Oxaliplatin Compromised CDK1 Activity Sensitizes BRCA-Procient Cancers to PARP Inhibition in Oxaliplatin Resistance Gastric Cancer

Background: Oxaliplatin resistance is one of the most important problems in the treatment of cancer. The successful culture of tumor organoid in gastric cancer can help us to study oxaliplatin resistance and its mechanism. Thus, it is convenient for us to successfully solve oxaliplatin resistance and improve the prognosis of patients. Methods: Two oxaliplatin resistant patients and two oxaliplatin sensitive patients were enrolled through our Gastric Cancer Center of Sun Yat-sen University. Core genes of oxaliplatin resistant and non-resistant patients were analyzed by sequencing. The overexpression and knockdown of core genes were carried out by organoid in vivo, combined with oxaliplatin-resistant cell lines AGS, MKN74 and SNU719 for cell viability, WB and immunouorescence, etc., to verify the role of core genes in oxaliplatin resistance. Again, in vivo experiments were veried by subcutaneous tumor formation in vitro. Results: Through sequencing, we found that PARP1 is an important core gene leading to oxaliplatin resistance. In vivo organoids, oxaliplatin resistant cell lines and subcutaneous tumor formation in vivo. We found that PARP1 was an important cause of oxaliplatin resistance. Oxaliplatin can inhibit CDK1 activity and make cancer with normal BRCA1 function sensitive to PARP inhibition. Through the combination of oxaliplatin and PARP1 inhibitor olaparib, we can effectively kill tumor cells. Through the patients' follow-up data, we found that the expression level of PARP1 was signicantly correlated with oxaliplatin resistance. Conclusion: Our results indicate that PARP1 is an important core gene leading to oxaliplatin resistance. Combined oxaliplatin and PARP1 inhibitor olaparib can effectively kill tumor cells. Oxaliplatin can inhibit CDK1 activity and make cancers with normal BRCA1 function more sensitive to PARP inhibitors. (radiological evaluation), and evaluated in accordance with the response evaluation criteria in the solid tumor (RECIST) guidelines (7). In the validation phase, patients with worsening symptoms, new lesions, or radiologically assessed tumor regeneration ≥ 25% were assigned to the progressive disease (PD) group (n = 45) and the remaining non-PD group (n = 55). PFS is dened as the duration from tumor resection to PD. Follow-up was performed every 3 months (for the initial 0-2 years), 6 months (subsequent 2-4 years), and once a year until death or February 2020. The follow-up study included abdominal computed tomography and postoperative physical examination. chemotherapy and the patients with good chemotherapy response. We then veried that PARP1 have strong characteristics and Oxaliplatin resistance through in vivo organoids and Oxaliplatin resistant cell lines and in vivo BALB/C NUDE mice tumorigenesis. Then we found that PARP1 is the core gene of their common differences by looking for the difference genes between Oxaliplatin resistance patients and non-resistant patients. In order to explore the relationship between PARP1 and Oxaliplatin resistance, we inhibited PARP1 to signicantly enhance the ability of Oxaliplatin to kill cancer and Oxaliplatin resistance cells. The combined use of PARP1 inhibitors can signicantly inhibit activity of GC organoids, which affects their tumor initiation ability. In vivo experiments have also shown that inhibiting PARP1 can signicantly overcome the resistance to Oxaliplatin. Subsequently, we found that PARP1 mediates the DNA repair ability of Oxaliplatin resistance cells by regulating the DNA repair pathway BER, and after the combination of PARP1 inhibitor, Olaparib, the joint effect allowed the drugs to effectively cause homologous recombination failure through CDK1 and BRCA1 functions, eventually leading to tumor cell apoptosis. Gastric cancer resistance 1; GCR2: Gastric cancer resistance 2; ATP: Adenosine triphosphate; IACUC: Institutional Animal Care and Use Committee; OXA: Oxaliplatin; OLP: Olaparib; CON: control group; AGSR: AGS Oxaliplatin resistance; SNU719R: SNU719 Oxaliplatin resistance; MKN74R: MKN74 Oxaliplatin resistance; CISP: cisplatin; PCDK1: CDK1 phosphorylation antibody; PBRCA1: BRCA1 phosphorylation antibody; TOM: Topological overlap measure; GS: Gene signicance; MS: Module signicance; MCODE: Molecular Complex Detection; DAVID: Database for Annotation, Visualization and Integrated Discovery; GO: Gene ontology; KEGG: Kyoto Encyclopedia of Genes and Genomes.

