Small molecule probe suitable for in situ profiling and inhibition of protein disulfide isomerase

Article abstract of DOI:10.1021/cb4002602

Proper folding of cellular proteins is assisted by protein disulfide isomerases (PDIs) in the endoplasmic reticulum of mammalian cells. Of the at least 21 PDI family members known in humans, the 57-kDa PDI has been found to be a potential therapeutic target for a variety of human diseases including cancer and neurodegenerative diseases. Consequently, small molecule PDI-targeting inhibitors have been actively pursued in recent years, and thus far, compounds possessing moderate inhibitory activities (IC₅₀ between 0.1 and 100 μM against recombinant PDI) have been discovered. In this article, by using in situ proteome profiling experiments in combination with in vitro PDI enzymatic inhibition assays, we have discovered a phenyl vinyl sulfonate-containing small molecule (P1; shown) as a relatively potent and specific inhibitor of endogenous human PDI in several mammalian cancer cells (e.g., GI₅₀ ～ 4 μM). It also possesses an IC₅₀ value of 1.7 ± 0.4 μM in an in vitro insulin aggregation assay. Our results indicate P1 is indeed a novel, cell-permeable small molecule PDI inhibitor, and the electrophilic vinyl sulfonate scaffold might serve as a starting point for future development of next-generation PDI inhibitors and probes.

Full text of DOI:10.1021/cb4002602

Articles
pubs.acs.org/acschemicalbiology
Small Molecule Probe Suitable for In Situ Profiling and Inhibition of
Protein Disulfide Isomerase
Jingyan Ge,† Chong-Jing Zhang,† Lin Li,† Li Min Chong,† Xiaoyuan Wu,† Piliang Hao,‡ Siu Kwan Sze,‡
and Shao Q. Yao*,†
†Department of Chemistry, National University of Singapore, 3 Science Drive 3, Singapore 117543
‡School of Biological Sciences, Nanyang Technological University, 60 Nanyang Drive, Singapore 637551
S
* Supporting Information
ABSTRACT: Proper folding of cellular proteins is assisted by
protein disulfide isomerases (PDIs) in the endoplasmic
reticulum of mammalian cells. Of the at least 21 PDI family
members known in humans, the 57-kDa PDI has been found
to be a potential therapeutic target for a variety of human
diseases including cancer and neurodegenerative diseases.
Consequently, small molecule PDI-targeting inhibitors have
been actively pursued in recent years, and thus far, compounds
possessing moderate inhibitory activities (IC50 between 0.1
and 100 μM against recombinant PDI) have been discovered.
In this article, by using in situ proteome profiling experiments in combination with in vitro PDI enzymatic inhibition assays, we
have discovered a phenyl vinyl sulfonate-containing small molecule (P1; shown) as a relatively potent and specific inhibitor of
endogenous human PDI in several mammalian cancer cells (e.g., GI50 ∼ 4 μM). It also possesses an IC50 value of 1.7 0.4 μM in
an in vitro insulin aggregation assay. Our results indicate P1 is indeed a novel, cell-permeable small molecule PDI inhibitor, and
the electrophilic vinyl sulfonate scaffold might serve as a starting point for future development of next-generation PDI inhibitors
and probes.
Proper folding of cellular proteins is a highly complex process,
and the disulfide bond formation is one of the key steps, which
normally takes place in the endoplasmic reticulum (ER) with
the assistance of a class of enzymes called protein disulfide
isomerases (PDIs).1 PDIs catalyze the formation (oxidation),
breakage (reduction), and rearrangement (isomerization) of
disulfide bonds between cysteine residues within proteins as
they fold. This allows proteins to quickly find the correct
arrangement of disulfide bonds in their fully folded state.
Currently, there are at least 21 PDI family members known in
humans.2 One of the best-known human PDIs is PDIA1
(sometimes simply referred to as PDI), a 57-kDa protein
residing mostly in the ER, although it can also be released to
function at the cell surface or extracellular matrix. Similar to
most other family members of PDIs, PDI contains four
thioredoxin-like domains (a, b, b′ and a′), two of which possess
a canonical CGHC motif, giving rise to the two PDI active sites
within the a and a′ catalytic domains (Figure 1A). The other
two noncatalytic domains (b and b′) are essential for the
noncovalent binding of incompletely folded protein substrates.3
linked to apoptosis of cells in several neurological disorders
including Huntington’s disease, Alzheimer’s disease, and
Parkinson’s disease.5 PDI is involved in the cellular immune
response by facilitating the loading of antigenic peptides onto
MHC class I molecules.6,7 PDI has also been found to take part
in the breaking of bonds on the HIV gp120 protein, which is
required for HIV infection of lymphocytes and monocytes.8
Several studies have shown ER stress and unfolded protein
responses can activate PDI expression.9 Consequently, elevated
PDI expression levels have been observed in a variety of human
cancers.2,10
As a potentially promising druggable target, PDI in recent
years has received some attention in drug discovery.5,11−17
Bacitracin, a dodecapeptide antibiotic, was first reported as an
PDI inhibitor in 198111 and has been found to inhibit PDI
activities in a variety of cellular processes.10,12,13 With an active
concentration in the high micromolar range, this compound is a
fairly weak and ineffective PDI inhibitor.13 Its clinical use was
further hampered by its nephrotoxicity, low specificity, and low
cell permeability.14 Its relatively large size also makes it difficult
to be further modified synthetically. Juniferdin (Figure 1C), an
PDI inhibitor discovered through an high-throughput screening
(HTS) effort using collections of natural product libraries, was
The four cysteine residues in PDI (Cys53/56 and Cys397/400
)
have been shown to possess unusually high/low pKa values,
which is critical in maintaining the catalytic activities of PDI
and the cycling between the oxidized and reduced states
(Figure 1B).4 In recent years, PDI has been found to play key
roles in a wide range of physiological and disease processes.2 As
the key enzyme involved in protein folding, PDI was recently
Received: April 16, 2013
Accepted: September 16, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/cb4002602 | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
Figure 1. (A) Domain overview of human protein disulfide isomerase (PDI), with the catalytic a and a′ domains (in orange) and the catalytically
́
inactive b and b domains (in green). (B) Schematic showing of the PDI-catalyzed reaction. (C) Some known small molecule inhibitors of PDI, as
well as P1 (boxed), the newly discovered PDI inhibitor from the current study.
