ManuscriptsubmittedtoeLife
CharacterizationoftheABC
1
methioninetransporterfrom
2
Neisseriameningitidis
revealsthat
3
lipidatedMetQisrequiredfor
4
interaction
5
NaimaG.Sharaf
1,2,
†
,*
,MonaShahgholi
1
,EstherKim
1
,JeffreyY.Lai
1,2
,David
6
VanderVelde
1
,AllenT.Lee
1,2
,DouglasC.Rees
1,2,*
7
*Forcorrespondence:
dcrees@caltech.edu
(DCR);
ngsharaf@stanford.edu
(NGS)
1
CaliforniaInstituteofTechnology,DivisionofChemistryandChemicalEngineering
8
114-96,CaliforniaInstituteofTechnology,Pasadena,CA91125USA;
2
HowardHughes
9
MedicalInstitute,CaliforniaInstituteofTechnology,Pasadena,CA91125,USA;
†
Current
10
address:DepartmentofBiology,StanfordUniversity,Stanford,CA94305,USA
11
12
Abstract
NmMetQisasubstrate-bindingprotein(SBP)from
Neisseriameningitidis
thathas
13
beenidentifiedasasurface-exposedcandidateantigenformeningococcalvaccines.However,
14
thislocationforNmMetQchallengestheprevailingviewthatSBPsinGram-negativebacteriaare
15
localizedtotheperiplasmicspacetopromoteinteractionwiththeircognateABCtransporter
16
embeddedinthebacterialinnermembrane.ToelucidatetherolesofNmMetQ,wecharacterized
17
NmMetQwithandwithoutitscognateABCtransporter(NmMetNI).Here,weshowthatNmMetQ
18
isalipoprotein(lipo-NmMetQ)thatbindsmultiplemethionineanalogsandstimulatestheATPase
19
activityofNmMetNI.Usingsingle-particleelectroncryo-microscopy,wedeterminedthe
20
structuresofNmMetNIinthepresenceandabsenceoflipo-NmMetQ.Basedonourdata,we
21
proposethatNmMetQtetherstomembranesviaalipidanchorandhasdualfunctionand
22
localization,playingaroleinNmMetNI-mediatedtransportattheinnermembraneand
23
moonlightingonthebacterialsurface.
24
25
Introduction
26
Thesubstrate-bindingprotein(SBP)NmMetQfromthehumanpathogen
Neisseriameningitidis
has
27
beenidentifiedasasurface-exposedcandidateantigenforthemeningococcalvaccine(
Pizzaetal.,
28
2000
). Subsequently, NmMetQhasbeenshowntointeractwithhumanbrainmicrovascularen-
29
dothelialcells(
Kánováetal.,2018
),potentiallyactingasanadhesin.However,thesurfacelocaliza-
30
tionofNmMetQchallengestheprevailingviewthatSBPsresideintheperiplasmofGram-negative
31
bacteria(
ThomasandTampé,2020
),bindinganddeliveringmoleculestocognateATP-BindingCas-
32
sette(ABC)transportersintheinnermembrane(IM).Severalquestionsarisefromthesestudies:
33
HasNmMetQlostitsABCtransporter-dependentfunctionintheIM?HowdoesNmMetQbecome
34
embeddedintheoutermembrane(OM)surfaceofthebacterium?
35
TheABCtransporter-dependentroleofSBPshasbeenwellcharacterizedformultipleABCtrans-
36
portersystems(
Hollensteinetal.,2007
;
Oldhametal.,2013
;
Sabrialabedetal.,2020
;
Liuetal.,
37
1of22
ManuscriptsubmittedtoeLife
2020
;
Nguyenetal.,2018
;
deBoeretal.,2019
). ThesestudiesrevealconservedSBP-dependent
38
characteristics,includingthattheSBPislargelyresponsibleforsubstratedeliverytotheABCtrans-
39
porter,withconcomitantstimulationofthetransportcoupledATPaseactivity. Structuralstudies
40
have shown that SBPs dock to the periplasmic surface of the transporter’s transmembrane do-
41
mains,withthesubstrate-bindingpocketjuxtaposedwiththetranslocationpathwayofthetrans-
42
porter. WhilemanySBPshaveonlybeenassignedABCtransporter-dependentfunctions, afew
43
SBPshavealsobeenshowntohavebothABCtransporter-dependentandABCtransporter-inde-
44
pendentfunctions(oftenreferredtoasmoonlightingfunctions)(
Adler,1975
). Forexample,the
E.
45
coli
maltoseSBP(MBP)bindsandstimulatesitscognateABCtransporterduringthemaltoseimport
46
cycle(
Davidsonetal.,1992
).Inaddition,theMBP-maltosecomplexisalsoaligandforthechemo-
47
taxisreceptor,triggeringthesignalingcascadeinvolvedinnutrientacquisition(
Hazelbauer,1975
;
48
Mansonetal.,1985
).OtherSBPshavealsobeenassignedABCtransporter-independentfunctions
49
(
Mülleretal.,2007
;
Castañeda-Roldánetal.,2006
;
Matthysseetal.,1996
),includingNspSfrom
50
Vibriocholerae
,whichhasbeenshowntoplayaroleinbiofilmformation(
Youngetal.,2021
)and
51
nottransport(
Cockerelletal.,2014
).Additionally,twoMetQproteins,
N.gonorrhoeae
(Ng)NgMetQ
52
and
Vibriovulnificus
(Vv)VvMetQhavealsobeenidentifiedasputativeadhesins,mediatingbacte-
53
rialadhesiontohumancervicalepithelialcells(
Semchenkoetal.,2016
)andtohumanintestinal
54
epithelialcells(
Leeetal.,2010
;
Yuetal.,2011
),respectively.EvidencethattheseMetQSBPsbind
55
andstimulatetheircognateABCtransporters,however,islacking. WhetherNmMetQhaslostits
56
ATPtransporter-dependentfunctionorwhetheritplaysrolesatboththeIMandOMcannotbe
57
determinedthroughaminoacidsequencealoneandmustbeexperimentallyverified.
