Outside the limit: questioning the distance restrictions for cooperative miRNA binding sites

Selection of cooperative miRNAs and targets

Cooperativity is suggested to be a broad, but so far little explored phenomenon, which may apply to many different miRNAs [12, 23]. As for the selection of miRNAs, we chose miR-21-5p and miR-155-5p that have been reported as potent regulators for a large number of targets with special relevance in various contexts of health and disease [24, 25]. In our recent study on the early human T cell activation process, we highlighted miR-21-5p and miR-155-5p as highly expressed miRNAs with a shared target network and a likely cooperative function [15]. Furthermore, a cooperatively regulated targeting in mice argues for the evolutionary conservation of a cooperative interaction of these two miRNAs [26].

As for miRNA target sequences, we set out to identify sequences that are targets to both miR-21-5p and miR-155-5p. We used the miRWalk 2.0 tool [14] for an in silico prediction of endogenous 3′UTRs binding sites for both, miR-21-5p and miR-155-5p. We considered only putative targets that were predicted by at least four of the 13 algorithms integrated in miRWalk 2.0 and identified 1295 shared target genes for miR-21-5p and miR-155-5p. We next related the expressional data of the two miRNAs and to their potential targets utilizing information from our former study (as mentioned above) that integrated miRnome and transcriptome data in context with early T cell activation process [15]. Since both miRNAs showed strong expressional increase upon T cell activation (Fig. 1A and B), we selected only genes, the mRNAs of which showed an according expressional decrease (Fig. 1C and D). Out of 355 identified potential targets, we chose a total of 15 genes with log2 fold changes ranging from − 0.66 to − 2.50 after T cell activation as compared to the value prior to activation (Fig. 1E). Furthermore, the 15 targets were selected to represent various 3′UTR binding site constellations for miR-21-5p and miR-155-5p, ranging from partially (1 nt of miR-155-5p/ 3 nt of miR-21-5p) overlapping binding sites to neighboring binding sites that were up to 1,300 nt apart from each other (Fig. 1F). As summarized in Additional file 2: Table S1 the sequences were also selected to contain multiple binding sites for either of the two miRNAs.

Fig. 1figure 1

Parameters of cooperative target genes that were selected for the analysis of cooperative miRNA effects by dual luciferase reporter assays. AE RNA expressional data from a time-course study on the early T cell activation process [15] were utilized for the selection of 15 putative joint targets of miR-21-5p and miR-155-5p to be tested by HiTmIR dual luciferase reporter assays. The hsa-miR-21-5p and hsa-miR-155-5p showed a strong expressional increase upon T cell activation (Donor 1 (A), Donor 2 (B)) and the mRNAs of the selected genes showed expressional decrease during the time-course in both analyzed donors (Donor 1 (C), Donor 2 (D)). Maximum mRNA expression changes compared to the 0 h time-point (log2 fold changes (FCs)) are depicted for the 15 selected genes (E). F The length of distances between in silico predicted binding sites for miR-21-5p and miR-155-5p are given for the 3’UTR sequences of the selected genes. The previously defined distance range for cooperative miRNA binding is highlighted in turquoise color

The effect of single and simultaneous miRNA expression

To screen for the effect of differing binding distances, we cloned corresponding 3′UTR sequences into pMIR-RNL-TK dual luciferase reporter plasmid. Resulting plasmids were employed in the semi-automated HiTmIR assay [13] to examine effects upon overexpression of only miR-21-5p or only miR-155-5p as well as co-expression of both miRNAs (Additional file 6: Table S5A), notably in the same experiment, thus allowing for better comparability between different conditions. The same amounts of transfected miRNA expression plasmids were used for comparisons between single and co-expression effects. Four independent experiments were performed in technical duplicates each. MiRNA induced effects were evaluated based on averaged values relative to an empty reporter plasmid control. All p-values were adjusted by the Benjamini–Hochberg procedure to control for false discovery rate (FDR) [20]. We categorized target sequences based on the distances between neighboring miR-21-5p and miR-155-5p binding sites within the 3′UTRs of the target genes. In detail, we differentiated between genes with miRNA binding sites that were either < 40 nt, 40–240 nt, > 240–600 nt, or > 600 nt apart from each other.

A first category of genes (CCDC96, F13A1 and LEMD3) (Fig. 2A) represented predicted miR-21-5p and miR-155-5p binding sites less than 40 nt apart from each other.

