Uracils at nucleotide position 9–11 are required for the rapid turnover of miR-29 family

MiRNA mimickers

Synthetic duplex of miRNA mimickers in our assay are completely complementary. The sequences of miRNA mimickers are listed in the Supplementary Table S2. Mimickers were synthesized by ShangHai GenePharma Co., Ltd.

Cell culture and transfection

HeLa cells were maintained in Dulbecco’s Modified Eagle Media (DMEM, Invitrogen) supplemented with 10% fetal bovine serum. For half life measurement, 2.5 × 10⁵ HeLa cells were seeded into each well of 6-well plate. Twenty hours later, 40 nM miRNA mimickers were transfected by Lipofectamine 2000 (Invitrogen). Four hours after transfection, the medium was removed, cells were washed twice with PBS and were cultured with fresh medium. The time point at which the medium was replaced was set as 0 h and then cells were harvested at the time points indicated.

Real-time PCR

About 100 ng of total RNA was reverse transcribed by Improm II (Promega) with both a miRNA-specific stem-loop primer and random primer. The reaction was carried out at 42°C for 1 h. The reaction mixture was 100-fold diluted and real-time PCR with SYBR green was performed. Expression of miRNAs was normalized to U6 and ΔΔC_t analysis was used to calculate the relative expression of miRNAs. The products were analyzed by melting curve and manual inspection by electrophoresis. Real-time PCR for each miRNA was repeated for three times.

Dual luciferase reporter assay

The multiple cloning sites of plasmid pGL3 (Promega) in the promoter region were removed and filled in, and then a synthetic oligonucleotide containing KpnI and XhoI sites was inserted at XbaI site which is in 3′-UTR region of luciferase coding sequence. Cdc42 3′-UTR was amplified by RT–PCR and then inserted into the modified pGL3 by KpnI and XhoI digestion. The forward primer is GATGGTACCTCTCCAGAGCCCTTTCTGC and reverse primer GTCCTCGAGCAAAGAATTGAGACATGAGAAAGC. About 100 nM miRNA mimickers or negative control RNAs, 80 ng pRL plasmid and 1 μg pGL3-cdc42 3′-UTR plasmid were transfected into HeLa cells by Lipofectamine 2000 (Invitrogen), with triple replications for each transfection. Twenty-four hours after transfection, cells were harvested and reporter activity was examined with the dual luciferease reporter assay kit (Promega).

Northern blotting

RNA was isolated by TRIZOL (invitrogen) and 6 μg RNA for each sample was analyzed in 15% polyacrylamide gel with 8 M urea. RNAs were transferred onto positively charged Nylon membrane (Ambion) and immobilized by ultraviolet light for 2 min. Oligonucleotide probes were 5′-end-labeled with [γ-³²P]ATP using T4 polynucleotide kinase (Takara). All probe sequences are listed in Supplementary Table S3. The membranes were probed at 39°C with the hybridization solution PerfectHyb (Toyobo) overnight. The membranes were washed by 2× SSC and 0.5% SDS twice at 37°C. Radioactive signals were quantified using Quantity ONE (Bio-Rad). MiRNA abundance was normalized to U6 and the half life was then calculated according to the signals at five time points (0, 2, 4, 8 and 12 h). Lifespan of miR-29 family (including the wild-type and mutants) was examined by three independent experiments and other miRNAs two independent experiments.

Analysis of developmental expression profiles of miRNAs

Expression profiles of worm, fly and fish miRNAs were collected from published data. Data during all developmental stages in worm, the egg stage in fly and the first 2 days embryonic stage of fish were analyzed. MiRNAs with no expression signal were excluded. Either continuous or transient decrease of expression was classed as ‘down-regulated’ and others were classed as ‘not down-regulated’. Each of both classes in three species was summed up. The difference between miRNAs with and without U-rich sequence was examined by the Fisher’s exact test.

