Upregulated LINC01088 facilitates malignant phenotypes and immune escape of colorectal cancer by regulating microRNAs/G3BP1/PD-L1 axis

Long intergenic non-coding RNA LINC01088 is a newly discovered long non-coding RNA (lncRNA). Its biological function in colorectal cancer (CRC) remains unknown. Here, 36 paired CRC and para-cancerous tissues were collected. In vitro, fluorescence in situ hybridization (FISH) assay, qPCR, western blotting analysis and cellular functional experiments, RNA immunoprecipitation (RIP) assay and dual-luciferase reporter system analysis were performed. In vivo, xenograft tumor mouse models were generated. Besides, patient-derived intestinal organoid (PDO) was generated ex vivo. We found that LINC01088 was significantly upregulated in colorectal cancer tissues and CRC cell lines compared to adjacent normal tissues and colonic epithelial cells. High LINC01088 levels were correlated with adverse outcomes in patients with CRC. LINC01088 was mainly located in the cytoplasm. LINC01088 knockdown suppressed the proliferation, migration, invasion, and immune escape of colorectal cancer cells. Mechanistically, LINC01088 bound directly to miR-548b-5p and miR-548c-5p that were significantly upregulated Ras GTPase-activating protein-binding proteins 1 (G3BP1) and programmed death ligand 1 (PD-L1) expression, altering CRC cell phenotypes. In mouse xenograft models, LINC01088 knockdown restrained CRC tumor growth and lung metastasis. Furthermore, G3BP1 overexpression reversed LINC01088-knockdown-mediated inhibitory effects on tumor growth. Notably, LINC01088 knockdown downregulated PD-L1 expression, while G3BP1 overexpression restored PD-L1 expression in xenograft tumors. Besides, LINC01088 knockdown repressed CRC organoid growth ex vivo. Overall, these findings suggested that LINC01088 directly targeted miR-548b-5p and miR-548c-5p, promoting G3BP1 and PD-L1 expression, which facilitated colorectal cancer progression and immune escape.


Introduction
Colorectal cancer (CRC) remains a serious threat to human health globally, with nearly two million new cases and approximately one million deaths caused each year (Dekker et al. 2019;Sung et al. 2021). Although the early diagnosis and clinical treatment of CRC have been greatly advanced in the last decades (Anania et al. 2019;Van Cutsem et al. 2014), it continues to be a significant burden on the individual and society. Recently, molecular targeted therapy and immunotherapy are considered to be scientific breakthroughs in treatment of tumors including colorectal cancer (Franke et al. 2019;Nappi et al. 2018). Nevertheless, not all 1 3 patients benefit from it. There are still unknown mechanisms in colorectal cancer progression.
Long non-coding RNAs (lncRNAs) with a length of more than 200 nucleotides are a class of non-coding RNAs discovered in recent years, including antisense long non-coding RNAs (antisense lncRNAs) and long intergenic non-coding RNAs (lincRNAs) (Quinn and Chang 2016;Ransohoff et al. 2018). Previously, these non-coding RNAs were considered as a large class of "junk" RNA produced during transcription, until recent studies have identified that they have important biological functions (Palazzo and Koonin 2020;Quinn and Chang 2016). And growing evidences support that long non-coding RNAs engage in a variety of physiological and pathological processes including organismal development and cancer progression (Dykes and Emanueli 2017;Fang and Fullwood 2016;Gutschner and Diederichs 2012;Sanchez Calle et al. 2018;Schmitz et al. 2016). We recently reported that several lncRNAs are implicated in colorectal cancer progression (Li et al. 2021a;Xu et al. 2019;Zhou et al. 2021), among which downregulated LncRNA SATB2-AS1 expression is associated with poor outcome in CRC patients. LncRNA SATB2-AS1 in the nucleus cis-activates SATB2 (specific AT-rich binding protein 2) transcription via regulation of histone H3 lysine 4 trimethylation (H3K4me3) deposition and DNA demethylation in the SATB2 promoter region, thereby affecting tumor immune microenvironment (Xu et al. 2019). LincRNAs belonging to lncRNAs do not overlap protein-coding transcripts. LincRNAs can attenuate protein, mRNA and miRNA activity via their sequestration (Ransohoff et al. 2018). For example, a recent study has reported that long intergenic non-coding RNA Linc00284 localized in the cytoplasm is highly expressed in CRC tissues and associated with tumor metastasis. Linc00284 regulates miR-27a/c-Met axis, resulting in malignant phenotypes of CRC cells (You et al. 2021). The role of long intergenic non-coding RNAs in the development and progression of various cancers including colorectal cancer is becoming clearer with the progressive research (Castellano et al. 2016;Li et al. 2021b), but with the increasing number of newly discovered lincRNAs, it is essential to understand precise mechanisms of these novel lincRNAs in cancer.
LINC01088 located on Chromosome 4 at q21.21 is a newly discovered long intergenic non-coding RNA. A bioinformatic analysis of two GEO datasets (GSE28619 and GSE143318) identified that LINC01088 was downregulated in liver tissues from patients with acute alcoholic hepatitis compared to samples derived from donor livers (Yan et al. 2021). However, another analysis of TCGA database demonstrated that LINC01088 was significant highly expressed risk candidate in primary lung squamous cell carcinoma compared with adjacent normal tissues and predicts poor prognosis (Liu et al. 2019a). In addition, work in non-small cell lung cancer (NSCLC) has indicated that LINC01088 was overexpressed in NSCLC tissues and cell lines, where LINC01088 promoted cancer cell proliferation by binding with EZH2 to repress p21 (Liu et al. 2020). Conversely, in ovarian cancer (OC), LINC01088 expression was markedly lower in OC tissues in comparison to adjacent noncancerous tissues, which is an independent predictor for overall survival (Ai et al. 2018;Zhang et al. 2018). These evidences suggest that the expression pattern of LINC01088 can be different in different types of cancers. In our previous screen, we found that LINC01088 was upregulated in CRC tissues and associated with multiple clinicopathological features in patients with colorectal cancer. Hence, we investigated the role of LINC01088 in CRC with the aim of gaining a deeper understanding the pathogenesis of colorectal cancer progression. It may provide a potential target for CRC diagnosis and treatment.