58
Since SBPs are not membrane proteins, the detection of NmMetQ at the cell surface of the
59
bacteriumsuggestsitmustbetetheredtotheOM.InGram-negativebacteria,theparadigmthat
60
SBPstranslocateintotheperiplasmwheretheydiffusefreelybetweentheIMandOMcanbetraced
61
backtoearlyexperimentsbyHeppelshowingthattheosmoticshockofGram-negativebacteria
62
leadstothereleaseofSBPs(
Heppel,1969
).WhilemanySBPsinGram-negativebacteriahavebeen
63
identifiedassecretedproteins(
WillisandFurlong,1974
;
Ahlemetal.,1982
),severalstudieshave
64
alsoidentifiedafewlipid-modifiedSBPs(lipo-SBP)(
Tokudaetal.,2007
)). However,thepresence
65
oflipo-SBPsinGram-negativebacteriahasnotbeengenerallyappreciatedandtherolethatlipid
66
modificationsplayinSBPsurfacelocalizationremainsunexplored.
67
AlthoughABCtransporter-dependentfunctionsofNmMetQ,VvMetQ,andNgMetQarenotwell
68
studied,thehomologousSBPfrom
E.coli
,EcMetQ,iswellcharacterized. Studiesshowthatthe
E.
69
coli
methionineuptakesystemconsistsofEcMetQanditscognateABCtransporterEcMetNI
Kadner
70
(
1974
,1977).StructuresofbothEcMetQandEcMetNIaloneandincomplexareavailable.(
Kadaba
71
etal.,2008
;
Johnsonetal.,2012
;
Nguyenetal.,2015
,
2018
). EcMetNIcomprisestwotransmem-
72
branedomains(TMD),whichformasubstratetranslocationpathway,andtwonucleotide-binding
73
domains(NBD),whichcoupletransporttothebindingandhydrolysisofATP.Intheabsenceof
74
EcMetQ,EcMetNIadoptstheinward-facingconformation,withtheTMDsopentothecytoplasm
75
andNBDsseparated. TheavailablecrystalstructuresofEcMetQrevealtwodomainsconnected
76
byalinkerthatformthemethionine-bindingpocket(
Nguyenetal.,2015
). Ofnote,EcMetQhas
77
beenexperimentallyverifiedtobealipoproteinbyradioactivepalmitatelabeling(
Tokudaetal.,
78
2007
). Additionally,Carlsonetal. foundthatwild-typeEcMetQremainsassociatedwithrecombi-
79
nantlyexpressedhis-taggedEcMetNIwhensolubilizedindetergentandpeptidiscsbutnotwhen
80
itsN-terminalcysteineismutatedtopreventlipidation(
Carlsonetal.,2019
).Thisstudyshowsthat
81
EcMetQassociationwithEcMetNIdependsonitsN-terminallipid. StructuresofEcMetQarealso
82
available,however,thelipidmodificationisnotpresentinEcMetQstructures. Astructureofthe
83
EcMetQ:EcMetNIcomplexisalsoavailableandshowsEcMetNIintheoutward-facingconformation,
84
withtheTMDsandNBDsclosetogether.Inthisstructure,EcMetQisdockedtotheperiplasmicsur-
85
faceoftheTMDswiththebindingpocketopentothecentralcavity(
Nguyenetal.,2018
). These
86
structures,togetherwith
invivo
functionalassays(
Nguyenetal.,2018
;
Kadner,1974
,
1977
),show
87
thatEcMetQisintimatelyinvolvedinEcMetNI-mediatedmethioninetransport.
88
2of22
ManuscriptsubmittedtoeLife
Whereas the interaction between EcMetQ and EcMetNI is well characterized, less is known
89
aboutthecorrespondingsystemin
Neisseriameningitidis
.Todate,therehavebeennobiochemical
90
orstructuralstudiesreportedforNmMetNI.RecentlydeterminedstructuresofNmMetQareinthe
91
ligand-free,L-methionine-,orD-methionine-boundstates,andbindingassaysshowL-methionine
92
bindsNmMetQwithgreateraffinitythanD-methionine(
Nguyenetal.,2019
). Thesestudieswere
93
carriedoutwithanNmMetQproteinthatlacksthenativeN-terminalsignalsequence,establish-
94
ingthattheN-terminalsignalsequenceisnotnecessaryforligandbinding. However, NmMetQ
95
ispredictedtobelipoproteinbasedontheN-terminalproteinsequence(UniprotentryQ7DD63)
96
(
Consortium,2019
). Experimentalevidenceconfirmingthismodification,however,hasnotbeen
97
reported. Thus, a full understanding of the post-translational modification of NmMetQ and its
98
interactions with NmMetNI are lacking. To better understand NmMetQ and the role it plays in
99
methioninetransport,adetailedcharacterizationofbothNmMetNIandNmMetQwithitsnative
100
N-terminalsignalsequenceisrequired.
101
Inthiswork,wecharacterizedNmMetQandNmMetNIusingmultiplebiophysicalmethods.Us-
102
ingmassspectrometryandsite-directedmutagenesis,wedemonstratethatfull-lengthNmMetQ,
103
recombinantly-expressedin
E.coli
,isalipoprotein(lipo-NmMetQ).Functionalassaysshowedthat
104
bothlipo-NmMetQandL-methioninearerequiredformaximalstimulationofNmMetNIATPase
105
activity. NmMetNIwasalsostimulatedtoalesserextentbypre-proteinNmMetQ(avariantwith
106
anunprocessedN-terminalsignalpeptide)withL-methionineandbylipo-NmMetQwithselectme-
107
thionineanalogs. WealsodeterminedthestructuresofNmMetNIinthepresenceandabsence
108
of lipo-NmMetQ to 6.4
Å
and 3.3
Å
resolution, respectively, using single-particle electron cryo-
109
microscopy (cryo-EM). Using a bioinformatics approach, we also identified MetQ proteins from
110
otherGram-negativebacteriathatarepredictedtobemodifiedwithlipids.Thisanalysissuggests
111
thatthelipidmodificationofMetQproteinsisnotrestrictedto
N.meningitidis
and
E.coli
.
