VPS9D1-AS1 overexpression amplifies intratumoral TGF-β signaling and promotes tumor cell escape from CD8+ T cell killing in colorectal cancer

  1. Lei Yang  Is a corresponding author
  2. Xichen Dong
  3. Zheng Liu
  4. Jinjing Tan
  5. Xiaoxi Huang
  6. Tao Wen
  7. Hao Qu  Is a corresponding author
  8. Zhenjun Wang  Is a corresponding author
  1. Medical Research Center, Beijing Chao-Yang Hospital, Capital Medical University, China
  2. Department of General Surgery, Beijing Chao-Yang Hospital, Capital Medical University, China
  3. Department of Cellular and Molecular Biology, Beijing Chest Hospital, Capital Medical University & Beijing Tuberculosis and Thoracic Tumor Research Institute, China

Abstract

Efficacy of immunotherapy is limited in patients with colorectal cancer (CRC) because high expression of tumor-derived transforming growth factor (TGF)-β pathway molecules and interferon (IFN)-stimulated genes (ISGs) promotes tumor immune evasion. Here, we identified a long noncoding RNA (lncRNA), VPS9D1-AS1, which was located in ribosomes and amplified TGF-β signaling and ISG expression. We show that high expression of VPS9D1-AS1 was negatively associated with T lymphocyte infiltration in two independent cohorts of CRC. VPS9D1-AS1 served as a scaffolding lncRNA by binding with ribosome protein S3 (RPS3) to increase the translation of TGF-β, TGFBR1, and SMAD1/5/9. VPS9D1-AS1 knockout downregulated OAS1, an ISG gene, which further reduced IFNAR1 levels in tumor cells. Conversely, tumor cells overexpressing VPS9D1-AS1 were resistant to CD8+ T cell killing and lowered IFNAR1 expression in CD8+ T cells. In a conditional overexpression mouse model, VPS9D1-AS1 enhanced tumorigenesis and suppressed the infiltration of CD8+ T cells. Treating tumor-bearing mice with antisense oligonucleotide drugs targeting VPS9D1-AS1 significantly suppressed tumor growth. Our findings indicate that the tumor-derived VPS9D1-AS1/TGF-β/ISG signaling cascade promotes tumor growth and enhances immune evasion and may thus serve as a potential therapeutic target for CRC.

Editor's evaluation

This research work uncovered the role of a long noncoding RNA VPS9D1-AS1(VPS) in mediating immune evasion of colorectal cancer cells, which is achieved via amplifying intra-tumoral TGF-β/ISG signaling to facilitate escape from cytotoxic T cells killing. Overall, the experiments were well-designed and the data were properly analyzed. The findings are of potential significance to gaining insight into treating colorectal cancer cells by enhancing the efficacy of immunotherapy

https://doi.org/10.7554/eLife.79811.sa0

Introduction

Colorectal carcinoma (CRC) is a major cause of cancer-related death worldwide and shows a high propensity for metastatic dissemination (Siegel et al., 2020). Microsatellite-stable (MSS) CRC is regarded as immunologically ‘cold’, meaning that it is scarcely infiltrated by T cells and possibly nonimmunogenic and therefore unlikely to benefit from immune therapies (Guan et al., 2021). Thus, immune checkpoint blockade (ICB) is more effective in microsatellite instability-high (MSI-H) CRC but not in MSS (Liao et al., 2019). The lack of a DNA mismatch repair mechanism in MSI patients results in a higher tumor mutation burden and thus high neoantigen exposure that favors ICB (Lu et al., 2021). However, important immune features, including the degree of T cell infiltration and the differentiation or activation state of T cells, remain to be elucidated (Benci et al., 2019).

Effective ICB relies on CD8+ T cell infiltration in the tumor microenvironment (TME). However, advanced-stage tumor cells secrete high levels of transforming growth factor (TGF)-β to reduce the activity of intratumoral cytotoxic T lymphocytes (CTLs), thereby inhibiting their antitumor effector functions (Katlinski et al., 2017). As a result, solid tumors evade anticancer immunity by establishing immune-privileged niches in the TME (Chongsathidkiet et al., 2018). Increased TGF-β in the TME limits the adaptive immune responses by inhibiting T effector cell functions and ushering exhausted T cells to apoptosis (Tauriello et al., 2018; Liu et al., 2020). The receptors of interferon (IFN) are found to regulate TGF-β signaling pathway and are associated with CD8+ T cell immunity (Mariathasan et al., 2018).

IFN signaling is essential for communication between tumor cells and their neighboring cells (Sistigu et al., 2014). Endogenous IFNs contribute to antitumor immunity by stimulating specific CD8α lineage dendritic cells to cross-present antigens to CTLs (Katlinski et al., 2017) and provide a ‘third signal’ to stimulate the clonal expansion of CD8+ T cells (Gracias et al., 2013). In contrast, high levels of tumor-derived IFN stimulating genes (ISGs) are associated with immunological resistance. For example, IFN alpha receptor (IFNAR)–1 knockout (KO) in mouse cancer cells provoked pronounced immune responses after ionizing radiation, and the cancer cells were more susceptible to CD8+ T cell-mediated killing (Chen et al., 2019). In tumor cells, PDL1 expression is promoted by IFN-γ secretion, which results in tumor cells escaping immune elimination (Cerezo et al., 2018). Thus, the IFN pathway plays contradictory roles in tumor cells and T lymphocytes.

VPS9D1-AS1 (also known as MYU), a long noncoding RNA (lncRNA) that has been proven to be overexpressed in multiple types of cancers (Kawasaki et al., 2016; Tan and Yang, 2018; Wang et al., 2020), was identified as a target of Wnt/c-Myc signaling and exhibited pro-oncogenic roles (Kawasaki et al., 2016). Recently, VPS9D1-AS1 was reported to enhance colon cancer progression through upregulating integrin subunit alpha 1 (Huang et al., 2022). VPS9D1-AS1 was demonstrated to upregulate the kinesin family member 11 through competitively sponging miRNA-30a (Liu et al., 2021). Here, we first report that VPS9D1-AS1 is an essential lncRNA that decreases CD8+ T cell infiltration by enhancing TGF-β and ISG expression in CRC. In addition, we propose that VPS9D1-AS1 might serve as a drug target to enhance the efficacy of ICB treatment against CRC.

Results

Increased VPS9D1-AS1 levels are positively associated with TGF-β signaling in CRC tissues

To study the clinical relevance of VPS9D1-AS1 expression, we used RNAscope to evaluate VPS9D1-AS1 levels in two independent cohorts. The OUTDO cohort enrolled 158 CRC subjects, and the BJCYH cohort enrolled 49 CRC patients. The levels of VPS9D1-AS1 were significantly higher in cancer tissues than in normal intestinal epithelial tissues (Figure 1A, Figure 1—figure supplement 1A). The survival and pathological characteristic analyses demonstrated that the levels of VPS9D1-AS1 were significantly associated with overall survival (OS), TNM stage, and tumor lymph node metastasis (Figure 1B, Figure 1—figure supplement 1B). We further confirmed the overexpression (OE) of VPS9D1-AS1 in cancer tissues using qRT-PCR assays (Figure 1C).

Figure 1 with 1 supplement see all
VPS9D1-AS1 is significantly upregulated in colorectal cancer (CRC) and activates the TGF-β signaling pathway.

(A) RNAscope stained VPS9D1-AS1 in CRC tissues that were enrolled in OUTDO (upper) and BJCYH cohorts (lower). Semiquantitative analyses of the levels of VPS9D1-AS1 in cancer and normal tissues of CRC patients (right). (B) Kaplan-Meier overall survival curves of VPS9D1-AS1-positive and VPS9D1-AS1-negative CRC patients. (C) qRT-PCR evaluation of the mRNA levels of VPS9D1-AS1 (upper). Expression of VPS9D1-AS1 was compared in paired normal and cancer tissues (lower). (D) Representative pictures of VPS9D1-AS1-negative (+, C11) and VPS9D1-AS1-positive (G15, ++++) and multispectral fluorescence immunohistochemistry (mfIHC)-stained TGF-β, TGFBR1, and SMAD1/5/9 in the same CRC patients. (E) Integrative analysis of RNAscope and mfIHC data indicates that cancer tissues with high levels of VPS9D1-AS1 had higher levels of TGF-β, TGFBR1, and SMAD1/5/9 than these with low levels of VPS9D1-AS1. (F) Pearson correlation analyses investigated the mRNA levels of VPS9D1-AS1, TGF-β, TGFBR1, SMAD1, and SMAD9. p-Values were obtained by chi-square (A), log-rank test (B), unpaired t nonparametric test (C, E), paired t test (C), and Pearson correlation test (F). Data are shown as data points with mean ± standard deviation of mean (SEM) (C), data are depicted by violin and scatter plots with mean value (E). * p<0.05, ** p<0.01, *** p<0.001.

Our previous study quantitatively investigated eight proteins (TGF-β, TGFBR1, TGFBR2, SMAD1/5/9, pSMAD1/5/9, SMAD2/3, pSMAD2/3, and SMAD4) involved in TGF-β signaling by multispectral fluorescence immunohistochemistry (mfIHC) staining (Yang et al., 2018; Yang et al., 2019). Because mfIHC and RNAscope assays were carried out on the same CRC tissue samples, we examined the relationships between VPS9D1-AS1 and TGF-β signaling. The protein levels of TGF-β signaling molecules were analyzed separately in the tumor and the cancer stroma. In tumor tissues, we found that CRC patients with positive VPS9D1-AS1 expression shown higher levels of TGF-β, TGFBR1, and SMAD1/5/9 (Figure 1D-E). We also detected the upregulation of mRNA encoding TGFBR1, SMAD1, and SMAD9 in tissue samples from BJCYH cohort using qRT-PCR (Figure 1—figure supplement 1C-D). In cancer stroma, VPS9D1-AS1 showed no effects on TGF-β signaling molecules (Figure 1—figure supplement 1E). At the mRNA level, VPS9D1-AS1 was positive associated with the levels of TGFB1 and SMAD1 (Figure 1F).

Overexpression of VPS9D1-AS1 negatively associates with the levels of infiltrated cytotoxic T lymphocytes

To further explore the role of VPS9D1-AS1, we compared their levels in The Cancer Genome Atlas (TCGA) datasets that included consensus molecular subtype (CMS) status (Guinney et al., 2015). Our analyses revealed that VPS9D1-AS1 was expressed predominantly in CMS2 patients (Figure 2—figure supplement 1A). The lymphocyte infiltration signature scores in CMS2 were significantly lower than those in CMS1, CMS3, and CMS4 (Figure 2—figure supplement 1B). Thus, we considered that the OE of VPS9D1-AS1 in CRC cells might be an important cause of the exclusion of T-infiltrating lymphocytes (TIL) from the TME.

To validate this hypothesis, we evaluated the levels of TILs in CRC tissue samples. In the OUTDO cohort, an mfIHC assay was carried out to calculate the percentages (%) of T lymphocytes in the total cancerous and cancer stromal cells, which represent the levels of T cell infiltration. TILs included CD4+, CD8+, and FOXP3+ T cells, and all subsets were significantly reduced in the cancerous tissues compared to the cancer stromal tissues (Figure 2A, Figure 2—figure supplement 1C). In the BJCYH cohort, IHC assays demonstrated that the levels of CD8+ T cells were decreased while FOXP3+ T cells were increased in cancer tissues in comparing with matched normal tissues (Figure 2B, Figure 2—figure supplement 1D).

Figure 2 with 1 supplement see all
VPS9D1-AS1 is associated with reduced T lymphocyte infiltration.

(A) Representative pictures of T cell infiltration (CD4, CD8, FOXP3) in colorectal cancer (CRC) tissues for VPS9D1-AS1 quantification. Tumor cells are marked by cytokeratin. (B) Representative pictures of CD8+, CD4+, FOXP3+ T cells, and TGFBR1 stained by immunohistochemistry (IHC) in the BJCYH cohort. (C) The overall survival curves depicting the percentage of surviving CRC patients stratified by the levels of CD8+ T cell infiltration in cancerous (Ca) tissues, cancer stroma (STM), and the Ca/STM ratio. (D) Ca/STM ratios of CD4+, CD8+, and FOXP3+ T cells were calculated to identify the difference between VPS9D1-AS1 negative and positive populations in the OUTDO cohort. (E) The numbers of CD4+, CD8+, and FOXP3+ T cells in cancer tissues of BJCYH cohort were compared between VPS9D1-AS1 negative and positive tissues. (F) Pearson correlation analyses investigated the relationships between T-infiltrating lymphocytes and TGF-β signaling in cancer tissues. Eight protein levels were investigated by multispectral fluorescence IHC assays in same samples, and fluorescence intensity of each protein level was transferred into quantitative data for Pearson correlation analyses. p-Values were obtained by log-rank test (C), unpaired t nonparametric test (D, E), and Pearson correlation test (F). Data are shown by mean ± SEM (D, E). * p<0.05, ** p<0.01, *** p<0.001.

In the OUTDO cohort, the percentages of CD4+ T and FOXP3+ T cells in cancerous tissues (Ca) and cancer stroma (STM) and their ratios (Ca/STM) did not show any statistically significant relationship with OS (Figure 2—figure supplement 1E). In contrast, our analyses revealed that the levels of CD8+ T cells in Ca and STM and the Ca/STM ratio were significantly associated with OS (Figure 2C). We next tried to investigate the relationship between VPS9D1-AS1 and TILs and found that the levels of TILs were not significantly different between patients with low levels of VPS9D1-AS1 and those with high levels of VPS9D1-AS1 (Figure 2—figure supplement 1F). Interestingly, the levels of VPS9D1-AS1 were related to the Ca/STM ratios of CD4+ and CD8+ T cells (Figure 2D), suggesting that VPS9D1-AS1 prevented T cells from entering cancer tissues. In the BJCYH cohort, the levels of CD4+ and CD8+ T cells were compared between VPS9D1-AS1 positive patients and these negative patients. High levels of VPS9D1-AS1 were demonstrated to associate with lower levels of TILs (CD8+ T cells) (Figure 2E). We further performed a Pearson correlation analysis to explore the relationships between the levels of TGF-β signaling molecules and TILs in OUTDO cohort. In tumor cells, the protein levels of TGF-β, TGFBR1, SMAD2/3, and pSMAD2/3 were negatively associated with the Ca/STM ratio of CD8+ T cells, and the levels of SMAD4 were negatively associated with the Ca/STM ratio of CD4+ T cells (Figure 2F), which is consistent with the role of TGF-β signaling in suppressing TILs. On the other hand, the protein levels of TGF-β, SMAD2/3, TGFBR1, and pSMAD1/5/9 in cancer stromal cells were positively associated with FOXP3+ T cell infiltration but negatively associated with CD8+ T cell infiltration (Figure 2—figure supplement 1G). Together, these results suggested that high expressions of VPS9D1-AS1 were positive associated with TGF-β signaling and negative associated with the levels of infiltrated CD8+ cytotoxic T cells.

VPS9D1-AS1 is a tumor driver and positively regulates TGF-β signaling

First, we determined the levels of VPS9D1-AS1 in 16 cell lines and found that CRC cells expressed higher levels of VPS9D1-AS1 than other cells (Figure 3A). We designed four small guide RNAs targeting VPS9D1-AS1 (sgVPS) and used CRISPR/Cas9 to generate stable KO CRC cell lines (Figure 3B, Figure 3—figure supplement 1A-B). VPS9D1-AS1 KO significantly downregulated TGF-β, TGFBR1, and SMAD1/5/9, did not affect SMAD2/3 and SMAD4, and increased SMAD6 expression, which acts as a negative regulator of TGF-β signaling (Figure 3C, Figure 3—figure supplement 1C-D). Furthermore, inferring RNA (siRNA) was used to disrupt the expression of VPS9D1-AS1 in HCT116 cells. We confirmed that VPS9D1-AS1 knockdown (KD) decreased TGF-β, TGFBR1, and SMAD1/5/9 expression (Figure 3—figure supplement 3E-F). Moreover, VPS9D1-AS1 KD (both sgRNA and siRNA) had no impact on the mRNA expression of TGFB, TGFBR1, and SMAD1, ~5, ~9 (Figure 3—figure supplement 2A-B).

Figure 3 with 3 supplements see all
VPS9D1-AS1 controls TGF-β signaling and drives cell proliferation and metastasis.

(A) The levels of VPS9D1-AS1 were determined by qRT-PCR in 16 cell lines. (colorectal cancer: HCT116, SW620, HT29, RKO, SW480; cervical cancer: Hela; lung cancer: A549, H1299; gastric cancer: BGC823; prostatic cancer: PC3, BPH1; leukemia: K562; pancreatic cancer: PANC1; live cancer: HepG2; HUVEC: human umbilical vein endothelial cell; HASMC: human atrial smooth muscle cell). (B) Northern blotting validated the knockout (KO) of VPS9D1-AS1. (C) Western blotting measured the levels of proteins involved in TGF-β, EMT, and ERK signaling pathways. (D) Representative pictures show the cell morphologies of HCT116 and SW480 cell lines. (E) The proliferation of HCT116/SW480 sgControl and sgVPS cells was determined by Cell counting kit-8 (CCK8) assays. (F) Proliferation of xenograft tumors derived from HCT116 sgControl and sgVPS cells. (G) Immunohistochemistry determined the levels of Ki67 and (H) PDL1 in xenograft tissues. (I) RNA fluorescence in situ hybridization (FISH)-immunofluorescence (IF) and (J) RNA immunoprecipitation (RIP) assays showed the interaction between VPS9D1-AS1 and proteins that included TGF-β, TGFBR1, and SMAD1/5/9. (K) RNA pulldown-Western blotting assays detected the interaction between VPS9D1-AS1 and the intended proteins. (L) Western blotting determined the changes in RPS3, TGF-β, TGFBR1, and SMAD1/5/9 in HCT116 control (CTRL) and VPS9D1-AS1 (VPS)-overexpressing (OE) cells treated with CX5461. RPD, RNA pull down probe. p-Values were obtained by two-way ANOVA (E, F) and paired or unpaired t tests (C, G, H, J). Data are shown as the mean ± SEM (C, E, F, G, H, J). *p<0.05, ** p<0.01, *** p<0.001.

