Electrophilic tuning of the chemoprotective

natural product sulforaphane

Young-Hoon Ahna, Yousang Hwanga, Hua Liua, Xiu Jun Wangb, Ying Zhangb, Katherine K. Stephensona, Tatiana N.
Boroninac, Robert N. Colec, Albena T. Dinkova-Kostovaa,b, Paul Talalaya,1, and Philip A. Colea,1
a
Department of Pharmacology and Molecular Sciences, The Johns Hopkins University School of Medicine, 725 North Wolfe Street, Baltimore, MD 21205;
b
Biomedical Research Institute, Ninewells Hospital and Medical School, University of Dundee, Dundee DD1 9SY, United Kingdom; and cMass
Spectrometry and Proteomics Facility, The Johns Hopkins University School of Medicine, 733 North Broadway, Baltimore, MD 21205

Contributed by Paul Talalay, March 26, 2010 (sent for review March 19, 2010)

Sulforaphane [1-isothiocyanato-4-(methylsulfinyl)butane], a natu- where it activates gene expression (Fig. 1A). However, other pro-
rally occurring isothiocyanate derived from cruciferous vegetables, tein targets for sulforaphane have also been proposed (4, 18), and
is a highly potent inducer of phase 2 cytoprotective enzymes and mechanistic questions remain.
can protect against electrophiles including carcinogens, oxidative Whereas sulforaphane has been shown to react directly with
stress, and inflammation. The mechanism of action of sulforaphane purified, recombinant Keap1 in solution (15, 19), it has not yet
is believed to involve modifications of critical cysteine residues of been established that sulforaphane modifies Keap1 in cells. Hu-
Keap1, which lead to stabilization of Nrf2 to activate the antioxi- man Keap1 is a 70-kDa cysteine-rich protein (624 amino acids
dant response element of phase 2 enzymes. However, the dithio- and 27 cysteines), comprising five domains: (i) N-terminal region
carbamate functional group formed by a reversible reaction (NTR), (ii) BTB (broad complex, tramtrack, or bric-a-brac), an
between isothiocyanate of sulforaphane and sulfhydryl nucleo- evolutionarily conserved protein-protein interaction motif that
philes of Keap1 is kinetically labile, and such modification in intact often dimerizes with other BTB domains (20), (iii) intervening
cells has not yet been demonstrated. Here we designed sulfo- region (IVR), a cysteine-rich region, (iv) double glycine region
raphane analogs with replacement of the reactive isothiocyanate (DGR), a domain that comprises six Kelch motifs and binds
by the more gentle electrophilic sulfoxythiocarbamate group that to the Neh2 domain of Nrf2 (21), and (v) C-terminal region
also selectively targets cysteine residues in proteins but forms (CTR). It is difficult to express soluble Keap1, which is highly
stable thiocarbamate adducts. Twenty-four sulfoxythiocarbamate prone to oxidation and oligomerizes easily. Sulforaphane con-
analogs were synthesized that retain the structural features impor- tains the highly reactive isothiocyanate functionality that forms
tant for high potency in sulforaphane analogs: the sulfoxide or dithiocarbamate products with sulfhydryl nucleophiles both enzy-
keto group and its appropriate distance to electrophilic functional
Downloaded from https://www.pnas.org by 103.168.155.60 on October 17, 2023 from IP address 103.168.155.60.

matically catalyzed by glutathione transferase and nonenzymati-

group. Evaluation in various cell lines including hepatoma cells, cally (Fig. 1B) (22, 23). Such dithiocarbamate adduct formation is
retinal pigment epithelial cells, and keratinocytes as well as in a reversible process, which complicates their isolation and char-
mouse skin shows that these analogs maintain high potency acterization (22).
and efficacy for phase 2 enzyme induction as well as the inhibitory In prior studies on coenzyme A (CoA)-utilizing enzymes, it was
effect on lipopolysaccharide-induced nitric oxide formation like sul- found that a sulfoxythiocarbamate-CoA analog (CoA-probe,
foraphane. We further show in living cells that a sulfoxythiocarba- Fig. 1C) was a relatively mild but selective electrophilic probe
mate analog can label Keap1 on several key cysteine residues as for acetyltransferases (24). Mass spectrometric analysis revealed
well as other cellular proteins offering new insights into the me- that cysteine residues were selectively targeted by this CoA probe,
chanism of chemoprotection. and stable thiocarbamate adducts could be readily isolated. We
considered the possibility that replacing the isothiocyanate moi-

