Introduction

DEAD box proteins are RNA helicases that contain the characteristic Asp-Glu-Ala-Asp (DEAD) motif1,2. In humans, 37 different DEAD-box RNA helicases play important roles in RNA metabolism, including translation, splicing, and ribosomal RNA processing3,4,5,6. DDX1 is a founder member of the DEAD protein family. It was identified as a co-amplified gene with MYCN in retinoblastoma and neuroblastoma cells7,8. The importance of DDX1 in testicular and colorectal tumors has also been reported9,10, suggesting that DDX1 is involved in tumorigenesis, although its functions remain unclear at the molecular level.

To date, the molecular functions of DDX1 in mammals have been extensively studied, revealing it plays a critical role in various processes, including mRNA splicing, microRNA processing, viral replication, and embryonic development11,12,13,14,15. Furthermore, we recently reported that loss of Ddx1 in mouse embryonic stem cells (mESCs) resulted in rRNA processing defects, thereby inducing apoptosis via ribosome stress-mediated p53 activation16. In the study, we employed the all-in-one conditional knockout (cKO) method to generate Ddx1 cKO mESCs16,17. This system enabled a highly efficient one-step generation of cKO cells in mESCs, as well as in the pseudo-diploid human fibrosarcoma cell line HT1080.

Among human tRNAs, Arg-TCT, Ile-TAT, Leu-CAA, and Tyr-GTA tRNAs contain introns, which are inserted at one base 3´ side from the anticodon sequence. The splicing mechanism of tRNAs is distinct from that of mRNAs18. The introns of pre-tRNAs are cleaved by tRNA splicing endonuclease (TSEN), which leaves 2´, 3´-cyclic phosphate on the 3´-end of the 5´-exon and a hydroxyl group at the 5´-end of the 3´-exon. In higher eukaryotes such as humans and mice, the cleaved pre-tRNAs (intermediate tRNAs) are then ligated by an RNA ligase RTCB. The active site of RTCB is first guanylated, and the GMP moiety is then transferred to the 2´, 3´-cyclic phosphate termini of intermediate tRNAs for the ligation reaction19,20,21,22. Popow et al. demonstrated that DDX1 cooperates with Archease (also known as ZBTB8OS) to enhance the guanylation of RTCB in vitro; however, it is unclear whether this mechanism works under physiological conditions11. In addition to intronic tRNAs, XBP1 mRNA is another target of non-spliceosomal splicing 23. Cleavage of the unspliced XBP1 mRNA (XBP1u) is initiated by ER stress-induced activation of the endoribonuclease ERN1 (endoplasmic reticulum to nucleus signaling 1; also known as inositol-requiring enzyme 1α [IRE1α]) on the ER membrane. Similar to intronic tRNA splicing, the cleaved XBP1 mRNA is then ligated by RTCB to generate spliced XBP1 (XBP1s), which encodes the transcription factor XBP124,25,26. In this study, we clarified whether DDX1 is required for the non-spliceosomal splicing of tRNAs and XBP1 mRNA in live cells using human cell lines.

Results

Establishment of the DDX1 cKO U2OS cells

We previously demonstrated a one-step and highly efficient generation of cKO cells via the all-in-one cKO method16. The all-in-one cKO method utilizes CRISPR/Cas9-assisted homologous recombination of the targeting vector containing the FRT-flanked promoter-less EGFP gene, followed by a P2A peptide sequence, a FLAG-tag sequence, and the coding sequence of the target gene upstream of the CRISPR/Cas9 cleavage site. The cKO cells expressed the endogenous promoter-driven FLAG-tagged target gene and EGFP, allowing the target gene expression monitoring. We demonstrated this method is effective in diploid mESCs and the human pseudodiploid cell line HT1080. However, general human cancer cell lines, which have been utilized for studying gene function, are primarily hyperploid. To enable the molecular analysis of DDX1 in hyperploid cell lines, we applied the all-in-one cKO system to the human osteosarcoma cell line U2OS (Fig. 1A, Supplementary Fig. 1). Fluorescence in situ hybridization (FISH) analysis showed that U2OS cells had four DDX1 loci in the genome (Supplementary Fig. 2A). Genomic PCR analyses of the isolated U2OS clones showed that all the clones had targeted alleles, and 10 out of 12 clones did not have non-targeted alleles (Fig. 1B). Furthermore, all clones expressed EGFP at comparable levels, suggesting that the EGFP unit was correctly inserted into the DDX1 locus (Supplementary Fig. 2B). Next, we randomly selected six clones and analyzed the DDX1 protein levels via western blotting (Fig. 1C). Genetically replaced DDX1 protein was detected as a slightly larger molecular weight band in the targeted clones due to the FLAG tag, whereas endogenous DDX1 protein was not detected in any of them. Consistently, FLAG-tagged DDX1 was detected only in the target clones. These data suggested that the all-in-one cassette was efficiently targeted to all DDX1 alleles. We further validated the targeting of DDX1 alleles using Southern blotting. While one of the clones (#8) showed aberrant bands, two out of three clones presented only the targeted band (Fig. 1D). The expression of DDX1 mRNA in the targeted clone was comparable to that in the parental U2OS cells (hereafter, the DDX1 cKO cell without FLP-induced deletion is referred to as DDX1 cKOF) (Fig. 1E). We verified that DDX1 KO cells could be separated after transient expression of FLP recombinase, followed by sorting of the EGFP-negative cells (Fig. 1F, G). Although DDX1 KO cells showed a significant growth defect, they were still viable without massive cell death at least by Day 13 after KO in contrast to mESCs (Fig. 1H)16. These results demonstrate that homozygously-targeted cKO cells can be efficiently isolated from hyperploid cells using all-in-one cKO methods.

