Structural investigation of heteroyohimbine alkaloid synthesis reveals active site elements that

14 Pages·2016·2.78 MB·English

by  Anna Stavrinides

ARTICLE
Received30Dec 2015 |Accepted 31May2016 |Published 15 Jul2016 DOI:10.1038/ncomms12116 OPEN
Structural investigation of heteroyohimbine alkaloid
synthesis reveals active site elements that control
stereoselectivity
Anna Stavrinides1,*, Evangelos C. Tatsis1,*, Lorenzo Caputi1, Emilien Foureau2, Clare E.M Stevenson1,
David M. Lawson1, Vincent Courdavault2 & Sarah E. O’Connor1
Plants produce an enormous array of biologically active metabolites, often with stereo-
chemical variations on the same molecular scaffold. These changes in stereochemistry
dramatically impact biological activity. Notably, the stereoisomers of the heteroyohimbine
alkaloids show diverse pharmacological activities. We reported a medium chain
dehydrogenase/reductase (MDR) from Catharanthus roseus that catalyses formation of a
heteroyohimbine isomer. Here we report the discovery of additional heteroyohimbine
synthases(HYSs),oneofwhichproducesamixtureofdiastereomers.Thecrystalstructures
for three HYSs have been solved, providing insight into the mechanism of reactivity and
stereoselectivity,withmutationofonelooptransformingproductspecificity.Localizationand
genesilencingexperimentsprovideabasisforunderstandingthefunctionoftheseenzymes
in vivo. This work sets the stage to explore how MDRs evolved to generate structural and
biological diversity in specialized plant metabolism and opens the possibility for metabolic
engineering of new compounds based on this scaffold.
1TheJohnInnesCentre,DepartmentofBiologicalChemistry,NorwichNR47UH,UK.2Universite´Franc¸ois-RabelaisdeTours,EA2106‘Biomole´culeset
BiotechnologiesVe´ge´tales’,Tours37200,France.*Theseauthorscontributedequallytothiswork.Correspondenceandrequestsformaterialsshouldbe
addressedtoV.C.(email:[email protected])ortoS.E.O’C.(email:[email protected]).
NATURECOMMUNICATIONS|7:12116|DOI:10.1038/ncomms12116|www.nature.com/naturecommunications 1
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms12116
H
eteroyohimbines are a prevalent subclass of the mono- theC.roseustranscriptome25,26basedonaminoacidsimilarityto
terpene indole alkaloids (Corynanthe type skeleton), this enzyme (Supplementary Table 1). Each of these candidates
having been isolated from many plant species, primarily was cloned from C. roseus cDNA and expressed in Escherichia
from the Apocynaceae and Rubiaceae families1. These coli, with the exception of Cr_017994, which could not be
alkaloids exhibit numerous biological activities: ajmalicine is an expressed and was not considered further. The remaining
a1-adrenergic receptor antagonist2–5, and mayumbine (19-epi- candidates were assayed with the substrate strictosidine
ajmalicine) is a ligand for the benzodiazepine receptor (Fig. 1)6. aglycone, and product formation was monitored by liquid
Oxidized beta-carboline heteroyohimbines also exhibit potent chromatography mass spectrometry (LC-MS). Of these, four
pharmacological activity: serpentine has shown topoisomerase (Cr_010119, Cr_021691, Cr_032583a, Cr_032583b) reduced
inhibitionactivity7andalstoninehasbeenshowntointeractwith strictosidine aglycone to a product corresponding to one of the
5-HT2A/C receptors and shows promise as an anti-psychotic heteroyohimbines(Fig.2,SupplementaryFig.1).Theproductsof
agent8–13. In addition, heteroyohimbines are biosynthetic theenzymaticreactionswereidentifiedbasedonLC-MSdataand
precursors of many oxindole alkaloids, which also display a comparison to authentic standards (Supplementary Fig. 2).
wide range of biological activities14. Although a total of 16 Enzymes that failed to produce a heteroyohimbine product
heteroyohimbine stereoisomers are possible, only 8 are reported were not studied further (Supplementary Fig. 1). Three of the
tobefoundinnature,atstereocentresC3,C19,C20(Fig.1)14–20. enzymes (Cr_021691, THAS2; Cr_010119, THAS3; Cr_032583a,
How and why the stereoselectivity is controlled in the THAS4) produced tetrahydroalstonine in B85% yield, with
biosynthesis of these alkaloids remains unclear. small amounts of 19-epi-ajmalicine (mayumbine) (o15%) also
The medicinal plant Catharanthus roseus produces three of observed in these reactions, similar to the previously reported
theseisomers,ajmalicine(raubasine),tetrahydroalstonineand19- THAS1. Notably, one enzyme (Cr_032583b, HYS) produced a
epi-ajmalicine (mayumbine) (Fig. 1)21. These heteroyohimbines, dramatically different product profile consisting of a mixture of
along with the majority of monoterpene indole alkaloids, ajmalicine/tetrahydroalstonine/mayumbine (55:27:15, at pH 6)
are derived from deglycosylated strictosidine (strictosidine (Fig. 2). The discovery of this enzyme, HYS, now provides a
aglycone)22. The removal of a glucose unit from strictosidine by molecular basis to understand the generation of stereochemical
strictosidine glucosidase (SGD) forms a reactive dialdehyde diversity in this alkaloid family.
intermediate that can rearrange to form numerous isomers23.
Thestabilizationoftheseisomersbyenzyme-catalyzedreduction
Crystallography of three HYSs. To understand the mechanism
is hypothesized to be the stepping stone for the extensive
of stereochemical control at this crucial biosynthetic branch
chemical diversity observed in the monoterpene indole alkaloids
point, we crystallized three HYSs. THAS1 and THAS2, which
(Fig. 1)21,22. We recently reported the first cloning of a
produce predominantly tetrahydroalstonine, were both crystal-
biosynthetic gene encoding an enzyme that acts on strictosidine
lized, since their amino acid sequence identity is relatively low
aglycone. This zinc-dependent medium chain dehydrogenase/
(55%) and the predicted active sites have numerous differences
reductase (MDR), named tetrahydroalstonine synthase (THAS),
(Fig.3).HYS,whichhasadistinctlydifferentproductprofile,was
produces the heteroyohimbine tetrahydroalstonine (Fig. 1)24.