Pulse–chase assay to examine the miRNA lifespan

Both transcription and post-transcriptional processing influence miRNAs generation, and hence transcriptional shutoff may not completely terminate the generation of miRNAs. We therefore chose to employ the pulse–chase assay with synthetic RNA duplex as miRNA mimickers to examine the miRNA turnover in cell culture (24). Synthetic duplex of miRNA mimickers in our assay are completely complementary. First, we examined whether these miRNA mimickers could steadily accumulate in cells and therefore minimize the interference by endogenous miRNA species. Five microRNAs with different endogenous expression level were selected (25), and their mimickers were transfected into HeLa and analyzed by real-time PCR. A synthetic duplex RNA, which does not target any RNAs, served as the negative control (NC). Each miRNA increased substantially and therefore allow a long-term chase (Figure 1A). Previously it has been proven that this type of synthetic miRNA mimickers could incorporate into the RNA-induced silencing complex (RISC) (13). Here to further validate whether these mimickers could simulate miRNAs in a cellular context, we examined the silencing effect of synthetic miRNA mimickers on target mRNAs. MiR-29 family, including miR-29a, b and c, was reported to directly target cdc42 mRNA in HeLa cells (26). Our reporter assay showed that the synthetic miR-29 mixture (equal amount of miR-29a/b/c) was able to repress the luciferase reporter fused to the cdc42 3′-UTR by 40% (Figure 1B). A miR-29 mutant with the substitution of uracil at nucleotide position 10 with cytosine (miR-29b-C10) was as effective as the wild-type.

MiRNAs compete in loading into the RISC. Saturated RISC may leave miRNAs free and therefore make them more vulnerable to degradation. To exclude this effect in our assay, we titrated the transfection dose of miRNA mimickers at different concentrations. Northern blotting showed that cellular abundance had a good linear correlation with the transfection dose for miR-29b mimicker (Figure 1C, left). As high as 100 μM of miR-29b mimicker, the abundance of cellular miR-29b paralleled with the transfection dose. In contrary, negative control RNA was hardly detected in the same assay (Supplementary Figure S1). Real-time PCR verified this correlation (Figure 1C, right). In the subsequent assays, we therefore adopted the transfection dose of 40 nM miRNA mimickers in the measurement of miRNA lifespan. We noted that due to global increase of miRNA abundance in response to cell contact as reported previously (27), cells in pulse–chase assay should be appropriately administrated to <80% confluence in order to avoid the artificial interference.

Lifespan of miRNAs are long but diverse

Five miRNAs with various expression levels in HeLa were selected for lifespan measurement, including miR-16 with high abundance, let-7a, miR-17 and miR-29b with mediate expression, and miR-34a with no detection (25). Pulse–chase assay was used to measure lifespan of these representative miRNAs mentioned above. Generally the miRNA half life was 12 h or longer (Figure 2). Furthermore, these representative miRNAs possessed distinct characteristic in their dynamic change. MiR-16 did not have apparent decrease in their abundance until the time duration limit (12 h); MiR-34a and miR-17 possessed a moderate turnover rate, with a half life of about 12 h; Let-7a and miR-29b rapidly disappeared, with a half life of 11 and 7 h, respectively (Figure 2). The diversity of decay dynamics of miRNAs raises a question of sequence determinant of their stability.

