Identification of acyl protein thioesterases 1 and 2 as the cellular targets of the ras-signaling modulators palmostatin B and M

Article abstract of DOI:10.1002/anie.201102967

Finding the target: Activity-based proteomic profiling probes based on the depalmitoylation inhibitors palmostatin B and M (see picture) have been synthesized and were found to target acyl protein thioesterase 1 (APT1) and 2 (APT2) in cells. Copyright

Full text of DOI:10.1002/anie.201102967

Communications
DOI: 10.1002/anie.201102967
Enzyme Inhibition
Identification of Acyl Protein Thioesterases 1 and 2 as the Cellular
Targets of the Ras-Signaling Modulators Palmostatin B and M
Marion Rusch, Tobias J. Zimmermann, Marco Bꢀrger, Frank J. Dekker, Kristina Gçrmer,
Gemma Triola, Andreas Brockmeyer, Petra Janning, Thomas Bçttcher, Stephan A. Sieber,
Ingrid R. Vetter, Christian Hedberg,* and Herbert Waldmann*
The S-palmitoylated and S-farnesylated signal-transducing
H- and N-Ras GTPases play important roles in cell growth,
division, and differentiation, and are frequently mutated in
melanoma, leukemia, and cancers of the bladder, liver, and
kidney.[1–3]
enzyme targets on the basis of their mechanism of enzymatic
catalysis. Typically, the reaction of a nucleophile in the
enzyme active site with an electrophile embedded in a
selective probe is employed. Previously, b-lactone-based
probes have successfully been used to profile bacterial,[10,11]
plant,[12] as well as mammalian proteomes.[13] A frequently
applied strategy to facilitate target isolation by means of
reactive proteomics is a bioorthogonal ligation of the
covalently modified enzyme with an additional affinity tag,
such as a suitably functionalized biotin. In particular, the
copper(I)-catalyzed Huisgen [3+2] cycloaddition has served
as an efficient tool for this purpose.[14,15]
Here we describe the development of ABPP probes based
on the Ras depalmitoylation inhibitors palmostatin B[5] and
M,[6] and the identification of APT1 and APT2 as their
targets, which has relevance to the dynamic Ras palmitoyla-
tion cycle.
For target identification, we designed two ABPP probes
based on our most potent compound palmostatin M (1) and
the previously reported APT1 inhibitor palmostatin B (2). In
the case of the probes based on 1, an alkyne was introduced at
the two sides of the molecule, thereby affording 3 and 4, to
facilitate isolation through a [3+2] cycloaddition (Sche-
me 1A). For the synthesis of an ABPP probe derived from
palmostatin B, the alkyne functionality was embedded in the
lipid chain (5; Scheme 1B; see the Supporting Information).
This site previously served well for introduction of a
fluorophore without compromising the inhibitory activity of
APT1.[5] A conceptually similar strategy as for the synthesis of
the focused b-lactone library in the development of palmos-
tatin M[6] was employed and only the more potent S,S isomers
were considered. In brief, an anti-aldol reaction between O-
acylephedrine derivative 6 and aldehyde 7 resulted in the b-
hydroxyester 8. S-Oxidation of 8 with oxone, followed by
saponification (LiOH) and subsequent b-lactonization with
PhSO2Cl in pyridine yielded ABPP probe 3 (Scheme 1C).
The regioisomeric alkyne derivative 4 was synthesized in an
analogous way by means of an anti-aldol reaction between O-
dodecylated ephedrine derivative 9 and alkyne-aldehyde 10,
which led to the corresponding b-hydroxyester 11, which was
further transformed to 4 by b-lactonization with PhSO2Cl in
pyridine (Scheme 1D).
