Human breast cancer cell lines, SKBR3, T47D and MCF7, and human colorectal cancer cell lines, DLD1, LS174T, HCT116 and RKO, were obtained from the American Type Culture Collection. T47D and DLD1 cells were cultured in RPMI medium (Gibco; Thermo Fisher Scientific, Inc.), MCF7 and LST174T cells in EMEM (Gibco; Thermo Fisher Scientific, Inc.), SKBR3 and HCT116 cells in McCoy's 5A (Gibco; Thermo Fisher Scientific, Inc.) and RKO cells in DMEM medium (Gibco; Thermo Fisher Scientific, Inc.), all containing 10% fetal bovine serum (Biosera, Ltd.) and antibiotic/antimycotic solution. The cells were incubated at 37°C under a humidified atmosphere containing 5% CO₂. Mycoplasma contamination was tested using MycoStrip (Thermo Fisher Scientific, Inc.), which indicated that all cell lines were free of mycoplasma.

Total cellular RNA was extracted from the cultured cells using a RNeasy Mini Kit (Qiagen, Inc.). cDNA was synthesized from 1 µg of total RNA with ReverTra Ace (Toyobo Life Science) according to the manufacturer's protocol. qPCR was performed using KAPA SYBR FAST qPCR Master Mix (2X) Kit (Sigma-Aldrich; Merck KGaA) with sets of primers for JUNB, JUN, FOS, JUND, activating transcription factor 3 (ATF3), ATF7, Jun dimerization protein 2 (JDP2), CAMP responsive element binding protein 1 (CREB1), CREB3L4, PVRL4 and hypoxanthine phosphoribosyltransferase 1 (HPRT1) (Table SI) and a StepOnePlus system (Thermo Fisher Scientific, Inc.). The thermocycler conditions were as follows: Denaturation at 95°C for 20 sec, followed by annealing and extension at 60°C for 20 sec, for a total of 40 cycles. Quantities of transcripts were determined by relative standard curves, and HPRT1 was used as the internal control. The quantification of the JUNB, JUN, FOS, JUND, ATF3, ATF7, JDP2, CREB1, CREB3L4, PVRL4 and HPRT1 transcripts were calculated from five-point standard curves prepared by amplifying the pooled control cDNA derived from SKBR3 cells according to Getting Started Guide of Applied Biosystems StepOne™ and StepOnePlus™ Real-Time PCR Systems Standard Curve Experiments (PN: 4376784; Thermo Fisher Scientific, Inc.). The standard curves were automatically generated by the StepOne Software v2.2.3. Baseline cycles and thresholds were set manually, and the quantification cycles were calculated automatically by the software using an in-built algorithm.

Pools of human specific siRNAs were obtained from Merck KGaA (Table SII). ON-TARGETplus Non-targeting Pool (cat. no. D-001810-10; GE Healthcare Dharmacon, Inc.) was used as the control. All gene silencing experiments were performed in SKBR3 cells excepting si-FOS, where T47D cells were also used. SKBR3 and T47D cells were seeded 1 day before siRNA treatment. Cells were transfected with 10 nM siRNA using Lipofectamine RNAiMAX (Thermo Fisher Scientific, Inc.). After 48 h of incubation, total RNA was isolated from the cells using an RNeasy Mini Kit according to the manufacturer's instruction. The silencing effect was evaluated by quantitative RT-qPCR, as aforementioned.

Putative promoter regions in the 5′-flanking sequence, intron, and 3′-flanking sequence of PVRL4 were amplified by PCR with specific primer sets (Table SIII and IV) containing recognition sites for the following restriction enzymes: XhoI, BglII, KpnI or HindIII. PCR products were digested with restriction enzymes and cloned into the pGL4.23 vector (Promega Corporation). Mutant versions of PVRL4 reporter plasmids were generated by site-directed mutagenesis. Briefly, PCR was performed using KOD Plus NEO enzyme (Toyobo Life Science), 100 ng of the pGL4.23 -PVRL4#10-III plasmid containing the enhancer region as a template and a set of primers for Mut-1 and −2 (Table SV), according to the manufacturer's protocol. The PCR products were digested with Dpn1 restriction enzyme for 2 h at 37°C and were transformed into Escherichia coli DH5α cells.

To generate plasmids expressing FOS, the entire coding sequence of FOS was amplified by PCR using KOD ONE (Toyobo Life Science) with a set of primers (Table SVI) containing EcoRI or XhoI restriction sites, according to the manufacturer's protocol. The PCR product was cloned into a pCMV-myc vector (Promega Corporation). Sanger sequencing [using Applied Biosystems 3500 (Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol] confirmed the full-length cDNA sequence of FOS was inserted into the plasmid.