alsocrystallizedtoexplorethestructuralbasisbehindthisdistinct
AlthoughthesestudiesdemonstratedthatTHASisakeyenzyme
stereochemical outcome. Structures (Supplementary Table 3)
in heteroyohimbine biosynthesis, the mechanism by which this
were obtained for THAS1 and THAS2 with NADPþ bound
enzyme controls the stereoselectivity of the reduction remained
(THAS1, 1.05Å resolution (Fig. 4a,b; Supplementary Figs 3–5);
unknown. Moreover, it is important to note that strictosidine
THAS2,2.10Åresolution(Fig.4d,SupplementaryFig.4))andin
aglyconeservesastheprecursorformanyalkaloidscaffolds,and
apo form (THAS1, 2.25Å resolution (Fig. 4c, Supplementary
therefore represents a central branch point in the monoterpene
Fig.5);THAS2,2.05Åresolution(Fig.4c)),whileHYScouldonly
indole alkaloid biosynthetic pathway. Therefore, we set out to
be crystallized in the apo form (2.25Å resolution, Fig. 4c).
identify additional heteroyohimbine synthases (HYSs) with
different stereochemical product profiles that would more
clearly define how structural diversity, in this case the StructuralfeaturesofHYSactivesites.ThefiveHYSstructures
formationofdifferentstereoisomers,iscontrolledinthissystem. described here are similar to sinapyl alcohol dehydrogenase
In this study, we assayed 14 MDR homologues identified from (SAD; PDB accession codes 1YQX and 1YQD) and the SAD
the C. roseus transcriptome25,26 that have homology to THAS homologue cinnamyl alcohol dehydrogenase (CAD; PDB acces-
(Cr_024553). This screen revealed three additional enzymes with sion codes 2CF5 and 2CF6)27–29, which reduce the aldehyde
THAS activity (Cr_010119, Cr_021691, Cr_032583a), and, moiety of lignin precursors. Indeed, pairwise superpositions of
importantly, an enzyme that produced a mixture of hetero- subunits from these structures gave RMSD values of o2Å
yohimbine diastereomers (Cr_032583b). Crystal structures of (Supplementary Table 4). The biological unit is an elongated
THAS (here referred to as THAS1), a second representative homodimer, with each subunit divided into a substrate and
THAS(Cr_021691,THAS2)andthestructureofthepromiscuous cofactor-binding domain; the latter also being responsible for
homologue(Cr_032583b,HYS)weresolvedandmutantsrevealed forming the dimer interface (Supplementary Fig. 3). The overall
key residues that control the stereochemistry of the product structure of THAS1, with active site, cofactor and loops
profiles.Notably,analysisofthesubcellularlocalizationofsomeof highlighted, is shown in Supplementary Fig. 3.
theseHYSsindicatesanunusualnuclearlocalizationpatternandan The activesitecavitiesofthe HYSs are framed by helix a2,the
interaction with the previous enzyme, SGD. These discoveries catalytic zinc coordination sphere, and loops 1 and 2, with the
provide insight into the mechanism and evolution of a crucial NADP(H)co-substratebindingatthebaseoftheactivesite(Figs3
branch point in a specialized metabolic pathway with both and 4a,b,d, , Supplementary Fig. 4). Loop 2, which is positioned
pharmacological and evolutionaryimportance. above the active site, is highly variable in length and sequence
(Figs3and4d).InbothTHAS1andTHAS2,anetworkofamino
acids holds NADPþ in place. Most notably, Glu59 of THAS1
Results anchors NADP(H) through a bidentate interaction with both
Discovery of HYSs. Guided by our initial discovery of THAS1 ribose hydroxyls, with His59 playing a comparable role in SAD,
(ref.24)weidentifiedcandidatesfromtheMDRproteinfamilyin although here the interaction is with the 30 OH only (Fig. 4b).
2 NATURECOMMUNICATIONS|7:12116|DOI:10.1038/ncomms12116|www.nature.com/naturecommunications
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms12116
Oxidized heteroyohimbine Oxidized heteroyohimbine
alstonine serpentine
+ +
N H N H N H
N H O N
H H
O O
H N O H H
MeO C H MeO C MeO C
2 2 2
Oxindole
mitraphylline
Tetrahydroalstonine Ajmalicine
3(S), 20(S), 19(S) 3(S), 20(R), 19(S)
N H N H
N H N H
H H
O O
H H
Reduction MeO2C Reduction MeO2C
+ +
21
N H 3 N N H
20
N H N N H
H H H 19 H
O O O
MeO C MeO C MeO C
2 2 2
Pro-tetrahydroalstonine Cathenamine Pro-ajmalicine
NH
O-Glc +
N H N
H O SGD N H
H
MeO C
2
Strictosidine MeO2C OH
(precursor for ~ 2000 Strictosidine aglycone
monoterpene indole alkaloids)
+ +
N H 3 N 21 N H
20
N H N N H
H H H 19 H
O O O
MeO C MeO C MeO C
2 2 2
Epi-cathenamine
Reduction Reduction
N H
N H N H
N H H
H H O
O
H MeO C
2
MeO2C Mayumbine/19-epi-ajmalicine
Rauniticine 3(S), 20(R), 19(R)
Figure1|Heteroyohimbinealkaloidbiosynthesis.Heteroyohimbineswith3(S)stereochemistryderivefromstrictosidineaglycone.Thethree
diastereomersfoundinCatharanthusroseus,arehighlightedwithredarrows.Alkaloidsderivedfromheteroyohimbinesarealsoshown.
Glu59isconserved inHYS, butan aspartate residue (THAS3 and THAS1 is B2Å further away from the cofactor relative to SAD
THAS2) or a tyrosine residue (THAS4) serves this role in other and thus may play no direct role in catalysis (Supplementary
homologues. MDRs usually contain two zinc ions30, a distal Fig.4).However,itmayhaveafunctioninmaintainingthetertiary
‘structural’ zinc ion, which in this case is coordinated by four structuresincethreeoftheligandingresiduesareinthesubstrate-
cysteineresidues,andaproximal‘catalytic’zincionneartheactive bindingdomainandthefourthisinthecofactor-bindingdomain.
site, which iscoordinated by two cysteines, one histidine and one SAD/CADutilizeanactivesiteserinethatprotonatesthealkoxide
glutamate residue (Figs 3 and 4b)27,31. The proximal zinc of that results from reduction of the aldehyde substrate27,29; this
NATURECOMMUNICATIONS|7:12116|DOI:10.1038/ncomms12116|www.nature.com/naturecommunications 3
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms12116
Ajmalicine position of strictosidine aglycone in THAS1 (Fig. 4b). To ensure
that the correct substrate tautomer was used for docking, we
Tetrahydroalstonine
identified the most predominant strictosidine aglycone isomer
Mayumbine thatformsinsolution.Althoughproductprecipitationprevented
monitoring the SGD reaction in situ under aqueous conditions
(Methods),1H,15N-HMBCNMRshowedthatanenaminespecies
was the predominant product in aqueous methanolic solution
(Supplementary Fig. 6). This is consistent with literature reports
that cathenamine is the major product of SGD, and is the
proposed precursor of ajmalicine and tetrahydroalstonine
(Fig.1)21,23.InsilicodockingwithTHAS1positionscathenamine
between the nicotinamide of the NADPþ and Tyr56, which is
located on helix a2 (Fig. 4b). THAS1 loop 1 contains Phe65
that also projects into the active site and may interact with the
aromatic cathenamine substrate.
