Long noncoding RNA pncRNA-D reduces cyclin D1 gene expression and arrests cell cycle through RNA m6A modification

pncRNA-D is an irradiation-induced 602-nt long noncoding RNA transcribed from the promoter region of the cyclin D1 (CCND1) gene. CCND1 expression is predicted to be inhibited through an interplay between pncRNA-D and RNA-binding protein TLS/FUS. Because the pncRNA-D–TLS interaction is essential for pncRNA-D–stimulated CCND1 inhibition, here we studied the possible role of RNA modification in this interaction in HeLa cells. We found that osmotic stress induces pncRNA-D by recruiting RNA polymerase II to its promoter. pncRNA-D was highly m6A-methylated in control cells, but osmotic stress reduced the methylation and also arginine methylation of TLS in the nucleus. Knockdown of the m6A modification enzyme methyltransferase-like 3 (METTL3) prolonged the half-life of pncRNA-D, and among the known m6A recognition proteins, YTH domain-containing 1 (YTHDC1) was responsible for binding m6A of pncRNA-D. Knockdown of METTL3 or YTHDC1 also enhanced the interaction of pncRNA-D with TLS, and results from RNA pulldown assays implicated YTHDC1 in the inhibitory effect on the TLS–pncRNA-D interaction. CRISPR/Cas9-mediated deletion of candidate m6A site decreased the m6A level in pncRNA-D and altered its interaction with the RNA-binding proteins. Of note, a reduction in the m6A modification arrested the cell cycle at the G0/G1 phase, and pncRNA-D knockdown partially reversed this arrest. Moreover, pncRNA-D induction in HeLa cells significantly suppressed cell growth. Collectively, these findings suggest that m6A modification of the long noncoding RNA pncRNA-D plays a role in the regulation of CCND1 gene expression and cell cycle progression.

. Several studies have reported the involvement of m 6 A modification in the cell cycle as well. Knockdown of FTO delays cell cycle at the entry of G 2 phase in adipogenesis (33), and knockdown of METTL3 results in cell cycle arrest at G 0 /G 1 phase (34). Moreover, WTAP participates in cell cycle progression by stabilizing cyclin A2 mRNA (35), and knockdown of YTHDC1 reduces HeLa cell proliferation (36). However, despite diverse comprehensive studies of m 6 A modification, there are very few reports that discovered the function of m 6 A modification in individual RNA, especially in lncRNAs.
In this study, we demonstrated that pncRNA-D was induced after osmotic stress by recruitment of RNA polymerase II to its promoter. We also confirmed that TLS was required for repression of CCND1 expression. Moreover, pncRNA-D was highly m 6 A-methylated in the usual state, but its level decreased after exposure to cellular stresses. m 6 A methylation of pncRNA-D was preferentially recognized by nuclear RNA-binding protein, YTHDC1. The m 6 A modification of pncRNA-D shortened its life time, and YTHDC1 blocked the interaction with TLS. In addition, knockdown of pncRNA-D successfully rescued the cell cycle arrest at the G 0 /G 1 phase caused by reduced m 6 A methylation. In conclusion, we elucidated the importance of pncRNA-D and TLS in the inhibition of CCND1, and this inhibition is due to increased expression and decreased RNA m 6 A modification of pncRNA-D.

Osmotic stress induces pncRNA-D by recruiting general transcription factor proteins to the promoter region
Because pncRNA-D is induced by irradiation (6), we tested whether other stimuli could also induce pncRNA-D. HeLa cells were treated with heat shock, osmotic stress, or etoposide, which induce DNA double-stranded break. A double-stranded break was detected by ␥H2AX antibody, but no difference was observed among the stimuli tested compared with irradiation ( Fig. S1A). Although heat shock and etoposide treatment showed slight increase of pncRNA-D, only osmotic stress by 0.4 M sorbitol treatment significantly induced pncRNA-D other than irradiation (Fig. 1A, bottom panel). Therefore, in terms of convenience and induction efficiency compared with irradiation, we decided to employ osmotic stress instead of irradiation in this study. HeLa cells were treated with 0.4 M sorbitol, and the expression levels of pncRNA-D and CCND1 were examined at 2 and 3 h after treatment. As a result, negative correlation between pncRNA-D and CCND1 expression was observed (Fig. 1B).
We then validated the importance of TLS in the pncRNA-D provoked CCND1 repression system. WT HAP1 cells and TLSknockout (KO) HAP1 cells were treated with 0.4 M sorbitol and increased expression of pncRNA-D was detected in both cells  ( Fig. 1C, left panel). On the other hand, CCND1 expression was dropped to 50% after osmotic stress in WT cells, but not in the TLS-KO cells (Fig. 1C, right panel). To observe whether the overexpression of TLS in TLS-KO cells could rescue the repression of CCND1, GFP-tagged TLS overexpression vector was transfected to the TLS-KO cells. This resulted in the decrease of CCND1 expression after osmotic stress (Fig. 1C, right panel, light gray box). These data confirmed the necessity of TLS in the CCND1 repression mechanism.