Uracils at nucleotide position 9–11 of miR-29b are required for its rapid decay

MiR-29a is stable, while miR-29b is rapidly decayed in HeLa cells although their sequence are highly similar (13), thus we took miR-29 family as a paradigm to study the sequence dependence of miRNA degradation. MiR-29a and miR-29b differ at two regions. One is at their 3′-end and the other is at nucleotide position 10, with cytosine in miR-29a and uracil in miR-29b (Figure 3A). It has already been proven that the sequence at miR-29b 3′-end alone is not responsible for its rapid decay (13). We asked whether the variance at position 10 together with near nucleotides is involved in miRNA inherent stability. At the beginning, uracil was mutated into cytosine for miR-29b (miR-29b-C10), and this mutation did slightly enhance miRNA stability (Figure 3B). We further questioned whether uracils near position 10 had a synergic effect with the central uracil (U10) to function. Mutation of uracils in both position 9 and 10 (miR-29b-C9C10) into cytosine prolonged the half life of miR-29b (Figure 3B). This effect was observed for mutation in both position 10 and 11 (miR-29b-C10C11) to a more significant degree. All three uracils from position 9 to 11 into cytosine (miR-29b-C9C10C11) abolished the rapid decay of miR-29b (Figure 3B). It is questionable whether the change of observed decay rate is simply attributed to the alteration of miRNA GC content. Mutation into adenosine (miR-29b-A9A10A11) was examined and the result demonstrated a significant decrease in the decay rate (Figure 3B). Furthermore, to test whether position at nucleotides 9–11 is critical for the effect of uracil stretch, the nucleotides at position 9–11 were exchanged with that at position 12–14 (miR-29b-U12U13U14, see Figure 3A). Northern blotting showed that the change in position stabilized the wild-type miR-29b (Figure 3B). To test whether our observation would hold generally, the lifespan of miR-29b and miR-29b-C9C10C11 was chased in HEK293T cells. Indeed, mutation of centric uracils impeded the rapid degradation of miR-29b (Supplementary Figure S2). Our observation therefore demonstrated that uracil at the middle region of miR-29b (from 9 to 11 nt) was required for its rapid turnover.

Due to nucleotides in positions 9 and 10 of miRNAs influence their sorting in Drosophila (28,29), we examined the star strand of miR-29b duplex. Northern blotting could hardly detect the existence of the star strand (Figure 3C, upper). To further explore the mechanism of the uracil-dependent decay, single strand but not duplex of miR-29b and miR-29b-C9C10C11 were chased. The result showed that the single strand of both miR-29b and miR-29b-C9C10C11 were rapidly decayed (Figure 3C, lower). This observation suggests that the uracil element function during the loading of miRNA into RISC or the unwinding of miR/miR* duplex.

A subset of miRNAs with uracils at nucleotide positions 9–11 possesses rapid turnover

Due to the above observation that uracils in positions 9–11 (U-rich sequence) was involved in miR-29b decay, the next question we asked was whether it is a general rule for the U-rich sequence to play a role as a cis-acting element in regulation of miRNA stability. MiRNAs are thought to contribute to the complexity of organism, thus majority of miRNAs emerge with cellular differentiation and development (30). We deduced that if U-rich sequence is truly involved in the acceleration of miRNA turnover, the expression of miRNAs with U-rich sequence will tend to be down-regulated during development. To this end, we listed all miRNAs with U-rich sequence and examined their expression profile during development in the published data which was summarized in Supplementary Table S1 (31–33). Factors other than internal stability of miRNAs can regulate their abundance, such as transcriptional regulation. To decrease this complication, the interval of developmental stage in expression data was restricted to no more than 24 h. There are 17 miRNAs with the U-rich sequence whose expression has been characterized (Table 1). As expected, 10 out of the 17 miRNAs exhibit the rapid decrease during development, while the majority of other miRNAs accumulates (Table 2), with a different expression dynamics (P = 0.004).

The decrease in expression level of miRNA does not necessarily attribute to miRNA degradation, thus we examined the stability of miRNAs in the above list. MiR-29c is another member of miR-29 family, and our assay showed it was much more unstable than miR-29a, with a half life of 10.6 h (Figure 4). The fruit fly miRNA bantam was sharply down-regulated during the development (34). The abundance of bantam decreased with a relative rapid rate in HeLa cell (Figure 4B).

Further, we test whether the U-rich sequence functions dependent or independent of sequence context. First, bantam 3′-end sequence was replaced with that of miR-29b (Dbantam-3 M), and northern blot demonstrated that the sequence alternation did not visibly accelerate the decay of the hybrid (Figure 4B). To place the U-rich sequence in a random sequence background, we constructed two mutants of miR-126 by removing the nucleotides in positions 9–11 with UUU and CAG, respectively. A U-rich element accelerates the decay of miR-126 compared to element with no uracil (Figure 4B). However, the lifespan of U-rich miRNA mutant is still beyond the duration limit of our assay (>12 h). This indicates that the U-rich sequence confers the rapid turnover rate in a context dependent manner.