HYS
Mechanism of reduction and heteroyohimbine formation. In
tetrahydroalstonine biosynthesis, we hypothesize that cathe-
naminetautomerizestotheiminiumformbyprotonationatC20,
THAS4 followed by addition of the hydride at C21. Protonation at C20
must occur from the bottom face to yield the S stereochemistry
observedatthisposition(Fig.5a).Whiletheredoesnotappearto
beanappropriatelypositionedactivesiteresiduetoperformthis
role,thecrystalstructuresoftheseenzymesrevealthepresenceof
numerouswatermoleculesintheactivesitethatcouldpotentially
protonatethiscarbon(Fig.4a).1H,15N-HMBCmeasurementsof
THAS3 strictosidine aglycone at different pH values show formation of
the iminium species in solution when the pH was reduced to
B3.5, indicating that this tautomer can readily form in the
presence of an acidic moiety (Supplementary Fig. 6).
THAS2
To elucidate the stereo and regioselectivity of reduction by
NADPH, we isolated tetrahydroalstonine from reactions using
THAS1 and pro-R-NADPD. Analysis by 1H-NMR showed that
tetrahydroalstonine is labelled with deuterium in the pro-R
position at C21,consistent withpreviously reported experiments
performed in crude cell extracts (Fig. 6a)32. It is possible that
THAS1couldreducetheenaminedirectly,inwhichcasehydride
additionwouldoccuratC21,followedbyprotonationatC20bya
watermoleculeasdescribedabove.Thepresenceofmayumbine/
THAS1 19-epi-ajmalicine in some of the enzymatic reactions
suggests that small amounts of cathenamine can open and form
Cr033062 19-epi-cathenamine, either in solution or in the active site.
In the case of HYS, which produces both ajmalicine (R C20)
and tetrahydroalstonine (S C20), protonation must also occur
8.00 9.00 10.00 11.00 12.00 13.00 14.00
Time (min) fromtheoppositefacetoyieldRstereochemistryatC20(Fig.5b).
Products of HYS generated with pro-R-NADPD were also
Figure2|LC-MSanalysisofactiveMDRcandidatesagainststrictosidine isolated and analysed by 1H-NMR, and, as for tetrahydroalsto-
aglycone.Cr033062exhibitedonlytraceactivity.SeeSupplementaryFig.1 ninefromTHAS1ineachcaseshoweddeuteriumlabellinginthe
forchromatogramsofassayswithinactiveenzymesandnegative pro-R position at C21 (Fig. 6a–c). Therefore, the stereochemical
controls. course of hydride addition is not altered in HYS compared with
THAS1.
serine is replaced with a tyrosine residue in THAS1 (Tyr56) and The major difference between HYS and THAS1/THAS2
HYS (Tyr53) (Fig. 3). In THAS2, this tyrosine on helix a2 is appears to be the extended loop over the HYS active site
replacedwith a tryptophan residue,but a tyrosineat position 120 (D125-GHFGNN-F132 in HYS and D128-SN-Y131 in THAS1,
points into the active site. Interestingly, a non-proline cis-peptide loop 2 in Fig. 3). The histidine residue in HYS loop 2 (His127)
is present in the THAS1-NADPþ and HYS apo structures appears to be positioned appropriately to provide an alternative
(Supplementary Fig. 5). Closer inspection of this region in protonsourcefortheopposite(‘top’)faceofthesubstrate,which
THAS1 shows that when this bond is in the trans conformation, couldexplaintheappearanceofajmalicineintheproductprofile
thesidechainofAsp340isprojectedintothecofactor-bindingsite of HYS. Reactions with THAS1 and HYS performed at different
such that itwould prevent NADPH binding. pH conditions (5–8) revealed that while changes in pH did not
substantially impact the product profile of THAS1, HYS
produced increased amounts of ajmalicine relative to tetrahy-
Strictosidine aglycone binding. Despite extensive efforts, both droalstonine at pH 6 compared with higher pH values
product and substrate failed to co-crystallize with any of the (Supplementary Fig. 7). The increased level of ajmalicine in
enzymes. Therefore, molecular docking was used to visualize the HYS at lower pH values is consistent with the pKa value of the
4 NATURECOMMUNICATIONS|7:12116|DOI:10.1038/ncomms12116|www.nature.com/naturecommunications
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms12116
α1 β1 β2 β3 α2 β4
HYS TT TT TT TT
1 10 20 30 40 50 60 70 80 90
HYS
THAS2
THAS1
SAD
Loop1
β5 β6 α3 η1 β7 β8 η2 β9 α4 η3 α5
HYS TT TT TT
100 110 120 130 140 150 160 170
HYS
THAS2
THAS1
SAD
Loop2
β10 α6 β11 α7 β12 α8 β13 α9
HYS TT
180 190 200 210 220 230 240 250 260
HYS
THAS2
THAS1
SAD
NLS
β14 β15 α10 β16 α11 β17 η4 α12 β18 α13
HYS
270 280 290 300 310 320 330 340 350
HYS
THAS2
THAS1
SAD
Figure3|SequencealignmentofCatharanthusroseusHYSenzymesandPopulustremuloidesSADNumberingcorrespondstoHYS.Identicalandsimilar
aminoacidsarehighlightedinredandyellow,respectively.SecondarystructureelementsoftheHYSapocrystalstructurearedisplayed.THAS1-andHYS-
activesiteaminoacids(Y56/53andE59/56)areindicatedbybluedots,andTHAS2activesiteaminoacids(Y120andD49)areindicatedbygreendots.
Ligandsforcatalyticandstructuralzincionsarehighlightedbyblackandgreydots,respectively.Thenuclearlocalizationsignalof(THAS1andHYS)and
loops1and2,respectively,areindicatedinred.Anon-prolinecis-peptidebondthatisobservedinTHAS1holo,inonesubunitofTHAS1apo
(SupplementaryFig.5),inHYSapo,andnotatallinTHAS2isindicatedwithanorangedot.Thesubstrate-bindingdomainandthecofactor-bindingdomain
areindicatedbyblueandpurplebars,respectively.
histidine side chain and supports the role of this histidine in stereoselectivity and relative activity of the mutant enzymes
ajmalicine biosynthesis, though attributing pH dependence to (Supplementary Table 6). However, the k (observed)
cat
specific residues must be approached with caution33. values could be measured for THAS1 (1.518±0.059s(cid:2)1),
THAS2(0.033±0.001s(cid:2)1),THAS3(0.102±0.004s(cid:2)1),THAS4
(0.044±0.006s(cid:2)1), HYS (0.083±0.005s(cid:2)1), HYS_loop2 swap
Switching stereoselectivity of HYSs. Since the major sequence (1.970±0.153s(cid:2)1), THAS1 Y56S (0.118±0.005s(cid:2)1) and
and structural difference between HYS and THAS1 is the THAS1 E59A (0.061±0.005s(cid:2)1).
