Ailanthone Enhances Gastric Cancer Cell Apoptosis Through Inducing DNA Damage and Repressing Base Excision Repair Via Downregulating P23

Background: Chemotherapy plays an irreplaceable role in the treatment of gastric cancer (GC), but currently available chemotherapeutic drugs are not ideal. Medicinal plants are an important direction for new drug discovery. Methods: We combined literature collection, organoid drug screening and PDX verication to search for new, effective drugs. The effect of AIL on the GC cells was investigated by CCK-8 assays, Colony Formation Assay and Cell Viability, Cell migration and invasion assay and PDX mice model. The expression of indicated RNA and proteins were measured by qRT-PCR ,IHC and western blotting. The proteins HSP90 binding to P23 or XRCC1 were identied by Co-IP. Results: Through drug screening of GC organoids, we had determined that ailanthone (AIL) has an anticancer effect on GC cells in vitro and in vivo. We also found that AIL promoted apoptosis of GC cells through ow cytometry analysis and the expression of caspase 3 and H2AX were signicantly enhanced by AIL in GC cells and tumor tissues, suggesting that AIL can induce DNA damage and apoptosis of GC cells. Further transcriptome sequencing of the PDX tissue indicated that AIL inhibited the expression of XRCC1 which plays an important role in DNA damage repair, and the results were also conrmed by western blotting. In addition, we found that AIL inhibited the expression of P23 and inhibition of P23 decreased the expression of XRCC1 through western blotting analysis, indicating that AIL can regulate XRCC1 via P23. The results of co-immunoprecipitation showed that AIL can inhibit the binding of P23 and XRCC1 to HSP90 respectively. Conclusion: These ndings indicate that AIL can induce DNA damage and apoptosis of GC cells. Meanwhile, AIL downregulating P23 DNA damage The present study shed light to We reviewed the of patients undergoing gastric cancer resection from May 2012. The inclusion criteria were pathological TNM stages I–III gastric adenocarcinoma patients whom underwent radical surgery patients. Exclusion criteria: other history of malignant tumors, neoadjuvant therapy, loss to follow-up, no tissue samples. A total of 93 cases of tissue samples were collected from surgical and informed the This study patients study The tested by IHC, diagnosis and of used a semi-quantitative for scoring. Among the samples, more than 10% of tumor cells staining were positive. The staining intensity was divided into negative, weak, moderate and strong. Low P23 expression was detected in the negative and low intensity samples, and medium and strong intensity returned with high P23 expression. Subsequently, the disease-free survival of patients in the high P23 expression group (n = 42) and low P23 expression group (n = 51) were evaluated. Progression free survival is dened as the duration from tumor resection to progressive disease. Follow up was completed in intervals of 3 months (0–2 years), 6 months (2–4 years), and once a year until recurrence or June 2020. The follow-up study included abdominal computed tomography and postoperative physical examination.


