Systematic identification and evolutionary features of rhesus monkey small nucleolar RNAs

It is widely accepted that up to 90% of the human genome is transcribed into various types of RNAs [1–4]. However, only a very small proportion of transcripts (~2-3%) encode proteins. Although there is a possibility that many transcripts are simply noise [5], a considerable number of non-protein-coding RNAs (npcRNAs/ncRNAs) are produced [1–4]. The increasing numbers of ncRNAs found by systematic genome-wide screening have also demonstrated the widespread existence of ncRNAs in nature [6–9]. The ncRNAs can be categorized by length as 19~35 nt small ncRNAs such as miRNAs and piRNAs [10–12]; intermediate-size ncRNAs, ranging between 50 and 500 nt, such as the small nucleolar RNAs (snoRNAs) [13]; and long mRNA-like ncRNAs with sizes larger than 500 nt [14–18].

snoRNAs function mainly as modulators of ribosomal RNAs (rRNAs) [19], and represent the largest group of functional ncRNAs. Based on sequence and structural features, snoRNAs can be classified into two families-box C/D snoRNAs and box H/ACA snoRNAs-which guide site-specific 2'-O-ribose methylation and pseudouridylation of rRNA, respectively [20, 21]. The spectrum of snoRNA targets is continuously growing. Some snoRNAs control methylation of tRNAs [22, 23]. Small Cajal body RNAs (ScaRNAs), a subset of snoRNAs with box C/D and/or box H/ACA, regulate post-transcriptional modification of RNA polymerase II-transcribed snRNAs [24]. Recent findings have demonstrated that snoRNA can also target mRNA, to guide alternative splicing [25]. Another interesting discovery is that snoRNAs may be precursors of microRNAs and possess microRNA-like functions [26, 27]. Together, available evidence suggests that snoRNAs may have broader functions than previously appreciated.

The genomic organization of snoRNA genes displays great diversity in different organisms. Unlike yeast and plants, in which snoRNAs are usually transcribed from independent polymerase II transcription units with dedicated promoters [28], most vertebrate snoRNAs reside in the introns of protein-coding or non-protein-coding genes and are generated by splicing-dependent processing [29, 30]. Intron-encoded snoRNAs may also have special promoters to drive snoRNA transcription [31]. Many snoRNA genes have multiple paralogs derived from one or more duplications [32]. In nematodes, the paralogs of intron-encoded snoRNA genes were likely generated by cis- and trans-duplication mechanisms [23]. Luo and Li demonstrated that most human box H/ACA snoRNAs were retrogenes produced by L1 integration [33]. Weber reported that many mammalian snoRNAs were mobile genetic elements designated as snoRNA/scaRNA retroposons (snoRTs, scaRTs) [34]. Recently, Schmitz and colleagues discovered a platypus-specific snoRNA retroposon with powerful transposable activity that replicated a single snoRNA to form about 40,000 paralogs in the whole genome [35]. It is therefore possible that retroposition of snoRNA genes may have played an important role during evolution of mammalian genomes.

Based on these recent findings, it is likely that ncRNAs have important functions in almost every aspect of eukaryotic growth regulation. However, only a limited number and classes of ncRNAs have been discovered to date. Therefore, systematic identification of ncRNAs from various organisms is a critical primary step in the provision of a road map for functional studies of ncRNAs in various organisms. The rhesus macaque (Macaca mulatta) is the most thoroughly studied primate apart from humans. Although phylogenetically separated by more than 70 million years of evolution [36, 37], rhesus macaques and humans are closely related and share a common ancestor dating back to about 25 million years ago [36, 38]. Therefore, study of rhesus monkeys assists primate evolutionary research and modern biomedical programs [38, 39]. A total of 21,905 protein-coding genes and 5,253 non-protein coding genes (including 715 predicted snoRNA loci) have been identified in the rhesus monkey genome by the ENSEMBL genome annotation group [6]. Although the expression pattern and possible functions of many protein-coding genes have been reported, identification of non-protein-coding genes of the rhesus monkey has relied only on computational predictions, by searching for sequences similar to those of known ncRNAs identified in other species. Such an approach is obviously inappropriate for identification of novel ncRNAs. Here, we conducted a systematic experimental identification of rhesus monkey ncRNAs by constructing a cDNA library derived from RNA fragments with sizes of 50 to 500 nt. We identified 117 non-redundant ncRNAs, including 80 snoRNAs and eight unclassified ncRNAs. We found that some of our identified ncRNAs were lineage- or species-specific. Further analysis of the genomic organization of these ncRNAs demonstrated that the majority represented snoRNAs with multiple paralogs in the rhesus monkey genome. Detailed analysis of the flanking sequences of each of the snoRNA paralogs revealed that some snoRNAs were retrogenes generated through L1-mediated integration machinery, suggesting that retroposon-mediated trans-duplication may have been a driving force for expansion of novel snoRNAs in the rhesus monkey genome.