extended loop over the HYS active site, loop 2 (Fig. 3), we
swapped these loop regions in THAS1 and HYS to determine
whether the stereochemical product profile could be switched In planta localization of HYSs. Plants use spatial organization
(Fig. 7a). Loop1,whichisnear theactive site,was alsoswapped on the organ, tissue and intracellular levels to control product
(Fig. 7a). While the THAS1 mutant containing the swaps distribution (Supplementary Fig. 12). Atthesubcellular level, we
displayed reduced activity rather than an altered product profile previously showed that THAS1 has an unusual nuclear localiza-
(Fig. 7c), the HYS mutant containing the shorter THAS1 loop 2 tion pattern24, which is also where SGD, the enzyme that
resulted in a product profile similar to that of THAS1 (Fig. 7b, synthesizes strictosidine aglycone, is localized34. Physical
SupplementaryFig.8).SinceHis127istheonlyionizableresidue interactions using Bimolecular Fluorescence Complementation
inthisloop,wehypothesizedthatthisresidueprotonatesC20,as (BiFC)wereobservedbetweenthesetwoenzymes24.THAS2and
discussed above. Therefore, we mutated this histidine to alanine HYSlocalizationweresimilarlyinvestigatedbyexpressingyellow
or asparagine in HYS. Both of these mutants gave the same fluorescentprotein(YFP)fusionsinC.roseuscells.Microscopyof
THAS-like profile, suggesting that His127 is required for pro- transiently transformed cells revealed that THAS2–YFP(Fig. 8a)
ducing the ajmalicine (R C20) stereochemistry (Supplementary was located in both the cytosol and the nucleus while HYS–YFP
Fig. 9). Mutation of other conserved ionizable residues (Fig. 8e), similar to THAS1, displayed a preferential nuclear
in the THAS1 active site (Tyr56, Ser102 and Thr166) did not localization (Fig. 8 and Supplementary Fig. 13). As reported for
result in substantial changes in the distribution of products THAS1, this localization relies on the presence of a class V
(Supplementary Table 6, Supplementary Figs 10 and 11). Muta- nuclear localization sequence in HYS (215-KKKR-218) that is
tionstoGlu59,whichanchorstheNADPHcofactor,resultedina absent from THAS2 (Fig. 3). BiFC assays revealed that both
slight increase in product promiscuity (Supplementary Fig. 11), THAS2andHYSarecapableofself-interactions(Supplementary
perhapsbycausingashiftinthecofactorposition.Thereactivity Fig. 14).
of the substrate34 and precipitation at high concentrations BiFCassayswereusedtodeterminewhetherTHAS2andHYS
during the assay makes obtaining accurate kinetic constants also interact with SGD (Fig. 9). C-terminal split-YFP fragment
challenging24, so end-point assays were used to assess fusions of both enzymes (THAS2–YFPC and HYS–YFPC) were
NATURECOMMUNICATIONS|7:12116|DOI:10.1038/ncomms12116|www.nature.com/naturecommunications 5
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms12116
a b
LLLLLLLoooooooooooopp1111 LLLLLLLLLLLLLoooooooooooooooooooooooooooppppppppppp22222222
LLoooopp2222
LLLLLLLLLLLLLLoooooooooooooooooopppppp1111111111
CCCCCCCCCCaaaaaaaaattttttttthhhhhhhhhhheeeeeeennnnnnnaaaaaaaammmmmiinnee CCCCCCCCCCCCCCCaaaaaaaaaaaaaatttttttttttttttttthhhhhhhhhhhhhheeeeeeeeeeeeeennnnnnnnnnnnnnaaaaaaaaaaaaammmmmmmmmmmiiiiiiiiinnnnnneeeeeeeeeee
c d
Loop2
Loop2
NADP+
Figure4|CrystalstructuresofheteroyohimbinesynthasesTHAS1,THAS2andHYS.(a)Sampleofautomaticallyderivedexperimentallyphased
electrondensityfromTHAS1(at1.12Åresolution)superimposedonthefinalmodelshowingtheactivesiteregionwiththeNADPþ cofactor(green
carbons)togetherwithneighbouringresidues(magnoliacarbons)andwatermolecules(smallredspheres).(b)THAS1dockedwithcathenamine(pale
bluecarbons)withtheproteinshowninbothcartoon(left)andspacefilling(right)modes.TheNADPþcofactorisshownwithgreencarbons;loop1isin
orangeandloop2isincyan.Zincionsaredisplayedasmagentaspheres.Theactivesiteislargelycontainedwithinasinglesubunit(magnoliasurface),
althoughthemouthofthechannelleadingtotheactivesiteispartiallyboundedbythesecondsubunitofthebiologicaldimer(greysurface)(c)
SuperpositionoftheapostructuresofTHAS1(gold),THAS2(pink)andHYS(blue).(d)Superpositionoftheholo(NADPþ containing)structuresof
THAS1(gold)andTHAS2(pink),withthecofactorofTHAS1shownasvanderWaalsspheresforemphasis.Forcandd,thestructuresweresuperposed
ontotheTHAS1structure,basedontheuppersubunitalone;onlypartofthelowersubunitoftheTHAS1structureisshowningreyforreference(see
SupplementaryFig.3forimagesofthefullTHAS1dimer).Theinsetsemphasizethedifferinglengthsofloop2betweenthevariousstructures;thecentral
portionofloop2inapoTHAS2wasdisordered.
co-transformed with SGD that was fused to a N-terminal split- recovered,confirmingthespecificityofTHAS1–,THAS2–,HYS–
YFP fragment (YFPN–SGD). The formation of a nuclear BiFC SGD interactions (Fig. 9o,p).
complexsuggeststhatbothoftheseenzymesinteractwithSGDin
the nucleus (Fig. 9a–d). Interestingly, the emitted fluorescent
signal exhibited a punctated, sickle-shaped aspect as previously In planta silencing of HYSs. The expression levels of the HYS
observedfortheTHAS1/SGDinteraction(Fig.9e,f)andforSGD transcripts do not suggest whether a specific HYS is the most
localization34. In contrast, no interactions were detected when biologically relevant (Supplementary Fig. 12). To establish
the BiFC assay was conducted with a downstream enzyme whetheranyoftheseenzymessynthesizetheexpectedmetabolic
from this biosynthetic pathway, 16-hydroxytabersonine 16-O- productinplantathegenesencodingactiveHYSsweresilenced.
