SnoRNAs from the filamentous fungus Neurospora crassa: structural, functional and evolutionary insights

Identification of 55 box C/D and 20 box H/ACA snoRNAs from N. crassa

The snoRNA candidates identified by cDNA cloning or the snoscan program were subsequently confirmed by northern blot and/or reverse transcription analyses. The expression of 27 box C/D and all 20 box H/ACA snoRNAs were positively detected as expected (Figure 1 and 2). Among these snoRNAs, the sequence of CD31 snoRNA obtained from the cDNA cloning appears uncompleted; it corresponds to the second half of CD31 full-length which is further validated by the northern blotting.

Together, through the combination of computational analysis and construction of the specialized cDNA libraries, 55 box C/D and 20 box H/ACA snoRNAs were identified and all the snoRNAs are denominated according to the order of identification (Table 1 and 2).

In most cases (86%) the C and D boxes in snoRNAs are highly conserved when compared to the consensus sequence (UGAUGA and CUGA, see Additional file 1). However, the C' and D' box are nonconserved and exhibit much more sequence-degeneracy than their C and D box counterparts. In some instances, the C' and D' box cannot be unambiguously inferred as a result of the absence of the functional region. Generally, the box C/D snoRNAs from N. crassa are similar to their metazoan and yeast counterparts in size. However, the sizes of box H/ACA snoRNAs from N. crassa are peculiar. Almost all of them are larger than 160 nt (Figure 2), reminiscent of some irregular box H/ACA snoRNAs in S. cerevisiae. These observations show that the N. crassa snoRNAs have their own unique sequence and structural features (see Additional file 2 and 3).

Functional properties of the N. crassa box C/D and box H/ACA snoRNAs

Based on the modification rules of snoRNAs [2] , 55 box C/D snoRNAs from N. crassa were predicted to direct 71 methylations. These include 64 methylations on rRNAs which comprise 43 methyls on 26S rRNA, 20 methyls on 18S rRNA and one methyl on 5.8S rRNA (see Additional file 4A). The remnant seven methylations consist of four methyls on snRNAs and three methyls on tRNAs (see Additional file 4B and 4C). Furthermore, the structure and function elements of U14 which participate in the processing of pre-rRNA were unambiguously identified. Interestingly, two different methylated sites were predicted to be guided by the same functional element of a single snoRNA CD27. Two box C/D snoRNAs (CD53 and CD55) lack the sequences complementary to either rRNAs or snRNAs and therefore belong to orphan snoRNAs. Fourteen box H/ACA snoRNAs were predicted to guide 17 pseudouridine sites of rRNAs (see Additional file 5), and no pseudouridine sites on snRNAs were predicted. The remaining six box H/ACA snoRNAs were also classified into an orphan snoRNA family lacking functional region complementary to rRNA, tRNA or snRNA. A different modification pattern appears in N. crassa as compared to the two yeasts S. cerevisiae, and S. pombe (see discussion)[38, 39].

Interestingly, a novel snoRNA, CD29, possesses two guide elements that can form duplexes with U2 snRNA and 5.8S rRNA for 2'-O-methylation. Primer extension mapping of 2'-O-methylated nucleotides of the U2 snRNA and 5.8S rRNA in the presence of low concentration of dNTPs resulted in stop signals at the G32 and A43 residues, indicating that U2-A31 and 5.8S-A42 are methylated (Figure 3). We have identified cognates of CD29 in other filamentous fungi, however, these cognates only possess the guide sequence for the methylation of U2 snRNA. This suggests that CD29 evolves from the snoRNA with a single guide function. This is reminiscent of the human small Cajal body-specific RNAs (scaRNAs) that can guide modification of the RNA polymerase II-transcribed snRNAs such as U2 snRNA. The comparative analyses revealed that CD29 and its homologs in fungi have one functional element similar to that of human SCARNA9 which was first known as Z32 (GeneBank accession no. AJ009638), and therefore was homologous to this human scaRNA. In addition, we characterized a multi-function box C/D snoRNA, CD11, in N. crassa . CD11 has the potential to direct a methylation in U6 snRNA, and two methylations in 18S and 26S rRNAs, respectively (Figure 3). Interestingly, the CD11 is also partially similar to mgU6-47 in mammals [40], but possesses a novel function that can guide a N. crassa-specific methylation on 26S rRNA at A356.

