Integrated phenotypic screening and activity-based protein profiling to reveal potential therapy targets of pancreatic cancer

Article abstract of DOI:10.1039/c8cc08753a

Pancreatic cancer has been defined as one of the most complex and challenging cancers to treat, but very few valid therapeutic targets have been identified to date. To address this issue, a 61-compound library was readily created by Ugi reaction followed by phenotypic screening, leading to the discovery of two most potent inhibitors, P21 and P26, which significantly impair BxPC-3 pancreatic cancer cell survival. A series of interacting protein hits, such as GSTO1, FAM213A, RAB6A/6B/39A and USMG5, were subsequently identified by quantitative chemoproteomics studies. The main cellular target, GSTO1, was further validated as a novel pancreatic cancer therapeutic target.

Full text of DOI:10.1039/c8cc08753a

ChemComm
View Article Online
View Journal
COMMUNICATION
Integrated phenotypic screening and activity-
based protein profiling to reveal potential therapy
targets of pancreatic cancer†
Cite this:DOI: 10.1039/c8cc08753a
Received 2nd November 2018,
Accepted 10th January 2019
a
a
a
b
a
Wenyan Liu, Zhang Zhang, Zhi-Min Zhang, Piliang Hao, Ke Ding and
a
Zhengqiu Li
*
DOI: 10.1039/c8cc08753a
rsc.li/chemcomm
6
Pancreatic cancer has been defined as one of the most complex and this disease are known to date. Accordingly, in this study, we
challenging cancers to treat, but very few valid therapeutic targets endeavoured to integrate phenotypic screening with activity-based
have been identified to date. To address this issue, a 61-compound protein profiling (ABPP) to reveal potential targets that could be
library was readily created by Ugi reaction followed by phenotypic used in chemotherapy of pancreatic cancer.
screening, leading to the discovery of two most potent inhibitors, P21
To achieve this goal, a 61-member library was efficiently created
and P26, which significantly impair BxPC-3 pancreatic cancer cell by an Ugi reaction (Fig. 1A and Table S1 (ESI†)). Different substi-
1
2
3
4
survival. A series of interacting protein hits, such as GSTO1, FAM213A, tuents (R /R /R /R ) were embedded in the scaffold to promote
RAB6A/6B/39A and USMG5, were subsequently identified by quanti- interactions with different protein targets in cells. The cysteine-
tative chemoproteomics studies. The main cellular target, GSTO1, was reactive warheads, acrylamide or chloroacetamide, in these mole-
further validated as a novel pancreatic cancer therapeutic target.
cules offer a way to trap interacting proteins in situ through covalent
reactivity (Schemes S1–39, ESI†). Amongst these 61 molecules,
Therapeutic target discovery, an essential first step in drug 37 bioorthogonal compounds bear an alkyne handle that makes
discovery and development, plays an important role in the treat- them suitable for subsequent target enrichment.
1
ment of human diseases. Although as many as 5000 potential
With this compound library in hand, we screened them
2
new targets are currently available for drug intervention, the against human pancreatic cancer cell line, BxPC-3, by using a
main challenge facing the pharmaceutical industry is still to CCK-8 assay. As shown in Fig. 1 and Table S1 (ESI†), P21 and
identify and validate novel disease-related protein targets. P26 exhibit the most potent antiproliferative activity with IC50
Recently, phenotypic screening coupled with chemoproteomics values of 2.58 mM and 1.36 mM, respectively, suggesting that the
has been developed as an efficient way to reveal druggable proteins that interact with these molecules might be therapeutic
hotspots for cancer therapy. For example, Nomura et al. identified targets of pancreatic cancer. This drives us to further identify the
ALDH3A1 as a target for lung cancer therapy by coupling pheno- interacting protein targets. Both P21 and P26 have an alkyne handle,
3
typic screening with isoTOP-ABPP. Cravatt et al. combined which makes them suitable for reporter tag conjugation to visualize
fragment-based chemoproteomics with phenotypic screening to and identify labeled proteins without further modifications. Two
identify small molecules that promote adipocyte differentiation by nonclickable analogues, P27 and P28, were also created for use
4
engaging the protein PGRMC2. We have recently disclosed in competitive proteome profiling by replacing the alkyne group
several potential druggable targets for breast cancer therapy by with alkyl groups as competitive inhibitors. Antiproliferation assay
tandem phenotype screening and affinity-based protein profiling demonstrated that these competitors possess a similar inhibitory
(
AfBP) studies; at the same time, a specific probe targeting a potency to P21 and P26, respectively (Fig. 1A and B). Notably, P28
5
cancer biomarker was discovered. Pancreatic cancer is the fourth displayed a submicromolar IC50 value of 0.6 mM. The cell cycle
leading cause of cancer death and very few therapeutic targets for analysis revealed that P28 markedly increased the proportion of
BxPC-3 cells in G2/M phases (Fig. 1C and Fig. S2, ESI†), and
a
significantly induced cancer cell apoptosis (Fig. 1D). These data
confirm the potent anticancer effects of P21 and P26 and
corresponding competitors, P27 and P28.
School of Pharmacy, Jinan University; Guangzhou City Key Laboratory of Precision
Chemical Drug Development; International Cooperative Laboratory of Traditional
Chinese Medicine Modernization and Innovative Drug Development, Ministry of
Education (MOE) of People’s Republic of China, 601 Huangpu Avenue West,
Guangzhou, 510632, China. E-mail: pharmlzq@jnu.edu.cn
Next, chemoproteomics coupled with bioimaging was carried
out to identify the target proteins of P21 and P26. The labeling
b
School of Life Science and Technology, ShanghaiTech University,
7
profile was first carried out using previously reported procedures.
3
93 Middle Huaxia Road, Shanghai 201210, China
†
Electronic supplementary information (ESI) available. See DOI: 10.1039/c8cc08753a After incubation of the probes with live BxPC-3 cells for 2–4 h,
This journal is ©The Royal Society of Chemistry 2019
Chem. Commun.