Background
Gastric cancer (GC) is the 5th most common cancer in the world, with a 5-year survival rate of 29%, it is 3rd in cancer-related deaths [1] .Currently, the most effective treatment is known to be surgical intervention.
However, due to the fact that most GC patients were diagnosed in the late stage, they have already lost the window of opportunity for surgery. For advanced cancer, chemotherapy and perioperative chemotherapy or adjuvant chemoradiotherapy is extremely signi cant in improving their survival [2] . Still, chemotherapy drugs have toxic side effects, and currently, researches are showing that not all patients are sensitive to the traditional chemotherapeutic drugs [3] . Thus, developing a new, effective yet low toxicity drug is of utmost importance to these patients.
Medicinal plants have a long history of use in cancer treatment. Countries such as China and Japan have had thousands of years of history in cancer treatment through traditional medicine [4,5] . Today, many drugs in the clinic settings are in fact derived from such medicinal plants. For example, vincristine, paclitaxel and docetaxel are common place in modern clinical practice [6,7] . Therefore, the identi cation of natural compounds from medicinal plants and the development of new anticancer drugs have been gaining increasing traction over the years. The process of discovery of new drugs is inseparable from effective drug screening tools, thus we here as the largest GC diagnosis and treatment center in Southern China have established a large GC specimen banks and have successfully established multiple GC organoids and Patient-Derived Xenograft(PDX) models over the years.By using the advantages above, we combined literature collection, organoid drug screening and PDX veri cation to search for new, effective drugs. Fortunately,AIL turned out to be an effective compound in suppressing GC organoid growth. AIL is a medicinal plant extracted from Ailanthus altissima. It has been used in traditional Chinese medicine (TCM) to treat various diseases since ancient times. In recent years, it has been con rmed by modern scienti c methods to have signi cant anticancer activity [8][9][10][11][12][13][14][15] .In the GC PDX model, AIL was veri ed to be a better suppressor of tumor growth with a lower cell toxicity. Further analysis revealed that AIL can effectively induce DNA damage in GC cells and downregulate P23 which suppresses DNA damage repair in the base excision repair (BER) pathway through regulating XRCC1.
AIL had been proven to treat castration-resistant prostate cancer by down-regulating P23.The results of our bio-information analysis showed that the high expression of P23 enhanced the DNA repair pathways, mainly the BER pathway, when compared to low expression of P23. P23 is known as an important cochaperone of HSP90 [16] . HSP90 and its associated co-chaperones play key nucleic functions including chromatin remodeling, DNA transcription, RNA processing, DNA replication, telomere maintenance, and DNA repair [17,18] . In fact, HSP90 is involved in DNA repair and mainly depends on its co-chaperone P23, or phosphorylation to interact with the DNA metabolism proteins [17][18][19] .
In this study, we found that AIL can signi cantly inhibit tumor growth in GC cells, organoids and PDX models both in vivo and in vitro. AIL mainly inhibited DNA repair of XRCC1 by inducing DNA damage and down-regulating P23, thereby exerting anticancer effect on GC. This nding indicated that AIL could effectively kill tumors through DNA damage and inhibition of DNA repair.

Methods And Materials
Cell Culture Human gastric cancer cell lines SGC7901, SNU719, AGS were provided by Nanjing Kegen Biotechnology Co., Ltd. The cells were cultured in RPMI-1640 medium containing 10% fetal bovine serum and 1% penicillin and streptomycin, placed in a humidi ed incubator containing 5% CO 2 and 95% air, subculture and all experiments were performed at 37 °C. AIL stock solution was prepared with DMSO, stored at 4 °C, and diluted with RPMI-1640 medium to the required concentration immediately before use; the nal concentration of DMSO in the medium was 0.1%. Cells in the control group were treated with DMSO (0.1%) without AIL.

Human tissues and organoids
Human gastric cancer tissues were taken from patients who underwent gastric cancer surgery in the First A liated Hospital of Sun Yat-sen University. They agreed and signed a donation and research consent form. It was approved by the clinical scienti c research and animal experiment ethics committee of the After the tumor was excised, it was placed in 50 U/ml penicillin-streptomycin (Thermo Fisher) frozen G solution. The tissue was minced on ice and incubated in DMEM containing 1 mg/ml collagenase V (Sigma-Aldrich) for 1 hour at 37 °C. Iced PBS was added to stop the digestion, and subsequently centrifuged at 4 °C (300 rcf, 5 min). The sample was further digested with TrypLE (Thermo Fisher) at 37 °C for 5 min, and then stopped with a large amount of PBS. The suspension was ltered through 40 nylon meshes, centrifuged, and the cells were xed in the matrix. It was then passaged with TrypLE every 2 weeks. The medium for establishing and culturing human GC organoids was as described in the literature [20] .
Colony Formation Assay and Cell Viability AGS, SNU719, and SGC7901 cells were seeded in 6-well plates with 500 cells per well, and the experiment was repeated three times. AIL was added to the cells at different nal concentrations (0, 0.05, 0.1 µM) and it was then incubated for 12 days. The cells were xed for 30 minutes with 4% paraformaldehyde, stained with crystal violet for 30 minutes, and nally counted.
In a 96-well transparent bottom black board, 3,000 cells were seeded into each well (organoids were seeded in Matrigel). Medications were added in each well according to a 5-fold or 10-fold concentration gradient. After 72 hours, a luminometer (PerkinElmer Life and Analytical Sciences, Boston, MA) was used to determine the level of adenosine triphosphate (ATP) by CellTiter-Glo Luminescent Cell Viability Assay (Promega, Madison, WI).

