The identification of the ATR inhibitor VE-822 as a therapeutic strategy for enhancing cisplatin chemosensitivity in esophageal squamous cell carcinoma

Qi Shi, Luyan Shen, Bin Dong, Hao Fu, Xiaozheng Kang, Yongbo Yang, Liang Dai, Wanpu Yan, Hongchao Xiong, Zhen Liang, Keneng Chen

PII: S0304-3835(18)30408-7
DOI: 10.1016/j.canlet.2018.06.010
Reference: CAN 13943 To appear in: Cancer Letters
Received Date: 24 February 2018
Revised Date: 12 May 2018
Accepted Date: 6 June 2018

Please cite this article as: Q. Shi, L. Shen, B. Dong, H. Fu, X. Kang, Y. Yang, L. Dai, W. Yan, H. Xiong,
Z. Liang, K. Chen, The identification of the ATR inhibitor VE-822 as a therapeutic strategy for enhancing cisplatin chemosensitivity in esophageal squamous cell carcinoma, Cancer Letters (2018), doi: 10.1016/ j.canlet.2018.06.010.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Abstract
The activation of ATM (ataxia-telangiectasia mutated) and ATR (ataxia telangiectasia and Rad3-related), is essential for DNA damage repair and the maintenance of genomic stability. Therefore, ATM or ATR inhibition is considered as a promising strategy for sensitizing cancer cells to chemotherapy. This study is aimed to explore the effect of ATR inhibitor on sensitizing ESCC (esophageal squamous cell carcinoma) cells to cisplatin, and whether ATM deficiency could impact the sensitization. We found that 21.5% of ESCC cases had ATM deficiency and that patients with ATR activation after neoadjuvant chemotherapy had worse chemotherapy response and poorer overall survival than that without ATR activation (32 mons vs. >140mons).

Then, it was shown that VE-822 inhibited CHK1 activation, leading to the accumulation of cisplatin-modified DNA. And it inhibited cell proliferation, induced cell cycle arrest in G1 phase and enhanced cell apoptosis. Moreover, VE-822 significantly sensitized ESCC cells to cisplatin, and these two drugs had synergistic effects, especially in ATM-deficient cells, in vitro and in vivo. Our results suggest that ATR inhibition combining with cisplatin is a new strategy for managing patients with ESCC, especially patients with ATM-deficiency.

1 The identification of the ATR inhibitor VE-822 as a therapeutic strategy
2 for enhancing cisplatin chemosensitivity in esophageal squamous cell
3 carcinoma
4 Running title: ATR inhibitor sensitized ESCC cells to chemotherapy
5
6 Qi Shi1*, Luyan Shen1*, Bin Dong2, Hao Fu1, Xiaozheng Kang1, Yongbo Yang1,
7 Liang Dai1, Wanpu Yan1, Hongchao Xiong1, Zhen Liang1, Keneng Chen1.
8 Key Laboratory of Carcinogenesis and Translational Research (Ministry of
9 Education), 1Department of Thoracic Surgery I, 2Department of Pathology,
10 Peking University Cancer Hospital & Institute, Beijing, People’s Republic of
11 China
12 *Qi Shi and Luyan Shen contributed equally to this work
13 Corresponding author: Keneng Chen, Department of Thoracic Surgery I, Key
14 Laboratory of Carcinogenesis and Translational Research (Ministry of
15 Education), Peking University Cancer Hospital & Institute, No. 52, Fucheng Rd,
16 Haidian Dist, Beijing 100142, P.R. China. Phone: 86-10-88196536; Fax:
17 86-10-88196526; E-mail: [email protected].
18
19

20 Abstract
21 Inducing DNA damage is known to be one of the mechanisms of cytotoxic
22 chemotherapy agents for cancer such as cisplatin. The endogenous DNA
23 damage response confers chemoresistance to these agents by repairing DNA
24 damage. The initiation and transduction of the DNA damage response (DDR)
25 signaling pathway, which is dependent on the activation of ATM
26 (ataxia-telangiectasia mutated) and ATR (ataxia telangiectasia and
27 Rad3-related), is essential for DNA damage repair, the maintenance of
28 genomic stability and cell survival. Therefore, ATM or ATR inhibition is
29 considered as a promising strategy for sensitizing cancer cells to
30 chemotherapy. This study is aimed to explore the effect of ATR inhibitor on
31 sensitizing ESCC (esophageal squamous cell carcinoma) cells to cisplatin,
32 and whether ATM deficiency could impact the sensitization. We found that
33 21.5% of ESCC cases had ATM deficiency and that patients with ATR
34 activation after neoadjuvant chemotherapy had worse chemotherapy response
35 and poorer overall survival than that without ATR activation (32 mons
36 vs. >140mons). Then, it was shown that VE-822 inhibited ATR-CHK1 pathway
37 activation, leading to the accumulation of cisplatin-modified DNA. And it
38 inhibited cell proliferation, induced cell cycle arrest in G1 phase and enhanced
39 cell apoptosis. Moreover, VE-822 significantly sensitized ESCC cells to
40 cisplatin, and these two drugs had synergistic effects, especially in
41 ATM-deficient cells, in vitro and in vivo. Our results suggest that ATR inhibition
42 combining with cisplatin is a new strategy for managing patients with ESCC,
43 especially those with ATM-deficiency. However, this is an idea that requires
44 further validation.
45
46 Highlights:
47  ATR activation was associated with efficacy of neoadjuvant

48 chemotherapy.
49  VE-822 promoted cell apoptosis and induced cell cycle arrest in ESCC
50 cells.
51  VE-822 sensitized ESCC cells to cisplatin, in vitro and in vivo.
52  VE-822 along with cisplatin induces DNA-cisplatin adduct accumulation.
53
54 Key words: DNA damage response; chemoresistance; ATM deficiency; ATR
55 inhibitor
56
57 1. Introduction
58 Esophageal carcinoma (EC) is the 8th most common cancer worldwide[1].
59 EC ranks as the 6th most common cause of cancer-related morbidity and the
60 4th most common cause of cancer-related mortality in China[2]. EC comprises
61 the following two major pathological types: adenocarcinoma and squamous
62 cell carcinoma. The majority of ESCC cases occur in Asia, particularly in north
63 China. Surgery is known as the main treatment for ESCC, but long-term
64 survival of patients with advanced disease remains poor and unsatisfactory.
65 Currently, the development of comprehensive perioperative therapies has
66 greatly improved the efficacy of ESCC treatment, especially for long-term
67 survival. Platinum-based combination regimen is used most frequently in the
68 clinical practice; however, previous studies have demonstrated that the
69 improved efficacy associated with neoadjuvant therapy is limited to patients
70 who respond to chemotherapy, and the prognosis of non-responders is worse
71 compared with that of patients who received surgery alone[3, 4], probably
72 because of chemotherapy resistance, which is inevitable. Thus, new
73 approaches to conquering chemoresistance to improve chemotherapy
74 effectiveness are urgently needed.