Figure 3—source data 1

TGF-β, TGFBR1, SMAD4, SMAD1/5/9, SMAD6, SMAD2/3, β-actin, N-cadherin, E-cadherin, vimentin, CK, ERK, and p-ERK western blot for Figure 3C.

https://cdn.elifesciences.org/articles/79811/elife-79811-fig3-data1-v2.zip
Figure 3—source data 2

RPS3 RNA-pull-down western blot for Figure 3K.

https://cdn.elifesciences.org/articles/79811/elife-79811-fig3-data2-v2.zip
Figure 3—source data 3

SMAD1/5/9, RPS3, TGF-β, and TGFBR1 western blot for Figure 3L.

https://cdn.elifesciences.org/articles/79811/elife-79811-fig3-data3-v2.zip

We next asked whether there was feedback between VPS9D1-AS1 and TGF-β signaling. Human recombinant (h)TGF-β protein and SB431542 were used to treat SW480 and HCT116 cells. These treatments had no significant effect on VPS9D1-AS1 levels in these cell lines (Figure 3—figure supplement 2C-D). On the other hand, the downregulation of TGF-β, TGFBR1, and SMAD1/5/9 by siRNAs reduced the levels of VPS9D1-AS1 by 40–60% compared with the controls (siNC) (Figure 3—figure supplement 2E). These results indicated that loss of the endogenous TGF-β signaling molecules altered the expression of VPS9D1-AS1 through a feedback loop. However, manipulating the TGF-β signaling pathway with exogenous stimuli had no effects.

We next addressed the oncogenic roles of VPS9D1-AS1 such as promoting cell proliferation, migration, and clone formation. First, we found that stable VPS9D1-AS1 KO cells exhibited morphological changes (Figure 3D) and decreased clone formation capacity (Figure 3—figure supplement 2F). VPS9D1-AS1 KO significantly inhibited cell proliferation and migration (Figure 3E, Figure 3—figure supplement 2G). Consistent with these observations, mechanistic analyses revealed that VPS9D1-AS1 KO reduced the levels of ERK, pERK, N-cadherin, vimentin, and cytokeratin but increased the level of E-cadherin (Figure 3C, Figure 3—figure supplement 2H). In xenograft models, VPS9D1-AS1 KO significantly reduced the tumor volumes compared with the controls and significantly decreased the Ki67 and PDL1 levels in xenograft tumors (Figure 3F-H, Figure 3—figure supplement 2I). Specifically, SW480 VPS9D1-AS1 KO cells did not form xenograft tumors in mice (Figure 3—figure supplement 2I). Taken together, these findings support the notion that VPS9D1-AS1 acts as the driver of tumor progression by activating the ERK and EMT pathways.

VPS9D1-AS1 scaffolds TGF-β signaling-related proteins

We predicted that VPS9D1-AS1 might act as scaffolding lncRNA in tumor cells. To validate this hypothesis, RNA FISH-immunofluorescence (IF) assays were conducted, and the colocalization of VPS9D1-AS1, TGF-β, TGFBR1, and SMAD1/5/9 was confirmed in SW480 cells (Figure 3I). RNA immunoprecipitation (RIP) assays further showed that VPS9D1-AS1 directly bound to TGF-β, TGFBR1, and SMAD1/5/9 (Figure 3J).

We next predicted the subcellular localization of VPS9D1-AS1 by lncLocator (Lin et al., 2021) and found that most transcripts of VPS9D1-AS1 were localized in ribosomes (Figure 3—figure supplement 3B). To map the protein binding regions in VPS9D1-AS1, we synthesized four biotinylated RNA probes targeting VPS9D1-AS1 transcript (Figure 3—figure supplement 3A). Our RNA pulldown (RPD) assay also proved that VPS9D1-AS1 bound with RPS3, one of the proteins constituting the small ribosomal subunit (Figure 3K). Thus, we sought to determine whether preventing ribosome biogenesis plays a role in regulating the translation of TGF-β, TGFBR1, and SMAD1/5/9 (Devlin et al., 2016). We found that VPS9D1-AS1 OE prevented RPS3 degradation caused by CX5461 (an inhibitor of RNA polymerase I transcription of ribosomal RNA genes) treatment. However, CX5461 treatment immediately increased the levels of TGF-β, which declined over time. In VPS9D1-AS1 OE cells, the levels of TGFBR1 and SMAD1/5/9 were decreased after treatment with CX5461, but the TGF-β levels did not decrease following the degradation of RPS3 (Figure 3L). These findings suggest that VPS9D1-AS1 scaffolds the TGF-β protein and regulates its translation in ribosomes.

IFN signaling activation induced by VPS9D1-AS1 expression acts downstream of TGF-β signaling

RNA sequencing was performed to identify the mRNAs differentially expressed between HCT116 sgControl and sgVPS cells. A total of 705 differentially expressed genes were identified, which included 203 upregulated genes and 502 downregulated genes (Figure 4A, Figure 4—figure supplement 1A). VPS9D1-AS1 KO significantly inactivated IFNα/β signaling and the cell death pathway as well as immune system processes (Figure 4B, Figure 4—figure supplement 1B-C). Seventeen genes involved in IFNα/β signaling were validated in HCT116, RKO, and SW480 cells. IFI27 and OAS1 were the most significantly downregulated genes upon VPS9D1-AS1 KO (Figure 4C, Figure 4—figure supplement 1D). Analyses in TCGA datasets demonstrated that OAS1 and IFI27 were significantly overexpressed in CRC cancer tissues (Figure 4—figure supplement 1E). We also analyzed the mRNA expression of OAS1 and IFI27 in tissue samples from 26 CRC cancer tissues and 10 normal colon tissues (Figure 4D). Pearson correlation analysis revealed that the levels of OAS1, but not IFI27, were significantly related to VPS9D1-AS1 levels (Figure 4E).

Figure 4 with 1 supplement see all
VPS9D1-AS1 regulates interferon signaling.

(A) Heatmap illustrating the results of RNA sequencing of the genes regulated by VPS9D1-AS1. (B) VPS9D1-AS1 regulated the pathways associated with interferon signaling. (C) Differential expression of 17 genes in the IFNα/β signaling pathway was validated in HCT116 cells. (D) Colorectal cancer (CRC) tissue mRNA levels of IFI27 and OAS1 were determined by qRT-PCR, and (E) their relationships with VPS9D1-AS1 were calculated with Pearson correlation analysis. (F) Effect of VPS9D1-AS1 on STAT1 pathway activation induced by human recombinant IFNα. (G) VPS9D1-AS1 overexpression (OE) increased the expression of IFI27 and OAS1 through activated TGF-β signaling. (H) OAS1 and (I) IFI27 levels exhibited disparate changes upon IFNα stimulation in VPS9D1-AS1 OE cells and control (CTRL) cells. (J) Chromatin immunoprecipitation (ChIP) assays demonstrated the interactions between SMAD4 and the promoter regions of OAS1. (K) Pearson correlation analyses investigated the relationships among VPS9D1-AS1, OAS1, IFNAR1, and TGFBR1 in CRC tissues. (L) Immunohistochemistry assays showed the levels of OAS1, IFNAR1, and TGFBR1 in patients with negative or positive expression of VPS9D1-AS1. p-Values were obtained by unpaired t test (D, J), two-way ANOVA (G, H, I), and Pearson correlation (K). Data are shown as the mean ± SEM (D, G, H, I, J). * p<0.05, ** p<0.01, *** p<0.001. N.S., not significant.

STAT1 is a well-known transcription factor activated by various ligands, including IFNα (Cerezo et al., 2018). After hIFNα stimulation, we found that VPS9D1-AS1 KO resulted in the downregulation of STAT1 and pSTAT1 (Figure 4F). When the cells were treated with hTGF-β, the phosphorylation of STAT1 induced by hIFNα stimulation was inhibited (Figure 4—figure supplement 1F). We further performed siRNA-mediated KD of TGF-β, TGFBR1, and SMAD1 and confirmed that blocking TGF-β signaling reduced OAS1 and IFI27 expression (Figure 4G). The expression levels of OAS1 were more significantly affected than IFI27 (Figure 4G). In contrast, VPS9D1-AS1 OE restored OAS1 and IFI27 expression (Figure 4G). Surprisingly, OAS1 and IFI27 mRNA levels were significantly reduced in VPS9D1-AS1 OE cells, although VPS9D1-AS1 was stably upregulated by ~180.40 times in HCT116 cells and ~42.28 times in SW480 cells (Figure 4—figure supplement 1G-H). However, hIFNα stimulation significantly increased OAS1 and IFI27 in VPS9D1-AS1 OE cells (Figure 4H-I). These results led us to hypothesize that OAS1 activated IFN signaling that was dependent on VPS9D1-AS1.

SMAD4 enters the nucleus and acts as transcription factors and induced transcription factor which regulates gene expression (Derynck et al., 2021). We confirmed that SMAD4 antibodies immunoprecipitated the promoter regions of both the OAS1 (–77 to +284) and IFI27 (–1933 to –1843) genes (Figure 4—figure supplement 1I). Importantly, VPS9D1-AS1 OE significantly enhanced the binding between SMAD4 and the promoter regions of OAS1 but failed to enhance this binding in the IFI27 promoter region (Figure 4J). Moreover, IHC analysis indicated that the levels of OAS1, IFNAR1, and TGFBR1 were consistently elevated in CRC tissue samples that had positive VPS9D1-AS1 expression (Figure 4K~L). Collectively, OAS1 involved in IFN signaling acts downstream of TGF-β signaling and increased by VPS9D1-AS1 OE.

VPS9D1-AS1 mediates the crosstalk between tumors and T cells depending on IFNAR1 expression

The receptor for IFNα/β is one of the downstream targets of ISGs. When we examined the changes in IFNAR1 on the surface of HCT116 cells by flow cytometry (FCM), we found that VPS9D1-AS1 KO reduced the expression of IFNAR1 (Figure 5A). Conversely, VPS9D1-AS1 OE significantly upregulated the expression of IFNAR1 in tumor cells (Figure 5B). To further delineate the upstream pathway essential for IFNAR1 upregulation in tumor cells, we applied the CRISPR/Cas9 technique to abolish OAS1 expression (Figure 5—figure supplement 1A). Our results demonstrated that the deletion of OAS1 decreased the expression of IFNAR1, although VPS9D1-AS1 OE cells were resistant to this effect (Figure 5B). In addition, the deletion of OAS1 inhibited STAT1 and TGF-β expression (Figure 5—figure supplement 1A).

Figure 5 with 1 supplement see all
VPS9D1-AS1 mediates crosstalk between T cells and cancer cells by regulating IFNAR1.

(A) Flow cytometry (FCM) revealed the decrease in IFNAR1 after VPS9D1-AS1 knockout in HCT116 cells and (B) in HCT116 sgControl and sgVPS cells after CRISPR/Cas9-mediated inhibition of OAS1. Experiments were repeated three times. (C) Levels of IFNAR1 on the surface of CD8+ T cells cocultured with HCT116 control (CTRL), VPS overexpression (OE), sgControl, and sgVPS cells were detected by FCM in three independent assays. (D), (E) FCM determination demonstrated that colorectal cancer patients with positive VPS9D1-AS1 expression had lower levels of IFNAR1 in CD4+ and higher levels of PD1 in CD8+ T cells in peripheral blood than these patients with negative VPS9D1-AS1 expression. (F) T cell cytotoxicity assays against CTRL and VPS9D1-AS1-OE MC38-OVA cell lines by OT-1 CD8+ T cells. VPS9D1-AS1 OE reduced the cytotoxicity of activated OT-1 CD8+ T cells. (G) FCM determined the IFN-γ levels in OT-1 CD8+ T cells after exposure to CTRL and VPS-OE MC38 cells. Antibodies against TGF-β and PD1 were added to the medium. The experiments were repeated three times. p-Values were obtained by unpaired t test (A, B, C, E, F, G). Data are shown as the mean ± SEM (A, B, C, E, F, G).

To investigate the interaction of T cells and tumor cells, we prepared T cells from the peripheral blood of healthy donors. CD8+ T cells and total lymphocytes were separately cultured and primed with human interleukin 2 and antibodies against CD3 and CD28. In vitro-primed T cells were cultured with HCT116 sgControl and sgVPS cell lines. Coculture with VPS9D1-AS1 KO cells increased IFNAR1 expression in CD8+ and CD4+ T cells (Figure 5C, Figure 5—figure supplement 1B). We further detected the CD4+ IFNAR1 levels and CD8+ PD1 levels of peripheral blood lymphocytes in 15 patients with VPS9D1-AS1 positive expression and 16 patients with VPS9D1-AS1 negative expression in cancer tissues (Figure 5D-E). Although there was no statistically significant difference, we found that CRC patients with tissue positive VPS9D1-AS1 expression shown lower levels of IFNAR1 in CD4+ T cells. However, VPS9D1-AS1 was found to be positively associated with the expression levels of PD1 of peripheral blood CD8+ T cells.

We further developed a T cell cytotoxicity assay using MC38-OVA (ovalbumin) cells. Our models successfully demonstrated that primed CD8+ T cells from OT-1 mice suppressed the proliferation of MC38-CTRL-OVA and MC38-VPS9D1 OE-OVA tumor cells (Figure 5F, Figure 5—figure supplement 1C). Although there were no statistically significant differences, we observed that supplied anti-TGF-β and anti-PD1 antibodies in media with OT-1 cells enhanced the cytotoxicity of CD8+ T cells in killing VPS9D1-AS1 OE cells (Figure 5F). FCM analysis showed that CD8+ T cells secreted more IFN-γ once they contacted tumor cells. However, VPS9D1-AS1 OE tumor cells inhibited CD8+ T cells from secreting IFN-γ. Furthermore, neutralizing antibodies against PD1 restored IFN-γ secretion by CD8+ T cells (Figure 5G).

These data support the idea that VPS9D1-AS1 upregulates OAS1 by enhancing TGF-β signaling derived from cancer cells to protect themselves from T cell-mediated cytotoxicity through regulation of IFNAR1.

Upregulation of VPS9D1-AS1 inhibits antitumor immune cell infiltrations in immune-competent mice

The human VPS9D1-AS1 gene is located on the plus strand, while the protein coding gene VPS9D1 is located on the minus strand of human chromosome 16. NR045849 is a mouse lncRNA located on the plus strand near the Vps9d1 gene (Figure 6—figure supplement 1A). When we ectopically expressed full-length VPS9D1-AS1 in MC38 and CT26W cell lines (Figure 6—figure supplement 1B), we observed an increase in the expression levels of TGF-β, TGFBR1, SMAD1, and STAT1 (Figure 6A). NR045849 OE did not increase the levels of TGF-β in MC38 cells (Figure 6A). These findings indicated that VPS9D1-AS1 shared similar biological functions in murine and human cells. Thus, we decided to explore the roles of VPS9D1-AS1 in vivo by OE in murine tumor cells.

Figure 6 with 1 supplement see all
VPS9D1-AS1 overexpression (OE) cells inhibited T cell function in vivo.

(A) VPS9D1-AS1 OE promoted the expression of TGF-β, TGFBR1, SMAD1, and STAT1 in murine cells. (B) MC38 VPS9D1-AS1 OE and control (CTRL) cells were determined proliferation using CCK8 assays. (C) Growth curves of MC38 allograft tumors (n=7 per group). (D) Plots represent the percentages of CD8+ and CD4+ T cells and PD1 frequencies in allograft tumors and spleens. (E) Levels of CD8+ and CD4+ T cells were compared in the tumor and spleen. (F) Levels of PD1 in CD4+ T and CD8+ T cells were compared between CTRL and VPS OE allograft tumors (left) and spleens (right). p-Values were obtained by two-way ANOVA (B) and unpaired t nonparametric tests (E, F). Data are shown as the mean ± SEM (E, F). **p<0.01, ***p<0.001.

Figure 6—source data 1

TGF-β, SMAD1, STAT1, TGFBR1, and β-actin western blot for Figure 6A.

https://cdn.elifesciences.org/articles/79811/elife-79811-fig6-data1-v2.zip

We further upregulated VPS9D1-AS1 more than 600-fold through 3 transfection cycles of lenti-VSP9D1-AS1 in MC38 cells. VPS9D1-AS1 OE cells exhibited a higher speed of growth than control cells in vitro (Figure 6B). VPS9D1-AS1 OE MC38 cells as well as CTRL cells were subcutaneously injected into C57BL/6 mice (Figure 6—figure supplement 1C). MC38 VPS OE cell-derived tumors grew faster than MC38 CTRL cell-derived tumors (Figure 6C, Figure 6—figure supplement 1D-F). We also investigated T cells in mouse spleens and infiltrating T cells in tumor allograft tissues (Figure 6D). The splenic CD8+ T and CD4+ T cells showed no significant differences between mice injected with MC38 VPS OE and CTRL cells (Figure 6E). When allograft tumors were digested into single cells and analyzed by FCM, however, we found that VPS9D1-AS1 OE prevented CD4/8+ T cell from infiltrating into the tumor tissue (Figure 6E). These data underscore the inhibitory roles of VPS9D1-AS1 in T cell tumor infiltration.

We further assessed T cell surface markers that included CD44, CD107, CD62L, CCR7, and PD1 (Figure 6E, Figure 6—figure supplement 1G). The levels of PD1 on CD8+ T cells were elevated in mice bearing MC38 VPS OE tumors (Figure 6F), while CD44, CD107, CD62L, and CCR7 expression were not significantly different in either spleens or allografted tumor tissues (Figure 6—figure supplement 1H). These results indicated that VPS9D1-AS1 regulated the pathway that controls the activation or differentiation of CD8+ T cells.

VPS9D1-AS1 promotes tumorigenesis in AOM/DSS-induced intestinal cancer and acts as a therapeutic target

Using the VPS9D1-AS1 transgenic (VPS Tg) mouse model, we generated an azoxymethane/dextran sulfate sodium salt (AOM/DSS) model to further address the oncogenic roles of VPS9D1-AS1 in vivo (Figure 7—figure supplement 1A). VPS9D1-AS1 was found to specifically express in intestinal epithelium of VPS Tg mice (Figure 7—figure supplement 1B). VPS Tg and wild-type (WT) mice from the same founder were subjected to AOM/DSS treatment cycles (Figure 7A). After AOM/DSS treatment, VPS Tg mice showed markedly higher intestinal tumor growth than WT mice (Figure 7B-D, Figure 7—figure supplement 1C). The VPS Tg mice had a shorter OS time than the WT mice (Figure 7E). Interestingly, VPS Tg mice showed lower CD8+ T cell infiltration (Figure 7F-G). We further determined the levels of Ifnar1 in murine tumors and found that there were no differences in Ifnar1 levels between cancer and normal tissues (Figure 7—figure supplement 1D). However, Ifnar1 levels in tumor tissues were increased in VPS Tg mice (Figure 7H), indicating that VPS9D1-AS1 OE promoting Ifnar1 expression in tumor cells enhanced tumor growth.