Reduced m 6 A modification of lncRNA inhibits CCND1 expression
To address whether the increased expression level of pncRNA-D by osmotic stress was due to enhanced expression and/or RNA stability, HeLa cells were treated with actinomycin D to stop the nascent transcription. The expression level of pncRNA-D was measured 2 h after actinomycin D treatment, but we could not detect dramatic increase in the expression even after sorbitol treatment (Fig. 1D).
Because lncRNAs are transcribed by general transcription factors (GTFs) like mRNAs (37), we searched for TATA-binding box around the pncRNA-D transcriptional start site but could not find any typical sequences. However, pulldown assay with biotinylated pncRNA-D promoter region incubated with HeLa nuclear extract (NE) successfully exhibited that GTFs could bind pncRNA-D promoter ( Fig. 2A). Thus, the enrichment of GTFs were examined before and after osmotic stress. The enrichment of RNA polymerase II (Pol II), TATA-binding protein (TBP), and transcription factor IIB (TFIIB) was decreased at CCND1 promoter after the stress (Fig. 2B). In contrast to CCND1 promoter, the enrichment of Pol II increased after osmotic stress at pncRNA-D promoter (Fig. 2B). Although they were not significant, the recruitments of TBP and TFIIB to pncRNA-D promoter tended to increase after the stress. Taken together with the result of actinomycin D treatment, increased expression level of pncRNA-D after osmotic stress is caused by activated transcription caused by recruitment of GTFs to pncRNA-D promoter.

pncRNA-D is highly m 6 A-modified, and this methylation destabilizes pncRNA-D
Because the interplay with TLS is a pivotal step for CCND1 repression by pncRNA-D, we speculated that the interaction was influenced by the osmotic stress. RNA immunoprecipitation (RIP) assay was performed using antibody against TLS and HeLa cells with or without osmotic stress. As a result, the interaction between pncRNA-D and TLS was increased by 2-fold after the stress (Fig. 3A). In contrast to the enhanced expression level of pncRNA-D by osmotic stress, the amount of TLS in the nucleus was decreased after the treatment (Fig. 3B). TLS is known to change localization to cytoplasmic stress granules after sorbitol treatment (38), and we also confirmed the change in localization. We further quantified the alteration of arginine methylation of TLS and detected a slight decrease after the stress (Fig. 3B). This was in good agreement with our previous finding that arginine methylation of TLS blocks RNA binding (8).
Then we assumed that modification in pncRNA-D might also be influenced by osmotic stress and examined m 6 A modification, which is one of the most abundant RNA modifications that occurs in mRNA and in lncRNA. According to the meRIPseq data obtained from the NCBI's Gene Expression Omnibus database, transcripts from CCND1 promoter are likely to be modified with m 6 A, including the region of pncRNA-D (Fig.  S1B). RIP assay with anti-m 6 A antibody confirmed that pncRNA-D was highly methylated in normal condition, but its methylation dramatically decreased after osmotic stress or irradiation (Fig. 3C). We speculated on the cause of the decrease in pncRNA-D m 6 A modification and measured expression levels of m 6 A related proteins after osmotic stress or irradiation. As a result, WTAP, a component of m 6 A modification complex, was halved by both treatments, but other proteins did not show much difference or had different effects among the two treatments (Fig. S1C), at least in mRNA level. Therefore, we assume

