The HCC-9903 cell line was presented by Professor Liu Chunsheng from the Qilu Hospital of Shandong University Bioengineering Center (Shandong, China). The cells were taken from liquid nitrogen and then thawed in a 37°C water bath. After centrifugation at 1,000 × g for 5 min, the cells were suspended in RPMI-1640 complemented with 10% FBS (Invitrogen-Life Technologies, Carlsbad, CA, USA) and seeded in 6-well plates at a density of 1×10⁵/cm².

Extracted RNA was treated with DNase I (Fermentas, Thermo Fisher Scientific, Waltham, MA, USA) to exclude the possibility of DNA contamination. The RNA concentration in each sample was quantified using a spectrophotometer at 260 nm and the purity of RNA was assessed by measuring the OD260/280 ratio (1.85–1.96). The integrity of RNA was checked by electrophoresis on a 1.0% agarose gel with ethidium bromide staining. Subsequently, 2 µg of DNase I-treated RNA was applied in the reverse transcription reaction with 2 pM specific RT primer and SuperScript III reverse transcriptase (Invitrogen-Life Technologies). PCR reactions were performed on 1 µl of strand-specific cDNA by mixing 2X Thermo Pol Taq buffer (New England Biolabs, Ipswich, MA, USA), 0.25 mM forward primer, 0.25 mM reverse primer, 0.2 mM dNTPs (Amersham, Buckinghamshire, UK), and 1.25 units Taq polymerase (New England Biolabs). The PCR conditions used were: 95°C for 5 min followed by 30 cycles of 95°C for 30 sec, 48°C for 30 sec and 72°C for 45 sec with a final incubation at 72°C for 10 min. The product was checked by electrophoresis on a 1.5% agarose gel with ethidium bromide staining.

RT-qPCR reactions were carried out in a final volume of 25 µl, using SYBR Premix Ex Taq (Takara), 0.4 mM of each primer, and 200 ng of cDNA template. Each individual sample was run in triplicate wells. PCR amplification cycles were performed using the iQ™5 Multicolor Real-Time PCR detection system (Bio-Rad, Hercules, CA, USA) and SYBR Premix Ex Taq II kit (Takara, Dalian, China). The reactions were initially denatured at 95°C for 3 min followed by 40 cycles of 95°C for 10 sec, 50°C for 30 sec, 72°C for 20 sec. The change in transcript abundance of all tested genes was calculated using the 2^−ΔΔCt method. Gene mRNA amounts were normalized to β-actin.

Northern blotting was performed according to a previous study (?). Specific hybridization probe complementary to SIRT1-AS was designed and synthesized by GenScript Co. (Nanjing, China). Total RNA (20 µg) was analyzed on a 12% polyacrylamide denaturing gel containing 7.5 M urea and transferred onto a Hybond N⁺ nylon membrane (Amersham). The membranes were cross-linked using ultraviolet light for 30 sec at 1,200 mJ/cm². Hybridization was performed according to the manufacturer's instructions. Following washing, the membranes were exposed for 20–40 h to Kodak XAR5 films (Sigma-Aldrich, St. Louis, MO, USA). As a control, the membranes were hybridized with a human U6 snRNA probe. The probes used were: SIRT1-AS, 5′-TAACACCAAATCCTCCCAAT-3′; SIRT1 V1, 5′-AGGCAGTTGGAAGATGGCGGACGAGG-3′; SIRT1 V2, 5′-CCTGAGGTTGAGGGCGGCTGGG-3′; and U6, 5′-GCAGGGGCCATGCTAATCTTCTCTGTATCG-3′.

Proliferation of HCC-9903 was determined using an Apollo^® 567 EdU staining kit (Roche, Basel, Switzerland) according to the manufacturer's instructions, and the nuclei were stained by Hoechst 33342.

Cell viability was performed using a CellTiter-Blue H cell viability assay kit (Promega, Promega, WI, USA) according to the manufacturer's instructions.

The full-length cds sequence of SIRT1 mRNA, WT SIRT1-AS and the mutant SIRT1-AS were amplified by PCR using primers containing KpnI (in forward primer) and XhoI (in reverse primer) sites. The primers used were: SIRT1 forward, 5′-CGGGGTACCTATGCTATGAACAATGGAAG-3′ and reverse, 5′-CCGCTCGAGTTGCCTGTTGAGGATTTGGT-3′; and SIRT1-AS forward, 5′-CGGGGTACCAGTCAAATGACAATTTTAATAGAC-3′ and reverse, 5′-CCGCTCGAGTTAGTGCCTGCCTGGA-3′. The products were cleaved and ligated onto corresponding sites of the pcDNA3.1 plasmid, which was confirmed by sequencing. The constructs and miR-29c mimic were transfected with X-tremeGENE DNA transfection reagents (Roche) according to the manufacturer's instructions.