Figure 7 with 3 supplements see all
Transgenic mice validated the inhibitory role of VPS9D1-AS1 on CD8+ T cell infiltration and demonstrated the antitumor effects of targeting VPS9D1-AS1 with an antisense oligonucleotide (ASO) drug.

(A) Treatment scheme of the azoxymethane/dextran sulfate sodium salt (AOM/DSS) colorectal cancer (CRC) model. Endpoint at 12 weeks. (B) Representative images of colorectal tumors in mice of the indicated genotypes. (C) Hematoxylin (H) and eosin (E) staining showing the AOM/DSS-induced murine tumors. (D) Tumor volumes are plotted for wild-type (WT) and VPS Tg mice. (E) The overall survival (OS) curve depicts the difference in survival rate for WT and VPS Tg mice. (F), (G) VPS9D1-AS1 suppressed CD8+ T cell infiltration in AOM/DSS-induced CRC tissues. (ROI, region of interest). (H) The levels of Ifnar1 were upregulated in VPS Tg tumors compared with WT tumors. (I) Tumors, their weights, and (J) growth curves of HCT116 xenograft tumors after injecting ASO-VPS (n=6) and ASO-NC (n=6). (K) Western blotting assays of the proteins involved in TGF-β, IFN, ERK, and EMT signaling in xenograft tumors. (L) Treatment scheme of ASO-treated mice with AOM/DSS-induced CRC. (M) ASO-VPS treatment decreased tumor volumes and increased OS of VPS Tg mice with AOM/DSS-induced CRC. (N) Ifnar1 expression upon ASO-VPS and ASO-NC treatment. p-Values were obtained by unpaired t nonparametric test (D, G, H, I, K, M, N), log-rank test (E, M), and two-way ANOVA tests (J). Data are shown as the mean ± SEM (D, G, H, I, K, M, N). * p<0.05, ** p<0.01, *** p<0.001.

Figure 7—source data 1

TGF-β, TGFBR1, SMAD1, OAS1, IFI27, ERK, E-cadherin, vimentin, and β-actin western blot for Figure 7K.

https://cdn.elifesciences.org/articles/79811/elife-79811-fig7-data1-v2.zip

To address the therapeutic potential of targeting VPS9D1-AS1, three 2-O-methyl antisense oligonucleotides (ASOs) specifically targeting VPS9D1-AS1 and one negative control were developed. Transfection of ASOs targeting VPS9D1-AS1 impaired the proliferation of HCT116, SW480, and MC38 VPS OE cells in culture (Figure 7—figure supplement 1E). Among them, ASO-siVPS3 showed the highest inhibition efficiency and was further used in an in vivo assay. We treated BALB/c nude mice bearing HCT116 xenograft tumors with intratumoral injection of ASO drugs and demonstrated that ASO-VPS significantly inhibited tumor growth in vivo (Figure 7I~J, Figure 7—figure supplement 1F). Moreover, the levels of TGF-β, TGFBR1, SMAD1, OAS1, IFI27, ERK, N-cadherin, and vimentin were decreased, while E-cadherin was increased in ASO-VPS-treated xenograft tumors (Figure 7K).

VPS Tg mice were subjected to the AOM/DSS cycle to investigate whether VPS9D1-AS1 acts as a therapeutic target (Figure 7L, Figure 7—figure supplement 1G). At the 12th week, mice were divided into two groups and treated with ASO-NC and ASO-VPS drugs. In the ASO-NC-treated group, four (4/6) mice died because of aggressive tumor progression in the intestinal tract (Figure 7M). In contrast, all six mice treated with ASO-VPS survived until the final follow-up at the 27th week. Among these six mice, five mice developed CRC, but their body weights were heavier than those of ASO-NC-treated mice from the 12th week to the 27th week (Figure 7—figure supplement 1H). ASO-VPS significantly reduced the tumor volumes (Figure 7M) and reduced Ifnar1 expression in murine tumors compared to ASO-NC treatment (Figure 7N). To investigate the function of IFNAR1 on VPS9D1-AS1 promoting cell proliferation, stable cell lines with IFNAR1 KD were produced in VPS9D1-AS1 OE cells and CTRL cells (Figure 7—figure supplement 2A). The IFNAR1 KD significantly inhibited HCT116 VPS OE cell proliferation (Figure 7—figure supplement 2B). Collectively, our in vivo analyses indicated that VPS9D1-AS1 was the driver of CRC, inhibited CD8+ T cell infiltration, and could serve as a therapeutic target.

Discussion

Our present study shows that VPS9D1-AS1 synergizes with the TGF-β and IFN signaling pathways to form a signaling axis in which high intratumor cell activation prevents CTL evasion in the TME. Mechanistically, VPS9D1-AS1 enhances TGF-β signaling by increasing the expression of TGF-β, TGFBR1, and SMAD1/5/9 at the protein level through scaffold mechanisms. Meanwhile, VPS9D1-AS1 KO in tumor cells downregulates the expression of many ISG genes, thus inactivating IFN signaling, and VPS9D1-AS1 KO in tumor cells increases the sensitivity of tumor cells to T cell cytotoxicity, which is illustrated by high levels of secreted IFN. Thus, VPS9D1-AS1/TGF-β/ISG signaling consists of the pathway that crosstalks between tumors and CD8+ T cells. Our in vivo assays support the use of VPS9D1-AS1 as a therapeutic target for CRC treatment (Figure 7—figure supplement 3).

The dual roles of IFN signaling in tumor immune reactions have recently received intense attention (Benci et al., 2016). Autocrine type I IFNs from activated T cells augment T cell-mediated tumor regression (Chen et al., 2021) and are essential for the proliferation and differentiation of T cells (Lopes et al., 2021). In the TME, both tumor cells and infiltrating T cells express IFNA/GR, which sense and bind IFN molecules and then activate the expression of a series of ISGs (Grasso et al., 2021; Jiang et al., 2018). On the other hand, the suppression of IFN signaling in tumor cells stimulates the production of chemokines, such as CXCL13, which results in tumor infiltration of NK cells and subsequent inhibition of tumor progression (Muthalagu et al., 2020). Our data show that VPS9D1-AS1 OE in CRC cells increases IFNAR1 through upregulated OAS1, which is an ISG gene. Hundreds of ISG genes have been identified, and they act either as tumor drivers, such as ADAR1 (Fritzell et al., 2019), or tumor repressors, such as UBA7 (Fan et al., 2020). We also found that both OAS1 and IFI27 are ISGs highly expressed in CRC. Consistent with these findings, OAS1 is a prognostic factor in breast cancer (Zhang and Yu, 2020), while IFI27 is overexpressed in cholangiocarcinoma (Chiang et al., 2019). In addition, OAS1 (but not IFI27) is the downstream ISG gene that is activated by VPS9D1-AS1.

IFN-activated STAT1 promotes PDL1 expression in tumors, which further accelerates tumor progression (Lv et al., 2021). This result is consistent with our finding that STAT1 is activated by VPS9D1-AS1. However, the effect of VPS9D1-AS1 on promoting IFN-induced phosphorylation of STAT1 does not persist for a long time, and it might be associated with the role of pSTAT1 in activating the downstream cGAS-STING pathway, which is essential for defective-mismatch-repair-mediated immunotherapy (Guan et al., 2021).

Previous work has shown that TGF-β blocks IFN production, which is secreted in a paracrine fashion by activated CTLs and limits tumor regression (Guerin et al., 2019). Antibodies reduce TGF-β signaling in cancer stromal cells, facilitate T cell penetration into the center of tumors, and provoke antitumor immunity (Mariathasan et al., 2018). In the present study, we propose a model of autocrine signaling by cytokines in which VPS9D1-AS1 synergistically activates TGF-β and IFN signaling and induces cytokines in tumor cells to regulate CD8+ T cell infiltration and differentiate into an exhausted phenotype.

AOM/DSS-induced murine colorectal cancers frequently carry mutations in Ctnnb1 and Kras and chronic inflammation-related MSI (Sharp et al., 2018), the Wnt pathway is constitutively activated, closely mirroring human CRC at the molecular level (Schulz-Heddergott et al., 2018). The Tg model we studied here highlights the driving role of VPS9D1-AS1, which might have been activated by mutations caused by AOM/DSS. Chronic inflammation contributes to intestinal symptoms, such as bleeding (Zhang et al., 2020). ASOs are chemically synthesized single-stranded oligonucleotides that selectively inhibit the target mRNA, and it is covalently linked to hepatocyte asialoglycoprotein receptor-binding N-acetylgalactosamine (GalNAc) to achieve cell-selective gene silencing (Yu et al., 2020). ASO drugs targeting VPS9D1-AS1 inhibited tumor-associated symptoms such as bleeding and prolonged survival time. Thus, this preclinical model demonstrates that suppressing VPS9D1-AS1 is an effective way to treat CRC.

In summary, we described the mechanism by which VPS9D1-AS1 promoted tumor immune evasion through the TGF-β signaling pathways and ISGs and provided compelling evidence that the VPS9D1-AS1/TGF-β/ISG axis might serve as a drug target to enhance the efficacy of ICB treatment against CRC. These findings expand our current mechanistic understanding of CRC progression and provide potential therapeutic approaches by targeting VPS9D1-AS1 to enhance immunotherapy in patients with CRC.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
AntibodyTGF-β (Rabbit polyclonal)Cell Signaling TechnologyCat# 3711 smfIHC(1:800)
WB(1:200)
IF (1:100)
AntibodyTGFBR1 (Rabbit polyclonal)AbcamCat# ab31013IF (1:1000) mfIHC(1:2000)
WB(1:500)
IHC(1:400)
AntibodyTGFBR2 (Mouse monoclonal)AbcamCat# ab78419WB(1:500)
AntibodySMAD1/5/9 (Rabbit polyclonal)AbcamCat# ab66737mfIHC(1:1000)
WB(1:400)
AntibodypSMAD1/5/9 (Ser 463/Ser465) (Rabbit polyclonal)Santa Cruz BiotechnologyCat# sc-12353mfIHC(1:500)
AntibodySMAD4 (Mouse monoclonal)Cell Signaling TechnologyCat# 46535mfIHC(1:1000)
AntibodySMAD2/3 (Rabbit polyclonal)Santa Cruz BiotechnologyCat# sc-8332mfIHC(1:400)
AntibodypSMAD2/3 (Ser 423/425) (Rabbit polyclonal)Santa Cruz BiotechnologyCat# sc-11769mfIHC(1:1000)
AntibodySMAD6 (Rabbit polyclonal)Santa Cruz BiotechnologyCat# sc-25321WB(1:200)
AntibodyCytokeratin (CK) (Mouse monoclonal)Santa Cruz BiotechnologyCat# sc-57004mfIHC(1:200)
WB(1:500)
AntibodySTAT1 (Rabbit polyclonal)Cell Signaling TechnologyCat# 14994WB(1:500)
AntibodypSTAT1(Tyr701) (Rabbit polyclonal)Cell Signaling TechnologyCat# 9167WB(1:500)
AntibodyERK (Rabbit polyclonal)Cell Signaling TechnologyCat# 4695 SWB(1:200)
AntibodypERK (Rabbit polyclonal)Cell Signaling TechnologyCat# 4376 SWB(1:200)
AntibodyVimentin (Rabbit monoclonal)AbcamCat# ab92547WB(1:100)
AntibodyE-cadherin (Rabbit polyclonal)AbcamCat# ab15148WB(1:100)
AntibodyN-cadherin (Rabbit polyclonal)AbcamCat# ab12221WB(1:100)
AntibodyGAPDH (Mouse monoclonal)BeyotimeCat# AG019-1WB(1:1000)
Antibodyβ-actin (Mouse monoclonal)BeyotimeCat# AF0003WB(1:1000)
AntibodyCD4 (Rabbit monoclonal)AbcamCat# ab133616IHC(1:200) mfIHC(1:1000)
AntibodyCD8 (Mouse monoclonal)Santa Cruz BiotechnologyCat# sc-53212IHC(1:500) mfIHC(1:200)
AntibodyFOXP3 (Rabbit polyclonal)Cell Signaling TechnologyCat# 98377IHC(1:200) mfIHC(1:800)
AntibodyOAS1 (Rabbit polyclonal)Cell Signaling TechnologyCat# 14498IHC(1:100)
AntibodyIFI27 (Rabbit polyclonal)AbclonalCat# A14174WB(1:100)
AntibodyRPS3 (Rabbit monoclonal)AbcamCat# ab128995WB(1:200)
AntibodyIFNAR1 (Rabbit polyclonal)AbcamCat# ab45172WB(1:100)
AntibodyPDL1 (Rabbit monoclonal)AbcamCat# ab205921WB (1:100)
AntibodyKi67 (Rabbit polyclonal)ZSGB-BioCat# ZM-0166IHC (1:1000)
AntibodyHuman CD3 (OKT3) FG (Mouse monoclonal)eBioscienceCat# 16-0037-85(1:100)
AntibodyHuman CD28 FG 16-0289-85 (Mouse monoclonal)eBioscienceCat# 16-0289-85
AntibodyHuman FITC CD3 (Mouse monoclonal)BD PharmingenCat# 555332FCM(1:100)
AntibodyHuman APC CD4 (Mouse monoclonal)BD PharmingenCat# 555349FCM(1:100)
AntibodyHuman PE CD8 (Mouse monoclonal)BD PharmingenCat# 555637FCM(1:100)
AntibodyUltra-LEAF purified anti-mouse CD3 Antibody (Rabbit monoclonal)BioLegendCat# 100239FCM (1:100)
AntibodyUltra-LEAF purified anti-mouse CD28 Antibody (Mouse monoclonal)BioLegendCat# 122021(1:100)
AntibodyMouse FITC CD3 (Rabbit monoclonal)BioLegendCat# 100204FCM(1:100)
AntibodyMouse PE CD8a (Rabbit monoclonal)BioLegendCat# 100708FCM(1:100)
AntibodyMouse APC CD44 (Rabbit monoclonal)BioLegendCat# 103012FCM(1:100)
AntibodyMouse BV421 CD62L (Rabbit monoclonal)BioLegendCat# 104435FCM(1:100)
AntibodyMouse APC CD279 (PD-1) (Rabbit monoclonal)BioLegendCat# 135209FCM(1:100)
AntibodyMouse PE/Cy7 CD107a(LAMP-1) (Rabbit monoclonal)BioLegendCat# 121619FCM(1:100)
AntibodyMouse PE/Cy7 CD197 (CCR7) (Rabbit monoclonal)BioLegendCat# 353211FCM(1:100)
AntibodyInVivoPlus Anti-mouse PD-1 (Mouse monoclonal)BiXcellCat# BP0273(1:100)
AntibodyInVivoPlus Anti-mouse/human TGF-β (Mouse monoclonal)BiXcellCat# BP0057(1:100)
AntibodyHuman IFN-alpha/beta R1 PE-conjugated antibody (Mouse monoclonal)R&DCat# FAB245PFCM(1:100)
Biological samples (Homo sapiens)Human colon cancer tissue microarrayOUTDOHCol-AS-180Su-14
Biological samples (H. sapiens)Human rectal cancer tissue microarrayOUTDOHRec-Ade180Sur-05
Biological samples (H. sapiens)Human colorectal cancer tissue cDNAOUTDOHcolA30CS01
Cell lines (H. sapiens)HCT116Dr. L. Yang (Beijing Chaoyang Hospital)
Cell lines (H. sapiens)SW480Dr. L. Yang (Beijing Chaoyang Hospital)
Cell lines (H. sapiens)RKODr. L. Yang (Beijing Chaoyang Hospital)
Cell lines (H. sapiens)HEK293TDr. L. Yang (Beijing Chaoyang Hospital)
cell lines (Mus musculus)MC38PUMC
Cell lines (M. musculus)CT26WPUMC
Commercial assay or kitQuantiNova SYBR Green PCR kitQiagenCat# 208054
Commercial assay or kitQuantiTect Reverse Transcription kitQiagenCat# 205314
Commercial assay or kitPoerce BCA Protein Assay KitThermoCat# RB230514
Chemical compound, drugLipofectamine 3000InvitrogenCat# L3000015
Recombinant proteinsHuman Interleukin-2 (hIL-2)Cell Signaling TechnologyCat# 8907SC
Commercial assay or kitPerkinElmer Opal 7-color fIHC KitPerkinElmerCat# NEL797001KT
Commercial assay or kitFicoll reagentSolarbioCat# P8900
Commercial assay or kitTumor Dissociation Kit, mouseMiltenyi BiotecCat# 130-091-376
Commercial assay or kitCD8a+ T Cell Isolation Kit, mouseMiltenyi BiotecCat# 130-104-075
Commercial assay or kitAnti-Human CD8 Magnetic Particles-DMBD PharmingenCat# 557766
Commercial assay or kitHiScribeTM T7 Quick High Yeild RNA Synthesis KitNEBCat# E2040S
Commercial assay or kitMonarch RNA Cleanup KitNEBCat# T2030S
Commercial assay or kitIP Lysis BufferThermo FisherCat# 87787
Commercial assay or kitPierce RNA 3’ Desthiobiotinylation kitThermo FisherCat# 20163
Commercial assay or kitPIERCE MAGNETIC RNA PULL-DOWN KITThermo FisherCat# 20164
Chemical compound, drugCX-5461MCECat# HY-13323
Recombinant proteinsHuman Interferon-α1Cell Signaling TechnologyCat# 8927SC
Commercial assay or kitDIG Northern StarterRocheCat# 12039672910
Commercial assay or kitVPS9D1-AS1 FISH ProbesBiosearchN/A
Commercial assay or kitMagna RIP RNA-Binding Protein
Immunoprecipitation Kit
MilliporeCat# 17–700
Commercial assay or kitPositively Charged Biodyne Nylon membraneInvitrogenCat# 10100
Commercial assay or kitChIP-IT Express Enzymatic Magnetic Chromatin Immunoprecipitation kitActive MotifCat# 53009
Commercial assay or kitStellaris RNA FISH Wash Buffer ABIOSEARCHCat# SMF-WA1-60
Commercial assay or kitStellaris RNA FISH Wash Buffer BBIOSEARCHCat# SMF-WA1-20
Commercial assay or kitStellaris RNA FISH Hybridization buffersBIOSEARCHCat# SMF-HB1-10
Commercial assay or kitStellaris FISH Probes with Quasar 570 DyeBIOSEARCHCat# SMF-1063–5
Commercial assay or kitFITC labeled Goat anti-rabbit secondly antibodyByotimeCat# 0652
Commercial assay or kitFITC labeled Goat anti-mouse secondly antibodyByotimeCat# A0568
Commercial assay or kitAntifade mounting medium with DAPISolarbioCat# S2110

RNAscope in situ hybridization assay

RNA in situ hybridization assays were performed using an RNAscope kit (Advanced Cell Diagnostics, Hayward, CA, USA). VPS9D1-AS1 mRNA molecules were detected with single-copy detection sensitivity. Single-molecule signals were quantified on a cell-by-cell basis by manual counting. The signals per cell were evaluated to quantitate the levels of VPS9D1-AS1. The signals were graded as 0 (0–1 dots/10 cells), + (1–3 dots/cell), ++ (4–10 dots/cell), +++ (>10 dots/cell with <10% of dots in clusters), and ++++ (>10 dots/cell with >10% of dots in clusters) (Zhao et al., 2018). The scores were classified as follows: ‘−, +’ represented negative expression and ‘++, +++, ++++’ represented positive expression of VPS9D1-AS1.