Reduced m 6 A modification of lncRNA inhibits CCND1 expression
that the decrease in pncRNA-D m 6 A modification is caused by a reduced expression level of WTAP.
Next, we asked which of the m 6 A-binding protein(s) recognize the m 6 A modification of pncRNA-D. m 6 A modification is recognized by YTHDF1 or YTHDF2 in the cytoplasm and YTHDC1, hnRNP A2B1, or hnRNP C in the nucleus. RIP assay with antibodies against the above five proteins revealed that YTHDF1, YTHDC1, hnRNP A2B1, and hnRNP C exhibited interaction with pncRNA-D (Fig. 3D). Unexpectedly, the cytoplasmic YTHDF1 was able to bind pncRNA-D. This is because despite predominant nuclear localization of pncRNA-D, a small portion of them localize in the cytoplasm. However, we focused on YTHDC1, which bound more intensely than other m 6 A recognition proteins.
The major role of RNA m 6 A methylation is to influence the stability of RNA (39,40). Therefore, we tested whether the modification could affect the stability of pncRNA-D by knockdown of METTL3, a m 6 A modification protein, or YTHDC1. Knockdown efficiency of METTL3 or YTHDC1 by siRNAs were detected by Western blotting analysis, and their protein levels decreased dramatically after 48 h (Fig. S1D). As a result of actinomycin D treatment, the expression level of pncRNA-D in the control HeLa cells reduced down to 50% in approximately 2 h, but the knockdown of METTL3 increased stability of pncRNA-D (Fig. 3E). Meanwhile, knockdown of YTHDC1 had no effect on pncRNA-D stability. Collectively, m 6 A modifica-tion itself destabilizes pncRNA-D, but YTHDC1 is not involved in destabilization directly.

m 6 A modification also interrupts the interaction between pncRNA-D and TLS
Because YTHDC1 did not influence the stability of pncRNA-D, we expected that YTHDC1 and m 6 A modification have some other functions; thus, we examined the effect on the interaction between pncRNA-D and TLS. RIP assay by TLS antibody after knockdown of METTL3 or YTHDC1 resulted in increased interaction between TLS and pncRNA-D by ϳ2-fold (Fig. 4A). The result indicated that m 6 A modification and YTHDC1 interrupts the binding between pncRNA-D and TLS.
To determine which adenosine residue(s) of pncRNA-D has m 6 A modification, we searched for RRACH motif in the pncRNA-D sequence and found 12 motifs (Fig. S1B, blue and red boxes). m 6 A modification especially tends to occur in the consensus sequence GGACU (41)(42)(43)(44), and among the 602 nt of the full-length pncRNA-D, we could only find one motif at 465-469 nt (Fig. 4B, top panel, gray box). Notably, we previously identified that the sequence around GGACU of pncRNA-D strongly bound to RGG2-ZF-RGG3 domains of TLS, which were responsible for RNA binding (Fig. 4B, middle panel) (9). The meRIP-seq data also implied that m 6 A modification occurs at the region around 3Ј end of pncRNA-D (Fig. S1B); therefore we predicted that the GGACU motif at 465-469 nt is m 6 A- Figure 3. pncRNA-D is highly m 6 A-modified, and the modification destabilizes pncRNA-D. A, the interaction between pncRNA-D and TLS was detected by RIP assay using TLS antibody with HeLa cells before and after osmotic stress (n ϭ 5). B, Western blotting analysis using HeLa NE with or without 0.4 M sorbitol treatment. The arginine-methylated TLS level was compared with the total TLS level. ACTB was used as a loading control. A representative image of three independent experiments is shown. C, RIP assay with HeLa cell after osmotic stress or irradiation by m 6 A antibody (n ϭ 5). D, RIP assay as in A with indicated antibodies of m 6 A recognition proteins. pncRNA-D bound to each recognition protein was measured by RT-qPCR (n ϭ 3). E, relative expression level (normalized to GAPDH) of pncRNA-D after actinomycin D treatment with HeLa cells treated with siRNAs of negative control, METTL3, or YTHDC1. Expression levels at 1, 2, and 3 h after actinomycin D treatment are shown. Expression levels at 0 h in control cells were set to a value of 1.0 (n ϭ 5). *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.005.

