Long noncoding RNA 01534 maintains cancer stemness by downregulating endoplasmic reticulum stress response in colorectal cancer

Abstract Background Studies have shown that cancer stemness and the endoplasmic reticulum (ER) stress response are inversely regulated in colorectal cancer (CRC), but the mechanism has not been fully clarified. Long noncoding RNAs (lncRNAs) play key roles in cancer progression and metastasis. In this study we investigated lncRNA 01534 (LINC01534) as a possible modulator between cancer stemness and ER stress response. Methods In vitro experiments using CRC cell lines were performed to explore a possible role of LINC01534. The expression of LINC01534 in clinical CRC samples was assessed by quantitative reverse transcription‐polymerase chain reaction (qRT‐PCR) and in situ hybridization. Results Silencing LINC01534 led to suppression of cell proliferation, invasiveness, and cell cycle progression at the G2‐M phase, and promoted apoptosis. Moreover, we found that silencing LINC01534 suppressed cancer stemness, while it activated the ER stress response, especially through the PERK/eIF2α signaling pathway. In situ hybridization revealed LINC01534 was expressed in tumor cells and upregulated in CRC tissues compared with normal epithelium. A survival survey indicated that high LINC01534 expression was significantly associated with shorter overall survival in 187 CRC patients. Conclusion This is the first report on LINC01534 in human cancer. Our findings suggest that LINC01534 may be an important modulator of the maintenance of cancer stemness and suppression of the ER stress response, and that it could be a novel prognostic factor in CRC.


| INTRODUC TI ON
The incidence of colorectal cancer (CRC) has increased over recent years, and CRC is one of the most common gastrointestinal cancers worldwide, with the third highest morbidity (10.2%) and the second highest mortality (9.2%) in 2018. 1 Despite advances in therapeutic strategies-including surgical technology, radiotherapy, chemotherapy, and molecular-targeted therapy-the 5-y survival rate remains <65%; thus, the molecular basis of CRC must be further investigated. 2,3 Protein coding genes account for <2% of human genes, and the nonprotein-coding portion of the human genome was long considered "junk DNA." 4,5 However, noncoding RNAs (ncRNAs) are now the focus of increasing attention. 5 Long noncoding RNAs (lncRNAs) are defined as transcripts of over 200 nucleotides, in contrast to the small ncRNA group, which includes microRNAs (miRNAs). 6 LncRNAs have recently been found to play key roles in biological processes and pathological conditions related to cancer development. 7,8 LncRNAs regulate gene expression through many mechanisms, including acting as scaffolds for chromatin modifiers, microRNA sponges, transcriptional regulators, protein decoys, and enhancers. 4,6 Several lncRNAs, including HOTAIR and LUCRC, have been reported as prognostic factors in CRC. 4,8 These lncRNAs control tumor cell growth, migration, and invasion by affecting cell cycle progression and the epithelial-mesenchymal transition through altered signaling in multiple pathways including those of p53, AKT/ mTOR, JAK/STAT, and NF-kB. 4,6,9 However, the functions of most lncRNAs remain unclear. 7 A recent report demonstrated high expression of lncRNA 01534 (LINC01534) in the inflammatory disease osteoarthritis, and showed that LINC01534 promotes an inflammatory response through proinflammatory factors, such as TNFα and IL-6. 10 To date, however, we are not aware of any reports that describe LINC01534 in human cancer. We previously demonstrated that low proteasome activity cells (LPACs) that were visualized by ZsGreen fused to the carboxyl terminal degron of ornithine decarboxylase (ODC) can serve as a cancer stem cell (CSC) model in colon cancer cells. 11 Using ODC degron-transduced HCT116 cells as LPACs, we found by RNA sequencing that LINC01534 was one of the upregulated lncRNAs.
Notably, RNA sequence analysis revealed that siRNA treatment for LINC01534 restored the endoplasmic reticulum (ER) stress response in CRC cells. This particularly drew our attention because several studies have reported that stemness and the ER stress response are inversely regulated in colonic epithelium and CRC cells. 12,13 The ER stress response is an adaptation to many stresses-including nutrient and lipid deprivation, hypoxia, and acidic extracellular pH-and its role in cancer remains largely unknown. [14][15][16] In an effort to reveal a role of LINC01534 in CRC, we performed in vitro experiments including cancer stemness and the ER stress response and assessed its clinical significance in the survival of 187 CRC patients. Overall, the present study reveals many important aspects of LINC01534 in CRC in terms of its expression, function, and clinical relevance.