Colorectal cancer tissues
Thirty-six cancer tissues and paired tissues were collected from CRC patients with complete pathological and clinical data who were hospitalized and treated surgically in Nanjing First Hospital from January 2015 to January 2020. Adjacent tissue was taken from a distance of more than 5 cm from cancerous tissue, and intraoperative pathological slices were pathologically confirmed to be free of cancer. There were 21 males and 15 females, aged (61.6 ± 7.7) years, with an age range of 51-72 years. Inclusion criteria: ① surgically resected specimens were pathologically confirmed; ② all patients with colorectal cancer did not receive anti-cancer treatment such as chemotherapy, radiotherapy, immunotherapy, and targeted therapy before surgery; ③ clinical and pathological data of CRC patients were complete; and ④ all patients gave informed consent. Besides, CRC patients with other malignant tumors, systemic infections, immunodeficiency, immune diseases, coagulation dysfunction, serious heart, liver, or kidney diseases were excluded. All patients signed informed consent form prior to surgery, and the study was approved by the Ethics Committee of Nanjing First Hospital. Table 1 lists patient characteristics.

Nucleo-cytoplasmic separation experiment
Nucleo-cytoplasmic fractions were isolated using PARIS™ Kit following manufacturer instructions. Briefly, adherent cells (10 7 cells per experiment) were harvested until they were grown to ~ 90% confluence, transferred to 1.5 mL microfuge tubes, and centrifuged at 180 × g for 4 min at 4 °C. Next, the supernatant was discarded. 500 μl of icecold Cell Fractionation Buffer was added. Tubes were then placed on ice for 7 min. After centrifugation at 500g for 4 min at 4 °C, the supernatant (cytoplasmic fraction) was transferred to a new RNase-free 1.5 mL EP tube. The pellet contained nuclear fraction. Total nuclear or cytoplasmic RNA was extracted, respectively. RNA samples were stored at − 80 °C or kept on ice when in use.

Cell counting kit-8 (CCK-8) assay
The cells were inoculated into 96-well dishes at a density of 3 × 10 4 cells per well, and five replicate wells were set up in each group. After 24 h, 48 h, 72 h and 96 h of incubation, respectively, CCK-8 reagent (Sigma-Aldrich, cat no.96992) with a volume fraction of 10% was added to the cells following incubation for 2 h at 37 °C. The plates were placed on a shaker for 5 min, and absorbance values were measured at 490 nm using Microplate reader (Bio-Rad). Growth curve was plotted with time as the horizontal coordinate and absorbance value as the vertical coordinate.

