In the post-genome era, there has been a major re-evaluation of the importance of RNA genes. Currently, despite decades of not seriously investigating RNA genes, the number of recognized RNA genes is already on a par with the number of protein-coding genes (see Table 1 for analyses of the GRChp38.12 reference genome sequence). And it is now clear that RNA genes perform a wide variety of functional roles, with many RNAs working as regulators of specific target genes—see Figure 1 for a brief illustration of some RNA functions.
Table1. HEADLINE HUMAN GENE STATISTICS FROM THE GRCh38.p12 REFERENCE SEQUENCE (NOVEMBER 2017 FREEZE)
Fig1. An illustration of the functional diversity of RNA. Different RNAs within the white boxes include individual RNAs (red font) or RNA families (black font). 7SL RNA is the RNA component of the signal recognition particle that regulates protein export from cells. RNAs involved in RNA maturation include spliceosomal small nuclear RNAs (snRNAs), small nucleolar RNAs (snoRNAs), small Cajal body RNAs (scaRNAs), and RNA components of two RNA ribonucleases: RNase P, which cleaves tRNA precursors, and RNase MRP, which cleaves rRNA precursors and also has a crucial role in initiating mtDNA replication. The Y RNA family is involved in replicating chromosomal DNA. Telomere DNA synthesis is performed by a ribonucleoprotein complex of TERC (telomerase RNA component) and a reverse transcriptase, and TERRA regulates certain telomere characteristics, including length. Some RNAs have general accessory roles in transcription, but many regulate specific target genes. Three classes of tiny RNA use RNA interference pathways to act as regulators: individual microRNAs (miRNAs) regulate the expression of defined target genes; piRNAs (PiWi protein-interacting RNAs) regulate the activity of transposons in germ-line cells; and endogenous short interfering RNAs (endo-siRNAs) act as gene regulators and also regulate some types of transposon. A large number of long noncoding RNAs regulate genes, often at the transcriptional level; some are involved in epigenetic gene regulation, in imprinting, X-inactivation, and so on.
Short and long noncoding RNA
It has become a convention to divide RNA genes into two classes, according to whether their mature RNA products are up to 200 nucleotides in length (short noncoding RNA) or over 200 nucleotides in length, when the RNA is described as a long noncoding RNA (lncRNA). This extremely arbitrary division was designed simply as a way of separating most of the long-familiar ubiquitous RNAs (which were thought to assist in general mechanisms used to express genes) from long RNAs that acted as regulators of specific target genes. In many ways this is a very unsatisfactory separation. First, many important regulatory RNAs are now known to be small noncoding RNAs. Additionally, this convention would split the ribosomal RNAs into two classes, identifying the longer ribosomal RNAs (cytoplasmic 28S and 18S rRNA; mitochondrial 16S and 12S rRNA) as long noncoding RNAs, while classifying the cytoplasmic 5.8S and 5S rRNAs as short noncoding RNAs. If we disregard rRNAs for the moment, the compositions of the two classes are listed below.
• Short noncoding RNA. There are seven major subfamilies (Table 2). Four are well-understood classes of ubiquitous RNAs; three are recently discovered classes of tiny regulatory RNAs. MicroRNAs are often tissue-specific and/or developmental stage-specific regulators; they work by binding to short recognition sequences in RNA transcripts produced by the target genes. piRNAs (PiWi protein- interacting RNAs) and endogenous siRNAs (short interfering RNAs) are important in germ line cells as part of genome defense mechanisms.
• Long noncoding RNA (lncRNA). The interest in lncRNAs is predominantly because this noncoding RNA class contains many important regulatory RNAs. Some lncRNAs work in the nucleus to regulate target genes, either through chromatin modification (epigenetic regulation), or through transcriptional regulation (often as natural antisense RNA transcripts). Many lncRNAs resemble mRNAs in having 5′ caps and 3′ poly(A) tails, but others do not, including many thousands of circular RNAs (which can be produced in different ways, including “back splicing” where the 5′ splice acceptor preceding an upstream exon splices to the 3′ splice donor of a downstream exon) and stable linear RNAs processed from intronic RNA sequences. Both of the latter lncRNA categories include some examples that have been claimed to regulate target genes.
