MicroRNAs hsa-miR-495-3p and hsa-miR-486-5p suppress basal and rifampicin-induced expression of human sulfotransferase 2A1 (SULT2A1) by facilitating mRNA degradation

Abstract

Drug metabolizing enzymes mediate biotransformation of drugs and play an essential role in drug efficacy and toxicity. Human sulfotransferases are a superfamily of Phase II detoxification enzymes that metabolize a wide spectrum of endogenous compounds and xenobiotics. SULT2A1 is one of the most abundant hepatic sulfotransferases and it catalyzes the sulfate conjugation of many endogenous substrates, such as bile acids and steroids. In the current study, we utilized a systematic approach by combining a series of computational analyses and in vitro methods to identify miRNAs that repress SULT2A1 expression post-transcriptionally. Our in silico analyses predicted miRNA response elements for hsa-miR-495-3p and hsa-miR-486-5p within the 3'-UTR of SULT2A1 mRNA and the levels of these miRNAs were inversely correlated with that of SULT2A1 mRNA in human liver. Using fluorescence-based RNA electrophoretic mobility shift assays, we found that hsa-miR-495-3p and hsa-miR-486-5p interacted directly with the SULT2A1 3'-UTR. The activity of a luciferase reporter gene construct containing sequences from the SULT2A1 3-UTR was suppressed by hsa-miR-486-5p and hsa-miR-495-3p. Furthermore, gain- and loss-of-function assays demonstrated that hsa-miR-486-5p and hsa-miR-495-3p negatively modulate basal and rifampicin-induced expression of SULT2A1 in HepG2 cells by decreasing mRNA stability.

Keywords: Drug metabolism; Epigenetics; MicroRNA; SULT2A1; Sulfotransferase (SULT); miR-486-5p; miR-495-3p.

Fig. 1.

In silico identification of human miRNA candidates that target the SULT2A1 3′-UTR. A. Schematics of mRNA sequences of SULT2 family members. SULT2A1 mRNA (NM_003167) contains a 988-bp long 3′-UTR, while the transcripts of the SULT2B1 gene, SULT2B1a (NM_004605) and SULT2B1b (NM_177973), have short 3′-UTRs that are 49 bp and 17 bp long, respectively. Transcript schematics were drawn proportional to lengths of the mRNA sequences. B. Venn diagram comparing the miRanda and TargetScan predictions of human miRNAs that target the SULT2A1 3′-UTR. miRanda identified12 miRNAs that were conserved with a good miRSVR score (< −0.1), while 309 miRNAs of various conservation levels were predicted by TargetScan. 10 miRNAs were among both prediction results. C. Seed match positions of the 10 miRNAs predicted by both miRanda and TargetScan within the SULT2A1 3′-UTR. miR-383-5p, miR-411-5p, and miR-129-5p were predicted to have two MREs in the SULT2A1 3′-UTR, whereas miR-324-5p, miR-494-3p, miR-196a, miR-196b, miR-495-3p, miR-590-3p, and miR-486-5p were shown to have one MRE in the SULT2A1 3′-UTR. Seed match sequences for all 10 miRNAs but miR-590-3p were in shade; seed match sequence of miR-590-3p was underlined to be distinguished from that of miR-486-5p.

Fig. 2.

Correlation analysis of the expression levels of SULT2A1 mRNA and the 10 miRNA candidates in non-tumor human liver samples. Scatter plots showing the correlation between the level of each of the 10 miRNA candidates and that of SUL2A1 mRNA were shown in an order with the r value from the lowest to the highest, and concomitantly with the p value from the lowest to the highest. All 10 miRNAs have a negative r value in correlation with SULT2A1 mRNA on the expression level; however, only hsa-miR-495-3p (r = −0.4267, p = 0.0022) and hsa-miR-486-5p (r = −0.3480, p = 0.0143) were in significantly inverse correlation with SULT2A1 mRNA on the expression level, and they have the highest inverse correlation reflected by the lowest r values.

Fig. 3.