Colony-formation assay
The cells (500 cells per well) were inoculated into 6-cm culture dishes. After 2 weeks of incubation at 37 °C, cell colonies were observed and then culture medium was discarded. Adherent cells were gently washed twice with PBS (Gibco) and fixed with 4% paraformaldehyde for 15 min at

Cell cycle analysis
Cell cycle was evaluated by flow cytometry. The cells in logarithmic growth phase were digested using EDTA-free trypsin (Gibco) and washed three times with pre-chilled PBS. Thereafter, 400 μl of Binding Buffer and 10 μl of PI were added sequentially to cell suspension, following reaction for 30 min under protection from light at RT. Cycle distribution was assessed using CytoFlex Flow Cytometer (Beckman Coulter).

Transwell migration and invasion assays
For transwell invasion assay, Matrigel (Corning, cat no.356234) was removed from the − 20 °C refrigerator and placed in an ice-water mixture to dissolve. All tips and microfuge tubes used in the experiments were pre-chilled in a 4 °C refrigerator. Matrigel and serum-free basal medium were mixed thoroughly at a 1:3 ratio and then added into transwell upper chambers (Costar, 8-μm pore size) for incubation for 1 h at 37 °C. In addition, cell concentration was adjusted to 5 × 10 5 cells/ml. 200 μl of the cell suspension was slowly added to the upper chambers and the lower chambers contained 500 μl of complete culture medium. Twenty-four hours later, the non-invasive cells in transwell upper chamber were wiped off with wet cotton swabs. Next, the filter was fixed with 4% formaldehyde for 10 min, stained with 0.5% crystal violet for 20 min, rinsed with PBS and air dried. The number of invasive cells was observed microscopically, and 10 randomly chosen fields of view were imaged and counted. Transwell chambers pre-coated without Matrigel were used for transwell migration assay and workflow was followed as indicated above.

Propidium iodide (PI) staining
After 48 h of co-culture, the cells were incubated with PI staining solution (Sigma) for 5 min. Next, PI-positive cells were observed and counted under a fluorescent microscope (Nikon), while the total number of cells was observed under white light. The percentage of PI-positive cells was calculated.

Dual-luciferase reporter assay
The pmirGLO dual-luciferase vectors and dual-luciferase reporter system (Promega) were used for evaluating the interaction between LINC01088 and microRNAs according to the manufacturer's instructions. Predicted binding sites were analyzed by https:// starb ase. sysu. edu. cn/ starb ase2/ and http:// starb ase. sysu. edu. cn/. The desired primer sequences were cloned into luciferase reporter plasmid as described in the instruction manual. Empty vector plasmid was as the control. 293 T cells were seeded into 24-well plates and allowed to grow to 80% confluence. Reporter plasmid was co-transfected with the expression plasmid into 293 T cells using Lipofectamine ™ 2000 Reagent (Invitrogen). Seventytwo hours later, the proteins were extracted and Dual-Luciferase Reporter Assay System Kit (Promega) was used to test luciferase activity.

RNA immunoprecipitation (RIP) assay
RIP assay was performed using Imprint ® RNA Immunoprecipitation Kit according to the manufacturer's instructions (Sigma-Aldrich). Briefly, CRC cells with LINC01088 knockdown in logarithmic growth phase were harvested and resuspended using RIP Lysis Buffer, mixed thoroughly, and set aside on ice. After incubation for 30 min, the samples were centrifuged at 2500g for 10 min at 4 °C. Besides, magnetic beads conjugated anti-Ago antibody or IgG for pre-clearance were washed with RIP Wash Buffer (provided with the kit). 900 µl of RIP Immunoprecipitation Buffer was added to each tube and mixed with 100 µl supernatant following incubation overnight at 4 °C with rotation. After precipitating the magnetic beads, removed the supernatant, added 500 µl of RIP Wash Buffer to wash the beads, vortexed, and then collected the precipitate. 10 µl of the cell lysate supernatant was taken as "Input" and store it temporarily at − 80 °C. Next, 150 µl of Proteinase K Buffer was added to the precipitate obtained in the previous step. In addition, RIP Wash Buffer, 10% SDS, and Proteinase K mixture were added to the "Input" after thawing, followed by incubation for 30 min at 55 °C with shaking to digest