Dynamic S-palmitoylation and S-depalmitoylation of
H- and N-Ras by palmitoyl transferases and thioesterases
determines proper Ras localization and signaling in cells.[4]
Interference with the Ras-de/reacylation cycle by inhibition
of the depalmitoylation through the use of the b-lactone acyl
protein thioesterase 1 (APT1) inhibitors palmostatin B[5] or
palmostatin M[6] (Figure 1) disturbs the precise H- and N-Ras
steady-state localization, and results in down-regulation of
global Ras signaling. Palmostatin B does not inhibit several
phosphodiesterases with relevance to Ras signaling.[5] How-
ever, it remains unclear whether APT1 is the only Ras-
depalmitoylating enzyme in cells and whether palmostatin B
and M target additional proteins relevant to the Ras cycle, in
particular the related isoenzyme APT2.[7]
Since the palmostatins are acylating reagents and, as in the
case of APT1, inhibit their target proteins through reversible
acylation of active-site nucleophiles,[5] their chemical nature
offers an opportunity for the identification of the target by
means of activity-based proteome profiling (ABPP) through
slow reactive proteomics.[8,9] In ABPP, chemical probes with
balanced reactivity are employed to tag, isolate, and identify
[*] Dipl.-Chem. M. Rusch, Dipl.-Chem. T. J. Zimmermann,
M. Sc. M. Bꢀrger, Dr. K. Gçrmer, Dr. G. Triola, A. Brockmeyer,
Dr. P. Janning, Dr. I. R. Vetter, Dr. C. Hedberg, Prof. Dr. H. Waldmann
Max-Planck-Institut fꢀr Molekulare Physiologie
Abt. Chemische Biologie
Otto-Hahn-Strasse 11, 44227 Dortmund (Germany)
E-mail: christian.hedberg@mpi-dortmund.mpg.de
herbert.waldmann@mpi-dortmund.mpg.de
Dipl.-Chem. M. Rusch, Dipl.-Chem. T. J. Zimmermann,
Dr. K. Gçrmer, Dr. G. Triola, A. Brockmeyer, Dr. P. Janning,
Dr. C. Hedberg, Prof. Dr. H. Waldmann
Technische Universitꢁt Dortmund, Fakultꢁt Chemie
Lehrbereich Chemische Biologie
Otto-Hahn-Strasse 6, 44227 Dortmund (Germany)
Dr. F. J. Dekker
Univ. of Groningen, University Centre for Pharmacy
A. Deusinglaan 1, 9713 AV Groningen (The Netherlands)
Investigation of APT1 inhibition by the ABPP probes
with the established assay[6] revealed inhibition constants of
IC50 = 4.4 nm for 3, IC50 = 10.0 nm for 4, and IC50 = 6.0 nm for
5, which are in a similar range to the values reported for the
corresponding non-alkyne-functionalized APT1 inhibitors. To
Dr. T. Bçttcher, Prof. Dr. S. A. Sieber
Technische Universitꢁt Mꢀnchen, Organic Chemistry II
Lichtenberg Strasse 4, 85747 Garching (Germany)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201102967.
9838
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 9838 –9842

subjected to separation by sodium dodecylsulfate polyacryl-
amide gel electrophoresis (SDS-PAGE). In the case of
incubation with ABPP probe 3, in-gel fluorescence detection
revealed three major fluorescent bands which are not present
in the negative control (Figure 1A). In the case of incubation
with ABPP probe 4, the labeling and identification of the
bound proteins was found to be considerably less efficient
(Figure 1A). Excision of the bands, followed by in-gel tryptic
digestion and subsequent identification of the proteins from
the resulting peptide fragments by means of nano-HPLC MS/
MS revealed the labeled proteins to be APT1, APT2, and with
some probability, palmitoyl protein thioesterase 1[7,16] (PPT1).
