Divergent synthesis and identification of the cellular targets of deoxyelephantopins

Abstract

Herbal extracts containing sesquiterpene lactones have been extensively used in traditional medicine and are known to be rich in α,β-unsaturated functionalities that can covalently engage target proteins. Here we report synthetic methodologies to access analogues of deoxyelephantopin, a sesquiterpene lactone with anticancer properties. Using alkyne-tagged cellular probes and quantitative proteomics analysis, we identified several cellular targets of deoxyelephantopin. We further demonstrate that deoxyelephantopin antagonizes PPARγ activity in situ via covalent engagement of a cysteine residue in the zinc-finger motif of this nuclear receptor.

ARTICLE

Received 11 Jan 2016 | Accepted 23 Jun 2016 | Published 19 Aug 2016

Herbal extracts containing sesquiterpene lactones have been extensively used in traditional medicine and are known to be rich in a,b-unsaturated functionalities that can covalently engage target proteins. Here we report synthetic methodologies to access analogues of deoxyelephantopin, a sesquiterpene lactone with anticancer properties. Using alkyne-tagged cellular probes and quantitative proteomics analysis, we identied several cellular targets of deoxyelephantopin. We further demonstrate that deoxyelephantopin antagonizes PPARg activity in situ via covalent engagement of a cysteine residue in the zinc-nger motif of this nuclear receptor.

DOI: 10.1038/ncomms12470 OPEN

Divergent synthesis and identication of the cellular targets of deoxyelephantopins

Roman Lagoutte1,*, Christelle Serba1,*, Daniel Abegg1,*, Dominic G. Hoch1, Alexander Adibekian1

& Nicolas Winssinger1

1 Department of Organic Chemistry, School of Chemistry and Biochemistry, NCCR Chemical Biology, University of Geneva, 30 quai Ernest-Ansermet, Geneva 1211, Switzerland. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to A.A. (email: mailto:[e-mail]

Web End =[e-mail] ) or to N.W. (email: mailto:[e-mail]

Web End =[e-mail] ).

NATURE COMMUNICATIONS | 7:12470 | DOI: 10.1038/ncomms12470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12470

The contribution of natural products to our current pharmacopeia and to the identication of important therapeutic targets is well recognized1,2. While natural

products are the result of a long evolutionary optimization, a number of examples have demonstrated that synthetic modications beyond the biosynthetically accessible analogues can bring about important pharmacological improvements. Success stories starting with the semisynthetic derivatization of 6-aminopenicillanic acid to enhance b-lactam activity, to the conversion of erythromycin into azithromycin or baccatin III into taxotere have inspired tremendous efforts in natural product synthesis. While a signicant portion of bioactive natural products are endowed with reactive functionalities that can engage in covalent interactions with their target, the historic reluctance to develop covalent inhibitor has curtailed interest in this subset of natural products. In a number of cases, these mildly reactive groups are pivotal to the compounds bioactivity. Despite the potential for promiscuous covalent engagement through unspecic reactions, a number of covalent inhibitors display useful selectivity with regards to their targeted protein3,4 by virtue of the fact that at low inhibitor concentration (mM), the kinetics of unspecic reaction are slow compared with the reaction resulting from a specic inhibitortarget interaction (that is, high effective concentration of reagents). The preponderance of such reactive groups amongst secondary metabolites would suggest that there is an evolutionary advantage to covalent inhibition. For instance, a covalent inhibitor may also be important in displacing an otherwise unfavourable equilibrium with an endogenous ligand5. The declining pipeline of traditional small-molecule drugs coupled to the benet of covalent binding to overcome resistance/selectivity

issues in kinase inhibition, or efcacy in protease inhibition, have led to a recent reconsideration of covalent inhibitors68. Natural products have played a key role in the drug-discovery process and as probes in chemical biology9. This privileged role has inspired many efforts to access natural-product-like libraries by conventional or diversity-oriented synthesis1013. Terpenoids and sesquiterpene lactones certainly stand out for their historical use in medicine and are rich in mildly reactive functionalities that can engage in a covalent interactions14,15. Indeed, functional groups such as a-methylene-g-butyrolactone, a,b-unsaturated reactive ester chain and epoxides are preponderant in this natural product family and are at the source of its rich biological activity1618. For example, both helenalin (Fig. 1) and parthenolide inhibit the NF-kB pathway by covalently inactivating their target19. In the case of helenalin, this inhibition has been proposed to result from a covalent crosslinking of two cysteines in p65. Helenalin is broadly used as an anti-inammatory drug in the form of its natural extract from Arnica. Thapsigargin is widely used in cellular biology and covalently inhibits SERCA (Sarco/endoplasmic reticulum Ca2 ATPase)20. Arglabin inhibits protein farnesylation without affecting protein geranylation. On the basis of the critical role of farnesylation for H-Ras function (an important oncogenic driver), this compound has been shown to be an effective antitumour agent21 and a dimethyl amine prodrug of this natural compound is currently used therapeutically. Most recently, a-methylene-g-butyrolactones also showed promising antibacterial activity by covalently binding to critical transcriptional regulators and inhibiting the virulence of Staphylococcus aureus22.

Extracts of the plant Elephantopus scaber have long been used in traditional medicine with deoxyelephantopin being the most

Selected bioactive sesquiterpene lactones

O

H

O

O

2

O OH

HelenalinNF-B inhibitor O

OH OH

Thapsigargin SERCA inhibitor

O

Arglabin

O

O

O

O

O

O

O

O

O

O

O

O

Deoxyelephantopin

O

Isodeoxyelephantopin

Elephantopin

Carbonylation

OH

I

Br

Barbier reaction

RCM

O

Figure 1 | Bioactive covalent sesquiterpenes and retrosynthetic analysis of deoxyelephantopins. Selected examples of bioactive sesquiterpene interacting covalently with their target; structure and retrosynthetic analysis of deoxyelephantopins.