Figure 2. (A) In situ proteome reactivity profiles of various electrophilic probes in live MCF-7 cells. For detailed structures of the probes, see SI
Figure S1. Upon labeling by each probe (P1/16F16A, 5 μM; V1/V2/TR/TL/AZ, 20 μM), MCF-7 cells were lysed and clicked with Rh−N3,
separated by SDS-PAGE gels followed by in-gel fluorescence scanning (FL). The 57-kDa fluorescent band corresponding to human PDI was labeled
(*). (B) In situ labeling of MCF-7 cells by P1, followed by click chemistry with Rh-Biotin-N3, pulled-down (PD) by avidin agarose beads, then gel-
separated before FL and Western blotting (WB). Control pull-down (PD) was done with DMSO in place of P1. (*) The three enriched fluorescent
bands were identified as different human PDIs (PDI, PDIA4 and PDIA6) using the corresponding antibodies. (C) LC-MS/MS analysis of the three
fluorescent bands (labeled with *) to confirm their identities. (D) Concentration-dependent in vitro labeling of purified bovine PDI (25 ng) by P1.
SS = silver-stained gel. (E) In-gel fluorescence scanning showing both concentration- and time-dependent in situ labeling of MCF-7 cells by P1. (F)
Competitive in situ labeling of MCF-7 cells by P1 (5 μM), with and without pretreatment of 16F16.
found to block the PDI-catalyzed reduction of disulfide bonds
in the HIV-1 envelope glycoprotein gp120, thereby inhibiting
the entry of HIV-1 virus into cells.15 The compound showed an
IC50 value of 156 nM in an in vitro insulin aggregation assay.
16F16, another recently discovered PDI inhibitor, was found to
suppress apoptosis caused by misfolded proteins in a cell model
of Huntington‘s disease.5 It possesses an electrophilic
chloroacetamide structure that reacted covalently with one
(or more) of the four highly nucleophilic cysteine residues
within the PDI active sites. In an in vitro fluorescence self-
quenching assay that uses diabz-GSSG and purified PDI from
bovine liver, it exhibited a reported IC50 value of ∼70 μM.5,18 In
another study, a compound named PACMA 31, which contains
an electrophilic propynoic amide in its core structure, was
found to significantly inhibit the growth of ovarian cancer cells
through the irreversible inhibition of PDI activity. Subsequent
B
dx.doi.org/10.1021/cb4002602 | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
testing of the compound in vitro using recombinant PDI in an
insulin aggregation enzymatic assay revealed an IC50 value of
∼10 μM.16 Importantly, this compound also exhibited in vivo
antitumor activities without causing detectable toxicity to
normal tissues. While our manuscript was in preparation,
Weerapana and co-workers reported the discovery of RB-11-ca,
a trifunctionalized 1,3,5-triazine capable of covalently labeling
endogenous human PDI (presumably at the Cys53 residue) in
live HeLa cells.17 This compound also contains a chloroace-
tamide moiety and was found to possess similar in vitro
inhibitory activity as 16F16 against bovine PDI using the
fluorescence self-quenching assay, as mentioned above (e.g.,
IC50 values of between 30 and 50 μM).17 Moreover, RB-11-ca.
showed a GI50 value of 23.9 μM against HeLa cells in an MTT
antiproliferation assay.
We noticed the common structural features of 16F16,
PACMA 31 and RB-11-ca, all of which contain electrophilic
moieties (e.g., propynoic amide and chloroacetamide), and
were therefore highly susceptible to covalent attack by the
highly nucleophilic cysteine residues present in the PDI active
sites under physiological conditions.5,16,17 We wondered if
other pharmacophores possessing similar kinds of electrophilic
groups (e.g., vinyl sulfones,19,20 β-lactones,21,22 azanitriles,23,24
etc.), all of which are known irreversible inhibitors of cysteine-
reactive enzymes including cysteine proteases, lipases, and
protein tyrosine phosphatases (PTPs), could provide a
convenient chemical toolbox for identification of novel PDI
inhibitors. We noted that some other recently discovered,
biologically interesting compounds might serve the same
purpose,25−27 but they were not included in the current
study due to their inaccessibility (to us). Herein, we report the
serendipitous discovery of P1 (boxed compound in Figure 1C),
a phenyl vinyl sulfonate-containing small molecule, as a
relatively potent and specific inhibitor of endogenous human
PDI, capable of killing numerous mammalian cancer cells (GI50
∼4 μM). In vitro inhibition assay indicated that this compound
has an IC50 value of 1.7 μM against bovine PDI using insulin
aggregation assay and is 40-fold more potent than 16F16 under
identical assay conditions. These findings would place P1
among some of the more potent, cell-permeable small molecule
PDI inhibitors discovered to date.
exclusive reactivity against human PDI at 5 μM concentration.