Animal experiments
To test proliferative ability of CRC cells in vivo, tumor cells in logarithmic growth phase were resuspended in PBS and cell concentration was adjusted to 1 × 10 7 cells/ml. BALB/c female nude mice of age 5 weeks and weighing about 20 g were acclimated for one week prior to experiments. 100 μl of LoVo cell suspension was subcutaneously injected into the right axilla of nude mouse (N = 5 or 6 mice per group). Tumor diameters of xenograft mice were measured weekly after inoculation. Tumor volume was calculated by the formula: tumor volume = width 2 × length/2 Mice were killed under anesthesia 35 or 42 days after inoculation, and tumor tissues were photographed and weighed after stripping and removing the surrounding connective tissues. Besides, to measure lung metastatic capability of LINC01088-knockdown Caco2 cells in vivo, cell concentration of each group was adjusted to 1 × 10 6 cells/ml. 50 μl of the cell suspension was injected into the mice through the lateral tail vein of the nude mice (N = 6 mice per group) and lung metastatic nodules were examined at the desired time point after deep anesthesia with isoflurane. Lung tissues were collected for HE staining. All animal experiments were conducted under the approval of Animal Care and Use Committee of Nanjing First Hospital.

Human colorectal cancer tissue-derived organoids
This study was approved by the Ethics Committee of Nanjing First Hospital. The patient (female, 58 years old, sigmoid colon cancer) signed informed consent form prior to surgery. Fresh colorectal cancer tissues were collected after surgical resection and maintained in ice-cold DMEM/F-12 medium with 15 mM HEPES (Gibco) within 2 h. Tissues were cut into 1-3 mm 3 pieces, washed ten times using prechilled PBS, and digested with Gentle Cell Dissociation Reagent (GCDR) (Stemcell Technologies, cat no.07174) for 50 min at 37 °C with shaking. After centrifugation at 300 × g for 5 min, the supernatant was discarded. The tissues were resuspended using DMEM/F-12 medium with 15 mM HEPES and 2% BSA (Solarbio Life Science, cat no.9048-46-8) and filtered through 70 μm strainers. The number of intestinal crypts were counted. Next, Matrigel Matrix Growth Factor Reduced (GFR), Red-free (Corning, cat no.356231) was diluted with DMEM/F-12 complete medium (as indicated above) at a 1:1 ratio and the crypts were resuspended. Matrigel containing intestinal crypts was dropped onto the center of a pre-warmed 24-well plate without medium (50 μl/well). The plate was placed in an incubator at 37 ℃ for 10 min. Subsequently, human organoid medium IntestiCult OGM (Stemcell Technologies, cat no. A8010) was added and was changed every 2-3 days. Passaging was performed 8 days after culturing.

Isolation, characterization and cultivation of human CD8 + T cells
Fresh blood from a healthy donor was transferred into a blood draw tube containing heparin anticoagulant and mixed upside down. PBS supplemented with 2% FBS (PBST) was added to dilute the blood at a 1:1 ratio.

Bioinformatic analysis
To explore downstream targets of LINC01088, microRNAs expression in CRC tissues were extracted from The Tumor Cancer Genome Atlas (TCGA) dataset. In addition, the relationship between tumor G3BP1 expression and immune infiltration in colorectal cancer was analyzed using Tumor IMmune Estimation Resource (TIMER) (https:// cistr ome. shiny apps. io/ timer/) dataset (Li et al. , 2017.

Statistical analysis
Statistical analysis was performed using SPSS 21.0 software. Graphs were made using the GraphPad Prism 8 (GraphPad Software, San Diego, CA). All data were expressed using mean ± standard deviation and data conforming to normal distribution were compared between two groups using twotailed Student t test. For multi-group comparisons, statistical significance was determined by one-way analysis of variance (ANOVA) followed by Fisher's Least Significant Difference (LSD) test. Kaplan-Meier method was used to analyze the relationship between gene expression and prognosis of CRC patients. A P < 0.05 was considered as statistically significant. *P < 0.05, **P < 0.01, ***P < 0.001, and NS: not significant.

