The mice (C57/BL6) were maintained at 25°C with a photoperiod of 14:10 h light:dark with free access to food and water. Testes were isolated from 7- to 56-day-old mice and stored at −80°C until use. Experimental procedures used in this study were approved by the Institutional Animal Use and Care Committee at Hokkaido University.

GC-2spd(ts) and its derivatives were cultured in Dulbecco Modified Eagle medium containing 10% fetal bovine serum, 100 U/ml penicillin, 100 μg/ml streptomycin, and 292 μg/ml l-glutamine (Invitrogen, Carlsbad, CA, USA) (38, 40). We used two types of GC-2spd(ts)-derived cell clones: lncRNA-6.9kb-EGFP clones that contained the intact sequence of the lncRNA-Tcam1 locus (full sequence) and ΔCNS1-lncRNA-EGFP clones that lacked the lncRNA-Tcam1 promoter (Δpromoter) (Figure 1A).

Isolation of total RNAs, treatment with DNase, and qRT-PCR were done as previously described (38, 49). To detect lncRNA-Tcam1, we used KOD SYBR qPCR Mix (Toyobo, Osaka, Japan), and the other transcripts were measured with Power SYBR Green Master Mix (Applied Biosystems, Foster City, CA, USA). Primer pairs are listed in Table 1.

Total RNAs were isolated from a lncRNA-6.9kb-EGFP clone (full sequence) and a ΔCNS1-lncRNA-EGFP clone (Δpromoter) using ISOGEN II (Nippongene, Tokyo, Japan) according to the manufacturer’s instruction. The RNAs were further purified by phenol/chloroform–isoamylalcohol extraction and ethanol precipitation, and treated with TurboDNase (Applied Biosystems). The quality of the RNA samples was evaluated using BioAnalyzer (Agilent Technologies, Palo Alto, CA, USA) with RNA6000 Nano Chip. A 500-ng aliquot of total RNA from each sample was used to construct cDNA libraries using TruSeq Stranded mRNA Sample preparation kit (Illumina, San Diego, CA, USA), according to the manufacturer’s instructions. The resulting cDNA library was validated using BioAnalyzer with DNA1000 Chip and quantified using Cycleave PCR Quantification Kit (Takara Bio Inc., Otsu, Japan). 101-cycle of single-end sequencing was performed using HiSeq1500 (Illumina) in the rapid mode. Total reads were extracted with CASAVA v1.8.2 (Illumina). Then, PCR duplicates, adaptor sequences and low quality reads were removed from the extracted reads as follows. Briefly, if the first 10 bases of the two reads were identical and the entire reads exhibited >90% similarity, the reads were considered PCR duplicates. Remaining reads were then aligned with Bowtie version 2.2.3 to the mouse genome, mm10, which was downloaded from UCSC (https://genome.ucsc.edu/). The expression level of each gene was calculated as gene-specific reads per million total reads (RPM).

A PiggyBac-based plasmid for induction of lncRNA-Tcam1 transcription was generated as follows. A full length of lncRNA-Tcam1 was amplified by PCR with mouse genome DNA and KOD FX Neo (Toyobo) using a primer pair, 5′-CCGGAGGAGCGGGAGCGGAA-3′ and 5′-CACCGGAAAACAGCTTTAAT-3′. We used the genome DNA as a template because lncRNA-Tcam1 contained no intron. The product was subcloned into a pBluescript II vector (Stratagene, La Jolla, CA, USA) at EcoRV site, and the resulting plasmid was checked by DNA sequencing method. This plasmid was digested with SalI, blunted by T4 DNA polymerase (Takara), and further digested with NotI. The lncRNA-Tcam1 fragment was purified and inserted into pPBhCMV*1-cHA-pPA at the NotI/blunted XhoI site. The completed plasmid was checked by DNA sequencing and named pPBhCMV*1-lncTcam1-pPA. All plasmids for establishing the Tet-off system were kindly gifted by Dr. Kazuhiro Murakami (50).

GC-2spd(ts) cells were co-transfected with pPBhCMV*1-lncTcam1-pPA, pPBCAGtTA-IN, pPyCAG-PBase, and pKO-Select-Puro plasmids using GeneJuice transfection reagent (Merck, Darmstadt, Germany) and selected with 3 μg/ml puromycin for 14 days. The selected cells were kept in the presence of 1 μg/ml doxycycline (Dox). For inducing lncRNA-Tcam1 transcription, the cells were spread onto 35-mm dishes, and Dox was removed on the next day.