Cells seeded in 12-well plates were transfected with 0.3 µg of pGL4.23 reporter plasmid and 0.05 µg of pRL-TK plasmid (Promega Corporation) by FuGENE 6 reagent (Promega Corporation). After 48 h of transfection, a PicaGene Dual Sea Pansy Luminescence Kit (TOYOB-Net) was utilized to measure the activities of firefly and Renilla luciferase according to the supplier's protocol. The Renilla activity was normalized to the firefly activity.

To identify DNA sequences for putative transcription factor binding sites, ChIP-seq data from the ENCODE project [http://www.genome.ucsc.edu; The University of California Santa Cruz Genome Browser Database (UCSC); accession nos. GSM733646, GSM733674 and GSM733771] and JASPER (http://jaspar.genereg.net/) were used. In the present study, a JASPAR score >15.5 was deemed a putative-binding motif.

ATAC-Seq of breast and colorectal cancer cells expressing either high or low levels of PVRL4 was performed using an ATAC-Seq Kit (Active Motif, Inc.), according to the manufacturer's instructions. Briefly, the cells were resuspended in 100 µl of ATAC lysis buffer to extract the nuclei. The nuclei were collected by centrifugation at 500 × g for 5 min at 4°C, resuspended in 50 µl of Tagmentation Master Mix (including assembled transposomes) and then incubated at 37°C for 30 min with shaking. The tagmented DNA was purified using 250 µl of DNA Purification Binding Buffer and 5 µl of 3 M sodium acetate. The DNA library was collected using SPRI beads, which binds to a magnet, and eluted with 20 µl of DNA Purification Elution Buffer. The quality of library was verified using a bioanalyzer (Agilent Technologies, Inc.). The libraries were sequenced (primers listed in Table SVII) with HiSeq Rapid SBS Kit v2 (Illumina, Inc) and HiSeq Rapid SR Cluster Kit v2 (Illumina, Inc) using 60 bp single-end reads on the HiSeq2500 platform (cat. no. SY-401-2501; Illumina, Inc). The raw sequencing reads were analyzed for quality using FastQC and then aligned to the human genome (hg19) using Bowtie2 (v2.4.1; http://bowtie-bio.sourceforge.net/bowtie2/index.shtml). Integrative Genomics Viewer was used to visualize the sequencing data (https://igv.org/doc/desktop/).

The 3C assay was performed as previously described (22,23). Briefly, SKBR3 cells were cross-linked with 1% formaldehyde for 10 min at room temperature and then quenched with 125 mM glycine. The cross-linked chromatin was digested at 37°C overnight with 400 U BgIII (Takara Bio, Inc.), which was then heat-inactivated for 25 min at 65°C with 1.6% SDS. Subsequently, DNA fragments were ligated with T4 DNA ligase (New England Biolabs, Inc.) for 8 h at 16°C. Samples were treated with 300 µg of proteinase K (Merck KGaA) at 37°C overnight to remove the cross-link and then with 300 µg of RNase A (Merck KGaA). The ligated DNA was purified by phenol/chloroform extraction and ethanol precipitation. The first PCR reaction was amplified with the outer primer sets. The first PCR parameters for relative quantification were as follows: 2 min at 98°C, followed by 35 cycles at 98°C for 10 sec, 60°C for 5 sec and 68°C for 10 sec. After the products of the first PCR were purified, nested PCR (KOD One; Toyobo Life Science) was performed using these products with the inner primer sets to investigate a possible interaction between the promoter and enhancer regions of PVRL4. The nested PCR parameters for relative quantification were as follows: 2 min at 98°C, followed by 30 cycles at 98°C for 10 sec, 62.5°C for 5 sec and 68°C for 10 sec. The sequences of the PCR primers used are shown in Table SVIII. The PCR products were confirmed by gel electrophoresis using 1.5% agarose gels and visualized with ethidium bromide staining (Sigma-Aldrich; Merck KGaA).