for 5 minutes, and then stopped with a large quantity of PBS. The suspension was ltered through 40 nylon meshes, centrifuged, and the cells were xed in the medium. It was passaged with TrypLE every 2 weeks. The medium for establishing and culturing human GC organoids was as described in the literature (5).
lentivirus production and infection of organoids Control and shRNA_PARP1-expressing pLKO vectors were purchased from Sigma. PARP1 overexpression vectors were designed to generate the lentivirus by Shanghai Genechem Co., Ltd. All lentiviral particles were produced in HEK293T cells by standard procedures, concentrated by ultracentrifugation at 100,000g for 2h and resuspended in sterile PBS. Organoids were extracted from Matrigel using TrypLE Express (Thermo Fisher), resuspended in OptiMEM with 10 µgml−1Polybrene, and then mixed with the virus solution in an incubator for 6h. Cells were plated back into Matrigel and split 3 to 7 days later when antibiotic selection was started.

Quantitative real-time PCR
When the number of cells were few, MagMAX-96 Total RNA Isolation Kit (Ambion) or RNeasy Mini Kit (Qiagen) was used to extract RNA according to the manufacturer's instructions. Random hexamer primers (Invitrogen) was used to SuperScript III First-Strand cDNA synthesis kit or use iScript cDNA synthesis kit (BioRad) according to the manufacturer's instructions to generate cDNA. 5 times dilution was performed on the cDNA with distilled water, diluted 2 mol/L to use each RT-qPCR reaction, and the measurement was performed on the Express SYBR GreenER (ThermoFisher) ABI7500 (Applied Biosystems). The primers were designed using the Universal Probabilistic Analysis and Design Center (Roche) to ensure that they span the exon-exon junction. The family gene transcription level (actin) was used for normalization. The RT-qPCR primers were listed in Supplementary Table 2.

Flow cytometry
Annexin V-PI apoptosis assay were performed using Annexin V-FITC Apoptosis Detection Kit (Sigma-Aldrich) according to the manufacturer's protocol. FlowJo 10 software was used to analyze the data.
In a 96-well transparent bottom blackboard, 3000 cells were planted in each well (organoids were planted in Matrigel). The drug was then added to each well according to a 10-fold concentration gradient.