A side-by-side comparison between P1 and 16F16A, which is
an alkyne-containing analog of 16F16 and had identical
inhibitory activities against PDI,5 showed that P1 was at least
as potent as, but more specific than, 16F16A in labeling
endogenous PDI under our assay conditions. Interestingly, β-
lactone probes such as TR and TL,21 as well as the azanitrile-
containing probe AZ,23 previously shown to strongly label
endogenous human fatty acid synthase (FAS) and cathepsin L
(a cysteine protease) in mammalian cells, respectively, failed to
label human PDI either. We therefore concluded P1 might
constitute a novel, cell-permeable small molecule inhibitor that
could covalently modify endogenous PDI potently and
selectively in human cancer cells. In order to unequivocally
establish the 57-kDa band labeled by P1 as endogenous human
PDI, we carried out pull-down (PD) experiments on the P1-
labeled MCF-7 proteome followed by Western blotting (WB)
and LC-MS/MS analysis (Figure 2B and C); both results
confirmed the 57-kDa band as human PDI. Two additional
bands were also significantly enriched from the PD experiment,
and they were subsequently identified as PDIA6 (48 kDa) and
PDIA4 (73 kDa). Both proteins were previously known
isoforms of human PDIs and possess similar active-site cysteine
residues in their catalytic domains.2 While PDIA4 is known to
be localized to ER, the plasma membrane, and mitochondria,
PDIA6 is localized predominantly to the nucleus of mammalian
cells.36 Further quantitative PD experiments showed that,
although the endogenous expression levels of PDIA4 and
PDIA6 were similar to that of PDI in MCF-7 cells, these two
proteins were not labeled as efficiently by P1 (SI Figure S9).
We next used purified bovine PDI, which shares >95%
sequence homology with human PDI, to establish the
sensitivity of P1 in labeling PDI in vitro (Figure 2D); as little
as 50−100 nM of P1 was sufficient to generate a strong
fluorescent band. We further confirmed this labeling reaction
was dependent upon PDI enzymatic activity, as small-molecule
additives such as H2O2, DTT, and cystamine, all of which were
previously known to modulate PDI enzymatic activity,5,37,38
appeared to up/down-regulate the labeling intensity (SI Figure
S2). We also directly labeled endogenous human PDI in situ in
concentration- and time-dependent manners (Figure 2E); as
low as 1−5 μM of P1 was able to specifically label human PDI
in as short as 15 min, indicating the PDI/P1 reaction was rapid
under physiological environments. Finally, in a competitive
labeling experiment (Figure 2F), it was shown that the addition
of excessive 16F16 to MCF-7 cells prior to P1 was able to
completely abolish PDI labeling, indicating that both 16F16
and P1 likely targeted the same active-site cysteine residue(s) in
human PDI.
RESULTS AND DISCUSSION
■
Screening. In order to identify compounds that directly
target human PDI in situ, we carried out screening in live MCF-
7 cells (a human breast cancer cell line) by using the well-
established activity-based protein profiling (ABPP) ap-
proach.28−31 This necessitates all compounds to contain a
small, chemically benign tag (e.g., a terminal alkyne) that
minimizes perturbation to bind PDI in situ and at the same time
provides a tractable tag for subsequent bioconjugation to
reporter-containing azides using click chemistry.32−35 Of the
numerous compounds screened, several representatives were
found to possess interesting in situ proteome reactivity profiles
(Figure 2A; see Figure 1C and Supporting Information (SI)
Figure S1); different vinyl sulfone/sulfonate-containing com-
pounds (P1, V1 and V2) had strikingly different reactivity
against endogenous human PDI (the 57-kDa fluorescent band
labeled with * in lane 1). V1 and V2, two previously reported,
highly potent irreversible inhibitors of parasitic/human cysteine
proteases,20 showed no apparent reactivity against human PDI
even at 20 μM probe concentrations, whereas P1 (an intended
PTP inhibitor; Figure 1C; vide infra) showed potent and almost
Structure−Activity Relationship of P1 Analogs. Our
newly discovered PDI-targeting compound P1 was originally
designed as a potential inhibitor of PTP1B (a human PTP
linked to diabetes and obesity; see Figure 3A);39 its phenyl
vinyl sulfonate moiety was expected to serve as a phosphotyr-
osine mimic that might irreversibly label the nucleophilic
cysteine residue in PTP1B active site.19 Therefore, its exquisite
in situ labeling profiles against human PDIs (PDI, PDIA4, and
PDIA6) in MCF-7 cells were unexpected. To ensure that P1
did not label endogenous PTP1B, we carried out labeling
experiments both in situ using MCF-7 cells and in vitro using
bovine PDI/recombinant PTP1B (Figure 3B and C); upon
pull-down enrichment of the in situ P1-labeled MCF-7
proteome, we were unable to detect any P1-labeled
C
dx.doi.org/10.1021/cb4002602 | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
coumarin analog (R1d), and reduced vinyl sulfonate (R1e).
These variables were expected to delineate the effect of the
electrophilic moiety in R1a toward the labeling reaction. The
Boc-(S)-Phe in P1 (e.g., R2a), another important structural
component of the probe for PDI recognition, was replaced with
four other aromatic acid building blocks (e.g., R2b to R2e).
Lastly, the terminal alkyne-containing, five-carbon “click” tag in
P1, R3a, was replaced with either a smaller terminal alkyne
linker (R3b) or biotin (R3c). In total, 10 additional compounds
(P2−P11) were synthesized in this first library using schemes
similar to what was shown in Figure 4A. We first compared the
in situ proteome reactivity profiles of these newly synthesized
probes in MCF-7 cells (Figure 4B). It became apparent that the
phenyl vinyl sulfonate in P1 was one of the most important
determinants for PDI labeling, as P3/P4/P5 did not show any
noticeable fluorescent band at ∼57 kDa (lanes 3−5). P5 was
structurally identical to P1, except with a reduced vinyl
sulfonate, which was incapable of being covalently modified by
cysteine residues in the PDI active sites. Both P3 and P4
contain a 2-fluoromethyl aromatic phosphate, which would
generate a highly reactive, electrophilic quinone methide
intermediate only upon endogenous PTP hydrolysis.40,41 It
was previously shown that similarly designed, amino acid-
containing probes were able to label many endogenous cellular
proteins, including human PDI, but with little specificity in
Hepatoma cells infected with Hepatitis C Virus (HCV).42 The
complete abolishment of PDI labeling by both P3 and P4
under our labeling conditions indicates that the presence of an
electrophilic group in these probes alone was insufficient to
achieve potent and specific in situ PDI targeting. Interestingly,
P2, a coumarin analog of P1, showed a similar, but overall
weaker, in situ labeling profile as P1, indicating the phenyl
group in R1a was optimized for PDI binding. Probes P6 to P10,
all of which showed a strongly fluorescent 57-kDa PDI band,
produced similar in situ labeling profiles as P1, albeit with
varying degrees of potency (judged by the fluorescent intensity
of the 57-kDa band) and selectivity (judged by the presence of
other fluorescent bands). This indicates both R2 and R3 groups
in P1 were nonessential elements, but provided additional
recognition for specific in situ PDI binding/labeling (vide infra).