Fig. 1: Establishment of DDX1 cKO U2OS cells via all-in-one cKO method.
figure 1

A Strategy to generate U2OS DDX1 cKO cells. An arrow indicates the CRISPR/Cas9 target site. White triangle, FRT sequence; arrowhead, primers for genome PCR; N, NdeI recognition sequence. The blue box represents exon 3 upstream of the CRISPR/Cas9 cleavage site. The red box represents exon 3 downstream of the CRISPR/Cas9 cleavage site. The yellow box represents the ORF of exon 2. B Twelve isolated clones were analyzed by genome PCR. The upper band represents the targeted allele, while the lower band represents the non-targeted allele. The non-targeted band of clone 7 is larger than that of control, which is likely due to a large insertional mutation in the non-targeted allele caused by the function of CRISPR/Cas9. C Expression of DDX1 in the indicated clones was analyzed via western blotting. C, U2OS cells. D NdeI-digested genomic DNA of clones #2, #8, and #10 was analyzed by Southern blotting. C, U2OS cells. Clone #2 was used as U2OS DDX1 cKO cells hereafter. E The expression of DDX1 was analyzed by qRT-PCR (n = 3; NS, not significant). F The expression of the fluorescent protein was analyzed via flow cytometry. Cells were analyzed on day 2 after sorting. G Expression levels of DDX1 in EGFP-positive (cKOF) and EGFP-negative (KO) cells were analyzed via western blotting. H The number of DDX1 cKOF and KO cells were counted on the indicated days after sorting (n = 3; *P < 0.05; ***P < 0.005).

Full size image

Function of DDX1 in non-spliceosomal splicing

Splicing of intron-containing tRNAs is initiated by TSEN complex-mediated digestion of the intronic sequence. The RNA ligase RTCB then ligates the cleaved tRNA to generate a mature tRNA (Fig. 2A). Since DDX1 reportedly forms a complex with RTCB21,27,28, we investigated whether DDX1 is involved in the splicing of tRNA in vivo. We analyzed the splicing of intron-containing tRNAs (Tyr-GTA, Ile-TAT, Leu-CAA, and Arg-TCT) by northern blotting with oligo-probes designed on 5´ exon of tRNAs and found that intermediate tRNAs were markedly increased in DDX1 KO U2OS cells (Fig. 2B, C). The accumulation of intermediate tRNA was also verified with oligo-probes designed on 3´ exon of Tyr-GTA tRNA (Supplementary Fig. 3). The intermediate tRNA was not detected in the intron-less Val-CAC tRNA, suggesting that the intermediate tRNA was generated through the interruption of tRNA splicing. We also confirmed that the basal level of intermediate tRNA in DDX1 cKOF cells was similar to that of parental U2OS cells (Supplementary Fig. 4A). To rule out the possibility that the severe cell growth inhibition of the DDX1 KO cells caused the tRNA splicing defect (Fig. 1H), we analyzed the tRNA splicing in serum-starved U2OS cells (Supplementary Fig. 4B, C). While serum-starved U2OS cells showed severely reduced cell growth, the tRNA splicing was not affected in these cells. The protein levels of RTCB, family with sequence similarity 98 (FAM98B), and RNA transcription, translation, and transport factor (RTRAF, commonly known as CGI99), which form a complex with DDX1, were similar in DDX1 KO cells (Fig. 2D). These results suggest that DDX1 is required for tRNA splicing in the U2OS cells.

