Identication of SPINT3 as a Metastatic Suppressor in Gastric Cancer Through Induction of TGF-β1 Secretion

Background: Cancer metastasis is responsible for most cancer-related deaths. SPINT3 (serine peptidase inhibitor, Kunitz type, 3) is a newly identied gene, and its relationship with gastric cancer has not been well studied. Methods: We rst designed an sgRNA sub-library to screen for metastatic suppressors, and chose SPINT3. We then used QPCR and western blotting to detect the expression level of SPINT3 in gastric cancer tumor tissues. In addition, we used annexin V/propidium iodide (PI) to detect changes in apoptosis. PI staining and wound healing assays were used to detect the cell cycle and changes in cell migration, respectively. Finally, CyTOF and T cell proliferation experiments were performed to detect changes in immune cells in the tumor microenvironment. Results: SPINT3 sgRNA (based on the sgRNA library used) was present in every metastatic tumor. The expression of SPINT3 was low in gastric cancer tissues, and this increased the metastatic ability of gastric cancer cells. In addition, high expression of SPINT3 induced cell apoptosis, downregulated cell proliferation, and impaired cell migration. Furthermore, low expression of SPINT3 induced the secretion of TGF-β1 and an immunosuppressive signal, which could be blocked by apatinib and SB431542. Conclusion: We conrmed that SPINT3 is a metastasis suppressor in gastric cancer.


Introduction
Gastric cancer is the fth most common cancer in the world after lung, breast, colorectal, and prostate cancer [1]. In 2012, there were approximately 952,000 new cases and 723,000 gastric cancer-related deaths worldwide [2]. Gastric cancer is characterized by geographical distribution, and is common in East Asia, Central Europe, and South America [3]. The number of gastric cancer-related deaths in China in 2012 was 320,000, accounting for 45% of total deaths in the country [4]. This shows that the disease burden of gastric cancer is extremely heavy, and it makes the search for new biomarkers and therapeutic targets a top priority. SPINT3 (serine peptidase inhibitor, Kunitz type, 3) is a relatively new gene, and its function might be related to serine-type endopeptidase inhibitor activity. In this study, we chose SPINT3 as our candidate gene owing to the wide presence of SPINT3 sgRNA in a metastatic mouse model. Studies have shown that SPINT3 is expressed at low levels in gastric cancer tissues, and that downregulation of SPINT3 activates tumor cell metastasis and generates an immunosuppressive signal.

Ethics Statement
This the study was carried out in compliance with the ARRIVE guidelines. All experiments were approved by the First A liated Hospital of Zhejiang University. All the patients involved in this study provided informed consent, and all the sample extraction protocols followed the Declaration of Helsinki and were authorized by the Institutional Ethics Board of The First A liated Hospital, School of Medicine, Zhejiang University. All mouse experiments were performed at the First A liated Hospital of Zhejiang University in accordance with the relevant guidelines and regulations formulated by the hospital. Mice were anesthetized using avertin and sacri ced by applying excessive carbon dioxide.

Patient samples
Tumor tissues and adjacent normal tissues of ve cases of gastric adenocarcinoma were collected from The First A liated Hospital, School of Medicine, Zhejiang University. The samples were divided into two parts and subjected to QPCR and western blotting, depending on the size of the sample.

Mouse experiment
The SCID, nude, and 615 mice used in this study were housed in the animal facility of the First A liated Hospital of Zhejiang University. (1) For the library experiment, nude mice (6-8 weeks old) were injected subcutaneously with the CON or SPINT3 KO MKN-45 cell line, and then the primary tumors were harvested, cut into 1 mm × 1 mm × 1 mm slices, and implanted in the gastric mucosa of SCID mice. (2) To detect the tumor microenvironment, 5 million CON or SPINT3 KO MFC cells were injected subcutaneously into 615 mice; after 2 weeks, the primary tumors were harvested and digested into single cells. For, Mice were anesthesia by avertin in the implantation of tumor cells, and sacri ced by excessive carbon dioxide inhalation.