2 NATURE COMMUNICATIONS | 7:12470 | DOI: 10.1038/ncomms12470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12470 ARTICLE

active component23. Recently, deoxyelephantopin has been shown to be more effective than paclitaxel in suppressing tumour growth and metastasis in a murine orthotopic breast cancer model24. At the cellular level, deoxyelephantopin has been shown to be cytotoxic at doses of 0.52 mg ml 1 in several human cancer cell lines. While there is evidence that deoxyelephantopin inhibits the NF-kB pathway24,25, proteomics analysis of up- and downregulated proteome in treated cells suggested it also suppressed proteasome activity26. Moreover, SPR experiments suggest that deoxyelephantopin can act as a partial agonist of PPARg27, a nuclear receptor that is well known to be involved in pathologies of obesity, diabetes and atherosclerosis and thus represents a major pharmacological target. However, it is not clear whether this natural product can also engage PPARg directly in cells. While these activities could be rationalized by diverse covalent target engagement, a proteome-wide identication of direct cellular targets of deoxyelephantopin has not been performed to date. The impressive in vitro and in vivo activities reported for deoxyelephantopin coupled to its historical use as a traditional remedy demands a better understanding of its reactivity prole in a cellular setting and covalent protein target(s). Perhaps, due to its abundance from natural extracts, there is no total synthesis of deoxyelephantopin reported to date nor structure-activity relationship for this promising therapeutic. Here we report synthetic methodologies to access analogues of deoxyelephantopins, including alkyne-tagged cellular probes for quantitative proteomics analysis. We identied several cellular targets of deoxyelephantopin and demonstrated that deoxyelephantopin antagonizes PPARg activity through covalent engagement.

ResultsSynthesis of deoxyelephantopins and analogues. The synthesis of deoxyelephantopin analogues was envisioned to proceed as shown in Fig. 1 making use of a Barbier reaction (conditions are known for syn28 or anti29 addition products) and a ring-closing metathesis (RCM) to join two readily available fragments (1 and 2). Mindful that there are few precedents for the RCM of strained 10-membered rings30 with a triply substituted alkene, we reasoned that the strategy might provide a rapid entry into the hitherto unknown nordeoxyelephantopin and its analogues.

The synthesis commenced with the alkylation of butenal 3 with lithiated alkyne 4 affording the alcohol required in the subsequent hydroxyl-directed alkyne conversion to vinyl iodide 5 (Fig. 2). Nickel-catalysed carbonylation31 yielded the rst key intermediate 6. The second fragment 2 was assembled from the pentadienol 7. Acylation with acryloyl chloride followed by a BaylisHillman reaction with formaldehyde provided the cyclization substrate 8 that was engaged with Grubbs II followed by a conversion of the allylic alcohol to the bromide under Appel conditions to yield 2 as a racemate. While the prochiral substrate 8 might be desymmetrized via enantioselective RCM with the newly developed ruthenium-based catalysts3234, we anticipated to separate diastereomers arising from this racemic fragment at a latter stage. Conversion of the dimethyl acetal 6 to the corresponding aldehyde using hydrated iron trichloride followed by Barbier coupling with allylic bromide 2 afforded the addition product 9 with 49:1 relative stereochemistry for the two newly established stereocenters at C-7 and C-8 and a 1:1 diastereomeric mixture based on the relative stereochemistry of both fragments. With the intermediate in hand, we proceeded to experiment with the RCM. All attempts to carry out a RCM directly on 9 failed using various catalysts. However, introduction

of the methacrylate side chain (10) provided an intermediate that did afford nordeoxyelephantopin in the RCM with the desired E-alkene geometry. Interestingly, the RCM proceeded only under the action of rst generation catalyst (Grubbs I). Furthermore, only the diastereomer corresponding to deoxyelephantopin relative stereochemistry afforded the cyclization product, the C-2 epimer failed to give cyclization. A high correlation between nordeoxyelephantopin and deoxyelephantopin NMR coupling constants suggests that the two products have very similar dihedral angles along the 10-membered ring and hence, a comparable conformation. Furthermore, comparison of NOESY spectra showed similar interactions between the protons on either face of the 10-membered ring for both compounds (Supplementary Figs 1 and 2). With these ring-closing conditions in hand, the same strategy was pursued to access analogues wherein one of the exocyclic conjugate acceptors was reduced (11 and 12 respectively). While the same reaction starting with 3b afforded the cyclization precursor (not shown), no cyclization product was observed under a variety of metathesis conditions.

To control the stereochemistry at C-2, we rst explored recent chemistry for enantioselective alkyne addition35, however, substrate 3a proved problematic based on its propensity to isomerize under Lewis acidic conditions. As an alternative, substrate 13 was converted to the furan 14, which underwent a palladium-catalysed decarboxylative asymmetric allylic alkylation (DAAA)36,37 using Trosts ligand to afford 6 in either stereochemistry with 92:8 er.38 Enantiomerically enriched (R)-6 was used to obtain nordeoxyelephantopin in four steps while (S)-6 (obtained with the (R,R)-DACH-phenyl catalystnot shown) afforded the ent-nordeoxyelephantopin.

We reasoned that performing the cyclization before the carbonylation used to form the endocyclic lactone might relax the conformational bias and change the outcome of the cyclization. To this end, substrate 15 (Fig. 3) was prepared according to the same methodology as shown in Fig. 1. Concomitant silylation of the C2 hydroxyl and deprotection of the THP under the action of TESOTf39 followed by DessMartin periodinane oxidation afforded the desired aldehyde that was engaged in the Barbier coupling to yield 16. Introduction of the methacrylate side chain, or an acetate, followed by exposure to Grubbs II yielded the 10-member cyclized product however, exclusively as the Z-alkene (17a and 17b, respectively). As for the previous examples, only one of the C-2 epimers underwent RCM. Silyl deprotection under acidic conditions and nickel-mediated carbonylation afforded 19a and b, the Z-analogue of norisodeoxyelephantopin and its C-8 acetate analogue. Structural analysis revealed a close proximity of the allylic C-6 hydrogen to the p-orbital of C-1 conjugated double bond. Exposure of this compound to 254 nm light afforded 20 quantitatively (for related photochemical transformation in sesquiterpene lactones; see ref 40). This transformation presumably proceeds through an excitation of the C1 C10 double bond resulting in

C-6 hydrogen abstraction and C-6 C-10 bond formation.

To the best of our knowledge, this ring system is unprecedented. Alternatively, intermediate 17a was engaged in a palladium-catalysed coupling that resulted, after acidic deprotection, in an aromatization affording 18.