In addition, several other proteins were identified (p < 0.01,
see the Supporting Information). According to current
knowledge, none of these proteins display any enzymatic
activity linked to Ras signaling. To gain insight into the
efficiency of labeling by probes 3 and 4, and thus their binding
modes, we carried out a Western blot with a known amount of
recombinant hAPT1 on a separate gel line (Figure 1B). This
blot allowed us to assess the total amount of APT1 pulled
down with the two ABPP probes. Although, probes 3 and 4
display similar biochemical potency (IC50 = 4.4 nm and 10 nm,
respectively), the pull-down efficiency significantly differs in
terms of the amount of labeled APT1. This finding indicates
that the alkyne in pull-down probe 3 is accessible for the
dipolar cycloaddition and that the aminomethylsulfonyl
group consequently penetrates the binding site, whereas the
orientation might be reversed in the case of 4, so that the
alkyne is buried in the active site and no longer accessible
(Figure 1C). This difference in labeling is not only indicative
of the binding mode of the probes, but also in accordance with
their substrate-based design.[6]
Although palmostatin B and M display biochemical
IC50 values in a similar range (5.4 and 2.5 nm, respectively),
palmostatin M is several times more active in cells. To
elucidate the cellular enzyme inhibition profile of palmosta-
tin B versus palmostatin M, we subjected the alkyne-tagged
analogue of palmostatin B (5) to analogous concentration-
dependant affinity isolation experiments as compound 3 and
analyzed the total amount of APT1 pulled down by Western
blotting (Figure 2). Although the profiles of the fluorescent
gels (Figure 2A) were similar, quantification by Western
blotting for APT1 showed detectable activity down to 50 nm
for ABBP probe 3 derived from palmostatin M, whereas
ABPP probe 5 derived from palmostatin B required a
concentration of at least 100 nm to label detectable levels of
APT1 under identical conditions (Figure 2B).
Thus, the activity-based proteomic profiling identified
only APT1 and APT2 as target proteins with potential
relevance to Ras deacylation and signaling. APT2 has been
suggested as a relevant thioesterase for Ras depalmitoylation
before,[5,17] however, experimental proof has been lacking. To
demonstrate that palmostatin M and B indeed inhibit APT2,
and to explore differences in substrate specificity between the
APTs, we expressed recombinant human APT2 protein in an
analogous way as APT1.[6] To this end, the hAPT2 gene was
Gateway-cloned into a pGEX expression vector and modified
to contain a PreScission protease recognition sequence
between the APT2 and the GST gene (for details and
Figure 1. Cellular proteome labeling profile for probes 3 and 4 derived
from palmostatin M after 20 min incubation at various probe concen-
trations (50, 10, 1 mm). A) Fluorescent gel highlighting three major
bands corresponding to PPT1 (34.193 kDa), APT1 (MW=24.67 kDa),
and APT2 (MW=24.73 kDa) Conditions: Biotin-rhodamine-
N3 =20 mm, TCEP·HCl=0.5 mm, ligand=50 mm, CuSO4 =0.5 mm,
HeLa cells=2ꢂ106. B) Corresponding Western blot for APT1 visual-
ization using polyclonal Rabbit APT1 antibody which shows a stronger
affinity of probe 3 compared to 4 for APT1. Recombinant hAPT1
(25 ng) was used as the reference. C) Binding mode accounting for the
observed APT1-labeling efficiency based on the accessibility of the
alkyne group for the click reaction. TCEP=tris(2-carboxyethyl)phos-
phine.
characterize the protein targets of palmostatin M, HeLa cells
were incubated with ABPP probes 3 and 4 at concentrations
of 50 mm, 10 mm, and 1 mm. The cells were lysed and the
labeled proteome was tagged by a copper-catalyzed [3+2]
Huisgen reaction with an azide-biotin-rhodamine construct as
previously described.[11] After enrichment on magnetic strept-
avidin beads, the bound proteins were released by heating and
Angew. Chem. Int. Ed. 2011, 50, 9838 –9842
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 9839

Communications
DiFMUO-APT1 assay[6] was adjusted to
monitor APT2 activity by correcting for the
higher Km value of the substrate (see the
Supporting Information) and used to
screen our focused b-lactone library. The
IC50 values determined revealed a selectiv-
ity profile for APT2 very similar to that of
APT1, with IC50 values being higher in
general since the assay was performed at a
higher enzyme concentration (50 nm for
APT2 versus 5 nm for APT1). Sublibrary A
is equally potent for inhibition of APT2
and APT1. Enzyme kinetic parameters for
APT2 proved to be very similar to those of
APT1, with a pre-steady-state inactivation
rate higher than 4000 molÀ1 sÀ1. The inhib-
ition mode was found to be of mixed mode
(competitive-like) because of the slow
rehydrolysis of the enzyme–inhibitor com-
plex. None of the libraries A–C investi-
gated could discriminate between APT1
and APT2, thus providing no opportunity
to dissect APT1 from APT2 in the context
of intracellular depalmitoylation events
(Table 1).