The results were expressed as means ± SD. The data were analyzed by one-way analysis of variance (ANOVA) followed by Dunnett’s test (Figures 2A, and 3B,C) or by Student’s t-test (Figures 2B,C, 3A, and 4B) using Microsoft Excel statistical analysis functions (Microsoft Corporation, Redmond, WA, USA).

To identify potential target genes that were regulated by lncRNA-Tcam1, we used GC-2spd(ts) cell clones we established in a previous study (38). In these lines, we stably transfected a construct, which contained a 6.9-kb sequence encompassing the Tcam1 promoter and the lncRNA-Tcam1 region (full sequence) or the same sequence without the lncRNA-Tcam1 promoter (Δpromoter), connected to the enhanced green fluorescent protein (EGFP) gene (Figure 1A). lncRNA-Tcam1 was expressed at higher levels in the clones with “full sequence” than those with “Δpromoter” (38), and hereafter, we designated “GC-2spd-Full” and “GC-2spd-Δprom,” respectively. We chose a representative cell clone from each cell line, and the GC-2spd-Full and GC-2spd-Δprom clones showed 63-fold difference in expression levels of lncRNA-Tcam1 by qRT-PCR (Figure 1B). We expected that target genes of this lncRNA should be expressed at different levels between them.

To determine the genes that were differentially expressed between GC-2spd-Full and GC-2spd-Δprom, we performed the RNA-seq analysis. 101-cycle of single-end sequencing using HiSeq 1500 yielded 13,054,029 and 13,194,037 reads for GC-2spd-Full and GC-2spd-Δprom, respectively, and 92.25 and 90.23% each of which was mapped to the mouse genome (mm10). The resultant fastq files were deposited on NCBI SRA database (accession no. SRR5435658 and SRR5435659). Subsequently, we compared the gene expression levels between GC-2spd-Full and GC-2spd-Δprom, and picked protein-coding genes that were expressed at lower levels than 0.1-fold in GC-2spd-Full as candidate genes negatively regulated by lncRNA-Tcam1, and those expressed at higher levels than fivefold in GC-2spd-Full as positively regulated genes. We then selected the gene whose RPKM value was higher than 1.5 in GC-2spd-Δprom or GC-2spd-Full. As a result, 29 genes were chosen as candidates potentially suppressed by lncRNA-Tcam1, and 27 genes as those activated by the lncRNA. However, after this selection, we found that one potentially activated gene was not detected in GC-2spd-Full by qRT-PCR, and eliminated it from candidate genes. Consequently, a total of 55 genes (29 potentially suppressed and 26 activated genes, Tables S1 and S2 in Supplementary Material) were subject to further analysis. An outline of the screening of candidate genes is shown in Figure 1C.

The target genes should be expressed correlatively to lncRNA-Tcam1 in the testis. We thus examined the expression of these potential target genes during postnatal testis development. We collected testes at 7–56 days after birth and measured the lncRNA-Tcam1 level by qRT-PCR. The transcript of lncRNA-Tcam1 was detected in all samples, and a significant increase was observed between day 7 and day 14 after birth (Figure 2A). The absence of lncRNA-Tcam1 expression in somatic cells in the mouse testis (38) led to the presumption that the transcription of lncRNA-Tcam1 is initiated in spermatogonia and increased in primary spermatocytes. Then, we examined expression levels of the 55 candidate genes (29 potentially downregulated and 26 upregulated) in the testis at 7 days and 14 days after birth by qRT-PCR, to compare their levels in spermatogonia and primary spermatocytes. Among the genes that might be suppressed by lncRNA-Tcam1, 5 genes showed significant decreases during this period (Figure 2B), and 16 genes out of potentially activated candidates exhibited significant increases (Figure 2C). The expression patterns of genes that were not correlated with lncRNA-Tcam1 is presented in Figures S1 and S2 in Supplementary Material. Signals for four downregulated candidates, Cyclin D2 (Ccnd2), CD55 molecule, decay accelerating factor for complement (Cd55), Syndecan 3 (Sdc3), and uc009oek.1, and an upregulated candidate, 2′-5′ oligoadenylate synthetase 3 (Oas3), were not detected by this analysis. From these data, 21 genes (5 genes down and 16 genes up) correlated with in vivo lncRNA-Tcam1 expression were forwarded to the next analysis as target gene candidates.