75 Platinum-based drugs cross-link with double-stranded DNA and form
76 DNA-platinum abduct to induce DNA damage, leading to cell apoptosis[5]. Due
77 to the intrinsic DDR mechanism, the DNA damage could be repaired. DDR
78 initiation relies on the activation of two major kinase systems, namely,
79 ATR/CHK1 and ATM/CHK2 pathways. Sequentially activated ATR and ATM
80 directly phosphorylate the kinases CHK1 and CHK2, respectively, to activate
81 the downstream effectors such as p53 to upregulate cell cycle checkpoint
82 pathways and then repair the DNA damage[6].Therefore, DDR is an important
83 chemoresistance mechanism through which tumor cells escape from DNA
84 damage induced by genotoxic agents and thus avoid cell death[7-10]. It has
85 been reported that many malignant cancers are characterized by the functional
86 loss or deficiencies in key proteins involved in the DDR, most notably ATM and
87 p53[5, 6, 11-14]. ATM or p53 deficiency in cells leads to synthetic lethality in
88 the presence of ATR depletion[15-19]. Therefore, ATR blockades are
89 considered as promising therapeutic targets, as ATR inhibition may have
90 deleterious effects on cancer cells.
91 Previous studies have demonstrated ATR inhibition is effective for treating
92 cancers combining with chemotherapies in lung adenocarcinoma, gastric
93 cancer, HER2 positive breast cancer and chronic lymphocytic leukemia cells to
94 enhance chemotherapy sensitivity[16-18, 20-22]. VE-822 is an orally, highly
95 specific ATR inhibitor, which has been entered clinical trials. However, we
96 have less knowledge about not only the effect of ATR inhibition in ESCC, but
97 also whether it enhance the chemotherapy sensitivity.
98 In this study, we firstly examined ATM expression status in ESCC and
99 analyzed the association between ATR activation with chemotherapy
100 response evaluated with TRG and overall survival of patients who underwent
101 neoadjuvant chemotherapy. Then, we investigated the effect of VE-822 or
102 combination with cisplatin in ESCC in vitro and in vivo with the context of

103 endogenous ATM activation or ATM deficiency by CRISPR. It was
104 demonstrated that VE-822 could block the activation of ATR, which
105 consequently increase DNA damage and sensitize tumor cells to

106 chemotherapy, with presenting the synthetic lethality effect, especially when
107 ATM was deficient.
108
109 2. Materials and methods
110 2.1 Patients
111 All data of the patients included in this study were retrieved form our
112 prospective database for esophageal cancer, which is established beginning in
113 January 2000 at the Department of Thoracic Surgery, Peking University
114 Cancer Hospital (Beijing, China). From January 2000 to December 2012, 954
115 cases of esophageal cancer underwent surgery, of which, 651 cases were

116 diagnosed as ESCC. According to strict entry criteria, 144 patients who
117 underwent esophagectomy followed by adjuvant chemotherapy and 110
118 patients who underwent neoadjuvant chemotherapy followed by
119 esophagectomy were enrolled in this study. The detailed clinicopathological
120 characteristic of patients were listed in supplementary information. The study
121 was approved by the Ethics and the Academic Committees of Peking
122 University Cancer Hospital (Beijing, China) and informed verbal consent was
123 obtained from all patients.
124
125 2.2 Chemotherapy and surgery methods
126 One hundred and forty-four patients underwent esophagectomy first. After
127 4-6 weeks, they were treated with adjuvant chemotherapy including
128 platinum-based double drug regimen, mainly of which are the paclitaxel and
129 cisplatin at the proportion of 95%. One hundred and ten patients were treated
130 by neoadjuvant chemotherapy including platinum-based double drug

131 combination, mainly the paclitaxel and cisplatin with the proportion of 95%.
132 The curative effects of the treatment were evaluated by enhanced chest
133 computed tomography (CT) and esophagography. Approximately 1-4 cycles of
134 neoadjuvant chemotherapy were administered before surgery. Surgery was
135 carried out 3-5 wk. after neoadjuvant chemotherapy. Chemotherapy regimen
136 was as follows: On day 1, paclitaxel at a dose of 175 mg/m2 of body surface
137 area was administered intravenously. On day 1-3, cisplatin at a dose of 25
138 mg/m2 of body surface area was administered intravenously, a single course of
139 treatment lasted 21 days.
140
141 2.3 Tumor regression grade assessment (TRG)
142 H&E staining results of all the enrolled subjects were reviewed by two
143 experienced pathologists who were blinded to the clinical information and
144 associated issues. Tumors were graded by TRG which was a four-point scale
145 based on the histological tumor response assessment[23]. This assessment
146 was described as: grade I, no residual tumor cells; grade II, nearly complete
147 response with <10% vital residual tumor cells (VRTCs); grade III, 10-50%
148 VRTCs; and grade IV, >50% VRTCs.
149
150 2.4 Follow-up
151 Follow-up evaluation consisted of outpatient interviews at 3-month
152 intervals for 2 years, then at 6-month intervals for 3 years, and finally at
153 12-month intervals until death. Outpatient follow-up visits included recording of
154 symptoms and findings of body examinations such as CT, upper
155 esophagography, ultrasound, and gastroscopy, if necessary. After 2010, some
156 subjects underwent positron emission tomography-computed tomography
157 (PET-CT) examinations. Overall survival (OS) was measured from surgery

158 date until death or the last follow-up. The latest follow-up was June 1st, 2017
159 at the rate of 93%.
160
161 2.5 Immunohistochemistry (IHC)
162 Specimens of the included patients were retrieved from department of
163 pathology, Peking University Cancer Hospital. After routine deparaffinization
164 and hydration, tissue sections were treated with 3% hydrogen peroxide and
165 then heated in citrate solution for antigen retrieval. After antigen retrieval, the
166 sections were incubated with 10% normal goat serum to block any nonspecific
167 reaction. Then, the sections were incubated with rabbit monoclonal anti-ATR
168 (phospho S428) antibody (Abcam, ab178407, at 1:500) or mouse monoclonal
169 anti-ATM antibody (Abcam, ab78, at 1:1000) overnight at 4°C. Dako REAL
170 EnVision Detection System, Peroxisase/DAB, Rabbit/Mouse (K5007), was
171 used as the secondary antibody and for staining. The immunohistochemical
172 signals were scored by two independent pathologists. To evaluate ATR-pS428
173 and ATM expressions, immunohistochemical staining was classified into the
174 following four groups according to intensity. The staining intensity was
175 categorized by relative intensity as follows: 0, negative; 1, weak; 2, moderate;
176 and 3, strong.
177
178 2.6 Cell lines and cell culture
179 Human ESCC cell lines KYSE450, KYSE150, KYSE70, KYSE180, and
180 KYSE 510 were purchased from the Japanese Collection of Research
181 Biosources cell bank (Osaka, Japan). Identities of the cell lines were confirmed
182 by standard STR analysis matched with the American Tissue Culture
183 Collection (ATCC) and Deutsche Sammlung von Mikroorganismen und
184 Zellkulturen GmbH (DSMZ). All cells were passaged for less than 1 year
185 before use and cultured in RPMI-1640 medium (Hyclone; GE Healthcare,

214 Cisplatin modified DNA accumulation was examined using flow cytometry
215 (BD, Biosciences) by anti-cisplatin modified DNA antibody (Abcam, ab103261).
216 Cells in culture were treated with cisplatin, VE-822, or combination of these
217 two drugs, with un-treated cells as control. Cells were fixed with 70% ehtanol
218 for 30 min at 4°C and permeabilized with 0.3% TritonX-100 in PBS. The cells
219 were incubated with the primary antibody (1:200) for 18 hours at 4°C. A rabbit
220 anti-rat IgG/Alexa Fluor 488 (Bioss, 1/100) was used as the secondary
221 antibody. An isotype control used Rat IgG2a kappa monoclonal (Abcam,
222 ab18450), simultaneously.
223
224 2.10 Western blotting
225 The proteins were extracted by using RIPA lysis buffer and separated by
226 10% SDS-PAGE with 30µg protein per lane and transferred onto a
227 polyvinylidene uoride membrane, followed by western blot analysis. The
228 membrane was blocked using 5% bovine serum albinum at room temperature
229 for 1 h. It was then immunoreacted with, ATM, p-ATM (S1981), ATR, p-ATR
230 (Ser428), Chk1, p-Chk1, Chk2, p-Chk2, p53, p21, p-STAT3, STAT3, p-AKT,
231 AKT, caspase-3, p-histone H2A.X and PARP were acquired from Cell
232 Signaling Technology(America). GAPDH (ZSGB-BIO, China) were also
233 purchased. Goat anti-rabbit or anti-mouse polyclonal IgG was used as a
234 secondary antibody.
235
236 2.11 Cell proliferation assay (CCK-8 assay)