Preliminary experiments on the reactivity of the three conjugate systems of deoxyelephantopin revealed that the g-butyrolactone was most reactive (reaction with 5 equivalent of glutathione led to a single addition product onto the g-butyrolactone, see

Supplementary Figs 35 and Supplementary Methods for conditions). The endocyclic conjugate system proved unreactive. On the basis of this observation, we used the Barbier methodologies to prepare a set of minimal analogues tagged with an alkyne (23a-c and 24ac, Fig. 4). In addition, pentenal 26 and

NATURE COMMUNICATIONS | 7:12470 | DOI: 10.1038/ncomms12470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12470

OMe

O

O Li OMe

a.

b. LiAlH4, THF

4

OH I

c.

Ni(CO)2(PPh3)2

Et2O, 78 C, 45%

Et3N, PhMe, reflux

31%

OMe

R

reflux, then I2, 73%

OMe

5 6

3a: R = H 3b: R = Me

COCI, Et3N,

CH2Cl2, r.t., 95%

h. Grubbs II (20 mol%)

i.

f.

FeCl3-H2O

CH2Cl2, reflux, 75%

d.

e.

CH2Cl2, reflux

Zn, THF,

OH O

O

g.

aq. HCHO, DABCO dioxane, r.t., 43%

CBr4, PPh3

CH2Cl2, r.t., 88%

Br

aq. NH4Cl, r.t

7

2

36% over two steps

8

O

O

O

O

O

O

2

O

Grubbs I (20 mol%)

32% b.r.s.m.

CH2Cl2, reflux

k.

DMAP, Et3N CH2Cl2, 0 C

62%

O

O

OH

()-10

O

()-9

()-Nordeoxyelephantopin

m.

DMAP, Et3N

Cl

O

l.

CH2Cl2, 0 C, 44%

j.

NaBH4, MeOH CH2Cl2, 0 C, quant.

CH2Cl2, 0 C, 66%

k. Grubbs I (20 mol%)

38% b.r.s.m.

CH2Cl2, reflux

O

O

O

O

O

2, DMAP, Et

k. Grubbs I (20 mol%)

32% b.r.s.m.

O

CH2Cl2, reflux

O

O

H

H

O

()-11

O

()-12

O

O

O

o.

p.

Pd2dba3 (2.5 mol%) Trost ligand (5 mol%)

NMP, 20 C 180C (w), PhMe

60%

O

O

O

O

n. NaHMDS, AllocCl THF, 60 C

97%

OMe MeO

OMe

MeO

OMe

O NH HN

PPh2

O

Ph2P

(S,S)-DACH-phenyl Trost ligand

13

14

()-(R)-6

4 steps

()-Nordeoxyelephantopin

Figure 2 | Synthesis of nordeoxyelephantopin and related analogues.

2-vinylbenzaldehyde 28 were used to make simplied analogues of the deoxyelephantopin (26 and 28, respectively).

Cytotoxicity of deoxyelephantopins and analogues. We began our biological investigations by assaying the cytotoxicity of natural deoxyelephantopin (DEP) and its unnatural analogues in four different cancer cell lines (Fig. 5a, Supplementary Figs 68). As expected, deoxyelephantopin proved potently toxic (o30% cellular viability at 1 mM concentration). The cytotoxic effect was equally strong when MCF7 cells were treated with nordeoxyelephantopin

or the open ring derivative 10 (Supplementary Fig. 8). We conrmed that the cell death is caused by caspase-mediated apoptosis by treating MCF7 cells with 20 mM DEP and imaging caspase activation and apoptosis using the uorescent probe FITC-VAD-FMK or Annexin-V-FLUOS staining (Fig. 5b). In contrast, propidium iodide staining did not indicate any signicant necrotic death in treated cells (Supplementary Fig. 9).

Covalent interactome of deoxyelephantopin. In our efforts to determine the direct molecular targets of deoxyelephantopin we

4 NATURE COMMUNICATIONS | 7:12470 | DOI: 10.1038/ncomms12470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12470 ARTICLE

OH I

a. TESOTf, 2,4,6-collidine CH2Cl2, r.t

b. DMP, CH2Cl2, r.t

c. 2, Zn, THF, aq. NH4Cl, r.t

TESO

I

OTHP

15

O

O

O

OH

e. or Ac2O

DMAP, Et3N CH2Cl2, 0 C

2

f. Grubbs II (20 mol%) CH2Cl2, reflux

O

O h. TsOH, THF/H2O, r.t.

i. Ni(CO)2(PPh3)2 Et3N, PhMe, reflux

I

TESO

O

R

R

O

O

O

O

19a: R = C(Me)CH2, 52% 19b: R = Me, 49% (yields over two steps)

17a: R = C(Me)CH2, 33% 17b: R = Me, 30%

g. Pd(OAc)2 (20 mol%) dppb (20 mol%)

K3PO4, PhMe, 60 Ch. TsOH, THF/H2O, r.t

MM2 minimized structure

MM2 minimized structure

j. 254 nm light, MeCN, r.t. quant.

18 O

O

O

R

O

20a: R = C(Me)CH2 20b: R = Me

Figure 3 | Synthesis of deoxyelephantopin-related analogues 1820.

rst performed a gel-based competitive in situ proteomic proling assay41 (Fig. 6a, Supplementary Fig. 11) using the g-butyrolactone probe 24c that showed the highest cytotoxicity among the entire series of alkyne-tagged analogues 24 in all four tested cell lines (Fig. 5a). While the greater conformational exibility of probe 24c could potentially lead to more promiscuous target engagement relatively to deoxyelephantopin, it was favoured based on the fact that an alkyne moiety appended to the rigid scaffold of deoxyelephantopin may hinder some of its interactions. MCF7 cells were pretreated with various concentrations of deoxyelephantopin for 4 h, harvested, lysed and the lysates were treated with the uorogenic probe 24c-Cy3 (10 mM; Supplementary Fig. 10). The gel prole showed multiple labelled bands, but only some of them were selectively competed by twofold excess of the natural product. Furthermore, cells were pretreated with 20 mM concentration of deoxyelephantopin or 24c and the proteomic proles were compared (Fig. 6b, Supplementary Fig. 11). Multiple bands were detected that were

equally competed by both DEP and 24c, thus again conrming that 24c can be used as an acceptable clickable analogue of deoxyelephantopin in pulldown assays. Having established the optimal labelling conditions by SDSPAGE, we next carried out a mass spectrometry-based competitive proling assay42 coupled with the method of stable isotope labelling of amino acids in culture (SILAC)43. In total, 1,522 proteins were enriched from MCF7 cells by 10 mM probe 24c, but labelling of only 11 proteins was 470% competed by 20 mM deoxyelephantopin (SILAC ratioo0.30; Supplementary Fig. 12; Supplementary Data 1).