112
Basedonourdata,weproposethatlipo-NmMetQ,andmoregenerallylipo-MetQproteinsin
113
otherGram-negativebacteria,possessesdualfunctionandlocalization:ABCtransporter-dependent
114
rolesattheIMandamoonlightingABCtransporter-independentrole(orroles)attheOM.Ourfind-
115
ingshighlightthecomplexityofthecellenvelopeandthatmuchremainstobeunderstoodabout
116
therulesgoverningproteinlocalizationinGram-negativebacteriaandthemoonlightingfunctions
117
ofSBPsonthesurfaceofthecell.
118
Results
119
N.meningitidis
MetQisalipoprotein
120
Whilelipoproteinsandsecretedproteinsbothmusttraversetheinnercellmembraneduringbio-
121
genesis,theirmaturationoccursthroughdifferentmechanismsdependingontheN-terminalsig-
122
nal sequence (
Figure 1
A). Lipoproteins are synthesized in the cytoplasm as pre-prolipoproteins,
123
insertedintotheIM,andthenanchoredviatheirN-terminalsignalsequencetothecytoplasmic
124
membrane(
OkudaandTokuda,2011
).WhiletetheredtotheIMthroughthesignalsequence,pre-
125
prolipoproteinsaresubsequentlymodifiedbythreeenzymes:(1)phosphatidylglyceroltransferase
126
(Lgt),whichtransfersthediacylglycerolgrouppreferentiallyfromphosphatidylglycerol(PG)tothe
127
cysteineresidueviaathioesterbondofthepre-prolipoprotein,producingaprolipoprotein(
Mao
128
etal.,2016
); (2)signalpeptidaseII(LspA),whichcleavestheprolipoproteinN-terminalsignalse-
129
quence to yield a diacylated lipoprotein with the N-terminal cysteine (
Hussain et al., 1982
;
Vo-
130
geleyetal.,2016
); and(3)apolipoproteinN-acyltransferase(Lnt),whichN-acylatesthecysteine
131
residuepreferentiallyusinganacylgroupofphosphatidylethanolamine(PE)toproduceatriacy-
132
latedlipoprotein. (
Nolandetal.,2017
;
Wiktoretal.,2017
). Similartolipoproteins,secretedpro-
133
teinsaresynthesizedinthecytoplasmaspre-proteinswithanN-terminalsignalsequence.These
134
pre-proteinsserveassubstratesforsignalpeptidaseI(SpaseI),whichcleavestheN-terminalsignal
135
sequencetoyieldthematuresecretedprotein(
Karlaetal.,2005
;
Paetzeletal.,1998
).
136
NmMetQispredictedtobealipoproteinbySignalP5.0,adeepneuralnetworkalgorithmthat
137
3of22
ManuscriptsubmittedtoeLife
Figure1.
Massspectrometry(MS)analysisoflipo-NmMetQandNmMetQC20Aproteins.
A.
(Top)Schematicoflipoproteinmaturationpathway.
Insetcontainsaschematicofalipoproteinwithacylchaincomposition[16:0,16:0,16:0].Acylchainsaregroupedinadottedlineboxandtheir
averagemassesarecalculated.Belowtheschematicarethetheoreticalmassesforthelipo-NmMetQproteins(initalics)assumingtriacylation
occursviathecanonicallipoproteinmaturationpathwayduetothesequentialactionofthreeenzymes(Lgt,LspA,andLnt).Thenumbersinthe
bracketscorrespondtothetotalnumberofcarbonsanddoublebonds,respectively,presentinthefattyacylchainsofthelipid.(Bottom)
SchematicofvariousNmMetQC20Aproteinswithexampletheoreticalaveragemasses,showninitalics,assumingcleavageoccursbetweenA19
andA20,possiblybysignalpeptidaseI(SPaseI).N-terminalsignalpeptidesarerepresentedbyagreenrectangle.
B.
Characterizationof
lipo-NmMetQ.Size-exclusionchromatogramandmassspectraofpeak1.Themolecularmassesofthemajorspeciescorrespondwithin1Dato
thepredictedmassfortwotriacylatedNmMetQspecies,onewithacylchaincomposition[16:0,16:0,16:0](31,661Da)andtheotherwith[16:0,
16:0,18:1](31,685Da).
C.
CharacterizationofNmMetQC20A.Size-exclusionchromatogramandmassspectraofthemajorspeciesfrompeak2
andpeak3.Themolecularmassesofthemajorspeciesofpeak2and3correspondtothepre-proteinNmMetQ(32,802Da)andsecreted
NmMetQ(30,839Da),respectively.Thesemeasuredmassesarewithin3Daofthepredictedmassesforeachspecies.AssignedNmMetQ
speciesaredepictedincartoonformonthechromatograms.
Figure1–Figuresupplement1.
DLSandSEC-MALSmeasurementsofNmMetQproteins.
4of22
ManuscriptsubmittedtoeLife
analyzesaminoacidsequencestopredictthepresenceandlocationofcleavagesites(
Armenteros
138
etal.,2019
).Tovalidatethisprediction,weexpressedNmMetQusingan
E.coli
expressionsystem
139
with the native N-terminal signal sequence and a C-terminal decahistidine tag.