Fig. 2: Function of DDX1 in non-spliceosomal splicing.
figure 2

A Splicing pathway of intronic tRNAs in humans. TSEN cleaves off intron from pre-tRNA, and RTCB ligates the 5´ and 3´ exon of tRNAs. B Total RNA (2.5 µg) was loaded to analyze splicing of indicated tRNAs by northern blotting with oligo-probes designed on 5´ exon of tRNAs. The samples were harvested on day 8 after sorting KO cells. Black triangle, mature tRNA; white triangle, intermediate tRNA. C The ratio of intermediate and mature tRNA was quantified (n = 3; *P < 0.05; **P < 0.01; ***P < 0.005). D Expression levels of DDX1-binding factors in the DDX1 cKOF and KO cells were analyzed via western blotting. E Schematic representation of the primer position to detect XBP1u and XBP1s expression via endpoint RT-PCR and qRT-PCR. Blue arrowheads, primers for endpoint RT-PCR; black arrowhead, a primer for qRT-PCR of XBP1u; white arrowhead, a primer for qRT-PCR of XBP1s; gray arrowheads, a qRT-PCR primer common to XBP1s and XBP1u. F DDX1 cKOF, KO, and ERN1 KO cells were treated with 10 µg/ml tunicamycin for the indicated period, and the expression levels of XBP1 and XBP1u were detected via endpoint RT-PCR. (G) DDX1 cKOF, KO, and ERN1 KO cells were treated with 10 µg/ml tunicamycin for the indicated period, and the expression of XBP1s was analyzed via qRT-PCR (n = 3; NS, not significant; ***P < 0.005). H The ratio of XBP1s to XBP1u mRNA expression was calculated based on qRT-PCR data (n = 3; NS, not significant; *P < 0.05). I DDX1 cKOF, KO, and ERN1 KO cells were treated with 10 µg/ml tunicamycin for the indicated period, and the expression of ERDJ4 was analyzed via qRT-PCR (n = 3; NS, not significant; *P < 0.05).

Full size image

To investigate whether DDX1 also regulates non-spliceosomal splicing of XBP1 mRNA, we analyzed spliced and unspliced XBP1 mRNA expression in tunicamycin (ER-stress inducing reagent)-treated DDX1 KO cells via endpoint RT-PCR and qRT-PCR (Fig. 2E–H). We found that the loss of DDX1 did not affect the ER stress-induced expression of XBP1s and the ratio of XBP1s/XBP1u. We confirmed that genetic disruption of the ERN1 gene, whose product is known to cleave the intron of XBP1 mRNA, suppressed the generation of XBP1s (Fig. 2F–H and Supplementary Fig. 6A). Consistently, the expression of the XBP1 downstream gene, ERDJ4, was decreased by ERN1 disruption but not altered by the loss of DDX1 (Fig. 2I). These results indicate that DDX1 is dispensable for the splicing of XBP1 mRNA. We confirmed that DDX1-disrupted HEK293 cells also showed tRNA but not XBP1 splicing defect (Supplementary Fig. 5). These results indicate tRNA-specific non-spliceosomal splicing defect found in DDX1 KO U2OS cells are not cell type-specific.

We next investigated the function of RTCB in non-spliceosomal splicing using RTCB-deficient U2OS cells prepared by transfecting a CRISPR/Cas9 vector designed in the coding region of the RTCB locus (Supplementary Fig. 6B, C). Disruption of RTCB was confirmed by sequencing and western blotting analyses (Supplementary Fig. 6D, E). The protein expression of RTCB complex components almost completely disappeared, and the cells showed a dramatic growth defect similar to DDX1 KO cells, consistent with a previous report (Supplementary Fig. 6F)21. In contrast to DDX1 KO cells, RTCB KO cells exhibited defects in both tRNA and XBP1 mRNA splicing (Fig. 3A–E). Moreover, disruption of the Archease gene, whose product guanylates RTCB and is thus required for the ligase activity of RTCB, also resulted in splicing defects in both tRNA and XBP1 mRNA (Fig. 3C, D, F–H and Supplementary Fig. 6G–I)11,29. Of note, unlike DDX1 KO cells, RTCB KO and Archease KO cells did not show a reduction of the mature tRNA in this setting. This is because RTCB KO and Archease KO cells were harvested at an earlier time-point than DDX1 KO cells. We confirmed both KO cells showed a clear reduction of the mature tRNA when harvested at a later stage (day 8 after sorting) as DDX1 KO cells (Supplementary Fig. 7). These results indicate that RTCB activity is required for the splicing of both tRNA and XBP1 mRNA, whereas DDX1 is required specifically for tRNA splicing. Similar results were also obtained with RTCB-disrupted HEK293 cells (Supplementary Fig. 8).