We therefore concluded that a potent and selective PDI-
targeting probe should possess both PDI-binding and suitable
electrophilic moieties within the same molecular skeleton. We
next tested all probes, including P1-P11, 16F16, 16F16A, and
16F16DC (an 16F16 analog that does not contain the
electrophilic chloroacetyl group; see SI Figure S1), for their
inhibitory property against bovine PDI using an insulin
aggregation assay (inserted table in Figure 4A, and Figure
4C);5,16 P1 again emerged as one of the most potent inhibitors,
with an apparent IC50 value of 1.7 0.4 μM. Interestingly, in a
side-by-side comparison, 16F16 was at least 40 times weaker in
PDI inhibition (e.g., IC50 = 74 6 μM) than P1 under identical
assay conditions (see Figure 4C). The other two 16F16
analogs, 16F16A and 16F16DC, also showed weak (IC50 = 108
7 μM) or no inhibition against bovine PDI, respectively. Not
surprisingly, P3/P4/P5 showed minimal PDI inhibition even at
100 μM concentration, whereas P2 and P6−P11 showed IC50
values ranging from 1.1 to 5.3 μM. For example, P11, the biotin
analog of P1, showed a slightly better in vitro inhibition than P1
(IC50 = 1.1 0.8 μM). This might be explained by its strong
“avidity”, due to the biotin in its structure, in nonspecific
binding to bovine PDI in vitro. Whether or not the same
phenomenon can be observed in live mammalian cells will need
Figure 3. (A) Structure of a known PTP1B inhibitor from which P1
was originally designed.39 The group shown in red is a non-
hydrolyzable bioisostere of phosphotyrosine. (B) In-gel fluorescence
scanning (top) and Western blotting (bottom) of in situ P1 (10 μM)-
labeled MCF-7 cells, showing that endogenous PTP1B was not a
target of P1. “−”: negative control with cells treated with DMSO in
place of P1. PD = pull-down sample of P1-treated MCF-7 proteome.
In the WB (bottom), no PD was done on the negative control. The
band labeled with an arrow indicates the endogenous PTP1B. (C) In
vitro P1 (100 nM) labeling of purified PTP1B in the presence of a
fixed amount of bovine PDI (20 ng).
endogenous PTP1B by Western blotting using anti-PTP1B
antibody (lane 3 in bottom gel of Figure 3B). This was despite
the fact that PTP1B was highly expressed in MCF-7 cells (lane
1). In the in vitro experiments with purified proteins (Figure
3C), P1 started to label PTP1B only when this protein was
present in a much larger excess (>10×) than bovine PDI. On
the basis of these experiments, we concluded that P1 indeed
did not label endogenous PTP1B in live MCF-7 cells.
In an effort to explore the structure−activity relationship
(SAR) of P1 in labeling/inhibiting endogenous human PDI, we
next made two small libraries of P1 analogs (Figure 4 and Table
1). We first made systematic modifications of the three key
structural components, R1, R2, and R3, in P1 (Figure 4A).
Docking results of P1 binding to one of the catalytic domains in
human PDI (a′) indicated that the phenyl ring in the R1 group
of P1 engaged in π−π interaction with PDI’s Trp396 (insets in
Figure 4A). In addition, the vinyl sulfonate in P1 was projected
into a cavity near Cys397 of PDI, therefore positioning itself in
close proximity to initiate a covalent reaction. To further
support possible modification of P1 at Cys397/400, we carried out
LC-MS/MS protein sequencing experiments (SI Figure S10);
the results showed >84% sequence coverage with the in situ P1-
labeled human PDI (post-PD), and the only missing peptide
being the one containing Cys397/400. This indicates either one of
these two residues was modified by P1 as predicted by our
docking results. Docking results indicated, however, that Cys397
was closer in distance to P1 than Cys400. Finally, cysteine-to-
alanine mutants of human PDI transiently overexpressed in
HEK293 cells were labeled by P1 (SI Figure S10); the results
unambiguously confirmed that Cys397 was indeed the site for
P1 modification. Subsequently, we replaced the phenyl vinyl
sulfonate moiety in P1 (e.g., R1a) with coumarin vinyl sulfonate
(R1b), 2-fluoromethyl phenyl phosphate (FMPP, R1c) and its
D
dx.doi.org/10.1021/cb4002602 | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
Figure 4. (A) Representative scheme showing the synthesis and structure of P1 and its analogs (P2 to P11; summarized in the inserted table), which
contain three points of diversity (R1, R2, and R3). The full structure of P1 is shown. For complete structures of other analogs, see SI Table S1. Insets:
docked structure of P1 binding to the a′-domain of human PDI (left), and a close-up view (right) showing the proximity of the phenyl vinyl
sulfonate group of P1 and Cys397 in human PDI. The structure of PDI was taken from the Protein Data Bank (3UEM). (B) In situ proteome
reactivity profiles of P1 to P10 (5 μM each) against MCF-7 cells. For P3 and P4, upon addition of the probes, cells were first UV-irradiated (2 min)
to uncage the photolabile group. For P11 profiles, see SI Figure S5. (C) Insulin aggregation assay showing IC50 inhibition curves of representative
compounds (P1, P5, 16F16, and 16F16DC; left) and the corresponding values (right). IC50 values of PACMA 31 and RB-11-ca. were extracted from
reported literatures.16,17 For IC50 value of other compounds, see the inserted table in part A. For other IC50 curves, see SI Figure S6.