View Article Online
Communication
ChemComm
Fig. 1 (A) Chemical structures of representative inhibitors and corresponding IC50 values against BxPC-3; a CCK-8 assay was used to evaluate the
inhibition. (B) IC50 values of P21/P26 and corresponding competitors, P27/P28. Cell cycle progression (C) and cell apoptosis (D) in BxPC-3 cells, treated
with 2 mM of P28 for 24–48 h, analysed by flow cytometry.
the cells were lysed; the resulting cell lysates were conjugated labeling profiles, the fluorescence signals of both probes
with TAMRA-N and separated by SDS-PAGE followed by in-gel decreased sharply in the presence of excess competitors, P27
3
fluorescence scanning. As shown in Fig. 2A and B, P21 and P26 can and P28. Control imaging experiments with DMSO under similar
specifically label an B25 kDa and B30 kDa protein, respectively, at conditions gave minimal background fluorescence when com-
1
mM or even lower probe concentrations (red box, Fig. 2A and B). pared to the labeled cells (Fig. S4, ESI†). These competitive in situ
Higher probe concentrations led to more labeling bands (Lane 1 of labeling and live-cell imaging experiments proved that both P21
left gel in Fig. 2A and B), which were presumed to arise from non- and P26 were able to efficiently capture their intended cellular
specific labeling. P26-treated samples produce the major labeling targets, and have highly selective cellular targets.
band at probe concentrations as low as 0.2 mM (Fig. 2B, left gel),
Finally, we proceeded to identify the interacting protein targets
and time-dependent labeling experiments showed that the major of P21 and P26 by chemoproteomics experiments. Similar to the
labeling band can be visible in as little as 5 min, indicating high procedures described above, SILAC (stable isotope labeling by
labeling efficiency. In the presence of a ten-fold excess of the amino acids) labeled BxPC-3 cells were treated with the probes
competitors, P27 and P28, all fluorescent bands disappeared for 2–4 h (1 mM). Upon cell lysis, the resulting probe-labeled
3
(Fig. 2C), demonstrating that they were probe-targeted proteins proteomes conjugated to a biotin-N tag. After enrichment by
rather than results of non-specific labeling. streptavidin beads, bound proteins were digested with trypsin.
To track the subcellular locations of P21 and P26 in live Control experiments with DMSO- and competitor-treated samples
cells, cellular imaging was next carried out. The cells were were carried out concurrently, which served to distinguish between
first treated with the probes (1 mM) for 2–4 h. Subsequently, real targets and background labeling. The identified proteins were
the cells were fixed, permeabilized and reacted with TAMRA-N
under previously optimized click chemistry conditions. Strong ment ratio (probe/DMSO) and competition ratio (probe/probe with
3
analyzed by corresponding volcano plots as (log
2
À) of both enrich-
7
fluorescence signals were observed in the cell cytosol from excess competitors) against statistical significance (Àlog10 p-value).
2
probe-treated samples (Fig. 2D). Consistent with the results of Proteins with enrichment (log À) and Àlog10 p-value ratios higher
Chem. Commun.
This journal is ©The Royal Society of Chemistry 2019