Mice

Conditional floxed human VPS9D1-AS1 knock-in alleles with LoxP sites were introduced into the Rosa26 gene locus to construct C57BL/6J-Gt (ROSA)26Sorem(CAG-VPS9D1-AS1)1Smoc mice. A mixture of the construct and CRISPR/Cas9 vector was microinjected into zygotes. The zygotes were implanted into foster mice. The above procedures were performed by Shanghai Model Organisms. Successful integration in the founder mice was identified by PCR analyses of genomic DNA using primers targeting VPS9D1-AS1. After screening, the positive founder (bearing R26-eCAG-VPS9D1-AS1) was crossed with Cre-Villin mice to produce conditional VPS9D1-AS1 transgenic (VPS Tg) mice in the intestinal epithelium. VSPTg mice were backcrossed to C57BL/6 mice for at least six generations. The VPS Tg mice did not show any differences in comparing with WT mice in lifespan and body weight.

To induce colorectal cancer with AOM/DSS, C57BL/6 VPS Tg mice received a single intraperitoneal injection of AOM (Sigma Aldrich) at a dose of 10 mg/kg body weight. One week later, animals were exposed to 1–3 cycles of 2% DSS (MW = 36,000–50,000, MP Biomedicals, CAT# 216011050). For in vivo inhibition of VPS9D1-AS1, mice in the treatment group received 10 nmol (100 μl, 100 mM ASO in PBS) ASOs i.p., once a week. Intestinal tumor volumes were calculated by the formula: π × (1/2 width)2 × length.

For the xenograft tumor model, 2×106 HCT116 or SW480 cells were implanted subcutaneously in 6-week-old female BALB/c nude mice (Charles River). For MC38 cells, 4×106 cells were subcutaneously injected into the lower abdominal region of 6-week-old C57BL/6 WT mice (Charles River). Tumors were assessed two to three times a week by a caliper measurement, and the volumes were calculated with the following formula: 1/2 × (length × width2). MC38 cells were stably transfected with luciferase to label tumor progression and were monitored using a Carestream in vivo imaging system (MS FX Pro). Animal experimental protocols were approved (AEEI-2021–105) according to the guidelines of the Ethics Committee for Animal Testing of Capital Medical University.

Patient samples

There were two independent cohorts in this study. OUTDO cohort enrolled the colon and rectal cancer tissue microarrays (TMA) and cDNA sample array, which were provided by Shanghai OUTDO Biotech (Shanghai OUTDO Biotech Co., Shanghai, China). The HCol-Ade 180Sur-07 TMA consisted of tumor tissues and matched normal tissues from 90 patients with colon cancer who underwent surgery from January 2008 to November 2009 with follow-up information available until May 2014. The Hrec-Ade180Sur-04 TMA included tumor tissues and matched normal tissues from 90 patients with rectal cancer who underwent surgery from October 2007 to August 2008 with follow-up information available until September 2014. Colon cancer cDNA sample array enrolled 80 colon patients and consisted with 80 cDNA samples from tumor tissues and 15 cDNA samples from normal tissues. Bei Jing Chao Yang Hospital (BJCYH) cohort enrolled the tissue and blood samples to use for validation. Tumor samples were fresh or formalin-embedded and collected from CRC patients from 2020 to 2021 at Beijing Chao-yang Hospital.

All sample donors provided informed consent, and the study was conducted under the approval of the Institutional Ethics Committee from Beijing Chaoyang Hospital of Capital Medical University between 2018 and 2020 (2018-ke-24). Samples were collected from patients with CRC who did not receive chemotherapy or radiotherapy before surgery. All procedures were performed in accordance with the relevant guidelines and regulations.

RNA fluorescence in situ hybridization (FISH) and immunofluorescence (IF)

Cells were seeded in 24-well plate with glass coverslips (Solarbio), fixed with 4% (w/v) paraformaldehyde (Solarbio) for 10 min at room temperature, washed with 1× PBS, 70% ethanol for 1 hr, and washed with Stellaris Buffer A (Biosearch). Then cells were incubated with Stellaris FISH probes which were diluted by Stellaris FISH Hybridization Buffer (1:100) at 37℃ overnight. Cells were next washed with Buffer A for 30 min at 37℃ and B for 5 min at room temperature. Primary antibodies TGF-β (dilution, 1:100), TGFBR1 (dilution, 1:1000), and SMAD1/5/9 (dilution, 1:500) with 5% BSA in PBS were, respectively, incubated with cells for 2 hr at room temperature. Then, cells washed three times with PBS and incubated with fluorescein isothiocyanate (FITC)-labeled goat anti-rabbit/mouse secondary antibody (Beyotime) for 30 min at room temperature. Finally, coverslips were mounted onto slides with antifade mounting medium supplemented with DAPI (Solarbio). Confocal images were acquired using a Lecia TCS SP8 laser scanning microscope.

CRISPR/Cas9 interference

CRISPR/Cas9 (all-in-one lentiviral) system was employed to achieve KO of target genes. Briefly, the predicted gRNAs were designed by online tools (http://crispr.mit.edu/) and individually cloned into the all-in-one lentiviral vector (LnetiCRISPRv2). To obtained KO lentiviruses, lentiviral vector along with the packing plasmids psPAX2 and pMD2.G (VSV-G envelope) was transfected into HEK293T cells using Lipofectamine 3000 (Invitrogen) in serum/antibiotic-free media. Media were collected at second and third days post-transfection. The lentiviral media were used to transduce target cells after filtering through a 0.45-μM filter to remove cells or debris and supplied 10 μg/ml Polybrene (Santa Cruz Biotechnology). To produce stable KO models, cells were selected with puromycin for at least 14 days. The KO efficiencies were confirmed by qRT-PCR, northern blotting, or western blotting. The sgRNAs sequences were listed in Supplementary file 1a.

RNA immunoprecipitation

Tumor cells were harvested at 80% confluence that were seeded in a 10-cm dish for RIP. RIP kit (Magna RIP RNA-Binding Protein Immunoprecipitation Kit, Millipore) was applied to detect the interactions between VPS9D1-AS1 and intended antibody-protein complex. The procedures were carried out according to the manufacturer’s instructions. Mouse or Rabbit normal IgG (Millipore), which was served as negative control. InPut and co-immunoprecipitated RNAs were extracted by the reagents in RIP kit for further analysis by qRT-PCR.

Northern blot

For synthesize probes that targeted VPS9D1-AS1, PCR reaction was conducted to clone a 722 bp sequence from pcDNA3.1-VPS9D1-AS1 templated. Primers were listed in Supplementary file 1b. RNA probes were transcript by T7 RNA polymerase and labeling by digoxigenin-11-UTP. To visualize VPS9D1-AS1, a total of 10 μg of the indicated RNA from HCT116 or SW480 cells were subjected to formaldehyde gel electrophoresis and transferred to a positively charged Biodyne Nylon membrane. After fixation, the membranes were hybridized with VPS9D1-AS1 probes at 68℃ for 12 hr. At last, anti-digoxigenin-AP was incubated with the membrane for 30 min at room temperature. Washes were performed as described in DIG Northern Starter Kit, and the membrane was placed in CDP-Star solution and visualized by a Bio-Rad ChemiDoc XRS + system.

RNA isolation, cDNA synthesis, qRT-PCR

Total RNA was extracted from cell or tissue samples using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). For determining mRNA, total RNA (1 µg) was reverse transcribed using the QuantiTect Reverse Transcription kit (Qiagen). qRT-PCR was performed on cDNA using the Applied Biosystems 7500 Real-Time PCR System (Life Technologies, Gaithersburg, MD, USA) and the SYBR Select Master Real-Time PCR assay (Qiagen). For determining microRNA, reverse transcriptions were performed by miScript II RT KIT (Qiagen), and qRT-PCR reactions were carried out by miScript SYBR Green PCR Kit (Qiagen). The PCR program was as follows: predenaturation at 95°C for 5 min, 40 cycles of denaturation at 95°C for 5 s, and annealing 60°C for 30 s and elongation at 72°C for 1 min. The primers are listed in Supplementary file 1b.

Small interfering RNA (siRNA), microRNA mimic transfection

siRNAs or microRNA mimics used to silence targets are listed in Supplementary file 1c. Indicated siRNAs or microRNA mimics were transient transfected into targeted cells by Lipofectamine 3000.

Western blotting

Cells were lysate and determined the protein concentrations using BCA method (Thermo). Then, these lysates were prepared to protein solution and separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). SDS-PAGE gel was subjected to semi-dry transfer and transferred proteins to a polyvinylidene fluoride membranes (Millipore, MA, USA). Follow blocking using 5% milk, membrane was incubated with various primary antibodies for overnight at 4℃. After washing, goat anti-mouse/rabbit IgG secondary antibodies (LI-COR) were incubated the membranes for 1 hr at room temperature. Membrane was visualized using an LI-COR odyssey imager. The relative levels of intended proteins were calculated using ImageJ software (Version 1.52 a).

CCK8, transwell migration, and soft agar clone forming assays

For cellular proliferative ability assays, 800 tumor cells seeded in 96-well plates and incubated with 100-μL culture medium for 5 days. Cellular viability was determined by Cell Counting Kit (CCK8, Dojindo) with dilution 1:100 at each day. For cell migration assays, 5×104 cells were seeded in the upper transwell chambers (8 μm pore size; Coring) using the medium with fetal bovine serum while supplied complete medium in 24-well plate. After culturing for 48 hr, medium and non-migrated cells in upper chambers were wiped off with cotton swaps, and the filter was stained for 30 min with 0.04% crystal violet and rinsed with PBS for three times. For cell clone forming assays, 2×104 tumor cells were seeded in 12-well plates that were coated with dulbecco's modified eagle medium (DMED) containing 1.2% (w/v) soft agar and culture for 30 days. Colonies were stained using 0.04% crystal violet and counted.

mRNA expression profile sequencing

All sequencings were performed by Novogene. Total RNA was extracted from three HCT116sgVPS and three HCT116sgControl cell samples. The quality of purified RNA was tested on an Agilent 2100 Bioanalyzer system (Agilent technologies, Santa Clara, CA, USA). The ribosomal RNAs were removed and were fragmented into small pieces. The first strands were reverse-transcribed into first cDNA by random primer, RNAs were digested by RNase H, followed by second strand cDNA synthesis using DNA polymerase I. A single ‘A’ base adapter was added to the fragments. AMPure XP beads were used filtered 370–420 bp small pieces. The products were removed by second strands with U base using USER enzyme and amplified by PCR to create the final cDNA library. The raw reads were cleaned and mapped against the human reference genome using Hisat2. The differential level of genes was determined based on the value of Fragments Per Kilobase of transcript sequence per Millions base pairs sequenced, which was calculated by cuffdiff. GSEA was performed to enrich pathway of VPS9D1-AS1 based on the differential genes using RNA sequencing.

Immune subtype and consensus molecular subtype (CMS) analyses using TCGA

RNAseq data for TCGA COAD and READ tumor tissues were downloaded from Broad GDAC Firehose portal (https://gdac.broadinstitute.org/). The transcripts per kilobase of exon model per million mapped reads (TPM) values were used to validate the associations for immune subtypes and CMS of indicated genes. The status of immune subtype for each TCGA sample was recognized according to the criterion from Thorsson et al., 2018. CMS information was derived from http://www.synapse.org (Guinney et al., 2015).

RNA pulldown assay

Four probes (RNA probes for RPD) that targeted four regions of VPS9D1-AS1 transcript were designed to perform RPD assay. The primers were listed in Supplementary file 1d. RNA fragments were transcribed with HiScribeTM T7 Quick High Yeild RNA Synthesis Kit (NEB) in vitro. RNA transcript products were treated with RNAase-free DNase I on column with Monarch RNA Cleanup Kit (NEB). These RNAs were labeled using Pierce RNA 3’ Desthiobiotinylation kit (Thermo). 1×107 cell samples were lysed by IP Lysis Buffer (Thermo). Pulldown assays were performed following the Pierce Magnetic RNA-Protein Pull-Down Kit (Thermo). Briefly, RNA probes were bound to streptavidin magnetic beads. Then, these beads were mixed with cell extract and incubated at 4℃ for 60 min with rotation. The binding protein complexes were heated at 95–100℃ for 10 min for purifying intended proteins and further analyzed by western blotting.

Chromatin immunoprecipitation (ChIP) and quantitative PCR

A ChIP assay for various nuclear proteins was performed using ChIP-IT Express Enzymatic Magnetic Chromatin Immunoprecipitation kit (Active Motif). In brief, 1×107 cells were fixed with fixation solution at room temperature for 10 min and stopped by adding glycine buffer. The cells were resuspended in ice-cold lysis buffers provided in the kit and homogenized to aid in nuclei release. To sheared DNA (200–400 bp), nuclei solutions were resuspended and digested for 5 min at 37℃ by enzyme buffers provided in the kit. For immunoprecipitation reactions, samples were incubated with 1 μg of intended antibodies or isotype control (rabbit/mouse IgG) and Protein G Magnetic Beads for overnight at 4℃. The magnetic beads were washed by buffers in the kit. The DNA-protein complex was eluted by heating at 95℃ for 15 min in a thermocycler. The DNA were isolated by adding proteinase K to digest binding proteins and stopped by stopping buffer in kit. The DNA was then subjected to real-time PCR or routine PCR analysis. The primers for ChIP-PCR were listed in Supplementary file 1e.

Multispectral fluorescence immunohistochemistry (mfIHC) and scoring multispectral images

For quantitatively determining the infiltrated lymphocytes, multispectral imaging was employed to stain antibodies in a TMA slide. The mfIHC assays were performed using PerkinElmer Opal 7-color fIHC Kit (PerkinElmer) according to manufacturer’s introduction. Slide was baked 2 hr at 65℃ to prevent tissue samples fall out from slide. After that, slide was deparaffinized to water following the cycle, xylene 15 min two times, 100% ethanol, 95% ethanol, 85% ethanol, and 75% ethanol. Then, samples were antigen retrieval using microwaving method (90 s on 100% power, followed by 15 min on 20% power). The primary antibodies against CD4 (Abcam, dilution 1:1000), CD8 (Santa Cruz, dilution 1:200), FOXP3 (CST, 1:800), and CK (Santa Cruz, dilution 1:200), respectively, stained for 1 hr at room temperature and then incubated with horseradish peroxidase-labeled mouse/rabbit secondary antibody for 10 min at room temperature. Tyramide (TSA)-conjugated fluorophore, a series Opal reagent, mocked the slide at 1:100 dilution. Opal 690 labeled CD4, Opal 570 labeled CD8, Opal 620 labeled FOXP3, and Opal 520 labeled CK. Slide was heated using microwave to remove primary antibody and leaves Opal reagent to show target protein at the last of each Opal staining procedure. Finally, slide was counterstained with Spectral DAPI to show nuclei and cover slips followed by sealing with mounting medium supplied with antifade solution (Applygen Technologies Inc Beijing, China). Multispectral TMA images were scanned using Vectra Polaris Imaging System (Perkin Elmer) through ×10 objective lens.

InForm 2.1.1 software (Perkin Elmer) was used to batch analysis of multispectral images. The scanned image of each tissue core was loaded to InForm and substrate background using the image derived from a slide without staining in the same exposure setting. The segmented tissues of parenchymal neoplasm and mesenchyme stroma were trained according to CK signals intensity. Cells were segmented to count the total number of cells. The percentages for CD4, CD8, and FOXP3 T cells were calculated using the number for each stained by Opal signal dividing by total cell number in tumor or stroma, respectively.

Immunohistochemistry (IHC) assays

The primary antibodies used in immunohistochemistry (IHC) for tissue slides were used for IHC of tissue slides: anti-IFNAR1 (Abclonal, 1:200 dilution), anti-CD4 (Abcam, 1:200 dilution), anti-CD8 (Abcam, Cat# ab93278, 1:500 dilution), anti-FOXP3 (CST, 1:200 dilution), anti-OAS1 (CST, 1:100 dilution), and anti-TGFBR1 (Abcam, 1:400 dilution). The slices were deparaffinized and rehydrated, pretreated with a citric acid antigen retrieval solution (pH = 6.8), and rinsed in PBS. The sections were blocked in 2% goat serum and incubated with the primary antibody overnight at 4°C. The streptavidin-peroxidase method (ZSGG-BIO, CAT# PV9001) was used to show the levels of stained proteins. The number of CD4, CD8, and FOXP3 cells were calculated using K-Viewer V1 system (Version, Code 1.5.3.1, KONFOONG BIOTECH INTERNATIONAL CO., LTD). The levels of OAS1, IFNAR1, and TGFBR1 were scored as follows: 1 (0–25%), 2 (26–50%), 3 (51–75%), and 4 (>75%). The intensity of positive staining was classified into four scales as follows: 0 (negative), 1 (weak), 2 (moderate), and 3 (strong). The levels were semiquantitatively determined as percentages multiplied by intensity.