Genomic organization and expression of the snoRNAs in N. crassa

The genomic organization of the snoRNA genes in N. crassa exhibits great diversity. Among the 55 box C/D snoRNAs, forty five snoRNA genes are intron-encoded in protein-coding or non-coding host genes. The remaining nine were found in the intergenic regions with a putative polymerase II promoter upstream and appeared independently transcribed. Meanwhile, six gene clusters that only encode box C/D snoRNAs were identified from N. crassa . Interestingly, an exon-encoded snoRNA (CD6) was identified in the snoRNA gene cluster III in contrast to another two intron-encoded snoRNAs (CD9 and CD17) in the same cluster (Figure 5). Of 20 box H/ACA snoRNA genes, 16 are located in intergenic regions and two are intron-encoded. In particular, two snoRNA genes (ACA10 and ACA16) are located in the 3' UTR of two hypothetical protein genes, one of which is similar to phosphoglycerate mutase. Obviously, different strategies dominate in the expression of the two families of snoRNA genes in N. crassa .

In accordance with the mode of one snoRNA per intron in vertebrates [4], a large proportion of the box C/D snoRNA genes (45 of 55) are located within introns of the host genes. The distances from the intronic snoRNA genes to the 3' splice sites of introns, which has been proven to be important for the effective processing of intronic snoRNAs from their host mRNA precursors [41, 42], resemble those in D. melanogaster [32, 41, 42]. The distances from the snoRNA genes to the 5' splice sites appear to mainly be between 41 to 60 nt, similar to those in human[41] (Figure 4).

Remarkably, five (cluster I to V) of the six box C/D snoRNA gene clusters arehighly conserved between yeast and N. crassa (Figure 5). Although these host genes were not well annotated for their introns and exons in the N. crassa genome, canonical intron splicing sequences were observed flanking every cluster of snoRNA genes. To further confirm this observation, the mature RNA transcripts were identified with the expected sizes by cloning and sequencing of RT-PCR products. It is worth noting that two snoRNA genes, CD16 and CD37, in the cluster V are validated to be co-transcribed by RT-PCR and sequencing, though each of the snoRNA genes in the cluster has a putative promoter upstream. Intriguingly, the putative promoter upstream of CD37, a homologue of U14, would play a role in guaranteeing and promoting the function of U14 that has been demonstrated vital in diverse eukaryotes. Our results further revealed that the genomic organization of the host genes for these five clusters is most like the UHG gene in animals. The host genes of Cluster I to V only contain short open reading frames with length ranging from 159 bp to 267 bp, suggesting the little potential for protein coding just like the gas 5 [43].

Unexpectedly, various alternative splicing events were found in the processing of polycistronic transcripts from the snoRNA gene clusters I and II by analyzing cDNA sequences from RT-PCR of the transcripts (Figure 6). In cluster I, two alternatively spliced transcripts, differing by the absence of exon 2 or exon 2 plus exon 3 were detected. The pattern of alternative splicing in the expression of cluster II was contingent on an alternative 3' splice site that allows the lengthening or shortening of exon 3.

High diversity of post-transcriptional modification predicted by snoRNAs in fungi