The ChIP and ChIP-qPCR analysis were performed as previously described (24). A total of 2×10⁷ SKBR3 cells were cross-linked with 1% formaldehyde for 10 min, followed by quenching with 125 mM glycine for 5 min. DNA fragmentation was performed using a UD-201 homogenizer (Tomy Seiko Co., Ltd.) with the following settings: Output level 5, 50% duty, 15 sec, 3 cycles and on floating ice to obtain 200–500 bp DNA fragments. The fragmented DNA samples were confirmed by gel electrophoresis using 1.5% agarose gels and visualized with ethidium bromide staining (Sigma-Aldrich; Merck KGaA). An aliquot of the sample was kept as an input and the remaining sample was used for ChIP analysis. The fragmented samples from the SKBR cells were incubated with 8 µg of anti-phospho-FOS antibody (Cell Signaling Technology, Inc.; cat. no. 5348; 1:100) or 8 µg of anti-IgG antibody (normal mouse IgG; Santa Cruz Biotechnology, Inc.; cat. no. sc-2025; 1:400) and bound to protein G-sepharose beads (GE Healthcare) at 4°C overnight. The beads were separated with a column and washed sequentially for 5 min with 1 ml of the following buffers: 1X low salt wash buffer [1% Triton X-100, 1 mM EDTA, 150 mM NaCl, 20 mM Tris-HCl (pH 8.0)], 1X high salt wash buffer [0.1% SDS, 1% Triton X-100, 2 mM EDTA, 500 mM NaCl, 20 mM Tris-HCl (pH 8.0)], 1X LiCl wash buffer [0.25 M LiCl, 0.5% Nonidet P-40, 0.5% sodium deoxycholate, 1 mM EDTA, 10 mM Tris-HCl (pH 8.0)] and 2X Tris EDTA buffer [10 mM Tris-HCl (pH 8.0), 1 mM EDTA]. The beads were then eluted with 200 µl of elution buffer (100 mM NaHCO₃, 1% SDS, 5 mM NaCl) and 2 µl of proteinase K (10 mg/ml; Sigma-Aldrich; Merck KGaA). The immunoprecipitated and input DNA were purified by using a QIAquick PCR Purification Kit (Qiagen, Inc.). The concentration of two purified DNA samples was measured using Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, Inc.) with a Qubit fluorometer (Thermo Fisher Scientific, Inc.). qPCR was performed using KAPA SYBR FAST qPCR Master Mix (2X) Kit (Sigma-Aldrich; Merck KGaA) with primers shown in Table SIX. Amplification of the glyceraldehyde-3-phosphate dehydrogenase gene was used as a negative control, and relative quantification was performed using StepOne Software v2.2.3. The PCR parameters for relative quantification were as follows: 2 min at 98°C, followed by 40 cycles at 95°C for 1 sec and 60°C for 20 sec.

To identify genes regulated by PVRL4, RNA-seq analysis was performed using an Ion Proton system™ (Thermo Fisher Scientific, Inc.). Total RNA was extracted from SKBR3 cells treated with siPVRL4#1, #2 or control siRNA using a RNeasy Mini Kit (Qiagen, Inc.), and the quality of RNA was assessed using an Agilent bioanalyzer device (Agilent Technologies, Inc). Libraries were prepared using all samples that had a RNA integrity number >7.0. RNA-seq libraries were prepared with 100 ng total RNA using the Ion AmpliSeq Transcriptome Human GeneExpression Kit (Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. The libraries were sequenced using an Ion Proton system with an Ion PI Hi-Q Sequencing 200 kit and Ion PI Chip v3 (Thermo Fisher Scientific, Inc.). The sequencing reads were aligned to hg19_AmpliSeq_Transcriptome_ERCC_v1 using the Torrent Mapping Alignment Program (https://github.com/iontorrent/TMAP), and the raw count data were generated using the AmpliSeqRNA plug-in (v5.2.0.3), both from the Torrent Suite Software (v5.2.2; Thermo Fisher Scientific, Inc.). The DESeq2 package (v1.26.0; http://bioconductor.org/packages/release/bioc/html/DESeq2.html) was used to normalize the read count data and test for differential gene expression. False discovery rate-adjusted P-values (q-values) <0.5 were considered to indicate a statistically significant difference. Functional enrichment analysis was performed using Metascape (https://metascape.org/gp/index.html#/main/step1) (25). Metascape pathway enrichment analysis uses GO, Kyoto Encyclopedia of Genes and Genomes (https://www.genome.jp/kegg/kegg_ja.html), Reactome (https://reactome.org/) and MSigDB (https://www.gsea-msigdb.org/gsea/msigdb). In brief, pairwise similarities between two enriched terms were calculated based on a κ test score. Then the similarity matrix was clustered hierarchically, and a similarity threshold score of 0.3-κ test score was applied to trim the result into separate clusters.

Unpaired two-tailed Student's t-test and one-way ANOVA followed by Dunnett's test were applied for statistical analyses. P<0.05 was considered to indicate a statistically significant difference. Data obtained from three independent experiments are presented as the mean ± SD.