Fig. 2figure 2

Analysis of miRNA cooperativity by HiTmIR assays. The 3’UTR sequences of 15 putative cooperative targets of miR-21-5p and miR-155-5p were cloned into pMIR-RNL-TK reporter plasmid and analyzed by semi-automated HiTmIR dual luciferase reporter assays. MiRNA induced effects on firefly luciferase activity were determined upon overexpression of only miR-21-5p or only miR-155-5p or upon co-expression of both miRNAs. For representation, the analyzed genes were categorized into four groups regarding the distances between neighboring miR-21-5p and miR-155-5p binding sites (< 40 nt (A), 40-240 nt (B), > 240–600 nt (C), > 600 nt (D)). Relative luciferase units (RLU) were related to the activity of an empty reporter control (pMIR-RNL-TK; under the respective miRNA condition) and are shown as the mean result (± standard error of the mean (± SEM)) of four independent experiments that were conducted in technical duplicates. Statistical evaluation was performed in comparison to the empty reporter control (gray asterisks) or for the co-expressional conditions in comparison to the effects upon single miRNA overexpression of miR-21-5p and miR-155-5p (black asterisks), respectively. Significant p-values after FDR adjustment are denoted (*p ≤ 0.05, **p < 0.01, *** p < 0.001)

The 3′UTR of CCDC96 showed partially overlapping binding sites for miR-21-5p and miR-155-5p. Neither miR-21-5p, nor miR-155-5p or co-expression of both miRNAs produced a significant regulatory effect. That result is consistent with the idea that very close proximity of miRNA binding sites causes mutual steric hindrance of RISCs [11].

The F13A1 3′UTR was characterized by an 8 nt distance between the two miRNA binding sites. Although that distance is assumed to be sufficient for cooperative effects, there were no regulatory effects from single expression or co-expression of both miRNAs.

In contrast, the LEMD3 3′UTR, with the two miRNA binding sites laying in 9 nt distance from each other, showed no effect for miR-155-5p, but an effect for miR-21-5p and an even more pronounced effect for co-expression of both miRNAs. Specifically, relative luciferase reporter activity, measured as relative light units (RLU), was not significantly affected by miR-155-5p expression compared to empty reporter control (RLU = 98.98 ± 6.35; p = 0.998). Overexpression of miR-21-5p, however, significantly reduced the RLU to 83.64% ± 4.64 (p = 2.27 × 10–2) and co-expression of both miRNAs further reduced the RLU to 53.76% ± 4.66 (p = 6.34 × 10–6). The effect of the co-expression of both miRNAs was also significantly different from the effects measured separately for each of the miRNAs (miR-21-5p: p = 4.61 × 10–3; miR-155-5p: p = 1.02 × 10–3).

A second category of genes, including PELI1, EHD1, DRAM2, and GALNT12 (Fig. 2B), displayed distances between neighboring miR-21-5p and miR-155-5p binding sites ranging from 40 nt to 240 nt.

The PELI1 3′UTR showed no effects for neither miR-21-5p nor miR-155-5p (miR-21-5p: RLU = 93.62 ± 5.00; p = 0.439; miR-155-5p: RLU = 101.46 ± 6.84; p = 0.894), but significant decrease of RLU to 70.51% ± 5.14 (p = 7.59 × 10–4) was detectable, when co-expressing both miRNAs. The effect of the co-expression of both miRNAs was also significantly different from the effects measured for each of the miRNAs (miR-21-5p: p = 2.45 × 10–2; miRNA-155: p = 1.86 × 10–2). In contrast to our results, separate regulatory effects have formerly been described of each the miR-21-5p and the miR-155-5p on PELI1 in mice [27, 28]. These contrary findings may be explained by experimental differences regarding the size of the analyzed 3′UTR sequences, by the different species and by a different miRNA dose [27, 29].

The EHD1 3′UTR showed no effect for miR-21-5p, but a significant effect for miR-155-5p, and an even more pronounced effect when co-expressing both miRNAs. However, this effect that could indicate some cooperative interaction, was not significantly verified when compared to the single miRNA´s effects on EDH1 3′UTR. Likewise, DRAM2 3′UTR showed significant effect on RLU upon co-expression of both miRNAs, but no significant difference to the effect of the single miRNAs. As for the GALNT12 3′UTR, neither expression of single nor co-expression of both miRNAs resulted in a significant regulatory effect.

A third category of genes, including RNF103, MYBL1, RECK, and LHFPL2 (Fig. 2C), displayed distances between neighboring miR-21-5p and miR-155-5p binding sites ranging from 240 to 600 nt.