237 Five thousand cells per well were plated in 96-well plates overnight at
238 37°C and treated with gradient dilution of cisplatin (0.125, 0.25, 0.5, 1, 2, 4, 8,
239 16 and 32 µg/ml) or VE-822 (0.125, 0.25, 0.5, 1, 2, 4, 8, 16 and 32 µM), or
240 combination of cisplatin (0.125, 0.25, 0.5, 1, 2, 4, 8, 16 and 32 µg/ml) and
241 VE-822 (2 µM). After 48h of incubation, 10µl CCK-8 reagent (Dojindo

242 Molecular Technologies Inc., Kumamoto, Japan) was added to each well
243 about 2h at 37°C. Then, the absorbance of each well was examined at 450 nm
244 by the Microplate reader (iMark, Bio-rad, USA).
245
246 2.12 Cell cycle analysis
247 Cells at a density of 1*106 per well were plated in 6-well plates overnight at
248 37°C and treated with cisplatin, or VE-822 or combination for 24h until cells
249 were harvested. Cells at a density of 1*106 were collected and fixed with 70%
250 cold ethanol overnight at 4°C. After fixation, cells were washed 3 times in PBS.
251 Then, the PI-staining solution with RNase A (BD Biosciences) was added
252 about 30 min in room temperature to stain samples and were run on the
253 FACScan cytometry (BD Biosciences, America), and data were analyzed
254 using FlowJo software (Tree Star).
255
256 2.13 Apoptosis analysis in vitro
257 Cells at a density of 1*106 per well were plated in 6-well plates overnight at
258 37°C and treated with cisplatin, or VE-822 or combination for 8h until cells
259 were harvested. Cells were washed 3 times in PBS and incubated in trypsin
260 (without EDTA) at 37°C for 10min. Cells were rinsed 3 times in PBS and
261 re-suspended in binding buffer (Dojindo, Japan). Annexin V-FITC antibody
262 (Dojindo, Japan, 5 µl) and PI (Dojindo, Japan, 5 µl) were added in cells (1 ×
263 105cells/100µl) and incubated for 15 min at room temperature in the dark.
264 Then, the samples were analyzed by flow cytometry (BD Biosciences, America)
265 within 1 h.
266
267 2.14 Immunofluorescence（γ-H2AX and p53 foci formation）

268 Before the assay, the cells treated with 2µg/ml cisplatin, 2µM VE-822 and
269 the combination of these two drugs for 24h. The coverslips were rinsed 3 times

270 in PBS, fixed in 3.7% paraformaldehyde for 15min, permeabilized with PBST
271 (0.5% Triton X-100 in PBS) for 5min and blocked specimen in blocking buffer
272 (5% normal serum in PBST) for 60min at room temperature. After washing 3
273 times in PBS, the coverslips were incubated with phospho-Histone H2A.X
274 (Ser139) primary antibody (#9718, Cell Signaling Technology; America) at a
275 dilution of 1:200 and p53 primary antibody at a dilution of 1:2000 (#2524, Cell
276 Signaling Technology; America) overnight at 4°C. Then, the coverslips were
277 washed 3 times in PBS and incubated with the appropriate
278 fluorophore-conjugated secondary antibody: TRITC for γ-H2A.X
279 (tetramethylrhodamine goat anti-rabbit IgG, Invitrogen, American) and FITC for
280 p53 (fluorescein goat-mouse IgG, Invitrogen, American) for 1h at room
281 temperature in the dark. Finally, the cells were counterstained with DAPI (300
282 nmol/l; Invitrogen, American). Immunofluorescence was visualized by Zeiss
283 scanning microscope(Germany).
284
285 2.15 RNA-seq library construction and sequencing
286 Total RNA from KYSE450 and KYSE450 treated with cisplatin (1µg/ml)
287 and VE-822(1µM) (named KYSE450.1) with three replications was isolated
288 using Trizol for the construction of a RNA-seq library and sequencing. The
289 construction of RNA-seq library was performed using the KAPA Stranded
290 mRNA-Seq Kit (Illumina® platform) (product codes KK8420 and KK8421,
291 Boston, Massachusetts, United States) following the manufacturer’s
292 instructions. Briefly, mRNA was extracted and purified from total RNA ,then
293 fragmented and primed for cDNA synthesis. Double-stranded cDNAs were
294 synthesized and then purified with 1.8× Agencourt AMPure XP beads
295 (Beckman Coulter, Beverly, USA) followed by the end 2nd Strand Synthesis.
296 After A-Tailing, Illumina adapter oligonucleotides were ligated to cDNA
297 fragments, and the 1X SPRI® cleanup was performed. Suitable cDNA

298 fragments were selected as templates for PCR amplification using the KAPA
299 Library Amplification Primer Mix and KAPA HiFi Hot Start Ready Mix. Products
300 were purified with the AMPure XP bead system and quantified using a
301 Bioanalyzer (Agilent high sensitivity chip). Finally, RNA-seq libraries were
302 sequenced using an Illumina HiSeq at Beijing Microread Genetics Co Ltd
303 (Beijing, China). Raw data were processed with Fastp using recommended
304 parameters.The filtered reads were mapped to Hg19 by hista2 . The bam files
305 were processed with samtools. Feature Counts was used to calculate gene
306 expression. A list of differential expression genes (DEGs) was identified using
307 the R packages” EdgeR”. P-value of 0.05 and |log2(foldchange)|>2 were set
308 as the threshold for significantly differential expression by default. Gene
309 Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG)
310 enrichment analysis of differentially expressed genes were implemented with
311 KOBAS3.0. GO classification was done by the R packages “TopGO”.To
312 further validate the RNAseq results, we selected 10 DEGs of our interests to
313 examine their expression in both samples by using qRT-PCR. The
314 gene-specific primers for these twenty genes are listed in Table S1.
315
316 2.16 Primary ESCC xenograft models
317 BALB/c nude mice were raised under the care of the Laboratory Animal
318 Unit of First Affiliated Hospital of PLA General Hospital, China. KYSE70,
319 KYSE70 ATM (-), KYSE450, and KYSE450 ATM (-) (3 × 106) cells suspending
320 in 200 µl PBS were injected into the right groin of 8 weeks old female Balb/C
321 nude mice. All mice were housed and raised under specific pathogen-free
322 conditions. The sizes of tumors and body weight of each mouse were
323 measured every 3 days. Tumor volumes were calculated using the following
324 formula: tumor volume = [(length) × (width) × (width)] / 2. When the tumor
325 volume reached about 200 mm3, the mice were divided into two groups

326 randomly (6 mice per group). First group of mice were given 5mg/kg cisplatin
327 on days1, 4, 7, 10 via intravenous injection. The second group were given 60
328 mg/kg VE-822 on days 1, 3, 5 via oral gavage. The third group were given both
329 of cisplatin and VE-822, the last group were given PBS as control. At the end
330 of the measurement period, excised tumors were measured by a slide caliper
331 for volume and weighted by an electronic analytical balance. All experiments
332 were done in accordance with institutional standard guidelines of Peking
333 University Cancer Hospital and Unit of First Affiliated Hospital of PLA General
334 Hospital for animal experiments.
335
336 2.17 Statistical analysis
337 SPSS software (version 24.0; IBM SPSS, Armonk, NY, USA) was used to
338 perform the statistical analysis. The relationships between ATR-pSer428 and
339 ATM expression and clinicopathologic characteristics were tested using
340 Chi-square test. Survival curves were plotted by Kaplan-Meier method and
341 compared by log rank test. The association between gene expression and
342 TRG were evaluated using a  test. All in vitro experiments were performed at
343 least 3 times with triplicates. Comparisons between groups for statistical
344 significance were performed with a 2-tailed unpaired Student’s t test. Bars and
345 error bars on the graphs as well as data in the text represent the mean ± SD. P
346 < 0.05 was considered statistically significant.
347
348 3. Results
349 3.1 ATM protein was deficient in 21.5% ESCC patients
350 Many studies have reported that various tumors had ATM deficiency with
351 different degrees. Therefore, we examined the expression rate of ATM through
352 IHC in 144 ESCC patients’ samples without preoperative treatment(Fig 1a).
353 The result showed that 21.5%(31/144) of cases were ATM expression

