All the collected human gastric cancer tissues came from The Affiliated Cancer Hospital of Nanjing Medical University between 2016 and 2017, and the patients’ informed consents were obtained. Our hospital’s ethics committee evaluated and authorized this experiment prior to its implementation (NYDLS-2019-919). At the same time, we carefully read and abided by the Declaration of Helsinki. After the gastric cancer tissues were obtained, we strictly abided by the relevant tissue handling specifications and froze them in the refrigerator at -80°C.

In our high-throughput sequencing files, the expression data of tRFs and tiRNAs were presented in counts and normalized to total aligned reads (CPM). The R package edgeR was utilized to screen the tRFs and tiRNAs. We used Fold Change (cutoff 1.5) and P-value (cutoff 0.05) for screening differentially expressed tRFs and tiRNAs. Principal Component Analysis (PCA), Correlation Analysis, Venn plots, Hierarchical clustering, Scatter plots, and Volcano plots were analyzed in R software or Perl environment. We utilized the GO and KEGG analysis on the potential targets of tRNA derivatives. Analysis of Biological Process (BP), Cellular Component (CC), and Molecular Function (MF) was presented with R software in GO analysis. P < 0.05 was considered significantly enriched in a certain process. Pathway clustering was conducted using KEGG analysis. We applied the enrichment score to indicate the significance of correlation. To elucidate the GC-related pathways more comprehensively, we acquired microarray profile GSE65801 in the GEO database and TCGA-STAD database. We then undertook the KEGG and GSEA analysis to further investigate the possible biological characteristics of GC.

We obtained the GC cell lines NCI-N87, HGC-27, BGC-823, SGC-7901, and GES-1 from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). GES-1 cell line was grown in 10% FBS DMEM medium. Medium for HGC-27, BGC-823, and SGC-7901 was RPMI 1640 with 10% FBS, for NCI-N87 was RPMI 1640 medium with 15% FBS. All media were purchased from ThermoFisher. We incubated the cell lines in a humidified incubator at 37°C with 5% CO2. The GC cells were seeded in a 6-well plate at a cell density of 20%-40%. 24 hours later, we used Lipofectamine 2000 (ThermoFisher, USA) to transfect tRF mimics or inhibitors (Ribobio, Guangzhou, China) (50 nM) into GC cells. We selected p38 MAPK pathway inhibitor (p38 MAPK-IN, MCE, USA) to act as the positive control of tRF-Glu-TTC-027. The detailed steps are available in our previous study (14). The sequences of the tRF mimic and corresponding control group were listed in Table 1 .

We put the preserved GC tissue into the TRIzol solution (Life Technologies, USA) for RNA extraction. After the process of chloroform, isopropanol, ethanol, we got the relatively pure RNA. During the process, the RNA solution was centrifuged several times. Then we applied nanodrop to determine the RNA concentration and the OD values of 260/280 absorbance ratios were between 1.8 and 2.0. Next, we performed RNA reverse transcription (RT) using the riboSCRIPT Reverse Transcription Kit (RiboBio, Guangzhou, China). Reaction conditions for RT were 60 min at 42°C and 10 min at 70°C. Finally, we applied LightCycler 1.5 (Roche, Switzerland) to manipulate PCR in a 20 μL reaction system (2 μL cDNA, 6.4 μL DEPC, 10 μL SYBR Green Mix, 0.8 μL forward primer, and 0.8 μL reverse primer). PCR reaction conditions were 95°C for 30 sec, 40 cycles of 95°C for 5 sec and 60°C for 20 sec. U6 was used as an internal control. The detailed steps are available in our previous study (14). The primers used in qPCR were presented in Table 1 . The relative expression was analyzed with the 2^-ΔΔCt method.

After the cells were treated with 4% paraformaldehyde, the cell membranes were permeabilized by Triton-100 (Beyotime, Shanghai, China) for 5 min at 4°C. Next, we added the hybridization solution mixed with Cy3-labelled tRF-Glu-TTC-027 to the permeabilized cells to locate the tRF-Glu-TTC-027 in GC cells. We stained the GC nuclei with DAPI and captured the pictures with a fluorescence microscope.