Cell migration and invasion assay
Transwell migration assay and matrix invasion assay were used to determine cell migration and invasion ability. In the Transwell migration assay, a small chamber (8 µm pore size; Corning) was placed in a 24well plate, and then 5 × 10 4 cells were suspended in 200 ul serum-free RPMI-1640 medium in the chamber; 20% pre-warmed fetal bovine serum was used in the well.The cells were incubated for a period of time (AGS/16 hours, SNU719/24 hours, SGC7901/42 hours) in 5% CO 2 at 37 °C, and then xed with 4% paraformaldehyde. In the matrix invasion assay, matrix gel and serum-free 1640 medium were injected into the transwell chamber at a ratio of 1:8 for pretreatment for 2 hours. 1 × 10 5 cells was suspended in 200 µL of serum-free RPMI-1640 medium, and pre-warmed medium containing 20% fetal bovine serum was then added to the well. The cells were incubated for a period of time (AGS/16 hours, SNU719/24 hours, SGC7901/42 hours) in 5% CO 2 at 37 °C, and then xed with 4% paraformaldehyde. A cotton swab was used to gently remove cells that have not migrated or invaded the membrane. Cells were stained with 0.1% crystal violet ( Balb/c nude mice (female, 8 weeks old, 19-21 g in mass) were purchased from GemPharmatech Co., Ltd (Nanjing, Jiangsu) Experimental Animal Co., Ltd. and raised in an SPF environment. PDX method was as mentioned before [21] .When the tumor volume reached about 150 mm 3 , they were randomly divided into three groups, and different reagents were given by intraperitoneal injection (IP). The tumor volume and body mass were measured every 3 days. The calculation formula of tumor volume is as follows: V = Πa b2/8 (where V is the tumor volume, a is the largest tumor diameter, and b is the smallest tumor diameter).
Nude mice were sacri ced on day 30, and tumors were taken for measurement and weighing.

Biochemistry, histology and immunohistochemical staining
In order to evaluate the toxicity of Ailanthone, mice with subcutaneously transplanted tumor were sacri ced at the end of the experiment, and blood serum, liver and kidney were harvested. Serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) were measured. Tumor or mouse tissue samples were immediately xed in 10% neutral buffered formalin for 24 hours, gradually dehydrated in a solution with increasing ethanol content (75, 85, 95 and 100%, v/v), and nally embedded in para n blocks. Liver and kidney were xed by para n-embedded formalin and cut into 3 mm sections. H&E staining was performed by the experimental histopathology laboratory in accordance with standard procedures. Observations were made under light microscope (100X magni cation). Immunohistochemical (IHC) staining: Collect tissues after xation, dehydration, embedding and sectioning. After heat-mediated antigen repair in 10 mM sodium citrate buffer (pH 6.2) or Tri-EDTA, 1.6% hydrogen peroxide was used to block endogenous peroxidase. Ki67 Quantitative real-time PCR AGS, SNU719, and SGC7901 cells were sown in 6-well plates and incubated with different concentrations of AIL for 12 hours. AG RNAex Pro reagent (Accurate Biology, CAT#AG21102) was used to extract total RNA from tissue samples or cell lines. Evo M-MLV reverse transcriptase premix (Accurate Biology CAT#AG11706) was used to synthesize cDNA from 2 µg RNA of each sample. SYBR Green Premix Pro Taq HS qPCR kit (Accurate Biology, CAT# AG11701) was used for qRT-PCR. Each sample was tested 3 times. The data was analyzed using the 2-ΔΔCT calculation method. The primers were as follows:

Western blotting
The cells were processed differently according to their corresponding requirements, and the protein concentration is determined by BCA Protein Assay Kit (KGI Biosciences), using SDS-PAGE Sample Loading Buffer, 5X (Beyotime) at 95° for 8 minutes. The lysate is separated on a polyacrylamide gel and transferred to nitrocellulose. The blot was detected with a speci c antibody, and then the membrane was detected with the LI-COR Odyssey CLx Infrared Imaging System (LI-COR Biotechnology, Lincoln NE). Antibodies were H2AX (1: 1000, Proteintech, 10856-1-AP), XRCC1 (1: 1000, Abcam, ab44830), Caspase3

Immuno uorescent(IF) staining
The collected organoids were xed for 16-20 hours, then embedded, sliced, baked, and depara nized. After heat-mediated antigen retrieve in 10 mM sodium citrate buffer (pH 6.2) or Tri-EDTA, 1.6% hydrogen peroxide was used to quench endogenous peroxidase, then blocked with PBS containing 5% BSA and 0.2% Triton-X. The primary antibody was incubated in blocking buffer at 4 °C for 16-20 hours. The uorescent secondary antibody was incubated for 1 hour at 20 °C, and then incubated in DAPI for 10 minutes.

RNA isolation and microarray
Total RNA was extracted from tissue samples, the concentration and purity of the extracted RNA were detected by Nanodrop 2000, RNA integrity was detected by agarose gel electrophoresis, and RIN value was determined by Agilent 2100. A single library construction requires that the total amount of RNA is not less than 5 µg, concentration ≥ 200 ng/µL, and the OD260/280 is between 1.8 and 2.2. mRNA capture and library preparation were completed by the advanced sequencing equipment of Shanghai Origingene Bio-pharm Technology Co., Ltd. using KAPA mRNA HyperPrep kit (Roche). Three biological library was 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. The high-quality DNA sequencing obtained after quality control was compared with the designated reference genome. For the PDX sample, it was rst compared to the mouse reference genome. After eliminating the mouse data, the remainder was then compared to the human reference genome retrieved from Ensembl database (genome version GRCh38, gene annotation information Ensemble 92). Before alignment, cutadapt (version 1.9.1) was used for quality trimming and adaptor removal of the original reading. Using the annotation release 86 as a reference, read was sequenced on human genome GRCh38 using RSEM 1.3.0 and STAR 2.5.2, and the subsequent gene levels were counted. In R program (version 3.6.1), 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.
Gene Set Enrichment Analysis (GSEA) GSEA was performed with the software (GSEA V4.0.3) developed by the Broad Institute of Massachusetts Institute of Technology and Harvard University (https://www.gsea msigdb.org/gsea/index.jsp). For the RNA-seq datasets of low and high expressionof P23, OXA group and AIL group, the normalized RNA read count was used for analysis, and the following settings were applied: number of permutations = 1000, type of permutation = gene set, enrichment statistics = weighting, measurement of gene ranking = signal noise. For the TCGA gastric cancer data set, 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 set as signi cant enrichment.

Screening of differentially expressed genes (DEGs)
The expectation-maximization algorithm of 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 normalized by quantiles and were normally distributed [22] . In this study, the R program package limma v3.28.14 was used to analyze the differential genes in the gene expression data, and its mRNA satis ed P < 0.01, false discovery rate (FDR) was < 0.01 and |log2 fold change (FC)| was > 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 null hypotheses rejected. FC is 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.