The cells were lysed in lysis buffer (Beyotime Co., Shanghai, China) supplemented with 1 mM PMSF. Protein concentration was determined with the BCA protein assay kit (Tiangen). Protein (20 µg) of each sample were separated by 12% SDS-PAGE and electro-transferred to a PVDF membrane (Millipore, Billerica, MA, USA) for immunoblot analysis. The primary antibodies used were: anti-SIRT1 (1:300), anti-p53 (acetyl K382, 1:200) and anti-p53 (1:400) (all from Abcam, Cambridge, MA, USA), and anti-β-actin (1:800; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) which was used as the internal reference. After incubation with the appropriate HRP-conjugated secondary antibody, the proteins were detected using a ChemiDoc XRS imaging system and analysis software Quantity One (Bio-Rad).

To detect the impact of SIRT1-AS on SIRT1 mRNA half-life, after 48 h of transfection of each constructs or miR-29c mimic, HCC-9903 cells were treated with 2 µg/ml actinomycin D (Sigma-Aldrich), which had an inhibitory effect on transcription. The cells were collected at 0, 0.5, 1, 2, 4, 6, 8, 10 and 12 h after treatment, and then total cell RNA was isolated to measure the residual mRNAs by RT-qPCR. GAPDH mRNA was applied as an internal control, which was relatively stable within 32 h.

According to the sequencing result of SIRT1-AS, full-length WT and mutant SIRT1-AS were synthesized, both of which contained the complimentary sequence of the miR-29c binding site and the no. 622 mutation site. Additionally, 1 µg of each fragment was incubated with synthetic SIRT1 mRNA (a part complementary to SIRT1-AS) at 37°C for 3 h under simulated physiological conditions (10 mM Tris-HCl pH 7.4, 15 mM NaCl). Then, 20 U RPA-grade RNase A (Applied Biosystems) was added to each mixture to remove the single-strand RNAs. Four microliters of each residue was added to the 20 µl volume double-strand reverse transcription reaction with 5 pM random primers adding SuperScript III reverse transcriptase (both from Invitrogen-Life Technologies). The system was incubated at 60°C for 60 min and terminated at 75°C for 15 min, and the resultant double-strand cDNAs were amplified in a 25 µl PCR reaction system as templates. The primers used were: forward, 5′-ATT CTG CTA TTA CAA GTT-3′ and reverse, AAG TAC ATG TCT CCC AT. After a 35-cycle amplification, the product was checked by electrophoresis on a 1.5 % agarose gel with ethidium bromide staining.

In the in vivo study of the association between SIRT1-AS level with HCC risk, 52 patients (26 male and 26 female; average age 49.2±18.3 years) were invited to participate in the study. The patients were diagnosed as primary HCC and treated with tumor resection in our hospital from May, 2012 to June, 2014. The patients did not suffer any complications. Their tumor sizes ranged from 3 to 6 cm. There was no angiogenesis and invasion to peripheral tissues. Twenty-six male and 26 female healthy volunteers (average age 49.4±18.6 years) also participated in the investigation. Each individual in the control group was gender- and age-matched with the patients in a one-on-one manner.

The study protocol was approved by the Ethics Committee of The Affiliated Hospital of Inner Mongolia Medical College. The subjects provided written informed consent prior to their inclusion in the study.

The odds ratio was used as a precise estimation for the risk of 622U>C mutation in SIRT1-AS contributing to HCC in this study. The 2,000 primary HCC patients and 1,500 volunteers were from The Inner Mongolia Autonomous Region and Northeast China. Male and female subjects were equally divided, and had an age ranged of 22–58 years. Among the 2,000 HCC patients, the proportion of each stage (HCC staging according to the TNM Standard published by the Union for International Cancer Control in 2003) was approximately the same. The odd ratio was calculated as: Odd ratio T = (number of patients possessing allele T x number of volunteers possessing allele C)/(number of patients possessing allele C x number of volunteers possessing allele T), and the odd ratio C = (number of patients possessing allele C x number of volunteers possessing allele T)/(number of patients possessing allele T x number of volunteers possessing allele C).

SIRT1-AS was predicted using information obtained from the Natural Antisense Transcript Database (http://natsdb.cbi.pku.edu.cn). SIRT1 mRNA variant sequences were obtained from GenBank (http://www.ncbi.nlm.nih.gov) and UCSC Genome (http://genome.ucsc.edu/index.html). The Coding Potential Calculator (http://cpc.cbi.pku.edu.cn/) was used to calculate the coding potential of the sequences. The secondary structures of WT SIRT1-AS and the mutant were predicted by the mFold online server (http://mfold.rna.albany.edu/?q=mfold).

Data were obtained from at least three independent experiments. Values were presented as means ± SEM. Statistics were calculated using SPSS statistics v19.0 software. Multiple comparisons were assessed by one-way ANOVA followed by Dunnett's tests. The difference between groups was considered statistically significant when P<0.05.