Immuno uorescent staining
Immuno uorescence staining of organoids and cell lines: the body was placed in a glass bottom tissue culture plate (ibidi, lot:191218/2), xed with 5% NBF for 10 minutes, and blocked with PBS containing 10% FCS, 1% BSA (Sigma) and 0.2% Triton-X. The primary antibody was incubated in blocking buffer at 4°C for 16h. The uorescent secondary antibody was incubated with 3 DAPI in blocking buffer at 20°C for 1-6h. Fluorescence staining was imaged on a Zeiss LSM 780 confocal microscope. Tissues were prepared as above and the same antigen retrieval procedure was applied. Secondary antibodies were uorophoreconjugated and incubated with 3µM DAPI in the dark. Before mounting, slides were incubated in 0.1% (w/v) Sudan black B (Sigma) in 70% ethanol to reduce background signal. For the other 6 mice, PBS were injected intraperitoneally. The cancer-bearing BALB/C NUDE mice were sacri ced 4 weeks later, and tumors were harvested for measuring and weighing. In order to study the drug resistance of PARP1 expression, we inoculated 100,000 cells of plko and PARP1-sh1 (GCR1 and GCR2) into BALB/C NUDE mice with Matrigel (BD, 354230). After 25 days, 6 mice with organoids transplantation tumors received a treatment of Oxaliplatin (Selleckchem, s1224) at a dose of 5 mg/kg twice a week for 4 weeks. The cancer-bearing BALB/C NUDE mice were sacri ced 4 weeks later, and tumors were harvested for measuring and weighing. we inoculated 200,000 cells of control and PARP1 overexpression (GC1 and GCR2) into BALB/C NUDE mice with Matrigel (BD, 354230). After 25 days, 6 mice with organoids transplantation tumors received a treatment of Oxaliplatin (Selleckchem, s1224) at a dose of 5 mg/kg twice a week for 4 weeks. The cancer-bearing BALB/C NUDE mice were sacri ced 4 weeks later, and tumors were harvested for measuring and weighing.
The organoids of GCR1 and GCR2 were digested into single cells by TrypLE (glibco, 12604-013) and then counted, and 100,000 cells were placed on Matrigel (BD, 354230) and inoculated subcutaneously into BALB/C NUDE mice (6 per group). After 25 days, the organoids transplated BALB/C NUDE mice received intraperitoneal injection of either Oxaliplatin (Selleckchem, s1224) + Olaparib (Selleckchem, AZD2281, s1060), Oxaliplatin, Olaparib, or PBS. Oxaliplatin dose was 5 mg/kg, Olaparib dose was 50 mg/kg, combined group dose was 5 mg/kg of Oxaliplatin and 25 mg/kg of Olaparib twice per week, each treatment lasting for 4 weeks. The tumor size and body mass were measured every three days. The mice were sacri ced one month later, and tumor tissues were prepared for histological examination. Tumor volume (mm 3 ) = 0.5 × width 2 × length. All animal experiments were carried out in accordance with health guidelines, and the protocol was set by the Sun Yat-sen University Animal Protection and Use Committee. When the mice reached the end point, the tumor was photographed and the tumor was weighed.

RNA isolation and microarray
Total RNA was extracted from tissue samples, and Nanodrop 2000 was used to detect the concentration and purity of the RNA. Agarose gel electrophoresis was used to detect RNA integrity, and Agilent 2100 was used to determine the RIN value. A single library construction required that the total amount of RNA was no less than 5μg, the concentration ≥ 200ng/μL, and the OD260/280 between 1.8 and 2.2. The mRNA capture and library preparation were completed by the advanced sequencing equipment of Shanghai Origin-gene Biomedical Technology Co., Ltd. using KAPA mRNA HyperPrep kit (Roche). The biological triplicate libraries were sequenced on the Illumina Truseq TM RNA sample prep Kit platform of the facility, and each sample produced an average of 25 million single-ended reads of 75 bp. Align the post quality control, high-quality sequence with the designated reference genome. The PDOX sample was rst compared with mice reference genome. After removing mice-related data, it was then compared with human reference genome. The human reference genome was obtained from Ensembl database, genome version GRCh38, gene annotation information was Ensemble 92. Before alignment, cutadapt (version 1.9.1) is used for quality control and adaptor trimming of the original reading. Use the annotation release 86 to sequence the Reads of the human genome GRCh38 using RSEM 1.3.0 and STAR 2.5.2, and count the subsequent gene levels. In version 3.6.1 of R package, the DESeq2 package (version 1.24.0) was used for normalization and differential expression analysis of raw count data. Regularized logarithmic transformation was performed on the rlog function.
Clinical GC patient samples  Table 1. The tumor response to chemotherapy was evaluated by the three-dimensional volume reduction rate or tumor response rate (radiological evaluation), and evaluated in accordance with the response evaluation criteria in the solid tumor (RECIST) guidelines (7). In the validation phase, patients with worsening symptoms, new lesions, or radiologically assessed tumor regeneration ≥25% were assigned to the progressive disease (PD) group (n = 45) and the remaining non-PD group (n = 55). PFS is de ned as the duration from tumor resection to PD. Follow-up was performed every 3 months (for the initial 0-2 years), 6 months (subsequent 2-4 years), and once a year until death or February 2020. The follow-up study included abdominal computed tomography and postoperative physical examination.

Table1
Demographics of GC patients of SYSU.