In silico analysis
In this study, all in silico analyses were completed using R 3.4.3 software. (1) Extraction of 22-gene list: the transforming growth factor beta binding pathway (GO:0050431) was selected, and the genes involved in this pathway were extracted using AnnotationDbi and annotated using the org.Hs.eg.db package. (2) Other analysis: TCGA-STAD (Stomach adenocarcinoma dataset from The Cancer Genome Atlas Program) data were downloaded and normalized using the GDCRNATools package, and the gene expression data, follow-up time, and survival status of patients were extracted and analyzed using the univariate COX algorithm. For pathway enrichment analysis, samples were rst divided into SPINT3 low and high expression groups based on the expression of SPINT3; then, the GSEA function from the clusterPro le package was used to detect pathway enrichment in SPINT3 low expression samples and compare it with that in SPINT3 high expression samples.

Sub-library design
(1) Twenty-two genes belonging to GO:0050431 were selected, and the sgRNA that target these genes were designed; the sgRNA oligo sequences are listed in Table 1. (2) Oligos were cloned into the lentiCRISPR v2 vector (Addgene 52961). (3) The vectors were packaged as lentiviruses using the 2 nd lentivirus package system.

QRT-PCR
(1) Firstly, cells were lysed with 500 µL TRIzol and centrifuged at 12000 rpm for 10 min to remove the cell fraction. (2) The solution was added to 200 µL chloroform, shaken up and down for 15 s, and centrifuged at 12000 rpm for 15 min. (3) 100 µL of the supernatant was drawn, mixed with 100 µL isopropanol, and centrifuged at 12000 rpm for 10 min. (4) The precipitate was washed with 75% ethanol, and the concentration of total RNA was detected using the OD at 260 nm/280 nm. (5) Total mRNA was reverse transcribed using the PrimeScript RT Reagent Kit with gDNA Eraser (R047A, Takara Bio) and TB Green Premix Ex Taq (Tli RNase H Plus) (RR420A, Takara Bio) to detect the expression of the internal reference gene and target gene; the primer sequences used in this study were human SPINT3 F-5′-GATCTCCTCCCAAATGTATGCGC-3′, R-5′-TCCGCAGCCTCCGTAAGCAAAT-3′, and human GAPDH F-5′-GTCTCCTCTGACTTCAACAGCG-3′, R-5′-ACCACCCTGTTGCTGTAGCCAA-3′. The relative expression level of SPINT3 was determined using the 2 -△△CT value.

Western blotting
(1) Bulk tissue was lysed using RIPA buffer on ice for 30 min, mixed with loading buffer, and boiled for 15 min. (2) 20ug of samples were loaded on polyacrylamide gels and proteins were separated based on molecular weight using electrophoresis. (3) The proteins were then transferred to a PVDF membrane and blocked with 5% BSA solution. (4) the membrane was labeled with different primary antibodies, including GAPDH (ab8245), TGF-β1 (ab215715), TGF-β type 1 receptor (ab31013), TGF-β type 2 receptor (ab186838), Smad3 (ab40854), and p-Smad3 (ab52903), which were all purchased from Abcam. The SPINT3 antibody was synthesized by Beyotime Company (Shanghai, China), dilution is 1:1000. After culturing overnight, the membrane was washed thrice and labeled with secondary antibodies, and protein expression was quanti ed using ECL solution.

Cell apoptosis, proliferation, migration, and cell cycle
For cell apoptosis, proliferation, and cell cycle experiments, MKN-45 cells were rst transfected with the CoN and SPINT3 overexpression vectors for 48 h. (1) Cell apoptosis was detected using the APCconjugated Annexin-V/propidium iodide (PI) kit (BD Biosciences, San Jose, CA, USA). Cells were rstly harvested and washed with PBS, and then suspended in annexin V binding buffer; 5 µL of annexin V APC antibody and 5 µL of PI were added, and after culturing for 15 min, the BD Calibur machine was used to detect positive cells from the FL2 and FL4 channels. The ratio of apoptotic cells was quanti ed using the percentage of APC-positive cells. (2) Cell proliferation was detected using Cell Counting Kit-8 (CCK-8) (Dojindo Molecular Technologies, Rockville, MD, USA); rstly, cells were dispersed in a 96-well plate, each well was seeded with 2000 cells, 10 µL of CCK-8 solution was added, and the cells were cultured for 2 h. Then, a microplate reader was used to measure the OD at 450 nm, and the proliferation index was quanti ed using the ratio OD450 nm at other time point/OD450 nm at 0 h. (3) Cell migration: Cells were rstly dispersed in a 24-well plate, and each plate was seeded with 100000 cells; after cell attachment, white tips were used to scratch the bottom, and pictures were taken at the 0 and 12 h time points. (4) Cell cycle: After transfection of the SPINT3 vector for 48 h, cells were washed with PBS and xed in 70% ethanol overnight. The cells were then stained with PI and cultured for 20 min; the BD Calibur machine was used to detect changes in the cell cycle.