A certain number (2 x 10³) of transfected GC cells were transferred to a 96-well plate, and then we added 10ul CCK8 solution (Dojindo, Japan) into each well to measure the OD value in 450 nm by the microplate reader after 2 hours of incubation at 0, 24, 48, 72 and 96 hours. We added 1,000 transfected cells into each well of the 6-well plate, and these GC cells were cultured for 10-14 days. Subsequently, we terminated the cultivation with 4% paraformaldehyde as soon as the macroscopic clones were observed. The 1% crystal violet was used to stain the clones and the numbers were counted using Image J software.

Cell-Light EdU Apollo In Vitro Kit (Ribobio, Guangzhou, China) for EdU assay was utilized to compare the growth ability of the transfected GC cell. NCI-N87 and HGC-27 cell lines were transfected with tRF-Glu-TTC-027 mimics or mimics control, then 4 x 10³ transfected cells were transferred into each well of the 96-well plate. We diluted the EdU solution with complete medium to prepare enough 50 μM EdU medium. And EdU medium was then added into each well to incubate the cells for two hours. After the fixation, permeabilization, and staining, we captured the images with a fluorescence microscope.

For wound healing assay, GC cells were seeded in a 6-well plate with 1.5×10⁶ cells/well 24 hours before the experiment, and the tRF-Glu-TTC-027 mimics group and mimics control group were plated in three wells each. On the day of the experiment, we used a pipette tip to make a “+” mark in the 6-well plate and washed the exfoliated cells with PBS buffer. Then we took a picture and recorded the width of the scratch, and added 2 mL of complete medium to each well for culture. After 24h, we captured a picture and recorded the width of the scratch again. Percentage of gap closure = (0h scratch width - 24h scratch width)/0h scratch width x 100%. For transwell migration and invasion assays, we took GC cells in the logarithmic growth phase to prepare the cell suspension with a cell density of 2 x 10⁶/mL using the serum-free medium. The upper layer of the transwell chamber with or without matrigel was then filled with 200 μL cell suspension, and the lower chamber was filled with 600 μL complete medium. After the incubation for 24 h in the incubator, we removed the liquid in the chamber’s upper layer and carefully cleaned off the non-penetrated cells with a cotton swab. The polycarbonate membrane of the chamber was stained with hematoxylin, and we counted the number of penetrated cells under an optical microscope. Our previous research has described the specific steps in detail (14).

NCI-N87 and HGC-27 cells were first harvested after 48 hours of transfection and the cell suspension was then centrifuged. Subsequently, we washed the cell pellet three times with ice-cold PBS and conducted the fixation with 70% ethanol. Afterward, we resuspended the cells in the PI-staining solution to perform the flow cytometry analysis.

The supernatant of the cell lysates was transferred to another clean centrifuge tube, and we used the BCA protein assay kit (ThermoFisher, USA) to detect the protein concentration. Then we performed gel electrophoresis (10% SDS-PAGE) and transferred the protein to a polyvinylidene fluoride membrane. The membrane was immersed in 5% skimmed milk for one hour and treated with the primary antibodies at 4°C with gentle tremors overnight. Next, the secondary antibody was added to block the membrane for two hours. We used the chemiluminescence method to develop and measure the gray band. Antibody dilutions and manufacturers are provided in Table 2 .

We obtained the five-week-old Balb/c female nude mice from the Shanghai Experimental Animal Center of the Chinese Academic of Sciences (Shanghai, China). The mice were randomly assigned to NS, tRF-Glu-TTC-027 NC, tRF-Glu-TTC-027 agomir groups (tRF-Glu-TTC-027 overexpression), and 1.2 x 10⁶ treated NCI-N87 cells were then subcutaneously injected into their right flanks. tRF-Glu-TTC-027 NC and tRF-Glu-TTC-027 agomir were designed by Ribobio company (Ribobio, Guangzhou, China). We measured the tumor sizes every three days. 40 days later, mice were sacrificed and the tumors were resected and sliced for IHC analysis and immunofluorescence assay. The tumor weight was measured and the volume was determined using the formula: 1/2 x (length x width²).

The paraffin-embedded tissue sections were deparaffinized and dehydrated. Then the sections were blocked with immunostaining blocking solution for 15 minutes at RT. Afterward, they were incubated in the Ki67 primary antibody ( Table 2 ) and secondary antibody for one hour, respectively. DAB substrate was used for staining. We counter-stained the cell nucleus with hematoxylin for two minutes and dehydrated the sections with ethanol. Images were captured by a microscope.