APT1 has a broad substrate tolerance
and in vitro deacylates several palmitoy-
lated proteins, such as Gia1, eNOS, and H-
Ras.[18–20] However, previous results
showed that the catalytic efficiency of
APT1 for Gia1 is threefold higher than for
H-Ras, thus indicating some substrate pref-
[19]
erence for palmitoylated Gia1
.
In agree-
ment with this finding, studies carried out
on yeast strains in which the APT1 gene
was disrupted showed that the Gia1-depal-
mitoylating activity was completely abol-
ished in these strains.[19] However, depal-
mitoylation of H-Ras was similar to that of
wild-type strains, hence suggesting the
existence of another depalmitoylating
enzyme in yeast.[19] Although a direct
equivalent for APT2 could not be found
in yeast, it has been hypothesized that
APT2 is a depalmitoylating enzyme with a
different substrate preference than APT1.
To characterize APT2 and to gain novel
insights into its possible involvement in Ras
signaling we monitored the APT1- and
APT2-mediated depalmitoylation of a bio-
Scheme 1. A) Design of ABPP probes 3 and 4 derived from palmostatin M. B) Design of the
ABPP probe 5 derived from palmostatin B. C) Synthesis of the b-lactone ABPP probe 3: a) 6
(1.2 equiv), NEt3 (2.5 equiv), (CyHex)2BOTf (2.2 equiv), À788C (2 h), CH2Cl2, 66%; b) oxone
(3 equiv), RT (7 h), quantitative; c) LiOH (4 equiv), dioxane/H2O (8:1), 408C, 2 days, 85%;
d) PhSO2Cl (3 equiv), dry pyridine, 08C (24 h), 31%. D) Synthesis of the b-lactone ABPP
probe 4: e) 9 (1.2 equiv), NEt3 (2.5 equiv), (CyHex)2BOTf (2.2 equiv), À788C (2 h), CH2Cl2,
40%; f) oxone (3 equiv), RT (7 h), 97%; g) LiOH (4 equiv), dioxane/H2O (8:1), 408C, 2 days,
30%; h) PhSO2Cl (3 equiv), dry pyridine, 24 h, 08C, 30%. CyHex=cyclohexyl, OTf=trifluoro-
methanesulfonyl, Bn=benzyl, Mes=mesityl=2,4,6-trimethylphenyl.
references see the Supporting Information). Escherichia coli
BL21-CodonPlus(DE3)-RIL strain cells were transformed by
the plasmid and expressed as GST fusion protein. After
bacterial lysis, APT2 was separated from the GST by over-
night incubation of the GSH column with PreScission
protease. The resulting recombinant protein was purified to
homogeneity by size-exclusion chromatography (see the
Supporting Information). In general, 1.3 mg of protein was
obtained from one liter of the bacterial culture. The
logically active semisynthetic N-Ras protein in vitro.[21] The
palmitic acid liberated in the thioester hydrolysis was
detected by means of the ADIFAB (acrylodated intestinal
fatty acid binding protein) assay. ADIFAB is an established
fluorescent indicator for the detection and measurement of
free fatty acids.[22] Indeed, both APT1 and APT2 successfully
depalmitoylate N-Ras in vitro at similar concentrations.
Kinetic studies revealed comparable KM values for both
APT1 and APT2. Interestingly, the turnover number of
9840 www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 9838 –9842

Figure 2. Concentration-dependent APT1 pull-down profile for probe
5 derived from palmostatin B and probe 3 derived from palmosta-
tin M after 20 min incubation at probe concentrations ranging from
10 mm to 60 nm. A) Fluorescent gel highlighting the bands corre-
sponding to APT1 (MW=24.67 kDa) and APT2 (MW=24.73 kDa).
Conditions: Biotin-rhodamine-N3 =10 mm, TCEP·HCl=0.5 mm,
ligand=50 mm, CuSO4 =0.5 mm, number of HeLa cells=7ꢂ105.
B) Corresponding Western blot for APT1 visualization (polyclonal
Rabbit APT1 antibody) showing a stronger affinity of probe 3 than 5
for APT1.