Clinical gastric cancer samples
The sample data were procured from the prospectively collected GC database of the First A liated Hospital of Sun Yat-sen University, Guangzhou, China. We reviewed the treatment records of patients undergoing gastric cancer resection from January 2010 to May 2012. The inclusion criteria were pathological TNM stages I-III gastric adenocarcinoma patients whom underwent radical surgery patients. Exclusion criteria: other history of malignant tumors, neoadjuvant therapy, loss to follow-up, no tissue samples. A total of 93 cases of tissue samples were collected from surgical specimens, and informed consent was obtained from the patients. This study was approved by the Ethics Committee of the First A liated Hospital of Sun Yat-sen University. The detailed clinical characteristics of the patients were shown in Table 1. The endpoint of the study was recurrence. The specimen tissues were tested by IHC, and the results were interpreted by two independent pathologists. They were blinded to the speci c diagnosis and prognosis of each case, and used a semi-quantitative method for scoring. Among the samples, more than 10% of tumor cells staining were positive. The staining intensity was divided into negative, weak, moderate and strong. Low P23 expression was detected in the negative and low intensity samples, and medium and strong intensity returned with high P23 expression. Subsequently, the diseasefree survival of patients in the high P23 expression group (n = 42) and low P23 expression group (n = 51) were evaluated. Progression free survival is de ned as the duration from tumor resection to progressive disease. Follow up was completed in intervals of 3 months (0-2 years), 6 months (2-4 years), and once a year until recurrence or June 2020. The follow-up study included abdominal computed tomography and postoperative physical examination.  Statistical analysis SPSS 22.0 software was used for statistical analysis. Data were expressed as mean ± standard deviation (s.d.), and P < 0.05 was considered statistically signi cant. Tumor volume was measured repeatedly using a general linear model. The difference between categorical variables was tested using χ2 test, and the difference between two groups was tested using student's t test. Kaplan-Meier curve and log-rank test were used to analyze the clinical data of gastric cancer patients.

AIL has a good inhibitory effect on GC
TCM has played an important role throughout human history before development of modern medicine took over. So far, many drugs are monomers that were rst inspired by it and subsequently extracted from plants, such as vincristine, which has been widely used in clinical treatment of cancer [7] , and artemisinin which saved countless lives from malaria [23] .From the above, it is fair to say seeking inspiration from ancient methodologies like TCM is still an important way to discover new medicines. From the 166 kinds of medicinal plants with anticancer effects recorded in the Pharmacopoeia of the People's Republic of China (2000 edition) [24] ,we have selected 9 kinds of medicinal plants( Table 1)that had been proven through modern science in recent years to have relatively good anticancer effects ( Figure S1-D). We then conducted drug susceptibility tests on three strains of GC organoids (Figure S1-ABC) we had established.
Fortunately, AIL excelled above the rest (Fig. 1-A),. AIL is an herbal monomer extracted from the bark of Ailanthus altissima [25] , which in recent years has been proven to have obvious anticancer effect on multiple cancer types. In order to verify whether AIL has the potential to inhibit the growth of GC cells, we evaluated its effect on GC cell growth and colony formation. The results showed that AIL signi cantly inhibited the growth of AGS, SNU719 and SGC 7901 cell lines ( Figure S2-AB). We also evaluated its effect on the spheroidizing ability of GC organoids, and the results showed that AIL inhibited the spheroidizing ability of GC organoids signi cantly ( Fig. 1-BCD). We further veri ed its effect on GC PDX models and found that it also had a good inhibitory effect in vivo ( Fig. 1-EFG, Figure S6-BC). There was no signi cant weight loss in mice during treatment as well ( Figure S3-B), with no signi cant difference in serum ALT and AST levels post treatment ( Figure S3-A). There was no damage found on liver and kidney function ( Figure S3-C). Additionally, AIL exerted a signi cant inhibitory effect on GC cell migration and invasion in transwell cell migration and invasion assays ( Figure S4-ABCD). In summary, these results indicated that AIL could signi cantly inhibit GC cell in vitro and in vivo, while maintaining a low cell toxicity edge. However, we do not know the mechanism by which AIL inhibits the proliferation of GC cells. To achieve this end, we had further explored its mechanisms.