CD8+ T cell sorting and culturing

Anti-human CD8 magnetic particles (BD) were used to positive selection of human CD8-bearing leukocytes. Human peripheral blood mononuclear cells (PBMC) were isolated by density gradient centrifugation using Ficoll reagent (Solarbio). 50-μl CD8 magnetic particles were mixed with per 1×107 PBMCs and incubated at room temperature for 30 min. The particle labeling volume was placed on the cell separation magnet and washed with buffers in kit. The positive fractions were resuspended with RPMI1640 media. CD3 antibodies, CD28 antibodies, and hIL2 were added to media to expand CD8+ T cells.

Splenocytes were harvested from the spleen of OT-I mice using Mouse CD8+ T Cell Isolation Kit (Miltenyi). In brief, 10-μl Biotin-Antibody Cocktail were mixed with per 107 cells and incubated for 5 min at 4℃. Then, 20-μl Anti-Biotin MicroBeads were added and incubated for 10 min at 4℃. Above cell suspensions were subsequent applied to LS Column in the magnetic field of Magnetic Activated Cell Sorting (MACS) Separator (Miltenyi). The flow-through that contained unlabeled CD8+ T cells was collected and cultured with mouse CD3 and CD28 antibodies for an additional 3 days before use in co-culture.

In vitro cytotoxicity assays

5×105 tumor cells were seeded per well in 12-well culture plate (Excell Bio). At first, CD8+ T cells were admixed in dilutions (twofold, starting at a 1:1 ratio) to determine the cytotoxic efficiencies. After co-culture for 24 hr, T cells were washed away and collected for FCM sorting. After further 3 days, plates were fixed and stained for 30 min using a 0.04% crystal violet solution (Solarbio). When indicated, blocking TGF-β and PD-1 antibodies (BioXcell) was added in media (5 μl per well).

Flow cytometry

For in vitro cell lines, 5×105 cells were digested by 0.25% pancreatin enzymes (Biological Industry) to single-cell suspensions. Cell surface IFNAR1 staining was done for 1 hr at room temperature. Intracellular IFNAR1 staining was fixed using 4% (w/v) paraformaldehyde (Solarbio) and permeabilized using 0.1% (v/v) Trxion-100 (Amresco). For infiltrated T cells in xenograft tumors, single-cell suspensions were prepared by mouse Tumor Dissociation Kit (Miltenyi) according to the manufacturer’s instructions. Mouse spleens were cut into pieces and filtered using a Coring cell strainer to prepare single-splenocyte suspensions. Both splenocyte and PBMC suspensions were lysed using 1× lysis buffer (BD) to remove red cells. A BD FACScanto II was used for data acquisition, and analysis was performed using FlowJo (TreeStar).

Statistical analysis

Statistical analysis was performed using GraphPad Prism software and R software. Student’s t test or ANOVA (one- or two-way) with Bonferroni post hoc test was used to evaluate the statistical significance. The survival curves were plotted according to the Kaplan-Meier method and evaluated by the log-rank test. A p-value less than 0.05 was considered statistically significant.

Data availability

RNA sequencing data set of HCT116 sgControl and sgVPS cells were deposited in Sequence Read Archive (PRJNA716724) and Dryad Digital Repository (https://doi.org/10.5061/dryad.qnk98sfk6).

The following data sets were generated
    1. Yang Lei
    2. Dong X
    3. Liu Z
    4. Tan J
    5. Huang X
    6. Wen Tao
    7. Qu Hao
    8. Wang Z
    (2021) NCBI BioProject
    ID PRJNA716724. VPS9D1-AS1 regualtes differential mRNA in HCT116 cells.
    1. Yang L
    2. Dong X
    3. Liu Z
    4. Tan J
    5. Huang X
    6. Wen T
    7. Qu H
    8. Wang Z
    (2022) Dryad Digital Repository
    Overexpression of VPS9D1-AS1, an activator of transforming growth factor β signaling, upregulates interferon-stimulated-gene expression to regress CD8+ T cell infiltration in the microenvironment of colorectal cancer.
    https://doi.org/10.5061/dryad.qnk98sfk6

References

    1. Huang G
    2. Yang Y
    3. Lv M
    4. Huang T
    5. Zhan X
    6. Kang W
    7. Hou J
    (2022)
    VPS9D1-AS1, a novel long-non-coding RNA, acts as a tumor promoter by regulating the mir-324-5p/ITGA2 axis in colon adenocarcinoma
    American Journal of Translational Research 14:955–966.

Decision letter

  1. Caigang Liu
    Reviewing Editor; Shengjing Hospital of China Medical University, China
  2. W Kimryn Rathmell
    Senior Editor; Vanderbilt University Medical Center, United States
  3. Philippe Krebs
    Reviewer; University of Bern, Switzerland

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "VPS9D1-AS1 overexpression amplifies intratumoral TGF-β signaling and promotes tumor cell escape from CD8+ T cell killing in colorectal cancer" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, 0one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by W Kimryn Rathmell as the Senior Editor. The following individual involved in the review of your submission has agreed to reveal their identity: Philippe Krebs (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) The mechanistic study regarding VPS9D1-AS1/TGF-β/ISG was mainly carried out in vitro, while this part shall be tested in cell line-derived xenografts or mouse tumors to support the claims.

2) The data volume is large and the methodology has covered a broad range in the work, while the arrangement of the manuscript could be improved to optimize the presentation and enhance the readability.

3) Some details of the figures conflict with the statement in the manuscript, and corrections are warranted.

Reviewer #1 (Recommendations for the authors):

Here are some major concerns:

Figure 1/S2: The authors shall explain or discuss why FOXP3 T cells showed a different trend in OUTDO and BJCYH cohorts.

Figure 2: Which cohort(s) was analyzed in 2F? Did OUTDO and BJCYH show the same or different patterns?

Figure 4/S6: 4H-I showed that at 0 hours, OAS1 and IFI27 did not change between CTRL and VPS OE; however, in S6H, OAS1 and IFI27 were much lower in VPS OE, conflicting with 4H-I.

Figure 5/S7: In S7A, VPS OE+sgOAS1 (last three lanes) show a substantial OAS1 expression. This made the claims on OAS1 (5B) less convincing. How did VPS KO cells impact IFNAR1 level of T cells in coculture?

Figures 6 and 7: The mechanistic model proposed by in vitro studies (Figures 4 and 5) shall be tested in cell line-derived xenografts or mouse tumors. For example, the IFNAR1 level of VPS OE tumors was examined (7H) but the IFNAR1 level of infiltrated T cells was not. Moreover, CRC patient-derived xenografts shall be used and treated with VPS ASO, which will be more physiologically relevant than the cell line with 600-fold VPS upregulation used in the study.

Reviewer #2 (Recommendations for the authors):

My comments are detailed below:

(Figure 1)

1. Panels A and B (and S1A) are not useful for the conclusions of the paper – the histology is difficult to distinguish and it is not clear of levels (of expression) or frequencies (of positive cells) were quantified in panel B. Those data are better presented in panel D.

2. Panel F, for instance, doesn't actually corroborate what is indicated in panel A, as the authors already gate on "high" populations, and there are still some samples with almost no VPS9D1-AS1, this suggests the staining protocol is not a good replacement for qRT PCR.

3. In the text the authors write that VPS expression doesn't correlate with the expression of genes in the TGFb pathway (pp4, lines 18-20), but do not actually separate by gene expression, but rather by cancer and normal tissue. A better analysis may actually be a xy graph with regression curves.

(Figure 2)

1. There seem to be artefacts or resolution issues in panel A; in panel B, the inlays seem to have different magnifications (size bars are not provided).

2. It seems data were mixed up data in panel E; in addition, panel D shows Ca/STM ratios while panel E indicates cell numbers, making it difficult to compare the two cohorts. The same applies to Sfig2 panel C versus F.

3. Panel F needs to be more clearly explained, it is difficult to interpret these data with the explanation given and certain p values are missing. The same applies to SFigure panel G.

4. Panel A doesn't show what is indicated in the text (cancer vs stromal pp 5 lines 2-3). It shows neg. vs pos. according to the figure and it is difficult to distinguish healthy from tumor tissue.

(Figure 3)

1. Is SMAD1/5/9 in Figure 3D indeed the top band?

2. In panels E and M (and several other western blot data), automated quantification would help support the statements by the authors on increased / lower expression of specific proteins, which are not always convincing as stated.

3. L – the oligonucleotide design is not fully clear to me; how can these differences between probes be justified, e.g. between RPD1 and RPD2? Why not compare to the entire VPS9D1-AS1 molecule?

4. O – these data are not convincing.

Figure 4

1. In panel E it seems the labels have been swapped.

2. In panel F: there are differences in the two cell lines after VPS-KO; are the labels correct?

3. Panel J and references indicated in the text: I could not find in the literature evidence for a nuclear function of TGFb.

4. In panel K: how were the decimal values for OAS1, IFNAR1, and TGFBR1 determined?

(Figure 5)

1. Panel B: some of the samples seem to have been swapped. Please cross-check.

2. Panel D: How were patients selected to be either VPS neg. or pos.? This is also relevant due to the apparent intra-tumor heterogeneity of VPS9D1-AS1 expression. FACS plot, VPS positive condition, upper and lower rows: both T cell and non T cell populations are shifted down, respectively up, suggesting a problem or changes in the settings of the flow cytometer instrument. The quality of the flow cytometry data in panel G is also not convincing (several cells are out of the depicted area). Panel E and F: the authors should provide information on the cell types that have been gated on.

3. One concern about the T cell killing (panel F) is that the differences could be due to cell growth changes in the VPS OE cells as seen previously. Is there a way to account for/eliminate that in this setting? Check the units on the y axis of the graph. There is no clear effect of neutralizing antibodies (the conditions including MC38 cells + OT-1 cells should be compared).

(Figure 6).

1. The murine counterpart of VPS9D1-AS1 is mentioned in the text, but all experiences in vivo are performed using (human) VPS9D1-AS1. The rationale for not using NR045849 in wild-type mice should be better justified.

2. Panel B: are there differences in vitro in the growth of control and VPS OE MC38 cells? This should be indicated to better understand the in vivo phenotype of these cells.

3. In general, the flow cytometry data are not convincing. Some cell populations are squished on the top of the graph, there is no indication of what cells are gated on (no gating path or gating strategy indicated). Have isotype controls (or FMO) been used for PD1? The label of the X axis in E should be revised (2 markers on the same axis). If Figure 6D tumor left panel: CD8 numbers are close to 0; therefore, I am not sure how confident one can be about %PD1 in Figure 6F.

(Figure 7)

1. There should be an indication of the phenotype of VPS9D1-AS1 transgenic mice at steady-state; are these animals normal? And is VPS9D1-AS1 indeed only expressed in epithelial cells (this is not clear from FigS9B).

2. "However, Ifnar1 levels in tumor tissues were increased in VPS Tg mice (Figure 7H), indicating that the tumor-promoting effect of VPS9D1-AS1 OE was dependent on Ifnar1 expression": this is a strong statement or a conclusion that is not supported by the current data. Only a blockade of IFNAR (pharmacologically or genetically) would actually prove this statement.

3. What are the units on the Y axis for G? and how were the cells quantified?

4. Panel H: there seem to be differences in the quality of these stainings. It would be helpful to show healthy intestinal tissue (note that the current magnification does not allow to discern well the malignant tissue). How were IFNARI levels quantified? Note that the same comments also apply to panel N.

5. I think the author over-interpret the panel K data; again (semi-)quantitative densitometry would help to support their statements.

The authors do not discuss how the T cells and the VPS9D1-AS1 expressing tumors interact in vivo. Is this also mediated via TGFb (note that this is not shown)? And how do they explain that IFNARI becomes increased on T cells in the tumor micro-environment?

As mentioned in the public review section, the paper could be better served in several ways:

(1) The paper could be split into two parts. The paper has plenty of data to support the hypothesis of VSP importance in cancer with a less exhaustive look at the binding partners/etc. In particular, the data on downstream miRNAs (Figure 3O, currently not convincing) would benefit from a more in-depth analysis, possibly in a follow-up study that may as well include some of the data presented in Figure 4.

(2) The data could be slimmed.

A few additional suggestions:

– Introduction: the already reported roles of VPS9D1-AS1 for cancer in general and colon cancer, in particular, could be presented in greater detail.

– Figure 3A: For non-specialists, it would be helpful to indicate which cells originate from CRC tissue.

– Figure 3: why swap the colors from F and G? It's confusing.

– Figure 3: K – the units seem to be wrong in the y axis.

– Sfig4G: data on SW480 are too faint to be assessed.

– Fig6 and Fig7: these data are not from xenograft experiments since murine cells were injected into mouse recipients (of the same haplotype).

– Fig7 panel C (and several of the histology shown): larger magnification should be provided and grading by a pathologist would be beneficial.

– Overall, the data are very dense as presented. Their presentation could be improved to enhance readability.

Reviewer #3 (Recommendations for the authors):

1. Figures to present the expression difference of VPS9D1-AS1 between benign and malignant tissue should be consistent in Figure 1D and 1E for better visualization;

2. The P value between VPS 0 and VPS1 was more prominent than that between VPS 0 and VPS 2-4 in Figure 1F, while the Figure 1G only compares the difference between stroma and VPS 4, please explain the reason;

3. In Figure 1I, the difference in the expression level of TGFBR1, SMAD1/9, VPS9D1-AS1, and TGF-β seems not significant between the normal and cancer tissue, which is not consistent with the description in the manuscript;

4. The quantification of immune cell subsets was presented with two staining methods in Figure 2A and 2B, please specify the reason;

5. The number of CD8+ T cells in VPS9D1-AS1 positive tissue overweighs that in VPS9D1-AS1 negative tissue, which is contrary to the description in the manuscript;

6. Tumor pictures would be better provided along with the Figure 3G, as was done in Figure 7I;

7. On page6, line 11: what is your reason for deducing that there was feedback between VPS9D1-AS1 and TGF-β, and what is the impact of this feedback.

8. The data volume is sufficient enough to support most of the conclusions, while the logicality of the Results warrants improvement, so does the diagram;

9. Since IFN plays a double role in suppressing or promoting tumor progression, the specific circumstances when employing IFN or target IFN would better be discussed in the Discussion section for better comprehension.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "VPS9D1-AS1 overexpression amplifies intratumoral TGF-β signaling and promotes tumor cell escape from CD8+ T cell killing in colorectal cancer" for further consideration by eLife. Your revised article has been evaluated by W Kimryn Rathmell (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

1) Figure 2A: the artefact previously mentioned (obvious quadrant-type shapes, possibly from stitching scanned images together) have not been addressed. These images should be rescanned. Also, it is unclear how these issues may have impacted the analysis of their data.

2) Figure 2E: this panel has not changed compared to the last time, and contrary to what the authors claim in their reply. And contrarily to what is written on lines 147-149, they are changes in CD8 T cells between VPS9D1-AS1 positive versus negative samples in this cohort.

3) Figure 3 J and K, (and maybe I). The data from the pull-down assays are not fully convincing as the binding of the probes seems to be arbitrary; there is a large difference for RPD1 versus RPD2 in binding RPS3, while these probes are quite similar. Compared to RPD1, RPD2 is missing 1 bp at the 5' end from 1 but has 15 residues added at the 3' end… What is the rationale for this different but comparable design?

4) Along these lines: the short RPD3 construct is the best binder of all the tested compounds, but RPD4 still binds RPS3 and SMAD even having no overlap with RPD3. These data would need to be refined to be publishable and would be better served by being moved to the other study mentioned.

5) Line 256: this sentence claims that TGFb can enter the nucleus to act as a transcription factor. Is this what the authors mean? They should be more explicit in the text. If this is not what they mean, they should rephrase the text accordingly.

6) Figures 5D and 5G: these flow cytometry data need further improvement to be more convincing. E.g. for Figure 5D/upper panels, the gates to indicate IFNAR1 positivity do not seem to be appropriate. Did the authors use an isotype or FMO control for these data to set their gates (this aspect was not mentioned for IFNAR1 in their reply)? Figure 5D: the axes of the flow plots need to be labelled. Also, why the different stainings for CD4 and CD8 T cells in the top and bottom panels? This seems needlessly confusing.

7) Figure 5 Panels E and F: the authors should provide information on the cell types that have been gated on in the graph/axis labels; e.g.: % IFNAR1 of CD4 and CD8, … this would facilitate the understanding of these data. In general, it would be much better and informative to distinguish the expression of these molecules on CD8 and CD4 T cells taken individually (and not together), in particular, if they want to make the point that PD1/PD-L1 interaction has a functional impact on CD8 + T cells in their model. Are the values plotted in panel F directly taken from the FACS plots in panel E (it should be the case to better compare the data)?

8) Figure 5G: to make clearer which cells have which compounds added, I would increase the font a bit, and also add the labels to the bottom panels. Are the authors sure they didn't accidentally switch any of the bottom panels? The percent values indicated in the quadrant do not seem to fit the data plots. Why do the authors detect 2.45% of IFNγ+ OT-1 cells in naïve / unstimulated conditions?

9) Text lines 296-300 ("VPS9D1-AS1 OE reduced the cytotoxicity of activated OT-1 CD8+ T cells"): my interpretation of figure 5F differs from the authors. I see the only differences in survival of cancer cells as being due to increased growth of VPS9D1-AS1 OE MC38 cells (as also clearly shown in Figure 6B). This is still a major issue for the paper. I don't think the authors have adequately addressed that the increased tumorigenic potential of VPS9D1-AS1 OE could just be due to enhanced growth, not immune evasion, as posited by the authors. In addition, and as previously mentioned, the effect of PD1 and TGFb blockade are not convincing; the significance appears to come from the higher variation of the data in the condition w/o blockade (and did they add an isotype control there?).

10) Figure 7: There is still no indication in the manuscript methods of the phenotype (and not genetic background) of VPS9D1-AS1 transgenic mice at steady-state / in untreated conditions. In other terms: are these animals entirely normal? i.e. do they show a weight, lifespan, reproductivity, etc that are comparable to their non-transgenic counterparts?

11) Figure 7E: Did 80% of the transgenic mice really die by week 12 of this experiment? And why do in Figure 7L / M transgenic mice in the ASO-NC group start dying at around 10w of AOM/DSS treatment, while there is no death in the ASO-VPS group, and considering the ASO-NC or ASO-VPS treatment only starts from week 14 of AOM/DSS treatment on?

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "VPS9D1-AS1 overexpression amplifies intratumoral TGF-β signaling and promotes tumor cell escape from CD8+ T cell killing in colorectal cancer" for further consideration by eLife. Your revised article has been evaluated by W Kimryn Rathmell (Senior Editor) and a Reviewing Editor.

