1. nc886 Had Been Misidentified
nc886 is a 101 nucleotide (nt) long ncRNA whose aliases include vault RNA2-1, pre-miR-886, etc. As one of its aliases indicates, nc886 was mis-annotated as an miRNA precursor, and its mature miRNAs, miR-886-5p and -3p, had been registered in the miRNA database. In fact, my initial interest in nc886 came from miRNA array data, which yielded miR-886-5p and -3p to be most differentially expressed between malignant and non-malignant cells
[1]. However, experimental evidence from my laboratory led to a conclusion that nc886 is not an miRNA
[1]. The issue of whether they are functional miRNAs is very important, because most papers on miR-886-5p and -3p employed miRNA study tools (to be elaborated later). If they are not miRNAs, those data should be reexamined.
miRNAs are defined to be ~21 nt-sized small RNAs that are produced from a hairpin precursor (pre-miRNA) by Dicer
[2]. A long primary transcript is processed by Drosha to generate a pre-miRNA, which is cleaved by Dicer and is loaded to Argonaute (Ago) family proteins to suppress the expression of target mRNAs (reviewed in
[3]). When recognizing target mRNAs, the critical sequence element of an miRNA is a so-called “seed sequence”, which is 6–8 nt at the 5′-side of an miRNA (position 1~2 to 7~8) (reviewed in
[4]). Small RNA sequencing (small RNA-seq) by NGS captured RNA fragments corresponding to miR-886-5p and -3p sequences, providing evidence that they exist
[5][6][7][8]. However, the existence of RNA fragments does not guarantee that they are real miRNAs, because there are various types of small RNAs and miRNA is one of them (reviewed in
[9]). Currently available data, such as Northern blot and miRNA activity assays, unequivocally indicate that nc886 does not produce miR-886-5p or -3p at a level that could be significant as a functional miRNA
[1][10][11][12][13][14][15]. Northern blot in all these papers failed to detect a ~21-nt band. Even when detected, the bands were extremely minute in quantity and appeared to be smeared
[16][17], indicating that they are degradation products of nc886. In fact, the steady-state level of nc886 is very abundant, despite its short half-life
[1][18], leading to an expectation that a significant amount of degradation products are continuously produced.
In conclusion, RNA fragments corresponding to miR-886-5p or -3p exist, but they are most likely to be degradation products whose level is negligible compared to the intact, 101 nt long nc886. Nonetheless, there are a number of papers about miR-886-5p or -3p (40 articles retrieved in the PubMed database when searched by “miR-886-3p OR miR-886-5p”). I will discuss here whether data in the literature could draw a conclusion that miR-886-5p or -3p are functional miRNAs.
2. What Is Measured versus What Really Exists; Are They the Same?
The majority of the papers on miR-886-5p or -3p came from an attempt to find an miRNA that played a role in a biological situation of their interest. miRNA study typically begins with array experiments to screen an miRNA whose expression levels are altered. In hybridization-based array platforms, probes for miR-886-5p or -3p will also detect nc886. PCR-based arrays, which usually employ the TaqMan PCR technique
[19], cannot distinguish nc886 from miR-886-3p in principle, since they share an almost identical 3′-end. In agreement with this notion, a significant fraction of studies employing TaqMan array platforms claimed to preferentially detect miR-886-3p rather than -5p
[20][21][22][23][24]. What they detected was probably nc886, as inferred from the aforementioned Northern blot results
[1][10][11][12][13][14][15][16][17]. Hence, a positive signal in array platforms cannot prove the existence of miR-886-5p or -3p. This would be easily understandable if compared with conventional array experiments to measure gene expression. Array probes are an oligonucleotides complementary to the target mRNA. From the positive hybridization signal, everyone notices the mRNA expression level, but nobody would insist that an RNA fragment corresponding to the probe sequence exists naturally. In many cases of hybridization-based arrays, miR-886-5p and -3p exhibited a similar expression pattern
[25][26][27][28]. These data strongly indicate that nc886 was what they detected, similar to the case of conventional arrays, where an mRNA is detected by multiple probes.
3. The Mistaken Identity Misleads Functional Approaches
When researchers have selected an miRNA of their interest, the next step is typically to examine cellular phenotypes and target mRNAs by performing overexpression or knockdown (KD) experiments. The most prevalent method is to transfect an miRNA mimic or inhibitor. Caution is required, especially when using an miRNA mimic. It should be noted that miRNA mimics are a chemically synthesized RNA duplex, whereas natural miRNAs are single-stranded.