AIL promotes apoptosis through inducing DNA damage
In order to clarify the question of how AIL inhibits the proliferation of GC cells, we conducted a series of exploratory experiments. Cell cycle analysis and apoptosis analysis were performed via ow cytometry on the GC organoids after AIL treatment. The results showed that AIL did not induce cycle arrest on the cells ( Figure S5-AB), but it did signi cantly induced apoptosis ( Fig. 2-AB). We also explored whether AIL increases DNA damage in GC cells. Western blotting and IF analysis found that AIL did induce DNA damage in GC cells (Fig. 2-CDEFGH), and it was concentration (Fig. 2-EFG) and time dependent (Fig. 2-H).
It was revealed that AIL has a stronger ability to induce DNA damage (Fig. 2-CDEFGHIJ). In this regard, we speculate that AIL may be able to inhibit DNA repair, thereby causing continuous DNA damage.

AIL inhibits DNA repair in GC cells
To prove the above hypothesis, we sequenced the transcriptomes of the PDX tissues. The pathway enrichment analysis revealed that the BER pathway associated with DNA repair was mainly regulated by AIL (Fig. 3-A); the differential gene expression analysis suggested that the AIL group had a signi cant inhibitory effect on the key gene XRCC1 of the BER pathway ( Fig. 3-B) [26][27][28] .Subsequent experiments also con rmed that AIL has a good inhibitory effect on XRCC1 of GC cells (Fig. 3-CDFG). In previous study, AIL had been proven to treat castration-resistant prostate cancer by down-regulating P23 [8,29] .
Bioinformatics analysis showed that P23 is closely related to DNA repair-related pathways, that is to say that high expression of P23 leads to higher enrichment of the BER, HR, NHEJ, and MMR pathways when compared with low expression of P23, indicating that P23 promoted DNA repair ( Figure S6-A). Western blotting also con rmed that AIL can signi cantly inhibit the expression level of P23 in GC cells (Fig. 3-CD). As an important co-chaperone of HSP90, P23 plays a key role in HSP90's participation in chromatin remodeling, DNA transcription, RNA processing, DNA replication, telomere maintenance and DNA repair. In fact, involvement of HSP90 in DNA repair requires its co-chaperone P23 or phosphorylation to interact with DNA metabolism proteins. However, our results showed that AIL did not increase the expression of HSP90 (Fig. 3-D).Instead, we noted that AIL is not an ATP-competitive inhibitor of HSP90 like 17-AAG, which increases the expression of HSP90 protein [30] . Interestingly, AIL could su ciently inhibit the AKT protein expression of GC (Fig. 3-D). We further compared the inhibitory effects of 17-AAG, CEL and AIL on XRCC1, and found that AIL has signi cant advantages as well (Fig. 3-E).The IHC of PDX tissue also con rmed that AIL inhibits the expression of P23, HSP90 and AKT (Fig. 3-FG, Figure S6-BC). These results indicated that the DNA damage of GC cells induced by AIL was mainly related to the inhibition of the BER pathway.
AIL regulates XRCC1 through P23 to inhibit DNA repair Although we had con rmed the effects of AIL on P23, HSP90, and XRCC1, the regulatory relationship between them was still unclear. In the past, the HSP90 chaperone system was thought to occur in the cytoplasm [31] .It is becoming increasingly clear that HSP90 and its associated chaperone proteins have important functions in the nucleus, including chromatin remodeling, DNA transcription, RNA processing, DNA replication, telomere maintenance and DNA repair [32,33] .The involvement of HSP90 in DNA repair depended upon its co-chaperone P23, or phosphorylation to interact with DNA metabolism proteins [32][33][34] .In cancer cells, HSP90 and P23 bind stably to form a super-chaperone complex, which is the activated state of the protein [35] .The existence