The cell line 293T and the human colon carcinoma cell lines SW620 and Caco2 were purchased from the American Type Culture Collection and cultured in Dulbecco's Modified Eagle's medium (DMEM) or L-15 medium (Leibovitz), respectively, supplemented with 10% fetal bovine serum. Human umbilical vein endothelial cells (HUVECs) were purchased from PromoCell and cultured in endothelial cell Growth Medium 2 (PromoCell).

The CDR1-AS-expressing plasmid, pcDNA3.1-laccase2-MCS-ciRS7, whose transcript naturally circularizes due to DNAREP1_DM cassettes, and its control plasmid were obtained from Addgene, Inc. The lentiviral vector pCDH-CDR1-AS was constructed by subcloning the corresponding sequences from pcDNA3.1 into pCDH vector, using the EcoRI and NotI sites. Lentiviruses produced from the pCDH control vector were used as a negative control.

To generate stably expressed polyclonal cells, the Lentivirus Packaging System (System Biosciences) was used according to the manufacturer's protocol. The viruses were transduced into SW620 cells and Caco2 cells using polybrene (Santa Cruz Biotechnology, Inc.), followed by selection with 6 µg/ml puromycin to obtain polyclonal cells stably expressing CDR1-AS RNA.

RNA was extracted using ISOGEN II (Nippon Gene Co., Ltd.) according to the manufacturer's protocol. After RNA extraction, 6 µg RNA was treated with RNase R (AR Brown Co., Ltd.) at 37°C for 15 min, followed by purification using the RNeasy MinElute Cleanup kit, according to the manufacturer's protocol (Qiagen).

Northern blotting was performed as described previously (9). Briefly, 5 µg of RNAs were separated in 1% formaldehyde-denatured agarose gel and hydrostatically transferred to a Hybond N⁺ membrane (GE Healthcare Life Sciences). Membranes were UV-crosslinked and hybridization was performed overnight at 42°C in ULTRAhyb Buffer (Ambion; Thermo Fisher Scientific, Inc.) containing 10 ng/ml biotin-labeled RNA probe. The membranes were stringently washed and bound probe was visualized using a BrightStar BioDetect kit (Ambion) according to the manufacturer's protocol. The probes for detecting CDR1-AS were generated using in vitro transcription MEGAscript T7 kit, and the pre-made probe for β-actin was obtained using a DIG Northern Starter kit (Roche Diagnostics).

The Cell Counting Kit-8 (CCK-8; Dojindo Molecular Technologies, Inc.) was used to evaluate 2D cell growth according to the manufacturer's instructions. Cells were seeded at 1×10³ cells/well in 96-well plates. CCK-8 solution was added at 24-, 48-, 72-, and 96-h time-points. The absorbance at 450 nm was measured using a microplate reader (Thermo Fisher Scientific, Inc.).

For the 3D cell culture, SW620 cells (5×10⁴ cells/well) were seeded in hanging droplets in a volume of 40 µl medium/droplet using the GravityPlus system (InSphero AG). After 8 days, the spheroids were transferred to a 96-well microtissue receiver plate with a non-adhesive surface (GravityTRAP; InSphero) in a volume of 70 µl medium/well. CellTiter-Glo 3D cell reagent (Promega Corp.) was added at 24-, 48-, 72-, and 96-h time-points. Following 30 min of incubation at room temperature, luminescence was quantified on a GloMax 96 Microplate Luminometer (Promega Corp.).

An invasion assay was performed using a CytoSelect 96-well cell invasion assay kit (Cell Biolabs, Inc.) with 1×10⁵ cells placed in the upper chamber of the assay plate. Chemoattractant was added to the feeder tray. The assay plate was cultured for 24 h, and cells that migrated were collected and quantified using CyQUANT GR dye solution (Cell Biolabs, Inc.). Invasion values were reported as the mean relative fluorescence units (RFUs) measured at 480 nm/520 nm using a GloMax Discover Multimode Microplate Reader (Promega Corp.).

An Angiogenesis Assay Kit (PromoCell) was used to evaluate endothelial tube formation according to the manufacturer's instructions. HUVECs were seeded onto 96-well plates coated with basement membrane extract at 1.0×10⁴ cells/well. The culture supernatant of the negative control and the CDR1-AS-expressing SW620 cells were added to each well, and the plate was incubated for 6.5 h. The tubular networks were analyzed using ImageJ software 1.51J8 (National Institutes of Health).

cDNA microarray analysis was performed using cDNA oligo chips (Toray Industries, Inc.). The data and the protocols were deposited in a public database (GEO; accession no. GSE125687).