to be further investigated. We therefore concluded that, at least
under in vitro settings, P1 was clearly more potent than 16F16,
and, to our knowledge, one of the more potent small molecule
inhibitors reported to date, against recombinant PDI.
of P1 against six different cancer cell lines. Weerapana et al.
recently reported an GI50 value of 23.9 μM for RB-11-ca.
against HeLa cells in an MTT antiproliferation assay.17 The
GI50 values for PACMA 31 was reported to be between 0.3 and
1.4 μM in various cancer cell lines. The exact GI50 value for
16F16, however, was not available, based on published data.5,16
Again, in order to confirm PDI targeting by the probes (P1 and
16F16A), and ensure accurate comparison of the tumor cell-
killing activities of P1 versus 16F16, both the in situ proteome
profiles and the XTT antiproliferation activity of the probes
were simultaneously determined (Figure 5A and B); in situ
proteome reactivity profiles of P1 in all six cell lines showed
predominantly PDI labeling with some varying degrees of
fluorescence, likely indicating the differences in the endogenous
expression level of PDI in these cell lines. 16F16A, on the other
hand, continued to show strong labeling of several other
cellular proteins in addition to PDI, in all cell lines tested.
Interestingly, P1 and 16F16 showed comparable, low-micro-
molar inhibition of cell proliferation against all six cancer cells
(Figure 5B), with apparent GI50 values of ∼4 μM for both
compounds. This finding is in contrast with our earlier in vitro
results from the insulin aggregation assay, in which 16F16
showed comparatively weaker PDI inhibition. Our earlier in situ
In order to further delineate the SAR of P1 in its inhibition
against PDI, a secondary small molecule library was synthesized
(Table 1), containing 14 compounds (P12 to P25) with
structural variations at the R2 position. These compounds were
again tested for their inhibition against bovine PDI using the
insulin aggregation assay; the results indicate the IC50 values of
all compounds were similar to P1 (e.g., raging between 1.0 to
5.7 μM), which further confirms that even dramatic changes
such as replacement of the chiral amino acid at the R2 position
in P1 had marginal effects on the compound’s overall inhibition
against PDI in vitro.
Antiproliferation Activity of P1 in Different Cancer
Cell Lines. Having successfully confirmed that P1 as an in vitro
PDI inhibitor was more potent than 16F16 under identical
insulin aggregation assay conditions (e.g., > 40 times), and
approximately 10 and 30 times more potent than PACMA 31
and RB-11-ca, respectively, based on previously reported
values,16,17 we next determined its cellular activities by
measuring the GI50 (e.g., inhibition of 50% cell growth) values
E
dx.doi.org/10.1021/cb4002602 | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
due to its cellular targeting of other yet-to-be-identified cellular
proteins. It is also possible that 16F16 inhibited endogenous
PDI more effectively under native cellular environments.
To further delineate specific cellular targeting of PDI by P1/
16F16, we used confocal fluorescence microscopy to visualize
the colocalization of P1/16F16A with endogenous PDI in live
MCF-7 cells (Figure 5C); upon treatment with the probe (1
μM), the labeled cells were fixed, followed by click chemistry
with Rh-PEG-N3, then imaged (pseudocolored in red; P1,
panels 2 and 6; 16F16A, panels 10 and 14). The same cells
were subsequently treated with anti-PDI antibodies following
standard immunofluorescence (IF) protocols and imaged again
(pseudocolored in green; panels 3 and 11). Merged images
indicated that most of the fluorescence signals arisen from the
P1/16F16A labeling indeed colocalized well with those from
ER-localized human PDI (panels 4 and 12), as one might have
expected. In addition, weaker but clear fluorescence signals
from P1/16F16A were also detected within the cell nuclei. For
P1, these imaging results appeared to be consistent with our
earlier findings that P1 labeled not only endogenous human
PDI but also PDIA4 and the nucleus-localized PDIA6 (Figure
2B). For 16F16, on the other hand, our earlier in situ proteome
profiling results had already confirmed the presence of other
unknown cellular targets. The question of whether or not these
unknown targets include other human PDIs such as PDIA6,
however, remains to be further investigated. Notwithstanding,
the potency and specificity of P1 against human PDI both in
vitro and in living cells indicate this probe might be further
developed into a suitable small molecule-based, PDI-imaging
agent.43−45
Table 1. Structures of R2-focused Library (P12-P25) and the
Corresponding IC50 Values
proteome profiling results (Figure 5A) pointed to one possible
explanation, that is, the improvement from in vitro-PDI
inhibition to cell-proliferation inhibition by 16F16 might be
Conclusion. We have discovered a phenyl vinyl sulfonate-
containing small molecule probe, P1, which showed relatively
potent and selective covalent labeling of endogenous human
Figure 5. (A) In situ proteome profiling of P1 and 16F16A (5 μM) against six different cancer cell lines. (B) GI50 values of P1 and 16F16 against six
cancer cell lines. See SI Figure S7 for original GI50 plots. (C) Confocal microscopy showing the cellular localization of P1 (panels 1−8) and 16F16A
(panels 9−16).
F
dx.doi.org/10.1021/cb4002602 | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
incubated with 1 μM P1 or 16F16A in fresh growth medium (400 μL).