| Cell lines and culture conditions
The human CRC cell lines HCT116 and RKO were purchased from the American Type Culture Collection. Cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 100 U/ml penicillin/streptomycin. For glucoseor glutamine-deprived cultures, DMEM/no glucose (Thermo Fisher Scientific) or DMEM/low glucose/pyruvate/no glutamine/no phenol red (Thermo Fisher Scientific) were used. Cultures were maintained at 37°C in a humid incubator with 5% CO 2 .

| Retroviral transduction of the degron reporter
The degron sequence of ornithine decarboxylase (ODC) is recognized directly by proteasomes, leading to the destruction of the involved protein. 11

| Small interfering RNA and transfection
We purchased siLINC01534 from Thermo Fisher Scientific, and negative control siRNA from Gene Design (Osaka, Japan). The target sequences were as follows:

| Cell proliferation assay
Cells were seeded at a density of 3000-5000/well in 96-well plates, and cultured for 24-72 h. The viable cell number was counted using Cell Counting Kit-8 (Dojindo Laboratories).

| Invasion assay
Invasion assays were performed using polyethylene terephthalate membranes (8μm pore size) in a BioCoat Matrigel Invasion Chamber (Corning Inc.), as previously described. 18 Briefly, a 0.5-ml cell suspension (DMEM including 2 × 10 5 cells/ml for HCT116 cells and 3 × 10 5 cells/ml for RKO) was added to the upper well chamber. After removal of noninvading cells from the upper side of the membranes, the invading cells on the lower side were fixed and stained with Diff-Quik (Sysmex) for counting.

| Quantitative reverse transcriptionpolymerase chain reaction (qRT-PCR)
TRIzol RNA Isolation Reagent (Thermo Fisher Scientific) was used to extract total RNA from the cells. 19 cDNA was synthesized from 10 ng total RNA using the Rever Tra Ace qRT Master Mix (Toyobo Life Science). Quantitative PCR was performed in a Light Cycle 2.0 System (Roche Applied Science). The amplification conditions were as follows: initial denaturation at 95°C for 10 min, followed by 45 cycles of denaturation at 95°C for 10 s, annealing at 60°C for 10 s, and extension at 72°C for 10 sec, as previously described. 20 Data were normalized to the expression of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The sequences of the primers are listed in Table S1.

| Sphere formation assay
HCT116 cells were seeded in 96-well ultralow attachment plates (Corning) at a density of 1000 cells/well. These cells were transfected with negative control siRNA or siLINC01534, and cultured in DMEM/ F-12 serum-free medium supplemented with 20 ng/ml epithelial growth factor and 10 ng/ml fibroblast growth factor-2 at 37°C. At 6 d after seeding, the number of spheres of diameter ≥150 μm was counted.

| RNA sequencing
Total RNA was extracted from cells using the miRNeasy Mini Kit cgi?acc=GSE17 9624, enter ilktygmgdpglxex into the box). GSEA version 3.0 was used to identify gene sets that were significantly altered by the addition of siLINC01534 from the Gene Ontology database. Gene sets were considered activated if the false discovery rate (FDR) q-value was less than 0.05.

| Western blot analysis
Whole cells were lysed in RIPA buffer (Thermo Fisher Scientific) including a phosphatase and protease inhibitor cocktail (Thermo Fisher Scientific). Lysates were separated by electrophoresis, incubated with primary antibody overnight at 4°C, and then incu- The primary antibodies are listed in Table S2.

| RNAscope in situ hybridization (ISH) assay
RNAscope ISH assays were performed using an RNAscope 2.0 Highdefinition Assay Kit (Advanced Cell Diagnostics). In tissue sections, the LINC01534 expression levels were scored based on the number of dots or clusters derived from RNA molecules in the cells, as follows; Score 0: 0 dot/cell; score 1: 1-3 dots/cell; score 2: 4-9 dots/ cell, occasional cluster formation present; and score 3: ≥10 dots/cell, frequent cluster formation present. We also obtained gene expression and clinical data from The Cancer Genome Atlas Project (TCGA; http://tcga-data.nci.nih.gov/) for colon cancer patients containing a total of 367 clinical samples.
The TCGA dataset indicated that high expression of LINC01534 (n = 235) was significantly associated with shorter OS when compared with low LINC01534 expression (n = 132) (p = 0.0213; Figure S1B).