hsa-miR-495-3p and hsa-miR-486-5p directly bind to the SULT2A1 3′-UTR for gene silencing. A. IDT OligoAnalyzer and RNAhybrid calculation of minimum free energy for RNA hybridization. IDT OligoAnalyzer and RNAhybrid showed variation in the minimum free energy for miRNA and MRE binding. Solid lines indicate the base pairs considered for energy calculation, while semi-colons indicate the base-pairing that was excluded for calculation. B. FREMSA. Cy5.5-tagged miRNA probes (red) and IRdye800-tagged SULT2A1 mRNA sequences (green) containing the predicted MREs were incubated alone or together, in the absence or presence of cold miRNA probes or cold-NC. RNA binding reaction mixtures were then subjected to a native PAGE gel. The incubation of hsa-miR-495-3p and miR-495-MRE-1 resulted in a yellow band (white arrow, lane 4), indicating the merge of the red hsa-miR-495-3p and green miR-495-MRE-1 due to the duplex formation. The yellow band was reduced in intensity in the presence of the 50-fold cold hsa-miR-495-3p probe (lane 6) but not cold-NC (lane 5). Hollow arrowheads pointed to the two different conformations of the hsa-miR-495-3p probe (lane 1). No yellow bands were observed when hsa-miR-495-3p was incubated with miR-495-MRE-2 (lanes 7–9), suggesting hsa-miR-495-3p specifically binds to miR-495-MRE-1 on SULT2A1 mRNA. hsa-miR-486-5p directly interacted with its predicted MRE on the SULT2A1 3′-UTR, which led to a yellow band (lane 3, white arrow) that was reduced in intensity by the addition of cold hsa-miR-486-5p probe (lane 5) instead of the cold-NC (lane 4). C. hsa-miR-495-3p and hsa-miR-486-5p overexpression significantly decreased luciferase activity in HepG2 cells via SULT2A1 3′-UTR. HepG2 cells were transfected with pmiRGLO-SULT2A1-3′-UTR or the empty vector along with mimics for hsa-miR-495-3p, hsa-miR-486-5p, and mimic-NC, respectively. Three independent experiments were conducted, each in triplicate. Data are shown as mean with standard deviation. *p < 0.05 compared to mimic-NC. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 4.

hsa-miR-495-3p and hsa-miR-486-5p overexpression down-regulated SULT2A1 mRNA and protein expression in a concentration-dependent manner. HepG2 cells were transfected with mimics for hsa-miR-495-3p, hsa-miR-486-5p, and mimic-NC, respectively. The levels of hsa-miR-486-5p (A), hsa-miR-495-3p (C), SULT2A1 mRNA (B and D), SULT2B1a mRNA (E and G), and SULT2B1b mRNA (F and H) were quantified using RT-qPCR. SULT2A1 protein levels were detected and measured via western blot (I-L). Three independent experiments were conducted. Data are shown as mean with standard deviation. *p < 0.05 compared with mimic-NC or 5 nM hsa-miR-495-3p and hsa-miR-486-5p mimic treatment.

Fig. 5.

hsa-miR-495-3p and hsa-miR-486-5p inhibition upregulated SULT2A1 mRNA and protein expression in a concentration-dependent manner. HepG2 cells were transfected with inhibitors for hsa-miR-495-3p, hsa-miR-486-5p, and inhibitor-NC, respectively. The levels of hsa-miR-486-5p (A), hsa-miR-495-3p (C), SULT2A1 mRNA (B and D), SULT2B1a mRNA (E and G), and SULT2B1b mRNA (F and H) were quantified using RT-qPCR. SULT2A1 protein levels were detected and measured via western blot (I-L). Three independent experiments were conducted. Data are shown as mean with standard deviation. *p < 0.05 compared with NC or 10 nM of inhibitors for hsa-miR-495-3p and hsa-miR-486-5p treatment.

Fig. 6.

hsa-miR-495-3p and hsa-miR-486-5p decreased the stability of SULT2A1 mRNA and inhibited rifampicin-induced SULT2A1 expression. A. The stability of SULT2A1 mRNA was reduced upon overexpression of hsa-miR-495-3p and hsa-miR-486-5p in HepG2 cells. HepG2 cells were transfected with mimics for hsa-miR-495-3p, hsa-miR-486-5p, or mimic-NC, 24 h before the onset of actinomycin D treatment (5 μg/mL) for 0, 8, 12, and 24 h. SULT2A1 mRNA was less stable in cells transfected with mimics for hsa-miR-486-5p or hsa-miR-495-3p. *p < 0.05 compared to NC. B. Overexpression of hsa-miR-495-3p and hsa-miR-486-5p inhibited induction of SULT2A1 by rifampicin. HepG2 cells were transfected with mimics for hsa-miR-495-3p, hsa-miR-486-5p, or mimic-NC, 24 h before the start of rifampicin treatment at 50 μM. 24 h of rifampicin treatment induced an increase of SULT2A1 mRNA, which could be significantly inhibited by overexpression of hsa-miR-495-3p and hsa-miR-486-5p. Three independent experiments were conducted. Data are shown as mean with standard deviation. *p < 0.05 compared to NC treated with DMSO or rifampicin.

Fig. 7.

Schematics of the integrated strategy for miRNA identification. Various tools and resources were used to compare different computational analyses to ensure a high confidence during miRNA prediction. Wet-lab experimental validation was carried out based on the in silico analyses results, which showed that hsa-miR-495-3p and hsa-miR-486-5p target SULT2A1 for gene repression via mRNA degradation.