Table 1: IC50 values for the inhibition of APT2 by focused b-lactone
libraries A–C.
Compound (S,S) IC50 APT2 [nm]
Compound (R,R) IC50 APT2 [nm]
Figure 3. Deacylation of semisynthetic S-palmitoylated S-farnesyl-N-Ras by
APT1 and APT2. A) APT1 and APT2 depalmitoylate semisynthetic N-Ras
in vitro. B) Palmitoylated N-Ras was used at different concentrations as the
substrate for APT1 (squares) and APT2 (circles). Reactions were initiated by
addition of APT1 or APT2 (50 nm). Release of palmitic acid was followed by
measuring the fluorescence spectra (400–650 nm) in the presence of
200 nm ADIFAB at room temperature. Fitted curves show the resulting
palmitic acid detected versus time in each case.
12: n=2
13: n=3
14: n=4
15: n=5
40.11Æ4.87
21.03Æ0.65
36.74Æ2.35
20.80Æ1.24
16: n=2
17: n=3
18: n=4
19: n=5
231.49Æ16.19
321.88Æ28.69
206.37Æ22.01
79.98Æ5.30
20: n=3
21: n=4
22: n=5
27.58Æ1.64
31.57Æ2.02
34.10Æ3.50
23: n=3
24: n=4
25: n=5
206.51Æ13.15
210.39Æ15.21
283.30Æ27.22
Table 2: Kinetic parameters of the depalmitoylation of semisynthetic
S-palmitoylated S-farnesyl-N-Ras with APT1 and APT2.
K
M [mm]
Kcat [sÀ1
]
Kcat/KM [sÀ1 mmÀ1
]
APT1
APT2
1.16Æ0.015
1.43Æ0.278
5.62ꢂ10À3
9.58ꢂ10À3
4.855ꢂ10À3
6.693ꢂ10À3
26: n=2
27: n=4
28: n=5
39.62Æ3.17
60.56Æ15.25
66.15Æ3.89
29: n=2
30: n=4
31: n=5
268.20Æ13.44
394.80Æ60.61
206.79Æ17.03
design principles and shown that they target APT1 and APT2
in cells. These inhibitors attenuate Ras signaling by interfer-
ing with the dynamic Ras de/reacylation cycle and induce
phenotypic reversal of H-Ras-transformed cells.[6] Therefore,
we suggest and provide a first experimental proof that both
APT1 and APT2 are thioesterases relevant to Ras depalmi-
toylation and signaling. Notably, we could not identify other
intracellular esterases (such as phospholipases A1, A2, C, and
D) relevant to Ras signaling as targets for the palmostatins,
which proves the selectivity of the probes. Taken together,
these findings suggest that no further hydrolases employing a
similar mechanism of catalysis, and possibly no further
ABPP probes
palm. B
19.58Æ0.87
3:
4:
5:
32.44Æ4.82
64.44Æ7.13
23.72Æ2.00
APT2 was double that of APT1 under identical conditions
(Figure 3 and Table 2).
In conclusion, we have developed potent inhibitors of acyl
protein thioesterases on the basis of substrate similarity
Angew. Chem. Int. Ed. 2011, 50, 9838 –9842
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
9841

Communications
[9] M. J. Evans, B. F. Cravatt, Chem. Rev. 2006, 106, 3279 – 3301.
[10] T. Bçttcher, S. A. Sieber, J. Am. Chem. Soc. 2008, 130, 14400 –
14401.
[11] T. Bçttcher, S. A. Sieber, Angew. Chem. 2008, 120, 4677 – 4680;
Angew. Chem. Int. Ed. 2008, 47, 4600 – 4603.
[12] Z. Wang, C. Gu, T. Colby, T. Shindo, R. Balamurugan, H.
Waldmann, M. Kaiser, R. A. L. van der Hoorn, Nat. Chem. Biol.
2008, 4, 557 – 562.
[13] P.-Y. Yang, K. Liu, M. H. Ngai, M. J. Lear, M. R. Wenk, S. Q.
