Testing the "proto-splice sites" model of intron origin: evidence from analysis of intron phase correlations

A few nucleotide sites of nuclear exons that flank introns are often conserved. A hypothesis has suggested that these sites, called "proto-splice sites," are remnants of recognition signals for the insertion of introns in the early evolution of eukaryotic genes. This notion of proto-splice sites has been an important basis for the insertional theory of introns. This hypothesis predicts that the distribution of proto-splice sites would determine the distribution of intron phases, because the positions of introns are just a subset of the proto-splice sites. We previously tested this prediction by examining the proportions of the phases of proto-splice sites, revealing nothing in these proportion distributions similar to observed proportions of intron phases. Here, we provide a second independent test of the proto-splice site hypothesis, with regard to its prediction that the proto-splice sites would mimic intron phase correlations, using a CDS database we created from GenBank. We tested four hypothetical proto-splice sites G / G, AG / G, AG / GT, and C/AAG / R. Interestingly, while G / G and AG / GT site phase distributions are not consistent with actual introns, we observed that AG / G and C/AAG / R sites have a symmetric phase excess. However, the patterns of the excess are quite different from the actual intron phase distribution. In addition, particular amino acid repeats in proteins were found to partially contribute to the excess of symmetry at these two types of sites. The phase associations of all four sites are significantly different from those of intron phases. Furthermore, a general model of intron insertion into proto-splice sites was simulated by Monte Carlo simulation to investigate the probability that the random insertion of introns into AG / G and C/AAG / R sites could generate the observed intron phase distribution. The simulation showed that (1) no observed correlation of intron phases was statistically consistent with the phase distribution of proto-splice sites in the simulated virtual genes; (2) most conservatively, no simulation in 10,000 Monte Carlo experiments gave a pattern with an excess of symmetric (1, 1) exons larger than those of (0, 0) and (2, 2), a major statistical feature of intron phase distribution that is consistent with the directly observed cases of exon shuffling. Thus, these results reject the null hypothesis that introns are randomly inserted into preexisting proto-splice sites, as suggested by the insertional theory of introns.     

Insertion of spliceosomal introns in proto-splice sites: the case of secretory signal peptides

Analysis of the exon-intron structures of 2208 human genes has revealed that there is a statistically highly significant excess of phase 1 introns in the vicinity of the signal peptide cleavage sites. It is suggested that amino acid sequences surrounding signal peptide cleavage sites are significantly enriched in phase 1 proto-splice sites and this has favored insertion of spliceosomal introns in these sites.     

Proto-splice site model of intron origin

It is proposed that nuclear pre-mRNA introns (classical introns) were first generated as by-products during the evolution of alternative splicing. They were formed whenever two splice sites within the coding sequence of ancestral genes were used at a frequency that removed the coding constraint from the intervening sequence. Once introns had evolved, it is suggested that they were spread by the splicing machinery which inserted them into proto or cryptic-splice sites of other genes by reverse splicing, so giving rise to genes that have introns yet are not alternatively spliced. It is argued that 5' and 3' splice sites evolved from common ancestral splice sites, referred to as proto-splice sites, that were bidirectional and had a core consensus sequence of C or A, A, G, R, which remains today as the immediate flanking sequence of most introns. The ancestral splicing machinery, although inefficient, would have been capable of generating vast mRNA diversity by splicing between proto-splice sites. Natural selection would be expected to have preserved mutations that increased the amounts of advantageously spliced mRNA. It is argued that this process drove the evolution of present 5' and 3' splice sites from a subset of proto-splice sites and also drove the evolution of a more efficient splicing machinery. The positions of most introns that evolved directly from the coding sequence would be expected to correlate with protein structure.     

Formation of new genes explains lower intron density in mammalian Rhodopsin G protein-coupled receptors

Mammalian G protein-coupled receptor (GPCR) genes are characterised by a large proportion of intronless genes or a lower density of introns when compared with GPCRs of invertebrates. It is unclear which mechanisms have influenced intron density in this protein family, which is one of the largest in the mammalian genomes. We used a combination of Hidden Markov Models (HMM) and BLAST searches to establish the comprehensive repertoire of Rhodopsin GPCRs from seven species and performed overall alignments and phylogenetic analysis using the maximum parsimony method for over 1400 receptors in 12 subgroups. We identified 14 different Ancestral Receptor Groups (ARGs) that have members in both vertebrate and invertebrate species. We found that there exists a remarkable difference in the intron density among ancestral and new Rhodopsin GPCRs. The intron density among ARGs members was more than 3.5-fold higher than that within non-ARG members and more than 2-fold higher when considering only the 7TM region. This suggests that the new GPCR genes have been predominantly formed intronless while the ancestral receptors likely accumulated introns during their evolution. Many of the intron positions found in mammalian ARG receptor sequences were found to be present in orthologue invertebrate receptors suggesting that these intron positions are ancient. This analysis also revealed that one intron position is much more frequent than any other position and it is common for a number of phylogenetically different Rhodopsin GPCR groups. This intron position lies within a functionally important, conserved, DRY motif which may form a proto-splice site that could contribute to positional intron insertion. Moreover, we have found that other receptor motifs, similar to DRY, also contain introns between the second and third nucleotide of the arginine codon which also forms a proto-splice site. Our analysis presents compelling evidence that there was not a major loss of introns in mammalian GPCRs and formation of new GPCRs among mammals explains why these have fewer introns compared to invertebrate GPCRs. We also discuss and speculate about the possible role of different RNA- and DNA-based mechanisms of intron insertion and loss.     