354 negative. ATM expression was not associated with the clinicopathologic
355 factors such as age, gender, pathologic stage and tumor location as well as
356 overall survival (Fig S1, Table S2).
357
358 3.2 ATR activation was associated with efficacy of neoadjuvant
359 chemotherapy
360 ATR-pSer428 is the active form of ATR protein. ATR-pSer428 expression
361 was examined in a cohort of 110 ESCC patients underwent neoadjuvant
362 chemotherapy. Then, we evaluated the relationship between the activation of
363 ATR (substituted by the expression of ATR-pSer428) in resected specimens
364 and overall survival or TRG. The expression of ATR-pSer428 in ESCC mainly
365 occur in nucleus. Among the 110 subjects in the study, 25.5% (28/110) cases
366 were ATR-pSer428 negative, whereas 74.5%(82/110) cases were
367 ATR-pSer428 positive. Then, we divided the patients with ATR-pSer428
368 expression into subgroups according to ATR-pSer428 expression intensity,
369 which was graded 0, 1, 2, or 3 (Fig 1b). We found that 25.5% (28/110), 17.3%
370 (19/110), 32.7% (26/110), and 24.5% (27/110) of patients displayed grade 0, 1,
371 2 and 3, respectively. Kaplan-Meier analysis of the 110 subjects showed that
372 more than half of the subjects without ATR-pSer428 expression survived until
373 the follow-up endpoint (144 months), whereas the median survival time for
374 ESCC patients with ATR-pSer428 expression was only 32 months (P < 0.05).
375 The median survival time of patients with grade 1, 2, and 3 were 38, 28 and 28
376 months, respectively. More than half of the subjects with grade 0 expression
377 survived until the follow-up endpoint (144 months) (P < 0.05) (Table 1, Fig 1c).
378 We further defined grades 0 and 1 as low-level expression and grades 2 and 3
379 as high-level expression. Thus, 42.7% (47/110) of patients had low-level
380 expression of ATR-pSer428, and 57.2% (63/110) of patients had high-level
381 expression of ATR-pSer428. Half of the patients with low-level expression of

382 ATR-pSer428 survived until the follow-up endpoint (144 months). The median
383 survival time of the patients with high-level expression of ATR-pSer428 was 28
384 months (P < 0.05) (Table 1, Fig 1c).
385 TRGs is currently the standard pathological indicators of neoadjuvant
386 chemotherapy responsiveness. In this study, we found that ATR-pSer428
387 expression was significantly associated with TRGs. ATR-pSer428 expression
388 in subjects with TRGs 2/3/4 was higher than that in subjects with TRGs 1 (P <
389 0.001) (Table 2). That is, patients with ATR activation after chemotherapy
390 display an unfavorable response to chemotherapy compared with those
391 without ATR activation, indicating that ATR-pSer428 expression can be used
392 to determine chemotherapy sensitivity.
393
394 3.3 DDR signaling pathway was activated in ESCC cell lines and was
395 inhibited by VE-822, which also inhibited cell proliferation
396 To assess the baseline of ATM/ATR signaling pathway activation, we
397 performed western blotting to examine the expression of phosphorylated ATM,
398 ATR, CHK1, and CHK2, which serve as surrogate markers for ATR and ATM
399 pathway activation, and the expression of the downstream target p53 in
400 KYSE70, KYSE150, KYSE180, KYSE450, and KYSE510 cell lines. The
401 results showed that p-ATM, p-ATR, p-CHK1, p-CHK2 and p53 were
402 endogenously activated in all ESCC cell lines (Fig 2a and Fig S2). Based on
403 these results, we inferred that ATM/ATR signaling pathway activation is
404 essential for overcoming replication stress and sustaining tumor cell genomic
405 stability. In these five cell lines, the baseline ATR activation was stronger in
406 KYSE450 and KYSE70 than other cell lines.Then, we used CCK-8 assay to
407 investigate the effect of VE-822 on cell viability. Forty-eight hours of treatment
408 with VE-822 robustly inhibited ESCC cell viability. The half maximal inhibitory
409 concentration (IC50) were 3.982, 11.870, 2.606, 6.922 and 9.387 in KYSE450,

410 KYSE150, KYSE510, KYSE180 and KYSE70 cells, respectively (Fig 2b). We
411 chose two ESCC cell lines, namely, KYSE70 and KYSE450, for the ATR
412 inhibition experiments. We found that p-ATR and p-CHK1 activation was
413 significantly inhibited by the ATR inhibitor VE-822. We also found that DNA
414 damage remarkably accumulated and γ-H2AX fluorescence intensity amplified,
415 as detected by Western Blotting (Fig 2c). In addition, VE-822 at least partially
416 inhibited cancer cell proliferation through Stat-Akt signaling pathway inhibition
417 (Fig 2d).
418
419 3.4 VE-822 promoted cell apoptosis and induced cell cycle arrest in
420 ESCC cells
421 ATR activation is a key step in DDR initiation, as ATR-pSer428 induces
422 activation of DNA damage checkpoint signaling, which induces cell cycle
423 arrest. ATR plays an important role in the G2-M phase transition. Consistent
424 with this finding, we found that the cell cycle was arrested in G1 phase in a
425 concentration-dependent manner after the cells being exposed to increasing
426 concentrations of VE-822 (1.0 µM, 2.0 µM, and 4.0 µM) for 24 h. As Fig 2e
427 shown, VE-822 increased the fraction of cells in G1 phase to 68.48%
428 (compared with baseline 52.62%) in KYSE70 cells and 42.22% (compared
429 with baseline 37.48%) in KYSE450 cells when administered at a dose of 4.0
430 µM. VE-822-induced cell cycle arrest also led to the downregulation of p21
431 expression (Fig 2f).
432 To determine the effect of VE-822 on cell apoptosis, we examined the ratio
433 of apoptotic cells to live cells using flow cytometry. VE-822 induced a
434 significant concentration-dependent increase in cell apoptosis in both KYSE70
435 and KYSE450 cells after 8 h of treatment (Fig 2g). We also examined the
436 expression of caspase-3, cleaved PARP by western blotting, as the indicated