Patient information in public databases
The transcriptome data of patients with gastric adenocarcinoma con rmed by pathology was datasets of Oxaliplatin resistance patients, normalized RNA read counts were used for analysis, and the following settings were applied: permutation number = 1000, permutation type = gene set, enrichment statistics = weighting, a measure of gene ranking = signal noise. For the TCGA gastric cancer dataset, the samples were grouped according to their expression above or below the median value. The normalized RSEM read count was used for analysis, and the following settings were applied: number of permutations = 1000, permutation type = phenotype, enrichment statistics = weighting, measurement of gene ranking = signal 2 noise. Recognized marker gene set 40, KEGG pathway or gene ontology (GO) terms, and false discovery rate (FDR q) <0.05 were considered signi cant enrichment.

Screening of differentially expressed genes (DEGs)
The expectation-maximization method RNA-Seq was used to normalize the 3-level transcriptome data of the data set, and the logarithmic transformation of all gene expression values was performed.
Approximate data were normally distributed after normalization by quantiles (8). In this study, the R package limma program v3.28.14 was used to analyze the differential genes of gene expression data, and its mRNA satis ed P<0.01, false discovery rate (FDR)<0.01 and |log2 fold change (FC)|>1.5, where P <0.05 indicated that the hypothesis test was statistically signi cant. FDR is a control indicator for the error rate of the hypothesis test. As an evaluation index of the selected differential genes, the number of false rejections was proportional to the number of rejected invalid hypotheses. FC was usually used to describe the degree of change from the initial value to the nal value. In this study, the ratio of tumor tissue gene expression value to normal tissue gene expression value was used, also known as the fold change. The heatmap and volcano map of the differential genes were constructed in R language for visual comparison.

WGCNA Co-expression Network Construction
Gene expression data (mRNA-seq data) was downloaded from the TCGA database. A total of 24,991 genes were identi ed in each sample. Analysis of variance was performed and then sorted from largest to smallest. The SD value of each gene was calculated and sorted from largest to smallest, and then the top 5000 genes were selected for WGCNA. WGCNA package in R software was used to construct a gene coexpression network from the expression data map of these 5000 genes (9). Using the adjacency function in WGCNA, an adjacency matrix was constructed by calculating the Pearson correlation between all pairs of genes in the selected sample. In this study, β = 7 (scale-free R2 = 0.9) was used as the soft threshold parameter to ensure a scale-free network. In order to further identify the functional modules in the coexpression network of these 5000 genes, the adjacency matrix was used to calculate the Topological Overlap Measure (TOM), which represents the overlap in the shared neighborhood. We identify related modules by calculating the correlation between MEs and PARP1 expression levels. Then the log10 transformation of the p value (GS=lgP) in the linear regression of gene expression and clinical PARP1 expression level information was de ned as gene signi cance (GS). In addition, module signi cance (MS) is de ned as the average GS of all genes in a module. In general, among all the selected modules, the module with the highest absolute value of MS was considered to be the module related to the level of PARP1 expression.

PPI network construction of key module gene
The Hub gene, which is highly interconnected with the nodes in the module, is considered to have important functions. We selected the top 30 Hub genes in the module network as candidate genes for further analysis and veri cation. The STRING data set is an online biological resource that can decode the interaction between proteins and proteins to obtain the actual precise functions of proteins (10). The candidate gene was submitted to the protein interaction of STRING, and the binding con dence interval of the cutoff value was set to 0.4. In the plugin, Molecular Complex Detection (MCODE), the signi cant models with strong protein-protein connection were calculated and selected with the default parameters (degree cut ≥ 2, node score cut ≥ 2, K-core ≥ 2, maximum depth = 100). P<0.05 was considered statistically signi cant.

Statistical analysis
The images and graphs shown represent several experiments repeated on different individuals at different times. Each experiment was repeated independently. All statistics were performed using SPSS and R software. The statistical test was explained in the gure legend. All results were statistically different based on the mean ± SD, P <0.05.