Reduced m 6 A modification of lncRNA inhibits CCND1 expression
modified and might play an important role in TLS-pncRNA-D interaction. To determine the effect of m 6 A modification at the site, we generated a 20-nt biotinylated RNA fragment around GGACU with or without m 6 A modification (fragment m 6 A and fragment A, respectively; Fig. 4B, bottom panel). Unpredictably, although it was not significant, RNA pulldown assay with GSTtagged TLS showed stronger binding with fragment m 6 A than fragment A (Fig. 4C).
We further investigated the effect of m 6 A modification on pncRNA-D and TLS interaction using HeLa NE. RNA pulldown assay with fragment m 6 A or fragment A was conducted with HeLa NE, and as expected, endogenous YTHDC1 bound to fragment m 6 A but not to nonmethylated fragment A (Fig. 4D, top panel). On the other hand, TLS in HeLa NE bound strongly to fragment A compared with fragment m 6 A (Fig. 4D, middle panel), which conflicts with the data of in vitro assay (Fig. 4C). We anticipated that YTHDC1 in the HeLa NE might block the interaction between TLS and fragment m 6 A. To test this hypothesis, NE was prepared from YTHDC1 knockdown HeLa cells, and the same experiment was conducted. As a result, TLS bound to fragment m 6 A at a comparable level with fragment A (Fig. 4D, bottom  panel), exhibiting the competitive effect of YTHDC1 on TLS-pncRNA-D interplay.
Next, we evaluated the inhibitory effect of YTHDC1 on interaction between fragment m 6 A and TLS using Escherichia coli. The results of Western blotting analysis after RNA pulldown assay were cited from our previous paper in 2016 (9). 20-nt biotinylated fragment around GGACU was generated with (fragment m 6 A) or without (fragment A) m 6 A modification. C and D, RNA pulldown assay using fragment A or m 6 A incubated with GST-TLS (C) or HeLa nuclear extract (D). RNA-bound proteins were detected by the indicated antibodies. Representative images of three individual experiments are shown. The band intensity of TLS bound to fragment A or fragment m 6 A in HeLa NE or YTHDC1 KD HeLa NE was quantified by ImageJ (RRID:SCR_003070) (D, right) (n ϭ 3); *, p Ͻ 0.05. E, RNA pulldown assay using fragment m 6 A incubated with GST-TLS and various concentration of strep-YTHDC1. Fragment m 6 A was incubated with GST-TLS and strep-YTHDC1 at the indicated amount, and the RNA bound GST-TLS was detected by Western blotting analysis.

Reduced m 6 A modification of lncRNA inhibits CCND1 expression
overexpressed strep-YTHDC1. strep-YTHDC1 was incubated with fragment A or fragment m 6 A, and RNA pulldown assay confirmed that it preferentially bound to fragment m 6 A (Fig.  S1E), which resembles the characteristic of endogenous YTHDC1. Then various concentration of strep-YTHDC1 were incubated with GST-TLS and fragment m 6 A. We observed decreased interaction of GST-TLS with fragment m 6 A from the strep-YTHDC1 concentration at one-fourth of GST-TLS (Fig. 4E, lane 3). By incubating same concentration of GST-TLS and strep-YTHDC1, GST-TLS was not able to bind fragment m 6 A at all (Fig. 4E, lane 5). Collectively, YTHDC1 binds to m 6 A-methylated pncRNA-D and inhibits the interaction between pncRNA-D and TLS.
To confirm the importance of GGACU motif of pncRNA-D, we removed the sequence around the motif with the CRISPR/ Cas9 system. HAP1 cells were selected for the assay, because they are haploid cells, and the CCND1 inhibitory system by TLS and pncRNA-D works in these cells as demonstrated in Fig. 1. We obtained two clones that lack the GGACU motif with a 20or 46-nt deletion, named ⌬20 or ⌬46, respectively (Fig. 5A). According to the results of the RIP assay, the m 6 A modification level was halved in both ⌬20 and ⌬46 cells (Fig. 5B). The expression level of pncRNA-D was slightly higher in mutated cells, and that of CCND1 was ϳ60% of the WT (Fig. 5C). Luciferase reporter assay was conducted to validate the effect of deleted genomic regions on promoter activity. As a result, the sequence deleted in ⌬20 or ⌬46 cells had no significant effect on CCND1 promoter activity (Fig. 5D), suggesting the reduced CCND1 expression was due to transcripts lacking GGACU motif but not the deleted genomic sequences. Because ⌬20 cells had shorter deletion, we further analyzed ⌬20 cells for elucidating the function of the pncRNA-D GGACU motif.

Reduced m 6 A modification of lncRNA inhibits CCND1 expression
Reduced m 6 A modification level stabilized pncRNA-D and altered interaction with RNA-binding proteins in HeLa cells (Figs. 3 and 4); therefore we inquired whether deletion of GGACU motif affect the stability and/or the interactions. Actinomycin D was treated to WT and ⌬20 HAP1 cells, but the amount of pncRNA-D decreased similarly in both cells (Fig. 5E). Next, the binding between pncRNA-D with TLS or YTHDC1 was detected in WT or ⌬20 cells. The results clearly showed that GGACU deletion enhanced the binding with TLS by 1.8fold and reduced the interaction with YTHDC1 (Fig. 5F), which supports the YTHDC1 inhibitory effect observed in Fig. 4. The above results apparently indicate the GGACU motif at positions 465-469 of pncRNA-D is m 6 A-modified and responsible for TLS and YTHDC1 binding but not stabilization.