Identification of guide snoRNAs in diverse organisms can provide valuable information towards understanding RNA modification patterns and their function [18]. It is interesting to compare the pattern of modifications on target RNAs of N. crassa to those described in the two yeasts, S. cerevisiae and S. pombe. Among 71 methylations predicted by the guide snoRNAs in N. crassa , 32 represent the most highly conserved modifications shared by the multicellular fungi and the yeasts, and 31 (43.7%) are modifications that have not yet been reported in other fungi when compared with the two unicellular yeasts(Figure 7). In the yeasts, only ten and eight methylations are S. cerevisiae-specific and S. pombe-specific, respectively. Our results imply a more complex modification pattern in multicellular fungi than in unicellular yeasts. They also reveal the high diversity of post-transcriptional modification of RNAs in the fungus kingdom as it has been shown that about 40% of methylations are species-specific in a protozoan Trypanosoma [17]. The species-specific modifications highlight the different modification patterns and their peculiar importance. Although eliminating a single modification does not have a dramatic effect on the ribosome [44], loss of three to five modifications in an intersubunit bridge of the ribosome (helix 69) impairs growth and causes broad defects in ribosome biogenesis and activity [45]. On the other hand, early studies have demonstrated that ribosome modifications play roles in determining antibiotic resistance or sensitivity [15, 46]. Thus the species-specific modifications have potential use in finding therapeutic targets for prevention and treatment of diseases caused by some eukaryotic pathogens.

Another interesting observation in this study was the presence of duplexes between box C/D snoRNAs and tRNAs (tRNA^Trp and tRNA^Leu from N. crassa Database). Duplexes between tRNA and snoRNAs have been also found in C. elegans [24] and recently in Plasmodium falciparum [47]. tRNA modification guided by snoRNAs has been also reported in Archaea [23]. This study provides for the first time a prediction of fungal snoRNAs and their potential target sites in tRNAs, although these remain to be confirmed by further experiments.

Structural and functional evolution of snoRNAs in fungi

Our study demonstrates the extensive separation and recombination of functional regions occurring during the evolution of snoRNA genes in fungi. For instance, the CD5 snoRNA in N. crassa possesses two conserved guiding elements. In S. cerevisiae, however, the conserved function of CD5 is executed by two independent snoRNAs, snR72 and snR78, with a single functional element [48] (Figure 8). This suggests that CD5 may have evolved as a double-guide snoRNA through recombination of two different halves of two ancestral single-guide snoRNAs. The other possibility is that a gene duplication of a double-guide snoRNA gene in S. cerevisiae led to specialization of each paralog to only target one modification site followed by loss of the other guide element for both paralogs. Another example is CD50 and CD51 that carry a conserved guiding function for U24 and U24b in S. pombe, respectively. In contrast, the U24 in S. cerevisiae has two guiding functions. Comparative analyses revealed that the structure and function of U24 are well conserved among the budding yeast and the flowering plants A. thaliana and rice, but the homologues of the S. cerevisiae U24 exist as two independent snoRNAs in other distant eukaryotes, such as human and mouse [49]. This suggests that U24 snoRNA gene has evolved in two pathways, with one leading to a dual functional snoRNA gene and the other separating the guiding functions and giving rise to two independent snoRNA genes.

It has been demonstrated the reciprocal evolutionary change between snoRNA complementary region and their rRNA target sequence in plants and nematodes[9, 24]. Our analyses indicate that co-evolution between snoRNAs and rRNAs exists widely in N. crassa (Figure 9) and plays an important role in preservation of phylogenetic conserved methylated sites of rRNAs which is essential for protein synthesis.

RIP may impact on the generation of snoRNA isoforms by gene duplication and transposition