E. coli
has been
140
previously used to produce lipid-modified
N. meningitidis
proteins (
Fantappiè et al., 2017
). We
141
purified NmMetQ in the detergent n-dodecyl-
훽
-D-maltopyranoside (DDM) using an immobilized
142
nickel affinity column followed by size-exclusion chromatography (SEC). The SEC elution profile
143
showsonemainpeakwithanelutionvolumeof66mL(
Figure1
A).Ananalysisofthepeakfraction
144
byliquidchromatographymassspectrometry(LC/MS)revealedtwomajordeconvolutedmassesof
145
31,662and31,682Da(
Figure1
B).Thesemassescorrespondwellwiththetheoreticalmassesoftwo
146
lipoproteinNmMetQproteins: onewithatriacylchaincompositionof16:0,16:0and16:0(31,661
147
Da)andanotherwithatriacylchaincompositionof16:0,16:0and18:1(31,685Da), respectively
148
(
Figure1
A,top).Wecalculatedtheintactmassesofthelipo-NmMetQproteinsusingacombination
149
of16:0and18:1acylchainsbecausethesewerethemajorspeciesfoundinpreviousstudiesof
150
recombinantlyexpressedlipoproteins(
HantkeandBraun,1973
;
Luoetal.,2016
).
151
ToconfirmthatlipidattachmentsiteoccursattheN-terminalCys20onNmMetQ,wegenerated
152
aCys-to-AlaNmMetQmutant(NmMetQC20A).Wehypothesizedthatthismutationwouldprevent
153
lipidattachmentandleadtotheaccumulationofpre-proteinNmMetQ,containinganunprocessed
154
N-terminal signal sequence and the C20A mutation. The NmMetQC20A protein was expressed
155
andpurifiedinDDMaspreviouslydescribed.TheSECelutionprofilerevealstwomajorpeakswith
156
distinctelutionvolumes,78mland100mLforpeak1and2,respectively(
Figure1
C).Forpeak1,
157
analysisofthefractioncontainingthehighestpeakrevealedadeconvolutedmassof32,804,which
158
correlateswellwiththetheoreticalintactmassofthepre-proteinNmMetQ(32,802Da).Forpeak2,
159
thedeconvolutedmasswas30,840,whichagreeswiththetheoreticalintactmassofasecretedNm-
160
MetQproteincleavedbetweenAla19andAla20(30,839Da)(
Figure1
A,bottom).Theproductionof
161
thesecretedNmMetQwassurprisingsinceweonlyexpectedtheaccumulationofthepre-protein
162
NmMetQ.However,thesedatasuggestthattheCys-to-Alamutationcreatedanoncanonicalcleav-
163
agesite,possiblyallowingSpaseItoinefficientlycleavethepre-proteintoyieldsecretedNmMetQ.
164
Together,thesedataclearlydemonstratethatthemajorspeciesofrecombinantly-expressedNm-
165
MetQisheterogeneouslytriacylatedatCys20. MutatingCys20toAlapreventstheproduction
166
oflipoproteinNmMetQ,leadingtotheformationofpre-proteinNmMetQandsecretedNmMetQ.
167
Thelocationofcleavagesite,positionoflipidattachment,andhetergoenoustriacylchaincomposi-
168
tionofNmMetQinthisstudyareconsistentwithpreviousstudiescharacterizingotherlipoproteins
169
producedin
E.coli
(
Luoetal.,2016
;
Kwoketal.,2011
).
170
ThesedataalsorevealaninterestingpropertyofeachDDM-solubilizedNmMetQvariant:lipo-
171
NmMetQ, pre-protein lipo-NmMetQ, and secreted NmMetQ proteins elute at different volumes
172
despitetheirsimilarmolecularmasses(between31and33kDa). Specifically,lipo-NmMetQand
173
pre-protein NmMetQ proteins elute at much higher apparent mass than secreted NmMetQ on
174
a HiLoad 16/600 Superdex 200 (GE healthcare) column (
Figure 1
B,C). To further investigate the
175
propertiesoftheNmMetQproteins,weuseddynamiclightscattering(DLS)tomeasuretheirhy-
176
drodynamicradii(R
ℎ
)andcalculatetheirtheoreticalmolecularweightsassumingafoldedglobular
177
protein.WefoundthattheR
ℎ
valuesandmolecularweightestimateswerelargerforlipo-NmMetQ
178
(R
ℎ
=7.9
±
0.2nm, Mw-R=430
±
20kDa)andpre-proteinNmMetQ(R
ℎ
7.7
±
0.06nm, Mw-R=
179
400
±
7kDa)thanforsecretedNmMetQ(R
ℎ
3.0
±
0.013nm, Mw-R=43.0
±
0.3kDa)(
Figure 1
–
180
FigureSupplement1
). TheseproteinswerealsoanalyzedusingSizeExclusionChromatography
181
with Multi-Angle Light Scattering (SEC-MALS). For both lipo-NmMetQ and pre-protein NmMetQ,
182
estimatedmolarmasseswerelowerwithSEC-MALSwhencomparedtoDLS,withlipo-NmMetQ
183
measurementsof111
±
0.3versus430
±
20kDaandpre-proteinNmMetQmeasurementsof105
184
±
0.3versus400
±
7kDa,respectively.Thesedatasuggestthatbothlipo-NmMetQandpre-protein
185
NmMetQaggregateandthatthemassoftheaggregatedependsonthepreciseconditionofthe
186
experiment. However,molarmassesforthesecretedNmMetQproteinweremoresimilar(26
±
187
1versus43.0
±
0.3kDa),suggestingthatunlikelipo-NmMetQandpre-proteinNmMetQ,secreted
188
5of22
ManuscriptsubmittedtoeLife
NmMetQdoesnotassociatewithDDMmicelles.Basedonthesize-exclusionchromatograms,DLS,
189
andSEC-MALSdata,weproposethatbothlipo-NmMetQandpre-proteinNmMetQaggregatewith
190
DDMtoformprotein-DDMmicelle-likecomplexes.
