Angewandte
Communications
Chemie
ure S6), whereas mass spectroscopy measurements provided
Subsequently, by using a relative quantitative proteomic
monoisotopic masses are consistent with cysteamine–1 and
cysteine–1 adducts (see Figure S7). These findings confirmed
that the a,b-unsaturated moiety of 1 is essential for reaction
with the sulfhydryl group of the thiol species, even though the
reaction occurs sluggishly. To demonstrate whether the
Michael acceptor of 1 was the prerequisite for potency, we
approach, we calculated the label-free quantification (LFQ)
ratio of streptavidin-enriched 5 to vehicle-labeled proteomes
(LFQ5:vehicle). From LFQ analysis (see Table S1 in the
Supporting Information; LFQ5/LFQvehicle > 20;), after ruling
out mismatched molecular weights, GAPDH, MDH2, and
ETFA were believed to be the potential targets. Validation
through RNA interference and cDNA overexpression of
GAPDH, MDH2, and ETFA revealed that only the knock-
down of GAPDH showed a survival benefit in response to
1 (ca. sixfold; see Figures S3a,d and S8; GAPDH-depleted
cells were continuously cultured in pyruvate-containing
media to avoid toxicity[15]).
=
reduced the C C bond in 1 to give 22a and 22b (Figure 1a)
for cytotoxicity assays. The results showed that both 22a and
22b were significantly less potent than 1 (Table 1). Taken
together, these results indicated that 1 contains an off-target
Michael acceptor as the pharmacophore.
To identify the target of 1, we modified 1 by incorporating
a biorthogonal azido group to facilitate target visualization or
enrichment to give probes 5 and 6 for TCI-oriented activity-
based proteome profiling (TCI-ABPP; see Figure S9a and
Scheme S1b,c). Ideally, these TCI-based probes would bind
to the target, remain for a residence time, and reorient
themselves to a suitable position for Michael addition with
neighboring nucleophilic side chains (usually cysteine). After
in situ labeling, the given proteome should be covalently
modified by the probe to allow subsequent Staudinger
ligation by a phosphine–biotin reporter tag (p-biotin; see
Figure S9a). By streptavidin blot analysis, one clear band
ranging from 35 to 40 kDa was visualized (Figure 3a), while
the labeling efficiency of each probe revealed a high corre-
lation with their corresponding IC50 value (5: 5.12 mm; 6:
12.20 mm). Since positive staining in a streptavidin blot reflects
the number of covalent adducts present, these results again
demonstrated that the formation of a Michael adduct plays
a critical role in the inhibitory activity of 1.
The exact modification site of 1 in GAPDH was identified
by incubating 1 with purified GAPDH at the physiological
pH value. After trypsin digestion, the peptides were analyzed
by MS/MS, which revealed that peptides with monoisotopic
masses that were consistent (< 5 ppm error) with the non-
catalytic C247 modification were observed (Figure 3c). A
parallel experiment performed using wild-type (WT) and
C247A GAPDH for 5-mediated phosphine–Cy5 ligation (p-
Cy5; see Figure S9b) provided the same conclusion that C247
was the only modification site (see Figure S9c). Interestingly,
human GAPDH contains three cysteine residues (Figure 3b),
yet the intramolecularly competing C152 residue, as the most
reactive thiolate in GAPDH, formed no adduct with 1, as
neither enzymatic activity nor a glycolytic network change
was observed (see Figure S10a,b), in close agreement with
thiolate-tolerating nature of 1 (Figure 2c). Hence, based on
the concept of TCI, the affinity for GAPDH should be the key
to lower the reaction barrier to give the final GAPDH–
1 adduct. By isothermal titration calorimetry (ITC), we
determined the affinity of 1 for GAPDHC247A (avoiding bond
formation heat). By ITC, this binding event was characterized
as an entropy-driven reaction with nanomolar affinity (KD =
0.26 mm; see Figure S9d). To date, all covalent inhibitors of
GAPDH, such as koningic acid, 3-bromopyruvate, and
orlistat (see Figure S10c–e),[16–18] rely on the alkylation of
C152 to modulate the function of GAPDH with their highly
reactive electrophiles. Collectively, these results suggested the
Michael adduct of GAPDH–1 was modulated by an affinity-
driven alkylation on C247 to overcome the energy barrier of
the thiolate-tolerating warhead in 1.
GAPDH was initially believed to be solely a glycolytic
enzyme in most cells. However, subsequent studies have
shown that GAPDH participates in many unexpected pro-
cesses, such as vesicle trafficking, cytoskeletal dynamics,
RPL13a protection, telomerase inhibition, and cell
death.[19,20] Based on GAPDH–1 adduct formation, we
covalently docked 1 with C247 of GAPDH. Consistent with
entropy-driven binding (ÀTDS dominance;[21] see Fig-
ure S9e), 1 was buried in a hydrophobic pocket located in
the middle of the a/b two-layer sandwich (Figure 4a), where it
revealed hydrophobic contacts to two functional motifs. For
clarity, we divided the pocket into 1,3-benzodioxole (BzD;
Figure 4b) and 3,4,5-trimethoxybenzyl subpockets (MOBz;
Figure 4c) related to SIAH1-dependent apoptosis (see Fig-
ure S11a) and the cytosol-restricted trans activation of AR
through interaction with a polyQ tract,[22,23] respectively.
Figure 3. a) Streptavidin blot of lysate treated with p-biotin-ligated 1, 5,
or 6, or a vehicle (see Figure S9a for a TCI-ABPP flowchart). a) pKa
values of each cysteine residue in human GAPDH. H179 deprotonates
C152 (blue dashed line; active site). c) Representative MS2 spectrum
of 1-derived peptide; Dm1 +398 Da (+C22H24O7).
Angew. Chem. Int. Ed. 2018, 57, 1 – 7
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3
These are not the final page numbers!