methyltransferase(16OMT),thatisnotexpectedtointeractwith For many medicinal plants, including C. roseus, virus-induced
SGD (Fig. 9g,h). gene silencing (VIGS) is the only established method to silence
Double BiFC assays were performed to combine the study of genes in the whole plant. In C. roseus, the effect of VIGS is
THAS2 and HYS interactions, as well as their interactions with temporally and spatially limited to the first two leaves that
SGD. After transformation into plant cells (16h), we noted the emerge immediately after infection35. Each of the genes
formation of a dual fluorescent signal for THAS2, both in encoding a biochemically active enzyme (THAS1, THAS2,
thecytosol and as punctates in thenucleus that may correspond THAS3, THAS4 and HYS) was subjected to VIGS in C. roseus
to the superposition of the signal observed for THAS2 self- seedlings and the effect on alkaloid production was monitored
interactionsandTHAS2–SGDinteraction(Fig.9i,j)asconfirmed by MS. Since HYS and THAS4 were too similar to silence
by multicolour BiFC (mBIFC) assays (Supplementary Fig. 15). separately, one common gene fragment was used for silencing
Increased time of expression (36h) progressively resulted in the both genes simultaneously. Successful silencing of the genes
disappearance of the cytosolic signal, and it is intriguing to was confirmed by quantitative reverse transcription–PCR
speculate that this implies a recruitment of THAS2 by SGD (qRT–PCR) (Supplementary Fig. 16). Aside from a small
(Fig. 9q,r). A similar phenomenon was observed for HYS and degree of cross-silencing between THAS2 and THAS3 (12%),
THAS1 (Fig. 9k–n,s–v- Supplementary Fig. 15). While self- all of the target genes were silenced selectively, as measured by
interactions of 16OMT were detected, no nuclear signal was qRT–PCR (Supplementary Fig. 16).
6 NATURECOMMUNICATIONS|7:12116|DOI:10.1038/ncomms12116|www.nature.com/naturecommunications
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms12116
a
Tetrahydroalstonine biosynthesis mechanism in THAS and HYS
Tyr 56
Tyr 56 Tyr 56 Tyr 56
O
H O Pro-tetrahydroalstonine Oδ– HO
R1H HNH R1HHHNH iminium form R1HHHNH R1H HNH
O N O S N+ O Nδ+ O RN Tetrahydroalstonine
CaHth3CeHnaOmHinHeSHRNCRO2NH2 H3C HHH NCORN2H2 H3C HHH NCORN2H2 H3C H H N+CORN2H2
NADPH NADPH NADPH NADPH
R1: -COOCH3
R2: Adenine dinucleotide phosphate
b
Ajmalicine biosynthesis mechanism in HYS
His127 Tyr 53 His127 Tyr 53 His127 Tyr 53 His127 Tyr 53
H H
O O
H N
HO-H N NH H O N H O Pimroi-naiujmma flircoimne-HO H N N H Oδ– H2O N NH HO
R1H HNH R1HH HNH R1HHHNH R1H HNH
O N O O Ajmalicine
O N R + δ+ R
CathenHam3Cine HHR CONH2 H3CH HH NCONH2 H3C HHH NCONH2 H3C H H NCONH2
S N R2 R2 R2 + R2
NADPH NADPH NADPH R1: -COOCH3 NADPH
R2: Adenine dinucleotide phosphate
Figure5|Mechanistichypothesisforheteroyohimbinesynthases.(a)Proposedmechanismforformationofthetetrahydroalstonine(SC20)
diastereomer.(b)Proposedmechanismofformationoftheajmalicine(RC20)diastereomerthatisobservedinHYS,whichcontainsahistidineresidue
neartheactivesite.
Due to the inherent reactivity of the HYS substrate enzymes make it difficult to draw firm conclusions from these
(strictosidine aglycone), changes in the level of this compound VIGSdata.Inaddition,regulatoryfactorsthatimpacttheratioof
in planta are difficult to accurately detect. Instead, the effect of ajmalicine and tetrahydroalstonine, such as transport
silencing must be established by quantitatively measuring mechanisms and/or further derivatization to other products,
decreases in heteroyohimbine levels. Previously for THAS1, we cannot be ruled out. Additional silencing systems, in different
measured the combined peak for heteroyohimbines, since the tissues,willberequiredtomorefirmlyestablishthephysiological
diastereomers were difficult to resolve on a reverse phase LC function of these enzymes. Nevertheless, we can state that
column under the reported conditions24. However, after silencing of HYS and THAS1 impacts alkaloid production in C.
substantial optimization (see Methods), an LC-MS method was roseus leaves.
developedtoseparateajmalicineandtetrahydroalstonineincrude
leaf extracts (19-epi-ajamlicine/mayumbine is not observed in
C. roseus leaves, Supplementary Fig. 17). Ajmalicine and Discussion
particularly tetrahydroalstonine are observed in low levels even Here we report several medium chain dehydrogenases/
in empty vector control samples compared with the major leaf reductases that produce the heteroyohimbine stereoisomers
monoterpene indole alkaloids vindoline and catharanthine, and ajmalicine and/or tetrahydroalstonine, thereby providing a
in addition, heteroyohimbine composition varied substantially framework to understand the enzymatic control over stereo-
among individual leaves. Therefore, accurate measurement of selectivity in this metabolic pathway. It is notable that we have
decreases in ajmalicine and tetrahydroalstonine levels is identified four enzymes that generate tetrahydroalstonine, yet
challenging. There was no evidence for a decrease of ajmalicine ajmalicine is the more abundant isomer in planta
or tetrahydroalstonine when THAS2 and THAS3 were silenced. (Supplementary Fig. 17). Expression profile data of the genes
However, for HYS, there was a statistically significant decrease identified in this study suggest that HYS, which produces
(t-test 0.0275) in ajmalicine, and no change in tetra- ajmalicine, is not expressed at higher levels than the other
hydroalstonine levels. Surprisingly, a statistically significant synthases (Supplementary Fig. 12). There may be additional
decrease in ajmalicine, as well as tetrahydroalstonine (t-test ajmalicine synthases that are not related to the MDR super-
0.0277 and 0.0276, respectively) was also noted for THAS1 family homologues identified in this study. Alternatively,
(Supplementary Fig. 16). While the results are statistically tetrahydroalstonine could be shuttled into another pathway or
significant, the leaf-to-leaf variability, the low level of degraded, thereby resulting in the observed lower levels that
endogenous production, and the catalytic redundancy of these accumulate in planta.