LINC01088 is overexpressed in colorectal cancer and associated with adverse outcome
We firstly collected 36 paired CRC tissues and adjacent normal tissues. qPCR results confirmed that LINC01088 expression was significantly elevated in colorectal cancer tumor tissues (Fig. 1A). Clinicopathological analysis showed that LINC01088 expression was associated with TNM stage, metastasis, and recurrence of patients with colorectal cancer, where LINC01088 expression was higher in Stage III/IV tissues than in Stage I/II tissues (Fig. 1A). Survival analysis revealed that CRC patients with high LINC01088 levels had worse overall survival compared with CRC patients with lower levels (Fig. 1B). Besides, LINC01088 expression was also markedly higher in CRC cells compared to normal colorectal epithelial cells and LINC01088 was located in the cytoplasm ang nucleus of tumor cells (Fig. 1C, D). Consistently, fluorescence in situ hybridization (FISH) assay further identified its localization (Fig. 1E).

LINC01088 knockdown significantly reduces proliferative ability of colorectal cancer cells and patient-derived organoids
Considering high level of LINC01088 in Caco2 and LoVo cells, these two cell lines were used to construct CRC cells with stable LINC01088 knockdown by lentiviral infection. qPCR analysis validated that both shRNAs (sh1-01088 and sh2-01088) lentivirus were able to significantly reduce LINC01088 expression levels ( Fig. 2A). Cell proliferation assay and colony formation assay confirmed that proliferative and colony-forming abilities were significantly attenuated after silencing of LINC01088 in CRC cells ( Fig. 2B-D). It seemed that the higher LINC01088 silencing efficiency, the lower CRC cell proliferative and colony-forming capabilities. Results of cell cycle showed that LINC01088 knockdown contributed to a significant increase in the proportion of CRC cells in G0/G1 phase and a decline in G2/M phase (Fig. 2E, F). Consistent with the in vitro results, subcutaneous xenograft experiments in nude mice further proved that LINC01088 knockdown significantly inhibited CRC cell growth (Fig. 2G-I).

Fig. 4 Effects of LINC01088 knockdown on the migration and invasion, lung metastatic ability of CRC cells in vitro and in vivo. A, B
Migratory and invasive capabilities of LINC01088 knockdown (A) Caco2 and (B) LoVo cells were tested by transwell assays. C Schematic illustration of animal experiments. D Representative images and quantification of lung metastatic nodules in xenograft mouse model. E HE staining of lung metastatic nodules. F Survival curves of CRC-bearing nude mice. G, H After co-culturing for 12 h, LINC01088-knockdown CRC cells and CD8 + T cells were imaged under the bright field. I, J After co-culturing for 12 h, surviving CRC cells were stained with crystal violet. K, L After co-culturing for 12 h, apoptotic CRC cells were stained by PI. Data were expressed using mean ± standard deviation. Comparison between two groups using two-tailed Student t test. For multi-group comparisons, statistical significance was determined by one-way ANOVA followed by Fisher's LSD test. *P < 0.05; **P < 0.01; ***P < 0.001. HE hematoxylin-eosin staining, PI propidium iodide, CRC colorectal cancer (Fig. 3B). Knockdown efficiency of organoids transduced with lentivirus was measured by visualizing green fluorescence. Under white light conditions, it could be seen that the LINC01088 knockdown resulted in smaller diameter and more cellular debris (Fig. 3C). PI staining suggested that more PI-positive cells were found in the organoid after LINC01088 knockdown (Fig. 3D).