Intriguingly, none of the 11 proteins were previously described as targets of deoxyelephantopin or related natural products. Moreover, we performed a targeted competitive proling assay following an in situ 4 h pretreatment of MCF7 cells with DEP and conrmed that all 11 identied targets are also engaged and bound by deoxyelephantopin directly in living cells (Fig. 6c, Supplementary Data 2). It should be noted that some of the identied targets, including CTTN44, CSTB45 and CBS46, are

NATURE COMMUNICATIONS | 7:12470 | DOI: 10.1038/ncomms12470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12470

a.

DIAD, PPh3, Et2O, r.t.b. Grubbs II (20 mol%) CH2Cl2 reflux

HO

O Br

O

c.

Zn, THF,aq. NH4Cl, r.t

d.

DMAP, Et CH2Cl2, 0 C

O

O

O

OH

O

O

OH

Br

O

O

O

O

O

21a: R = H 21b: R = iPr 21c: R = Ph

22a: R = H, 14% 22b: R = iPr, 27% 22c: R = Ph, 21%

(over two steps)

23a: R = H, 57% 23b: R = iPr, 50% 23c: R = Ph, 67%

24a: R = H, 23% 24b: R = iPr, 25% 24c: R = Ph, 13%

O

O

O

O

3 steps 3 steps

O

O

25 26 27 28

Figure 4 | Synthesis of deoxyelephantopin-related probes 23 and 24 and simplied analogues 26 and 28.

DEP

a

19a 23a 23b 23c 24a 24b 24c

25 20 15 10

6

4

Cell viability EC 50 (M)

2

0

**

***

***

**

**

***

*** **

MCF7 Caco-2 HeLa MDA-MB-231

20 M DEP

1 M staurosporine

DMSO

b

FITC-VAD-FMK Annexin-V-FLUOS

Figure 5 | Assessing the cytotoxicity of deoxyelephantopin (DEP) and analogues. (a) Cell viability measurements in four different cancer cell lines. Cells were treated with compounds for 3 days, the nuclei were stained with DAPI and counted (EC50 valuess.d. shown; n 3). **Po0.005 and ***Po0.0005

by two-sided Students t-test. (b) Detection of caspase activation and apoptosis in MCF7 cells using FITC-VAD-FMK uorescent probe (top) or Annexin V staining (bottom). The cells were treated with 1 mM staurosporine (positive control) or 20 mM deoxyelephantopin for 8 h. Confocal microscopy images are shown. EC50, effector concentration for half-maximum response. Scale bars, 30 mm.

6 NATURE COMMUNICATIONS | 7:12470 | DOI: 10.1038/ncomms12470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12470 ARTICLE

a b

24c-Cy3 (10 M)

24c-Cy3 (10 M)

0 0.3 1 3 10 20 0 DEP (M)

24c

DMSO

DEP (20 M)

kDa

kDa

250

250

150

150

100

100

75

75

50

50

37

37

c

Gene name

CBS

HGH1

ZNF346

GRHPR

TXNDC12

HMOX2

SEC24C

FLYWCH2

CTTN

NUP54

CSTB 11.14

50.78

Protein name

Mol. weight [kDa]

LFQ ratio DEP/DMSO s.d.

0.000

0.00

0.000

0.00

0.000

0.00

0.000

0.00

0.001

0.04

0.002

0.14

0.031

0.18

0.010

0.19

0.067

0.22

0.034

0.24

0.012

0.28

Cystathionine beta-synthase

60.59

42.13

Protein HGH1 homologue

Zinc finger protein 346

Glyoxylate reductase/hydroxypyruvate reductase

Thioredoxin domain-containing protein 12

Heme oxygenase 2

Protein transport protein Sec24C

FLYWCH family member 2

Src substrate cortactin

Nucleoporin p54

Cystatin-B

29.27

35.67

19.21

41.67

118.32

14.56

61.59

Figure 6 | Proteome-wide identication of deoxyelephantopin targets. (a) Gel-based competitive in situ proling of deoxyelephantopin (DEP) targets. MCF7 cells were treated with various concentrations of DEP for 4 h, lysed and treated with 10 mM 24c-Cy3. Asterisks indicate targets competed by DEP.

(b) Comparison of in situ proteomic proles of DEP versus derivative 24c after treatment of MCF7 cells for 4 h. (c) In situ proteomic targets of DEP (470% competition) in MCF7 cells. Quantication was performed using label-free quantication (LFQ) method (n 4).

known to cause cell death on depletion and thus potentially explain the cytotoxic activity of deoxyelephantopins. Mindful that another sesquiterpene lactone (ainsliadimer A) bearing an a-exomethylene-g-butyrolactone has been shown to covalently inhibit IKKa/b (ref. 47), we tested the probe 24c against recombinant IKKb but found no evidence of covalent interaction supporting the orthogonal target selectivity of these natural products.

On the basis of the reported interaction between deoxyelephantopin and PPARg and because MCF7 cells are known to express very low levels of this protein, we investigated in a separate experiment whether deoxyelephantopin indeed covalently binds PPARg. The recombinant transcription factor was spiked into

MCF7 lysates to 30 nM nal concentration and successfully enriched by the methacrylate ester probe 24c, as evidenced from LCMS/MS-based label-free quantication (LFQ)48. Moreover, PPARg was also successfully enriched with probe 23c yielding comparable LFQ value, indicating that the g-butyrolactone group is involved in the covalent bond formation with the target protein (Supplementary Fig. 13; Supplementary Data 3). To investigate

whether DEP is also capable of binding to endogenous PPARg directly in living cells, we performed a targeted competitive proteomics experiment in Caco-2, a colon cancer cell line known to express larger amounts of PPARg49. The cells were treated in situ with dimethyl sulfoxide (DMSO) or 20 mM deoxyelephantopin, lysed and treated with 24c. Gratifyingly, endogenous PPARg was successfully enriched and quantied from the DMSO-treated Caco-2 cells, whereas DEP pretreatment efciently competed PPARg enrichment (495%; Fig. 7a). In yet another control experiment, we expressed hPPARg in HeLa cells and successfully conducted a gel-based competition experiment once again using DEP as an in situ competitor and 24c-Cy3 as reporter probe (Supplementary Fig. 14). Encouraged by these results, we sought to investigate the exact mode of DEP action on PPARg receptor.