As for the RNF103 gene, miR-21-5p on its own did not have an effect on the RLU, but co-expression with miR-155-5p contributed to increase the effects of the miR-155-5p. As for the EDH1 3′UTR and the DRAM2 3′UTR potential cooperative effects could not be statistically verified within the scope of our measurements.

Similar to above described LEMD3, RECK showed no effect for miR-155-5p (RLU = 97.40 ± 4.68; p = 0.692), but an effect for miR-21-5p (RLU = 60.94% ± 2.00; p = 7.46 × 10–10), and an even more pronounced effect, when co-expressing both miRNAs (RLU = 43.33% ± 4.68; p = 6.64 × 10–8). The effect of the co-expression was also significantly different from the effects measured for each of the miRNAs (miR-21-5p: p = 3.81 × 10–2; miR-155-5p: p = 2.14 × 10–5). The other genes in this category, specifically MYBL1 and LHFPL2, did not show a cooperative effect upon combined overexpression of miR-21-5p and miR-155-5p.

The fourth category of genes, including OGT, DDX17, FBXL17, and IKZF5, displayed distances greater than 600 nt between miR-21-5p and miR-155-5p binding sites (Fig. 2D).

Only the DDX17 3′UTR showed a relatively mild regulatory effect for miR-155-5p and a likewise mild cooperative non-significant effect for the co-expression of both miRNAs.

The other genes in this category, specifically OGT, FBXL17, and IKZF5, showed no regulatory effect from single or co-expression of miRNAs.

Several of the above described genes were newly identified as targets for individual miR-155-5p and miR-21-5p. EHD1 and RNF103 were identified as new targets of miR-155-5p and MYBL1 as a target of both miR-155-5p and miR-21-5p, respectively. Likewise, previously reported effects were consistent with our data, including the formerly described regulation of RECK by miR-21-5p [30] and the regulations of LHFPL and DDX17 by miR-155-5p [15, 31]. The increases of RLU that were detected for some genes, including LHFPL upon miR-21-5p overexpression and FBXL17 upon individual overexpression of both miRNAs, may be attributable to some secondary effects, potentially from the regulation of endogenous targets affecting the reporter construct activity.

Validation of cooperative effects by reporter assays and potential impact of RNA secondary structures

We chose two of the identified cooperative targets for exemplary validation of the identified binding effects. We also evaluated the possible impact of secondary structures on these binding effects. Thereto, we predicted secondary structures for corresponding 3′UTR sequences using RNAfold as implemented in Geneious v2022.1 [22]. In detail, we analyzed LEMD3 and RECK 3′UTRs that both displayed cooperative effects for miR-21-5p and miR-155-5p within the initial luciferase assays. While LEMD3 sequence included neighboring binding sites of miR155-5p and miR-21-5p (5′ → 3′) at a distance of 9 nt, corresponding to the former defined limits, RECK 3′UTR included neighboring binding sites at a distance of 414 nt. Besides, RECK 3′UTR included an additional miR-155-5p binding site (BS1) at a distance of 584 nt in 5′ direction to miR-21-5p binding site.

Predicting RNA secondary structures for LEMD3 indicates that both miRNA's binding sites locate within the same hairpin structure. Hence, the sequence proximity and the orientation could allow for an interaction between two binding RISCs (Fig. 3A).

Fig. 3figure 3

In silico prediction of 3’UTR secondary structures and exemplary validation miRNA cooperative effects. A, C: RNA secondary structures were predicted for LEMD3 (A) and RECK (C) 3′UTR sequences that showed cooperative binding of miR-21-5p and miR-155-5p as determined by the HiTmIR dual luciferase assays. Corresponding 3′UTR sequences (as cloned for the underlying assays) are shown in a linearized (upper panel; 5′→3′) and in a folded representation (lower panel), respectively. The included miRNA binding sites are indicated. B, D For the validation of miRNA cooperative effects, miR-21-5p and miR-155-5p binding sites were mutated within pMIR-LEMD3 (B) and pMIR-RECK (D) 3′UTRs. The reporter constructs were tested by HiTmIR dual luciferase reporter assays. Comparative measurements were conducted with both, the wild type and mutated reporter constructs. MiRNA induced effects were determined upon overexpression of only miR-21-5p or only miR-155-5p or upon co-expression of both miRNAs. Relative luciferase units (RLU) were related to the activity of an empty reporter control (pMIR-RNL-TK; under the respective miRNA condition) and are shown as the mean result (± standard error of the mean (± SEM)) of three independent experiments that were conducted in technical duplicates. Statistical evaluation of the cooperative effects was performed in comparison to the effects upon single miRNA overexpression of miR-21-5p or miR-155-5p (black asterisks). Significant p-values after FDR adjustment are denoted (*p ≤ 0.05, **p < 0.01, ***p < 0.001)