Fig. 3: RTCB activity is required for both tRNA and XBP1 mRNA splicing.
figure 3

A Total RNA (2.5 µg) was loaded to analyze splicing of Tyr-GTA tRNAs in RTCB WT and KO cells by northern blotting with an oligo-probe designed on 5´ exon of Tyr-GTA tRNAs. The samples were harvested on day 3 after sorting transfected cells. Black triangle, mature tRNA; white triangle, intermediate tRNA. B The ratio of intermediate and mature tRNA was quantified (n = 3; *P < 0.05). C WT, RTCB KO, and Archease KO cells were treated with 10 µg/ml tunicamycin for the indicated period, and the expression of XBP1s was analyzed via qRT-PCR (n = 3; ***P < 0.005). D The ratio of XBP1s to XBP1u mRNA expression was calculated based on qRT-PCR data (n = 3; NS, not significant; **P < 0.01; ***P < 0.005). E RTCB WT and KO cells were treated with 10 µg/ml tunicamycin for the indicated period, and the expression of ERDJ4 was analyzed via qRT-PCR (n = 3; NS, not significant; **P < 0.01). F Total RNA (2.5 µg) was loaded to analyze Splicing of Tyr-GTA tRNAs in Archease WT and KO cells by northern blotting with an oligo-probe designed on 5´ exon of Tyr-GTA tRNAs. The samples were harvested on day 3 after sorting transfected cells. Black triangle, mature tRNA; white triangle, intermediate tRNA. G The intermediate to mature tRNA ratio was quantified (n = 3; *P < 0.05). H Archease WT and KO cells were treated with 10 µg/ml tunicamycin for the indicated period, and the expression of ERDJ4 was analyzed via qRT-PCR (n = 3; NS, not significant; ***P < 0.005).

Full size image

Requirement of enzymatic activity of DDX1 in tRNA splicing

To analyze whether the RNA helicase activity of DDX1 is necessary for tRNA splicing, wild-type (WT) or ATP-binding defective mutant (K52N) DDX1 was introduced into DDX1 KO U2OS cells11. The expression of WT or K52N mutant DDX1 protein was verified by western blotting and flow cytometry analyses (Fig. 4A, B). We observed that WT but not K52N DDX1 rescued the tRNA splicing defect in DDX1 KO cells (Fig. 4C, D). Interestingly, pull-down analysis revealed that the loss of DDX1 induced decreased FAM98B and CGI99 in the RTCB complex (Supplementary Fig. 9A). Thus, there was a possibility that DDX1 is required for tRNA splicing just to maintain the RTCB complex. However, this is not the case since the DDX1 K52N mutant protein retained an ability to maintain the RTCB complex similar to the DDX1 WT protein, although the mutant protein did not rescue the tRNA splicing defect (Fig. 4C, D, Supplementary Fig. 9B). Taken together, these results indicate that the DDX1 activity is required for tRNA splicing.

Fig. 4: Enzymatic activity of DDX1 is required for tRNA splicing.
figure 4

A Expression of exogenously expressed DDX1 in the DDX1 KO cells was analyzed by western blotting. B Infection efficiency was verified with the expression of Ruby, a red fluorescent protein, linked to the introduced gene of interest via the P2A peptide sequence. Red, DDX1 KO U2OS cells; gray, lentiviral vector infected DDX1 KO U2OS cells. C Total RNA (2.5 µg) was loaded to analyze splicing of Tyr-GTA tRNA in DDX1 wt or K52N mutant introduced DDX1 KO U2OS cells by northern blotting with an oligo-probe designed on 5´ exon of Tyr-GTA tRNAs. Black triangle, mature tRNA; white triangle, intermediate tRNA. D The ratio of intermediate and mature tRNA was quantified (n = 3; NS, not significant; ***P < 0.005).