Deoxyelephantopin antagonizes PPARc. Caco-2 cells were treated with DMSO, PPARg agonist rosiglitazone, antagonist T0070907 or DEP for 24 h, the cells were collected and the

NATURE COMMUNICATIONS | 7:12470 | DOI: 10.1038/ncomms12470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12470

a b

2

Rosiglitazone T0070907 DEP

100

75

Normalized LFQ intensity

1

0

50

25

0

10 M 24c

1

*** ***

20 M DEP/

10 M 24c

Log 2ratio cpd/DMSO

2 RBP4

GSTA2

SLC26A3

LIPA

ANXA1

c

24c-Cy3 (10 M)

DMSO

23a

23c

23b

24a

24c

24b

DEP

10

ent-10

18

26

28

DMSO

NorDEP

ent-NorDEP

11

12

19a

19b

20a

20b

(20 M)

hPPAR

100

75

Normalized intensity cpd/DMSO

50

25

DMSO

0 23a

23c

23b

24a

24c

24b

DEP

10 ent-10

18 26 NorDEP ent-NorDEP

11

28

12 19a

19b

20a

20b

Compounds (20 M)

Figure 7 | Identication of PPARc as target of deoxyelephantopin and screening of DEP derivatives as PPARc antagonists. (a) Quantication of endogenous PPARg enriched from Caco-2 cells using 10 mM 24c following in situ treatment with DEP (20 mM) or DMSO (normalized LFQ valuess.d. are shown; n 6). (b) Changes in expression levels of ve PPARg-dependent proteins following 24 h treatment of Caco-2 cells with DMSO, rosiglitazone

(10 mM), T0070907 (50 mM) or DEP (20 mM; LFQ ratioss.d. are shown; n 3). **Po0.005 and ***Po0.0005 by two-sided Students t-test.

(c) Gel-based competitive screening of DEP and analogues as PPARg antagonists against recombinant human enzyme (top). hPPARg was treated with each compound (20 mM) for 1 h followed by treatment with 24c-Cy3 (10 mM) as probe. Quantication of uorescence intensity in the competitive screening for

PPARg antagonists (bottom; valuess.d. are shown; n 3).

proteomes were comparatively analysed by LCMS/MS. Using LFQ as quantication method, we identied 25 upregulated and 18 downregulated proteins following treatment with both DEP and T0070907 and these proteins were either not, or differently affected by rosiglitazone treatment (Supplementary Data 4), indicating that DEP indeed acts in cells as an antagonist of PPARg. Furthermore, the list of 43 dysregulated proteins was analysed and classied using the GOrilla pathway analysis tool (Supplementary Fig. 15). Further analysis revealed ve previously reported PPARg-dependent proteins RBP4, GSTA2, SLC26A3,

LIPA and ANXA1 among all dysregulated targets (Fig. 7b). Most signicantly, RBP4, a well-studied adipokine and contributor to insulin resistance in obesity and type 2 diabetes50 and thus a major therapeutic target, was more strongly downregulated following treatment with DEP than T0070907, suggesting that DEP or a derivative may indeed nd clinical use as means to

reduce insulin resistance in various metabolic diseases. Interestingly, T0070907 was also reported to suppress breast cancer cell proliferation and motility in a PPARg-dependent manner51, meaning that at least some part of the cytotoxic activity of deoxyelephantopin may indeed originate from antagonizing this receptor.

Ranking of covalent PPARc binders. Wondering whether we would be able to identify more potent PPARg binders, we also screened other synthetic analogues of deoxyelephantopin in a gel-based competitive format using 24c-Cy3 (Fig. 7c, Supplementary Fig. 16). Briey, recombinant human PPARg was pretreated with compounds at 20 mM concentration followed by the probe 24c-Cy3. Analysis of labelling revealed that the Z-analogue of nordeoxyelephantopin (19a) is the most effective

8 NATURE COMMUNICATIONS | 7:12470 | DOI: 10.1038/ncomms12470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12470 ARTICLE

covalent binder to PPARg followed by the tricyclic analogue 28. Assuming covalent bond formation with PPARg, we applied the method of Kitz and Wilson52 to determine the kinetic values Ki and kinact for deoxyelephantopin and compound 19a (Supplementary Fig. 17). We observed clear time-dependent shift in half-maximal inhibitory concentration (IC50) values

characteristic for an irreversible mode of binding. Calculated kinact/Ki values of 680 M 1 min 1 for deoxyelephantopin and 1,241 M 1 min 1 for 19a conrmed that the latter compound is indeed a more potent PPARg binder.

Deoxyelephantopin and related analogues react with PPARcs zinc nger. Finally, we sought to understand the molecular basis for PPARg binding by the synthetic analogues of deoxyelephantopin through identication of the precise covalently modied amino acid site. There is literature precedent for oxidized fatty acids containing an a,b-unsaturated ketone to form a covalent bond with a cysteine in the ligand-binding site of PPARg53. More recently, synthetic ligands binding reversibly to a distinct site were identied54. Pretreatment of PPARg with the general cysteine-reactive probe iodoacetamide (IAA) completely abolished gel-based uorescence labelling with 24c-Cy3, implying that probe 24c and deoxyelephantopin covalently engage a cysteine residue (Supplementary Fig. 18). Bearing in mind that the structurally simplied, and thus less prone to MS/MS fragmentation, probe 23c binds PPARg as efciently as 24c (Supplementary Fig. 13), we treated the recombinant nuclear receptor with the former compound, digested the protein and analysed the resulting peptides by LCMS/MS. Peptide mass adduct search and detailed interrogation of MS2 spectra resulted in unambiguous identication of Cys190 as the protein site modied by 23c (Supplementary Fig. 19). Moreover, we also performed an IAA-competitive labelling experiment, where iodoacetamide labelling of cysteines on PPARg was competed through pretreatment with compounds 23c, 24c and DEP. MS2 spectra inspection revealed efcient (475%) competition of labelling of the Cys190-containing tryptic peptide with all three compounds, thus additionally conrming the identied binding site (Supplementary Data 5). Interestingly, analysis of the high-resolution X-ray structure of PPARg (PDB: 3DZU)