T he increasing societal burden of cancer and other chronic de-

generative diseases has led to an intense interest in the devel-
opment of strategies designed to reduce the risk of these
ety of sulforaphane with a sulfoxythiocarbamate group (Fig. 1B)
might allow retention of the natural product’s chemoprotective
properties and confer enhanced potential for mechanistic anal-
conditions. Chemoprotection with natural or synthetic agents of- ysis. Below, we report the design and synthesis of a series of
fers an attractive approach to boost the body’s defenses to ward sulfoxythiocarbamate sulforaphane analogs and describe their
off environmental and endogenous insults (1, 2). The plant pro- chemopreventive and protein labeling properties.
duct sulforaphane [1-isothiocyanato-4-(methylsulfinyl)butane],
derived from glucosinolates present in broccoli and other cru- Results
ciferous vegetables (3), has served as a prototype for our Design and Synthesis of Sulfoxythiocarbamate Sulforaphane Analogs.
understanding of chemoprotection by induction of phase 2 cyto- The simple cellular bioassay of measuring NQO1 enzyme activ-
protective enzymes including NAD(P)H:quinone oxidoreductase ities of murine hepatoma cells has accurately predicted and quan-
(NQO1) (4). A broad series of animal and human studies has de- tified the induction potency of many compounds. These potencies
monstrated the potential of sulforaphane to protect against the have been expressed as CD (Concentrations required to Double
onset, or reduce the severity, of cancer (5–8), retinal disease (9), the NQO1 activity) values (25, 26). With this assay, several struc-
and skin damage (10–12) resulting from oxidative or electrophile tural derivatives of sulforaphane have been compared for their
damage (5, 6, 8), UV irradiation (10, 11), or genetic predisposi-
tion (9, 13).
Like other known phase 2 enzyme inducers (14), sulforaphane Author contributions: Y.-H.A., Y.H., A.T.D.-K., P.T., and P.A.C. designed research; Y.-H.A.,
Y.H., H.L., X.J.W., Y.Z., K.K.S., T.N.B., and A.T.D.-K. performed research; Y.-H.A., Y.H.,
is an electrophilic compound that covalently modifies cysteine re- H.L., X.J.W., Y.Z., A.T.D.-K., P.T., and P.A.C. analyzed data; and Y.-H.A., R.N.C., A.T.D.-K.,
sidues in proteins (15, 16). Such enzyme induction likely involves P.T., and P.A.C. wrote the paper.
the Keap1-Nrf2-antioxidant response element pathway (15, 17). The authors declare no conflict of interest.
The prevailing hypothesis for sulforaphane’s cellular mechanism 1
To whom correspondence may be addressed. E-mail: [email protected] or ptalalay@
is that the natural product covalently modifies Keap1 on one or jhmi.edu.
more of its 27 cysteine residues, altering the Keap1-Nrf2-Cullin-3 This article contains supporting information online at www.pnas.org/lookup/suppl/
protein complex, and allowing Nrf2 to translocate to the nucleus doi:10.1073/pnas.1004104107/-/DCSupplemental.

9590–9595 ∣ PNAS ∣ May 25, 2010 ∣ vol. 107 ∣ no. 21 www.pnas.org/cgi/doi/10.1073/pnas.1004104107

 Fig. 1. Proposed chemoprotective mechanism of sulforaphane for phase 2 gene activation involves covalent Cys modification of Keap1. (A) Mechanism of
phase 2 enzyme induction. In the basal state, the transcription factor Nrf2 is efficiently ubiquitylated and targeted for proteasomal degradation by forming a
complex with Keap1 and Cul3 E3 ligase. Inducers such as sulforaphane modify cysteines of Keap1, which leads to stabilization and translocation of Nrf2 to the
nucleus, and its binding to ARE and stimulation of phase 2 gene transcription. (B) Reactions of sulforaphane and sulfoxythiocarbamate analog with a thiol. The
isothiocyanate group of sulforaphane reacts with sulfhydryl groups on protein to form dithiocarbamate adduct that appear to be kinetically labile (Top),
whereas sulfoxythiocarbamate analog forms a relatively stable thiocarbamate derivative (Bottom). (C) Structure of sulfoxythiocarbamate CoA probe devel-
oped previously (24).

potencies (3, 27). These findings highlight the importance of the bamate functionality. Synthesis of the keto analogs began with
sulfoxide group of sulforaphane in addition to the isothiocyanate 6-chloro-2-hexanone that, after protection of the ketone, was
group. Thus n-hexyl (CD 15 μM), n-cyclohexyl (CD 56 μM), n- derivatized in a similar fashion to provide the target com-
phenyl (inactive), and n-benzyl isothiocyanate (2–3 μM) are much pounds, 8a–8f.
less potent inducers when compared to sulforaphane (CD ¼
0.22 μM) (27). This sulfoxide group could be replaced by a polar Comparison of Reactivities of Sulforaphane and Sulfoxythiocarba-
functional group, ketone, without losing its activity while the cor- mate Analogs with a Thiol. To examine the relative electrophilicity
responding sulfone (CD 0.82 μM) and sulfide (CD 2.3 μM) of the sulfoxythiocarbamate and the isothiocyanate functional-
showed somewhat lower potencies (27). In addition, the length ities, we compared the reaction rates of β-mercaptoethanol with
of the methylene bridge between two important functional 4i, 8a, and sulforaphane. These reactions could be readily fit to a
groups, isothiocyanate and sulfoxide, is an important determinant pseudofirst order kinetic model with the sulfoxythiocarbamates
Downloaded from https://www.pnas.org by 103.168.155.60 on October 17, 2023 from IP address 103.168.155.60.

of potency since a shorter or longer bridge resulted in lower po- showing approximately 70-fold slower rate constants compared
tencies (3). Interestingly, the structure-potency studies of potent to sulforaphane (Figs. S1 and S2). This finding is consistent with
triterpenoid Michael acceptor inducers also illustrated the critical the hypothesis that the sulfoxythiocarbamate is a significantly
importance of two electrophilic functional groups separated by a weaker electrophile toward thiols relative to the isothiocya-
potentially similar distance to that found in sulforaphane (28). nate group.
Considering these structural features, sulfoxythiocarbamate ana-
logs were designed to retain the 4-methylsulfinylbutyl group moi- NQO1 Induction by Sulfoxythiocarbamate Analogs in Cells. Next, our
ety of sulforaphane with replacement of the isothiocyanate by a synthetic sulforaphane analogs were assayed in a murine hepato-
sulfoxythiocarbamate group (Fig 1B). In order to substitute the ma cell line (Hepa1c1c7) to quantify the induction of NQO1 ac-
isothiocyanate functionality of sulforaphane with a sulfoxythio- tivity as a marker of phase 2 enzyme induction (25, 26). Cellular
carbamate group, two significant chemical constraints had to be
overcome. First, sulfoxythiocarbamates, unlike isothiocyanates,
require alkyl substitution on the sulfur atom (R2 group, Fig. 1B)
to provide chemical stability. Second, the nitrogen atom in the
sulfoxythiocarbamate must also be alkyl-substituted to reduce hy-
drolytic breakdown (R1 group, Fig. 1B). These chemical require-
ments led us to explore a range of derivatives (Fig. 2) whose
substitution differences could influence potency, toxicity, and
protein labeling. Since analogs of sulforaphane in which the sulf-
oxide is replaced by a carbonyl group show very similar inducer
potencies (27), we also explored the effects of replacement of the
sulfoxide moiety with a carbonyl functionality (8a–8f, Fig. 2). A
BIOCHEMISTRY