SnoRNA gene isoforms or variants exist widely in diverse organisms, particularly in plants. For example, 97 box C/D snoRNAs with a total of 175 different gene variants were identified in the A. thaliana genome [50], and 346 gene variants encoding 120 box C/D snoRNAs were found in Oryza Sativa [9]. Compared with the plant snoRNAs, only a paucity of yeast snoRNA paralogs was detected because of a relatively small compact genome (~12 Mb for S. cerevisiae). The N. crassa genome (~ 40 Mb) is three-fold larger than that of the yeast; however, most snoRNA genes in this species are singleton. Why are the snoRNA genes devoid of isoforms in the N. crassa genome? It is known that a mutagenic process termed repeat-induced point mutation (RIP) has a profound impact on N. crassa genome evolution, which has greatly slowed the creation of new genes through genomic duplication and resulted in a genome with an unusually low proportion of closely related genes [51]. Of the predicted 10082 protein-coding genes, only six pairs (12 genes) share >80% nucleotide or amino-acid identities in their coding sequences [36]. RIP identifies duplications that are greater than ~400 bp (~1 kb in the case of unlinked duplications) and induces C:G to T:A during the sexual cycle [52, 53]. Early studies have provided clear evidence of retrotransposons inactivated by RIP [54, 55]. The analysis of the N. crassa genome sequence also revealed a complete absence of intact mobile elements [36]. Therefore the creation of new genes including snoRNA genes or their host genes through gene duplication and transposition seems to be impeded. It has been proposed that most, if not all paralogs in N. crassa duplicated and diverged before the emergence of RIP [51]. We have identified three U3 snoRNA gene variants, NcU3A, NcU3A-2 and NcU3A-3 in N. crassa (37). The sequence analysis revealed that these molecules have undergone nucleotide substitutions rather than RIP according to the calculation method previously reported [36]. In the case of CD46A and CD46B, we speculate that the two snoRNA gene isoforms may have duplicated and diverged before the emergence of RIP.

Alternative splicing in the expression of non-coding RNA genes with introns

It is well known that alternative splicing is an important and widespread process where one gene produces more than one type of mRNA which is then translated into different proteins in multicellular organisms [56]. Bioinformatic analysis indicates that 35-65% of human genes are involved in alternative splicing, which contributes significantly to human proteome complexity [57, 58]. However, alternative splicing is rarely reported for non-coding RNA genes which encode multiple introns. In this study, we identified several alternative splicing events that occurred in the processing of RNA precursors transcribed from the snoRNA gene cluster I and II of N. crassa. It has been reported that the mouse gas5 gene, a non-coding RNA and snoRNA host gene, had four alternative splicing transcripts [43]. Although different in snoRNA composition, the snoRNA gene clusters in N. crassa are most like UHG genes resembling gas5. Our results show that alternative splicing occurs frequently in the expression of snoRNA host genes in lower eukaryotes. This lends support to the concept that alternative splicing may be an ancient mechanism in regulating the expression of both protein-coding and non-coding RNA genes with introns. More work is necessary to elucidate the biological significance of the alternative splicing in the expression of non-coding RNA genes.

Strains and Media

The N. crassa wild-type strain (As 3.1604, purchased from the China General Microbiological Culture Collection Center) was used for the construction of the cDNA library and all RNA analyses. The strain was grown in PSA medium (2% sucrose, 20% extract of potato) at 30°C. The Escherichia coli strain TG1 grown in 2YT (1.6% Bacto tryptone, 1% yeast extract, 0.5% NaCl) liquid or solid medium was used for cloning procedures.

Construction and screening of cDNA library

We prepared total RNA from N. crassa culture according to the guanidine thiocyanate-phenol-chloroform procedure described by Chomoczynski et al [59]. Small RNA (~20 μg) was fractionated by 50% PEG-8000 and 0.5 M NaCl. The construction of cDNA library were performed as described previously with minor modifications (see Additional file 6) [60]. After randomly sequencing clones, we employed dot hybridization to screen the colony PCR products with P47 and P48 as described by Liu et al. [37] We sequenced clones exhibiting the lowest hybridization signal.