437 proteins are biomarkers for cell apoptosis. The results showed that caspase-3
438 and cleaved PARP expression was increased by exposure to VE-822 (Fig 2g).
439
440 3.5 VE-822 sensitized ESCC cells, especially ATM-deficient cells, to
441 cisplatin in vitro and in vivo
442 Cisplatin is commonly used as a first-line chemotherapy for patients with
443 ESCC. However, chemoresistance inevitably occurs spontaneously or
444 develops during treatment. Previous studies have demonstrated that ATR
445 kinase inhibitors could enhance the sensitivity of cancer cells to DNA
446 damaging agents, such as cisplatin, in solid-tumor models both in vitro and in
447 vivo. To validate the hypothesis that ATR inhibition enhances the efficacy of
448 cisplatin in ESCC, we performed CCK-8 assay to evaluate the viability of cells
449 exposed to cisplatin, VE-822 or the combination of cisplatin and VE-822 and to
450 determine the effect of these treatments on tumor growth in ESCC mouse
451 xenograft models. The combination of ATR inhibition and cisplatin
452 synergistically inhibited cell viability in KYSE 450 and KYSE70 cells. The
453 combination index(CI) values were calculated according to the
454 Chou-Talalay[24] median-effect principle. As shown in the figure 3a, there is
455 significant synergistic effect between VE-822 and cisplatin. Then, to
456 investigate whether ATM deficiency affect the efficacy of VE-822, we knockout
457 ATM expression by using CRISPR and established stable cell strain by
458 screening (Fig 3b). And we found that IC50 of VE-822 was decreased
459 significantly in ATM (-) cells (Fig 3c) and more striking synergistic inhibition
460 combination treatment was observed in ATM-deficient cells than in cells
461 expressing ATM (Fig 3d). Then, we treated the mouse xenograft models with
462 cisplatin (5 mg/kg) on days 1, 3, and 5; VE-822 (60 mg/kg) on days 1, 2, and 3;
463 or combination therapy, with PBS as control. The growth of tumors treated with
464 combination therapy was significantly slower than that of tumors treated with

465 cisplatin or VE-822 alone. On day 14, the TGI of KYSE70 xenograft tumor
466 were 28.8%、25.9% and 73.3% for cisplatin-alone group, VE-822-alone group
467 and combination group respectively. And on day 16, the TGI of KYSE450
468 xenograft tumor were 58.7%, 30.6% and 83.2%, respectively. However, the
469 TGI of ATM-deficient KYSE450 xenograft tumor were 38.0%, 48.1% and 85.2%
470 respectively (Fig 3e). The result indicated that ATM-deficiency magnified the
471 effect of VE-822 in KYSE450 cell line. However, body weight loss in mice
472 treated with combination therapy was not significantly greater than that in mice
473 treated with monotherapy (Fig 3f). Furthermore, p-CHK1, p-AKT activation
474 was inhibited, and cleaved-PARP expression was increased as detected by
475 Western Blotting (Fig 3g). The accumulation of cells in S phase was greater in
476 ATM-deficient KYSE450 and KYSE70 cells than control cells treated with
477 VE-822 (Fig 4a). VE-822 reinforced the effects of cisplatin to induce cell
478 apoptosis, especially for ATM-deficient cells. As Fig 4b shown, the
479 combination of VE-822 and cisplatin caused a significant increase in cell
480 apoptosis in both ATM-deficient KYSE450 and KYSE70 cells compared with
481 control cells. Accordingly, caspase-3 and cleaved-PARP expression
482 remarkably increased (Fig 4c). Then, we observed that VE-822 induced
483 accumulated DNA damage and amplified γ-H2AX and p53 fluorescence
484 intensity in wild-type cells (Figure 5a and 5b) and ATM-deficiency cells (Fig 5c
485 and 5d). These results suggested that the combination of VE-822 and cisplatin
486 has therapeutic potential, especially in ATM-deficiency cells.
487
488 3.6 ATR inhibition along with cisplatin induces DNA-cisplatin adduct
489 accumulation, especially in ATM (-) ESCC cells
490 Cisplatin cross-links with DNA to form cisplatin-DNA adducts, causing cells
491 to experience replication stress and undergo apoptosis; however, cells have
492 an endogenous repair mechanism, known as the DDR, to conquer this type of

493 stress. Inhibitors targeting the DDR can theoretically block the DDR and
494 enhance cisplatin-DNA adduct formation. To explore the mechanism by which
495 VE-822 sensitizes ESCC cells to cisplatin, we used an anti-cisplatin-modified
496 DNA antibody to detect cisplatin-modified DNA accumulation in cells treated
497 with cisplatin, VE-822 or the combination of the two drugs. The results showed
498 that VE-822 combined with cisplatin induced greater accumulation of modified
499 DNA than either agent alone, especially in ATM-deficient ESCC cells (Figure
500 6). In conclusion, VE-822 sensitized ESCC cells to cisplatin by increasing the
501 DNA damage induced by cisplatin and inducing the enrichment of cisplatin in
502 cells.
503
504 4. Discussion
505 Chemoresistance to platinum-based chemotherapy has been a huge
506 challenge with respect to achieving optimal treatment outcomes in ESCC.
507 DNA damage induced by cisplatin-DNA adducts causes DNA replication
508 stress and triggers the DDR, which is considered an important mechanism for
509 the development of chemoresistance [6]. Previous studies have shown that
510 DNA replication stress induced by genotoxic agents, such as cisplatin, allows
511 inhibitors targeting the DDR pathway to serve as an effective therapy for
512 ESCC. ATM and ATR inhibitors have been entered into clinical trials pertaining
513 to some types of solid cancers. However, whether DDR inhibitors can be used
514 in ESCC treatment is not known. In our study, we found that endogenous DNA
515 replication stress occurs in ESCC cells. p-ATM and p-ATR were endogenously
516 expressed in ESCC cells, and their activation was significantly enhanced upon
517 exposure to cisplatin. This finding serves as an important clue indicating that
518 DDR pathway inhibitors can be used in ESCC treatment.
519 Interestingly, we found that ATR activation (represented by ATR-pSer428
520 expression) was associated with a pathologic response to chemotherapy and

521 poorer long-term outcomes in patients who underwent neoadjuvant
522 chemotherapy, indicating that p-ATR expression may be used as a biomarker
523 for chemoresistance in ESCC. Therefore, we hypothesized that the
524 combination of ATR inhibition and platinum is an effective therapy for ESCC,
525 which relies on ATR signaling to facilitate DNA repair. We found that ESCC
526 cells were sensitized to cisplatin upon exposure to an ATR inhibitor in vitro and
527 in vivo, a finding supported by the data pertaining to cell viability. Additionally,
528 we demonstrated that STAT3 may play a critical role in VE-822-mediated
529 effects, as p-STAT3 expression was remarkably inhibited in KYSE450 and
530 KYSE70 cells treated with the combination of cisplatin and VE-822. In previous
531 study, it has been shown that STAT3 was essential for efficient repair of
532 damaged DNA, which was suppressed when DNA damage response was
533 inhibited[25-27]. To further learn the effect of combination of VE-822 and
534 cisplatin on the transcriptional profiling, we performed RNA-seq in KYSE450
535 exposure to combination of VE-822 and cisplatin. DEG analysis identified 1163
536 genes significantly altered in cells bearing treatment, with 555 genes
537 up-regulated and 608 genes down-regulated(Fig S3a,b). These genes were
538 then classified based upon their signaling pathways and biological functions
539 using KEGG analysis(Fig S3c,d) and GO analysis(Fig S3e,f). We found that
540 treatment with the combination of VE-822 and cisplatin caused a wide of
541 pathways changes including Hippo pathway, MAPK pathway, JAK-STAT
542 pathway, PI3K-AKT pathway and platinum resistance pathway, etc. To
543 validate the RNA-seq results, we selected 10 DEGs to examine their
544 expression by RT-PCR. These 10 genes, were enriched into drug resistance
545 related signaling pathways, including p53 pathway, apoptosis, platinum drug
546 resistance, PI3K-Akt signaling pathway[28-31] and JAK-STAT signaling
547 pathway[32]. The expression patterns of selected DEGs in the RNA-seq and
548 RT-PCR were highly similar(Fig S4). It indicated that our RNA-seq results was