key conserved feature of our analogs is a backbone containing an

electrophilic carbonyl group attached via a butyl linker to the ke-
tone or sulfoxide moiety. The distance between the sulfoxide that
contributes to inducer potency and the electrophilic reactive car-
bon center of sulforaphane is thus preserved in sulfoxythiocarba-
mate analogs.
The basic synthetic approach to the designed analogs is shown
in two related sequences outlined in Scheme S1 in SI Text.
Preparation of the sulfoxide-containing analogs, 4a–4r, began
by controlled oxidation of 4-hydroxybutyl-methyl thioether. Me- Fig. 2. Structures of sulfoxythiocarbamate sulforaphane analogs. Two
sylation followed by amine displacement, and thiocarbamylation classes of analogs, one with sulfoxide group (4a–4r) and another with
gave the penultimate precursor that could be oxidized with 3- ketone-group (8a–8f), were generated with alkyl or aryl modifications on
chloroperbenzoic acid or Oxone to generate the sulfoxythiocar- R1 and R2 positions.

Ahn et al. PNAS ∣ May 25, 2010 ∣ vol. 107 ∣ no. 21 ∣ 9591
 toxicity of the sulfoxythiocarbamates was also determined by to that of sulforaphane (33) (Fig. S4C). These results suggest that
measuring cell survival based on protein concentration (29). All glutathione can antagonize sulfoxythiocarbamate action by chem-
sulfoxythiocarbamate compounds tested were found to be indu- ical modification through thiol attack.
cers of NQO1, and dose-response curves were used to obtain CD We further examined the specific importance of Nrf2 and
values (Table 1) (26). As illustrated, benzyl substituents (4f–4j) in Keap1 in NQO1 induction by bioassays with 8f in mouse embry-
R1 have higher potencies when compared to an aliphatic alkyl onic fibroblasts (MEF) derived from wild-type, Nrf2-knockout, or
chain or phenyl group (4a–4e). Interestingly, benzyl isothio- Keap1/Nrf2-double knockout mice (Fig. S5). Like sulforaphane,
cyanate is also a more potent inducer than hexyl or phenyl iso- compound 8f stimulates NQO1 activity in wild-type (WT) MEF
thiocyanate (27). In addition, compounds with hydrophobic sub- in a dose-dependent manner (Fig. S5A). In sharp contrast, in
stitutions on sulfur, nitrogen, or both (4g–4i, 4o, 4r, and 8d) Nrf2-knockout and in Keap1/Nrf2-double knockout MEF, the ba-
tended to give the lowest CD values (2–6 μM). However, larger sal levels of NQO1 activity are much lower than in WT cells and
alkyl and bulky benzyl substituents on sulfoxythiocarbamate ana- are not affected by either sulforaphane or 8f. Furthermore, im-
logs (4i, 4q, 4r, and 8d) tended to increase the cellular toxicities munoblot analysis revealed that these differences in NQO1 en-
(LD50 ). We thus focused on a set of compounds (4f, 8a, 8f, and zyme activity between WT and knockout MEF correspond to
structures in Fig. S3D) that showed relatively high inducer poten- differences in protein levels (Fig. S5 B and C). Taken together
cies (low CD) and low toxicities (LD50 ). The chemoprotective in- with the ARE-reporter assays, these experiments demonstrate
dex (CI ¼ LD50 ∕CD) (26, 30) of these three compounds in that, in both human and murine cells, sulforaphane and its sul-
hepatoma cells was in the range of 8–20, within one order of sul- foxythiocarbamate analogs induce the expression of cytoprotec-
foraphane (CI ¼ 80), which while more potent, is also more toxic tive (phase 2) genes by targeting the Keap1/Nrf2/ARE pathway.
than some of the sulfoxythiocarbamates. NQO1 induction by 4f,
8a, and 8f was also analyzed in two other cell lines, human retinal Inhibition of Lipopolysaccharide-Induced Nitric Oxide Formation by
pigment epithelial cells (ARPE-19) (31) and murine keratino- Sulfoxythiocarbamate Analogs in Macrophage-Like Cells. In addition
cytes (PE) (32) (Table 1 and Fig. S3). We found that 8a (CD to inducing phase 2 enzymes, sulforaphane has demonstrated
1.5 μM) was essentially equipotent to sulforaphane (1.4 μM) antiinflammatory action including inhibition of the formation of
in ARPE-19 cells, and this sulfoxythiocarbamate was at least fi- reactive oxygen species in mouse peritoneal macrophage and
vefold less toxic than the natural product. Compounds 8f and 4f murine macrophage-like cell line (RAW264.7) (34). To explore
were just slightly less potent than 8a in this cell line. The results in the effects of the sulfoxythiocarbamates on macrophages, we uti-
the keratinocyte line showed that the CI values for 8a and sulfor- lized a lipopolysaccharide macrophage activation assay and mea-
aphane were essentially identical. Furthermore, the maximal de- sured nitric oxide (NO) levels in the presence of 8a and 4f. These
gree of NQO1 induction by 8a was greater than that of compounds blocked NO generation with IC50 values of 2.5 and
sulforaphane in keratinocytes (Fig. S3C). 8 μM, respectively, modestly less potent than sulforaphane in this
assay (IC50 0.5 μM) (Fig. S6).
Downloaded from https://www.pnas.org by 103.168.155.60 on October 17, 2023 from IP address 103.168.155.60.