View Article Online
ChemComm
Communication
Fig. 2 Time- and concentration-dependent labelling profiles of P21 (A) and P26 (B) with BxPC-3 cells. (C) Competitive labelling profiles of the probes
with BxPC-3 cells. (D) Cellular imaging of the probes in the presence or absence of the parent inhibitors (1 mM probe concentration). Scale bar = 10 mm.
(E) Quantitative mass spectrometry-based profiling of P21 and P26 binding proteins (1 mM).
than 1.5 were considered to be significant hits. As a result, 2 early diagnosis of esophageal cancer by targeting the GSTO1
9
protein hits, FAM213A and RAB6A/6B/39A, were identified by biomarker. The labeling profile of P26 with GSTO1 can be
P21; 3 protein hits, GSTO1, USMG5, and RAB6A/6B/39A, were blocked by treatment with excess P28 (Fig. 3B, lower gel), which
identified by P26 (Fig. 2E). These protein hits also showed high again proved that P26 and P28 interact with the same set
(
log
2
À) and Àlog10 p-value ratios in competition samples (Tables S3 of protein targets. Thermal shift binding assays proved that
and S4, ESI†). Further analysis revealed that FAM213A (Family both P26 and P28 induced a negative shift in the melting
with sequence similarity 213 member A) and GSTO1 (glutathione temperature, indicating protein destabilization on binding,
1
2
S-transferase omega 1) can match the major labeling bands of P21 which is consistent with the covalent binding mechanism.
and P26 at B26 and B30 kDa, respectively, which was further Docking experiments showed that P26 covalently bound to the
validated by pull-down/western blotting (WB) experiments with the C32 in GSTO1 (Fig. 3D), which coincides well with a previously
1
2a
corresponding antibodies (Fig. 3A). It was found that both FAM213A reported binding model of GSTO1 inhibitors. Knockdown of
and GSTO1 proteins locate in the cell cytosol, which is in-line with GSTO1 markedly decreased the apoptosis-inducing effects of P28
the imaging results (Fig. 2D). Both GSTO1 and FAM213A have against pancreatic cancer cells (Fig. 3E and Fig. S5, ESI†),
8
already been validated as potential new targets in oncology, and implying that targeting GSTO1 is necessary to retain the inhibi-
GSTO1 has been identified as a biomarker of esophageal squamous tory activity of P28. As shown in Fig. 3F and G, P28 is capable of
9
13
cell carcinoma (ESCC). Up-regulated during skeletal muscle growth enhancing the cytotoxic effects of cisplatin and 7rh, two known
protein 5 (USMG5) plays a critical role in maintaining the ATP pancreatic cancer inhibitors. Clear synergistic effects were
synthase population in mitochondria and is implicated in the observed from the combination of P28 with cisplatin (Fig. 3G).
10
development of heart failure. Ras-related proteins Rab-6A/6B/39A These lines of evidence together proved that GSTO1 is a ther-
RAB6A/6B/39A) are members of the RAB family, which are involved apeutic target for pancreatic cancer.
in matrix and immune modulation. It is possible that P21/P26/
P27/P28 produce their antiproliferative effects through the combi- tors, P27/28, are potent inhibitors against pancreatic cancer by
nation of these protein interactions. phenotypic screening. A series of interacting protein targets,
Inspired by previous studies that GSTO1 is a therapeutic target in such as GSTO1, FAM213A, USMG5 and RAB6A/6B/39A were
(
11
In conclusion, we discovered that P21/26 and the competi-
12a
colorectal cancer and is implicated in drug resistance, we further identified by an activity-based protein profiling method.
validated the therapeutic utility of GSTO1 in pancreatic cancer. As Further validation revealed that both P26 and P28 directly
shown in Fig. 3B (upper gel), as little as 0.005 pmol GSTO1 can be target GSTO1, and the presence of GSTO1 is essential to retain
successfully detected by P26 at a 1 mM probe concentration, showing the anticancer effects of these molecules, which proves that
that this probe possesses excellent sensitivity toward GSTO1. This GSTO1 is a therapeutic target in pancreatic cancer. Moreover,
result underscores the excellent labelling efficiency of P26 with live these potent inhibitors serve as valuable tools to investigate the
cells (Fig. 2B) and implies that P26 could be a useful reagent in role of GSTO1 in cancer and other pathologies, and can be
This journal is ©The Royal Society of Chemistry 2019
Chem. Commun.