Computational identification of box C/D snoRNA genes

Genomic sequences of N. crassa [36] available at http://www.broad.mit.edu/annotation/genome/neurospora/Home.html (N. crassa assembly 7) were downloaded and searched for potential box C/D snoRNAs target rRNA/snRNA using snoscan [12] with default parameters. Methylated sites prepared for the snoscan included the conserved methylated nucleotides of S. cerevisiae (yeast snoRNA database), H. sapiens (snoRNA-LBME-db), and D. melanogaster [32]. The snoscan results were processed by an in-house developed perl program for candidate selection. A sequence with the following characteristics was considered as candidate: ① box C motif bit score ≥ 7.48, box D motif bit score ≥ 8.05, ② the guide bit score ≥ 18.65, the guide sequence and the target sequence can form a concatenated 10 bp duplex with at most 1 GU pair allowed, or can form a concatenated 9 bp duplex with high GC content. ③ if the guide region is adjacent to the D' box, the length of spaces between box C and guide sequence must be ≤ 20 bp. If the guide region is adjacent to the D box, the length of spaces between box C and guide sequence must be between 40 and 85 bp. ④ total sequence length between 75 bp and 130 bp, total overall bit score ≥ 20. The candidates within CDS region predicted by Broad/Whitehead Institute automatic gene calling software (a combination of manual annotation, FGENESH, GENEID, and GENEWISE) [36] were removed. The BLAST program [61] was used to search gene variants of all novel snoRNA genes to establish the snoRNA gene isoforms. About 1 kb of flanking sequences of the snoRNA gene candidates was searched further for possible box C/D snoRNA genes and additional non-canonical C/D gene candidates.

Northern blot analysis

An aliquot of 30 μg total RNA was separated by electrophoresis on an 8% polyacrylamide gel containing 8 M urea and electrotransferred onto nylon membrane (Hybond-N+; Amersham) using semi-dry blotting apparatus (BioRad). After immobilizing RNA using a UV cross-linker, northern blot hybridization was performed as previously described [49].

Reverse transcription and mapping of ribose methylation

Reverse transcription was carried out in a 20 μl reaction mixture containing 15 μg of total RNA and a corresponding 5'-end-labeled primer. After denaturation at 65°C for 5 min and then cooling to 42°C, 200 units of M-MLV reverse transcriptase (Promega) were added and extension carried out at 42°C for 1 hour. The cDNA was separated on an 8% polyacrylamide gel (8 M urea) and then analyzed with an imager.

The mapping of rRNA methylated sites was determined by primer extension at low dNTP concentrations as described previously [40, 62]. Briefly, the N. crassa 18S and 26S rDNA were amplified by PCR with the primer pair Nc18SF/Nc18SR and Nc26SF/Nc26SR, respectively, and then cloned into the pMD-18T vector (Takara). The plasmid DNA insert was directly sequenced with the same primer used for reverse transcription and run in parallel with the reverse transcription reaction on an 8% polyacrylamide gel (8 M urea).

RT-PCR analysis

15 μg of total RNA was reverse transcribed with 200 U of M-MLV reverse transcriptase (Promega) using the box C/D snoRNA gene cluster specific reverse primers (see Additional file 7) in a 20 μl reaction mixture as described above for reverse transcription and mapping of ribose methylations. The negative RT control was carried out without M-MLV reverse transcriptase. We designed two specific antisense oligonucleotides: the first reverse primer used in the reverse transcription reaction overlaps the last several nucleotides of the second reverse primer used in the PCR reaction to help avoid non-specific PCR products. After 1 h at 42°C, 2 μl of RT reaction was used for PCR amplification with the second reverse primer and the corresponding forward primer (see Additional file 7) in a final volume of 20 μl. The positive PCR control was performed on N. crassa genomic DNA with the same pair of primers. Negative PCR control was performed on 2 μl of the negative control RT reaction with the same pair of primers. The PCR program: 30 cycles of denaturation (30 s, 94°C), annealing (30 s, 50-55°C), and extension (1-2 min, 72°C), following by a final extension (10 min, 72°C). The PCR product was purified from a 1.5% agarose gel with the QIAquick Gel extraction Kit (QIAGEN) and cloned into pMD-18T vector (Takara) and transformed into the strain TG1 of E. coli . Positive clones were subsequently chosen for sequencing.

Oligonucleotides

Oligonucleotides used for construction of the cDNA library, northern blot analyses of novel snoRNAs and the primers for reverse transcription and RT-PCR experiments are not shown (see Additional file 7).

Database accession numbers

The sequences of all snoRNAs determined in this work have been deposited in the GenBank Nucleotide Sequence Databases under accession numbers EU780925 - EU780999 and EU526091-EU526095.