It was reported that PVRL4 expression was elevated in ~61% of ductal carcinomas of the breast (11). However, genetic amplification of the PVRL4 gene was observed in ~10% of breast invasive carcinoma according to The Cancer Genome Atlas Pan Cancer Atlas data (https://gdc.cancer.gov/about-data/publications/pancanatlas), suggesting that different mechanisms may play a role in the elevated PVRL4 expression in breast cancer tissues.

To clarify the regulatory mechanisms of its expression in cancer cells, transcriptional regulatory elements in the PVRL4 genomic region were searched for. Analysis of the histone modifications, H3K4me1 and H3K27ac, in the UCSC genome database identified seven regions high in H3K4me1 and five regions high in H3K27ac, and the five high H3K27ac-regions overlapped with the high H3K4me1-regions (Figs. 1A and S1A). A further search of the database identified five DNase high sensitivity regions within the overlapped high H3K4me1 and high H3K27ac regions. To identify open chromatin regions, ATAC-seq was performed using breast cancer cells expressing high levels of PVRL4 (SKBR3 and T47D) and colon cancer cells expressing low levels of PVRL4 (HCT116 and RKO) (Fig. S1B). Subsequently, four putative open chromatin regions were identified with high ATAC-seq peaks in SKBR3 and T47D cells but not in HCT116 and RKO cells (Fig. 1B). Of these four regions, two were located in intron 4 and intron 6 and the remaining two were in the 3′-flanking region. Notably, three of the four regions were localized in the overlapped high H3K4me1 and high H3K27ac regions.

To identify the enhancer region within PVRL4, 14 regions were selected from the high H3K4me1 regions (Fig. 1A) and cloned into the pGL4.23 reporter plasmid. Transcriptional regulation is restricted by the physical constraints imposed by topologically associating domain (TAD). It is of note that these 14 regions were localized within the same TAD of the PVRL4- transcription start site according to the Topologically Associating Domain Knowledge Base (http://dna.cs.miami.edu/TADKB/), suggesting that the candidate regions may physically associate with the PVRL4 promoter region (Fig. S2). To examine the transcriptional activity of the candidate regions, a dual-luciferase assay using the 14 reporter plasmids was conducted in SKBR3 and T47D cells. The reporter plasmids containing regions #4, #9, #10 or #13 and those containing region #10 or #13 showed significantly higher luciferase activities compared with the control (mock) plasmids in SKBR3 and T47D cells, respectively (P<0.01; Figs. 1C and S3A). Since region #13 was localized in intron 1 of the Rho GTPase activating protein 30 (ARHGAP30) gene, this region was excluded from further analysis. Region #10 became the focus of further study as it demonstrated the highest transcriptional activity among the candidate regions. To elucidate the important region within region #10, three deletion mutants of #10 reporter plasmids (#10-I, #10-II and #10-III) were prepared and their reporter activities were compared with that of wild-type #10 reporter plasmids (Figs. 1D and S3B). Notably, deletion of the 5′-region (#10-I) increased the reporter activity in both SKBR3 and T47D cells. Furthermore, reporter plasmids containing #10-III also had a higher activity than the wild-type #10 plasmid in both cell lines, suggesting that this region may include transcriptionally important elements. Since the #10-III region was within an overlapping region (with high H3K4me1, high H3K27ac, high DNase sensitivity and open chromatin), it may play a role in the transcription of PVRL4 as a distant enhancer region.

Using the ChIP-seq data from the ENCODE project, transcription factors that may interact with the candidate enhancer region (#10-III) and putative-binding motifs were searched for using the JASPER database. Subsequently, 11 candidate binding-motifs of transcription factors were identified (Table I). Since FOSB and FOS-like 2 (FOSL2) act as heterodimers with JUN, knockdown of JUN by siRNA will decrease the activity of the FOSB-JUN and FOSL2-JUN complexes (26). However, FOS is known to function both as a homodimer and a heterodimer with JUN (27). Therefore, FOSB and FOSL2 were excluded from the subsequent knockdown experiments and the effect of the other nine transcription factors (JUN, JUNB, JUND, JDP2, FOS ATF3, ATF7, CREB1 and CREB3L4) were examined. siRNAs targeting the nine transcription factors were generated and the knockdown effect of each siRNA on the expression of its target gene was confirmed by qPCR analysis (Fig. S4). The effect of each siRNA on the reporter activity of plasmids containing #10-III was then investigated in SKBR3 cells. FOS siRNA significantly reduced the reporter activity (P<0.005), but the remaining eight siRNAs did not have a significant effect (Fig. 2A). Since the #10-III region contained a putative FOS binding motif (TGACGTCA), mutant reporter plasmids carrying nucleotide substitutions in the binding motif (mut1, CAACGTCA; mut2, TGACGCAA; mut1+2, CAACGCAA) were constructed (Fig. 2B). As expected, these substitutions significantly reduced the reporter activity compared with the wild-type plasmid (pGL4.23-PVRL4#10-III) in SKBR3 cells (Figs. 2B and S5A).