NATURECOMMUNICATIONS|7:12116|DOI:10.1038/ncomms12116|www.nature.com/naturecommunications 7
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms12116
a thoughthesideeffectscausedbyalmitrineresultedinwidespread
(1) 1H-NMR [21α-2H]-tetrahydroalstonine withdrawalofthedrugin2013(ref.4).Whiletetrahydroalstonine
H-21β has no reported pharmacological function, its oxidized product,
alstonine (Fig. 1), has recently been shown to act by a unique
mechanismformodulatingdopamineuptakeandshowspotential
asananti-psychoticdrug13.Theheteroyohimbineshaveexcellent
H-20 promiseasascaffoldforpharmacologicalactivity.Thediscovery
of the HYSs, along with recently developed heterologous
production platforms for monoterpene indole alkaloids36, now
allowsthepossibilityofgenerating thesealkaloidsand unnatural
derivatives through metabolic engineering/synthetic biology
(2) 1H-NMR tetrahydroalstonine
H-21α strategies.
The crystal structures of three HYSs reveal the potential of
H-21β biosynthetic machinery to generate stereochemical variation.
Flexible loop regions can be the key to unlocking chemical
H-20
diversity: as we have demonstrated here, mutating the extended
loop over the HYS active site (loop 2 in Fig. 3) impacts
stereochemical outcome. Notably, the MDRs that we have
3.1 3.0 2.9 2.8 2.7 2.6 2.5 1.7 1.6 p.p.m. identified demonstrate high variability at this region (Fig. 3).
Phylogeneticanalysis(SupplementaryFig.18)suggeststhatthese
b HYSs,whichappeartohaveoriginatedfromacommonancestor,
(1) 1H-NMR [21α-2H]-ajmalicine may have undergone neo-functionalization through mutation in
this loop region. This loop could potentially be harnessed in
H-21β protein engineering efforts to generate novel catalytic activity.
Whiletheinplantafunctionofheteroyohimbinesisunknown,
H-20
deglycosylated strictosidine is toxic and may act as a defense
compound34, similar to the defense roles of the aglycones of the
iridoidsfromwhichstictosidineisderived37,38.SGDisexpressed
inmosttissues(SupplementaryFig.12),suggestingthattheplant
must have evolved mechanisms to control the levels of the toxic
(2) 1H-NMR ajmalicine H-21α
H-21β strictosidine aglycone. In directed overflow metabolism, excess
reactive intermediates are converted into non-reactive
H-20 byproducts39. It is intriguing to speculate that monoterpene
indole alkaloid biosynthesis may have initially arisen as a
mechanism for handling overflow of strictosidine aglycone. The
HYSs perform a single, chemically straightforward reduction
3.0 2.9 2.8 2.7 2.6 2.5 2.4 2.3 2.2 p.p.m. reaction that immediately neutralizes the reactivity of
strictosidine aglycone/cathenamine. The co-localization and
c
interaction of THAS1, THAS2 and HYS with SGD reinforces
(1) 1H-NMR [21α-2H]-mayumbine
the hypothesis of an evolutionary mechanism deployed by
H-21β
strictosidine-accumulating plants to manage the reactivity of
thestrictosidineaglycone.Italsoraisesthequestionofapossible
H-20 competitionbetweenHYSsforrecruitmentbySGDwhendistinct
enzymes are co-expressed in the same tissue/cells. Whether the
heteroyohimbinesserveanactivebiologicalfunctionintheplant,
or whether they are simply the end product of directed overflow
metabolism, or both, remains to be investigated. Regardless, it is
clear that MDRs play an important role in the generation of a
(2) 1H-NMR mayumbine
wide variety of chemical structures. Duplication of the
H-21β
evolutionary dehydrogenase ancestor may have given rise to
H-21α
multiple HYSs, along with MDRs with other biosynthetic
H-20 activities, such as tabersonine-3-reductase that is involved in
the biosynthesis of the anti-cancer alkaloid vinblastine
(Supplementary Fig. 18)40.
3.2 3.1 3.0 2.9 2.8 2.7 2.6 2.5 2.4 2.3 2.2 2.1 2.0 1.9 1.8 p.p.m.
Methods
Figure6|DeuteriumlabellingofTHAS1andHYSproductsusingpro-R- SelectionandcloningofcandidateMDRs. Thenucleotideandtheprotein
NADPD.Comparisonofselectedregionsof1H-NMRspectraoflabelled sequencesofTHAS1weresubjectedtoaBLASTsearchagainsttheC.roseus
(a)tetrahydroalstonine,(b)ajmalicine,(c)mayumbine.Thespectra SunstormApricotV1.0Transcriptsequences(http://medicinalplantgen-
omics.msu.edu)andtheMDRsequenceswiththehighestidentitytoTHAS1atthe
indicatethatC21islabelledwithdeuteriuminthepro-Rposition.
activesiteandwhichshowednon-negligibleexpressionlevelsinyoungandmature
leaveswereselectedascandidatesforcloningandexpression.Theproteinsequence
ofSADwasblastedagainstthesamedatabaseandMDRswerealsoselectedbased
Importantly,thepharmacologicalactivityofheteroyohimbines ontheiractivesitesimilaritytothatofSAD.ThegenescodingthecandidateMDRs
wereamplifiedfromC.roseusleafcDNAandclonedintotheE.coliexpression
isimpactedbythestereochemistry.Ajmalicinehasrecently been vectorpOPINFusingtheIn-Fusioncloningkit(ClontechTakara)41byusing
used in combination with almitrine in post-stroke treatments, primersdesignedbasedonthetranscriptsequences(SupplementaryTable1).
8 NATURECOMMUNICATIONS|7:12116|DOI:10.1038/ncomms12116|www.nature.com/naturecommunications
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms12116
a b c
Loop1 THAS1 HYS
Loop2 HYS loop1+2 swap
THAS1 loop1 swap
THAS1 loop1+2 swap HYS loop2 swap
Tyr THAS1 loop2 swap HYS loop1 swap
8.00 9.00 10.00 11.00 12.00 13.00 14.00 8.00 9.00 10.00 11.00 12.00 13.00 14.00
Time Time
Figure7|ProductprofilesofTHAS1andHYSloopswapmutants.(a)ShownistheapoTHAS1structure(magnolia)withloops1and2highlightedin
orangeandcyan,respectively.Forclarity,onlythecorrespondingloopsoftheHYSapostructureareshowninyellowaftersuperposition.Similarly,onlythe
cofactorfromthesuperposedholoTHAS1structureisshownforreference(greencarbons).Thesidechainsofimportantresiduesarealsoshown.
(b)RepresentativeLC-MSchromatogramsofassayswithTHAS1mutantsinwhichloop1,loop2orbothhavebeenswappedwiththecorresponding
sequencesfromHYS.(c)RepresentativeLC-MSchromatogramsofassayswithHYSmutantsinwhichloop1,loop2orbothhavebeenswappedwiththe
correspondingsequencesfromTHAS1.
a THAS2–YEP b Nucleus–CFP c Merge d DIC
e HYS–YFP f Nucleus–CFP g Merge h DIC
i YFP j Nucleus–CFP k Merge l DIC
Figure8|THAS2displaysnucleocytosoliclocalizationwhileHYSispreferentiallytargetedtothenucleus.C.roseuscellsweretransiently
co-transformedwithplasmidsexpressingeitherTHAS2–YFP(a),HYS–YFP(e)orYFP(i)andtheplasmidencodingthenuclear-CFPmarker(b,f,j).