PARP1 is an Important Core Gene Leading to Oxaliplatin Resistance
In order to nd the cause of chemotherapy resistance, we used four PDOs (GC1, GC2, GCR1, GCR2), of which GCR1 and GCR2 were from patients whose GC recurred after postoperative chemotherapy, while GC1 and GC2 were patients with without recurrence after postoperative chemotherapy. In a viability assay, GCR1 (IC50 = 19.95um/L) and GCR2 (IC50 = 63.09um/L) were found to be more resistant to Oxaliplatin than GC1 (IC50 = 0.93um/L) and PT4 (IC50 = 3.03um/L) (see Figure S1A). In order to explore the differences between oxaliplatin resistant and non-resistant patients in organoids and the regulation of core genes. We performed mRNA sequencing on organoids of oxaliplatin-resistant and non-resistant patients. Figure 1A shows their signi cantly different genes. Compared with non-drug-resistant patients, the main enrichment pathways for drug-resistant patients include Homologous Recombination, DNA Replication, Base Excision Repair, Cell Cycle (Fig. 1B). Finally, we searched for the core gene through String and found that PARP1 was the core gene affecting drug resistance (Fig. 1C, D).

PARP1 is Upregulated in Gastric Cancer Oxaliplatin Resistance Organoid
Through our experiments on oxaliplatin resistance of GCR1, GCR2, GC1 and GC2 in vitro and in vivo, it was found that the tolerance of GCR1 and GCR2 to oxaliplatin was signi cantly higher than that of GC1 and GC2 ( Figure S1D, Fig. 2A-F). Moreover, it was found that the expression level of PARP1 in GCR1 and GCR2 of oxaliplatin-resistant organoid was signi cantly higher than that in GC1 and GC2 of oxaliplatinsensitive organoid. To illustrate the important role of PARP1 in oxaliplatin resistance.

PARP1 plays an important role in maintaining oxaliplatin resistance
Our previous study found that PARP1 expression was signi cantly elevated in oxaliplatin resistance. To study the role of PARP1 in oxaliplatin resistance. In vitro, through PARP1 overexpression in oxaliplatinsensitive organoid, it was found that oxaliplatin-sensitive organoid showed signi cantly increased tolerance to oxaliplatin, and after PARP1 knockdown in oxaliplatin-resistant organoid, it was found that oxaliplatin-resistant organoid showed decreased tolerance to oxaliplatin (Fig. 3A-C, Figure S3A). In vivo, we also further veri ed that the subcutaneous tumor-forming experiment was conducted after the overexpression of PARP1 in oxaliplatin-sensitive organoid, and it was found that the tolerance of subcutaneous tumors to oxaliplatin was signi cantly increased after intraperitoneal injection of oxaliplatin. Moreover, PARP1 knockdown was performed on oxaliplatin resistant organoid, and it was found that the tolerance of tumors to oxaliplatin in vivo was reduced (Fig. 3D-F). These results indicate that PARP1 plays a very important role in oxaliplatin resistance in vitro and in vivo.

PARP1 inhibition by Olaparib sensitizes gastric cancer to Oxaliplatin
Since PARP1 might be an important gene for Oxaliplatin resistance we had veri ed. To verify whether PARP1 inhibitor combined with oxaliplatin can effectively inhibit oxaliplatin resistance. We will carry out further in vitro and in vivo experiments. First of all, by using both the PARP1 inhibitor, Olaparib, and Oxaliplatin, the drug combination was found to effectively inhibit the viability, size, cell count, and proliferation of the organoids of oxaliplatin resistance gastric cancer (ORGC, GCR1 and GCR2) ( Fig. 4A-C). The drug combination could signi cantly inhibit the activity and proliferation of Oxaliplatin resistance gastric cancer cell lines ( Figure S2A, B). BALB/C NUDE mice in vivo tumorigenesis experiments have also con rmed that these drugs when used together could effectively inhibit tumor growth when compared with their use individually (Fig. 4D-F), and can induce cell apoptosis and affect proliferation of tumor cells ( Fig. 4G-L). By comparing Olaparib + Oxaliplatin versus Oxaliplatin, it was found that the Olaparib + Oxaliplatin group was mainly enriched in oxidative phosphorylation and PPAR signaling pathway (see Figure S2C, D). These two pathways are primarily important enrichment pathways for tumor apoptosis after chemotherapy-induced DNA damage (11,12). The main pathways enriched in the Oxaliplatin group were JAK-STAT, MAPK, NOTCH and WNT signaling pathways (see Figure S2E-H). In fact, these pathways are not only related to drug resistance in tumor, but also closely related to proliferation. PARP1 is an important factor affecting oxaliplatin resistance, and inhibition of PARP1 by olaparib can signi cantly change oxaliplatin tolerance. Thus for the clinical combination of drug use to bring convenience.