To confirm that FOS transcriptionally regulates the expression of PVRL4, the reporter activity of pGL4.23-PVRL4#10-III was analyzed in the presence and absence of a plasmid expressing wild-type FOS (pCMV-FOS; efficiency of the overexpression of FOS is shown in Fig. S5B) in SKBR3 cells. As shown in Fig. 2C, the reporter activity increased 3.58-fold in the presence of pCMV-FOS. This increased reporter activity by pCMV-FOS was not observed with a mutant reporter plasmid containing two substitutions in the FOS-binding motifs (Mut1 + 2; Fig. 2C and S5C). Furthermore, the ChIP-qPCR assay demonstrated that the immunoprecipitation with p-FOS antibody enriched the DNA of region #10-III compared with the control IgG antibody (Fig. 2D), which is in agreement with the ENCODE results. These results suggested that FOS transcriptionally upregulated PVRL4 through its interaction with two FOS-binding motifs in intron 4.

To examine whether the enhancer region interacts with the PVRL4 promoter, a 3C assay was conducted. DNA from SKBR3 cells was cross-linked with formaldehyde and subsequently digested with a restriction enzyme, Bglll. Self-ligation of the DNA was expected to produce chromatin loops between the enhancer and promoter regions when the two were closely associated (Fig. 3A). In total, four sets of first and nested PCR primers were designed that could detect associations between the two regions (Table SVII). Subsequently, amplification of the 3C DNA with the four primer sets produced PCR products with the expected sizes, but the amplification of control SKBR3 DNA failed to produce PCR products (Fig. 3B). Additionally, sequence analysis of the nested PCR products was conducted with primers in Table SVIII, which confirmed a ligated DNA sequence of the enhancer region (#10-III) and the promoter region (Fig. 3C). These data suggested that the enhancer interacted with the promoter region through the formation of a chromatin loop.

To investigate the involvement of FOS in the regulation of PVRL4, the effect of FOS knockdown on PVRL4 expression in SKBR3 and T47D cells was analyzed by qPCR. The efficiency of FOS knockdown by FOS siRNA transfection is shown in Fig. S4I and J. The expression of PVRL4 was significantly decreased by transfection with the three different FOS siRNAs (siFOS#1, #2 and #3; P<0.01; Figs. 4A and S6).

To confirm the effect of FOS on PVRL4 expression, MCF7 cells were transfected with pCMV-FOS and the expression of PVRL4 was analyzed by qPCR. Consistent with the reporter assay, PVRL4 expression was significantly enhanced by the overexpression of FOS (P<0.005; Fig. 4B).

To clarify the function of PVRL4 in breast cancer cells, RNA-seq analysis using SKBR3 cells treated with control or PVRL4 siRNA (siPVRL4#1 and siPVRL4#2) was performed. The efficiency of PVRL4 knockdown following transfection with these siRNA is shown in Fig. S7A. A total of 596 and 1734 genes were identified whose expression levels were significantly altered by siPVRL4#1 and siPVRL4#2, respectively, compared with control siRNA (q<0.5; Figs. 5A and S7B). The number of genes was low since PVRL4 is a cell membrane receptor (4). The effect of these two siRNA on expression may be small compared with the siRNA of transcription factors. Since the inclusion of unmerged genes may have increased the likelihood of detecting GO of off-target effects, 133 genes that were commonly altered by the two siRNAs were used. GO analysis with the 133 genes found three ontology terms, ‘Cytokine Signaling in Immune System’, ‘Antigen processing-Cross presentation’ and ‘Regulation of Extrinsic Apoptotic Signaling Pathway’ (Fig. 5B). In both siPVRL4 treatments, the expression levels of IFIT1, IFI44, IFI44L, MX1, XAF1 and OAS2, related to ‘Cytokine Signaling in Immune System’, were upregulated 11.4–2.9-fold, compared with the control siRNA (Table SX). These six, IFIT1 (28), IFI44 (29), IFI44L (29), MX1 (30,31), XAF1 (30,32) and OAS2 (30,33), genes are known for their involvement in the defense against virus infection. Since PVRL4 is a member of the nectin family, altered expression of the genes involved in ‘Negative Regulation of Binding’ may suggest a decrease in cell adhesion. These results suggested that PVRL4 was involved in the cytokine response and immune system.