| Silencing of LINC01534 inhibited cell proliferation
QRT-PCR indicated that the CRC cell lines HCT116 and RKO expressed relatively high levels of LINC01534 ( Figure S2). We confirmed that siLINC01534-1 and siLINC01534-2 reduced LINC01534 expression by >85% compared with the level in cells treated with the negative control siRNA ( Figure 1A). Compared with negative control siRNA treatment, silencing of LINC01534 significantly suppressed proliferation of the HCT116 and RKO cell lines (p < 0.05, Figure 1B).

| Silencing of LINC01534 inhibits invasive capability and induces apoptosis
Silencing LINC01534 markedly reduced the invasiveness of HCT116 and RKO cells (p < 0.001; Figure 1D). The annexin V assay showed an increase in apoptosis in both cell lines ( Figure 1E).

| Silencing of LINC01534 activates the endoplasmic reticulum (ER) stress response
To explore the molecular basis of the effects of silencing LINC01534 in CRC cells, we performed RNA sequencing. Gene ontology analysis indicated that silencing LINC01534 led to the enrichment of multiple gene sets associated with the ER stress response ( Figure S5A). Among

| Localization of LINC01534 in colonic normal mucosa and tumor tissue
In situ hybridization for LINC01534 was performed using RNAscope in 12 paired normal and CRC tumor tissue samples. LINC01534 was not detected in normal colonic epithelium ( Figure 4A). LINC01534 expression was detected in tumor tissues, and was localized in epithelial tumor cells but not at the tumor stroma ( Figure 4B,C). When the LINC01534 expression level in CRC tissue was classified into four levels (scores 0-3), as described in METHODS, five CRC tissue samples were classified as score 1, and seven samples as score 2. By contrast, normal tissue samples were all classified as score 0.

| LINC01534 expression is associated with poor prognosis of CRC patients
As suggested by the TCGA dataset we tried to clarify the significance of LINC01534 RNA expression in the OS of the CRC patients who underwent surgery at Osaka University and its related hospitals. One hundred eighty-seven patients with CRC were divided into a LINC01534 high-expression group (n = 94) and a low-expression group (n = 93), based on the median LINC01534 value of 1.53 in tumor tissue ( Figure 5A) LINC01534 expression had a significantly poorer prognosis than patients with low LINC01534 expression (p = 0.0088; Figure 5B).
Univariate analysis showed that tumor depth, lymph node metastasis, lymphatic invasion, venous invasion, and LINC01534 expression were significant prognostic factors for OS (Table 2). Multivariate analysis revealed that LINC01534 expression was an independent prognostic factor for OS (p = 0.047) as well as for T4 tumors and the presence of lymphatic invasion ( Table 2).