Combined Oxaliplatin with Olaparib inhibits BER and HR repair pathways via blocking both CDK1-BRCA1 and PARP1related activities
Through the study above, the increase in PARP1 expression was found to be an important mediating factor for Oxaliplatin resistance. Now we will explore the role of PARP1 in Oxaliplatin resistance. PARP1 is usually used to repair single base breaks in DNA. Single base break is a type of commonly occurring DNA damage, and unrepaired single base breaks are not harmful to cells. However, when these broken bases are transcribed or replicated, they will destroy and cause damage to the new DNA copies. The activation of PARP1 can promote DNA base excision repair (BER) and inhibit the binding of transcription factors to single-stranded DNA, thus inhibiting the transcription of damaged DNA and DNA repair (13). Meanwhile, the main role of Oxaliplatin is to disrupt the DNA synthesis of cells, thereby affecting the cell cycle and leading to DNA damage. The sensitivity and resistance of cells to platinum-based chemotherapy are largely determined by the activity of the DNA damage response (14). PARP1 is highly likely to mediate Oxaliplatin resistance through its regulation of DNA repair mechanisms. First, the effect of Oxaliplatin on DNA damage in Oxaliplatin resistant cells and sensitive cells was studied and results showed that the resistant cells were able to effectively repair DNA after the damage (Fig. 5A). Next, the effect of PARP1activity on the repair of DNA damage caused by Oxaliplatin was studied and PARP1 inhibition was found to signi cantly inhibit the repair process (Fig. 5B, C). BER is an important pathway used in platinum-based drug. The role of BER in cancer drug resistance had been proposed by many studies (15)(16)(17)(18), and PARP1 plays an important role in the BER pathway (19). To this end, the effect of PARP1 inhibitor combined with Oxaliplatin on the BER pathway marker, XRCC1 was studied, and Olaparib + Oxaliplatin was found to signi cantly inhibit the BER pathway when compared to Oxaliplatin alone (Fig. 5A, D, E). However, the transcription levels of XRCC1 in Oxaliplatin resistance patients and nonresistant patients, and XRCC1 of Oxaliplatin resistance and non-resistant strains did not change signi cantly (Figure S3 B, C). It showed that PARP1 can cause Oxaliplatin resistance by affecting the BER signaling pathway. Oxaliplatin will cause single-stranded damage after acting on tumor cells. At this point, if PARP1 had su cient function, it will lead to BER pathway activation and thus DNA repair, which will eventually lead to drug resistance in tumor cells (20).
The role of PARP1 is to bind to DNA damage sites (mostly single-stranded DNA breaks) and catalyze the synthesis of poly ADP ribose chains on protein substrates (21). In order to study the core target of PARP1 interaction, weighted gene co-expression network analysis (WGCNA) method was used to nd the core gene that interacted with PARP1 and it was found that CDK1 played a very key role in the high expression of PARP1 ( Figure S4, Table S1). Cyclin-dependent kinase (CDK) 1 is the core component of the cell cycle mechanism, forming a complex with cyclin A and B to promote the progression of S phase, G2 phase and M phase. Recently, CDK1 and its other family members had been shown to be involved in the upstream of the DNA damage response pathway (22). Studies had found that CDK1 can inhibit homologous recombination by inhibiting the phosphorylation of BRCA1 (23,24). Thus, we veri ed whether Oxaliplatin can directly act on BRCA1 or CDK1 to inhibit BRCA1 and cause homologous recombination failure.