Sulfoxythiocarbamate Analogs Work Through the Keap1/Nrf2/ARE

Pathway. To confirm that, like sulforaphane, the sulfoxythiocarba- Inducer Potency of Sulfoxythiocarbamate Analog 8a in Mouse Skin.
mates stimulate NQO1 via the antioxidant response element We next investigated the in vivo properties of sulfoxythiocarba-
(ARE) transcriptional pathway, we examined eight sulfoxythio- mate 8a in a mouse epidermis NQO1 induction assay. Topical
carbamate analogs by using a previously established ARE-lucifer- application of three doses of 8a at 24-h intervals led to fivefold
ase reporter stably transfected in MCF7 cells (AREc32) (33). induction of NQO1 in the dorsal skin (Fig. 3), with the level of
Each of these compounds led to a significant increase in the lu- induction at 0.5 μmol comparable to that of sulforaphane.
ciferase signal, and the most powerful effects were caused by 8f
and 4o as shown in Fig. S4A. A time course over three days of Labeling Keap1 by Sulfoxythiocarbamate Analog 8f in Cells. Since
luciferase induction with 6.25 μM 8f and 4o revealed that the sul- alkynyl sulfoxythiocarbamate analog 8f was shown to be a rela-
foxythiocarbamates showed a 20-fold maximal induction and sus- tively potent, efficacious, and nontoxic chemopreventive agent,
tained increase in the level of reporter, comparing favorably with comparable to sulforaphane, we elected to use it as a mechanistic
the effects of 2 μM sulforaphane (Fig. S4B). As expected, cellular probe for cellular Keap1. Whereas sulforaphane covalently modi-
depletion of glutathione with L-buthionine-sulfoximine en- fies specific cysteine residues of recombinant Keap1 in solu-
hanced luciferase induction by 8f and 4o in a fashion comparable tion (15, 19), this has not been demonstrated in intact cells.

Table 1. NQO1 induction potency and toxicity of sulfoxythiocarbamate analogs in three cell lines including murine hepatoma cells
(Hepa1c1c7), human retinal pigment epithelial cells (ARPE-19), and murine keratinocytes (PE)
Hepa1c1c7 ARPE-19 PE
Name CD, μM LD50 , μM Name CD, μM LD50 , μM Name CD, μM LD50 , μM Name CD, μM LD50 , μM Name CD, μM LD50 , μM
4a 94 >200 4j 7.3 >50 8a 5.4 115 4f 4.8 >200 4f 13 >200
4b 35 >200 4k 9.8 >50 8b 5.4 120 4i 3.8 21 8a 2.5 128*
4c 43 >200 4l ND — 8c 12 >50 4o 1.7 29 8b 2.8 131*
4d 64 >200 4m 12.6 >50 8d 2.1 12 8a 1.5 120* 8f 3.8 50
4e 43 >200 4n ND — 8e 7.1 >50 8b 1.8 123* SF 0.25 13*
4f 13 >200 4o 4.3 37 8f 5.5 43 8f 5 38
4g 5.3 >50 4p 8.2 >50 SF 0.23 18 SF 1.4 14*
4h 3.8 32 4q 7 12.5
4i 2.8 16 4r 2.6 10
Cells were grown in 96-well plates for 24 h and incubated with serial dilutions of each compound for 48 h. NQO1 activity and total protein concentrations
were determined in cell lysates by Prochaska method (26) and the BCA assay, respectively. The specific NQO1 activity or total protein concentrations were
plotted vs. compound concentrations with values of mean
SEM. (n ¼ 8). The concentrations that give a twofold increase of NQO1 activity or 50%
decrease of total protein concentrations when compared to control were determined as CD or LD50. Standard errors were <
20% for values shown.
ND, not detemined; less than twofold induction.
*Values were determined by extrapolation.

9592 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1004104107 Ahn et al.

 modification with the non-alkyne-containing inducer (8a), sulfor-
aphane, and the extremely potent triterpenoid inducer (TP225)
(Fig. 4B) (28). Treatment of FLAG-Keap1-transfected cells with
each of the three inducers followed by incubation with 8f led to
reduced labeling of FLAG-Keap1 compared with 8f by itself, as
shown in Fig. 4A. It is noteworthy that TP225 blocked FLAG-
Keap1 labeling approximately 1,000-fold more potently than
either 8a or sulforaphane, consistent with the known potency
of this powerful pentacyclic inducer (28).