Co-localizationofthefluorescencesignalsappearsinyellowwhenmergingthetwoindividual(green/red)falsecolourimages(c,g,k).Cellmorphologyis
observedwithdifferentialinterferencecontrast(DIC)(d,h,l).Scalebars,10mm.
Site-directedmutagenesisofTHAS1andHYS.THAS1mutantsweregenerated (H127ACAT-4GCA;H127NCAT-4AAC;F128ATTT-4GCT;F128YTTT-
byoverlapextensionPCR.Briefly,thecodontobemutatedwasselectedandtwo 4TAC)andthepOPINFoverhangsincludedatthe30 and50extremities.The
primers,onereverseandoneforward(SupplementaryTable4),weredesignedto THAS1andHYSdoubleloopmutantsweregeneratedbyfirstmakingtheirloop1
overlapandintroducethemutation.AfirstPCRwascarriedoutusingthereverse mutantgenesandtheninsertingthesecondloop2swapfollowingthesame
mutantprimerandthe50 forwardgene-specificprimer(SupplementaryTable1), proceduredescribedabove.Mutantconstructsweresequencedtoverifythemutant
thusgeneratingthe50halfofthegenecarryingthemutation.Inparallel,the30half genesequenceandcorrectinsertion.
ofthemutatedgenewasgeneratedbyPCRusingtheforwardmutantprimerand
the30 reversegene-specificprimer(SupplementaryTable4).PCRproductswere
gelpurifiedandusedforthesecondPCRoverlapreactionforgenerationofthefull- Enzymeactivityassays. Allcandidateenzymesandmutantswereexpressed
lengthmutatedgene,wherethe50and30halvesofthemutatedgenesweremixedin inSoluBL21(DE3)E.colicells(Genlantis)grownin2(cid:3)YTmedium.Protein
aPCRreactioninequimolaramounts(B100ngperfragment)andfivecyclesof productionwasinducedbyadditionof0.2mMIPTGandthecultureswereshaken
PCRwerecarriedoutwithoutincludingprimers.Afterthe5overlapPCRcycles, at18(cid:2)Cfor16h.Cellswerecollectedbycentrifugation,lysedbysonicationin
theforwardandreversegene-specificprimerswereaddedtothemixandfurther30 BufferA(50mMTris-HClpH8,50mMglycine,500mMNaCl,5%v/vglycerol,
cycleswereperformed.Full-lengthPCRproductsweregelpurified,ligatedinto 20mMimidazole)supplementedwithEDTA-freeproteaseinhibitor(Roche
pOPINFexpressionvectorandtransformedintocompetentE.coliStellarstrain Diagnostics)and0.2mgml(cid:2)1lysozyme.Solubleproteinswerepurifiedon
cells(ClontechTakara).HYSpointmutantswereobtainedasgenefragments Ni-NTAagarose(Qiagen)andelutedwithBufferB(50mMTris-HClpH8,50mM
(IntegratedDNATechnologies,Belgium)withtheH127orF128codonsmutated glycine,500mMNaCl,5%v/vglycerol,500mMimidazole).Eluateswereanalysed
NATURECOMMUNICATIONS|7:12116|DOI:10.1038/ncomms12116|www.nature.com/naturecommunications 9
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms12116
a THAS2–YFPC b DIC c HYS–YFPC d DIC e THAS1–YFPC f DIC g 16OMT–YFPC h DIC
YFPN–SGD YFPN–SGD YFPN–SGD YFPN–SGD
i THAS2–YFPN j DIC k HYS–YFPN l DIC m THAS1–YFPN n DIC o 16OMT–YFPN p DIC
YFPC–THAS2 YFPC–HYS YFPC–THAS1 16OMT–YFPC
YFPN–SGD YFPN–SGD YFPN–SGD YFPN–SGD
q THAS2–YFPN r DIC s HYS–YFPN t DIC u THAS1–YFPN v DIC w 16OMT–YFPN x DIC
YFPC–THAS2 YFPC–HYS YFPC–THAS1 16OMT–YFPC
YFPN–SGD YFPN–SGD YFPN–SGD YFPN–SGD
Figure9|THAS2andHYSinteractwithSGDinthenucleus.THAS2/SGD(a,i,q)andHYS/SGD(c,k,s)interactionswereanalysedbyBiFCinC.roseus
cellstransientlytransformedbydistinctcombinationsofplasmidsencodingfusionswiththetwosplitYFPfragments,asindicatedoneachfluorescence
picture.THAS1/SGD(e,m,u)and16OMT/SGD(g,o,w)interactionswerestudiedtoevaluatethespecificityofTHAS2/SGDandHYS/SGDinteractions.
SingleBiFCassaysshowinginteractionswithSGD(upperrow)anddoubleBiFCassayshighlightingbothinteractionswithSGDandTHAS2,HYS,THAS1,
16OMTself-interactionswereconductedandobserved16h(middlerow)and36h(lowerrow)post-transformation.Cellmorphologyisobservedwith
differentialinterferencecontrast(DIC)(b,d,f,h,j,l,n,p,r,t,v,x).Scalebars,10mm.
bySDS–PAGEtoverifythepurityandthemolecularweightofthepurified Crystallizationscreenswereconductedbysitting-dropvapourdiffusionin
proteins.AllproteinsweredialysedinBufferC(50mMphosphatepH7.6,100mM MRC296-wellcrystallizationplates(Swissci)withamixtureof0.3mlwellsolution
NaCl)andconcentrated.ProteinconcentrationwasmeasuredwithBradford fromthePEGs(Qiagen),PACT(Qiagen)andJCSG(MolecularDimensions)suites
reagent(Sigma-Aldrich)accordingtothemanufacturer’sinstructions.Purified and0.3mlproteinsolution.Proteinconcentrationswereadjustedto7–10mgml(cid:2)1
proteinsweredividedin20mlaliquots,fast-frozeninliquidnitrogenandstoredat whileNADPþ (Sigma-Aldrich)wasaddedtoafinalconcentrationof1mMfor
(cid:2)20(cid:2)C. co-crystallizationstudies.SolutionsweredispensedeitherbyanOryxNanooran
CandidateMDRenzymesandtheselectedmutantswerescreenedforactivity Oryx8robot(DouglasInstruments).