pncRNA-D is involved in the normal cell cycle
Finally, we considered the participation of pncRNA-D and m 6 A modification in cell cycle and cell proliferation. HeLa cells were fractionated into G 0 /G 1 , S, and G 2 /M phase cells according to the amount of DNA by FACS (Fig. 6A, top panel). Expression levels of pncRNA-D and CCND1 were measured by RT-qPCR in each fraction, and again negative correlation between pncRNA-D and CCND1 was observed (Fig. 6A, bottom panel) as in response to the osmotic stress (Fig. 1B). These data raised the possibility that pncRNA-D is also involved in normal cell cycle, not only in response to cellular stresses.
HeLa cells were treated with siRNAs of pncRNA-D, METTL3, or YTHDC1 to investigate the effect of pncRNA-D and RNA methylation on cell cycle. By knockdown of pncRNA-D, we detected a slight increase in the population of S phase cells compared with control HeLa cells, although it was not significant (Fig. 6, B and C). Knockdown of METTL3 was implicated to arrest cell cycle at G 0 /G 1 phase (34), and we also validated that si-METTL3 increased cell population of G 0 /G 1 phase and loss of G 2 /M phase peak (Fig. 6, B and C). A similar effect was observed by knockdown of YTHDC1 (Fig. 6, B and C). In addition to the above effects, osmotic stress also increased the population of the G 0 /G 1 phase and abolished the G 2 /M phase peak (Fig. 6, B and D). Moreover, we evaluated the effect of pncRNA-D knockdown at the same time with the above treatments. As a result, double knockdown of METTL3 or YTHDC1 with pncRNA-D partially rescued the G 0 /G 1 phase cell cycle arrest (Fig. 6C) and exhibited recovery of G 2 /M phase peak (Fig. 6B). On the other hand, pncRNA-D knockdown in the osmotic stress-treated HeLa cells showed a slight shift of G 0 /G 1 peak (Fig. 6B) but had no significant effect (Fig. 6D).
Additionally, the expression level of CCND1 was measured after knockdown of METTL3, YTHDC1, and/or pncRNA-D. Knockdown of pncRNA-D increased CCND1 expression level by 2-fold, and the knockdown of METTL3 or YTHDC1 dramatically decreased its expression (Fig. 6E). Notably, the expression of CCND1 was recovered to control level by double knockdown of METTL3 or YTHDC1 with pncRNA-D. Taken together, pncRNA-D without m 6 A or YTHDC1 reduces CCND1 expression and subsequently arrest cell cycle.
Cell proliferation assay revealed that knockdown of pncRNA-D modestly promoted cell growth, although it was not significant (Fig. 7A). Knockdown of METTL3 or YTHDC1 reduced HeLa cell growth and by knockdown of pncRNA-D at the same time rescued the growth to some extent (Fig. 7A). HAP1 lacking GGACU motif (⌬20) proliferated slower compared with control cells (Fig. 7B), supporting the importance of m 6 A modification of this motif in cell proliferation. Furthermore, the growth was delayed by inducing full-length pncRNA-D into HeLa cells, but the antisense pncRNA-D did not show such suppression (Fig. 7C). In conclusion, pncRNA-D is also involved in ordinary cell cycle and proliferation through m 6 A modification.