Identification of Cysteine Residues of Keap1 Labeled by Sulfoxythio-

carbamate 8f. To identify specific modification site(s) in Keap1 la-
beled by 8f, immobilized Keap1 was biotinylated, eluted with
FLAG peptide, and subjected to exhaustive trypsin digestion. The
tryptic digest was exposed to avidin-coated beads, and the cap-
Fig. 3. Evaluation of sulfoxythiocarbamate analog for NQO1 enzyme induc-
tion on mouse skin. The back of each SKH-1 hairless mouse (n ¼ 3) was to-
tured peptides were eluted with acetonitrile and analyzed by
pically treated with three concentrations of 8a (indicated amounts in MALDI and separately by linear trap quadrupole (LTQ) mass
40 μL of 80% aq. acetone) and solvent only over ca. 2.5 cm2 area for three spectrometry. MALDI MS showed the appearance of two promi-
doses at 24-h intervals. Mice were euthanized 24 h after the last dose, and nent peaks not present in control samples that were in precise
dorsal skin was harvested. NQO1 specific activity was measured in superna- agreement with molecular weights for modified Cys 273 (m∕z
tant fractions of homogenates of skin sections treated with 8a or solvent 1569, C*HALTPR) and Cys 288 (m∕z 1790, C*EILQADAR)
(control). Means
SD are shown. peptides (Fig. 4C). Tandem MS analysis confirmed these assign-
ments and revealed further that Cys 613 (SGVGVAVTMEPC*R)
We transiently transfected FLAG-tagged Keap1 in HEK293 cells was also labeled by 8f (Fig. S7).
and showed that it could be efficiently immunoprecipitated with Transfected FLAG-Keap1 Cys151Ala, Cys273Ala, and Cys-
immobilized anti-FLAG antibody, irrespective of exposure of 288Ala proteins were studied as single mutants and in com-
binations to dissect the importance of these mutations in sulfoxy-
cells to 8f (bottom blot, Fig. 4A). Taking advantage of the click
thiocarbamate labeling. In transgenic mouse and cell transfection
reaction (35), we treated antibody-immobilized FLAG-Keap1
experiments (17, 37–39), each of these cysteines in Keap1 has
with biotin-azide (36) and then eluted it with FLAG peptide been shown to be important for repressing Nrf2 activity or con-
and subjected it to streptavidin blotting. As shown in Fig. 4A, cells ferring resistance to inducers. Whereas single and paired cysteine
treated with sulfoxythiocarbamate 8f produced a major and clean Keap1 mutants showed minimal or modestly diminished labeling
band at identical molecular weight to FLAG-Keap1, dependent
Downloaded from https://www.pnas.org by 103.168.155.60 on October 17, 2023 from IP address 103.168.155.60.

by 8f, the triple mutant showed a sharper reduction in modifica-

on treatment with 8f. This provides evidence of covalent modi- tion by 8f (Fig. 4D). This is consistent with the mass spectrometry
fication of transfected FLAG-Keap1 by 8f. To examine further findings and underscores that multiple cysteine sites are signifi-
the significance of this labeling, we attempted to compete the cant in the targeting of Keap1 by inducers.

BIOCHEMISTRY

Fig. 4. Sulfoxythiocarbamate analog conjugates on Keap1 cysteines as well as potential target proteins in cells. (A) Labeling FLAG-Keap1 in cells. HEK293 cells
transiently expressing FLAG-tagged Keap1 were treated with 8f for 0.5 h, or pretreated with 8a, sulforaphane, or TP225 for 0.5 h before incubation of 8f
for 0.5 h. FLAG-Keap1 was immunoprecipitated from cell lysates with anti-FLAG antibody, subjected to reaction with biotin azide and eluted with FLAG pep-
tide. Samples were immunoblotted with streptavidin (Top) and anti-FLAG antibody (Bottom). (B) Structure of compounds, 8f, 8a, sulforaphane, and TP225.
(C) MALDI-MS spectrum of FLAG-Keap1 tryptic digests labeled by 8f. Eluted FLAG-Keap1 was digested by trypsin, followed by incubation with avidin-coated
beads. Biotin-conjugated peptides were eluted with aq. acetonitrile and analyzed by MALDI mass spectrometry, which showed two peaks corresponding to
modified Cys 273 and Cys 288 peptides. (D) Labeling FLAG-Keap1 cysteine to alanine mutants. Cells transiently expressing FLAG-tagged WT or mutant Keap1
were treated with 8f for 0.5 h. FLAG-Keap1 proteins, eluted after immunoprecipitation with anti-FLAG antibody and conjugation with biotin azide, were
immunoblotted with streptavidin (Top) and anti-FLAG antibody (Bottom). (E) Schematic structure of Keap1 with five domains that include NTR, BTB, IVR,
DGR, and CTR. Cys 288 and Cys 613 modified peptides were found by LTQ mass analysis. (F) Labeling potential target proteins in cells. After incubation
of compounds as described in A, cell lysates were subjected to click reaction with biotin azide, and analyzed by SDS-PAGE and immunoblotting with strepta-
vidin HRP.

Ahn et al. PNAS ∣ May 25, 2010 ∣ vol. 107 ∣ no. 21 ∣ 9593
 Exploring the Target Proteins Labeled by Sulfoxythiocarbamate 8f in in cells. A number of mass spectroscopic studies have examined
Cells. To explore the potential scope of other protein targets of 8f, recombinant Keap1 labeling in vitro by dexamethasone mesylate
cells were treated with 8f for 30 min, and cell extracts were re- (15), iodoacetamide-biotin (40–43), and several other agents
acted with biotin azide and then blotted with streptavidin. As (44). Iodoacetamide-biotin has also been studied in cells (41,
shown in Fig. 4F, a large number of bands were detected in cells 42). There are major structural differences between sulforaphane
treated with 8f. Many of these bands could be effectively com- and these other agents, particularly with respect to the electro-
peted by 100 μM 8a and sulforaphane, and to a lesser degree with philic carbon, raising concerns about their mechanistic relevance
TP225 at 1 μM. These surprising results underscore that there are in predicting the behavior of the plant natural product. Moreover,
many potential pathways that can be influenced by these electro- the complexity of isolating and purifying recombinant Keap1 to-
philic compounds that could lead to complex pharmacology, not gether with the lack of a compelling bioassay to ensure its func-
readily recapitulated with single gene knockouts. tional integrity limit the potential significance of in vitro
We then employed a mass spectrometry-based proteomic ap- modification studies. Nevertheless, each of the sites identified
proach to identify candidate proteins modified by 8f. After treat- here in the cellular assays with 8f (Cys 273, Cys 288, and Cys
ing HEK293 cells with 8f, cell extracts were reacted with biotin 613) have been also identified in the in vitro dexamethasone me-
azide under click conditions, immobilized on streptavidin beads, sylate analyses (15). This convergence along with competitive re-
and subjected to trypsin followed by mass spectrometric analysis. duction in labeling by the established agents, sulforaphane and
Over 100 proteins were identified in this fashion (Table S1). TP225, as well as transgenic mouse studies highlighting the im-
Using immunoprecipitation-Western analysis (Fig. 5), we showed portance of Cys 273 and Cys 288 (38), strongly support the role of
that at least six of these appear to be bona fide cellular targets of these residues in mediating repressor activity. As pointed out
8f, including macrophage migration inhibitory factor (MIF), per- previously (15, 17), these cysteine side chains may have signifi-
oxiredoxin 3 (Prx3), histone acetyltransferase 1 (HAT1), A-kinase cantly increased chemical reactivity because of the presence of
anchoring protein 149 (AKAP149), KH-type splicing regulatory neighboring basic residues that can lower the pK a values of their
protein (KSRP), and thioredoxin (Trx) (Fig. 5A). We further interactive cysteine thiols (45). Left unsettled is the precise con-
showed that, like Keap1, labeling of MIF and AKAP149 by 8f sequence of these cysteine modifications on the interaction of
Keap1 with Nrf2. The current model that cysteine modification
can be blocked by pretreatment with sulforaphane or TP225,
of Keap1 influences protein-protein interactions in the Keap1-
whereas Prx3 labeling is not competed by these inducers (Fig. 5B).
Nrf2-Cullin-3 complex is consistent with the literature.
This is consistent with the concept that there are overlapping but
The structure-potency data in this paper show that the small
distinct pharmacodynamic profiles for 8f vs. other chemopreven-
variation of structural modifications on analogs (e.g., 8a and 8b)
tive inducers.
appears to give less significant difference in their potencies.
Discussion These data as well as the fact that many structurally distinct com-
Here we have developed a series of sulforaphane analogs that pounds are known to be inducers of NQO1 (14) may imply that
Downloaded from https://www.pnas.org by 103.168.155.60 on October 17, 2023 from IP address 103.168.155.60.

