiscussion.IntransfectedLLC-
PK1cells,theseepitopeswereaccessibletotheanti-HAantibody
andcouldbemappedtotheextracytoplasmicandintracytoplas-
micfaceofthemembrane,respectively.However,thislocaliza-
tiondoesnotagreeeitherwiththetopologicalmodelproposedby
hydropathyprofiling(Figures1and2)andthatissupportedby
topologymappinginotherfunctionalHAtaggedconstructs,nor
withtheproposedmembranestructureofthedistantbacterial
MntHrelative(27,32),norwiththestructuralmodelobtainedby
threadingandhomologymapping(Figure7andSupporting
InformationFigures1and2).Interestingly,thecorresponding
Slc11a2constructs8band9showedconsiderablyreduced
transportactivity,indicatingthatthetaginsertiondisrupted
oneormorestructuraland/orfunctionalpropertiesofthetwo
variants.Therefore,wecannotexcludethepossibilitythattag
insertionintheseconstructsdisruptednormaltopologyor
membraneinsertionoftheproteinsleadingto(a)reduced
transportactivityand(b)aberrantexposureoftheHAepitope
anderroneousassociatedtopologydata.Finally,hydropathy
predictionsusedtoderivesecondarystructuremodels(Figure2)
arelargelybasedonthepresumptionthatmostTMDslie
perpendicularlytotheplaneofthemembraneandconsistof
approximately19or20aminoacids.Thisassumptionmaynotbe
validforallTMDs.TMDs3and8inthetransportersLeuTand
Mhp1andtheircounterpartsTMDs4and9inSGLTconsistof
30ormoreaminoacids.Theyspanthemembraneatoblique
anglesandappeartotiltaroundacentralaxisparalleltothe
membraneplane,suggestingthattheirextremitiesmayalterna-
tivelybeexposedtothemembranehydrophobiccoreand/orbe
engagedininterhelicalcontactsduringtransport(50,51,53)-
(SupportingInformationFigure2).Anyepitopeinsertionnear
theboundariesofsuchlongTMDwouldobviouslycompromise
topologyandfunction.Additionalepitopemappingexperiments
withotherepitopetagsinsertedinneighboringpositionswillbe
requiredtoformallytestthishypothesis.ArticleBiochemistry,Vol.48,No.35,20098431
RegardingHAtag8b,suchadoubletagcould,ontheone
hand,weakenthetopogenicsignalthatprecedesTMD9by
dissipatingthenetpositivechargeofthishydrophilicsegment
andpreventingcorrectpositioningrelativetothetranslocon
machineryandthecytoplasmicsideofthemembrane.Onthe
otherhand,thiseffectmightbeamplifiedbythepeculiar
conformationofTMD8,whichismodeledforminganangle
closeto45degwiththemembraneplaneandlikelyinteracting
withotherhelices(Figure7).Therefore,HAtaginsertion
immediatelydownstreamofN403(inconstruct8b)mayindeed
bedetrimentaltofunction.Thisfourthpredictedintracellular
loopofSlc11a2(targetedinconstruct8b)followsasegment
highlyconservedamongSlc11orthologues,includingamuta-
tion-sensitiveandhighlyconserved20aminoacidsequencemotif
(positions384-403).Thisconservedsequencemotifshowsthat
manyinvariantresiduesandmutationsatseveralofthesesites
significantlyreduceorcompletelyabrogatetransport(26,28).
Likewise,theinsertionofanHAepitopetaginthecorresponding
segmentoftheyeastorthologueSMF1alsocausedalossof
function(55).Butwhetherthishighlyconservedsequencemotif
formsamembrane-reentrantloopfollowingTMD8orisinstead
integraltoalongerTMD8ascurrentlysuggestedbyhomology
modeling(Figure7)remainstobeclarified.
Theresultsyieldedbytheapparentlysymmetricconstructs4
and9weresomewhatunexpectedgiventhelevelsofsequence
conservationobservedatthecorrespondingsites.Modification
ofthewell-conservedloopseparatingTM4-5didnotalter
Slc11a2functionalexpression,providingunambiguousdatain
aregionthathadnotbeenpreviouslyprobedinanySlc11
members.Incontrast,insertionoftheHAtaginthemoderately
conservedloopseparatingTM9-10mayhaveimpairedmem-
braneinsertionofTMD9.ThisTMDisshortcomparativelyto
others,andithasapositivelychargedaminoend.Insertionof
twonegativecharges(fromtheHAtag)closetothecarboxylend
ofTMD9mayfurtherreducetheoverallhydrophobicityofthat
TMD,resultinginlesseffectivemembranepartitioningand
preventingproperoverallmembraneinsertion.Also,basedon
theroleinexternalgatingofthetranslocationpathwaythatwas
reportedforTMD10(Figure7,C-repeat,comparetheLeuTand
vSGLTconformations),anotherpossibleexplanationisthatHA
taginsertionatsite9mayimpedeaconformationalchangein
Slc11a2thatmaybecriticalfortransportfunction.
Wehavealsousedthesequenceofconstructs1,2,3,7,8b,9,
and10asqueriesforthreadingandhomologymodeling(datanot
shown).Theseanalysesshowedthatepitopetaggingatsites
experimentallyobservedtoseverelyaffectproteinfunctionhada
markedeffectonhomologymodeling(e.g.,muchlowerscore
predictioncomparedtowild-typeNramp2forconstruct1<
construct8b