revealed that this cysteine is coordinated to a Zn2 ion in a zinc-nger motif as part of the DNA binding domain of PPARg (Supplementary Fig. 20). While zinc ngers respond to oxidative stress through reaction at cysteines55,56 and have been targeted by

small molecules forming disulde bridges57, to the best of our knowledge, deoxyelephantopin is the rst example of a small molecule that engages a zinc nger through a Michael addition. The success of Michael acceptors as a cysteine trap in current clinical development of covalent inhibitors points to the privileged reactivity prole of this moiety. Having identied the exact binding site, we then performed covalent docking of 19a using the proteinligand docking program GOLD. The predicted optimal binding mode shows key interactions between the methacrylate carbonyl group of 19a and Asp174 as well as the g-butyrolactone group and Cys176 (Fig. 8a, Supplementary

Fig. 20). Following the docking result, we mutated these two residues, transiently expressed the wild-type and the two PPARg mutant forms in 293T cells and performed gel-based competitive experiments to determine relative binding afnities for 19a. We observed a shift in the IC50 value from 19 mM (wt PPARG) to 33 mM (C176A) and 58 mM (D174A; Fig. 8b, Supplementary Fig. 21) that further supports the suggested binding mode of 19a. Finally, it is interesting to note that SILAC experiment led to the identication of ZNF346, another zinc nger motif-containing protein.

In conclusion, the procedures developed in the context of the synthesis of deoxyelephantopin analogues provide a rapid access to diverse ring systems embedding an a-methylene-gbutyrolactone, an important moiety in sesquiterpene lactones. PPARg-targeting small molecules such as rosiglitazone and other thiazolidinediones58 are clinically approved drugs against diabetes. The present discovery that deoxyelephantopin and related synthetic analogues react with the zinc-bound Cys190 in a zinc-nger motif of PPARg offers a novel pharmacological mechanism for modulating PPARg activity and may serve as a blueprint for the development of a new generation of potent antagonists of PPARg (refs 59,60) or other transcription factors with zinc-nger motif. Finally, following the identication of cancer-related proteins CTTN, CSTB and CBS as novel proteomic targets of deoxyelephantopin, it is also tempting to speculate about potential biomedical application of synthetic analogues of this intriguing natural product as novel anticancer therapeutics.

Methods

Gel-based screening for PPARc binders using probe 24c-Cy3. Recombinant human PPARg (Cayman Chemical, 50 ng in 12.5 ml PBS) was treated with 20 mM of each compound (0.5 ml of 25 stock in DMSO), 10 mM iodoacetamide or DMSO

for one hour. Samples were then treated with 10 mM 24c-Cy3 (0.5 ml of 25 stock

in DMSO) for 1 h at room temperature in the dark. SDSPAGE reducing loading buffer (4 ) was added and proteins were separated using a 10% SDSPAGE gel.

a b

24c-Cy3 intensity (%)

100

75

50

Asp174

Cys176

Cys190

25 WT IC50 = 19 M

C176A IC50 = 33 M

D174A IC50 = 58 M

Concentration (M)

0 3 10 30 100 300

Figure 8 | Proposed model for 19a binding to PPARc. (a) Docking prediction of 19a covalently bound to Cys190 of human PPARg. The model was created using the proteinligand docking program GOLD. (b) Measurement of 19a binding to WT, C176A and D174A PPARg using gel-based competitive assay. The proteins were transiently expressed in 293T cells and the cellular lysates were treated with different concentrations of 19a for 1 h followed by 24c-Cy3 (10 mM). Doseresponse curves are shown.

NATURE COMMUNICATIONS | 7:12470 | DOI: 10.1038/ncomms12470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12470

Gels were visualized at 625 nm using a Hitachi FMBIO II Multi-View uorescence scanner, then stained using silver staining. Images were quantied with ImageJ.

Detection of a covalent adduct on human PPARc. Recombinant human PPARg (500 ng in 9 ml PBS) was treated with 100 mM of 23c (1 ml of 10 stock in DMSO) or

DMSO for 1 h. Samples were denatured with 6 M urea in 50 mM NH4HCO3, reduced with 10 mM TCEP for 30 min and alkylated with 25 mM iodoacetamide for 30 min in the dark. Samples were diluted to 2 M urea with 50 mM NH4HCO3, and digested with trypsin (0.25 ml of 0.05 mg ml 1) in the presence of 1 mM CaCl2 for 12 h at 37 C.

Samples were acidied to a nal concentration of 5% acetic acid, desalted over a self-packed C18 spin column and dried. Samples were resuspended in 0.1% FA in water and analysed by LCMS/MS. The theoretical mass of the 23c adduct was calculated and LCMS/MS was run with an inclusion list containing m/z and charge of the expected peptides. MS data were analysed in MaxQuant with 23c( 256.1099 Da) and carbamidomethylation as variable modications on cysteine.

Only modied peptides with PEP value r1% were considered.

Data availability. The data supporting the ndings of this study are available within the article and its Supplementary Information (Supplementary Figs 198; Experimental procedures for cellular and biochemical experiments; NMR comparison of deoxyelephantopin and nordeoxyelephantopin; NMR spectra and chiral GC chromatograms); Supplementary Data 1 (Supplementary Tables 15 of the proteomic data) and from the corresponding author on request.

References

1. Li, J. W. H. & Vederas, J. C. Drug discovery and natural products: end of an era or an endless frontier? Science 325, 161165 (2009).

2. Newman, D. J. & Cragg, G. M. Natural Products as sources of new drugs over the 30 years from 1981 to 2010. J. Nat. Prod. 75, 311335 (2012).

3. Drahl, C., Cravatt, B. F. & Sorensen, E. J. Protein-reactive natural products. Angew. Chem. Int. Ed. 44, 57885809 (2005).