The tissue sections were infiltrated with PBS and rewarmed for 20 minutes. Then permeabilization was conducted using 0.5% Triton X-100 (prepared in PBS) at RT for 20 minutes. Next, we applied goat serum to block the tissue section for 30 minutes, and then the diluted primary antibody (Ki67) and fluorescent secondary antibody were used to incubate with each slide successively. Afterward, we immersed the slides with PBST and stained the specimens with DAPI. All the collected images were observed under a fluorescence microscope.

We applied SPSS 19.0 (IBM, Chicago, USA) and GraphPad Prism 6 (San Diego, CA, USA) to deal with the data. Data were presented as mean ± SD. The Chi-square test or Fisher’s exact test was applied to compare the composition ratio of two or more samples and analyze the correlation of two categorical variables. The Kaplan-Meier technique was used to evaluate the follow-up data, and the log-rank test was performed to assess the differences between groups. To compare the means of numerical variables, we conducted the Student’s t-test or one-way ANOVA test. P < 0.05 was regarded as statistically significant.

Differentially expressed tsRNAs were provided in Supplementary Tables 1 and 2 . We performed hierarchical clustering for tsRNAs differential expression data in high-throughput sequencing files. The results indicated that tsRNAs were significantly differentially expressed between gastric cancer and non-tumor adjacent tissue specimens (NATs) ( Figure 1A ). The result of PCA analysis shows a distinguishable tRF & tiRNA expression profiling among the GC and NATs samples. The figure is an overview of samples correlations using the R scatterplot3d package ( Figure 1B ). In terms of the results of the scatter plot, 142 up-regulated tRFs & tiRNAs, 150 down-regulated tRFs & tiRNAs, and 96 not differential expressed tRFs & tiRNAs were screened out. The Pearson correlation between GC and NATs was 0.871 ( Figure 1C ). The volcano plot is based on CPM values of tsRNAs, and we found that 69 up-regulated tRFs & tiRNAs, 42 down-regulated tRFs & tiRNAs, and 277 not differential tRFs & tiRNAs were uncovered, which was taken into analysis with the criteria of the p-value (<0.05) ( Figure 1D ).

We then predicted the target gene of tRF-Val-CAC-016, tRF-Glu-TTC-027, tRF-Glu-TTC-026, tRF-Ser-TGA-011, tiRNA-Pro-TGG-001, tRF-Ser-GCT-113, tiRNA-Val-CAC-001, tiRNA-His-GTG-001, tRF-Glu-TTC-017, and tiRNA-Asp-GTC-001 through GO and KEGG pathways. We conducted GO and KEGG analysis separately according to the expression level of the tsRNAs. Ten most statistically significant targets of GO and KEGG analysis of the down-regulated group (tRF-Val-CAC-016, tRF-Glu-TTC-027, tRF-Glu-TTC-026, tRF-Ser-TGA-011, tiRNA-Pro-TGG-001, tRF-Ser-GCT-113) are presented in Figures 2A, C , and the results of the up-regulated group (tiRNA-Val-CAC-001, tiRNA-His-GTG-001, tRF-Glu-TTC-017, tiRNA-Asp-GTC-001) are presented in Figures 2B, D . The most enriched biological process (BP) was nervous system development (GO: 0007399) in the down-regulated group and nucleic acid-templated transcription (GO: 0097659) in the up-regulated group. Concerning the CC analysis, it indicated that enrichment mainly occurred at intracellular (GO:0005622) in both groups. For MF, the analysis presented that the enriched function was protein binding (GO: 0005515) in the down-regulated group, and metal ion binding (GO: 0046872) in the up-regulated group ( Figures 2A, B ). In the KEGG pathways analysis, AGE-RAGE signaling pathway in diabetic complications (hsa04933), Small cell lung cancer (hsa05222), Proteoglycans in cancer (hsa05205), Focal adhesion (hsa04510), Measles (hsa05162), Protein processing in the endoplasmic reticulum (hsa04141), MAPK signaling pathway (hsa04010), Herpes simplex virus 1 infection (hsa05168), Regulation of actin cytoskeleton (hsa04810), and Adherens junction (hsa04520) were significantly enriched in the down-regulated group ( Figure 2C ). Endocrine resistance (hsa01522), Hedgehog signaling pathway (hsa04340), Wnt signaling pathway (hsa04310), Herpes simplex virus 1 infection (hsa05168), Pancreatic cancer (hsa05212), Mucin type O-glycan biosynthesis (hsa00512), Colorectal cancer (hsa05210), FoxO signaling pathway (hsa04068), Pathways in cancer (hsa05200), and Chronic myeloid leukemia (hsa05220) were meaningfully enriched in the up-regulated group ( Figure 2D ). We then searched TCGA and GEO databases to compare the bioinformatic analysis with present sequencing data, and we discovered that proliferation-related pathways were significantly enriched in KEGG analysis ( Figures 2E, F ). Meanwhile, the G2M checkpoint and p53 pathway in the GEO database ( Figures 2G, H ), DNA replication and cell cycle pathway in the TCGA database ( Figures 2I, J ) were significantly enriched in GSEA analysis. Above all, the MAPK signaling pathway was directly selected according to the bioinformatic analysis and the following biological assays of tRF-Glu-TTC-027.