To understand this discrepancy, a story about miRNA and RNA interference (RNAi) needs to be told (reviewed in
[29]). In the RNAi pathway, long double-stranded RNA (dsRNA) is cleaved by Dicer to produce small interfering RNAs (siRNAs) that are a ~21 nt duplex with a 2 nt overhang at the 3′-ends. A duplex with such an end structure is exactly the Dicer/Drosha processing product in the miRNA pathway
[3]. The miRNA and RNAi pathways merge at the Dicer step to produce miRNA and siRNA duplexes with nearly identical structure, and then both of them are loaded to Ago proteins to suppress target mRNAs. When loaded to Ago proteins, only one strand (called the “guide strand”) survives to recognize target mRNAs, but the other strand (the “passenger strand) is degraded. Once loaded on Ago proteins, it is not distinguishable whether the RNA was originally from an miRNA or an siRNA. This is the reason for an siRNA’s off-target effect, as first documented by the Dutta laboratory
[30]. Although an siRNA is designed to be perfectly complementary to its target mRNA, it could suppress a number of unintended mRNAs (off-targets) with a complementary sequence to the seed region (positions 1~2 to 7~8 of an miRNA) via an miRNA mechanism
[29]. Since the seed region is only 6–8 nt long, the probability of appearance is once per ~4 to 64 kb, which estimates more than several tens of thousands of target sites in the human genome sequence. This estimation, albeit a simple arithmetic calculation, indicates that any siRNA is likely to have a significant number of off-targets.
The above story provided a theoretical background to justify the use of an siRNA-like duplex as the functional mimic of a miRNA
[31]. Although very briefly described above, there were a significant number of research endeavors to establish this miRNA overexpression method. In addition, the story gives an important note of strict warning that an siRNA-like duplex must be used only for bona fide miRNAs. If an siRNA-like duplex is used for overexpression of a small RNA which is not an miRNA, it will lead to suppression of off-target mRNAs, which is irrelevant to the small RNA’s genuine biological role. As a result, the experimental data will be entirely artifactual. miR-886-5p and -3p are such examples.
From overexpression data, several papers claimed that they found a cellular role of miR-886-5p or -3p and identified target mRNAs
[7][8][16][32][33][34][35][36][37]. However, any experimental data obtained from the transfection of miR-886-5p or -3p mimics do not reflect their natural role nor prove their existence. For easy understanding, I want to present a suppositional situation that a researcher uses an siRNA against a gene, for example, TP53 (encoding p53, a tumor suppressor protein). Transfecting the siRNA will lead to KD of TP53 and other off-target genes to elicit a resultant phenotype. However, this does not prove their biological existence nor a natural role of “a small RNA antisense to TP53”. Nobody would be interested in it, either. Likewise, an miR-886-5p or 3p mimic is an siRNA-like synthetic duplex; thus, it will suppress some off-target genes to cause a certain phenotype, when transfected into cells. However, this cannot be evidence for the natural existence or role of miR-886-5p or -3p, as in the case of “a small RNA antisense to TP53”. Some studies imply that a phenotype resulting from the transfection of an miR-886-5p or -3p mimic represents the gain-of-function of nc886, because they claim that miR-886-5p or -3p is derived from nc886
[7][8]. However, an miR-886-5p or -3p mimic might lead to a loss-of-function by acting as an siRNA targeting nc886. In fact, one report used an siRNA for nc886 KD
[38].
4. Features of ncRNA also Matter When Designing Gain-of-Function Experiments
Plasmid vectors are the most common tool for the ectopic expression of a gene. nc886 is silenced in a number of cancer cell lines. When attempting its ectopic expression in these cell lines, the vector-based method should certainly be the primary choice. In this strategy, knowledge of nc886 is needed for optimal experimental design.
nc886 is transcribed by RNA polymerase III (Pol III), but not by RNA polymerase II (Pol II)
[17][18][39][40][41]. Genes transcribed by Pol III (shortly, “Pol III genes” or “Pol III transcripts”) are classified into three types, according to cis-acting promoter elements that determine which initiation subunits of the Pol III enzyme to be recruited (reviewed in
[42]). Representative genes for type 1, 2, and 3 are 5S rRNA, tRNAs, and U6 snRNA, respectively. nc886 contains gene-internal promoter elements, A and B boxes, which resemble those of type 2. Some type 2 Pol III genes, such as vault RNAs (vtRNAs), 7SL RNA, and BC200, also require their 5′-upstream sequence, in addition to A and B boxes, and these genes are sub-classified as type 2H (reviewed in
[43]). nc886 is a paralog of vtRNAs and is supposed to belong to the type 2H. Actually, in my unpublished data, nc886 was efficiently expressed from a plasmid devoid of a mammalian promoter only when > 200 nt at the 5′-upstream was inserted together with the nc886 RNA region. This was one way to construct an nc886-expressing plasmid. Another way was to insert the nc886 RNA region under the U6 or H1 promoter, which is a type 3 Pol III promoter. Plasmids constructed according to these two strategies expressed nc886 correctly, as indicated by a single band at the identical size of the endogenous nc886 in Northern blot
[1][13][18][44]. Moreover, when stable cell lines expressing nc886 were made with these plasmids, the ectopic expression level of nc886 was comparable to the endogenous levels of cell lines naturally expressing nc886
[13]. These data ascertained the legitimacy of the ectopic expression and the reliability of the resultant phenotypes.