Several of the previous comments have still not been adequately addressed, but there are still several issues to be resolved, which are listed below. The same numbering is kept to facilitate the reading.

2) Figure 2E: this panel has still not changed compared to the last time, and contrarily to what the authors claim in their reply. Positive sample (please correct spelling in the Figure ) are in red in this Fig2E and show increased numbers of CD8 and CD4 lymphocytes (this is not in line with the text on line 148-149), while FigD indicates increases Ca/STM ratios for CD8 lymphocytes in negative cases (displayed in red).

4) Again: RPD1 and RPD2 are very much comparable, yet their respective ability to binds RPS3 is quite different. In addition, RPD2 entirely overlaps with the sequence of RPD4, yet it binds RPS3 with reduced efficacy. This, together with my previous comments raises concern on the robustness of these data, which the authors did not discuss.

5) There are no definitive date proving TGF-β co-localization in the nucleus; the staining in Figure 3I is difficult to assess. In addition, I could not find clear evidence in the literature that TGF-β can indeed translocate to the nucleus and I would therefore be cautious before making such statement. Did the authors entirely control for the quality of their reagents (specificity of anti-TGF-β antibody, etc.)?

6) In the data newly provided in their reply, the upper row appears to indicate isotypes controls for CD4 and CD3 (and not PD1, as claimed). For the lower row, apparently showing the isotype control, there is a distinct shift in all cell populations between the isotype staining and the staining with the anti-IFNAR1 antibody. This raises concern on the robustness of these data. Possibly, the antibodies need to be titrated and the gating needs to be readjusted. In addition, in Figure 5D, the data show CD4+ and CD4- cells (upper row) and CD8+ and CD8- cells (lower row). It is formally not correct to assume the CD4- cells are CD8+ T cells and vice-versa, since the authors have not combined these antibodies in the same staining panel.

9) Contrarily to what the authors claim, in their reply, there is in Figure 5F not obvious killing between the conditions with MC38+OT-1 without antibody versus MC38+OT-1 and PD1 or TGF-β antibody-mediated blockade (irrespectively of MC38 cells being used as controls or overexpressing VPS) – there is no p value <0.05 indicated for these comparison groups. In other terms, the authors do not compare the right groups to support their statement (lines 299-300). This needs to be addressed and corrected in the manuscript (lines 299-300).

Please improve the figures appropriately to supplement solid evidence, and provide more logical descriptions in the manuscript to fulfil the requirement of eLife, which would help accelerate the final decision towards your manuscript.

https://doi.org/10.7554/eLife.79811.sa1

Author response

Essential revisions:

1) The mechanistic study regarding VPS9D1-AS1/TGF-β/ISG was mainly carried out in vitro, while this part shall be tested in cell line-derived xenografts or mouse tumors to support the claims.

In our study, we had applied cell line derived xenograft or allograft tumor models (in vivo) to explore the regarding mechanism for VPS9D1-AS1/TGF-β/ISG. Our study had demonstrated that ASO inhibiting VPS9D1-AS1 decreased the levels of TGFBR1 and TGFβ through HCT116 transplanted xenograft tumor. IFNAR1 was one of an important ISG. In our revised manuscript, we further proved that IFNAR1 played a role on regulating tumor cell proliferation.

2) The data volume is large and the methodology has covered a broad range in the work, while the arrangement of the manuscript could be improved to optimize the presentation and enhance the readability.

According to your and other reviewers’ comments, we shorted our manuscript and made it more readability.

3) Some details of the figures conflict with the statement in the manuscript, and corrections are warranted.

We had carefully read our manuscript and revised these errors in conflicted figures.

Reviewer #1 (Recommendations for the authors):

Here are some major concerns:

Figure 1/S2: The authors shall explain or discuss why FOXP3 T cells showed a different trend in OUTDO and BJCYH cohorts.

OUTDO cohort only included the cancer tissue samples to subject multiplex multispectral fluorescence immunohistochemistry (mfIHC) analysis, cancer tissues contained tumor parenchyma and tumor stroma, we compared the levels of T cells between cancer tissues and tumor mesenchymal tissues. BJCYH cohort enrolled tumor tissues and normal tissues for mfIHC investigation, and compared the levels of T cells. These reasons caused a different trend for FOXP3 T cells in OUTDO and BJCYH cohorts.

Figure 2: Which cohort(s) was analyzed in 2F? Did OUTDO and BJCYH show the same or different patterns?

In 2F, OUTDO cohort was used to analyze the relationships between VPS and TGFβ signaling genes. We used IHC to stain TGFBR1 to represent TGFβ signaling in BJCYH. The correlation between VPS and TGFBR1 was calculated in ‘Figure 4K’ to illustrate their interrelation with OAS1 and IFNAR1. Both analyses indicated that VPS shown a positive relationship with TGFβ signaling.

Figure 4/S6: 4H-I showed that at 0 hours, OAS1 and IFI27 did not change between CTRL and VPS OE; however, in S6H, OAS1 and IFI27 were much lower in VPS OE, conflicting with 4H-I.

Our results indicate that OAS1 and IFI27 are regulated by VPS, this effect is depending on IFN stimulation (as shown in 4H-I). Once removed IFN in culturing media (S6H), overexpression of VPS would increase both OAS1 and IFI27 expressions. Therefore, these results are no confliction.

Figure 5/S7: In S7A, VPS OE+sgOAS1 (last three lanes) show a substantial OAS1 expression. This made the claims on OAS1 (5B) less convincing. How did VPS KO cells impact IFNAR1 level of T cells in coculture?

In control (CTRL) cells, we demonstrated that sgOAS1 deleted the expression of OAS1 proteins. The S7A result indicate that VPS OE (overexpression) prevent the OAS1 downregulation caused by sgOAS1.

We detected IFNAR1 levels of T cells after co-culturing with VPS OE or sgVPS cells, and proved that IFNAR1 would be effect by these cells. For example, down regulation of IFNAR1 in tumor stroma stimulated CRC development and growth, this finding has been proved by many studies (Inactivation of Interferon Receptor Promotes the Establishment of Immune Privileged Tumor Microenvironment, 2017, Cancer Cell). However, genetic elimination of IFNAR1 in tumor cells enhanced immunological response depended on CD8+ T cells and was more susceptibility to T cell-meditated kill (Type I IFN protects cancer cells from CD8+ T cell mediated cytotoxicity after radiation, 2019, JCI). Cancer-specific IFNAR1 engagement has been demonstrated to promote cancer stemness and higher expression levels of PDL1 (Cancer-specific type-I interferon receptor signaling promotes cancer stemness and effector CD8+ T-cell exhaustion, 2021, Oncoimmunology). Collectively, there are many studies indicate that IFANR1 exert dichotomous function in cancer cells and lymphocytes. Our present study demonstrates that IFNAR1 could be upregulated by VPS in cancer cells, these cells with low IFNAR1 are more susceptible to CD8+ T cell killing. On the other hand, we found that cancer-derived IFNAR1 exhibit a negative relationship with CD8+ T cells IFNAR1. In this study, we did not dig out the detail mechanism. We hypothesized that IFNAR1 overexpressed in tumor cells regulated many ISGs and then released by tumor cells in TME, which would regulate IFNAR1 in CD8+ T cells. This mechanism will be explored in our future study.

Figures 6 and 7: The mechanistic model proposed by in vitro studies (Figures 4 and 5) shall be tested in cell line-derived xenografts or mouse tumors. For example, the IFNAR1 level of VPS OE tumors was examined (7H) but the IFNAR1 level of infiltrated T cells was not. Moreover, CRC patient-derived xenografts shall be used and treated with VPS ASO, which will be more physiologically relevant than the cell line with 600-fold VPS upregulation used in the study.

Our study provides an idea that cancer cell-derived IFNAR1 promotes tumor progression and suppresses anti-tumor immunological reaction in TME. In contrast, these cancer cells with high levels of IFNAR1 are more resistant to CD8+ T cell killing, once they contact with CD8+ T cells will suppress their IFNAR1 expression. In other word, IFNAR1 mediates the crosstalk between tumor cells and CD8+ T cells.

To further validate mechanism regarding VPS promoting tumor through IFNAR1, we downregulated IFNAR1 in CRC cells with VPS overexpression. Our data indicated that IFNAR1 downregulation reversed the promotion for VPS OE on tumor progression.

Moreover, we had proved that VPS was one of drug target and plan to validate this hypothesis using patient-derived xenografts in future study. We thus did not discuss related result in present study.

Reviewer #2 (Recommendations for the authors):

My comments are detailed below:

(Figure 1)

1. Panels A and B (and S1A) are not useful for the conclusions of the paper – the histology is difficult to distinguish and it is not clear of levels (of expression) or frequencies (of positive cells) were quantified in panel B. Those data are better presented in panel D.

We revised this figure to easier understand. Our study applied both RNAscope and qRT-PCR assays to detect the VPS overexpression in CRC patients. In new figure 1, A shows the result of RANscope, we respectively compared their difference expressions in colon cancer and rectal cancer. Panel B shows the data of qRT-PCR, we demonstrated that VPS levels were significantly higher in cancer tissues than their levels in normal tissues. The lower figures of panel B represent the results of paired tissue samples for both BJCYH and OUTDO cohorts.

2. Panel F, for instance, doesn't actually corroborate what is indicated in panel A, as the authors already gate on "high" populations, and there are still some samples with almost no VPS9D1-AS1, this suggests the staining protocol is not a good replacement for qRT PCR.

RNAscope technique enabled direct counting of mRNA molecules in single cells in routine formalin-fixed tissue specimens using bright-field microscopy. RNAscope was considered to superior to real time qRT-PCR because the false negative results obtained from admixtures of many non-malignant cells can be overcome. Thus, our study applied both RNAscope and qRT-PCR to validate the overexpression of VPS using either formalin-fixed CRC tissues or fresh cryopreserved CRC tissues. Panel F tried to mutual confirmation the result derived from two assays. This analysis generally reaches a consensus that VPS is overexpressed in CRC cancer tissues. To more concise, we would like to delete panel F.

3. In the text the authors write that VPS expression doesn't correlate with the expression of genes in the TGFb pathway (pp4, lines 18-20), but do not actually separate by gene expression, but rather by cancer and normal tissue. A better analysis may actually be a xy graph with regression curves.

According this suggestion, Pearson correlation analysis was instead of heatmap clustering to investigate their relationships between VPS and TGFβ signaling genes on mRNA levels. This analysis revealed that VPS9D1-AS1 was positive associated with TGFB1 on mRNA levels (as shown in Figure 1F). Because five variables were enrolled these correlation analyses, we used R package ‘corrplot’ to plot correlated coefficients.

(Figure 2)

1. There seem to be artefacts or resolution issues in panel A; in panel B, the inlays seem to have different magnifications (size bars are not provided).

mfIHC tissue figures in Panel A were obtained using a Vectra Polaris Imaging System (Perkin Elmer) through 10 X objective lens (100X magnification). Upper pictures show the whole core for each tissue. We enlarged pictures (down) at region of interesting. The size bars were added.

IHC pictures in Panel B were obtained using a white light microscope coupled with scanning system. We show the pictures with 90x magnification for each core. Down pictures were obtained at 400x magnification.

2. It seems data were mixed up data in panel E; in addition, panel D shows Ca/STM ratios while panel E indicates cell numbers, making it difficult to compare the two cohorts. The same applies to Sfig2 panel C versus F.

We sorry for our mistake in panel E, and had revised this error. In panel D, this analysis enrolled cancer tissue samples (OUTDO) using mfIHC method, in contrast, panel E enrolled both cancer and paired normal tissues (BJCHY) using IHC method. mfIHC assay allow us to separate tumor tissues into cancerous tissue and tumor stroma tissue using InForm software. We thus compared the ratios between T cell number in cancerous (Ca) tissues and T cell number in tumor stroma (STM) tissues. For BJCYH, we calculated total number of T cells either whole cancer tissue or paired normal tissue.

3. Panel F needs to be more clearly explained, it is difficult to interpret these data with the explanation given and certain p values are missing. The same applies to SFigure panel G.

According this suggestion, Panel F and S2 panel G had been revised to clearly explain the relationships between TILs and proteins levels involving TGFβ signaling. P values were represented as follows: *P<0.05, **P<0.01, ***P<0.001.

4. Panel A doesn't show what is indicated in the text (cancer vs stromal pp 5 lines 2-3). It shows neg. vs pos. according to the figure and it is difficult to distinguish healthy from tumor tissue.

As mentioned in above explanation, 2A and S2C show the different levels of these T cells between cancer tissues and tumor stroma tissues in OUTDO cohort. 2B and S2D compared the different levels of T cells between cancer tissues and paired normal tissues in BJCYH cohort. The neg. and pos. referred to the expression of VPS in cancer tissues. T cells were compared between CRC patients with neg. VPS in cancer tissues and CRC patients with pos. VPS in cancer tissues (2D and E).

(Figure 3)

1. Is SMAD1/5/9 in Figure 3D indeed the top band?

We used an antibody targeting the complex of SMAD1, ~5, ~9. In this picture (actually, 3C), western blotting analysis indicated VPS downregulation decreased the levels of SMAD1/5/9. In Figure 1F, clinical samples analyses indicated that VPS was positively associated with SMAD1 mRNA, other than SMAD9. Besides, we detected the levels of SMAD1, SMAD5, and SMAD9 using specific antibodies for each protein respectively (data not shown), and found SMAD1 was more effect by VPS knockout. Therefore, we inferred that top band should be SMAD1.

2. In panels E and M (and several other western blot data), automated quantification would help support the statements by the authors on increased / lower expression of specific proteins, which are not always convincing as stated.

According this suggestion, we quantified the western blot band for each protein in panels E and M as well as other western blot data in our revised manuscript.

3. L – the oligonucleotide design is not fully clear to me; how can these differences between probes be justified, e.g. between RPD1 and RPD2? Why not compare to the entire VPS9D1-AS1 molecule?

The strategy for designing RNA pull down probes was summarized in S4A. In 3’ end of VPS9D1-AS1 sequence, there are many of repetitive sequence, which increased the difficult to transcript the mRNA of VPS9D1-AS1. In our study, RPD1 and RPD2 were designed to amplify the whole length of VPS9D1-AS1 without repetitive sequence, while RPD3 and RPD4 were used to amplify 400~500bp fragments from 5’ to 3’. The amplified DNA fragments for these probes were then transcript into mRNA using in vitro transcript kit, these mRNAs were further labeled with desthiobiotinylation for RNA pull down assay. This assay indicated that VPS9D1-AS1 bound with RPS3, TGF-β, TGFBR1, SMAD1/5/9 at different sites.

4. O – these data are not convincing.

We had deleted miRNA-related data in our new manuscript according suggestions from you and other reviewers. The mechanism regarding VPS acted as ceRNA will be further discussed in our future study.

Figure 4

1. In panel E it seems the labels have been swapped.

In panel E we made the wrong labels for IFI27 and OAS1, we are sorry for that. We had revised them.

2. In panel F: there are differences in the two cell lines after VPS-KO; are the labels correct?

HCT116 cells shown higher VPS9D1-AS1 levels than SW480 cells, might be the reasons of genetic background. HCT116 was derived from microsatellite instability CRC patient, in contrast, SW480 was derived from microsatellite stability CRC patient. Our western blot detection demonstrated that pSTAT1 expression was depending on IFN stimulation in SW480 cells. Deleting VPS9D1-AS1 resulted STAT1 disappear in HCT116, but only slightly lowered STAT1 in SW480 cells. Besides, this detection revealed that STAT1/pSTAT1 pathways shown different response to IFN stimulation. To clearly illustrate our viewpoint, we would like to delete the result of SW480.

3. Panel J and references indicated in the text: I could not find in the literature evidence for a nuclear function of TGFb.

Both canonical and noncanonical TGFβ pathway has been well characterized. SMAD4 acted as the transcript factor is the canonical pathway for TGFβ signaling regulating targeting genes. The noncanonical pathway regarding TGFβ activated a series of targeting genes through AKT, mTOR, NF-κB et al. (as mentioned in Reference Derynck, R., Turley, S. J., and Akhurst, R. J. (2021). TGFbeta biology in cancer progression and immunotherapy. Nat Rev Clin Oncol 18, 9-34.). However, the roles regarding tumor-derived TGFβ overexpression are remained to be elucidated. One possible clue for TGFβ controlling OAS1 transcription might be that tumor-derived TGFβ activated AKT, in tumor cells, newly synthesized latent(L)-TGFβ crosslinked to GARP controlled TGFβ signaling (Overcoming TGFβ- mediated immune evasion in cancer, 2022, Nature Reviews Cancer). GARP has been proved to localize in nuclear compartments (GARP as an Immune Regulatory Molecule in the Tumor Microenvironment of Glioblastoma Multiforme, 2019, Int J Mol Sci).

4. In panel K: how were the decimal values for OAS1, IFNAR1, and TGFBR1 determined?

The levels of OAS1, IFNAR1, and TGFBR1 were scored as follows: 1 (0–25%), 2 (26–50%), 3 (51–75%), and 4 (>75%). The intensity of positive staining was classified into 4 scales as follows: 0 (negative), 1 (weak), 2 (moderate), and 3 (strong). The levels were semiquantitatively determined as percentages multiplied by intensity. Therefore, there were decimal values for these proteins.

(Figure 5)

1. Panel B: some of the samples seem to have been swapped. Please cross-check.

We recomposed FCM plot panel to show the effects of OAS1 on regulating IFNAR1. Three sgRNAs were used to OAS1 knockout (as shown in S7A). In the sgOAS1 cells, IFNAR1 levels were found to be downregulated. On the other hand, VPS OE enhanced IFNAR1 expression. In 5B, FCM assays were carried out to detect the expressions of IFNAR1. To briefly describe these results, we calculated the difference between CTRL and sgOAS1 cells, as well as VPS OE and VPS OE sgOAS1 cells. All these analyses were based on the results of sgOAS1-1 cells.

2. Panel D: How were patients selected to be either VPS neg. or pos.? This is also relevant due to the apparent intra-tumor heterogeneity of VPS9D1-AS1 expression. FACS plot, VPS positive condition, upper and lower rows: both T cell and non T cell populations are shifted down, respectively up, suggesting a problem or changes in the settings of the flow cytometer instrument. The quality of the flow cytometry data in panel G is also not convincing (several cells are out of the depicted area). Panel E and F: the authors should provide information on the cell types that have been gated on.