549 reliable, and on the other hand, it implied that the dysregulated signaling
550 pathway may be the key to explore the mechanism underlying the
551 sensitization.
552 Moreover, we demonstrated that the combination of cisplatin and VE-822
553 was effective in vivo in KYSE450 and KYSE70 tumor xenograft models but did
554 not affect mouse body weights, suggesting that ATR inhibition is well tolerated.
555 Multiple in vivo studies have established that ATR inhibitors do not exacerbate
556 the toxic effects of multiple genotoxic agents but still synergize with genotoxic
557 therapies.
558 Given that DDR initiation depends on ATM and ATR pathway activation,
559 the presence of ATM may influence the effects of ATR inhibitors. Interestingly,
560 ATM deficiency was detected in 21.5% of patients with ESCC, which
561 motivated us to hypothesize that using an ATR inhibitor in patients with ATM
562 deficiency will lead to synthetic lethality and that patients with ATM deficiency
563 will benefit more from ATR inhibitor treatment. We obtained ATM-deficient
564 ESCC cells by knocking down ATM expression with the CRISPR method. In
565 ATM-deficient KYSE450 and KYSE70 cells, the combination of cisplatin and
566 VE-822 induced cell cycle aberrations. Specifically, more cells accumulated at
567 S phase. This was not the case in wild-type KYSE cells. Both cell lines
568 exhibited a greater apoptotic response following treatment with the
569 combination of VE-822 and cisplatin, as demonstrated by the data showing
570 that caspase-3 expression and PARP cleavage were increased in combination
571 therapy-treated cells. Similar results were observed in the VE-822-alone group.
572 Thus, ATM deficiency may be a predictive biomarker for tumor responses to
573 ATR inhibitor monotherapy and combination therapies. Additionally, previous
574 preclinical studies have identified a series of tumor-specific alterations that
575 affect sensitivity to ATR inhibition. These include defects in the ATM and p53
576 pathways. However, the usefulness of ATM deficiency as a biomarker has not

577 been extensively examined in ESCC; thus, additional works regarding this
578 issue are required.
579 Exposure to cisplatin with ATR inhibitor resulted in an increase in
580 cisplatin-DNA adducts, especially in cells with ATM deficiency. This finding
581 indicates that suppressing ATR-Chk1 signaling with VE-822 enhances
582 cisplatin activity by enabling the drug to form DNA adducts. Therefore, VE-822
583 may increase cisplatin-DNA accumulation by hindering DNA damage repair
584 induced by cisplatin-DNA. On the other hand, in the previous study, the cells
585 treated with VE-822 along with cisplatin had decreased expression of
586 p-glycoprotein. It inferred that VE-822 inhibited the expression of
587 p-glycoprotein to prevent cisplatin efflux, then increasing its concentration[33].
588 In conclusion, our study revealed that endogenous DDR signaling
589 activation plays a critical role in mediating the cisplatin resistance of ESCC.
590 We determined that the ATR inhibitor VE-822 may be an attractive treatment
591 for ESCC, especially when used in combination with cisplatin, as VE-822
592 induces significant sensitization to cisplatin. Specifically, VE-822 synergizes
593 with cisplatin in ATM-deficient models of ESCC. Additional works are
594 warranted to explore the possibility that ATM deficiency can serve as a
595 biomarker enabling the prospective identification of patients with ESCC who
596 will benefit most from the combination of ATR inhibition and cisplatin.
597
598 5. Acknowledgements
599 This work was supported by Beijing Municipal Administration of Hospitals 600 Incubating Program (PX2018044); National Natural Science Foundation for 601 Young Scholars (Grant 81301748); National High Technology Research and 602 Development Program of China (2015AA020403); and Beijing Municipal 603 Administration of hospitals Clinical Medicine Development of special funding 604 support (ZYLX201509).

605
606 6. Conflicts of Interest
607 No potential conflicts of interest were disclosed.
608
609 7. Author Contribution
610 Conceived and designed the experiments: Keneng Chen, Luyan Shen. 611 Performed the experiments: Qi Shi, Luyan Shen. Analyzed the data: Luyan 612 Shen, Qi Shi, Bin Dong. Wrote the paper: Luyan Shen, Qi Shi. Patients care, 613 cases provision and data collection: Keneng Chen, Wanpu Yan, Liang Dai, 614 Xiaozheng Kang, Yongbo Yang, Hongchao Xiong, Zhen Liang.
615
616 8. Reference
617 [1] J. Ferlay, I. Soerjomataram, R. Dikshit, S. Eser, C. Mathers, M. Rebelo, 618 D.M. Parkin, D. Forman, F. Bray, Cancer incidence and mortality worldwide: 619 sources, methods and major patterns in GLOBOCAN 2012, Int J Cancer, 136 620 (2015) E359-386.
621 [2] W. Chen, R. Zheng, P.D. Baade, S. Zhang, H. Zeng, F. Bray, A. Jemal, X.Q. 622 Yu, J. He, Cancer statistics in China, 2015, CA Cancer J Clin, 66 (2016) 623 115-132.
624 [3] A.R. Davies, J.A. Gossage, J. Zylstra, F. Mattsson, J. Lagergren, N. Maisey, 625 E.C. Smyth, D. Cunningham, W.H. Allum, R.C. Mason, Tumor stage after 626 neoadjuvant chemotherapy determines survival after surgery for 627 adenocarcinoma of the esophagus and esophagogastric junction, J Clin Oncol, 628 32 (2014) 2983-2990.
629 [4] Medical Research Council Oesophageal Cancer Working Party, Surgical 630 resection with or without preoperative chemotherapy in oesophageal cancer: a 631 randomised controlled trial, The Lancet, 359 (2002) 1727-1733.

632 [5] P.M. Takahara, A.C. Rosenzweig, C.A. Frederick, S.J. Lippard, Crystal 633 structure of double-stranded DNA containing the major adduct of the 634 anticancer drug cisplatin, Nature, 377 (1995) 649-652.
635 [6] A.M. Weber, A.J. Ryan, ATM and ATR as therapeutic targets in cancer,
636 Pharmacol Ther, 149 (2015) 124-138.
637 [7] L. Galluzzi, L. Senovilla, I. Vitale, J. Michels, I. Martins, O. Kepp, M. 638 Castedo, G. Kroemer, Molecular mechanisms of cisplatin resistance, 639 Oncogene, 31 (2012) 1869-1883.
640 [8] J. Jin, H. Fang, F. Yang, W. Ji, N. Guan, Z. Sun, Y. Shi, G. Zhou, X. Guan, 641 Combined Inhibition of ATR and WEE1 as a Novel Therapeutic Strategy in 642 Triple-Negative Breast Cancer, Neoplasia, 20 (2018) 478-488.
643 [9] S.B. Koh, Y. Wallez, C.R. Dunlop, S. Bernaldo de Quiros Fernandez, T.E. 644 Bapiro, F.M. Richards, D.I. Jodrell, Mechanistic distinctions between CHK1 645 and WEE1 inhibition guide the scheduling of triple therapy with gemcitabine, 646 Cancer Res, (2018) pii: canres.3932.2017.
647 [10] A. Desai, Y. Yan, S.L. Gerson, Advances in therapeutic targeting of the
648 DNA damage response in cancer, DNA Repair (Amst), 66-67 (2018) 24-29. 649 [11] L.C. Villaruz, H. Jones, S. Dacic, S. Abberbock, B.F. Kurland, L.P. Stabile, 650 J.M. Siegfried, T.P. Conrads, N.R. Smith, M.J. O'Connor, A.J. Pierce, C.J. 651 Bakkenist, ATM protein is deficient in over 40% of lung adenocarcinomas, 652 Oncotarget, (2016) 57714-57725.
653 [12] F. Bullrich, D. Rasio, S. Kitada, P. Starostik, T. Kipps, M. Keating, M. 654 Albitar, J.C. Reed, C.M. Croce, ATM mutations in B-cell chronic lymphocytic 655 leukemia, Cancer Res, 59 (1999) 24-27.
656 [13] H. Kim, B. Saka, S. Knight, M. Borges, E. Childs, A. Klein, C. Wolfgang, J. 657 Herman, V.N. Adsay, R.H. Hruban, M. Goggins, Having pancreatic cancer with 658 tumoral loss of ATM and normal TP53 protein expression is associated with a 659 poorer prognosis, Clin Cancer Res, 20 (2014) 1865-1872.