againstdeglycosylatedstrictosidine.Thesubstratewasgeneratedbydeglycosylating THAS1apocrystalswereobtainedfromHis6-tagcleavedTHAS1(3Cprotease)
strictosidine(300mM)bytheadditionofpurifiedSGDinthepresenceof50mM inasolutioncontaining0.1MMES,pH6.5,15%w/vPEG2000.THAS1-NADPþ
phosphatebuffer(pH6.5)atroomtemperaturefor10min.Thereactionswere crystalswereobtainedfromasolutioncontaining0.2Mpotassium/sodiumtartrate
startedbytheadditionofMDRenzyme(1mM)andNADPH(5mM).Caffeine with20%w/vPEG3350.THAS2crystals(withandwithoutNADPþ)were
(50mM)wasusedasinternalstandard.Allreactionswereperformedintriplicate. obtainedfromaconditioncontaining0.2Mlithiumchlorideand20%w/vPEG
Aliquotsofthereactionmixtures(10ml)weresampled1and30minafteraddition 3350.HYScrystalswereobtainedafterremovaloftheHis6-tag(using3Cprotease)
ofMDRenzyme.Thereactionswerestoppedbytheadditionof10mlof100% in0.1MMMTbuffer,pH5and15%w/vPEG3350.Allcrystalswere
MeOH.Sampleswerediluted1:5inmobilephase(HOþ0.1%formicacid)and cryoprotectedbysoakingincrystallizationsolutioncontaining25%v/vethylene
2
centrifugedfor10minat4,000gbeforeUPLC-MSinjection(1ml).Theactivityof glycolbeforeflash-coolinginliquidnitrogen.
MDRenzymesandmutantswasmeasuredbyUPLC-MS.Enzymesexhibitedaloss
ofactivityafteronefreezethawcycle.
Theinitialvelocitywasdeterminedforthewild-typeenzymesthatdisplayed Datacollectionandstructuredetermination. X-raydatasetswererecordedon
HYSactivity(THAS1-4andHYS),aswellasforthemutantsTHAS1Y56S,E59A oneofthreebeamlinesattheDiamondLightSource(Oxfordshire,UK)(2FI3,I04;
andHYSloop2swap.Reactionsweremonitoredat340nmatroomtemperature 2FI5,I03;5H81I04-1;5H82,I04-1;5H83,I04-1)atwavelengthsof0.9000–0.976Å
usinga(Cary50Bio,Varian)spectrophotometer.Strictosidine(50mM)was (2FI3,0.900Å;2FI5,0.976Å;5H81,0.920Å;5H82,0.920Å;5H83,0.920Å)using
incubatedwithpurifiedSGDin50mMphosphatebuffer(pH7.0)forB30minat eitheraPilatus6Mor2Mdetector(Dectris)withthecrystalsmaintainedat100K
30(cid:2)C.NADPH(100mM)wasadded,mixedbypipettingandtheabsorbanceat byaCryojetcryocooler(OxfordInstruments).Diffractiondatawereintegrated
340nmwasmonitoreduntilstabilized(minimumof2min).Apredetermined usingXDS42andscaledandmergedusingAIMLESS43viatheXIA2expert
amountofenzyme(10–400nM)wasthenadded,mixedandthereactionwas system44;datacollectionstatisticsaresummarizedinSupplementaryTable3.
monitoredforaminimumof5minat340nm.Theresultingslopewascalculated InitiallytheTHAS1-NADPþdatasetwasautomaticallyprocessedatthebeamline
duringthelinearreactionrange,usually0–60safteradditionoftheenzyme.The byfast_dp45to1.12Åresolutionandastructuresolutionwasautomatically
reactionswerereplicatedfivetimesandtheinitialvelocitywascalculatedforeach obtainedbysinglewavelengthanomalousdispersionphasingusingtheSHELX
replicateafteraccountingforthebackground.Theconcentrationofsubstrate suite46viathefast_eppipeline(Winter,manuscriptinpreparation).Despitebeing
(50mMstrictosidine)wasdeterminedtobesaturatingforallwild-typeenzymes collectedatawavelengthsomewhatremotefromthezincKX-rayabsorptionedge
undertheseconditions. (theoreticalwavelength1.284Å),theanomaloussignalwassufficientforfast_epto
locatefourzincsitesandcalculateaveryclearexperimentallyphasedelectron
densitymap(Fig.4a).ThiswasavailabletoviewatthebeamlineintheISPyB
Proteincrystallization.Proteinsforcrystallizationwerepurifiedfrom2lcultures database47viatheSynchWebinterface48withinafewminutesofcompleting
in2(cid:3)YTmedium.ProteinexpressionwasinducedbyadditionofIPTGandthe thedatacollection.Themapwasofsufficientqualitytoenable94%ofthe
culturesweregrownfor16hat18(cid:2)C.Cellswerecollectedbycentrifugationand residuesexpectedforaTHAS1homodimertobeautomaticallyfittedusing
lysedbysonicationin50mlBufferAsupplementedwithEDTA-freeprotease BUCCANEER49.ThemodelwasfinalizedbymanualrebuildinginCOOT50and
inhibitorand10mgofLysozyme.Lysateswereclarifiedbycentrifugationat restrainedrefinementusinganisotropicthermalparametersinREFMAC5(ref.51)
17,000gfor20min.Two-dimensionalautomatedpurificationwasperformedonan againstthesamedatasetreprocessedtoaresolutionof1.05Åasdescribedabove
AKTAxpresspurifier(GEHealthcare).TheIMACstepwasperformedonHisTrap (SupplementaryTable3),andcontained97%oftheexpectedresidues,withone
HP5mlcolumns(GEHealthcare)equilibratedwithBufferA.Proteinswerestep- NADPþmoleculeandtwozincionspersubunit.Alltheremainingstructureswere
elutedwithBufferBanddirectlyinjectedonagelfiltrationcolumnequilibrated solvedbymolecularreplacementusingPHASER.52Ineachcase,theasymmetric
withBufferD(20mMHEPES,150mMNaCl,pH7.5).Fractionswerecollected unitcorrespondedtothebiologicaldimerandthepreliminarymodelswere
andanalysedbySDS–PAGEandthosecontainingpureproteinwerepooledand obtainedbysearchingfortwocopiesofamonomertemplate.ForTHAS1apo,
concentratedina10kDamembranefilter—Milliporefilter(MerckMillipore). THAS2NADPþ andHYSapo,aTHAS1-NADPþ protein-onlymonomermodel
PurificationofHYSrequiredtheadditionof1mMDTTtoallpurification wasusedasthebasisforthetemplate,althoughinthelattertwocasesahomology
buffersanddialysisinBufferDcontaining0.5mMtris(2-carboxyethyl)phosphine modelofthetargetstructurewasgeneratedfromtheTHAS1templateusingthe
(TCEP)beforecrystallizationandstorage. Phyre2server53(http://www.sbg.bio.ic.ac.uk/Bphyre2)beforerunningPHASER.
10 NATURECOMMUNICATIONS|7:12116|DOI:10.1038/ncomms12116|www.nature.com/naturecommunications