4. Robertson, J. G. Mechanistic basis of enzyme-targeted drugs. Biochemistry 44,

55615571 (2005).

5. Swinney, D. C. The role of binding kinetics in therapeutically useful drug action. Curr. Opin. Drug Discov. 12, 3139 (2009).

6. Potashman, M. H. & Duggan, M. E. Covalent modiers: an orthogonal approach to drug design. J. Med. Chem. 52, 12311246 (2009).

7. Singh, J., Petter, R. C., Baillie, T. A. & Whitty, A. The resurgence of covalent drugs. Nat. Rev. Drug Discov. 10, 307317 (2011).

8. Sanderson, K. Irreversible kinase inhibitors gain traction. Nat. Rev. Drug Discov. 12, 649651 (2013).

9. Wender, P. A. & Miller, B. L. Synthesis at the molecular frontier. Nature 460, 197201 (2009).

10. Schreiber, S. L. Organic chemistry: Molecular diversity by design. Nature 457, 153154 (2009).

11. Shaw, J. T. Naturally diverse: highlights in versatile synthetic methods enabling target- and diversity-oriented synthesis. Nat. Prod. Rep. 26, 1126 (2009).12. Galloway, W. R., Isidro-Llobet, A. & Spring, D. R. Diversity-oriented synthesis as a tool for the discovery of novel biologically active small molecules. Nat. Commun. 1, 8086 (2010).

13. Wetzel, S., Bon, R. S., Kumar, K. & Waldmann, H. Biology-oriented synthesis. Angew. Chem. Int. Ed. 50, 1080010826 (2011).

14. Merfort, I. Perspectives on sesquiterpene lactones in inammation and cancer. Curr. Drug Targets 12, 15601573 (2011).

15. Chadwick, M., Trewin, H., Gawthrop, F. & Wagstaff, C. Sesquiterpenoids Lactones: Benets to Plants and People. Int. J. Mol. Sci. 14, 1278012805 (2013).

16. Hoffmann, H. M. R. & Rabe, J. Synthesis and biological-activity of alpha-methylene-gamma-butyrolactones. Angew. Chem. Int. Ed. 24, 94110 (1985).

17. Kitson, R. R. A., Millemaggi, A. & Taylor, R. J. K. The renaissance of alpha-methylene-gamma-butyrolactones: new synthetic approaches. Angew. Chem. Int. Ed. 48, 94269451 (2009).

18. Gersch, M., Kreuzer, J. & Sieber, S. A. Electrophilic natural products and their biological targets. Nat. Prod. Rep. 29, 659682 (2012).

19. Lyss, G., Knorre, A., Schmidt, T. J., Pahl, H. L. & Merfort, I. The anti-inammatory sesquiterpene lactone helenalin inhibits the transcription factor NF-kappa B by directly targeting p65. J. Biol. Chem. 273, 3350833516 (1998).

20. Lytton, J., Westlin, M. & Hanley, M. R. Thapsigarin inhibits the SERCA family of Ca-pump. J. Biol. Chem. 266, 1706717071 (1991).

21. Shaikenov, T. E. et al. Arglabin-DMA, a plant derived sesquiterpene, inhibits farnesyltransferase. Oncol. Rep. 8, 173179 (2001).

22. Kunzmann, M. H., Bach, N. C., Bauer, B. & Sieber, S. A. alpha-Methylene-gamma-butyrolactones attenuate Staphylococcus aureus virulence by inhibition of transcriptional regulation. Chem. Sci 5, 11581167 (2014).

23. Su, M., Wu, X., Chung, H. Y., Li, Y. & Ye, W. Antiproliferative activitiesof ve chinese medicinal herbs and active compounds in Elephantopus scaber. Nat. Prod. Commun. 4, 10251030 (2009).

24. Huang, C.-C., Lo, C.-P., Chiu, C.-Y. & Shyur, L.-F. Deoxyelephantopin, a novel multifunctional agent, suppresses mammary tumour growth and lung metastasis and doubles survival time in mice. Br. J. Pharmacol. 159, 856871 (2010).

25. Ichikawa, H. et al. Isodeoxyelephantopin, a novel sesquiterpene lactone, potentiates apoptosis, inhibits invasion, and abolishes osteoclastogenesis through suppression of nuclear factor-kappa B (NF-kappa B) activation and NF-kappa B-regulated gene expression. Clin. Cancer Res. 12, 59105918 (2006).

26. Lee, W.-L., Wen, T.-N., Shiau, J.-Y. & Shyur, L.-F. Differential proteomic proling identies novel molecular targets of paclitaxel and phytoagent deoxyelephantopin against mammary Adenocarcinoma cells. J. Proteome Res. 9, 237253 (2010).

27. Zou, G. et al. Deoxyelephantopin inhibits cancer cell proliferation and functions as a selective partial agonist against PPAR gamma. Biochem. Pharmacol. 75, 13811392 (2008).

28. Zhang, F. H., Yang, Y. X., Xie, L. G. & Xu, X. H. Pd-catalyzed diastereoselective allylation of aldehydes with 3-bromomethyl-5H-furan-2-one: stereoselective synthesis of beta-(hydroxymethylaryl/alkyl)-alpha-methylenec-gamma-butyrolactones with a syn conguration. Chem. Commun. 49, 46974699 (2013).

29. Hodgson, D. M., Talbot, E. P. A. & Clark, B. P. Stereoselective synthesis of beta-(Hydroxymethylaryl/alkyl)-alpha-methylene-gamma-butyrolactones. Org. Lett. 13, 25942597 (2011).

30. Furstner, A. & Muller, T. The rst synthesis of a 10-membered ring by olen metathesis: Jasmine ketolactone. Synlett 1997, 10101012 (1997).

31. Semmelhack, M. F. & Brickner, S. J. Intramolecular carbonylation of vinyl halides. J. Org. Chem. 46, 17231726 (1981).

32. Funk, T. W., Berlin, J. M. & Grubbs, R. H. Highly active chiral ruthenium catalysts for asymmetric ring-closing olen metathesis. J. Am. Chem. Soc. 128, 18401846 (2006).

33. Fournier, P. A. et al. Mechanistically inspired catalysts for enantioselective desymmetrizations by olen metathesis. Chem. Eur. J 14, 86908695 (2008).