| DISCUSS ION
In this study we found by lncRNA sequencing that LINC01534 was highly expressed in ODC degron transduced HCT116 cells. The TCGA dataset indicated that only the two lncRNAs, LINC01534 and HOTAIR, were significantly associated with shorter OS of CRC patients among the lncRNAs listed in Figure S1A. HOTAIR is known as an oncogenic lncRNA in various type of human cancer. 8   We previously demonstrated that LPACs, including HCT116, have cancer stem-like properties, such as high sphere formation ability, tumor growth, and drug resistance, by using the ZsGreen-ODC degron system. 11 Studies have shown that inhibition of proteasome activity led to activation of ER stress response, namely UPR. [26][27][28] This appears reasonable because the misfolded proteins could F I G U R E 3 SiLINC01534 treatment activated the unfolded protein response in the PERK/p-eIF2α signaling pathway under standard or glucose-or glutamine-deprived conditions. (A) Expression of the proteins PERK, HSPA5, ATF4, P-eIF2α, eIF2α, and actin was examined by western blot analysis 9 h after transfection with negative controls siRNA, siLINC01534-1, or siLINC01534-2, in HCT116 cells. (B) Expression of the proteins PERK, HSPA5, ATF4, P-eIF2α, eIF2α, and actin was examined by western blot analysis 24 h after transfection with the negative controls, siLINC01534-1, or siLINC01534-2 in RKO cells. Western blot analysis showed that silencing LINC01534 upregulated molecules involved in the PERK pathway, represented by phosphorylated eIF2α. In glucose-or glutamine-deprived conditions the activation was more evident. The β-actin bands served as a loading control. (C) qRT-PCR showed that treatment with a UPR inducer, thapsigargin (Tg), significantly suppressed the expression of stem cell markers including DCLK1, CD133, BMI1, and CD44v9. *p < 0.05. **p < 0.001. Data presented as mean ± SD. (D) qRT-PCR showed that treatment with Tg significantly increased the expression of the epithelial differentiation marker, CDX2. *p < 0.05. Data presented as mean ± SD F I G U R E 4 In situ hybridization using RNAscope in normal epithelium and CRC tissues. (A) In normal mucosa, LINC01534 expression was generally not detected (score 0). (B) In a tumor tissue, LINC01534 expression was noted mainly in tumor cells. The LINC01534 signal was 1-3 dots/cell (score 1). (C) In another tumor tissue, increased expression of LINC01534 was found mainly in tumor cells. The LINC01534 signal was 4-9 dots/cell (score 2). Scale bar, 10 μm F I G U R E 5 Association of LINC01534 expression with overall survival (OS) of CRC patients. (A) LINC01534 levels in 187 CRC tissue samples. LINC01534 expression in the CRC tissue sample was measured by qRT-PCR and GAPDH RNA expression was used for normalization. The patients were divided into high-and lowexpression groups, based on the median LINC01534 value of 1.53. (B) Kaplan-Meier OS curves according to the LINC01534 level. The OS rate was significantly lower in the LINC01534 high-expression group (n = 94) compared with the low-expression group (n = 93; log-rank test; p = 0.0088) accumulate in the cells under the low proteasome condition, which triggers on the UPR. Consistently, we found that LPACs in HCT116 displayed increased phosphorylation of translation initiation factor eIF2α as the indictor of UPR activation, 14 under cultures supplemented either with a standard medium or a glutamine-deprivation medium, as compared with non-LPACs ( Figure S6). We also confirmed that the phosphorylated eIF2α level increased when the proteasome inhibitor lactacystin was administered to the parental HCT116 cells ( Figure S7).
As for the effect of LINC01534 inhibition in proteasome activity in LPACs of HCT116, no significant effect was noted ( Figure S8A), but treatment with siLINC01534 still inhibited the expression of the CSC markers, DCLK1 and Bmi1 ( Figure S8B). It is therefore supposed that upregulation of LINC01534 in LPACs might suppress further activation of ER stress response to maintain cancer stemness. In any case, LPAC is just a model for CSC with low proteasome activity.
Thus, it would be better, irrespective of proteasome activity, to explore the role of LINC01534 in the parental HCT116 cells, which is known to retain cancer stem cell properties. [21][22][23] We found that siRNA treatment for LINC01534 resulted in reduced sphere formation and decreased expression of a series of CSC markers in HCT116 cells. These findings suggest that LINC01534 plays an essential role in maintaining cancer stemness. This notion is further supported by the findings that siRNA treatment induced an epithelial differentiation marker CDX2 and enhanced the sensitivity of HCT116 cells to 5-FU.
One of the major findings in this study is that LINC01534 appeared to suppress the ER stress response. ER stress is a condition in which unfolded or incorrect proteins accumulate in the ER due to various cellular stresses; for example, nutrient deprivation, hypoxia, oxidative stress, and anticancer drug treatment. 14,16,29 Under normal physiological conditions, the chaperone protein HSPA5 binds to three ER stress sensors (ER-spanning transmembrane proteins), such as PERK, activating transcription factor 6α (ATF6α), and inositol-requiring enzyme 1α (IRE1α), which remain sensors in their monomeric inactive states. 14,29 Upon exposure to ER stress, HSPA5 is released from the sensors and binds to unfolded proteins, starting the UPR. 29  In conclusion, we demonstrated for the first time that LINC01534 is a novel prognostic factor in CRC. LINC01534 facilitates the proliferation and invasiveness of CRC cells and has a unique function in the maintenance of cancer stemness and coordinated suppression of the ER stress response. Targeting LINC01534 is a potential therapeutic option against CRC that may improve the outcome in CRC patients.

DATA AVA I L A B I L I T Y S TAT E M E N T
Access to raw data concerning this study was submitted under Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/query/ acc. cgi?acc=GSE17 9624, enter ilktygmgdpglxex into the box). For the survival analysis of lncRNA expression is listed in Figure S1B, The Cancer Genome Atlas Program (TCGA) database was used (https:// www.cancer.gov/about -nci/organ izati on/ccg/resea rch/struc tural -genom ics/tcga/using -tcga).