Quantitative RT-PCR (qRT-PCR) was performed as described previously (10). All values were normalized to the mRNA levels of the β-actin. Relative expression was calculated according to the ΔΔC_q method as follows: ΔΔC_q=ΔC_qsample -ΔC_qβ-actin (11). The primers used were as follows: CDR1-AS forward, 5′-GCTGATCTTCTGACATTCAGG-3′ and reverse, 5′-GAGTTGTTGGAAGACCTTGAC-3′; PD-L1 forward, 5′-GGTGCCGACTACAAGCGAAT-3′ and reverse, 5′-AGCCCTCAGCCTGACATGTC-3′; CMTM4 forward, 5′-CTGGCGTCTTGCTGATTATG-3′ and reverse, 5′-ATTTCTGCTCCGGCTCTATG-3′; CMTM6 forward, 5′-ATGAAGGCCAGCAGAGACAG-3′ and reverse, 5′-GTGTACAGCCCCACTACGGA-3′; β-actin forward, 5′-TCCCTGGAGAAGAGCTACGA-3′ and reverse, 5′-AGCACTGTGTTGGCGTACAG-3′.

pGL4-miR7RE, a luciferase-based reporter construct to monitor miR7 function, was generated by inserting two tandem sequences of the miR7 responsive elements (ACAACAAAATCACTAGTCTTCCAACAACAAAATCACTAGTCTTCCA) into the FseI site at the 3′UTR of the pGL4.5 luciferase vector (Promega Corp.). To construct the control mutated miR7 reporter, the following insertion sequence was used: ACAACAAAATCACTAGTCAAGGTACAACAAAATCACTAGTCAAGGT, which has mutations in the seed sequences of the miR7-recognizing sites.

To determine the functional activities of miR7, the pGL4-miR7RE plasmid, along with a pGL4-TK plasmid-expressing sea pansy luciferase (Promega Corp.) as an internal control, were transiently transfected into the cells using FuGENE6 (Promega Corp.). When adding the miR7 mimic, locked nucleic acid (LNA)-based miR7 oligonucleotides (Qiagen) were transfected using oligofectamine (Invitrogen; Thermo Fisher Scientific, Inc.) at the same time as the reporter transfection. At 48 h after transfection, dual luciferase assays were performed using a Dual-Luciferase Reporter Assay System (Promega Corp.) as previously described (10).

Flow cytometric analyses were performed as previously described (12). Cells were hybridized with anti-PDL1 (1:1,000; cat. no. PA5-20343; Thermo Fisher Scientific, Inc.) or the isotype control IgG (1:1,000; cat. no. 2729; Cell Signaling Technology, Inc.) for 40 min at 4°C. After washing, the cells were incubated with goat anti-mouse Alexa Fluor 488 (1:1,000; Molecular Probes; Thermo Fisher Scientific, Inc.) for 20 min. Flow cytometry was performed, and the data were analyzed using Guava Easy Cyte Plus (GE Healthcare Life Sciences). To quantify the PD-L1-positive cells, cells with fluorescence intensity exceeding the negative control cells were counted, and the ratio was calculated.

Experimental protocols were approved by the Ethics Committee for Animal Experimentation at the University of Tokyo (approval no. P18-017) and conducted in accordance with the Guidelines for the Care and Use of Laboratory Animals of the Department of Medicine, University of Tokyo.

Four 8-week old male BALB/c (nu/nu) nude mice (weight ~30 g) were purchased from CREA Japan and maintained in 14/10-h light/dark cycle at 25°C. Briefly, 2×10⁶ control or CDR1-AS cells were suspended in 30 µl of PBS containing 1% Matrigel (BD Biosciences) and injected subcutaneously into two mice to establish xenograft models. Mice were placed in standard conditions (4/cage) under specific pathogen-free conditions in laminar flow cabinets. All animals received food and water ad libitum. At 4 weeks post-transplantation, xenograft tissues were collected for immunohistochemistry analyses.

Immunohistochemistry was performed as previously described (13). Tissues were incubated overnight at 4°C with anti-PDL1 antibody (cat. no. PA5-20343; Thermo Fisher Scientific, Inc.) diluted with Can Get Signal Immunostain Immunoreaction Enhancer Solution (Toyobo Life Science). Signals were enhanced using a VECTASTAIN ABC kit (Vector Laboratories, Inc.) according to the manufacturer's protocol and visualized with 3,3′-diaminobenzidine in a buffered substrate (Nichirei Biosciences, Inc.). To determine the relative intensity on the cell surface with PD-L1 staining, the average pixel intensity was determined from at least 20 randomly selected cells using ImageJ 1.51J8.

Putative miRs targeting the 3′untranslated region (UTR) sequences of CMTM4 and CMTM6 transcripts were determined using TargetScan 7.2 software (14).

Statistically significant differences between two groups were identified using Student's t-test when the variances were equal and Welch's t-test when the variances were unequal. When comparing multiple groups, ANOVA and Bonferroni post hoc test were used to determine the statistical significances. P-values <0.05 were considered to indicate statistical significance.