34. Kress, S. & Blechert, S. Asymmetric catalysts for stereocontrolled olen metathesis reactions. Chem. Soc. Rev. 41, 43894408 (2012).

35. Frantz, D. E., Fassler, R. & Carreira, E. M. Facile enantioselective synthesis of propargylic alcohols by direct addition of terminal alkynes to aldehydes.J. Am. Chem. Soc. 122, 18061807 (2000).36. Fournier, J., Arseniyadis, S. & Cossy, J. A modular and scalable one-pot synthesis of polysubstituted furans. Angew. Chem. Int. Ed. 51, 75627566 (2012).

37. Fournier, J., Lozano, O., Menozzi, C., Arseniyadis, S. & Cossy, J. Palladium-catalyzed asymmetric allylic alkylation of cyclic dienol carbonates: efcient route to enantioenriched gamma-butenolides bearing an all-carbon alpha-quaternary stereogenic center. Angew. Chem. Int. Ed. 52, 12571261 (2013).

38. Trost, B. M., Xu, J. Y. & Schmidt, T. Palladium-catalyzed decarboxylative asymmetric allylic alkylation of enol carbonates. J. Am. Chem. Soc. 131, 1834318357 (2009).

39. Fujioka, H. et al. Unexpected highly chemoselective deprotection of the acetals from aldehydes and not ketones: TESOTf-2,6-lutidine combination. J. Am. Chem. Soc. 126, 1180011801 (2004).

40. Stichnoth, D. et al. Photochemical formation of intricarene. Nat. Commun. 5,

5597 (2014).

41. Jessani, N., Liu, Y. S., Humphrey, M. & Cravatt, B. F. Enzyme activity proles of the secreted and membrane proteome that depict cancer cell invasiveness. Proc. Natl Acad. Sci. USA 99, 1033510340 (2002).

42. Kidd, D., Liu, Y. S. & Cravatt, B. F. Proling serine hydrolase activities in complex proteomes. Biochemistry 40, 40054015 (2001).

43. Li, J. et al. Simultaneous structure-activity studies and arming of natural products by C-H amination reveal cellular targets of eupalmerin acetate. Nat. Chem. 5, 510517 (2013).

44. Luo, M. L. et al. Amplication and overexpression of CTTN (EMS1) contribute to the metastasis of esophageal squamous cell carcinoma by promoting cell migration and anoikis resistance. Cancer Res. 66, 1169011699 (2006).

45. Feldman, A. S., Banyard, J., Wu, C. L., McDougal, W. S. & Zetter, B. R. Cystatin B as a tissue and urinary biomarker of bladder cancer recurrence and disease progression. Clin. Cancer Res. 15, 10241031 (2009).

46. Maclean, K. N. et al. Cystathionine protects against endoplasmic reticulum stress-induced lipid accumulation, tissue injury, and apoptotic cell death. J. Biol. Chem. 287, 3199432005 (2012).

47. Dong, T. et al. Ainsliadimer A selectively inhibits IKK alpha/beta by covalently binding a conserved cysteine. Nat. Commun. 6, 6522 (2015).

48. Cox, J. et al. Accurate Proteome-wide label-free quantication by delayed normalization and maximal peptide ratio extraction, termed MaxLFQ. Mol. Cell Proteomics 13, 25132526 (2014).

49. DuBois, R. N. et al. The nuclear eicosanoid receptor, PPAR gamma, is aberrantly expressed in colonic cancers. Carcinogenesis 19, 4953 (1998).

50. Yang, Q. et al. Serum retinol binding protein 4 contributes to insulin resistance in obesity and type 2 diabetes. Nature 436, 356362 (2005).

10 NATURE COMMUNICATIONS | 7:12470 | DOI: 10.1038/ncomms12470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12470 ARTICLE

51. Zaytseva, Y. Y., Wallis, N. K., Southard, R. C. & Kilgore, M. W. The PPAR gamma antagonist T0070907 suppresses breast cancer cell proliferation and motility via Both PPAR gamma-dependent and -independent mechanisms. Anticancer Res. 31, 813823 (2011).

52. Kitz, R. & Wilson, I. B. Esters of methanesulfonic acid as irreversible inhibitors of acetylcholinesterase. J. Biol. Chem. 237, 3245324 (1962).

53. Itoh, T. et al. Structural basis for the activation of PPAR gamma by oxidized fatty acids. Nat. Struct. Mol. Biol. 15, 924931 (2008).

54. Hughes, T. S. et al. An alternate binding site for PPAR gamma ligands. Nat. Commun. 5 Article number: 3571 (2014).

55. Isaac, M., Latour, J. M. & Seneque, O. Nucleophilic reactivity of Zinc-bound thiolates: subtle interplay between coordination set and conformational exibility. Chem. Sci. 3, 34093420 (2012).

56. Lebrun, V. et al. Efcient oxidation and destabilization of Zn(Cys)(4) zinc ngers by singlet oxygen. Angew. Chem. Int. Ed. 53, 93659368 (2014).

57. Rice, W. G. et al. Inhibitors of HIV nucleocapsid protein zinc ngers as candidates for the treatment of aids. Science 270, 11941197 (1995).

58. Spiegelman, B. M. PPAR-gamma: adipogenic regulator and thiazolidinedione receptor. Diabetes 47, 507514 (1998).

59. Lehmann, J. M. et al. An antidiabetic thiazolidinedione is a high-afnity ligand for peroxisome proliferator-activated receptor gamma(Ppar-gamma). J. Biol. Chem. 270, 1295312956 (1995).

60. Ahmadian, M. et al. PPAR gamma signaling and metabolism: the good, the bad and the future. Nat. Med. 19, 557566 (2013).

Acknowledgements

We thank the SNSF and the NCCR Chemical Biology for nancial support.

Author contributions

N.W. and A.A. conceived the study and designed the experiments. R.L. and C.S. synthesized the compounds. D.A. and D.G.H. performed the biological studies. N.W. and A.A. wrote the manuscript, all authors reviewed and edited the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Lagoutte, R. et al. Divergent synthesis and identication of the cellular targets of deoxyelephantopins. Nat. Commun.. 7:12470 doi: 10.1038/ ncomms12470 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2016

NATURE COMMUNICATIONS | 7:12470 | DOI: 10.1038/ncomms12470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11