Commonly used expression vectors have a strong promoter, which is usually adopted from viruses such as cytomegalovirus (CMV) or simian virus 40 (SV40). A couple of studies employed these vectors to construct nc886-expression plasmids
[36][45]. It should be pointed out that the CMV or SV40 promoter is for Pol II. Most likely, Pol II will ignore a TTTT sequence, the termination signal for Pol III transcription, at the 3′-end of nc886. It is also questionable whether the transcription driven by an entirely irrelevant promoter would start at the correct position. Consequently, those plasmids are presumed to yield longer transcripts harboring the nc886 sequence. These extended transcripts cannot be distinguished from the correct nc886 of 101 nt, if qRT-PCR is employed as a method to confirm overexpression. I do not rule out a possibility that those Pol II transcripts might be capable of mimicking nc886 functionally, because they contain the nc886 sequence. Nevertheless, certainly there exists a risk that it cannot present nc886′s function, for a number of plausible reasons. These long, Pol II transcripts are likely to undergo a path different from the Pol III-transcribed nc886. As a result, they might lack necessary post-transcriptional processing steps or be mis-localized within a cell. Appending sequences in the longer nc886 transcripts might interfere with the formation of the nc886′s correct secondary structure.
In one report, a sufficient length (>200 nt) of the 5′-upstream sequence, together with the nc886 RNA region (101 nt), was inserted downstream of a Pol II promoter
[36]. Such a plasmid is expected to drive transcription of the correct nc886 RNA, in addition to long Pol II transcripts. Nonetheless, there are still concerns. A potent Pol II promoter might keep recruiting the Pol II enzyme complex, which competes with the Pol III enzyme complex for occupation on DNA, resulting in low-level expression of the correct nc886. Longer Pol II transcripts might inhibit the correct nc886 in a dominant-negative manner. These concerns indicate a need to conduct Northern blot or any functional assay to confirm whether the correct nc886 is expressed at a sufficient quantity.
5. Choosing a KD Method
RNAi-based methods and modified antisense oligonucleotides (ASO) are commonly used for KD of a gene (reviewed in
[46]). RNAi-based KD is usually highly effective for protein-coding genes, but ineffective for some ncRNAs
[47]. In addition, for nc886, RNAi-based method does not seem to be effective according to a report
[38] in which they transfected an siRNA against nc886 and measured the activity of protein kinase R (PKR), a protein that nc886 normally suppresses
[1][48][49][50]. PKR is typically activated by long dsRNAs (>55 nt) that are generated during viral infection. Since PKR should be autophosphorylated for its kinase activity, Western blot analysis of a phosphorylated form of PKR (phospho-PKR) is indicative of its activation
[51]. The transfection of siRNA against nc886 led to an increase in phospho-PKR, indicating nc886 KD
[38], but the fold-increase was very modest as compared to those in dsRNA or legitimate KD of nc886 (see below).
The standard method for nc886 KD in my laboratory is to transfect a 20 nt long ASO, having five 2′-O-methyl ribonucleotides at both ends
[1]. The backbone is phosphorothioate-modified. This design is a DNA-RNA mixmer, based on RNaseH-mediated cleavage of the heteroduplex between a target RNA and the middle DNA portion of the ASO
[52][53]. The transfection of this ASO into various cell lines led to efficient KD, as indicated by the measurement of the nc886 and phospho-PKR. In most cases, the 101 nt nc886 band was clearly decreased in Northern hybridization, and phospho-PKR was robustly increased to a comparable degree by its canonical activator dsRNA
[1][10][11][13][54][55][56][57].
Because it was mis-identified as an miRNA, several papers intended the KD of miR-886-5p or -3p using commercially available miRNA inhibitors
[16][32][35][58]. These inhibitors are also modified ASOs and thus, could have suppressed nc886. However, in these papers, nc886 was not their purpose and thus, was not measured. These miRNA inhibitor ASOs are worth considering for nc886 KD, although the DNA-RNA mixmer works well in most cell lines.
This entry is adapted from the peer-reviewed paper 10.3390/ijms23084251