Recent studies have demonstrated that the functions of non-protein-coding RNAs may encompass almost every aspect of biological activity in normal development and disease biogenesis [21, 25, 42–45]. Rhesus macaques are a suitable primate model for basic and applied biomedical research [38, 39]. However, in contrast to the considerable literature on human and mouse ncRNAs, rhesus monkey ncRNAs have not previously been systematically characterized. Here, we performed a detailed screening of the rhesus monkey intermediate-size ncRNA transcriptome and cloned 117 rhesus monkey ncRNAs, including 80 snoRNAs, eight unclassified ncRNAs, and 29 known RNAs (snRNAs, Y RNA, and others). By comparative genomics analysis, we found several lineage- or species-specific snoRNAs. Genomic organization analysis showed that the majority of rhesus monkey snoRNAs have many paralogs in the rhesus monkey genome. By flanking sequence analysis, we found that SINE-like retroposon-mediated trans-duplication may have been an important mechanism in expansion of novel snoRNAs in the rhesus monkey genome.

Among the 117 identified rhesus monkey ncRNAs, eight unclassified ncRNA candidates could not be assigned to any known class of ncRNA. These eight unclassified ncRNAs were ubiquitously expressed in the six rhesus monkey tissues tested. Recently, we also identified nine unclassified ncRNAs from the chicken [46]. Previous reports also showed that some ncRNAs obtained from cDNA library sequencing did not belong to any known ncRNA family, and these ncRNAs were designated as unclassified or unknown [13, 31, 47]. Hüttenhofer and coworkers found 57 such unclassified ncRNAs, of length 50~500 nt, in mouse brain cDNA libraries [13]. Deng et al. reported 14 unclassified ncRNAs of length 70~200 nt in a C. elegans cDNA library [31]. Yuan identified 29 unclassified ncRNAs by constructing cDNA libraries from four developmental stages of Drosophila melanogaster [48]. These unclassified ncRNAs often show little sequence conservation and are less prevalent compared to known snoRNAs. However, these observations do not mean that the unclassified ncRNAs are non-functional [13, 31, 48]. The increasing number of newly identified unclassified ncRNAs suggests that other types/classes of ncRNA of intermediate size (50~500 nt) remain to be identified, and novel ncRNA families will likely be susceptible to classification using enhanced bioinformatic comparisons and extensive functional studies of the roles played by such ncRNAs.

Previous reports showed that the majority of known snoRNAs were conserved between human and mouse, at a level of 80~90% [49]. Most rhesus monkey snoRNAs identified in the present study show high homology to those of the human and mouse. However, 13 snoRNAs had a conservation score below 0.6 (Table 1), suggesting that some snoRNAs are less conserved between primates and rodents. Using comparative genomics analysis, we found several lineage- or species-specific snoRNAs. Fifteen snoRNA families were ancient, being present at an early stage of vertebrate evolution, whereas 11 snoRNA families appeared after the divergence of birds and mammals. Fourteen young snoRNA families arose during mammalian evolution and one of these (SNORA15) developed only after primates had arisen. Our findings are in line with recent studies in other species. Previously, we found 30 chicken/bird-specific ncRNAs [46], and Schmitz reported 49 platypus-specific snoRNAs [35]. Computer analysis of human genomic tiling array data revealed 300 putative candidates for classification as primate-specific ncRNAs [50]. Together with previous reports, our data show that ncRNAs may play important roles in lineage development, or speciation, during evolution.