Discussion
In this study, we demonstrate that lncRNA designated as pncRNA-D could be induced by osmotic stress and reduces expression of CCND1 by interacting with RNA-binding protein, TLS. pncRNA-D is highly modified by m 6 A in the usual state, but its level dramatically decreases after osmotic stress or irradiation. This methylation of pncRNA-D has effect on its stability and interaction with TLS. m 6 A modification of pncRNA-D is recognized by nuclear YTHDC1, and this binding prohibits the interaction between pncRNA-D and TLS. Moreover, reduced RNA methylation level leads to cell cycle arrest, and pncRNA-D knockdown successfully rescues this arrest. In summary, we propose a model in which pncRNA-D inhibits the expression of CCND1 via RNA methylation alteration and subsequently arrests the cell cycle (Fig. 7D).
Osmotic stress is not a very popular treatment, but it is one of the cellular stresses that induces ncRNAs. DoGs (downstream of genes) are a type of ncRNAs induced by osmotic stress and are transcribed from the 3Ј regions of many coding genes (45). In addition, osmotic stress could alter gene expression of coding and noncoding genes (46). Osmotic stress and irradiation caused a similar level of DNA double-stranded break (DSB), and hence, we postulated that signaling cascade activated by DNA DSB might recruit general transcription factors to the pncRNA-D promoter. One candidate is signaling pathway activated by ATM/ATR. ATM is activated by DNA DSB, causes cell cycle arrest at G 1 or G 2 , and activates DNA-repairing and stress-responsive genes. The signaling cascade downstream of ATM/ATR includes p53 and p21, of which responsible for G 1 cell cycle arrest (47,48), which we observed by osmotic stress and METTL3 or YTHDC1 knockdown.
pncRNA-D is not the only lncRNA induced by irradiation or osmotic stress at the CCND1 promoter, but there are several other lncRNAs induced by cellular stresses (e.g. pncRNA-A, -B, and -E) (6). This study only focused on pncRNA-D because it showed the highest induction level after irradiation among the pncRNAs, but there is a possibility that pncRNAs work coordinately to inhibit CCND1 expression. In fact, other group has reported that pncRNA-B, which is transcribed just before transcriptional initiation site of CCND1, is highly expressed in Ewing Sarcoma patients (49). This lncRNA also negatively regulates the expression of CCND1 and is involved in cell proliferation of Ewing sarcoma cell lines.
meRIP-seq data analysis and motif search indicated that the adenosine residue in the GGACU motif at nt 465-469 of pncRNA-D is likely to be methylated. The fact that most of the m 6 A modifications in mRNAs and several ncRNAs occur

Reduced m 6 A modification of lncRNA inhibits CCND1 expression
in regions around 3Ј end also supports this speculation (15,50). Experiments with HAP1 cells lacking the GGACU motif resulted in decreased m 6 A level and enhanced TLS interaction, but RNA stability was not affected. Because the reduced methylation level destabilized pncRNA-D, other RRACH motifs in pncRNA-D should be m 6 A-modified and functions in RNA stabilization. Nevertheless, our data show that m 6 A modification in pncRNA-D could alter its stability and interaction with RNA-binding proteins in HeLa and HAP1 cells. Normally, pncRNA-D is highly methylated, and this modification leads to pncRNA-D degradation, and interaction with YTHDC1 inhibits binding to TLS. Therefore, histone acetyltransferase activity of CBP is maintained, and CCND1 is activated. However, once the cell gets damaged by cellular stresses, pncRNA-D expression is enhanced, and decreased methylation stabilizes it and thereafter recruits TLS to the promoter. Interaction with pncRNA-D enables TLS to inhibit histone acetyltransferase (HAT) activity of CBP and subsequently reduces the expression of CCND1, which results in cell cycle arrest. Sama et al. (38) have reported that TLS changes its localization to cytoplasm, especially to stress granules, after 0.4 M sorbitol treatment. We also confirmed that population of nuclear TLS decreased after the osmotic stress, and moreover, the arginine methylation level of TLS in the nucleus was reduced after the treatment. Interestingly, other groups reported that down-regulated arginine methylation by knockdown of PRMT1 reduces cytotoxic TLS cytoplasmic localization (51,52). In other words, arginine-methylated TLS changes its localization to cytoplasm. Despite the decrease of TLS in the nucleus by osmotic stress, nonmethylated TLS in the nucleus is able to bind pncRNA-D and assumed to be enough for CCND1 repression by pncRNA-D.