To validate the previously detected effects on the wildtype sequence, miRNA binding sites within the LEMD3 3′UTR reporter constructs were mutated. As compared to the single miRNA´s effects, the reporter assays demonstrated reversal of the cooperative miRNA effect upon mutation of either the miR-21-5p or the miR-155-5p binding site (Fig. 3B; Additional file 6: Table S5B). These findings highlight the relevance of both binding sites for the induction of the miRNA cooperative effect. Notably, in the case of miR-21-5p mutation, even an opposite development towards a slight increase of RLU could be detected as compared to the single miRNA´s effects.

As for the structure prediction on RECK 3′UTR sequence (Fig. 3C), the binding sites of miR-21-5p and miR-155-5p at 414 nt distance (BS2) may be in a spatial proximity due to their localization on two small opposite hairpin structures that are connected by a small common circular base. Under the assumption of certain 3-dimensional motility of these arms, a cooperative interaction is conceivable. According to our prediction of RNA secondary structures, the additional upstream miR-155-5p binding site (BS1), which is located at an even greater nucleotide distance to miR-21-5p, lies at the upper end of a larger secondary structure. A potential 3-dimensional movement of this protruding structure may permit a binding site approximation leading to a cooperative interaction.

Reporter assays verified the cooperative miRNA binding on RECK wild type 3′UTR, as defined by a significant decrease of RLU in comparison to the two miRNA´s individual effects, and showed reversal of the miRNA cooperative effect upon mutation of the included miR-21-5p and miR-155-5p binding sites (Fig. 3D; Additional file 6: Table S5B). According to our reporter assay results, no clear statement can be made about the relevance of the two included miR-155-5p binding sites. A reversion of cooperativity was achieved upon all three conditions, mutation of miR-155-5p BS1, miR-155-5p BS2 and the mutation of both miR-155-5p binding sites (BS1&BS2). In principle, a kind of ping-pong mechanism would be imaginable, in which both of the miR-155-5p binding sites could provide alternating interaction with the miR-21-5p binding site.

Cooperative miRNA effects on protein targets with distant 3′UTR binding sites

We examined cooperative miRNA effects on endogenous protein levels using the abovementioned pSG5-based miRNA expression system in 293T human embryonic kidney cells as cellular model for mass spectrometry-based protein analyses. As for the luciferase assays, we analyzed the cellular protein extracts following co-expression of both miRNAs or following the overexpression of either miR-21-5p or miR-155-5p. Three independent experiments were performed and a transfection with the empty effector plasmid was carried out as control. As detailed in the methods section, cooperative effects were determined by the comparison of effect sizes.

Three genes (DDX17, EHD1 and LEMD3), for which cooperative effects or trends have been observed by the luciferase assays, were identified in the proteome analyses i.e., were among the detected proteins in HEK293T cells (Additional file 7: Table S6). As defined by the comparison of Cohen’s d (see methods), DDX17 showed a cooperative effect of miR-21-5p and miR-155-5p by the proteomics analysis (Fig. 4A and B). The proteome data also indicated cooperative effects for 21 additional proteins (Fig. 4A), including CNPY2 and UBE3A (Fig. 4C and D) that contained seed binding sites for both miRNAs within their 3′UTR transcript sequences. In detail, the 3′UTR sequence of CNPY2 contained three miR-21-5p binding-sites and one miR-155-5p site at a minimal distance of at least 158 nt. The 3′UTR sequence of UBE3A contained only one binding site for each miRNA at a distance of 208 nt. In both cases, the distances of the binding sites exceeded the previously defined limits of 39 nt.

Fig. 4figure 4

Analysis of miRNA cooperativity by proteome data. Whole cell protein extracts from 293T human embryonic kidney cells were analyzed by high-resolution mass spectrometry upon (i) control plasmid transfection (con), (ii) overexpression of only miR-21-5p (miR21), (iii) only miR-155-5p (miR155) or (iv) upon co-expression of both miRNAs (miR21/155). Three independent transfection experiments were performed. As defined by the comparison of effect sizes by Cohen’s d, 22 gene products indicated cooperative miRNA effects (A; clustering of z-scored expression data). For specific representation, relative protein expression data for DDX17 (B), CYPN2 (C) and UBE3A (D) were normalized to controls and are shown as averaged results with SEM

留言 (0)

沒有登入
gif