Although homologs of some rhesus monkey snoRNAs could be found in the mouse and human genomes, the expression of several snoRNAs was not detectable by northern blotting, suggesting that some snoRNA homologs might be pseudogenes without transcriptional potential in the human and/or mouse. Thus, we found only 14 potential primate-specific and eight rhesus monkey (or non-human primate)-specific transcripts (Table 1 and Figure 2). However, it remains possible that undetectable expression in the human or mouse might be attributable to transcriptional regulation by spatio-temporal, physiological, or pathological stimuli/stresses that were not present under the normal conditions prevalent when our tissue samples were taken. In support of this hypothesis, several examples of tissue-specific expression of ncRNAs have been reported in previous studies describing brain-specific snoRNAs or snoRNAs involved in neuronal development [51]. By analogy, some microRNAs and piRNAs display specific spatio-temporal expression patterns, and play functional roles in cell differentiation and organogenesis during development [11, 12, 52, 53]. In the present study, we also found that SNORA71, ubiquitously expressed in human and rhesus monkey tissues, is predominantly expressed in the brain of mouse.

In vertebrates, most snoRNAs are located within introns of protein-coding or non-protein-coding genes [21, 54]. Some snoRNAs are present as several copies, either in different introns of the same gene or within introns of different genes [32, 55]. Genomic organization analysis showed that the majority of the rhesus monkey snoRNAs identified in this study have multiple paralogs in the rhesus genome, suggesting redundancy arising from duplication, including transposition. Diverse molecular mechanisms may be involved in the creation of protein-coding genes, such as gene duplication and retroposition [56]. To investigate the mechanisms of rhesus monkey snoRNA expansion, we analyzed the flanking sequences of each snoRNA paralog and found that these sequences adjacent to some rhesus monkey snoRNAs have a typical SINE-like retroposon characterized by a poly(A) end and TSDs, suggesting that some rhesus monkey snoRNA paralogs are retrogenes formed by autonomous retroposon-mediated retroposition. In addition, the 5' flanking sequences of rhesus monkey SNORA76b and SNORA24b possess T2A4 motifs, which are preferentially recognized by the L1 retroposon-encoded nicking endonuclease, suggesting that SNORA76b and SNORA24b were generated from a parental copy by L1 integration machinery-mediated retroposition. Significantly, we found that six paralogs of SNORA25 also possess typical SINE-like retroposon characteristics, and contain multiple poly(A) sequences, indicating that SNORA25 underwent multiple duplication events during evolution. Thus, we propose a model involving retroposition for SNORA25 duplication. Recently, the mechanisms of snoRNA gene expansion in other species have been reported. In nematodes, some snoRNA paralogs were generated by cis- or trans-duplication [23]. Other data suggest that mammalian snoRNA genes are SINE-like retroposons (snoRTs/snoRTEs), and that retroposition mediated by snoRTs may have played an important role in snoRNA expansion during evolution of the mammalian genome [33–35]. The extensive expansion of snoRNA-encoding genes during mammalian evolution might ensure the presence of a functional copy when a parental gene loses function because of mutation. On the other hand, novel paralogs could independently evolve to generate isoforms with different targets/functions, for example the acquisition of new sites complementary to modification regions of rRNAs [34].

In the present study, we provide the first experimentally-derived catalog of rhesus monkey ncRNAs. Small nucleolar RNAs (snoRNAs) comprise one of the largest groups of functionally diverse ncRNAs currently known to exist in eukaryotic cells. By performing northern blotting and comparative genomic analysis on rhesus monkey snoRNAs, we determined several features of interest. First, we identified several lineage- or species-specific snoRNAs. Moreover, we observed that the majority of snoRNAs have multiple paralogs in the rhesus monkey genome. Based on the data from the ENSEMBL genome annotation project, the total number of snoRNA-encoding genes was shown to have increased during vertebrate evolution. Our results demonstrate that SINE-like retroposon-mediated trans-duplication may have been a driving force for the expansion of novel snoRNAs in the rhesus monkey genome.