LINC01088 knockdown suppresses migration, invasion, and immune escape of CRC cells
Next, we investigated the role of LINC01088 in migration and invasion of CRC cells. Transwell assays revealed that LINC01088 knockdown significantly reduced migration and invasion of Caco2 and LoVo cells (Fig. 4A, B). The higher LINC01088 silencing efficiency, the lower migratory and invasive capacities of the cells. This result was further reproduced in xenograft mouse model. Specially, the number and the lesion surface of the lung metastasis nodules are significantly reduced in LINC01088-knockdown groups (Fig. 4C-E). Survival analysis showed that silencing of LINC01088 significantly prolonged the survival of nude mice (Fig. 4F).
Given that tumor immune escape plays an important role in cancer progression, CRC cells were co-cultured with activated human CD8 + T cells for 12 h. Results of white light image and crystal violet staining showed that the density of adherent CRC cells with LINC01088 silencing was markedly reduced compared to the control, which was dependent on the number of CD8 + T cells (Fig. 4G-J). It was suggested that LINC01088-knockdown CRC cells were more sensitive to CD8 + T cell-mediated tumor killing relative to microRNAs that interacted directly with LINC01088 in Caco2 cells. Data were expressed using mean ± standard deviation. Comparison between two groups using two-tailed Student t test. For multi-group comparisons, statistical significance was determined by one-way ANOVA followed by Fisher's LSD test. *P < 0.05. RIP RNA immunoprecipitation assay Fig. 6 miR-548b-5p and miR-548c-5p are the main direct downstream targets of LINC01088 in colorectal cancer. A Cell viability of LINC01088-knockdown Caco2 cells transfected with microRNA inhibitor for 48 h. B Representative images and statistical analysis of the migration and invasion of LINC01088-knockdown Caco2 cells transfected with microRNA inhibitor. C After co-culturing for 12 h, LINC01088-knockdown Caco2 cells transfected with microRNA inhibitor were stained with crystal violet. D Two microRNAs were determined according to the Venn diagram. E, G Predicted binding sites of (E) miR-548b-5p or (G) miR-548c-5p on LINC01088 sequence. F, H The interaction between LINC01088 and (F) miR-548b-5p or (H) miR-548c-5p was determined by dual-luciferase reporter system analysis. Data were expressed using mean ± standard deviation. Comparisons among groups was analyzed by oneway ANOVA followed by Fisher's LSD test. *P < 0.05; **P < 0.01; ***P < 0.001; NS, not significant Fig. 7 G3BP1 is the common direct downstream target of miR-548b-5p and miR-548c-5p in colorectal cancer. A Screening process of the common downstream targets of miR-548b-5p and miR-548c-5p. B, C G3BP1 protein expression in Caco2 cells after transfection with (B) miR-548b-5p or (C) miR-548c-5p mimics for 48 h. D, E Predicted binding sites between G3BP1 mRNA and (D) miR-548b-5p or (E) miR-548c-5p. F, G The interaction between G3BP1 mRNA and (F) miR-548b-5p or (G) miR-548c-5p was determined by dual-luciferase reporter system analysis. Data were expressed using mean ± standard deviation. Comparison between two groups using two-tailed Student t test. Comparisons among groups was analyzed by one-way ANOVA followed by Fisher's LSD test. *P < 0.05; **P < 0.01; ***P < 0.001; NS, not significant . Data were expressed using mean ± standard deviation. Comparison between two groups using two-tailed Student t test. Comparisons among groups was analyzed by one-way ANOVA followed by Fisher's LSD test. **P < 0.01; ***P < 0.001 WT CRC cells. Similar results were observed in PI staining (Fig. 4K, L).

LINC01088 exerts its biological activity by directly binding to intracellular microRNAs
Long non-coding RNAs acts as competing endogenous RNAs (ceRNAs) to exhibit its activity when they localized in the cytoplasm. In this study, LINC01088 was mainly localized in the cytoplasm. Hence, we suspected that LINC01088 exerts its biological function via lncRNA-microRNA (miRNA)-messenger RNA (mRNA)-ceRNA networks. Therefore, miRNAs that were significantly downregulated in colorectal cancer and predicted to bind directly to LINC01088 were analyzed by bioinformatics methods. A list of 35 miRNAs was obtained (Fig. 5A). We further confirmed that 20 miRNAs were able to be directly downregulated by LINC01088 in colorectal cancer by RIP assay (Fig. 5B). The expression levels of these 20 miRNAs were examined in LoVo cells with LINC01088 knockdown, and miRNAs that were upregulated after silencing LINC01088 were taken into consideration (Fig. 5C, D). Finally, 9 miR-NAs were selected as targets for further analysis (Fig. 5E).