CyTOF experiment
(1) Firstly, single cells from primary tumors were isolated and the CyTOF experiment was completed by Novogene (Beijing, China). The markers used in CyTOF analysis included LY6G, CD8A, CD45, CD11B, CD28, CD3E, LY6C, CD19, CD24, CD14, F4-80, and CD326. (2) After obtaining the FCS le, the data were analyzed using R 4.0.2 software. The FCS le was read using the owCore package, and analyzed using the Rtsne package; the conditions used for the Rtsne function were tsne_expr, check_duplicates = FALSE, pca = T,pca_scale = T,perplexity = 120. Finally, the results were visualized using the ggplot2 package.  2) The labeled T cells were divided into two parts. One part was used to detect the expression of CFSE by ow cytometry, and the other part was used for co-culture with MDSCs. (3) After counting the MDSC cell suspension, the supernatant of the CoN group or SPINT3 sgRNA group was used for re-suspension, and 2 folds of CD8+ T cells were added. (4) CFSE expression in CD11b-cells was detected using ow cytometry. Cells with a lower uorescence intensity at 0 h were used as T cells in the proliferation phase.

Results
3.1 Identi cation of SPINT3 as metastatic suppressor using CRISPR library In this study, to determine the link between TGF-β binding and gastric cancer metastasis, we rst extracted 22 genes from the transforming growth factor beta binding pathway (GO:0050431), and then synthesized an sgRNA library to target these 22 genes, as shown in Figure 1A. After 12 weeks, all the SCID mice were dissected, and the presence of the GFP signal was detected. Among the 20 SCID mice implanted with gastric cancer tissue from the sgRNA library group, 12 developed lung metastasis, and 5 SCID mice implanted with MKN-45 primary tumor tissue did not have any metastatic tumors, suggesting increased invasion upon library treatment ( Figure 1B). Then, we checked the expression of sgRNA in the lung; using PCR, we con rmed that all the 12 library mice were positive for SPINT3 sgRNA. Moreover, VASN and TWSG1 were present in 5 mice ( Figure 1C, D).Based on these results, we chose SPINT3 as our candidate gene for further analysis.
According to the literature, the function of SPINT3 includes serine-type endopeptidase inhibitor activity; however, the exact role of SPINT3 in gastric cancer remains unknown. Hence, we rst extracted the gastric cancer dataset from the TCGA database and found that the expression of SPINT3 (relatively fold change) in healthy donor and gastric cancer tissues is 8.13 ± 1.49 vs. 6.23 ± 1.82; using the t-test, we con rmed that SPINT3 expression was signi cantly lower in gastric cancer tissue than in healthy tissue ( Figure 1E). In-house data also showed lower SPINT3 expression in 5 cases of gastric cancer compared with normal tissue at the mRNA level ( Figure 1F); furthermore, SPINT3 protein expression was low in the tumor tissue ( Figure 1G). From a clinical perspective, patients with low SPINT3 expression would have a poorer prognosis, further proving that SPINT3 is a possible tumor suppressor in gastric cancer ( Figure  1H).

Low expression of SPINT3 increases metastasis in gastric cancer cell line
Given that SPINT3 was consistently expressed at low levels in gastric cancer tissue, we infected the MKN-45 cell line with the CoN or SPNT3 sgRNA lentivirus (Figure 2A), and then used the model in part 1 to detect the changes in CoN and SPINT3 sgRNA cell metastasis. Lung metastasis nodules ( Figure 2B) and GFP intensity ( Figure 2C) in SPINT3 KO mice were signi cantly more than those in the CoN group. In addition, SPINT3 expression was lower in the lung metastasis nodules( Figure 2D), proving the link between low expression of SPINT3 and increased lung metastasis.