of P23 is not solely limited to HSP90, because P23 also ful lls a variety of functions unrelated to HSP90 in cells [17] .In a prostate cancer study, it was found that by inhibiting P23, the client protein, androgen receptor of HSP90 can be signi cantly inhibited. As an important client protein of HSP90, the important scaffold protein XRCC1 e ciently promotes the repair of DNA single-strand breaks (SSBs) [36,37] .If the SSBs are not properly repaired, they may convert into doublestrand breaks during DNA replication, eventually leading to genetic instability and apoptosis [38] .Therefore, we theorize that AIL inhibits the interaction between P23 and HSP90 by down-regulating P23, thereby affecting the interaction between HSP90 and its client protein XRCC1.In order to verify the regulatory axis of this molecular mechanism, we rst proved the regulatory effect of P23 inhibitor (CEL) on HSP90 and XRCC1 (Fig. 4-A). To verify that AIL functions through P23,we separated P23(+) GC organoid and P23(-) GC organoid strains through ow sorting (Fig. 4-B). qPCR analysis found that P23(+) GC organoids upregulated XRCC1, while P23(-) GC organoids inhibited XRCC1 (Fig. 4-CD). Subsequently, we veri ed the inhibitory effect of HSP90 inhibitors on XRCC1 in GC cells (Fig. 4-E). At this point, we had basically con rmed that AIL inhibited XRCC1 by downregulating P23. Further bioinformatics analysis found that HSP90 is positively related to DNA repair, that is, as HSP90 is upregulated, its DNA repair ability gets stronger ( Figure S7-A). In summary, AIL has a obvious tumor suppressing effect because it can induce DNA damage while inhibiting its repair in GC cells. For this reason, we further compared the spheroidization ability of GC organoid cells treated with 17-AAG, CEL and AIL respectively and the results showed that the anticancer effect of AIL was excellent. This showed that P23 was not only upstream of XRCC1, but it can also regulate XRCC1.
AIL inhibits the binding of P23 to HSP90 and HSP90 to XRCC1 As an important co-chaperone of HSP90, P23 binds stably with HSP90 in cancer cells to form a super chaperone complex [39] it could also bind with HSP90 without the presence of any client protein [40] ;unlike XRCC1, which requires to bind with HSP90 as a chaperone molecule in order to participate in DNA repair [41] .Since AIL can inhibit the expression of P23 and XRCC1, does it also inhibits the interaction between P23 or XRCC1 and HSP90 at the same time? Through immunocolocalization and coimmunoprecipitation, we proved that both of P23 and XRCC1 can bind with HSP90 in GC cells, and AIL can inhibit their combination (Fig. 5-ABC). The above results revealed that AIL not only inhibited the expression of P23 and XRCC1, but it also inhibited the binding of P23 to HSP90, and HSP90 to XRCC1.
High expression of P23 is related to GC recurrence P23 is proven to promote breast cancer metastasis and affect its prognosis [42] ; it also promotes the invasion and metastasis of prostate cancer [42,43] , and targeting P23 can treat castration-resistant prostate cancer [8] . In this study, we had proven that AIL can target P23 in GC cells. In order to explore the clinical impact and status of P23 on gastric cancer, we had enrolled patients with GC recurrence from Sun Yat-sen University's Gastric Cancer Research Center. We used IHC of tumor tissues to detect the expression level of P23 and found that P23 was highly expressed in the specimens of patients who relapsed ( Fig. 6-A). The recurrence time of patients with high expression of P23 was signi cantly shorter than that of patients with low expression (Fig. 6-B). It showed that P23 promoted GC recurrence, and P23 could be used as an important index for clinical prediction of postoperative recurrence in patients.