We used the gel electrophoresis to further clarify the PCR product for tRF-Glu-TTC-027 ( Figure 3A ). The results of Sanger sequencing confirmed the validity of the primers and the authenticity of the PCR product sequence of the tRF-Glu-TTC-027 ( Figure 3B ). Next, we evaluated 35 pairs of tissues with RT-PCR, and tRF-Glu-TTC-027 was tested a low expression of great significance in GC compared with NATs, and so is in the detection of GC cells (NCI-N87, HGC-27) compared with GES-1 ( Figures 3C, D ). FISH assay confirmed that tRF-Glu-TTC-027 was most concentrated in the cytoplasm ( Figure 3E ). Meanwhile, designed tRF-Glu-TTC-027 mimics were capable of efficiently up-regulating the expression of tRF-Glu-TTC-027 ( Figure 3F ). Subsequently, we uncovered that the expression level of tRF-Glu-TTC-027 was significantly associated with tumor size and histological grade in terms of clinicopathological studies ( Table 3 ). Through the analysis of the PCR data of 35 pairs of GC tissues and the follow-up data of corresponding patients, we found that tRF-Glu-TTC-027 was not significantly related to the prognosis of GC patients ( Figure 3G ).

We transferred tRF-Glu-TTC-027 mimics, mimics control, or p38 MAPK pathway inhibitor (p38 MAPK-IN as the positive control) into the GC cells to discover the biological characteristics of tRF-Glu-TTC-027 in GC cell lines. The CCK-8 assay indicated that tRF-Glu-TTC-027 and p38 MAPK-IN could effectively decline the GC proliferation of great significance ( Figures 4A, B ). In addition, tRF-Glu-TTC-027 was capable of suppressing the migration and invasion capacities of GC cell lines ( Figures 4C–E ). Then the wound healing assay was applied to further demonstrate the effect of tRF-Glu-TTC-027 and p38 MAPK-IN on the GC cells, and the result was consistent with the previous assays ( Figures 4F, G ). Moreover, EdU assays also showed that tRF-Glu-TTC-027 and p38 MAPK-IN formed less DNA replication level than the control group, which was reflected in the aspect of the fluorescence concentration of EdU ( Figures 4H, I ). To differentiate and analyze the impact of tRF-Glu-TTC-027 on the GC cell cycle, we transfected cells with tRF-Glu-TTC-027 mimics, mimics control, or p38 MAPK-IN and tested the distribution of cell cycle with flow cytometry. Compared with the mimics control group, the ratio in the S phase of the tRF-Glu-TTC-027 mimics group significantly increased from 19.45% to 31.07%, p38 MAPK-IN group increased from 19.45% to 37.77%, but in the G2 phase of tRF-Glu-TTC-027 mimics group significantly decreased from 29.98% to 19.19%, p38 MAPK-IN group decreased from 29.98% to 11.26% in HGC-27 cells. Analogously, the cells in the G1 phase of the tRF-Glu-TTC-027 mimics group significantly declined from 51.12% to 43.95%, p38 MAPK-IN group declined from 51.12% to 44.23%, the cells in the S phase of the tRF-Glu-TTC-027 mimics group increased from 21.63% to 37.71%, p38 MAPK-IN group increased from 21.63% to 33.06%, whereas the cells in the G2 phase decreased from 27.25% to 18.35%, p38 MAPK-IN group decreased from 27.25% to 22.71% in NCI-N87 cells ( Figures 4J–L ). Therefore, tRF-Glu-TTC-027 and p38 MAPK-IN can arrest the cell cycle of NCI-N87 and HGC-27 at the S phase to some extent. Then the obvious change in the number of cell clones in the colony formation assays fully confirmed the inhibitory effect of tRF-Glu-TTC-027 and p38 MAPK-IN on GC ( Figures 4M, N ).