191
TheATPaseactivityofNmMetNIismaximallystimulatedinthepresenceofboth
192
lipo-NmMetQandL-methionine
193
Figure2
Ashowsthatinthepresenceof1
휇
MNmMetNIalone(blacktrace)andinthepresence
194
of50
휇
ML-methionine(bluetrace),theATPaseactivitywaslow,demonstratingthatL-methionine
195
aloneisnotsufficienttostimulateNmMetNIATPaseactivity.However,inthepresenceofboth1
휇
M
196
lipo-NmMetQand50
휇
ML-methionine,amarkedstimulationofATPaseactivitywasobserved(
Fig-
197
ure2
A,greentrace).ToexcludethepossibilitythatthestimulationofATPaseactivityismediatedby
198
eitherthelipid-moietyortheunligandedNmMetQproteinsubunit,theexperimentwasrepeated
199
intheabsenceofL-methionine(NmMetNIandunligandedlipo-NmMetQonly)(
Figure2
A,magenta
200
trace). UndertheseconditionstheATPaseactivityislow,showingthatunligandedlipo-NmMetQ
201
isnotsufficienttostimulateNmMetNIactivity. Giventhesefindings,weconcludethatNmMetNI
202
ATPase activity is tightly coupled, requiring both L-methionine and lipo-NmMetQ for maximum
203
stimulation. Thisresultstronglysuggeststhatlipo-NmMetQplaysaroleinmethionine-mediated
204
NmMetNIATPhydrolysis.
205
Next,wecharacterizedtheeffectofdifferentNmMetQproteins(lipo-NmMetQ,pre-proteinNm-
206
MetQ,andsecretedNmMetQ)ontheATPaseactivityofNmMetNI.
Figure2
Bdemonstratesthatin
207
thepresenceof50
휇
ML-methionine,theNmMetNIATPaseactivityincreaseswithincreasingcon-
208
centrationoflipo-NmMetQupto2
휇
M,afterwhichtheactivitystartstoplateau(greentrace).The
209
sameprotocolwasperformedwithpre-proteinNmMetQ,whichcontainsanN-terminalsignalse-
210
quencebutwithoutthelipidmodification.Additionofpre-proteinNmMetQalsoledtostimulation
211
ofATPaseactivity,althoughtoalesserextentthanobservedforlipo-NmMetQ(orangetrace).Addi-
212
tionofsecretedNmMetQ,however,hadlittleeffectontheATPaseactivity(cyan).Together,these
213
dataestablishthatthelipidmoietyoflipo-NmMetQisrequiredformaximalNmMetNIstimulation,
214
althoughtheN-terminalsignalsequenceofpre-proteinNmMetQcouldpartiallymimicitsstimula-
215
Figure2.
ATPhydrolysisofNmMetNIinthepresenceandabsenceofL-methionineandNmMetQproteins.
A.
ATPhydrolysiswasmeasuredinthepresenceof1
휇
MofDDM-solubilizedNmMetNIalone(blacktrace),50
휇
M
L-methionine(bluetrace),1
휇
Mlipo-NmMetQ(magentatrace)andboth50
휇
ML-methionineand1
휇
M
lipo-NmMetQ(greentrace).Insertshowsrepresentativemeasurementsofabsorbanceversustime(black
dots)andthelinearfits(greenlines)forNmMetNIATPaseactivityinthepresenceoflipo-NmMetQand
L-methionineatincreasingATPconcentrations(0.2,0.4,0.8,1.2,1.6,2.0and4.0mM)
B.
Specificactivityof
NmMetNIwithincreasingconcentrationsofvariousNmMetQproteins:lipo-NmMetQ(greentrace),
pre-proteinNmMetQ(orangetrace),andsecretedNmMetQ(cyantrace)with50
휇
ML-methionine.Vmax
valuesweredeterminedbyfittingtheMichaelis-MentenequationtoaplotofATPaseactivityversusATP
concentration(0.2,0.4,0.8,1.2,1.6,2.0and4.0mM)atdifferentMetQproteinconcentrations(0.5,1,2,4,5,8
휇
M).N=3errorbarsrepresentstandarderrorofthemean(SEM).ThesedatashowtheNmMetNIATPase
activityistightlycoupled,requiringbothL-methionineandlipo-NmMetQformaximalNmMetNIATPase
stimulation.
6of22
ManuscriptsubmittedtoeLife
toryeffect.
216
AcomparisonofNmMetNI’sATPaseactivitywiththatofthepreviouslycharacterizedEcMetNI
217
revealsthatthesetransportershavedifferentligand-dependentATPaseactivities.WhenL-methionine
218
andSBPareabsent,NmMetNIhasnodetectablebasalATPactivity.HoweverEcMetNIhasabasal
219
ATPaserateof300nmolPi/min/mg(
Kadabaetal.,2008
).Thesetransportersalsodifferintheirre-
220
sponsetoL-methionine.InthepresenceofL-methionine,theATPaseactivityofEcMetNIdecreases
221
duetothebindingofL-methioninetotheC2domain,whichisresponsiblefortheregulatoryphe-
222
nomenonoftransinhibition. ForNmMetNI,however,nosucheffectwasdetected,asanticipated
223
fromtheabsenceoftheC2autoinhibitorydomaininNmMetNI.
224
AcomparisonofNmMetNISBP-dependentATPasestimulationtootherABCimportersalsore-
225
vealssomesimilaritiesanddifferences.ForNmMetNI,onlyliganded-SBPmaximallystimulatedNm-
226
MetNIATPhydrolysis.Maximalstimulationbyliganded-SBPsisalsoamechanisticfeatureshared
227
bytheABCimportersEcMalFGK
2
(
Davidsonetal.,1992
)andEcHisQMP
2
(
Amesetal.,1996
).Incon-
228
trast,fortheABCimporterEcYecSC-FliY(
Sabrialabedetal.,2020
),fullstimulationofATPasecan
229
beachievedinboththeliganded-SBPandtheunliganded-SBP.Althoughtheoriginofthesediffer-
230
encesareunclear,ourdatashowthatNmMetNIistightlycoupledandhighlightthemechanistic
231
differencesbetweenABCimporters.