Relation between expression of SPINT3 and biological characteristics of gastric cancer cells
To further validate the role of SPINT3 in gastric cancer, we designed a SPINT3 overexpression vector and transfected it into a gastric cancer cell line. Using western blot, we con rmed that the level of SPINT3 expression was elevated in the SPINT3 cDNA-transfected group ( Figure 3A). Using annexin V/PI double staining, we found that apoptosis was induced after transfection of SPINT3 for 48 h ( Figure 3B); moreover, cell growth was inhibited in SPINT3-overexpressing cells ( Figure 3C, D). Furthermore, cell migration was inhibited after transfection of SPINT3 cDNA ( Figure 3E). These results indicated that inhibition of SPINT3 was critical for the survival and migration of gastric cancer cell lines.
Using in silico analysis, we concluded that in the TCGA-STAD dataset, the TGF-β1-related pathway was highly enriched in SPINT3 low expression samples ( Figure 3F). Using western blots, we also con rmed that overexpression of SPINT3 decreased the TGF-β1 signal, indicated by downregulated TGF-β1, TGFβ receptor 1, and downstream signal Smad3 ( Figure 3G). SPINT3 knockout also increased the secretion of total TGF-β1 and active TGF-β1 ( Figure 3H), suggesting that the de ciency of SPINT3 might modify the tumor microenvironment via the secretion of TGF-β1.

Modi cation of tumor microenvironment by SPINT3 KO
In the previous section, we stated that knocking out SPINT3 could increase the secretion of TGF-β1. TGF-β1 is a well-known factor that regulates cell proliferation [5], EMT [6], and cancer metastasis [7], as well as the tumor microenvironment [8]. In the TGF-β-rich tumor microenvironment, CD8+ T cell proliferation and maturation of T helper cells are inhibited [9]. In our study, to determine the changes in immune cells in the primary tumor, we chose 615 mice owing to their complete immune system. Firstly, we infected the MFC cell line with the CON or SPINT3 sgRNA lentivirus ( Figure 4A), and then injected CON, SPINT3 shRNA, and SPINT3-overexpressing MFC cells into 615 mice subcutaneously. Consistent with previous results, knockout of SPINT3 increased tumor cell proliferation in vivo after 2 weeks ( Figure 4B). Single cells from the primary tumor were subjected to CyTOF analysis ( Figure 4C), and we con rmed that after SPINT3 was knocked out, CD8+ CD28+ T cells decreased in number, whereas M2 macrophages, PMN-MDSCs (CD11b + Ly6G + Ly6C low/int ), and MO-MDSCs (CD11b + Ly6G − Ly6C high ) increased in number ( Figure 4D). We further validated this change using FACS, and proved the decrease in the number of CD8+ T cells ( Figure  4E) and increase in the number of PMN-MDSCs (LY6G+) and MO-MDSCs (LY6C+) ( Figure 4F). Moreover, the levels of active markers of MDSCs, including NOS2 and ARG1, signi cantly increased in SPINT3 knockout tissues ( Figure 4G).
MDSCs are a bone marrow-derived heterogeneous cell population with immunosuppressive activity, and recent studies have shown that the expansion of MDSCs is capable of inhibiting T cell proliferation in vivo and in vitro. Therefore, in our experiment, we sorted MDSCs from the primary tumors in the CON and SPINT3 knockout groups, co-cultured them with T cells from the spleen of 615 mice, and found that the activity of the MDSCs was higher in the SPINT shRNA group than in the CON group; the activity was quanti ed using the percentage of ampli ed T cells ( Figure 4H). These results suggested that, in SPINT3 knockout primary tumors, the anti-tumor immune response was compromised, reducing cancer cell recognition and clearance, and promoting tumor cell invasion.