To establish stably-CDR1-AS-expressing colon cancer cells, laccase cassettes were used to construct lentiviruses containing CDR1-AS sequences, which efficiently produced circular RNAs because a pair of inverted DNAREP1_DM family transposons were located very close to the circularizing exon and these repeats are highly complementary to each other (15) (Fig. 1A). The viruses were infected into SW620 cells, a colon cancer cell line with relatively high miR7 expression (16). To confirm the circularity of the expressed CDR1-AS RNA, the expression after RNase R treatment, which digests all linear RNAs but does not digest lariat or circular RNA structures, was examined by Northern blotting with a probe against the back-splice junction of the CDR1-AS RNA (Fig. 1B). CDR1-AS was almost the expected length (1,480 nucleotides) in agarose gel electrophoresis, and the bands were visible even after RNase R treatment, whereas the bands corresponding to β-actin completely disappeared, indicating that CDR1-AS forms a circular RNA resistant to RNase R treatment (Fig. 1B). The high intensity levels of bands from the CDR1-AS-expressing constructs (from 2,000 to 4,000 nucleotides) disappeared after RNase R treatment. It was speculated that these bands corresponded to incompletely circularized transcripts and/or nicked circular RNAs. As RNase R is a 3′-5′ exoribonuclease that digests RNAs with 3′ends, completely circularized CDR1-AS transcripts were unaffected and remained even after RNase R treatment. RT-PCR determined that CDR1-AS RNA was expressed at levels ~300 times higher than those in the non-expressing control cells, and did not significantly diminish even after RNase R treatment (Fig. 1C). Consistent with a previous study (6), CDR1-AS expression restored miR7 function as determined by a reporter assay, which was reversed by the forced expression of miR7-mimic LNA oligonucleotides (Fig. 1D). Although these results confirmed that CDR1-AS acted as an inhibitor of miR7 function in our system, endogenous miR7 levels were not prominent in colon cancer cells (8).

Since CDR1-AS expression is closely linked with poor prognosis in various cancers, including colon carcinoma (5,17), the effects of CDR1-AS on the growth and invasiveness of SW620 cells were examined. The growth of SW620 cells in both 2D and 3D cultures was examined since a 3D culture may render different results from conventional 2D cultures; however, there was no significant growth advantage observed in CDR1-AS-expressing SW620 cells (Fig. 2A and B). Furthermore, a Transwell invasion assay did not reveal any increased invasiveness of CDR1-AS-expressing SW620 cells (Fig. 2C). Angiogenesis was also examined using supernatant from CDR1-AS-expressing SW620 cells to determine whether they promote angiogenesis. However, there were no significant differences in the length and branch counts in in vitro angiogenesis between the supernatants from the control and the CDR1-AS-expressing cells (Fig. 2D). These results appear to exclude the possibility that angiogenetic factors are produced by CDR1-AS expression and suggest that, although CDR1-AS expression in colon cancer cells is clinically linked with poor prognosis, cell growth, invasiveness, and angiogenesis are unaffected.

To gain insight into the biological causes of the poor prognosis of colon cancers with CDR1-AS expression, changes in transcript expression levels in CDR1-AS-expressing SW620 cells (the cDNA microarray results were deposited in GEO; accession no. GSE125687) were assessed. The simultaneous increases in CMTM4 and CMTM6 transcript levels were noteworthy since these genes were recently reported as stabilizers of PD-L1 protein at the cell surface (18,19). The PD-L1 ligand plays a crucial role in suppressing tumor-specific T-cell responses, and its upregulation on the surface of cancer cells is linked to enhanced inhibition of T cells and thus, poor prognosis (20). The upregulation of CMTM4 and CMTM6 transcript levels using RT-PCR was confirmed (Fig. 3A). Accordingly, cell surface PD-L1 levels, which were expressed only in a subpopulation, were upregulated in CDR1-AS-expressing SW620 cells (Fig. 3B and C), although PD-L1 mRNA levels were unchanged (Fig. 3A). The increased expression of PD-L1 on the surface of CDR1-AS-expressing cells was also detected in xenograft models established using SW620 cells (Fig. 3D and E), although tumour growth/size was not analysed in the present study in the presence/absence of CDR1AS expression due to inconsistencies in the growth rates rendering the appropriate analysis difficult. Similarly, in Caco2 cells, another colon cancer cell line, forced expression of CDR1-AS RNA (Fig. S1A) induced significantly increased expression of CMTM transcripts (Fig. S1B) and PD-L1 protein on the cell surface (Fig. S1C and D). These results inidcated that CDR1-AS expression leads to an increase in CMTM4 and CMTM6 expression levels, resulting in an increase in cell surface PD-L1 levels.

Since CDR1-AS functionally antagonizes miR7 due to sequence similarities (Fig. 1D) (6,7), miR7 mimic LNA oligonucleotides were ectopically expressed in CDR1-AS-expressing cells to examine the involvement of deregulated miR7 function in the upregulated cell surface PD-L1 expression levels. However, the increased CMTM4 and CMTM6 transcript levels (Fig. 3F) and the cell surface PD-L1 protein levels (Fig. 3G), which were expressed in a subpopulation of the cells, were not reduced by forced expression of miR7 mimic LNA oligonucleotides. These results were consistent with the fact that the 3′UTRs of CMTM4 and CMTM6 do not share sequence similarities with miR7 (Fig. 3H). Thus, the upregulated expression levels of CMTM4 and CMTM6 in CDR1-AS-expressing cells do not appear to be dependent on impaired miR7 function, but likely depend on other undefined factors induced by CDR1-AS expression.