Panels D, blood samples and tissues samples were obtained from same patients. RNAscope detected the expression of VPS9D1-AS1 in BJCYH cohort, and defined the patients with VPS neg. or pos. FACS plot had been changed to depict the way to calculate the frequencies of IFNAR1 or PD1 in both CD4 and CD8 T cells.

Panel G, the FACS plots were shown using Pseudocolor type with ‘Smoothing’ model. At new figure, Panel G of the FACS plots were shown without ‘Smoothing’ model.

Panel E and D are the results of FCM detection IFNAR1 and PD1 of CD4 and CD8 T cells. Panel F and G show the data of in vitro OT-1 CD8+ T cells killing assays.

3. One concern about the T cell killing (panel F) is that the differences could be due to cell growth changes in the VPS OE cells as seen previously. Is there a way to account for/eliminate that in this setting? Check the units on the y axis of the graph. There is no clear effect of neutralizing antibodies (the conditions including MC38 cells + OT-1 cells should be compared).

Method for T cell killing assays was summarized in ‘supplementary methods’. MC38 CTRL and VPS OE cells were respectively cocultured with OT-1 CD8+ T cells for 24 hours. Neutralizing antibodies were mixed in these cultured media. Then, OT-1 CD8+ T cells were collected to FCM detection. The remaining MC38 cells were further cultured until to the third days for quantifying cell number. MC38 CTRL was used as control and defined as 1.00, the rest of groups were compared with MC38 CTRL group. The y axis shows the ratios for these analyses.

(Figure 6).

1. The murine counterpart of VPS9D1-AS1 is mentioned in the text, but all experiences in vivo are performed using (human) VPS9D1-AS1. The rationale for not using NR045849 in wild-type mice should be better justified.

VPS OE in MC38 cells promoted the expression of TGFβ, TGFBR1, SMAD1, and STAT1, indicated that VPS had similar biological functions between murine cells and human cells. Until to now, no homologous murine gene of VPS was reported. We found NR045849 located similar chromosomal gene location with VPS, but NR045849 was not the homologous gene of VPS. Our western blot analyses indicated that NR045849 had similar biological functions. NR045849 promoted the expression of TGFBR1, SMAD1 and STAT1 in MC38 cells, but not TGFβ, indicated that NR045849 could not instead of VPS. In this context, NR045849 used as the biological control to highlight the biological role of VPS in MC38 cells.

2. Panel B: are there differences in vitro in the growth of control and VPS OE MC38 cells? This should be indicated to better understand the in vivo phenotype of these cells.

CCK8 cell proliferation assays were performed to analyze the growth of MC38 CTRL and VPS OE cells, indicated that VPS significantly promoted MC38 cell proliferation (6B).

3. In general, the flow cytometry data are not convincing. Some cell populations are squished on the top of the graph, there is no indication of what cells are gated on (no gating path or gating strategy indicated). Have isotype controls (or FMO) been used for PD1? The label of the X axis in E should be revised (2 markers on the same axis). If Figure 6D tumor left panel: CD8 numbers are close to 0; therefore, I am not sure how confident one can be about %PD1 in Figure 6F.

We provide the original FCM plots to show the gating strategy. During the FCM detection, an isotype control for each staining antibody was used, included anti-CD3-FITC, anti-CD8-PE, anti-PD1-APC.

Panel D, our FCM detection indicated that little infiltrated CD8+ T cell were found in some samples of VPS OE tumor (the CD8+ T cell frequencies for VPS OE tumor are 34.7, 2.66, 0.22, 1.5, 5.31, 25.5, 28.8, respectively).

Panel F, the labels for y axis had been changed to CD4/8+ PD1 (%), represent the PD1 frequency in CD4/8+ T cells.

(Figure 7)

1. There should be an indication of the phenotype of VPS9D1-AS1 transgenic mice at steady-state; are these animals normal? And is VPS9D1-AS1 indeed only expressed in epithelial cells (this is not clear from FigS9B).

VPS9D1-AS1 transgenic mice were founded based on C57BL/6 background. All mice used in the experiments were backcrossed to C57BL/6 mice for at least 6 generations for maintaining mice at steady-state. This statement has been added in our revised manuscript. In our study, conditional VPS9D1-AS1 transgenic (VPS Tg) mice were produced through crossed with Cre-villin C57BL/6 mice which allowed VPS9D1-AS1 gene expressed in the intestinal epithelium.

To prove the epithelial location of VPS, we performed RNA FISH assay and found that VPS mainly expressed in the part of intestinal epithelium tissue. The related Figure S9B had been revised and replaced with new pictures.

2. "However, Ifnar1 levels in tumor tissues were increased in VPS Tg mice (Figure 7H), indicating that the tumor-promoting effect of VPS9D1-AS1 OE was dependent on Ifnar1 expression": this is a strong statement or a conclusion that is not supported by the current data. Only a blockade of IFNAR (pharmacologically or genetically) would actually prove this statement.

We agree with reviewer’s comment about IFNAR1. We deleted this statement. We also tried our best to discovery the mechanistic relationships between VPS and IFNAR1 on regulating cell proliferation. Our results shown that VPS OE with IFNAR1 knockdown cells show a lower proliferation rate than VPS OE cells, indicated that IFNAR1 play a role on VPS promoting tumor cell proliferation.

3. What are the units on the Y axis for G? and how were the cells quantified?

Y axis represent the CD8+ T cells count. To calculate the CD8+ T cells count in mice cancer tissues, immunofluorescence (IF) assays were performed to stain CD8+ T cells. For each tissue, 3~5 region-of-interest (ROI) pictures were recorded and manual counted using ImageJ IHC tool, the mean value was represented the CD8+ T cell count.

4. Panel H: there seem to be differences in the quality of these stainings. It would be helpful to show healthy intestinal tissue (note that the current magnification does not allow to discern well the malignant tissue). How were IFNARI levels quantified? Note that the same comments also apply to panel N.

We provided the normal tissues pictures for panel H and N. The IFNAR1 levels were quantified using ImageJ IHC tool, the mean values for 3~5 pictures at 200X magnification were calculated for each mouse.

5. I think the author over-interpret the panel K data; again (semi-)quantitative densitometry would help to support their statements.

We had semi-quantitative the western blot results.

The authors do not discuss how the T cells and the VPS9D1-AS1 expressing tumors interact in vivo. Is this also mediated via TGFb (note that this is not shown)? And how do they explain that IFNARI becomes increased on T cells in the tumor micro-environment?

Reviewer proposed a key issue about VPS9D1-AS1. In this study, our in vivo models support the idea that VPS OE prevents CD8+ T cell infiltration. IFNAR1 expressed in T cells are related with its antitumor function, many studies had proved this viewpoint. To solve this issue, further studies need to apply TGFβ knockout mice or IFNAR1 knockout mice. We would like to explore their relationships in future studies.

As mentioned in the public review section, the paper could be better served in several ways:

1) The paper could be split into two parts. The paper has plenty of data to support the hypothesis of VSP importance in cancer with a less exhaustive look at the binding partners/etc. In particular, the data on downstream miRNAs (Figure 3O, currently not convincing) would benefit from a more in-depth analysis, possibly in a follow-up study that may as well include some of the data presented in Figure 4.

Accordingly, we short our manuscript. Downstream miRNA related findings were deleted and will be report in our next study. We also removed some results from Figure 4 to make our manuscript more concise and easier to understand.

2) The data could be slimmed.

Our manuscript had deleted some data and been slimmed.

A few additional suggestions:

– Introduction: the already reported roles of VPS9D1-AS1 for cancer in general and colon cancer, in particular, could be presented in greater detail.

Recent advances about VSP9D1-AS1 had been summarized in the part of ‘Introduction’ and the related references were cited.

– Figure 3A: For non-specialists, it would be helpful to indicate which cells originate from CRC tissue.

The source for these cell lines had been noted in figure legends.

– Figure 3: why swap the colors from F and G? It's confusing.

The color for panels F and G had been revised the consistent color.

– Figure 3: K – the units seem to be wrong in the y axis.

The y axis notes had been corrected. The y axis shows the percentages for intended RIP assay relative to VPS9D1-AS1 in InPut.

– Sfig4G: data on SW480 are too faint to be assessed.

The data related to SW480 had been removed in Figure 4.

– Fig6 and Fig7: these data are not from xenograft experiments since murine cells were injected into mouse recipients (of the same haplotype).

We had revised these related statements in revised manuscript.

– Fig7 panel C (and several of the histology shown): larger magnification should be provided and grading by a pathologist would be beneficial.

The related pictures had been enlarged. All these IHC slices had invited a pathologist to view and confirm.

– Overall, the data are very dense as presented. Their presentation could be improved to enhance readability.

Thanks for your suggestive comments.

Reviewer #3 (Recommendations for the authors):

1. Figures to present the expression difference of VPS9D1-AS1 between benign and malignant tissue should be consistent in Figure 1D and 1E for better visualization;

We revised Figure 1D and 1E, to consistent show the difference of VPS9D1-AS1 between benign and malignant tissue in BJCYH and OUTDO cohort, we compared their levels in all normal and cancer tissues, and further selected paired normal and cancer tissues from same patients in two cohort to compare VPS9D1-AS1 levels.

2. The P value between VPS 0 and VPS1 was more prominent than that between VPS 0 and VPS 2-4 in Figure 1F, while the Figure 1G only compares the difference between stroma and VPS 4, please explain the reason;

We tried to compare the VPS9D1-AS1 levels detected using RT-qPCR methods among the RNAscope stained VPS9D1-AS1 with value 0, 1, and 2~4, 0 and 1 represented negative VPS9D1-AS1 expression, 2, 3, 4 represented positive VPS9D1-AS1 expression. This analysis might be confused for reviewers, we deleted this picture in revised manuscript.

3. In Figure 1I, the difference in the expression level of TGFBR1, SMAD1/9, VPS9D1-AS1, and TGF-β seems not significant between the normal and cancer tissue, which is not consistent with the description in the manuscript;

According to your suggestion and other reviewer’s comments, we further analyzed the correlations these genes on mRNA levels using Pearson correlation methods.

4. The quantification of immune cell subsets was presented with two staining methods in Figure 2A and 2B, please specify the reason;

Multiplex multispectral fluorescence immunohistochemistry (mfIHC) assay was carried out to stain immune T cell subsets in OUTDO cohort and IHC assay was performed to show these cells in BJCYH cohort. We intent to cross-validation by using these methods, because the results of mfIHC assay were analyses using Inform software, while immune T cell counts of IHC staining were recorded based on IHC graphs using ImageJ IHC software. Both mfIHC and IHC analyses indicated that VPS9D1-AS1 prevented CD8+ T cell tumor infiltrations.

5. The number of CD8+ T cells in VPS9D1-AS1 positive tissue overweighs that in VPS9D1-AS1 negative tissue, which is contrary to the description in the manuscript;

Reviewer might refer to Figure 2F. In this figure, the negative and positive VPS were wrongly labeled, we had revised these errors. We are sorry for that.

6. Tumor pictures would be better provided along with the Figure 3G, as was done in Figure 7I;

Accordingly, xenograft tumor pictures for Figure 3G were added.

7. On page6, line 11: what is your reason for deducing that there was feedback between VPS9D1-AS1 and TGF-β, and what is the impact of this feedback.

We found that VPS9D1-AS1 interacted with TGFβ, TGFBR1, SMAD1/5/9 and enhanced their levels. On the other hand, siRNAs targeting these genes inhibited the expression of VPS9D1-AS1. Therefore, we deduced that there is a feedback pathway between VPS9D1-AS1 and TGFβ signaling. This feedback plays a role on maintaining the balance of TGFβ signaling pathway. In tumor cells, high activation of TGFβ signaling pathway prevents T cell killing, high levels of VPS9D1-AS1 help to hold the activation of TGFβ signaling pathway. Once inactivation TGFβ signaling pathway, VPS9D1-AS1 would be downregulated, thus, the inhibition would enable tumor cell to be with low levels of TGFβ signaling pathway and cleared by T cells.

8. The data volume is sufficient enough to support most of the conclusions, while the logicality of the Results warrants improvement, so does the diagram;

Thanks for your positive comments, we had revised to make our manuscript more logical and easier understand.

9. Since IFN plays a double role in suppressing or promoting tumor progression, the specific circumstances when employing IFN or target IFN would better be discussed in the Discussion section for better comprehension.

We mentioned that IFN pathway plays contradictory roles in tumor cells and T lymphocytes. Which indicated that some tumor derived ISG genes overexpression involved in IFN pathway promoting cancer progression, such as OAS1. On the other hand, ISG gene expressed by T lymphocytes, such as IFN-γ and IFN-α/β,inhibited tumor progression. We had discussed these roles in the part of ‘introduction’ (line 9-20, page 3) and ‘discussion’ (line 8-24, page 12) in our manuscript.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

1) Figure 2A: the artefact previously mentioned (obvious quadrant-type shapes, possibly from stitching scanned images together) have not been addressed. These images should be rescanned. Also, it is unclear how these issues may have impacted the analysis of their data.

We rescanned these tissue sample using Perkin Elmer Vetra microscope to avoid quadrant-type shapes and updated represent images in Figure 2A. Inform software was used to our analysis about CD4. CD8, and FOXP3 T cells in CRC tissues. To ensure the accuracy of our data, the CD4. CD8, and FOXP3 T cell frequencies for each tissue was analyzed at one time, this analysis enrolled all regions of each scanned picture.

2) Figure 2E: this panel has not changed compared to the last time, and contrary to what the authors claim in their reply. And contrarily to what is written on lines 147-149, they are changes in CD8 T cells between VPS9D1-AS1 positive versus negative samples in this cohort.

We carefully considered this issue proposed by reviewer. We compared our revised manuscript version with our first submitted manuscript, the errors for Figure 2E had been corrected (we wrongly labeled VPS negative and positive). In Figure 2E, our results indicated that patients with VPS positive expression shown lower levels of infiltrated CD8+ T cells and CD4+ T cells than these patients with negative expression of VPS. This result is consistent with Figure 2D. The color for Figure 2D were revised to consist with Figure 2E for VPS negative and positive. But we did not find the contrarily between our written on lines 147-149 and Figure 2E. Besides, we rephrased this sentence to more rational raise our viewpoint.

3) Figure 3 J and K, (and maybe I). The data from the pull-down assays are not fully convincing as the binding of the probes seems to be arbitrary; there is a large difference for RPD1 versus RPD2 in binding RPS3, while these probes are quite similar. Compared to RPD1, RPD2 is missing 1 bp at the 5' end from 1 but has 15 residues added at the 3' end… What is the rationale for this different but comparable design?

Figure 3J is RIP assay, use antibodies to bind with VPS9D1-AS1. Figure 3K is RNA pulldown assay. The probes used in RNA pulldown assay were four transcripts of VPS9D1-AS1, which were transcribed using an in vitro RNA Synthesis Kit. Then, the RNA products were labeled with desthiobiotinylation. After that, these labeled VPS9D1-AS1 sequences were mixed with cell lysis. The binding proteins were tested using western blotting at last. High repeat sequences of VPS9D1-AS1 result that the transcription of full length VPS9D1-AS1 is very difficult in vitro. To transcribe RNA transcripts of VPS9D1-AS1 from cDNA template, two paired primers (RPD1 and RPD2, 1bp is different for upper primer) were used to transcribe the long probes, which is similar to the full length of VPS9D1-AS1. RPD1 and RPD2 share some same sequence with RPD3.

4) Along these lines: the short RPD3 construct is the best binder of all the tested compounds, but RPD4 still binds RPS3 and SMAD even having no overlap with RPD3. These data would need to be refined to be publishable and would be better served by being moved to the other study mentioned.

Follow up last question, the probes for RPD3 and PRD4 were design to transcribed a short RNA sequence of VPS9D1-AS1 to identify the difference binding ability between VPS9D1-AS1 and targeted proteins. Four probes can bind with RPS3. RPD3 shown higher binding ability with RPS3, TGF-β and SMAD1/5/9. RPD2 show a higher binding ability with TGFBR1. These results indicated that VPS9D1-AS1 located in ribosome of cancer cell by interacting with RPS3 and then regulated the transcriptions of TGF-β, TGFBR1 and SMAD1/5/9 through different sequence. In our study, Figure 3J was designed to proteins binding with VPS9D1-AS1, meanwhile, Figure 3K was designed to VPS9D1-AS1 binding with proteins, each assay complementing each other.

5) Line 256: this sentence claims that TGFb can enter the nucleus to act as a transcription factor. Is this what the authors mean? They should be more explicit in the text. If this is not what they mean, they should rephrase the text accordingly.

According to your suggestion, we rephrase this sentence. The reference (Derynck et al., 2021, TGFbeta biology in cancer progression and immunotherapy. Nat Rev Clin. ) indicated that TGFβ-induced activation of ERK–MAPK, AKT and NF-κB signaling might initiate from the distinct receptor complexes. Our study tried to reveal both SMAD4 and TGF-β acting as regulators of OAS1 gene. SMAD4 has been prove to act as the transcription factor for many genes. Our study indicated that TGF-β might be of direct or indirect transcription factor to control OAS1 expression. IF staining (Figure 3I) indicate that TGF-β can be found to co-locate with nucleus of HCT116 cells. Our ChIP assay further demonstrated that TGF-β-antibody could IP the promotor region of OAS1. This might be refer to some unknown mechanism for TGF-β launching intracellular signaling in cancer cells.

6) Figures 5D and 5G: these flow cytometry data need further improvement to be more convincing. E.g. for Figure 5D/upper panels, the gates to indicate IFNAR1 positivity do not seem to be appropriate. Did the authors use an isotype or FMO control for these data to set their gates (this aspect was not mentioned for IFNAR1 in their reply)? Figure 5D: the axes of the flow plots need to be labelled. Also, why the different stainings for CD4 and CD8 T cells in the top and bottom panels? This seems needlessly confusing.

In the experiments of Figure 5D, isotypes for IFNAR1 and PD1 was carried out to quantitative detect the levels of IFNAR1 and PD1 in blood T lymphocytes. The positive expressed IFNAR1 as well as PD1 were determined according to isotypes for each FCM experiment (as shown in Author response image 1).

The axes labels for Figure 5D had been added.

The antibody panel for quantitative detecting IFNAR1 included CD3-FITC, CD4-APC, and IFNAR1-PE in CD4 and CD8 cells. While, CD3-FITC, CD8-PE, and PD1-APC antibodies consist the panel to quantitatively detect PD1 in CD4 and CD8 cells. Therefore, the FCM dots for these experiments were different.