LINC01088 targets miR-548b-5p and miR-548c-5p, regulating G3BP1 expression
MicroRNAs usually regulate mRNA expression by directly binding to them. By bioinformatics analysis, we searched for downstream target genes jointly predicted by miR-548b-5p and miR-548c-5p in multiple databases. miR-548b-5p had 9 target genes and miR-548c-5p had 41 target genes. By taking the intersection of these two sets of genes, we obtained a list of genes including BRWD1, G3BP1, CCDC47, DDIT4, RORA, UE4A, LY75, TRA2B (Fig. 7A). Combining survival and expression analysis in colorectal cancer, we concluded that Ras-GTPase-activating protein SH3 domain-binding protein 1 (G3BP1) may be a key downstream target of LINC01088/microRNAs axis. G3BP1 was proved to be overexpressed in a variety of cancers and targeting G3BP1 exhibited anti-tumor activity (Zhang et al. 2019). Further analysis from TCGA and TIMER databases revealed that in most human cancers including colorectal cancer G3BP1 expression is upregulated and associated with immune infiltration in colorectal cancer (Fig. S1). Next, we explored G3BP1 protein expression in LINC01088-knockdown LoVo cells. Western blotting analysis showed that treatment with miR-548b-5p or miR-548c-5p mimics contributed to a significant decrease in G3BP1 protein levels in CRC cells (Fig. 7B, C). Next, predicted binding sites between G3BP1 mRNA and miR-548b-5p (Fig. 7D), G3BP1 mRNA and miR-548c-5p (Fig. 7E) were detected by dual-luciferase reporter assay. Data revealed that three binding sites between G3BP1 mRNA and miR-548b-5p are available (Fig. 7F) and two binding sites between G3BP1 mRNA and miR-548c-5p were identified (Fig. 7G).

LINC01088/microRNAs/G3BP1/PD-L1 axis promotes colorectal cancer progression
As previously described, our results suggested that LINC01088 knockdown repressed CRC cell proliferation, migration, and invasion. Subsequently, we implemented rescue experiments to verify effects of LINC01088/microR-NAs/G3BP1 axis on malignant phenotypes of colorectal cancer. LINC01088 knockdown resulted in decreased G3BP1 expression at the transcriptional nor the translational levels ( Fig. 8A, B), reduced cell viability, migratory and invasive capabilities, and weakened CD8 + T cell-mediated tumor killing activity, whereas overexpression of G3BP1 could partially reverse G3BP1 expression and phenotypes caused by LINC01088 knockdown in Caco2 cells (Fig. 8C-G). Additionally, when LINC01088-knockdown Caco2 cells were co-cultured with activated CD8 + T cells for 12 h, G3BP1 overexpression reduced lactate dehydrogenase (LDH) levels that were significantly elevated as LINC01088 knockdown (Fig. S2). Specially, LINC01088 knockdown caused a significant decrease in PD-L1 mRNA expression and cell-surface PD-L1 expression in Caco2 cells (Fig. 8H, I). Further, tumor xenograft in nude mice revealed that G3BP1 overexpression restored LINC01088 knockdown-induced suppression of cell proliferation (Fig. 8J, K). Remarkably, LINC01088 knockdown downregulated PD-L1 expression, while G3BP1 overexpression could reverse PD-L1 expression in xenograft tumors. It implied that G3BP1 regulated PD-L1 expression in CRC (Fig. 8L). These findings suggested that upregulation of LINC01088 promotes malignant phenotypes and immune escape of colorectal cancer by regulating microR-NAs/G3BP1/PD-L1 axis (Fig. 9).