Discussion
Chemotherapy plays a pivotal role in GC treatment [2] , however, even with a variety of chemotherapeutics and targeted therapy currently available in clinic, its therapeutic effect is still far from satisfactory. As the fth most common cancer in the world, GC is still the third leading cause of cancer-related death with only a 5-year survival rate of merely 29% [1] . Therefore, it is crucial to nd new drugs for its treatment. In this study, we selected medicinal plants with anticancer effects from the "Pharmacopoeia of the People's Republic of China" (2000 edition), combining with literature meta-screening, and applications of organoids established by our center as a powerful drug screening tool [44][45][46] ,we had successfully screened out the monomer AIL which can signi cantly inhibit the proliferation of GC cells, induce apoptosis, and has a good anticancer effect in vivo.
Organoids, as multi-cell clusters constructed by three-dimensional culture in vitro, have the ability to selfrenew and self-organize, thus maintaining the physiological structure and function of their source tissues [47] . With this advantage, organoids quickly became powerful tools for tumor and stem cell biology research [44] .In this study, we used organoids as research tools and con rmed that AIL can effectively inhibit the proliferation of GC cells in vitro and in vivo. Although AIL has no cycle arrest effect on GC cells, it can signi cantly induce GC cell apoptosis. Through transcriptome sequencing of the PDX tissues, we can observe that the DMSO group was enriched in the BER and HR DNA repair pathways, thus proving that AIL could suppress DNA repair pathways; differential gene expression analysis suggested that the AIL group had a signi cantly strong inhibitory effect on XRCC1, the key gene in the BER pathway. Subsequent experiments also con rmed that AIL has a good inhibitory effect on XRCC1 in GC cells.
Therefore, we speculated that AIL could inhibit DNA damage repair while inducing DNA damage in GC cells.
XRCC1 as an important gene for DNA repair in the BER pathway [26,35] can interact with PARP1, DNA ligase III and Pol proteolysis to promote effective repair of DNA SSBs [36,37] . If SSBs are not repaired promptly and properly, they may convert into DSBs during DNA replication, leading to apoptosis [38] . XRCC1 function depends on the binding and stability of HSP90 [41] .By binding and stabilizing XRCC1, the phosphorylated form of HSP90 promotes the formation of additional XRCC1 complexes because in the absence of HSP90 or HSP90 binding, free XRCC1 will be removed by ubiquitin-mediated degradation [41] .In recent years, more and more evidence has shown that HSP90 played an important role in cell homeostasis, transcription regulation, chromatin remodeling and DNA repair [17,18,35] .In fact, P23 not only showed overlapping interactions with HSP90, but it also has more interactions independent of HSP90 [17,50] . Our study also found that P23 can promote DNA repair. It has been reported that AIL can treat castration-resistant prostate cancer by inhibiting P23 [8] . Surprisingly, our study also veri ed that AIL can signi cantly inhibit P23 in GC cells. At the same time, AIL can also inhibit the protein binding of P23 to HSP90, and HSP90 to XRCC1. Additionally, AIL can also inhibit AKT and binding of AKT to HSP90. AKT can mediate DNA damage repair [51] ,and act as a client protein of HSP90 while interacting with HSP90 [52] .It is possible that the DNA damage induced by AIL itself and its other effect of DNA repair inhibition by blocking multiple signal pathways of GC cells led to synthetic lethality [53] .
It is known that P23 plays a key role in DNA repair, and through this study, we con rmed that AIL can signi cantly inhibit P23. We further proved that downregulation of P23 can inhibit HSP90 and XRCC1, and interestingly, while dividing GC organoids into P23(+) and P23(-) groups through ow sorting, it showed that the P23(+)group had upregulated XRCC1, while the P23(-)group had downregulated XRCC1.
All these indicated that P23 plays a central role in the P23/HSP90/XRCC1 axis. Other researches have revealed that the expression of P23 is increased in some cancers. For example, highly expressed P23 promotes breast cancer progression and poor prognosis by increasing lymph node metastasis and drug resistance [54] ; while cells with high expression of P23 in prostate cancer acquires a more aggressive phenotype, which promotes its invasion and metastasis, thereby leading to disease progression and a worse prognosis [55] .However, research on clinical prognosis in GC has not been carried out yet. Our comparative study on the recurrence time of GC patients with low through high expression of P23 through immunohistochemical methods found that patients with high expression of P23 had a shorter relapse time, indicating that P23 is an important factor affecting the recurrence and development of GC. Therefore, it provides an important foundation for the targeted therapy of P23.

Conclusion
In summary, we clari ed the anticancer effect of AIL in gastric cancer through GC cell lines, organoids

Consent for publication
All authors agree to submit the article for publication.

Availability of data and materials
The datasets from the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare no con ict of interest.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementaryFigures.doc