Reduced m 6 A modification of lncRNA inhibits CCND1 expression
We have previously reported that the methylation of Arg 476 of TLS inhibits the interaction with RNA (8). The present study demonstrated that pncRNA-D m 6 A modification is decreased by osmotic stress, and both arginine and RNA methylation work negatively in the TLS-RNA interplay. Collectively, we propose that the diverse alteration of RNA and protein methylation in response to osmotic stress regulates RNA-protein interaction. m 6 A modification of pncRNA-D significantly shortened its half-life. Knockdown of m 6 A-binding protein YTHDF2 stabilizes mRNAs in cytoplasm (23,24,26), but not in the nucleus. The degradation of RNAs in the nucleus is caused by exosome including DGCR8. For example, Hsp70 is degraded by DGCR8 in the nucleus after heat shock (53). Of note, DGCR8 can recognize m 6 A modification (54), and consequently, pncRNA-D could be degraded by DGCR8. Further studies will be required to elucidate the interaction between pncRNA-D and DGCR8 and its involvement in pncRNA-D degradation.
YTHDC1 is a nuclear RNA-binding protein responsible for RNA splicing (28,55). However, its function by recognizing m 6 A of pncRNA-D is presumably not splicing, because pncRNA-D is a single exon lncRNA. Thus, we predicted that it may alter the binding between pncRNA-D and TLS, and the result clearly showed that YTHDC1 could inhibit the interaction. To best of our knowledge, this is the first report that proposes a competitive effect of YTHDC1 on protein-RNA interplay.
Osmotic stress had a similar effect on cell cycle as knockdown of METTL3 or YTHDC1, but the knockdown of pncRNA-D in addition to above treatments had different effects. Unlike double knockdown of METTL3 or YTHDC1 with pncRNA-D, knockdown of pncRNA-D could not rescue the cell cycle arrest at G 0 /G 1 phase caused by the osmotic stress. This is because osmotic stress could decrease cyclin D1 protein level directly (56). Therefore, cell cycle arrest could not be rescued by the knockdown of pncRNA-D, which was not enough to overcome the effect of cyclin D1 destabilization after osmotic stress. We should note that knockdown of METTL3 or YTHDC1, or osmotic stress might cause cell cycle arrest independent of pncRNA-D and CCND1 by altering expression of other cell cycle-related genes.
In the article in which we demonstrated the TLS-pncRNA-D inhibitory system in CCND1 expression (6), we indicated that pncRNA-D principally works in cis, but because the exogenous pncRNA-D suppressed the proliferation of HeLa cells, there is a possibility that it could also work in trans. Several lncRNAs have been reported to work both in cis and in trans (57,58), but generally, individual lncRNA regulate different genes in cis or in trans. However, there is lncRNA such as ANRIL, which is able to reduce expression of neighboring gene (CDKN2B) both in cis and in trans (59). We should further validate whether the pncRNA-D inhibitory effect is not specific to HeLa cells but could also work in cells derived from other types of cancers. If we could confirm the effect in various cancer-derived cell lines, this sequence could be applied for seed sequence for oligonucleotide therapeutics for suppressor of cancer progression.
In summary, we report the importance of RNA methylation in modestly expressed lncRNA, which controls the stability and interaction with RNA-binding proteins. Our data provide precise regulation of CCND1 expression by lncRNA and RNAbinding proteins after cellular stresses and also in the normal cell cycle.

Cell culture and nuclear extract preparation
HeLa cell culture and nuclear extract preparation were conducted as previously described (9,60). For nuclear extract with YTHDC1 knockdown HeLa cells, the cells were cultured in a 15-cm dish with 70% confluence, and AllStar negative control siRNA (1027280, Qiagen) or siRNAs targeting YTHDC1 (GGA UGA UAU UUU AAC UGA A, and CAG UAA AGA UCG GAC GUG A) were transfected by RNAiMAX (13778150, Invitrogen) at the final concentration of 10 nM. The cells were cultured for 48 h in 37°C with 5% CO 2 , and the nuclear extract was prepared as above. HAP1 cells (control and TLS-KO cells, HZGHC-001314c002 Horizon) were cultured in Iscove's modified Dulbecco's medium supplemented with 10% FBS and 1ϫ penicillin/streptomycin (PS) (26252-94, Nacalai Tesque). For GFP-TLS overexpression, the TLS coding sequence was cloned into pAcGFP1-C3 vector (632482, Clontech Laboratories) with the primers listed in Table 1, and the plasmid was transfected to TLS-KO HAP1 cells by Lipofectamine 3000 (L3000015, Invitrogen) according to the manufacturer's protocol.

Irradiation, heat shock, osmotic stress, and actinomycin D treatment
Irradiation was treated to HeLa cells as previously described (6). For heat shock, HeLa cells were cultured at 80 -90% confluence in 24-well plate and incubated at 43°C for 30 min. For osmotic stress, HeLa cells were cultured at 80 -90% confluence, and the media were exchanged to DMEM, 10% FBS, 1ϫ PS, 0.

cDNA preparation and RT-qPCR
Total RNAs were collected by the ReliaPrep RNA cell miniprep system (Z6012, Promega). cDNA was prepared by a PrimeScript RT reagent kit with gDNA Eraser (RR047A, TaKaRa) using 500 ng of total RNAs. RT-qPCR was performed with KOD-SYBR (QKD-201, Toyobo) according to the manufacturer's instructions using the primers listed in Table 1.

RNA pulldown assay and overexpression of GST-TLS and strep-YTHDC1
RNA pulldown assay was performed as previously described (9) with slight modifications. Biotinylated 200-nt RNA fragments used in the assays (Fragments A and m 6 A) were generated and purchased from Hokkaido System Science. Fragments A or m 6 A were incubated with GST-TLS and various concentrations of strep-YTHDC1. Overexpression of GST-TLS was performed as previously described (8). For strep-YTHDC1, the coding region was amplified by PrimeSTAR GXL DNA polymerase (R050A, TaKaRa) with primers listed in Table 1, and the fragment was inserted to pASK-IBA5plus vector (2-1404-000, IBA). Overexpression and purification were conducted as previously described (8).