660 [14] K.G. Wiman, Restoration of wild-type p53 function in human tumors:
661 strategies for efficient cancer therapy, Adv Cancer Res, 97 (2007) 321-338. 662 [15] P.M. Reaper, M.R. Griffiths, J.M. Long, J.D. Charrier, S. Maccormick, P.A. 663 Charlton, J.M. Golec, J.R. Pollard, Selective killing of ATM- or p53-deficient 664 cancer cells through inhibition of ATR, Nat Chem Biol, 7 (2011) 428-430.
665 [16] A. Schmitt, G. Knittel, D. Welcker, T.P. Yang, J. George, M. Nowak, U. 666 Leeser, R. Buttner, S. Perner, M. Peifer, H.C. Reinhardt, ATM Deficiency Is 667 Associated with Sensitivity to PARP1- and ATR Inhibitors in Lung 668 Adenocarcinoma, Cancer Res, 77 (2017) 3040-3056.
669 [17] A. Min, S.A. Im, H. Jang, S. Kim, M. Lee, D.K. Kim, Y. Yang, H.J. Kim, K.H.
670 Lee, J.W. Kim, T.Y. Kim, D.Y. Oh, J. Brown, A. Lau, O.C. MJ, Y.J. Bang,
671 AZD6738, a novel oral inhibitor of ATR, induces synthetic lethality with 672 ATM-deficiency in gastric cancer cells, Mol Cancer Ther, 16 (2017) 566-577. 673 [18] Kwok M, Davies N, Agathanggelou A, Smith E, Oldreive C, Petermann 674 E, Stewart G, Brown J, Lau A, Pratt G, Parry H, Taylor M, Moss P, Hillmen 675 P, Stankovic T, ATR inhibition induces synthetic lethality and overcomes 676 chemoresistance in TP53- or ATM-defective chronic lymphocytic leukemia 677 cells, Blood, 127 (2016) 583-595.
678 [19] G. Manic, M. Signore, A. Sistigu, G. Russo, F. Corradi, S. Siteni, M. 679 Musella, S. Vitale, M.L. De Angelis, M. Pallocca, C.A. Amoreo, F. Sperati, S. 680 Di Franco, S. Barresi, E. Policicchio, G. De Luca, F. De Nicola, M. Mottolese, A. 681 Zeuner, M. Fanciulli, G. Stassi, M. Maugeri-Sacca, M. Baiocchi, M. Tartaglia, I. 682 Vitale, R. De Maria, CHK1-targeted therapy to deplete DNA 683 replication-stressed, p53-deficient, hyperdiploid colorectal cancer stem cells, 684 Gut, 67 (2018) 903-917.
685 [20] H.J. Kim, A. Min, S.A. Im, H. Jang, K.H. Lee, A. Lau, M. Lee, S. Kim, Y.
686 Yang, J. Kim, T.Y. Kim, D.Y. Oh, J. Brown, M.J. O'Connor, Y.J. Bang,

687 Anti-tumor activity of the ATR inhibitor AZD6738 in HER2 positive breast
688 cancer cells, Int J Cancer, 140 (2017) 109-119.
689 [21] L.M. Karnitz, L. Zou, Molecular Pathways: Targeting ATR in Cancer
690 Therapy, Clin Cancer Res, 21 (2015) 4780-4785.
691 [22] E. Fokas, R. Prevo, E.M. Hammond, T.B. Brunner, W.G. McKenna, R.J. 692 Muschel, Targeting ATR in DNA damage response and cancer therapeutics, 693 Cancer Treat Rev, 40 (2014) 109-117.
694 [23] L.R. Chirieac, S.G. Swisher, J.A. Ajani, R.R. Komaki, A.M. Correa, J.S. 695 Morris, J.A. Roth, A. Rashid, S.R. Hamilton, T.T. Wu, Posttherapy pathologic 696 stage predicts survival in patients with esophageal carcinoma receiving 697 preoperative chemoradiation, Cancer, 103 (2005) 1347-1355.
698 [24] T.C. Chou, Drug combination studies and their synergy quantification
699 using the Chou-Talalay method, Cancer Res, 70 (2010) 440-446.
700 [25] L. Chen, L. Fu, X. Kong, J. Xu, Z. Wang, X. Ma, Y. Akiyama, Y. Chen, J. 701 Fang, Jumonji domain-containing protein 2B silencing induces DNA damage 702 response via STAT3 pathway in colorectal cancer, Br J Cancer, 110 (2014) 703 1014-1026.
704 [26] S. Koganti, J. Hui-Yuen, S. McAllister, B. Gardner, F. Grasser, U. 705 Palendira, S.G. Tangye, A.F. Freeman, S. Bhaduri-McIntosh, STAT3 interrupts 706 ATR-Chk1 signaling to allow oncovirus-mediated cell proliferation, Proc Natl 707 Acad Sci U S A, 111 (2014) 4946-4951.
708 [27] W.W. Deng, Q. Hu, Z.R. Liu, Q.H. Chen, W.X. Wang, H.G. Zhang, Q.
709 Zhang, Y.L. Huang, X.K. Zhang, KDM4B promotes DNA damage response via 710 STAT3 signaling and is a target of CREB in colorectal cancer cells, Mol Cell 711 Biochem, (2018).
712 [28] B. Liu, C. Wang, P. Chen, B. Cheng, Y. Cheng, RACKI induces
713 chemotherapy resistance in esophageal carcinoma by upregulating the

714 PI3K/AKT pathway and Bcl-2 expression, Onco Targets Ther, 11 (2018)
715 211-220.
716 [29] Z. Amini-Farsani, M.H. Sangtarash, M. Shamsara, H. Teimori, 717 MiR-221/222 promote chemoresistance to cisplatin in ovarian cancer cells by 718 targeting PTEN/PI3K/AKT signaling pathway, Cytotechnology, 70 (2018)
719 203-213.
720 [30] Y. Ma, G. Zhou, M. Li, D. Hu, L. Zhang, P. Liu, K. Lin, Long noncoding 721 RNA DANCR mediates cisplatin resistance in glioma cells via activating 722 AXL/PI3K/Akt/NF-kappaB signaling pathway, Neurochem Int, (2018) pii: 723 S0197-0186(18)30024-X.
724 [31] S.C. Tripathi, J.F. Fahrmann, M. Celiktas, M. Aguilar, K.D. Marini, M.K. 725 Jolly, H. Katayama, H. Wang, E.N. Murage, J.B. Dennison, D.N. Watkins, H. 726 Levine, E.J. Ostrin, A. Taguchi, S.M. Hanash, MCAM Mediates 727 Chemoresistance in Small-Cell Lung Cancer via the PI3K/AKT/SOX2 728 Signaling Pathway, Cancer Res, 77 (2017) 4414-4425.
729 [32] A.S. Nascimento, L.L. Peres, A.V.S. Fari, R. Milani, R.A. Silva, C.J. da 730 Costa Fernandes, M.P. Peppelenbosch, C.V. Ferreira-Halder, W.F. Zambuzzi, 731 Phosphoproteome profiling reveals critical role of JAK-STAT signaling in 732 maintaining chemoresistance in breast cancer, Oncotarget, 8 (2017)
733 114756-114768.
734 [33] C.C. Li, J.C. Yang, M.C. Lu, C.L. Lee, C.Y. Peng, W.Y. Hsu, Y.H. Dai, F.R.
735 Chang, D.Y. Zhang, W.J. Wu, Y.C. Wu, ATR-Chk1 signaling inhibition as a 736 therapeutic strategy to Dactinomycin enhance cisplatin chemosensitivity in urothelial bladder 737 cancer, Oncotarget, 7 (2016) 1947-1959.
738

Figure 1. ATM deficiency occurred in ESCC and p-ATR(Ser428) expression was significantly associated with the overall survival of patients underwent neoadjuvant chemotherapy.
(a). ATM-positive and ATM-negative case.
(b). p-ATR(Ser428) expression pattern by the intensity of positive tumor cells. 0, Negative; grade 1, weak; grade 2, moderate; grade 3, strong.
(c). The association between p-ATR expression and the survival of patients with ESCC underwent neoadjuvant chemotherapy. The Kaplan-Meier survival curve showed that the median survival time (MST) of patients with p-ATR positive expression was significantly shorter than that of patients with negative expression; in the further analysis, with the increase of positive expression cell proportion, the MST of patients was extending; when the p-ATR expression pattern was classified into high expression group and low expression group according to the proportion of positive cells, the MST of high group was significantly shorter than that of low group.