Restoration of T cell proliferation by inhibitor
To inhibit the increase in tumor cell proliferation and invasion caused by SPINT3 de ciency, we investigated a drug for target therapy, and con rmed that the expression of SPINT3 in SPINT3 shRNA cells was reversed by treatment with apatinib and the TGF-β1 inhibitor SB431542 ( Figure 5A). Further experiments proved that cell apoptosis and proliferation were inhibited ( Figure 5B,C), and in the primary tumors from 615 mice, the percentage of PMN-MDSCs and MO-MDSCs decreased after treatment ( Figure  5D). The activity of MDSCs decreased in the treatment group, as shown by the T cell proliferation experiment ( Figure 5E), and treatment with apatinib and SB431542 in vivo successfully blocked the metastasis process in 615 mice ( Figure 5F).

Discussion
In this study, we rst con rmed that SPINT3 was expressed at low levels in gastric cancer tissues and correlated it with the patient survival rate. Then, using multiple experiments, we proved that SPINT3 expression could in uence cell proliferation, migration, and apoptosis. These results show that SPINT3 is a tumor suppressor.
The function of SPINT3 is still unclear, and we deduced that it might be related to the TGF-β1 secretion signal as (1) SPINT3 is involved in the TGF-β1 binding pathway, and (2) in SPINT3 low expression gastric cancer samples, the TGF-β1 secretion and activation pathway was highly enriched. We used ELISA to detect the change in TGF-β1 concentrations in the supernatant of CoN and SPINT3 sgRNA tumor cells, and found that total TGF-β1 and activated TGF-β1 levels increased. TGF-β1 has a signi cant impact on the recruitment, activation, and function of immune cells; in macrophages, TGF-β1 inhibits the identi cation and clearance of cancer cells and induces polarization from the M1 to M2 type [10], which further induces the formation of cancer-associated broblasts (CAFs) [11]. In NK cells, TGF-β1 can downregulate the expression of NKp30 and NKG2D on the cell surface; it can also inhibit MHC I expression in cancer cells, and suppress the function of NK cells [12]. Elevated TGF-β1 levels in the tumor microenvironment can also regulate the function of myeloid-derived cells, mainly of 3 types: (1) tumorassociated macrophages [13], (2) tumor-associated neutrophils [14], and (3) unmatured CD11B+ GR1+ myeloid cells (myeloid immune suppressor cells, MISCs) or MDSCs [15]. These 3 types of cells are associated with poorer prognosis of cancer patients and resistance to target therapy. On the other hand, most of the recent studies have focused on TGF-β1 secreted by these cells, as opposed to TGF-β1 secreted by tumor cells, and on the changes in the EMT process; for example, SALL4 overexpression in gastric cancer cells can activate TGF-β/Smad and further promote EMT, thus stimulating the metastasis of gastric cancer cells [16].
In this study, we found that the downregulation of SPINT3 increased the number of MDSCs and M2 macrophages in the tumor microenvironment, and that the proliferation of CD8+ T cells was inhibited by the activation of MDSCs. Moreover, the upregulation of SPINT3 using drugs such as apatinib or the TGF-β1 inhibitor SB431542 could inhibit the function of MDSCs in primary tumors. Based on these results, we concluded that SPINT3 could function as a metastatic suppressor in gastric cancer through the secretion of TGF-β1.

Declarations
Con ict of interest

Figure 2
Low expression of SPINT3 increase the metastasis in gastric cancer cell line (A) Gastric cancer cell line MKN-45 were infected with CON and SPINT3 sgRNA for 48h, then using western blot to detect the change of SPINT3 expression in protein level. (B) Metastasis ability of CON and SPINT3 sgRNA cells were detected by the mouse model in Figure 1A. (C) GFP intensity in CON and SPINT3 sgRNA were detected by image J software. (D) Using western blot to detect the change of SPINT3 in CON and SPINT3 sgRNA.

Figure 3
Relation between expression of SPINT3 and biological characteristics of gastric cancer cell. Gastric cancer cells were divided into CON and SPINT3 group, then transfected with CON and SPINT3 cDNA, after  T cell(E) and MDSCs(F) from CON and SPINT3 sgRNA primary tumor were further validated by FACS. (G) CD11B+ GR1+ cells were sort out, and using QPCR to detect the change of SPINT3, NOS2 and ARG1. (H) MDSC from CON and SPINT3 sgRNA primary tumor were co-cultured with T cells from spleen, and using CFSE to detect the change of cell proliferation. Figure 5