show promising chemoprotective properties and proved to be the electrophile sensor protein Keap1 has evolved to respond to a
useful in establishing Keap1 as a direct target of these agents wide range of compounds with less defined structural specificity
in order to facilitate the protective response of cells to exoge-
nous oxidants and electrophile toxicity. Nevertheless, structural
modifications of sulfoxythiocarbamate analogs prepared here dis-
played a range of potencies for NQO1 induction (CD 2.1–94 μM),
which suggest that there are important structural requirements
for phase 2 enzyme induction.
The distinctive relative inducing potentials of sulforaphane vs.
the sulfoxythiocarbamates in the three cell lines, murine hepato-
ma, retinal pigment epithelial, and murine keratinocytes, illus-
trate the necessity for caution in drawing conclusions about
chemoprotective properties from single cell assays. Such differ-
ential effects may be due to the specific nature of the ARE path-
way in individual cell types, but are perhaps more likely to be
related to the range of protein targets modified by these com-
pounds. As revealed here, sulfoxythiocarbamate analogs are con-
siderably less reactive electrophiles with thiols than is
sulforaphane, which might give a more restricted set of adducts
in cells. Nevertheless, the widespread labeling by 8f in HEK293
cells provides an indication of the many proteins that can be car-
bamylated by this agent, and it is difficult to imagine that some of
these will not have significant functional implications. Using a
proteomics approach, we have discovered at least six protein tar-
gets of 8f other than Keap1, MIF, AKAP149, Prx3, Trx, HAT1,
and KSRP. Inhibition of MIF, a well-established chemokine
(46), has previously been associated with a derivative of sulfor-
aphane, phenethyl isothiocyanate (47), and may be important
in blocking inflammation. Modification of AKAP149 by 8f could
Fig. 5. Validation of selected target proteins labeled by sulfoxythiocarba- influence protein kinase cell signaling, whereas modification of
mate analog 8f in cells. (A) Target proteins labeled by 8f. HEK293 cells were Prx3 and Trx could affect the antioxidant response. The identifi-
treated with 8f or vehicle for 2.5 h. (B) Competition by sulforaphane or TP225.
HEK293 cells were pretreated with sulforaphane, TP225, or vehicle for 0.5 h
cation and characterization of such protein targets of 8f and other
before incubation with 8f for 0.5 h. Each protein was immunoprecipitated agents facilitates our understanding of the chemoprotection mo-
from cell lysates, subjected to click reaction with biotin azide on beads. Eluted saic associated with specific compounds. Finally, these studies
samples were immunoblotted with streptavidin as well as protein specific further highlight the utility of the sulfoxythiocarbamate function-
antibodies. ality for proteomic and pharmacologic investigation.

9594 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1004104107 Ahn et al.