232
N-formyl-L-methionine, L-norleucine, L-ethionine, and L-methionine sulfoximine
233
arepotentialsubstratesforthelipo-NmMetQ:NmMetNIsystem
234
ToidentifypotentialsubstratesoftheNmMetQ-lipoproteinMetQsystem,wedeterminedtherel-
235
ativebindingaffinitiesofseveralmethionineanalogstoNmMetQ.Forthesemeasurements,we
236
used
F
luorinechemicalshift
A
nisotropyande
X
changefor
S
creening(FAXS)incompetitionmode,
237
apowerfulsolutionNMRexperimentthatmonitorsthedisplacementofafluorine-containingre-
238
portermoleculebyacompetingligand.AnimportantfeatureofFAXSisthatfluorinemodification
239
ofthecompetingligandisnotrequired(
Dalvit et al.,2003
;
Dalvit and Vulpetti, 2018
). Asprevi-
240
ouslydiscussed(
Gerig,1994
;
DalvitandVulpetti,2018
),thefluorinenucleushasseveralproper-
241
tiesthatareadvantageousforNMR:
19
Fis100%abundant,possessesaspin1/2nucleus,andhas
242
high gyromagnetic ratio, which results in high sensitivity (83 % of
1
H). It also has a large chemi-
243
calshiftanisotropy(CSA),allowinghigherresponsivenesstochangeinmolecularweight,suchas
244
thosethatoccurduringaprotein-ligandbindingevent. Additionally,sincefluorineatomsarenot
245
presentinmostcommonlyusedbuffersystemsandvirtuallyabsentfromallnaturallyoccurring
246
biomolecules,backgroundinterferenceinfluorineNMRexperimentsisminimal.
247
TooptimizetheFAXSexperiment,weconsideredseveralfactors.Asshownin
Figure1
B,lipopro-
248
teinNmMetQmaymultimerize,possiblythroughanassociationwiththehydrophobicacylchains,
249
increasing its apparent molecular weight. Because FAXS is sensitive to the apparent molecular
250
weightoftheprotein,wechosetouseaNmMetQconstructlackingitsnativeN-terminalsignalse-
251
quenceandisthereforenotmodifiedwithlipids(referredtohereasNLM-NmMetQ).Trifluoromethyl-
252
methioninewasselectedasareportermoleculeandthefluorinesignalintensitywasmonitoredin
253
thepresenceofNLM-NmMetQandseveralmethionineanalogs(
Figure3
A).Forthesestudies,we
254
optimizedtheconcentrationofthereportermolecule,NLM-NmMetQ,andtherelaxationtime(T2)
255
fortheNMRmeasurement. Areportermoleculeconcentrationof2mMwaschosenheretode-
256
creaseacquisitiontime.Additionally,43
휇
MNLM-NMMetQwaschosenasacompromisebetween
257
using less protein and increasing the fraction of reporter bound to the protein. The relaxation
258
timeT2=120mswaschosenforitsabilitytostronglyattenuatebutnoteliminatethereporter
259
signalinthepresenceof43
휇
MNLM-NmMetQ.Aspreviouslydescribed(
Dalvitetal.,2003
;
Dalvit
260
and Vulpetti, 2018
), for all experiments, two fluorine spectra (1D and Car-Purcell-Meiboom-Gill
261
(CPMG)filtered)wereacquired. Theintensitysignalsofthereportermoleculemeasuredinboth
262
spectraandtheratio-ln(CPMG/1D)werecalculated.WeanticipatedthatanalogsthatbindtoNLM-
263
NmMetQwouldleadtothedisplacementofthereportermolecule,resultinginadecreaseinthe
264
-ln(CPMG/1D)ratio.
265
7of22
ManuscriptsubmittedtoeLife
Figure3.
CharacterizationoftheinteractionofmethionineanalogswithNmMetQusingFAXSandATPaseexperiments.
A.
Schematicdiagramof
theFAXSexperiment.TheintensityofthefluorinesignaldecreasesinthepresenceofNLM-NmMetQ.Additionofthemethionineanalogcauses
thefluorinesignalintensityofthereportermoleculetoincreaseduetoitsdisplacementfromNLM-NmMetQ.
B.
Chemicalstructuresofthe
methionineanalogsusedinthisstudy.
C.
(Top)OrderingofmethionineanalogsbytheirbindingaffinitytoNLM-NmMetQ.(Bottom)Schematic
representationofFAXSexperimentdepictedinbulksolution.Methionineanalogswithhigheraffinityarepositionedtowardtheleftsideofthe
plot,whileloweraffinitymethionineanalogsarepositionedtowardtheright.
D.
ATPaseactivityofNmMetNIat2mMATPinthepresenceof
lipo-NmMetQandmethionineanalogsat1:8:50molarratio,respectively.N=3errorbarsrepresentSEM.
Figure3–sourcedata1.
Themeasured-ln(Icpmg/I1D)values:NMR.xlsx
8of22
ManuscriptsubmittedtoeLife
Ourresultsforthecompetitionbindingexperimentsareshownin
Figure3
C.Theplotshowsthe
266
signalintensityratioofthereportermoleculeinthepresenceofeachmethionineanalog. Since
267
displacement of the reporter molecule by the analog correlates to the analog’s binding affinity,
268
methionineanalogswithhigheraffinitywillbepositionedtowardtheleftsideoftheplot, while
269
lower affinity methionine analogs will appear on the right side. As controls, we measured the -
270
ln(CPMG/1D)ratioswiththereportermoleculealoneandreportermoleculeplusNLM-NmMetQ.
271
Asexpected,thereportermoleculealonehasalow-ln(CPMG/1D)ratio,whilethereportermolecule
272
plusNLM-NmMetQhasahigh-ln(CPMG/1D)ratio(lessfreereportermoleculeduetoNLM-binding).
273
Next, we carried out the FAXS experiments in the presence of various methionine analogs.