Author response image 1
Isotype for IFNAR1 in CD4 and CD8 T cells.

7) Figure 5 Panels E and F: the authors should provide information on the cell types that have been gated on in the graph/axis labels; e.g.: % IFNAR1 of CD4 and CD8, … this would facilitate the understanding of these data. In general, it would be much better and informative to distinguish the expression of these molecules on CD8 and CD4 T cells taken individually (and not together), in particular, if they want to make the point that PD1/PD-L1 interaction has a functional impact on CD8 + T cells in their model. Are the values plotted in panel F directly taken from the FACS plots in panel E (it should be the case to better compare the data)?

According to your suggestion, we had revised the labels for Figure 5E.

In Figure 5F, we employed an in vitro Cytotoxicity Assay to test the cytotoxicity CD8+ OT-I cells on killing MC38 cancer cells. This is one kinds of pattern recognition model, which MC38 cells were transfected by OVA vector. OVA is the antigen can be recognized by CD8+ OT-I cells. In panel F (right), the remaining cell number were calculated and compared their ratios (relative to group MC38-CTRL). The results for Panel F are not taken from FACS plots.

8) Figure 5G: to make clearer which cells have which compounds added, I would increase the font a bit, and also add the labels to the bottom panels. Are the authors sure they didn't accidentally switch any of the bottom panels? The percent values indicated in the quadrant do not seem to fit the data plots. Why do the authors detect 2.45% of IFNγ+ OT-1 cells in naïve / unstimulated conditions?

Accordingly, the fonts for Figure 5G were enlarged. We did not switch the bottom panels. In this FCM detection experiment, OT-I cells were collected and staining with CD8-PE and IFN-γ-APC antibodies. We used isotype for CD8-PE to gate OT-I cells, to distinguish mixed MC38 cells in cultured media.

The right panel of 5G analyzed the IFN-γ levels (relative to untreated OT-I) in different groups.

The OT-1 cells were separated from OT-1 mice spleen and subjected to the media supplied IL-1 and CD3/CD28 antibodies to prime. After this treatment, the expression of IFN-γ should be activated, therefore, we could detect IFN-γ in these OT-I cells.

9) Text lines 296-300 ("VPS9D1-AS1 OE reduced the cytotoxicity of activated OT-1 CD8+ T cells"): my interpretation of figure 5F differs from the authors. I see the only differences in survival of cancer cells as being due to increased growth of VPS9D1-AS1 OE MC38 cells (as also clearly shown in Figure 6B). This is still a major issue for the paper. I don't think the authors have adequately addressed that the increased tumorigenic potential of VPS9D1-AS1 OE could just be due to enhanced growth, not immune evasion, as posited by the authors. In addition, and as previously mentioned, the effect of PD1 and TGFb blockade are not convincing; the significance appears to come from the higher variation of the data in the condition w/o blockade (and did they add an isotype control there?).

As shown in Figure 5F, this T cell cytotoxicity assay was used to determine the inhibited ability of CD8+ OT-I cells in killing tumor cell with supplied TGF-β or PD-1 antibodies. In Figure 5F (left), we really observed that CD8+ T OT-I cells killed MC38 CTRL-OVA cells as well as MC38-VPS OE-OVA cells. However, there is no significantly difference on proliferated rates between MC38 CTRL-OVA and MC38-VPS OE-OVA cells after treatment with CD8+ T OT-I. In view of this, we agree with reviewer’ viewpoint and revised this sentence.

We repeated three times for T cell cytotoxicity assay and observed that PD1 antibody and TGF-β antibody supplied in the cultured media significantly enhanced CD8+ T OT-I cells for each experiment. Besides, isotype control was applied to gate CD8+ OT-I cells in Figure 6G.

10) Figure 7: There is still no indication in the manuscript methods of the phenotype (and not genetic background) of VPS9D1-AS1 transgenic mice at steady-state / in untreated conditions. In other terms: are these animals entirely normal? i.e. do they show a weight, lifespan, reproductivity, etc that are comparable to their non-transgenic counterparts?

In our mice models, VPS9D1-AS1 were conditional knock in murine genome, which did not cause any changes in mouse traits (such as lifespan and bodyweight). According to this suggestion, we had described this indication in our manuscript.

11) Figure 7E: Did 80% of the transgenic mice really die by week 12 of this experiment? And why do in Figure 7L / M transgenic mice in the ASO-NC group start dying at around 10w of AOM/DSS treatment, while there is no death in the ASO-VPS group, and considering the ASO-NC or ASO-VPS treatment only starts from week 14 of AOM/DSS treatment on?

In Figure 7E, three cycles of DSS after one week of AOM treatment were treated WT and VPStg mice. In contrast, one cycle of DSS treated mice after one week of AOM treatment in Figure 7L. Therefore, more DSS treatment lead higher death rate in Figure 7E than Figure 7L/M. To view the therapeutic effect of ASO-VPS, we used one cycle of DSS induced tumor in Figure 7L/M. After DSS treatment for one week, ASO-VPS and ASO-NC drugs were applied to treat mice. This would be the reason for ASO-NC group start dying at 10th week.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Several of the previous comments have still not been adequately addressed, but there are still several issues to be resolved, which are listed below. The same numbering is kept to facilitate the reading.

2) Figure 2E: this panel has still not changed compared to the last time, and contrarily to what the authors claim in their reply. Positive sample (please correct spelling in the Figure ) are in red in this Fig2E and show increased numbers of CD8 and CD4 lymphocytes (this is not in line with the text on line 148-149), while FigD indicates increases Ca/STM ratios for CD8 lymphocytes in negative cases (displayed in red).

We deeply sorry for our mistake in Figure 2E and thanks for your kindly comment. At this time, we accordingly revised these figures. Previous figure 2E wrongly labeled ‘positive’ (red) and ‘negative’ (blue) group. We fixed these errors. The labels for Figure 2D were correct, the Ca/STM ratios for CD4/8 lymphocytes were increased in patient with negative VPS9D1-AS1 expression.

4) Again: RPD1 and RPD2 are very much comparable, yet their respective ability to binds RPS3 is quite different. In addition, RPD2 entirely overlaps with the sequence of RPD4, yet it binds RPS3 with reduced efficacy. This, together with my previous comments raises concern on the robustness of these data, which the authors did not discuss.

We firstly response to reviewer previous comment. In view that, the binding VPS9D1-AS1 with TGF-β, TGFBR1, and SMAD1/5/9 had been proved by several assays in our study, as shown in Figure 3I and J. We thus revised RNA-pull down related data and deleted the results of TGF-β, TGFBR1, and SMAD1. According reviewer suggestion, these issues should be moved to another research to be addressed. All related description and figure legend had been consistent revised. Among of all these detecting binding proteins for VPS9D1-AS1, the RNA-pull down assay for RPS3 was better than other protein. In this revised manuscript, we tried to preserve this result for VPS9D1-AS1 binding with RPS3, this would help to hint the subcellular location for VPS9D1-AS1.

About RPD probes concern, as be pointed out by reviewer, four designed probes show different binding ability with VPS9D1-AS1 in our RNA-pull down assay. We thought that the specific repetitive sequences (from 1298bp to 1753bp, total 279 bp) in the 3’ end of VPS9D1-AS1 play an important role during binding with RPS3 and other proteins. The most important difference between RPD1 and RPD2 are the sequences in 3’end, RPD2 contained 13 bp repetitive sequences. On the other hand, RPD3 is the 5’ end transcript of VPS9D1-AS1, shown the highest binding affinity with RPS3, indicated that VPS9D1-AS1 bound with RPS1 through 5’ end of RNA chain. Although RPD3 shared same sequence with RPD1 and RPD2, did not contained repetitive sequences, shown higher binding ability than RPD2, but lower than RPD1 and RPD3, which further indicated the binding mainly occurring in the 5’ end.

5) There are no definitive date proving TGF-β co-localization in the nucleus; the staining in Figure 3I is difficult to assess. In addition, I could not find clear evidence in the literature that TGF-β can indeed translocate to the nucleus and I would therefore be cautious before making such statement. Did the authors entirely control for the quality of their reagents (specificity of anti-TGF-β antibody, etc.)?

According this suggestion, we revised the related statement in our manuscript (line 256~262, page 8~9). Figure 4J and legend were also consistently revised to delete the results of TGF-β. The issue for TGF-β acting as transcription factor might be discussed in future study.

6) In the data newly provided in their reply, the upper row appears to indicate isotypes controls for CD4 and CD3 (and not PD1, as claimed). For the lower row, apparently showing the isotype control, there is a distinct shift in all cell populations between the isotype staining and the staining with the anti-IFNAR1 antibody. This raises concern on the robustness of these data. Possibly, the antibodies need to be titrated and the gating needs to be readjusted. In addition, in Figure 5D, the data show CD4+ and CD4- cells (upper row) and CD8+ and CD8- cells (lower row). It is formally not correct to assume the CD4- cells are CD8+ T cells and vice-versa, since the authors have not combined these antibodies in the same staining panel.

The strategies for gating IFNAR1 and PD1 in T cells were shown as follow picture. Our previous response indicated the method for gating IFNAR1. The lower row represented an isotype sample with staining ISO-PE antibody. On the other hand, right graph shown the CD3/CD4 staining cells. This might be the reason for the shift in cell populations between lower graphs. We further regrouped these FCM graphs and tried to show gating strategies. We used isotype as the negative controls for gating positive IFNAR1 and PD1 cells to represent their levels in CD4 and CD8 T cells. As our mentioned in our previous response, we used CD3/CD4 antibodies and CD3/CD8 antibodies to discern CD4 and CD8 cells, this method allowed us to reduce an antibody for each panel, this might be not rigorous enough. However, CD3/CD4/IFNAR1 antibodies staining could represent the levels of IFNAR1 in CD4 cells, while, CD3/CD8/PD1 antibodies staining could represent the levels of PD1 in CD8 cells. Therefore, we revised our results in Figure 5 D and E. The related statement had also been revised in manuscript.

9) Contrarily to what the authors claim, in their reply, there is in Figure 5F not obvious killing between the conditions with MC38+OT-1 without antibody versus MC38+OT-1 and PD1 or TGF-β antibody-mediated blockade (irrespectively of MC38 cells being used as controls or overexpressing VPS) – there is no p value <0.05 indicated for these comparison groups. In other terms, the authors do not compare the right groups to support their statement (lines 299-300). This needs to be addressed and corrected in the manuscript (lines 299-300).

Current comment help us to more precise understand reviewer’s intent. We labeled the p values for comparison among VPS OE, VPS OE+OT-1, VPS OE+OT-1+TGF-βab, and VPS OE+OT-1+PD1ab groups. TGF-βab and PD1ab supplied in media lowered MC38 VPS OE cell proliferation, but there are no statistically significant difference. We thus revised our statement in manuscript.

https://doi.org/10.7554/eLife.79811.sa2

Article and author information

Author details

  1. Lei Yang

    1. Medical Research Center, Beijing Chao-Yang Hospital, Capital Medical University, Beijing, China
    2. Department of General Surgery, Beijing Chao-Yang Hospital, Capital Medical University, Beijing, China
    Contribution
    Conceptualization, Data curation, Funding acquisition, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    yl6649084@mail.ccmu.edu.cn
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3718-2138
  2. Xichen Dong

    Medical Research Center, Beijing Chao-Yang Hospital, Capital Medical University, Beijing, China
    Contribution
    Validation, Investigation, Methodology
    Competing interests
    No competing interests declared
  3. Zheng Liu

    Medical Research Center, Beijing Chao-Yang Hospital, Capital Medical University, Beijing, China
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
  4. Jinjing Tan

    Department of Cellular and Molecular Biology, Beijing Chest Hospital, Capital Medical University & Beijing Tuberculosis and Thoracic Tumor Research Institute, Beijing, China
    Contribution
    Visualization, Methodology
    Competing interests
    No competing interests declared
  5. Xiaoxi Huang

    Medical Research Center, Beijing Chao-Yang Hospital, Capital Medical University, Beijing, China
    Contribution
    Methodology
    Competing interests
    No competing interests declared
  6. Tao Wen

    Medical Research Center, Beijing Chao-Yang Hospital, Capital Medical University, Beijing, China
    Contribution
    Conceptualization, Funding acquisition, Visualization
    Competing interests
    No competing interests declared
  7. Hao Qu

    Department of General Surgery, Beijing Chao-Yang Hospital, Capital Medical University, Beijing, China
    Contribution
    Conceptualization, Resources, Project administration
    For correspondence
    13701320206@163.com
    Competing interests
    No competing interests declared
  8. Zhenjun Wang

    Department of General Surgery, Beijing Chao-Yang Hospital, Capital Medical University, Beijing, China
    Contribution
    Conceptualization, Resources, Project administration
    For correspondence
    drzhenjun@163.com
    Competing interests
    No competing interests declared

Funding

Natural Science Foundation of China (81802349)

  • Lei Yang

National Science Foundation of China (8213234)

  • Tao Wen

Beijing Natural Science Foundation (7192070)

  • Lei Yang

Beijing Municipal of Hospitals Incubating Program (PX2018013)

  • Lei Yang

Scientific Research Project of Beijing Educational Committee (KM20190025016)

  • Lei Yang

Open Project of Key Laboratory of Cardiovascular Disease Medical Engineering, Ministry of Education (2019XXG-KFKT-03)

  • Lei Yang

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Dr. Jian Liu for providing CRISPR/Cas9 virus packaging vectors and Dr. Suliang Guo for his assistance in animal feeding.

This project was supported by grants from the National Natural Science Foundation of China (81802349, 82173234), Beijing Natural Science Foundation (7192070), Beijing Municipal Administration of Hospitals Incubating Program (PX2018013), Scientific Research Project of Beijing Educational Committee (KM201910025016), and Open Project of Key Laboratory of Cardiovascular Disease Medical Engineering, Ministry of Education (2019XXG-KFKT-03).

Ethics

Human subjects: All sample donors provided informed consent, and the study was conducted under the approval (2018-ke-24) of the Institutional Ethics Committee from Beijing Chaoyang Hospital of Capital Medical University between 2018 and 2020 samples were collected from patients with CRC who did not receive chemotherapy or radiotherapy before surgery.

Animal experimental protocols were approved (AEEI-2021-105) according to the guidelines of the Ethics Committee for Animal Testing of Capital Medical University.

Senior Editor

  1. W Kimryn Rathmell, Vanderbilt University Medical Center, United States

Reviewing Editor

  1. Caigang Liu, Shengjing Hospital of China Medical University, China

Reviewer

  1. Philippe Krebs, University of Bern, Switzerland

Publication history

  1. Received: April 27, 2022
  2. Preprint posted: May 13, 2022 (view preprint)
  3. Accepted: December 1, 2022
  4. Accepted Manuscript published: December 2, 2022 (version 1)
  5. Version of Record published: December 12, 2022 (version 2)

Copyright

© 2022, Yang et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 309
    Page views
  • 58
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Lei Yang
  2. Xichen Dong
  3. Zheng Liu
  4. Jinjing Tan
  5. Xiaoxi Huang
  6. Tao Wen
  7. Hao Qu
  8. Zhenjun Wang
(2022)
VPS9D1-AS1 overexpression amplifies intratumoral TGF-β signaling and promotes tumor cell escape from CD8+ T cell killing in colorectal cancer
eLife 11:e79811.
https://doi.org/10.7554/eLife.79811

Further reading

    1. Cancer Biology
    2. Microbiology and Infectious Disease
    Changkun Hu, Taylor Bugbee ... Nicholas Wallace
    Research Article

    Double strand breaks (DSBs) are one of the most lethal DNA lesions in cells. The E6 protein of beta-human papillomavirus (HPV8 E6) impairs two critical DSB repair pathways; homologous recombination (HR) and non-homologous end-joining (NHEJ). However, HPV8 E6 only delays DSB repair. How DSBs are repaired in cells with HPV8 E6 remains to be studied. We hypothesize that HPV8 E6 promotes a less commonly used DSB repair pathway, alternative end-joining (Alt-EJ). Using CAS9 based Alt-EJ reporters, we show that HPV8 E6 promotes Alt-EJ. Further, using small molecule inhibitors, CRISPR/CAS9 gene knockout, and HPV8 E6 mutant, we find that HPV8 E6 promotes Alt-EJ by binding p300, an acetyltransferase that facilitates DSB repair by HR and NHEJ. At least some of this repair occurs through a subset of Alt-EJ known as polymerase theta dependent end joining. Finally, whole genome sequencing analysis showed HPV8 E6 caused an increased frequency of deletions bearing the microhomology signatures of Alt-EJ. This study fills the knowledge gap of how DSB is repaired in cells with HPV8 E6 and the mutagenic consequences of HPV8 E6 mediated p300 destabilization. Broadly, this study supports the hypothesis that beta-HPV promotes cancer formation by increasing genomic instability.

    1. Cancer Biology
    Anthony Lozano, Francois-Régis Souche ... Damien Grégoire
    Research Article

    Quantitative differences in signal transduction are to date an understudied feature of tumour heterogeneity. The MAPK Erk pathway, which is activated in a large proportion of human tumours, is a prototypic example of distinct cell fates being driven by signal intensity. We have used primary hepatocyte precursors transformed with different dosages of an oncogenic form of Ras to model subclonal variations in MAPK signalling. Orthotopic allografts of Ras-transformed cells in immunocompromised mice gave rise to fast-growing aggressive tumours, both at the primary location and in the peritoneal cavity. Fluorescent labelling of cells expressing different oncogene levels, and consequently varying levels of MAPK Erk activation, highlighted the selection processes operating at the two sites of tumour growth. Indeed, significantly higher Ras expression was observed in primary as compared to secondary, metastatic sites, despite the apparent evolutionary trade-off of increased apoptotic death in the liver that correlated with high Ras dosage. Analysis of the immune tumour microenvironment at the two locations suggests that fast peritoneal tumour growth in the immunocompromised setting is abrogated in immunocompetent animals due to efficient antigen presentation by peritoneal dendritic cells. Furthermore, our data indicate that, in contrast to the metastatic-like outgrowth, strong MAPK signalling is required in the primary liver tumours to resist elimination by NK cells. Overall, this study describes a quantitative aspect of tumour heterogeneity and points to a potential vulnerability of a subtype of hepatocellular carcinoma as a function of MAPK Erk signalling intensity.