Yao, J. Am. Chem. Soc. 2010, 132, 656 – 666.
hydrolases in general, are involved in Ras depalmitoyation in
cells. For these reasons, inhibition of APT1 and APT2 may be
a viable approach to interfere with H- and N-Ras signaling in
malignant settings such as leukemia, melanoma, and bladder,
liver, and kidney cancer.
Received: April 29, 2011
Revised: July 27, 2011
Published online: September 9, 2011
[14] A. E. Speers, B. F. Cravatt, Chem. Biol. 2004, 11, 535 – 546.
[15] P. F. van Swieten, M. A. Leeuwenburgh, B. M. Kessler, H. S.
Overkleeft, Org. Biomol. Chem. 2005, 3, 20 – 27.
[16] PPT1 is a lysosomal enzyme involved in the degradation of
lipidated proteins, for details see J.-Y. Lu, S. L. Hofmann, J.Lipid
Res. 2006, 47, 1352 – 1357. PPT1 is not considered relevant to Ras
signaling due to its compartmental localization, although it has
been shown that PPT1 depalmitoylates the Ras isoforms in
biochemical assays, see L. Meng, N. Sin, C. M. Crews, Biochem-
istry 1998, 37, 10488 – 10492 and S. Cho, P. E. Dawson, G.
Dawson, J. Neurosci. Res. 2000, 59, 32 – 38.
[17] Recently, APT2 was reported to specifically depalmitoylate
GAP43 in cells, see V. M. Tomatis, A. Trenchi, G. A. Gomez,
J. L. Daniotti, PloS One 2010, 5, e15045.
[18] J. A. Duncan, A. G. Gilman, J. Biol. Chem. 1998, 273, 15830 –
15837.
[19] J. A. Duncan, A. G. Gilman, J. Biol. Chem. 2002, 277, 31740 –
31752.
[20] D. C. Yeh, J. A. Duncan, S. Yamashita, M. Michel, J. Biol. Chem.
1999, 274, 33148 – 33154.
[21] B. Bader, K. Kuhn, D. J. Owen, H. Waldmann, A. Wittinghofer,
J. Kuhlmann, Nature 2000, 403, 223 – 226.
[22] G. V. Richieri, R. T. Ogata, A. M. Kleinfeld, J. Biol. Chem. 1992,
267, 23495 – 23501.
Keywords: acyl protein thioesterase · inhibition · proteomics ·
Ras proteins · signal transduction
.
[1] J. F. Hancock, Nat. Rev. Mol. Cell Biol. 2003, 4, 373 – 385.
[2] A. Wittinghofer, H. Waldmann, Angew. Chem. 2000, 112, 4360 –
4383.
[3] P. G. Dangle, B. Zaharieva, H. Jia, K. S. Pokar, Recent Pat. Anti-
Cancer Drug Discovery 2009, 4, 125 – 136.
[4] O. Rocks, A. Peyker, M. Kahms, P. J. Verveer, C. Koerner, M.
Lumbierres, J. Kuhlmann, H. Waldmann, A. Wittinghofer,
P. I. H. Bastiaens, Science 2005, 307, 1746 – 1752.
[5] F. J. Dekker, O. Rocks, N. Vartak, S. Menninger, C. Hedberg, R.
Balamurugan, S. Wetzel, S. Renner, M. Gerauer, B. Schoeler-
mann, M. Rusch, J. W. Kramer, D. Rauh, G. W. Coates, L.
Brunsveld, P. I. H. Bastiaens, H. Waldmann, Nat. Chem. Biol.
2010, 6, 449 – 456.
[6] C. Hedberg, F. Dekker, M. Rusch, S. Renner, S. Wetzel, N.
Vartak, C. Geerding-Reimers, R. S. Bon, P. I. H. Bastiaens, H.
Waldmann, Angew. Chem. 2011, DOI: 10.1002/ange.201102965;
Angew. Chem. Int. Ed. 2011, DOI: 10.1002/anie.201102965.
[7] R. Zeidman, C. S. Jackson, A. I. Magee, Mol. Membr. Biol. 2009,
26, 32 – 41, and references therein.
[8] S. A. Sieber, B. F. Cravatt, Chem. Commun. 2006, 2311 – 2319.
9842 www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 9838 –9842