To do this, we investigated the interaction of Oxaliplatin with BRCA1 and CDK1. First, Olaparib + Oxaliplatin drug combination was compared with single drug Oxaliplatin. Oxaliplatin was seen to signi cantly inhibit the phosphorylation of BRCA1 and CDK1 (Fig. 6A, B), but Olaparib had no signi cant effect (Fig. 6A, 7A-C, 8A-F). In addition to affecting the functions of BRCA1 and CDK1, Oxaliplatin can also decrease the expression level of RAD51 (Fig. 6A, Fig. 7D, E). RAD51 is an important marker in homologous recombination. Oxaliplatin may be able to inhibit homologous recombination by affecting the function of BRCA1, which in turn leads to a decrease in RAD51 and ultimately aggravating DNA damage (such as increased expression of H2AX). But whether Oxaliplatin indirectly inhibited BRCA1 function by inhibiting CDK1 or directly inhibiting BRCA1 function remains unclear. So their relationship was compared by inhibiting the effect of CDK1. Figure 6A showed that CDK1 inhibitors signi cantly inhibited the phosphorylation of BRCA1, and the effect was similar to that of Oxaliplatin. In order to examine whether Oxaliplatin can bypass CDK1 and directly inhibit BRCA1, the functional effects of cisplatin, which is also a platinum-based drug, was used on CDK1 and BRCA1 and compared to that of Oxaliplatin and found that cisplatin cannot inhibit the functions of CDK1 and BRCA1 (Fig. 6A). Moreover, it was shown through proliferation and colony formation assay that the effect of cisplatin combined with PARP1 and CDK1 inhibitors was not signi cantly different from the effect of Oxaliplatin combined with PARP1 inhibition (Fig. 6C, D). It showed that CDK1 plays an important role in killing tumor cells in platinum-based chemotherapy drugs. In fact, although the principle of action of Cisplatin and Oxaliplatin is basically the same, but the effect of cisplatin is worse than that of Oxaliplatin (25). CDK1 may be the main reason.
PARP1 expression predicts the relapse of human gastric cancer after surgery It showed that PARP1 play an important role in Oxaliplatin resistance in gastric cancer. In order to clinically verify the importance of PARP1 in the recurrence of gastric cancer after curative surgery and adjuvant chemotherapy, we enrolled gastric cancer patients undergoing adjuvant chemotherapy in Sun Yat-sen University's Gastric Cancer Research Center. Through immunohistochemistry and recurrence status of patients after adjuvant chemotherapy, we found that PARP1 were highly expressed in the specimens of patients who relapsed after adjuvant chemotherapy ( Figure S5 A, B, C). Moreover, the recurrence time of patients with high expression of PARP1 was signi cantly shorter than that of patients with low expression (Figure S5D, E). It showed that PARP1 can be used as important indicators to clinically predict recurrence in postoperative adjuvant chemotherapy patients, and it also showed that PARP1 play an important role in chemotherapy resistance.

Discussion
While trying to de ne and understand the biological characteristics of Oxaliplatin resistance cell subsets in GC, we found that PARP1 was signi cantly increased in relapsed patient after postoperative chemotherapy through the use of the sequencing results of patients who relapsed after Oxaliplatin chemotherapy and the patients with good chemotherapy response. We then veri ed that PARP1 have strong characteristics and Oxaliplatin resistance through in vivo organoids and Oxaliplatin resistant cell lines and in vivo BALB/C NUDE mice tumorigenesis. Then we found that PARP1 is the core gene of their common differences by looking for the difference genes between Oxaliplatin resistance patients and nonresistant patients. In order to explore the relationship between PARP1 and Oxaliplatin resistance, we inhibited PARP1 to signi cantly enhance the ability of Oxaliplatin to kill cancer and Oxaliplatin resistance cells.    Representative images of comparison of IHC staining of CDK1 and its phosphorylation and BRCA1 and its phosphorylation in BALB/C NUDE mice after tumorigenesis under the effects of Olaparib+Oxaliplatin,