274
WefirstaddedL-methionine,aknownhighaffinityligandofNmMetQ(Kd0.2nM(
Nguyenetal.,
275
2019
)). Asexpectedforahigheraffinityligand, L-methioninecompletelydisplacedthereporter
276
molecule. We then examined two methionine analogs with amino group substitutions: (1) N-
277
formyl-L-methionine,whichisusedbybacteriatoinitiatetranslationand(2)N-acetyl-L-methionine,
278
whichispresentinbacteria(
Schmidtetal.,2016
)andhumanbraincells(
Smithetal.,2011
)(
Fig-
279
ure3
C,circles).Additionoftheseanalogsledtotherespectivecompleteornearcompletedisplace-
280
mentofthereportermolecule, indicatingthatchangestotheaminogroupdonotdramatically
281
affectthesubstrate’sabilitytobindtightlytoNLM-NmMetQ.D-methioninedisplacedlessreporter
282
thanL-methionine,consistentwithitslowerbindingaffinity(3.5
휇
M(
Nguyenetal.,2019
)),while
283
N-acetyl-Dmethioninefailedtodisplacethereportermolecule.Theseresultssuggestthatmodifi-
284
cationstoboththeaminogroupandstereochemistryleadtosignificantlyweakerbindingthanthe
285
singlymodifiedderivative.
286
SimilartoourobservationswithD-methionineandN-acetyl-Dmethionine,changestothecar-
287
boxyl group resulted in less displacement of the reporter molecule than L-methionine. Specifi-
288
cally,L-methioninol,withthecarboxylgroupreducedtoanalcohol,failedtodisplacethereporter
289
moleculewhileL-methionineethylesteronlypartiallydisplacedthereportermolecule(
Figure3
C,
290
circles).
291
Lastly,changestotheL-methionineside-chainexhibitedvaryingeffects. Methionineanalogs
292
with changes to the sulfur atom, including seleno-L-methionine, L-methionine sulfoximine, and
293
L-norleucine were well tolerated, with a greater displacement of the reporter molecule than D-
294
methionine,whichhasanestimatedKdof3.5
휇
M(
Nguyenetal.,2019
).However,L-ornithinefailed
295
todisplacethereportermolecule,suggestingthatbindingofligandswithachargedaminogroup
296
isenergeticallyunfavorable. Side-chainlengthalsoplaysaroleinmethionineanalogaffinityto
297
NLM-NmMetQ.Increasingtheside-chainlengthbyanadditionofamethylenegroup(L-ethionine)
298
wasbettertoleratedthandecreasingthelengthbyonecarbon(S-methyl-L-cysteine).Shorterthiol
299
derivatives(L-cysteineandL-homocysteine)wereineffectiveatdisplacingthereportermolecule.
300
Together, our data establish that NLM-NmMetQ can accommodate variability in the binding of
301
methionineanalogs,includingmodificationstotheaminogroup,D-stereochemistry,andchanges
302
intheside-chain(toalimitedextent),whileexhibitinglittletoleranceforvariationsinthecarboxyl
303
group.
304
To determine whether methionine analogs could serve as potential substrates for the lipo-
305
NmMetQ:NmMetNIsystem,wemeasuredNmMetNIATPaseactivityinpresenceoflipo-NmMetQ
306
andseveralmethionineanalogs.Fortheseassays,wechosemethionineanalogsidentifiedbyFAXS
307
tobindNLM-NmMetQwithanaffinitysimilarorhigherthanD-methionine,aknownsubstratefor
E.
308
coli
NmMetNI.Sincesubstrate-stimulatedATPaseactivityisahallmarkofABCtransporters(
Bishop
309
etal.,1989
;
Mimmacketal.,1989
),weexpectedmethionineanalogsthataresubstratesforthis
310
systemwouldstimulateNmMetNIATPaseactivity.
Figure3
Dshowstheresultsforthemethionine
311
analogstimulationofNmMetNIATPaseactivity.Asanegativecontrol,wetestedL-cysteine,where,
312
asexpected,nosubstrate-stimulatedATPasestimulationwasdetected. Ourdatashowthatthe
313
followingmethionineanalogsledtosubstrate-stimulatedATPaseactivity: N-acetyl-L-methionine,
314
N-formyl-L-methionine,L-norleucine,L-ethionine,andL-methioninesulfoximine.However,nocor-
315
relationwasseenbetweenaffinitytoNLM-NmMetQandNmMetNIstimulation.Thisdatasuggest
316
9of22
ManuscriptsubmittedtoeLife
thatbindingtoNmMetQisnecessarytoinitiatetransport;however,thisstepalonedoesnotde-
317
terminethemagnitudeofNmMetNIATPasestimulation. Takentogether, ourFAXSandATPase
318
experimentssuggestthatN-formyl-L-methionine,L-norleucine,L-ethionine,andL-methioninesul-
319
foximinearepotentialsubstratesforthe
N.meningitidis
lipo-NmMetQ:NmMetNIsystem.
320
Structuresof
N.meningitidis
MetNIintheinward-facingconformationand
N.menin-
321
gitidis
lipo-NmMetQ:NmMetNIcomplexintheoutward-facingconformation
322
Togaininsightintotheroleoflipo-NmMetQintheNmMetNItransportcycle,wedeterminedstruc-
323
turesofNmMetNIindifferentconformationalstatesbysingle-particlecryo-EM.Byvaryingthenu-
324
cleotideanalogandconcentration, multipleconditionswerescreenedtoidentifyonesthatpro-
325
moteddifferentconformationalstates. With1mMAMPPNP(belowtheKm),thestructuralanaly-
326
sisofanequimolarmixtureoflipo-NmMetQandNmMetNIrevealedthatNmMetNIwascaptured
327
intheinward-facingconformationat3.3
Å
resolution(
Figure4
A);nodensitiesforeitherAMPPNP
328
andlipo-NmMetQwereobserved. Forthisdataset,thetwodimensionalclassaveragesshowed
329
clearstructuralfeatures,suggestingahighlevelofconformationalhomogeneity(
Figure4
–
Figure
330
Supplement2
).TheoverallarchitectureofNmMetNIissimilartopreviouslydeterminedstructures
331
ofEcMetNI,comprisingtwocopiesofeachTMDandNBD,encodedby
MetI
and
MetN
,respectively
332
(
Kadabaetal.,2008
;
Johnsonetal.,2012
).EachMetIsubunitcontainsfivetransmembranehelices
333
permonomerforatotaloftentransmembranehelicespertransporter(
Figure4
B).