The cells were further incubated for 1 h at 37 °C/CO2. Then, the cells
were washed with PBS three times. Subsequently, cells were fixed with
3.7% formaldehyde in PBS for 20 min at 37 °C/CO2, washed twice
again, and permeabilized with 0.1% Triton X-100 in PBS for 15 min,
then washed twice again. Subsequently, cells were treated with a
freshly premixed click chemistry solution (150 μL; 2 μM Rh-PEG-N3,
4 μM TBTA, 40 μM TCEP, 40 μM CuSO4) for 2 h at RT with gentle
shaking. Cells were washed with 2× PBS, several times PBS containing
0.1% Tween-20 and 0.5 mM EDTA, and 2× PBS. Cells were then
blocked with 2% BSA, 0.05% Tween-20 in PBS for 1 h. For ER-
localization experiments, cells were further incubated with ER tracker
Green (Invitrogen, 0.3 μM final concentration) for 1 h and washed
with PBS twice. For immunofluorescence experiments, cells were
further incubated with PDI primary antibody (1:200 in 2% BSA, Santa
Cruz, sc-166474) for 1 h at RT, washed once with 2% BSA, washed
twice with PBS, and then incubated with FITC-conjugated antimouse
IgG secondary antibody (1:100 in 2% BSA, Santa Cruz, sc-2010), then
washed once with 2% BSA and twice with PBS. For both experiments
the cells were incubated with nucleus stain (Hoechst, 0.2 μg mL−1 in
final concentration) for 20 min and washed twice with PBS. Finally,
the cells were washed and imaged.
PDI in live mammalian cells. Preliminary findings indicate P1
was a fairly effective inhibitor when compared to some of the
recently discovered PDI inhibitors.5,14 We further showed P1
may be a useful imaging agent to visualize endogenous human
PDI in mammalian cells. During the course of our study, we did
side-by-side detailed comparison of P1 and 16F16, a compound
previously shown to suppress apoptosis in a model of
Huntington’s disease, presumably through the inhibition of
PDI.5 We found that, while both P1 and 16F16 possessed
comparable antiproliferation activities in numerous cancer lines,
P1 was more potent than 16F16 in inhibiting in vitro PDI
activities. This, together with the finding that P1 labeled
endogenous human PDI more specifically than 16F16A (an
16F16 analog) in live mammalian cells, points to the possibility
that the cell-based biological activities displayed by 16F16, in
contrast to previous report,5 might have originated in part from
other unknown cellular targets. Further studies, however, are
needed to substantiate our speculation. The use of electrophilic
chemical scaffolds demonstrated in the current study may
provide a general concept to target other human PDIs in future.
ASSOCIATED CONTENT
■
METHODS
■
S
* Supporting Information
Synthesis. The synthesis of all compounds used in this study is
reported in the Supporting Information.
General Procedures of In Situ Proteome Profiling, Target
Validation by PD, WB, and LC-MS/MS. These protocols are
modified based on previously reported procedures21,44 and are
provided in the Supporting Information.
Other relevant experimental sections, characterizations of new
compounds, and supplementary biological results. This material
is available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Bovine PDI Experiments. For labeling of PDI with the probes,
bovine PDI (25 ng) was incubated with different concentrations of a
probe for 2 h. After incubation, click cocktail reagents were added
(with Rh−N3). The mixtures were gently shaken for 2 h. SDS-loading
buffer was then added, and the solution was heated for 10 min at 95
°C. The proteins were resolved in a 10% SDS-PAGE gel, and the
labeled bands were visualized by in-gel fluorescence scanning. To
determine the effect of additives, PDI (100 ng) was treated with three
different concentrations of DTT, H2O2, and cystamine for 1 h first in
PBS buffer, respectively. P1 (1 μM final concentration) was then
added. All subsequent steps were the same as above-described. For
PTP1B-competitive labeling experiments, different amounts of
recombinant PTP1B were mixed with 20 ng PDI in Tris buffer.
After 30 min, 100 nM P1 was added, followed by click reaction. All
subsequent steps were the same as above-described.
For PDI inhibition, the assay was carried out in 384-well plates
according to literature procedures.46 Each well contained 100 mM
sodium phosphate and 0.2 mM EDTA, pH 7.0. PDI (10 ng) was first
incubated with different concentration of probes (4% DMSO) in 20
μL buffer for 30 min at 37 °C. Then, insulin (0.16 mM final
concentration) and dithiothreitol (1 mM final concentration) were
added. The enzyme reaction was monitored at 650 nm on a Bioteck
microplate reader. Experiments with 16F16 were done similarly side-
by-side.
XTT Cell Proliferation Assay. Mammalian cell lines were seeded
in 96-well plates at a concentration of 4000 cells per well. The cells
were grown for 24 h before treatment with P1 (or 16F16). After
removing the growth medium by suction, different concentrations of
P1 in growth medium (100 μL, 1% DMSO) were added and the cells
were incubated for 3 d at 37 °C/5% CO2. Experiments with 16F16
were done similarly side-by-side. Subsequently, 50 μL of XTT reagent
(1 mg mL−1; Invitrogen) and PMS (0.025 mM; Sigma) were added,
and the mixture was incubated at 37 °C/5% CO2 for 6 h. The
absorbance was measured at 450 nm, and background absorbance was
measured at 650 nm using a Bioteck plate reader. Cells incubated with
1% DMSO served as positive control. Experiments were conducted in
duplicate.
Corresponding Author
*E-mail: chmyaosq@nus.edu.sg.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Financial support was provided by National Medical Research
Council (NMRC/1260/2010) and the Ministry of Education
(MOE2012-T2-1-116) of Singapore. We also acknowledge the
financial support from the Singapore−Peking−Oxford Research
Enterprise (COY-15-EWI-RCFSA/N197-1). We thank E.
Weerapana (Boston College, U.S.A.) for the generous gifts of
the mammalian expression constructs of human PDI and
mutants.
REFERENCES
■
(1) Ellgaard, L., and Ruddock, L. W. (2005) The human protein
disulphide isomerase family: Substrate interactions and functional
properties. EMBO Rep. 6, 28−32.
(2) Benham, A. M. (2012) The protein disulfide isomerase family:
Key players in health and disease. Antioxid. Redox Signal. 16, 781−789.
(3) Nguyen, V. D., Wallis, K., Howard, M. J., Haapalainen1, A. M.,
Salo, K. E. H., Saaranen1, M. J., Sidhu, A., Wierenga1, R. K., Freedman,
R. B., Ruddock, L. W., and Williamson, R. A. (2008) Alternative
conformations of the x region of human protein disulphide-isomerase
modulate exposure of the substrate binding b′ domain. J. Mol. Biol.