Table2. THE ARBITRARY DIVISION OF NONCODING NONRIBOSOMAL RNAs INTO SHORT AND LONG CATEGORIES
Problems in identifying functional RNA genes
When it became clear that noncoding RNAs had diverse functions, searches for novel human RNA genes were stepped up. But identifying and verifying novel RNA genes has generally been much more difficult than identifying protein-coding genes for various reasons. Two general explanations for the difficulty in identifying novel RNA genes apply to all RNA genes: there are no open reading frames to search for, and by comparison with proteins, RNA sequences have often been poorly conserved during evolution.
Functional noncoding RNAs are comparatively poorly conserved for two rea sons. First, unlike any protein, the sequence of some functional RNAs may be largely unimportant for their function. Because both enhancers and promoters are now known to initiate transcription bidirectionally, many long noncoding RNAs might not perform sequence-specific functions. Most antisense RNAa, for example, might have the sequence-independent function of simply being transcribed to disrupt transcription of the sense RNA. Even where it is clear that the sequence of an RNA is important for its function, natural selection may be predominantly focused on a limited number of nucleotide positions that are important in maintaining important functional features (such as binding sites for other molecules) and structural features (such as the stems of stem-loop structures—the structures of RNAs are often highly conserved, if not their sequences).
An additional reason that hindered identification of novel RNAs is that some RNAs, such as microRNAs, are tiny. Using computer programs alone to scan the genome and transcriptomes to identify tiny RNAs was too challenging. Instead, human microRNAs were discovered only after experimental studies revealed that microRNAs were functionally important gene regulators in model organisms such as C. elegans and Drosophila melanogaster, prompting strenuous efforts to identify human sequences that resembled t hem.)
Problems in defining RNA genes
An additional problem concerns how one defines RNA genes. It is now accepted that the great majority of the genome—at least 80%—is transcribed to produce a rather complex profile of RNA transcripts dominated by noncoding transcripts. Transcription can also occur in DNA sequences that we would not regard as genes, such as highly repetitive centromere and telomere DNA sequences. It is also often difficult to classify transcription units that make natural antisense transcripts as RNA genes: some antisense transcripts can extend over very large regions corresponding to several genes transcribed on the opposing DNA strand.
Many other functional RNAs are produced from transcription units that are not easily viewed as genes. For example, hundreds of thousands of different tiny piRNAs are made as part of general genome defense systems, acting through RNA interference pathways to limit the mobilization of endogenous transposons in germ cells. Additionally, the traditional view that intronic RNA sequences are excised and then degraded has been challenged by the finding of many thousands of circular RNAs containing intronic sequence and some stable linear intronic RNA sequences. For both of these types of noncoding RNAs, some examples are known that appear to play a role in gene regulation. As described below, intronic RNAs produced during the expression of some protein-coding genes are processed to produce small, functional noncoding RNAs. And, finally, some protein-coding genes are known to produce both mRNAs and functional noncoding RNAs that significantly overlap in sequence (Table 3).
Table3. FEATURES THAT CAUSE UNCERTAINTY ABOUT WHETHER A GENE SHOULD BE LABELED AS PROTEIN-CODING OR NONCODING, OR BOTH
Partly because of the above complexities, it has been proposed that mature non coding transcripts should be considered the functional units, not the DNA sequences from which they are transcribed. But that inevitably raises difficult challenges. DNA sequences are at least stable entities, being present in all nucleated cells, and, difficult though it has been, one can imagine eventually producing a definitive catalog of genes. But the sheer complexity of noncoding transcripts, their variable expression in different cell types and developmental stages, and the huge range in the expression levels makes the challenge of producing a definitive list of functional noncoding RNAs a formidable one.