334
AcomparisonbetweenNmMetNIandEcMetNIrevealssimilarsubunitfolds,withtherootmean
335
squaredeviation(RMSD)of2.4
Å
over843C
훼
atoms.Aspredictedfromtheprimarysequence,the
336
NmMetNsubunitslacktheC2autoinhibitorydomain. Asaresult,theinterfacesofNmMetNIand
337
EcMetNIaredistinct. Intheinward-facingconformationofNmMetNI,theNBDsinteractdirectly.
338
Incontrast,inEcMetNI,theinward-facingconformationformsaninterfacethroughtheC2autoin-
339
hibitorydomains,withaslightseparationbetweentheNBDs(
Figure4
–
FigureSupplement1
A).A
340
similarincreaseinNBD:NBDdistance,definedastheaveragedistancebetweenC
훼
ofglycinesinthe
341
Ploopandsignaturemotifs,isobservedinthepreviouslydeterminedmolybdateABCtransporter
342
structures,
Methanosarcinaacetivorans
ModBC(MaModBC)and
Archaeoglobusfulgidus
ModBC(Af-
343
ModBC)(
Hollensteinetal.,2007
;
Gerberetal.,2008
)(
Figure4
–
FigureSupplement1
B).Todate,
344
thesearetheonlyotherreportedpairofhomologousstructures,onewithanautoinhibitorydo-
345
mainandonewithout. ForAfModBC,whichlackstheautoinhibitorydomain, theNBD:NBDdis-
346
tancesare
∼
17
Å
and21
Å
foreachAfModBCintheasymmetricunit.ForMaModBC,whichdoes
347
haveanautoinhibitorydomain, thisdistanceincreasesto
∼
27
Å
. Acomparisonofthesestruc-
348
turessuggeststhattypeIABCimportersshareacommonquaternaryarrangementintheinward-
349
facingconformationsuchthatthepresenceofaregulatorydomainincreasestheseparationof
350
theNBD:NBDdistanceincomparisontothehomologousstructurewithoutaregulatorydomain
351
Figure4
–
FigureSupplement1
D.
352
ReasoningthatincreasingthenucleotideconcentrationtoabovetheKmforMgATPwouldpro-
353
motecomplexformation,wemixedequimolarlipo-NmMetQandNmMetQinthepresenceof5
354
mMATP.Undertheseconditions,wewereabletodeterminethesingle-particlecryo-EMstructure
355
ofthecomplexto6.4
Å
resolution(
Figure4
).Despiteextensiveefforts,wewereunabletoimprove
356
theresolutionofthiscomplex. Thestructurewasmodeledbyrigid-bodyrefinementofbothNm-
357
MetNIintheinward-facingconformation(tracedfromthe3.3
Å
resolutionreconstruction)andthe
358
previouslydeterminedsolubleNmMetQstructureinthesubstrate-freeconformation(PDB:6CVA).
359
Ourmodelshowslipo-NmMetQdockedontotheNmMetIsubunitsandtheNmMetNsubunitsina
360
closeddimerstate.Nocleardensitywasseenforthelipidmoietyoflipo-NmMetQorATP(
Figure4
).
361
Ofnote,Liuet.al.werealsonotabletoobservethelipidmoietyoflipo-SBPincomplexwithanABC
362
transporter,despitethesecryo-EMstructuresresolvingathigherresolution(3.30,3.44,3.78
Å
)
Liu
363
etal.
(
2020
).AcomparisonbetweenNmMetNIandEcMetNIintheoutward-facingconformationin
364
complexwiththeirrespectiveMetQproteinsrevealstheyhavesimilarconformations,withRMSD
365
of 2.2
Å
over 1048 C
훼
atoms (
Figure 4
–
Figure Supplement 1
C). In contrast to the inward-facing
366
10of22
ManuscriptsubmittedtoeLife
Table1.
Cryo-EMdatacollectionandrefinementstatistics.
Inward-facing confor-
mation of the MetNI
methionine ABC trans-
porter
Outward-facing confor-
mation of the MetNI
methionine ABC trans-
porter in complex with
lipo-MetQ
PDB
7MC0
7MBZ
EMD
EMD-23752
EMD-23751
Datacollectionconditions
Microscope
TitanKrios
TitanKrios
Camera
GatanK3Summit GatanK3Summit
Magnification
105,000x
105,000x
Voltage(kV)
300
300
Recordingmode
counting
counting
Frames/Movies
40
40
TotalElectrondose(e
−
/
Å
2
) 60
60
Defocusrange(
μ
m)
1.0–2.8
1.0-2.8
Pixelsize(
Å
)
0.856
0.856
Micrographscollected
4,709
6,183
Micrographsused
3,968
5,494
Totalextractedparticles 1,684,719
2,874,862
Refinedparticles
322,171
58,434
Symmetryimposed
C1
C1
NominalMapResolution(
Å
)
FSC0.143(unmasked/masked) 3.4/3.3
6.4/6.4
RefinementandValidation
Initialmodelused
3TUJ
Numberofatoms
Protein
7,092
8,987
Ligand
0
0
MapCC(mask/box)
0.80/0.65
0.75/0.69
MapsharpeningB-factor 91.3
496
R.m.s.deviations
Bondlengths(
Å
)
0.012
0.012
Bondangles()
1.62
1.92
MolProbityscore
1.76
1.73
Clashscore(allatom)
7.56
6.77
Rotameroutliers(%)
1.19
1.04
Ramachandranplot
Favored(%)
95.77
95.09
Allowed(%)
3.90
4.91
Outliers(%)
0.33
0
11of22