383, 1144−1155.
(4) Karala, A.-R., Lappi, A.-K., and Ruddock, L. W. (2010)
Modulation of an active-site cysteine pKa) allows PDI to Act as a
catalyst of both disulfide bond formation and isomerization. J. Mol.
Biol. 396, 883−892.
(5) Hoffstrom, B. G., Kaplan, A., Letso, R., Schmid, R. S., Turmel, G.
J., Lo, D. C., and Stockwell, B. R. (2010) Inhibitors of protein disulfide
isomerase suppress apoptosis induced by misfolded proteins. Nat.
Chem. Biol. 6, 900−906.
Live-Cell Imaging Experiments. MCF-7 cells were seeded in
glass bottom dishes and grown to ∼60% confluence. The cells were
G
dx.doi.org/10.1021/cb4002602 | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
(6) Lee, S. O., Cho, K., Cho, S., Kim, I., Oh, C., and Ahn, K. (2010)
Protein disulphide isomerase is required for signal peptide peptidase-
mediated protein degradation. EMBO J. 29, 363−375.
(7) Dong, G., Wearsch, P. A., Peaper, D. R., Cresswell, P., and
Reinisch, K. M. (2009) Insights into MHC class I peptide loading from
the structure of the tapasin-ERp57 thiol oxidoreductase heterodimer.
Immunity 30, 21−32.
(8) Gallina, A., Hanley, T. M., Mandel, R., Trahey, M., Broder, C. C.,
Viglianti, G. A., and Ryser, H. J. (2002) Inhibitors of protein disulfide
isomerase prevent cleavage of disulfide bonds in receptor-bound
glycoprotein 120 and prevent HIV-1 entry. J. Biol. Chem. 277, 50579−
50588.
(9) Haefliger, S., Klebig, C., Schaubitzer, K., Schardt, J., Timchenko,
N., Mueller, B. U., and Pabst, T. (2011) Protein disulfide isomerase
blocks CEBPA translation and is up-regulated during the unfolded
protein response in AML. Blood 117, 5931−5940.
(10) Goplen, D., Wang, J., Enger, P. O., Tysnes, B. B., Terzis, A. J. A.,
Laerum, O. D., and Bjerkvig, R. (2006) Protein disulfide isomerase
expression is related to the invasive properties of malignant glioma.
Cancer Res. 66, 9895−9902.
(11) Roth, R. A. (1981) BacitracinAn inhibitor of the insulin
degrading activity of glutathione-insulin transhydrogenase. Biochem.
Biophys. Res. Commun. 98, 431−438.
(12) Ryser, H. J., Levy, E. M., Mandel, R., and DiSciullo, G. J. (1994)
Inhibition of human immunodefficiency virus infection by agents that
interfere with thiol-disulfide interchange upon virus−receptor
interaction. Proc. Natl. Acad. Sci. U.S.A. 91, 4559−4563.
(13) Dickerhof, N., Kleffmann, T., Jack, R., and McCormick, S.
(2011) Bacitracin inhibits the reductive activity of protein disulfide
isomerase by disulfide bond formation with free cysteines in the
substrate-binding domain. FEBS J. 278, 2034−2043.
(14) Karala, A. R., and Ruddock, L. W. (2010) Bacitracin is not a
specific inhibitor of protein disulfide isomerase. FEBS J. 277, 2454−
2462.
(15) Khan, M. M., Simizu, G. S., Lai, N. S., Kawatani, M., Shimizu, T.,
and Osada, H. (2011) Discovery of a small molecule PDI inhibitor that
inhibits reduction of HIV-1 envelope glycoprotein gp120. ACS Chem.
Biol. 6, 245−251.
(16) Xu, S., Butkevich, A. N., Yamada, R., Zhou, Y., Debnath, B.,
Duncan, R., Zandi, E., Petasis, N. A., and Neamati, N. (2012)
Discovery of an orally active small-molecule irreversible inhibitor of
protein disulfide isomerase for ovarian cancer treatment. Proc. Natl.
Acad. Sci. U.S.A. 109, 16348−16353.
(17) Banerjee, R., Pace, N. J., Brown, D. R., and Weerapana, E.
(2013) 1,3,5-Triazine as a modular scaffold for covalent inhibitors with
streamlined target identification. J. Am. Chem. Soc. 135, 2497−2500.
(18) Raturi, A., Vacratsis, P. O., Seslija, D., Lee, L., and Mutus, B.
(2005) A direct, continuous, sensitive assay for protein disulphide
isomerase based on fluorescence self-quenching. Biochem. J. 391, 351−
357.
(19) Liu, S., Zhou, B., Yang, H. H., He, Y., Jiang, Z. X., Kumar, S.,
Wu, L., and Zhang, Z. Y. (2008) Aryl vinyl sulfonates and sulfones as
active site-directed and mechanism-based probes for protein tyrosine
phosphatases. J. Am. Chem. Soc. 130, 8251−8260.
(20) Yang, P. -Y., Wang, M., He, C. Y., and Yao, S. Q. (2012)
Proteomic profiling and potential cellular target identification of
K11777, a clinical cysteine protease inhibitor, in Trypanosoma brucei.
Chem. Commun. 48, 835−837.
(21) Yang, P. -Y., Liu, K., Ngai, M. H., Lear, M. J., Wenk, M. R., and
Yao, S. Q. (2010) Activity-based proteome profiling of potential
cellular targets of orlistatAn FDA-approved drug with anti-tumor
activities. J. Am. Chem. Soc. 132, 656−666.
(22) Bottcher, T., and Sieber, S. A. (2008) β-Lactones as privileged
structures for the active-site labeling of versatile bacterial. Angew.
Chem., Int. Ed. 47, 4600−4603.