To further elucidate the regulative function of tRF-Glu-TTC-027 on GC, we undertook the rescue assays with tRF-Glu-TTC-027 inhibitor group, tRF-Glu-TTC-027 inhibitor + p38 MAPK-IN group, p38 MAPK-IN group, and control group. In the CCK8 assay, tRF-Glu-TTC-027 inhibitor facilitated the proliferation of HGC-27 and NCI-N87, and this stimulative effect could be reversed by p38 MAPK-IN ( Figures 5A, B ). We demonstrated that tRF-Glu-TTC-027 inhibitor could promote the DNA replication in GC cell lines, while p38 MAPK-IN significantly suppressed the EdU assimilation and could reverse the stimulative effect of tRF-Glu-TTC-027 inhibitor during the proliferative process in EdU assay ( Figures 5C–F ). Then flow cytometry was used to analyze the cell cycle, and we found that p38 MAPK-IN imposed the inhibitory effect on the GC cells to cause the cell cycle arrest and this was able to be relieved by tRF-Glu-TTC-027 inhibitor ( Figures 5G–I ).

To figure out the specific proteins influenced by tRF-Glu-TTC-027, we adopted immunoblotting to test the expression levels of proteins in the MAPK signaling pathway. And we found that tRF-Glu-TTC-027 inhibitor could upregulate the expression of p-p38 and c-Myc in HGC-27, and p-ERK, p-p38, and c-Myc in NCI-N87. p38 MAPK-IN could suppress the expression of p-ERK, p-p38, Elk1, p-Elk1, c-Myc, cyclinD1 in HGC-27 and NCI-N87 ( Figures 6A, B ). tRF-Glu-TTC-027 mimics could inhibit the expression of p-ERK, p-JNK, p-p38, c-Myc, CDK2, p-ELK1 in NCI-N87, and p38, p-p38, c-Myc, p-ELK1 in HGC-27 ( Figure 6C ). To further investigate the underlying mechanism that tRF-Glu-TTC-027 influenced the MAPK signaling pathway, we discovered 26 possible target genes by taking the intersection of tRF-Glu-TTC-027 target genes and genes in the MAPK signaling pathway ( Figure 7A ). Then we analyzed the TCGA database and the Kaplan Meier plotter website to screen the significant targets. TGFB2 and ELK4 were selected as their 3’ UTR owned the target sites with tRF-Glu-TTC-027 ( Figures 7B, C ), and were highly expressed in GC compared with NATs ( Figures 7D, E ). Meanwhile, TGFB2 and ELK4 were associated with poor prognosis ( Figures 7F, G ). Finally, PCR and immunoblotting were conducted, and we found that TGFB2 was regulated by tRF-Glu-TTC-027 in HGC-27 and NCI-N87 ( Figures 7H, I ).

The in vivo effect of tRF-Glu-TTC-027 was evaluated using NCI-N87 xenografts. The results showed that tRF-Glu-TTC-027 significantly reduced the capacity of the tumor growth ( Figures 8A, B ). However, no significant difference was observed among these three groups concerning the bodyweight of the mice ( Figure 8C ). The tumor volume in the NS group increased from 106.733 ± 2.400 mm³ to 659.044 ± 136.553 mm³, in the tRF-Glu-TTC-027 NC group from 106.929 ± 3.263 mm³ to 676.301 ± 130.231 mm³, and in the tRF-Glu-TTC-027 agomir group from 107.618 ± 8.491 mm³ to 324.807 ± 45.991 mm³ ( Figure 8D ). The volume of the explant tumors in the tRF-Glu-TTC-027 agomir group was smaller than those in the NS and tRF-Glu-TTC-027 NC group at the endpoint of the experiment significantly ( Figure 8B ). We then conducted the immunofluorescence assay and IHC assay for Ki67 to further confirm the proliferative activity of xenografts. As shown in Figure 8E , the fluorescence density of Ki67 decreased significantly in the tRF-Glu-TTC-027 agomir group compared with NS and tRF-Glu-TTC-027 NC group, and the result was consistent with the IHC assay.