Figure 2. VE-822 inhibited cell growth, induced cell cycle arrest and cell apoptosis in ESCC cells.
(a). The basal level of ATM-CHK2 and ATR-CHK1 activation examined using Western Blotting in a panel of ESCC cell lines, including KYSE70, KYSE150, KYSE450, and KYSE510, which indicated that baseline ATM-CHK2 and ATR-CHK1 activation was essential for cell survival. The densitometry quantification was presented in Fig S2.
(b). KYSE70, KYSE150, KYSE450 and KYSE510 were seed in 96-well plate, and then treated for 48h with gradually increasing concentration of VE-822. The cell viability was assessed using CCK8 staining. Results was presented as the mean percentage of viable cells (Mean±SD), averaged from 3 independent experiments, each with 4 replicates per condition.
(c). The p-ATR and p-CHK1 expression were inhibited and DNA damage accumulation was induced by treatment with VE-822 in a dose-dependent manner as measured by Western Blotting.
(d). VE-822 treatment inhibited STAT3-AKT pathway activation in both KYSE70 and KYSE450.
(e). The effect of VE-822 on cell cycle progression was assessed by flow cytometry using PI/RNase staining. Exactly, VE-822 treatment for 24h significantly increased the proportion of cells in G1 phase, decreased proportion of G2 phase in KYSE70 and KYSE450.
(f). Several cell cycle and apoptosis markers were analyzed using Western Blotting. VE-822 treatment significantly augmented the DNA damage presented by γ-H2AX; VE-822 treatment down-regulated p21 expression for cells to pass the G1/S checkpoint and progress into G2 phase and up-regulated the expression of caspase-3.
(g). VE-822 affected the cancer cell apoptosis. To further investigate whether VE-822 affect cellular apoptosis, cells was stained by Annexin V/PE and underwent flow cytometry analysis after incubated with VE-822 for 8h. The apoptosis rate for VE-822-treated cells was significantly higher than control

Figure 3. VE-822 sensitizes ESCC cells to cisplatin and synergies strongly with cisplatin in ATM-deficient cells.
(a). CI value was calculated according to the Chou-Talalay median effect principle. We observed shift in cisplatin sensitivity in either KYSE450 or KYSE70 cell line. The CI value ≤ 0.9 stands for the synergistic effect.
(b). Western Blotting analysis confirmed that ATM expression was knocked out well both in KYSE450 and KYSE70.
(c). Cells were treated with select doses of VE-822 (as indicated) for 48 hours and viability was assessed using CCK8 assay. In the ATM-deficient cells, the IC50 of VE-822 was smaller than control cells, especially for relatively insensitive cell line KYSE70.
(d). There was more significant synergistic effect in ATM-deficient cells than control cells for both of KYSE450 and KYSE70 as presented by lower CI value.
(e). VE-822 potentiated cisplatin efficacy in ESCC xenografts, and the combination causes rapid regression. Nude mice bearing KYSE70 or KYSE450 or ATM-deficient KYSE70 or ATM-deficient KYSE450 were treated with 5mg/kg cisplatin on days1, 4, 7, 10 via intravenous injection or 60 mg/kg VE-822 on days 1, 3, 5 via oral gavage or combination of these two reagents. Tumor growth curves indicated that combination of VE-822 and cisplatin slowed down the tumor growth significantly compared to the single drug, especially for relative insensitive KYSE70.
(f). Although combination of two drugs, the mice did not suffer more weight loss than those in single drug group. In the ATM-deficient xenografts, the gap the weight loss curve between combination group and single group was narrower.
(g). The inhibition of p-CHK1, p-AKT activation were more obvious in ATM-deficiency cells and cleaved PARP was increased as measured by Western Blotting.

Figure 4. In the ATM-deficient cells, VE-822 alone or combination with cisplatin induced more severe DNA damage and cell apoptosis than control cell, as well as induced cell cycle arrest.
(a). Cells were treated with indicated concentration of VE-822, cisplatin, or combination of VE-822 and cisplatin for 24h and underwent flow cytometry analysis by staining with PI/RNase staining for assess the cell cycle progression. Cell cycle was arrested in phase S for ATM-deficient cells by VE-822 treatment.
(b). Cells were treated with indicated concentration of VE-822(2µM),
cisplatin(2µg/ml), or combination of VE-822(2µM) and cisplatin(2µg/ml) for 8h, and then underwent flow cytometry analysis by staining with Annexin V/PE. we observed dramatic cell apoptosis in ATM-deficient cells either with VE-822 or combination treatment.
(c). Expression of several markers which reflects the cell proliferation or cell apoptosis or cell cycle progression or DNA damage was examined using western blotting. Caspase-3 and cleaved-PARP expression remarkably increased in cells exposure to combination of VE-822 and cisplatin. In the ATM-deficient cell, VE-822 combination with cisplatin more dramatic changes in the expression of p-AKT, p21, and γ-H2AX than control cells.

Figure 5. Combination treatment enhanced the DNA damage effect presented by examining fluorescence intensity of γ-H2A.X and p53 in ESCC cells.
Cells treated with cisplatin, VE-822, or combination of two drugs for 24 h were fixed and co-labeled with anti-γH2AX and anti-p53 antibodies. The γ-H2A.X and p53 foci were analyzed by immunofluorescence microscopy. combination treatment resulted in accumulation of DNA damage presented by stronger and more fluorescence staining of γ-H2AX and p53 (a,b). The enhancement of the DNA damage effect was stronger in ATM-deficiency cells (c,d).

Figure 6. Combination of VE-822 and cisplatin increased cisplatin-DNA accumulation. (a) Cells were treated with cisplatin, VE-822, or combination of two drugs for 24 h. Anti-cisplatin modified DNA antibody was used to specifically bind to Cisplatin-DNA，and the positive signal was examined by using flow cytometry. The combination of VE-822 and cisplatin induced augmentation of cisplatin-DNA adduct.
(b) The percentage of cisplatin modified DNA positive cells were measured by flow cytometry and are represented as mean ± SD of three replications.

Table 1. Association between ATR-pSer428 expression and Median survival time(months)/5-y survival rate in the study cohort (n=110)

Item No. (%)

Negative VS Positive
ATR-pSer428
non-expression ATR-pSer428
expression
Expression Grades
ATR-pSer428
non-expression ATR-pSer428
expression
Grade1 ATR-pSer428
expression
Grade2 ATR-pSer428
expression
Grade3 Low vs. High ATR-pSer428
low expression
ATR-pSer428
high expression
Median survival time,□months (95%CI)

Highlights:
 ATR activation was associated with efficiency of neoadjuvant chemotherapy.
 VE-822 promoted cell apoptosis and induced cell cycle arrest in ESCC cells.
 VE-822 sensitized ESCC cells to cisplatin, in vitro and in vivo.
 VE-822 along with cisplatin induces DNA-cisplatin adduct accumulation.