(24) Loser, R., Frizler, M., Schilling, K., and Gutschow, M. (2008)
Azadipeptide nitriles: Highly potent and proteolytically stable
inhibitors of papain-like cysteine proteases. Angew. Chem., Int. Ed.
47, 4331−4334.
(25) Wirth, T., Schmuck, K., Tietze, L. F., and Sieber, S. A. (2012)
Duocarmycin analogues target aldehyde dehydrogenase 1 in lung
cancer cells. Angew. Chem., Int. Ed. 51, 2874−2877.
(26) Kaschani, F., Clerc, J., Krahn, D., Bier, D., Hong, T. N.,
Ottmann, C., Niessen, S., Colby, T., van der Hoorn, R. A. L., and
Kaiser, M. (2012) Identification of a selective, activity-based probe for
glyceraldehyde 3-phosphate dehydrogenases. Angew. Chem., Int. Ed. 51,
5230−5233.
(27) Pace, N. J., Pimental, D. R., and Weerapana, E. (2012) An
inhibitor of glutathione S-transferase omega 1 that selectively targets
apoptotic cells. Angew. Chem., Int. Ed. 51, 8365−8368.
(28) Evans, M. J., and Cravatt, B. F. (2006) Mechanism-based
profiling of enzyme families. Chem. Rev. 106, 3279−3301.
(29) Uttamchandani, M., Li, J., Sun, H., and Yao, S. Q. (2008)
Activity-based profiling: New developments and directions in
functional proteomics. ChemBioChem 9, 667−675.
(30) Fonovic, M., and Bogyo, M. (2008) Activity-based probes as a
tool for functional proteomic analysis of proteases. Exp. Rev. Proteomics
5, 721−730.
(31) Heal, W. P., Dang, T. H. T., and Tate, E. W. (2011) Activity-
based probes: Discovering new biology and new drug targets. Chem.
Soc. Rev. 40, 246−257.
(32) Sletten, E. M., and Bertozzi, C. R. (2009) Bioorthogonal
chemistry: Fishing for selectivity in a sea of functionality. Angew.
Chem., Int. Ed. 48, 6974−6998.
(33) Kalesh, K. A., Shi, H., Ge, J., and Yao, S. Q. (2010) The use of
click chemistry in the emerging field of catalomics. Org. Biomol. Chem.
8, 1749−1762.
(34) Kolb, H. C., and Sharpless, K. B. (2003) The growing impact of
click chemistry on drug discovery. Drug Discovery Today 8, 1128−
1137.
(35) Meldal, M., and Tornøe, C. W. (2008) Cu-catalyzed azide-
alkyne cycloaddition. Chem. Rev. 108, 2952−3015.
(36) http://www.hprd.org (accessed March 15, 2013).
(37) Lundstrom, J., and Holmgren, A. (1990) Protein disulfide
̈
isomerase is a substrate for thioredoxin reductase and has thioredoxin-
like activity. J. Biol. Chem. 265, 9114−9120.
(38) Darby, N. J., Freedman, R. B., and Creighton, T. E. (1994)
Dissecting the mechanism of protein disulfide isomerase catalysis of
disulfide bond formation in a model peptide. Biochemistry 33, 7937−
7947.
(39) Liljebris, C., Larsen, S. D., Ogg, D., Palazuk, B. J., and Bleasdale,
J. E. (2002) Investigation of potential bioisosteric replacements for the
carboxyl groups of peptidomimetic inhibitors of protein tyrosine
phosphatase 1B: Identification of a tetrazole-containing inhibitor with
cellular activity. J. Med. Chem. 45, 1785−1798.
(40) Kalesh, K. A., Tan, L. P., Liu, K., Gao, L., Wang, J., and Yao, S.
Q. (2010) Small molecule probes that target Abl kinase. Chem.
Commun. 46, 589−591.
(41) Ge, J., Li, L., and Yao, S. Q. (2011) A self-immobilizing and
fluorogenic unnatural amino acid that mimics phosphotyrosine. Chem.
Commun. 47, 10939−10941.
(42) Blais, D. R., Brulotte, M., Qian, Y., Belanger, S., Yao, S. Q., and
Pezacki, J. P. (2010) Activity-based proteome profiling of hepatoma
cells during hepatitis C virus replication using protease substrate
probes. J. Proteome Res. 9, 912−923.
(43) Yang, K. S., Budin, G., Reiner, T., Vinegoni, C., and Weissleder,
R. (2012) Bioorthogonal imaging of aurora kinase A in live cells.
Angew. Chem., Int. Ed. 51, 6598−6603.
(44) Shi, H., Zhang, C. -J., Chen, G. Y. J., and Yao, S. Q. (2012) Cell-
based proteome profiling of potential dasatinib targets by use of
affinity-based probes. J. Am. Chem. Soc. 134, 3001−3014.
(45) Li, L., Ge, J., Wu, H., Xu, Q. -H., and Yao, S. Q. (2012)
Organelle-specific detection of phosphatase activities with two-photon
(23) Yang, P. -Y., Wang, M., Li, L., Wu, H., He, C. Y., and Yao, S. Q.
(2012) Design, synthesis, and biological evaluation of potent
azadipeptide nitrile inhibitors and activity-based probes as promising
anti-Trypanosoma brucei agents. Chem.Eur. J. 18, 6528−6541.
H
dx.doi.org/10.1021/cb4002602 | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology
Articles
fluorogenic probes in cells and tissues. J. Am. Chem. Soc. 134, 12157−
12167.
(46) Smith, A. M., Chan, J., Oksenberg, D., Urfer, R., Wexler, D. S.,
Ow, A., Gao, L., McAlorum, A., and Huang, S. -G. (2004) A high-
throughput turbidometric assay for screening inhibitors of protein
disulfide isomerase activity. J. Biomol. Screen. 9, 614−620.
I
dx.doi.org/10.1021/cb4002602 | ACS Chem. Biol. XXXX, XXX, XXX−XXX