 Methods supernatant fractions, which were analyzed for protein concentration (BCA)
NQO1 Assays. Cells were grown for 24 h in 96-well plates (10,000 per well for assay and NQO1 activity. Experiments were in compliance with the National
Hepa1c1c7 and ARPE-19 cells; 30,000 per well for PE cells). Culture media Institutes of Health Guidelines and were approved by The Johns Hopkins Uni-
were then replaced with fresh media containing serial dilutions of com- versity Animal Care and Use committee.
pounds dissolved in acetonitrile (final 0.5%). After incubation for 48 h,
the spent culture media were discarded, and adhered cells were washed Other Methods. Detailed experimental procedures for synthesis of com-
three times with PBS and lysed with 0.08% digitonin (75 μL∕well). A fraction pounds, cell culture, transfection and protein labeling, and mass analysis
of the cell lysates (20 μL) was used for determining protein concentration by can be found in SI Text.
the bicinchoninic acid (BCA) assay. The remaining cell lysates were used for
measuring NQO1 activity by the Prochaska method (26).
ACKNOWLEDGMENTS. We thank M. Yamamoto (Tohoku University) and
T. W. Kensler (Johns Hopkins University) for the generous gift of the FLAG-
NQO1 Induction of SKH-1 Hairless Mouse Skin. Eight-week-old female SKH-1 Keap1 plasmid and for the mouse embryonic fibroblasts, M. B. Sporn,
hairless mice (n ¼ 3) were treated three times at 24-h intervals on their backs T. Honda, and G. Gribble (Dartmouth School of Medicine) for providing tri-
with 0, 0.5, 2, and 8 μmol of 8a dissolved in 80% aqueous (aq.) acetone terpenoid TP225, and J. D. Hayes (University of Dundee) for the antibody
(vol∕vol, 40 μL) over ca. 2.5 cm2 area. Mice were euthanized 24 h after against NQO1. This work was supported by National Institutes of Health
the final dose. Each treated segment of their dorsal skin was removed, pul- Grants CA094076, GM62437, U54 RR020839, and CA093780, American Cancer
verized in liquid N2 , and the resulting powder was homogenized in 10 vo- Society Grants IRG5800543 and RSG-07-157-01-CNE, the Lewis B. and Dorothy
lumes of 0.25 M sucrose in 10 mM Tris buffer (pH 7.4). After three freeze- Cullman Foundation, the Kaufman Foundation, the American Institute for
thaw cycles, centrifugation at 14,000 rpm and 4 °C for 30 min yielded clear Cancer Research, and Cancer Research UK (C20953/A10270)

1. Talalay P, Fahey JW (2001) Phytochemicals from cruciferous plants protect against 25. Prochaska HJ, Santamaria AB (1988) Direct measurement of NAD(P)H—quinone reduc-
cancer by modulating carcinogen metabolism. J Nutr 131:3027S–3033S. tase from cells cultured in microtiter wells—a screening assay for anticarcinogenic
2. Surh YJ (2003) Cancer chemoprevention with dietary phytochemicals. Nat Rev Cancer enzyme inducers. Anal Biochem 169:328–336.
3:768–780. 26. Fahey JW, Dinkova-Kostova AT, Stephenson KK, Talalay P (2004) The “Prochaska”
3. Zhang Y, Talalay P, Cho CG, Posner GH (1992) A major inducer of anticarcinogenic pro- microtiter plate bioassay for inducers of NQO1. Method Enzymol 382:243–258.
tective enzymes from broccoli–isolation and elucidation of structure. Proc Natl Acad 27. Posner GH, et al. (1994) Design and synthesis of bifunctional isothiocyanate analogs of
Sci USA 89:2399–2403. sulforaphane—correlation between structure and potency as inducers of anticarcino-
4. Juge N, Mithen RF, Traka M (2007) Molecular basis for chemoprevention by sulfora- genic detoxication enzymes. J Med Chem 37:170–176.
phane: A comprehensive review. Cell Mol Life Sci 64:1105–1127. 28. Dinkova-Kostova AT, et al. (2005) Extremely potent triterpenoid inducers of the phase
5. Zhang Y, et al. (1994) Anticarcinogenic activities of sulforaphane and structurally 2 response: Correlations of protection against oxidant and inflammatory stress. Proc
related synthetic norbornyl isothiocyanates. Proc Natl Acad Sci USA 91:3147–3150. Natl Acad Sci USA 102:4584–4589.
29. Prochaska HJ, Santamaria AB, Talalay P (1992) Rapid detection of inducers of enzymes
6. Kensler TW, et al. (2005) Effects of glucosinolate-rich broccoli sprouts on urinary levels
that protect against carcinogens. Proc Natl Acad Sci USA 89:2394–2398.
of aflatoxin-DNA adducts and phenanthrene tetraols in a randomized clinical trial in
30. Gerhauser C, et al. (1997) Cancer chemopreventive potential of sulforamate, a novel
He Zuo township, Qidong, People’s Republic of China. Cancer Epidem Biomar
analogue of sulforaphane that induces phase 2 drug-metabolizing enzymes. Cancer
14:2605–2613.
Res 57:272–278.
7. Cornblatt BS, et al. (2007) Preclinical and clinical evaluation of sulforaphane for
31. Gao X, Talalay P (2004) Induction of phase 2 genes by sulforaphane protects retinal
chemoprevention in the breast. Carcinogenesis 28:1485–1490.
Downloaded from https://www.pnas.org by 103.168.155.60 on October 17, 2023 from IP address 103.168.155.60.