Cell cycle analysis
HeLa cells were transfected with AllStar negative control siRNA, si-pncRNA-D, si-METTL3, or si-YTHDC1 for 48 h or treated with 0.4 M sorbitol for 2 h and stained with 10 g/ml Hoechst 33342 (B2261, Sigma) for 1 h at 37°C. The cells were then analyzed by cell sorter (SH800Z, SONY). To collect RNAs from the G 0 /G 1 , S, and G 2 /M phase cells, the cells were fractionated by the DNA contents, and total RNA was isolated as described above.

Overexpression of pncRNA-D sense and antisense strand
602-nt full-length pncRNA-D and antisense pncRNA-D was in vitro transcribed by a MEGAscript kit (AM1334, Thermo Fisher Scientific) according to the manufacturer's instructions. Briefly, sense and antisense sequence of pncRNA-D were inserted to pcDNA 3.1 (ϩ) vector (V790-20, Thermo Fisher Scientific), and RNA was synthesized by T7 transcription.

Reduced m 6 A modification of lncRNA inhibits CCND1 expression Cell proliferation assay
HeLa cells were harvested in 96-well plate at 2.5 ϫ 10 3 cells/ well. The next day, the cells were transfected with AllStar negative control siRNA or siRNAs targeting pncRNA-D, METTL3, or YTHDC1 by RNAiMAX at the final concentration of 10 nM or induction of 5 pmol sense or antisense strand pncRNA-D by JetMessenger (150-001, Polyplus) according to the manufacturer's protocol. Cell proliferation rates were measured by Cell Count Reagent SF (07553-15, Nacalai Tesque) according to the manufacturer's instructions.

Me-RIP-seq data analysis
Me-RIP-seq data by anti-m 6 A antibody were obtained from GEO accession viewer (GSE46705). Sequence reads were subjected to filtration for quality check by FastQC, the low quality (Q Ͻ 33) and short reads were excluded, and adapter sequences were removed by fastx_toolkit. Subsequently, the reads were mapped to the human genome (hg18 assembly) using TopHat software.

Lentivirus production for removing GGACU motif in pncRNA-D
The guide RNA targeting pncRNA-D GGACU motif was designed by CRISPRdirect (RRID SCR_018186; Table 1), and the specific target sequence was cloned into lenti-CRISPR vectors and verified by DNA sequencing. To produce lentivirus, HEK 293FT cells seeded in 10-cm plates were transfected with 6 g of lenti-CRISPR-v2 (52961, Addgene), 4 g of psPAX2 (12260, Addgene), and 3 g of pLP-VSVG (Invitrogen) plasmids using Lipofectamine 2000 (11668027, Invitrogen) according to the manufacturer's protocol. HEK 293FT cells were cultured for 24 h in DMEM, 10% FBS, 1ϫ PS, and the media were exchanged to Iscove's modified Dulbecco's medium, 10% FBS. After 24 h of culture, the supernatant containing lentivirus was collected, and HAP1 cells were infected with the generated lentiviruses in the presence of 8 g/ml Polybrene and selected with 1 g/ml puromycin for 6 days. The cells were collected, and a single cell per well was seeded in a 96-well plate by cell sorter. The genome sequence of each clone was examined, and two clones lacking the GGACU motif (⌬20 and ⌬46) were obtained.

Luciferase reporter assay
CCND1 promoter regions of control or GGACT removed HAP1 cells (⌬20 and ⌬46) were cloned into pGL4. 10[luc2] vector (E6651, Promega) with the primers listed in Table 1. 50 ng of empty vector or the above constructs, and 50 ng of Renilla luciferase plasmid (pRL-TK, E2241, Promega) were transfected by Lipofectamine 3000 (Invitrogen) into HAP1 cells seeded in a 96-well plate, according to the manufacturer's protocol. The cells were cultured in 37°C, 5% CO 2 and lysed 48 h after transfection, and luciferase activity was measured using the Dual-Luciferase assay kit (E1910, Promega). Luciferase activities were normalized to Renilla luciferase activities.

Statistical analysis
The data are presented as means Ϯ standard deviations. Pairwise differences were identified using two-tailed Student t test and were considered significant when p Ͻ 0.05.