Full size image

Discussion

In this study, we demonstrated that DDX1 activity is important for the splicing of tRNA but not of XBP1 mRNA in human cells. Previous study reported DDX1 works cooperatively with Archease to increase ligation speed by enhancing guanylation of RTCB11. However, our results indicate that DDX1 is specifically required for the splicing of tRNA but not of XBP1 mRNA. This fact indicates that DDX1 is dispensable for RTCB guanylation in XBP1 splicing. DDX1 could work for RTCB guanylation specifically during tRNA splicing.

Many RNA helicases are shown to act as RNA chaperones in various RNA metabolism processes, including mRNA splicing1. Similarly, the helicase activity of DDX1 may be required for remodeling the cleaved tRNA terminal structure suitable for RTCB-catalyzed ligation reaction (Fig. 5). The structural differences of tRNA and XBP1 mRNA may account for the specific requirement of DDX1 for tRNA splicing. It is also possible that other RNA helicase proteins compensate for DDX1’s function in XBP1 splicing. Difference of cleaving endonucleases or sub-cellular localizations of the splicing reaction may also affect the requirement of DDX1.

Fig. 5: Proposed model of DDX1 function in tRNA splicing.
figure 5

DDX1 could be specifically required for remodeling the intermediate tRNA structure for the RTCB-catalyzed ligation reaction. Filled circles and a white circle represent anticodon sequence and 3´ terminus of 5´ exon of pre-tRNA, respectively.

Full size image

DDX1 KO cells showed decreased FAM98B and CGI99 in the RTCB complex. This is consistent with the report showing the requirement of DDX1, FAM98B, and CGI99 for the integrity of the complex, which could be due to the assembly defect and/or instability of the RTCB complex27. Further analyses are required to clarify this point.

Previously, we reported that the loss of Ddx1 resulted in apoptosis in mESCs16. Detailed analysis revealed that apoptosis was induced via ribosome stress-mediated p53 activation. However, in this study, we found that the deletion of DDX1 in U2OS cells did not induce apoptosis. This could be due to the sensitivity of mESCs to the ribosome stress pathway 30,31.

We observed inhibition of tRNA splicing in DDX1, RTCB, and Archease KO cells. However, a considerable amount of mature tRNAs remained in all of these KO cells even eight days after gene deletion/disruption (Fig. 2B, Supplementary Fig. 7A, B). The remaining mature tRNAs are also found in DDX1 KO or RTCB KO HEK293 cells. Although we cannot rule out the possibility that residual DDX1, RTCB, or Archease protein in each KO cell contributes to the persistent production of mature tRNA, these results would suggest the existence of an alternative pathway for the ligation of intermediate tRNAs. In fact, two tRNA ligation pathways are widely found: one is the RTCB-dependent 3´-phosphate ligation pathway, and the other is the Trl1 (another RNA ligase) dependent 5´-phosphate ligation pathway 32. Although tRNA splicing via the 5´-phosphate ligation pathway has not yet been clearly demonstrated in mammalian cells, a previous report showed the enzymatic activity of 5´-phosphate ligation in the nuclear extract of HeLa cells33. Furthermore, a recent study reported the discovery of C12ORF29, an RNA ligase that mediates 5´-phosphate ligation34. Thus, we speculate an alternative tRNA splicing mechanism, the mammalian 5´-phosphate ligation pathway, acts as a spare pathway, although the spare pathway cannot process whole tRNA splicing when the RTCB-dependent splicing is completely blocked. DDX1 cKO U2OS cells could be a valuable tool to clarify the unidentified ligation pathway in mammalian cells.