pigment epithelial cells against photooxidative damage. Proc Natl Acad Sci USA
8. Conaway CC, et al. (2005) Phenethyl isothiocyanate and sulforaphane and their
101:10446–10451.
N-acetylcysteine conjugates inhibit malignant progression of lung adenomas induced
32. Yuspa SH, et al. (1986) Cultivation and characterization of cells derived from mouse
by tobacco carcinogens in A/J mice. Cancer Res 65:8548–8557. skin papillomas induced by an initiation promotion protocol. Carcinogenesis
9. Kong L, et al. (2007) Delay of photoreceptor degeneration in tubby mouse by 7:949–958.
sulforaphane. J Neurochem 101:1041–1052. 33. Wang XJ, Hayes JD, Wolf CR (2006) Generation of a stable antioxidant response
10. Talalay P, et al. (2007) Sulforaphane mobilizes cellular defenses that protect skin element-driven reporter gene cell line and its use to show redox-dependent activation
against damage by UV radiation. Proc Natl Acad Sci USA 104:17500–17505. of Nrf2 by cancer chemotherapeutic agents. Cancer Res 66:10983–10994.
11. Dinkova-Kostova AT, et al. (2006) Protection against UV-light-induced skin carcinogen- 34. Liu H, Dinkova-Kostova AT, Talalay P (2008) Coordinate regulation of enzyme markers
esis in SKH-1 high-risk mice by sulforaphane-containing broccoli sprout extracts. for inflammation and for protection against oxidants and electrophiles. Proc Natl
Cancer Lett 240:243–252. Acad Sci USA 105:15926–15931.
12. Dinkova-Kostova AT, et al. (2007) Induction of the phase 2 response in mouse and 35. Wang Q, et al. (2003) Bioconjugation by copper(I)-catalyzed azide-alkyne [3 þ 2]
human skin by sulforaphane-containing broccoli sprout extracts. Cancer Epidem Bio- cycloaddition. J Am Chem Soc 125:3192–3193.
mar 16:847–851. 36. Sun XL, Stabler CL, Cazalis CS, Chaikof EL (2006) Carbohydrate and protein immobi-
13. Kong L, et al. (2009) Molecular mechanisms underlying cochlear degeneration in the lization onto solid surfaces by sequential Diels-Alder and azide-alkyne cycloadditions.
tubby mouse and the therapeutic effect of sulforaphane. Neurochem Int 54:172–179. Bioconjugate Chem 17:52–57.
14. Dinkova-Kostova AT, Fahey JW, Talalay P (2004) Chemical structures of inducers of 37. Zhang DD, Hannink M (2003) Distinct cysteine residues in Keap1 are required for
nicotinamide quinone oxidoreductase 1 (NQO1). Method Enzymol 382:423–448. Keap1-dependent ubiquitination of Nrf2 and for stabilization of Nrf2 by chemopre-
15. Dinkova-Kostova AT, et al. (2002) Direct evidence that sulfhydryl groups of Keap1 are ventive agents and oxidative stress. Mol Cell Biol 23:8137–8151.
the sensors regulating induction of phase 2 enzymes that protect against carcinogens 38. Yamamoto T, et al. (2008) Physiological significance of reactive cysteine residues of
and oxidants. Proc Natl Acad Sci USA 99:11908–11913. keap1 in determining Nrf2 activity. Mol Cell Biol 28:2758–2770.
16. Dinkova-Kostova AT, et al. (2001) Potency of Michael reaction accepters as inducers of 39. Kobayashi M, et al. (2009) The antioxidant defense system Keap1-Nrf2 comprises a
enzymes that protect against carcinogenesis depends on their reactivity with sulfhy- multiple sensing mechanism for responding to a wide range of chemical compounds.
dryl groups. Proc Natl Acad Sci USA 98:3404–3409. Mol Cell Biol 29:493–502.
40. Eggler AL, et al. (2005) Modifying specific cysteines of the electrophile-sensing human
17. Wakabayashi N, et al. (2004) Protection against electrophile and oxidant stress by
Keap1 disrupt binding to the protein is insufficient to Nrf2 domain Neh2. Proc Natl
induction of the phase 2 response: Fate of cysteines of the Keap1 sensor modified
Acad Sci USA 102:10070–10075.
by inducers. Proc Natl Acad Sci USA 101:2040–2045.
41. Hong F, Sekhar KR, Freeman ML, Liebler DC (2005) Specific patterns of electrophile
18. Heiss E, et al. (2001) Nuclear factor kappa B is a molecular target for sulforaphane-
adduction trigger Keap1 ubiquitination and Nrf2 activation. J Biol Chem
BIOCHEMISTRY

mediated anti-inflammatory mechanisms. J Biol Chem 276:32008–32015.

280:31768–31775.
19. Hong F, Freeman ML, Liebler DC (2005) Identification of sensor cysteines in human
42. Rachakonda G, et al. (2008) Covalent modification at Cys151 dissociates the electro-
Keap1 modified by the cancer chemopreventive agent sulforaphane. Chem Res Toxicol
phile sensor Keap1 from the ubiquitin ligase CUL3. Chem Res Toxicol 21:705–710.
18:1917–1926.
43. Eggler AL, Luo Y, van Breemen RB, Mesecar AD (2007) Identification of the highly
20. Zipper LM, Mulcahy RT (2002) The Keap1 BTB/POZ dimerization function is required to reactive cysteine 151 in the chemopreventive agent-sensor Keap1 protein is meth-
sequester Nrf2 in cytoplasm. J Biol Chem 277:36544–36552. od-dependent. Chem Res Toxicol 20:1878–1884.
21. Li XC, Zhang D, Hannink M, Beamer LJ (2004) Crystal structure of the Kelch domain of 44. Luo Y, et al. (2007) Sites of alkylation of human Keap1 by natural chemoprevention
human Keap1. J Biol Chem 279:54750–54758. agents. J Am Soc Mass Spectr 18:2226–2232.
22. Zhang Y, Kolm RH, Mannervik B, Talalay P (1995) Reversible conjugation of isothiocya- 45. Snyder GH, Cennerazzo MJ, Karalis AJ, Field D (1981) Electrostatic influence of local
nates with glutathione catalyzed by human glutathione transferases. Biochem Bio- cysteine environments on disulfide exchange kinetics. Biochemistry 20:6509–6519.
phys Res Commun 206:748–755. 46. Cooke G, Armstrong ME, Donnelly SC (2009) Macrophage migration inhibitory factor
23. Zhang Y, Talalay P (1994) Anticarcinogenic activities of organic isothiocyanates— (MIF), enzymatic activity and the inflammatory response. Biofactors 35:165–168.
chemistry and mechanisms. Cancer Res 54:S1976–S1981. 47. Brown KK, et al. (2009) Direct Modification of the Proinflammatory Cytokine Macro-
24. Hwang Y, et al. (2007) A selective chemical probe for coenzyme A—Requiring phage Migration Inhibitory Factor by Dietary Isothiocyanates. J Biol Chem
enzymes. Angew Chem Int Edit 46:7621–7624. 284:32425–32433.

Ahn et al. PNAS ∣ May 25, 2010 ∣ vol. 107 ∣ no. 21 ∣ 9595