To further validate the RNA-seq and postnatal testis data, the expression was also examined using GC-2spd(ts) clones that were different from those used for RNA-seq. This would help us to select the real targets, because the properties might be different between cell clones as a result of the process to establish stable cells and pick the clones. qRT-PCR demonstrated that lncRNA-Tcam1 was expressed at a 3.5-fold higher level in GC-2spd-Full than in GC-2spd-Δprom (Figure 3A). We assessed the expression of 21 candidate genes using these two clones and GC-2spd(ts) cells without being transfected. By comparing to cells without being transfected, false-positive genes (due to the establishment of cell clones) could be excluded. Among five genes potentially suppressed by lncRNA-Tcam1, H19, imprinted maternally expressed transcript (H19), Insulin-like growth factor binding protein 5 (Igfbp5), and Brain expressed myelocytomatosis oncogene (Bmyc) were expressed at much higher levels in GC-2spd-Δprom than in GC-2spd-Full, but they were expressed at very low levels in GC-2spd(ts) cells without transfection (Figure 3B), indicating that these three genes were artifacts. Expression of Renin 1 structural (Ren1) and Glutathione peroxidase 7 (Gpx7) genes was positively correlated with lncRNA-Tcam1 (Figure 3B), which was inconsistent with the genes suppressed by lncRNA-Tcam1. Therefore, the five genes were unlikely to be targets of lncRNA-Tcam1.

Among 16 genes, expression levels of five genes [2′-5′ oligoadenylate synthetase 2 (Oas2), Homeobox D13 (Hoxd13), MX dynamin-like GTPase 2 (Mx2), Fatty acid binding protein 5, epidermal (Fabp5), and Interferon-induced protein wit tetratricopeptide repeats 1 (Ifit1)] were not significantly higher in GC-2spd-Full than in GC-2spd(ts) cells (Figure 3C), excluding the possibility that these genes were targets of lncRNA-Tcam1. On the other hand, 11 genes were expressed at significantly higher levels in GC-2spd-Full than in GC-2spd(ts) cells, and their expression was higher in GC-2spd-Full than in GC-2spd-Δprom (Figure 3C). Collectively, these expression profiles detected 11 potential genes upregulated by lncRNA-Tcam1 and they were subjected to further analysis.

We finally examined whether lncRNA-Tcam1 could activate the 11 candidate genes in the Tet-off system. We constructed the plasmid in which the hCMV*1 promoter induced lncRNA-Tcam1 transcription in response to the Dox removal, and stably transfected GC-2spd(ts) cells with it. qRT-PCR of the stable transfectants confirmed that the expression of lncRNA-Tcam1 was 57-fold induced exclusively at 48 h after removal of Dox (Figure 4A).

We examined expression levels of the 11 candidate genes 48 h after the Dox removal by qRT-PCR. We set the value from the control cells that were cultured for 48 h with Dox to 1.0 and calculated the levels of candidates. six genes, Tripartite motif-containing 30A (Trim30a), Interferon-induced protein wit tetratricopeptide repeats 3 (Ifit3), T cell-specific GTPase 2 (Tgtp2), Interferon gamma inducible protein 47 (Ifi47), 2′-5′ oligoadenylate synthetase 1G (Oas1g), and Guanylate binding protein 3 (Gbp3), exhibited significantly higher expression in the cells without Dox, and their fold-increases were 1.46, 3.37, 2.43, 1.44, 1.49, and 1.69, respectively (Figure 4B). We confirmed significant levels of expression of the six genes in the germ cell fraction from adult mouse testes (Table S3 in Supplementary Material). Interestingly, five genes of the final six candidates, namely, Ifit3, Tgtp2, Ifi47, Oas1g, and Gbp3, were immune response genes, and Trim30a was a negative-feedback regulator of immune response (Table 2). Collectively, these data indicated that lncRNA-Tcam1 is responsible for the regulation of some immune response pathways during spermatogenesis.