Methods

Cell culture

U2OS cells, obtained from the European Collection of Authenticated Cell Cultures (ECACC), were cultured in McCoy’s 5A medium (Thermo Fisher Scientific) containing 10% fetal bovine serum (FBS) (Biosera) and penicillin/streptomycin (Sigma). Cell cloning of the DDX1 cKO U2OS via the all-in-one cKO method was carried out as described previously16. To enhance the homologous recombination of the targeting vector, 2 µM of M3814 (Selleck) was added after transfection. The gRNA sequences used in this study are summarized in Supplementary Table 1. To induce KO cells, pCAG-Flpo, a gift from Massimo Scanziani (Addgene plasmid # 60662; http://n2t.net/addgene:60662; RRID:Addgene_60662), was transfected using the Neon Transfection System35. Two days after transfection, EGFP-negative cells were sorted using FACSAriaIII (BD Biosciences) to prepare DDX1 KO cells. The empty vector pCAGEN, a gift from Connie Cepko (Addgene plasmid # 11160; http://n2t.net/addgene:11160; RRID:Addgene_11160), was transfected, and EGFP-positive cells were sorted to prepare DDX1 cKOF cells as a control36. ERN1 disrupted clones were isolated by transfecting the CRISPR/Cas9 vector targeting ERN1. To prepare RTCB and Archease KO cells, modified PX458a vector encoding the improved gRNA scaffold sequence were transfected using the Neon Transfection System16,37. Transfected cells expressing the plasmid-derived fluorescent gene ametrine were sorted using FACSAriaIII. Cells were harvested on day 3 after sorting unless otherwise stated. To verify gene disruption, the genomic sequence encompassing the CRISPR/Cas9 target site was amplified by PCR using the primers listed in Supplementary Table 1, and the amplicon sequence was analyzed by sequencing. Recombinant lentiviral vectors encoding DDX1 WT, DDX1 K52N, or HA-tagged RTCB were prepared as previously described16. The Ruby gene was linked to the introduced gene of interest via the P2A peptide sequence to verify the infection efficiency. For the rescue experiment, DDX1 KO cells were infected with the lentiviral vectors on day 4, and samples were harvested on day 7. For cell counting of DDX1 KO or RTCB KO cells, 5 × 104 sorted cells were seeded in a well of a 24-well plate, and the cell number was counted on the indicated day. Pull-down experiments were performed as described previously with some modifications16. Briefly, DDX1 cKOF or KO U2OS cells were infected with a lentiviral vector encoding HA-tagged RTCB. Cell lysates prepared from them were incubated with anti-HA-tag mAb-Magnetic Beads (MBL) to immunoprecipitate HA-RTCB, and RTCB-associated proteins were detected via western blotting.

Western blotting

The following antibodies were used to confirm protein expression by western blotting: anti-DDX1 (A300–521A, Bethyl Laboratories), anti-RTCB (A305-079A-T, Bethyl Laboratories), anti-FAM98B (22251-1-AP, Proteintech), anti-CGI99 (19848-1-AP, Proteintech), anti-Archease (28608-1-AP, Proteintech), anti-ERN1 (PAC460HU01, Cloud-Clone), anti-ACTB-HRP (A00730, GenScript), anti-FLAG M2-HRP (A8592, Sigma), and anti-HA (3F10, Roche).

FISH

FISH analysis was performed as previously described38. The Cy3-labeled human RP11-422A6 probe, covering the DDX1 locus, was prepared using Nick Translation Mix (Roche). Samples were counterstained with 4´,6-diamidino-2-phenylindole (DAPI) to visualize genomic DNA.

Northern blotting

The total RNA was purified using ISOGEN (Nippon Gene). Glycogen was added to the ethanol precipitate to enhance yield. Purified RNA was separated on a 15% polyacrylamide gel containing 8 M urea. RNA was then transferred to Hybond N+ (Cytiva). After transfer, RNA was UV-crosslinked to the membrane. The membrane was incubated in a hybridization solution [250 mM SDS, 50% formamide, 4× saline-sodium citrate (SSC), 0.1% N-lauroylsarcosine, 2× Blocking Reagent (Roche), 50 mM phosphate buffer (pH 7.0)] at 50 °C for 1 h and then hybridized with 9.6 nM digoxigenin (DIG)-labeled probe overnight at 50 °C with rocking. Dig-labeled oligo DNA probes were prepared with DIG using the DIG Oligonucleotide Tailing Kit, 2nd generation (Roche), according to the manufacturer’s protocol. The probe sequences are listed in Supplementary Table 1.

qRT-PCR

Purified RNA was treated with deoxyribonuclease (RT grade) (Nippon Gene) and converted to cDNA using PrimeScript RT Master Mix (Takara) according to the manufacturer’s protocol. qRT-PCR was performed using the THUNDERBIRD® SYBR® qPCR Mix (Toyobo) and Light Cycler 480 II System (Roche). Standard curves were generated for each primer set, and the expression level of target genes was normalized to GAPDH. Primers used for qRT-PCR are listed in Supplementary Table 1.

Statistics and reproducibility

Unpaired two-tailed Welch’s t-test was performed on three independent samples to evaluate statistical significance. Data shown are means ± standard deviations. Statistical significance was set at P < 0.05.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.