View Article Online
Communication
ChemComm
Fig. 3 (A) Target validation of FAM213A and GSTO1 by pull-down/Western blotting in the presence or absence of corresponding competitors.
B) Concentration-dependent labeling of recombinant GSTO1 with P26 in the presence or absence of P28 (final probe concentration is 1 mM), FL = in-gel
(
fluorescence scanning. CBB = Coomassie gel. (C) Thermal shift binding assay of P26 and P28 with BxPC-3 cells. (D) Docking experiments to predict the
binding mode of P26 with GSTO1. (E) Validation of GSTO1 knockdown by western blot analysis. Cells treated with scrambled siRNA are a negative control
(NC). Apoptosis of pancreas cells treated with 2 mM P28 for 48 h, was analysed by flow cytometry. (F) IC50 values of 7rh, a known pancreatic cancer
inhibitor in the presence or absence of different concentrations of P28. (G) P28 enhances cisplatin-induced cytotoxicity. BxPC-3 cells were treated with
the indicated concentrations of P28 and cisplatin for 72 h. (H) Synergistic effect analysis of combination of cisplatin with P28.
further optimized for drug development and early diagnosis of
esophageal cancer.
4 C. G. Parker, A. Galmozzi, Y. Wang, B. E. Correia, K. Sasaki, C. M.
Joslyn, A. S. Kim, C. L. Cavallaro, R. M. Lawrence, S. R. Johnson,
I. Narvaiza, E. Saez and B. F. Cravatt, Cell, 2017, 168, 527–541.
We thank National Natural Science Foundation of China
5
K. Cheng, J. S. Lee, P. Hao, S. Q. Yao, K. Ding and Z. Li, Angew.
Chem., Int. Ed., 2017, 56, 15044–15048.
(21602079, 21877050), Science and Technology Program of
6
A. M. Roberts, D. K. Miyamoto, T. R. Huffman, L. A. Bateman,
A. N. Ives, D. Akopian, M. J. Heslin, C. M. Contreras, M. Rape,
C. F. Skibola and D. K. Nomura, ACS Chem. Biol., 2017, 12, 899–904.
Guangdong Province (2017A050506028), Science and Technology
Program of Guangzhou (201704030060, 201805010007) for their
financial support. We also thank Prof. Shao Q. Yao (NUS) and
Prof. Jingyan Ge (Zhejiang University of Technology) for their
invaluable suggestions on this work.
7 (a) D. Zhu, H. Guo, Y. Chang, Y. Ni, L. Li, Z.-M. Zhang, P. Hao, Y. Xu,
K. Ding and Z. Li, Angew. Chem., Int. Ed., 2018, 57, 9284–9289;
(
5
b) H. Guo, J. Xu, P. Hao, K. Ding and Z. Li, Chem. Commun., 2017,
3, 9620–9623; (c) H. Guo and Z. Li, MedChemComm, 2017, 8, 1585–1591.
8
(a) Y. Xie, J. L. Dahlin, A. J. Oakley, M. J. Casarotto, P. G. Board and
J. B. Baell, J. Med. Chem., 2018, 61, 7448–7470; (b) Y. F. Chen,
C. C. Yang, S. Y. Kao, C. J. Liu, S. C. Lin and K. W. Chang, Cancer
Res., 2016, 76, 4872–4886.
Conflicts of interest
9
Y. Li, Q. Zhang, B. Peng, Q. Shao, W. Qian and J. Y. Zhang, Tumour
Biol., 2014, 35, 10871–10877.
There are no conflicts to declare.
10 Y. Nagata, M. Yamagishi, T. Konno, C. Nakanishi, Y. Asano, S. Ito,
Y. Nakajima, O. Seguchi, N. Fujino, M. A. Kawashiri, S. Takashima,
M. Kitakaze and K. Hayashi, Sci. Rep., 2017, 7, 17417.
Notes and references
11 F. Qadir, M. A. Aziz, C. P. Sari, H. Ma, H. Dai, X. Wang, D. Raithatha,
L. G. L. Da Silva, M. Hussain, S. P. Poorkasreiy, I. L. Hutchison,
A. Waseem and M. T. Teh, Mol. Cancer, 2018, 17, 97.
1
(a) T. Sharifnia, A. L. Hong, C. A. Painter and J. S. Boehm, Cell Chem.
Biol., 2017, 24, 1075–1091; (b) R. Santos, O. Ursu, A. Gaulton, 12 (a) K. Ramkumar, S. Samanta, A. Kyani, S. Yang, S. Tamura, E. Ziemke,
A. P. Bento, R. S. Donadi, C. G. Bologa, A. Karlsson, B. Al-Lazikani,
A. Hersey, T. I. Oprea and J. P. Overington, Nat. Rev. Drug Discovery,
J. A. Stuckey, S. Li, K. Chinnaswamy, H. Otake, B. Debnath, V. Yarovenko,
J. S. Sebolt-Leopold, M. Ljungman and N. Neamati, Nat. Commun., 2016,
7, 13084; (b) S. Pan, S. Y. Jang, S. S. Liew, J. Fu, D. Wang, J. S. Lee and
S. Q. Yao, Angew. Chem., Int. Ed., 2018, 57, 579–583.
2
017, 16, 19–34.
2
3
http://function.gnf.org/druggable.
J. L. Counihan, A. L. Wiggenhorn, K. E. Anderson and D. K. Nomura, 13 M. Gao, L. Duan, J. Luo, L. Zhang, X. Lu, Y. Zhang, Z. Zhang, Z. Tu,
ACS Chem. Biol., 2018, 13, 1970–1977.
Y. Xu, X. Ren and K. Ding, J. Med. Chem., 2013, 56, 3281–3295.
Chem. Commun.
This journal is ©The Royal Society of Chemistry 2019