We detected part of a transcript derived from the antisense strand of the SIRT1 3′UTR region by specific-strand RT-PCR in RNA samples that were extracted from the HCC-9903 cell line (Fig. 1B). Then, a specific probe was applied in the northern blotting to assay the transcript and a single band was obtained (Fig. 1A and C). The probed AS transcript was extracted and sent to BGI (Shenzhen, China) for high-precision sequencing. The sequencing results revealed the single-exon AS transcript with a size of 763 nt, designated as SIRT1-AS (Fig. 1D). The sequence was inserted into a coding potential calculator to calculate its coding potential. The outputs suggested that SIRT1-AS was probably non-protein-coding, because no available ORF structure was identified in the sequence (Fig. 1E). Therefore, we identified a single-exon AS transcript that overlapped with SIRT1 3′UTR and it was probably an AS lncRNA.

The pcDNA-SIRT1-AS or pcDNA-SIRT1 expression vectors were transfected into HCC-9903 cells to assess the role of SIRT1-AS in HCC cell proliferation. The specific-strand RT-qPCR and western blotting results showed that SIRT1-AS overexpression did not influence SIRT1 mRNA expression, but markedly elevated the SIRT1 protein level (Fig. 2A and B). As a result, the downstream anti-oncogene p53 was deacetylated by SIRT1 and SIRT1-AS overexpression (Fig. 2C). Edu staining and cell viability detection by the CellTiter-Blue H cell viability assay kit revealed that SIRT1 and SIRT1-AS overexpression promoted the proliferation of HCC-9903 cells (Fig. 2D and E). The results suggested that SIRT1-AS functioned positively in the regulation of SIRT1 expression at the post-transcriptional level.

In clinical diagnosis, our team identified a single-nucleotide mutation in a small number of HCC patient's SIRT1-AS by gene sequencing assistant diagnosis. The no. 622 base T (also U in RNA) was substituted by C in the minority of the patients (the site was marked in Fig. 1D). To verify the performance of the mutant SIRT1-AS, a pcDNA-SIRT1 mutant was additionally established. WT or mutant SIRT1-AS expression constructs were transfected into HCC-9903 cells. After an incubation for 48 h, western blot assay showed that the SIRT1 protein level was elevated in the WT transfection group, but did not significantly change in the group of mutant SIRT1-AS transfection (Fig. 3A). Additionally, as a proven SIRT1-targeting miRNA, miR-29c mimics reduced the protein level that was elevated by WT SIRT1-AS (Fig. 3A). The transfected cells were treated with 2 µg/ml actinomycin D to block the transcription activity in the cells. After treatment for 0, 0.5, 1, 2, 4, 6, 8, 10 and 12 h, the cells were collected and used to extract total RNA separately. The residual SIRT1 mRNA was quantified to evaluate the variation of the half-life between the groups. The results revealed that the half-life of SIRT1 mRNA was ~8 h under normal circumstances, which was not influenced by the mutant. As expected, the miR-29c mimic significantly shortened the half-life of SIRT1 mRNA (to <6 h, Fig. 3B). Notably, the WT SIRT1-AS greatly extended the period to ~11 h (Fig. 3B). In particular, WT SIRT1-AS was likely to antagonize the negative regulation of miR-29c.

As the miR-29c binding site was within the overlapping region of SIRT1-AS and SIRT1 mRNA, we hypothesized that WT SIRT1-AS was able to bind to SIRT1 mRNA, thus miR-29c was unable to target SIRT1 mRNA, whereas the mutant was not able to bind to SIRT1 mRNA. The sequences of WT SIRT1-AS and the mutant were, respectively, entered into the mFold web server to calculate the most probable secondary structures. The outputs showed that their secondary structures were entirely different (Fig. 4A), suggesting at their distinct ability to combine with SIRT1 mRNA. WT or the mutant synthetic SIRT1-AS was incubated with SIRT1 mRNA under simulated physiological conditions for 3 h, and then the mixture was digested with RPA-grade RNase A. RT-PCR was subsequently performed to detect whether an RNA duplex was formed. A 101-bp band, which included the mutant site and cDNA of the miR-29c binding site, was only detected in the residue of the WT SIRT1-AS/mRNA mixture. The RPA results showed that only WT SIRT1-AS formed a sense-antisense RNA duplex (Fig. 4B), which also supported our hypothesis.

The above results suggested that the mutation 622U>C 're-liberated' the miR-29c binding site via WT SIRT1-AS and downregulated SIRT1 expression at the post-transcriptional level. We randomly detected total SIRT1-AS levels in the liver tissues of 52 HCC patients and 52 non-hepatopathic volunteers. The strand-specific RT-qPCR results showed that SIRT1-AS was more highly expressed in HCC patients (Fig. 5). A survey was performed to assess the relationship between 622C and the risk of HCC. Among 2,000 HCC patients, the mutant ratio was 4.8% smaller than that among 1,500 non-hepatopathic volunteers (up to 16.3%). The statistical analysis showed that the OR value of 622C was only 0.253 (<1), which was lower than that of 622T (3.958>1) (Table I). These results suggested that the mutant C allele reduced the risk of HCC.