Discussion
Colorectal cancer is a great threat to human health globally. Explorations of the pathogenesis of colorectal cancer as well as clinical treatments have achieved significant progress in recent decades (Dekker et al. 2019;Nguyen et al. 2020). However, considering that long-term survival and quality of life of patients with colorectal cancer have been dismal, further researches on specific mechanisms are deserved to find novel therapeutic strategies. In the present study, we found that LINC01088 was significantly upregulated in colorectal cancer tissues and cells. LINC01088 promoted colorectal cancer progression and immune escape by mediating micro-RNAs/G3BP1/PD-L1 axis.
LincRNA is one of the long non-coding RNAs (Ransohoff et al. 2018), and mounting evidence has confirmed that dysregulated lincRNA expression in tumor tissues plays an important role in multiple disease (Ransohoff et al. 2018;Ulitsky and Bartel 2013). Previous studies found that LINC01088 inhibited trophoblast cell function and led to recurrent miscarriages by activating MAPK signaling pathway (Zhao et al. 2021). Work in lung squamous cell carcinoma (LSqCC) demonstrated that through analyzing transcriptome profiling of 1771 lincRNAs in the TCGA database from 549 samples of 501 LSqCC patients, 10 lin-cRNAs including LINC01088 were confirmed to be significant highly expressed risk candidates and were associated with poor prognosis (Liu et al. 2019a). Another study has identified that LINC01088 directly binds to EZH2 inhibiting the expression of p21, a well-known oncogenic factor, thereby suppressing proliferation of lung cancer cells (Liu et al. 2020). Similarly, our data demonstrated that elevated LINC01088 expression facilitated proliferation of colorectal cancer cells, contributing to colorectal cancer progression. Mechanistically, LINC01088 acting as a pro-cancer factor regulated the proliferative capacity of colorectal cancer cells by adsorbing a series of intracellular microRNAs. Conversely, there is also evidence that LINC01088 expression levels are significantly lower in ovarian cancer tissues relative to ovarian epithelial tissues ) and low LINC01088 levels are correlated with FIGO staging, grade and metastasis (Ai et al. 2018). LINC01088/miR-24-1-5p/p21-activated kinase 4 (PAK4) axis may be responsible for the oncogenic role of LINC01088 in ovarian epithelium . LINC01088 plays different roles in diverse cancers.
Invasive capacity (Novikov et al. 2021;Yeung and Yang 2017) and immune escape (Liu and Cao 2016;Picard et al. 2020) of cells are important factors for tumor development and metastasis. Our study confirmed that LINC01088 knockdown significantly reduced migratory and invasive capabilities and inhibited immune escape of colorectal cancer cells.
Specifically, LINC01088 positively regulates G3BP1 expression by directly binding to and negatively regulating miR-548b-5p and miR-548c-5p, two oncogenic factors in a variety of cancers including colorectal cancer (Pan et al. 2016;Wang et al. 2020;Xu et al. 2020). G3BP1 is an oncogenic gene and targeting G3BP1 may be a promising anti-tumor strategy (Zhang et al. 2019). Our study is the first time to confirm the correlation between miR-548b-5p, miR-548c-5p and G3BP1. Restoration of G3BP1 levels reversed malignant phenotypes including proliferation, migration and immune escape of LINC01088-knockdown CRC cells. LINC01088 is a promising target for colorectal cancer treatment.
It is an outstanding challenge for cancer treatment that cancer cells escape immune surveillance and then promote tumor progression (Reisländer et al. 2020). CD8 T cell-mediated tumor regression is often inhibited by its interaction with programmed death ligand 1 (PD-L1) that plays a vital role in anti-tumor immunity (Freeman et al. 2000). G3BP1 plays a key regulatory role in DNA sensing and activation of cyclic GMP-AMP synthase-stimulator of interferon genes (cGAS-STING) pathway (Liu et al. 2019b), contributing to the upregulation of PD-L1. Besides, G3BP1 is responsible to the activation of the NF-κB and STAT3 pathways (Omer et al. 2020), both of which promote PD-L1 expression in cancer (Antonangeli et al. 2020;Jahangiri et al. 2020). In this study, we found that G3BP1 was highly expressed in various cancer tissues including colorectal cancer through bioinformatic analysis and LINC01088 knockdown dramatically decreased G3BP1 and PD-L1 expressions. However, downregulated PD-L1 expression could be restored by G3BP1 overexpression in LINC01088knockdown xenograft tumor tissues, suggesting that G3BP1 regulates PD-L1 expression in colorectal cancer. Therefore, targeting LINC01088 by RNA interference (RNAi)-related strategies may be effective in improving the effectiveness of cancer immunotherapy targeting PD-L1.
In summary, our findings have confirmed that LINC01088 is significantly upregulated in colorectal cancer tissues and directly binds to miR-548b-5P/miR-548c-5P to regulate G3BP1 and PD-L1 expression, thereby promoting colorectal cancer progression. LINC01088 exhibited tumor-promoting property. This work provides novel insights into the pathogenesis of colorectal cancer.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.

Conflict of interest The authors declare no conflict of interest.
Ethical approval All patients signed informed consent form prior to surgery, and the study was approved by the Ethics Committee of Nanjing First Hospital. All animal experiments were conducted under the approval of Animal Care and Use Committee of Nanjing First Hospital.