An analysis of intron positions in relation to nucleotides, amino acids, and protein secondary structure

We present an analysis of intron positions in relation to nucleotides, amino acid residues, and protein secondary structure. Previous work has shown that intron sites in proteins are not randomly distributed with respect to secondary structures. Here we show that this preference can be almost totally explained by the nucleotide bias of splice site machinery, and may well not relate to protein stability or conformation at all. Each intron phase is preferentially associated with its own set of residues: phase 0 introns with lysine, glutamine, and glutamic acid before the intron, and valine after; phase 1 introns with glycine, alanine, valine, aspartic acid, and glutamic acid; and phase 2 introns with arginine, serine, lysine, and tryptophan. These preferences can be explained principally on the basis of nucleotide bias at intron locations, which is in accordance with previous literature. Although this work does not prove that introns are inserted into genomes at specific proto-splice sites, it shows that the nucleotide bias surrounding introns, however it originally occurred, explains the observed correlations between introns and protein secondary structure.     

Evidence that introns arose at proto-splice sites.

The unexpected discovery of introns raised many questions about gene evolution. We provide evidence that actin and tubulin introns were gained between the G and R of the conserved coding sequence C/AAGR that is known to flank introns in general and which we call a proto-splice site. We conclude that the tubulin and actin introns are less ancient than the coding sequence and so could not have been involved in the primary evolution of the tubulin and actin genes.     

Analysis of ribosomal protein gene structures: implications for intron evolution

Many spliceosomal introns exist in the eukaryotic nuclear genome. Despite much research, the evolution of spliceosomal introns remains poorly understood. In this paper, we tried to gain insights into intron evolution from a novel perspective by comparing the gene structures of cytoplasmic ribosomal proteins (CRPs) and mitochondrial ribosomal proteins (MRPs), which are held to be of archaeal and bacterial origin, respectively. We analyzed 25 homologous pairs of CRP and MRP genes that together had a total of 527 intron positions. We found that all 12 of the intron positions shared by CRP and MRP genes resulted from parallel intron gains and none could be considered to be “conserved,” i.e., descendants of the same ancestor. This was supported further by the high frequency of proto-splice sites at these shared positions; proto-splice sites are proposed to be sites for intron insertion. Although we could not definitively disprove that spliceosomal introns were already present in the last universal common ancestor, our results lend more support to the idea that introns were gained late. At least, our results show that MRP genes were intronless at the time of endosymbiosis. The parallel intron gains between CRP and MRP genes accounted for 2.3% of total intron positions, which should provide a reliable estimate for future inferences of intron evolution.     

Spliceosomal intron insertions in genome compacted ray-finned fishes as evident from phylogeny of MC receptors, also supported by a few other GPCRs

Background

Insertions of spliceosomal introns are very rare events during evolution of vertebrates and the mechanisms governing creation of novel intron(s) remain obscure. Largely, gene structures of melanocortin (MC) receptors are characterized by intron-less architecture. However, recently a few exceptions have been reported in some fishes. This warrants a systematic survey of MC receptors for understanding intron insertion events during vertebrate evolution.

Methodology/Principal Findings

We have compiled an extended list of MC receptors from different vertebrate genomes with variations in fishes. Notably, the closely linked MC2Rs and MC5Rs from a group of ray-finned fishes have three and one intron insertion(s), respectively, with conserved positions and intron phase. In both genes, one novel insertion was in the highly conserved DRY motif at the end of helix TM3. Further, the proto-splice site MAG↑R is maintained at intron insertion sites in these two genes. However, the orthologs of these receptors from zebrafish and tetrapods are intron-less, suggesting these introns are simultaneously created in selected fishes. Surprisingly, these novel introns are traceable only in four fish genomes. We found that these fish genomes are severely compacted after the separation from zebrafish. Furthermore, we also report novel intron insertions in P2Y receptors and in CHRM3. Finally, we report ultrasmall introns in MC2R genes from selected fishes.

Conclusions/Significance

The current repository of MC receptors illustrates that fishes have no MC3R ortholog. MC2R, MC5R, P2Y receptors and CHRM3 have novel intron insertions only in ray-finned fishes that underwent genome compaction. These receptors share one intron at an identical position suggestive of being inserted contemporaneously. In addition to repetitive elements, genome compaction is now believed to be a new hallmark that promotes intron insertions, as it requires rapid DNA breakage and subsequent repair processes to gain back normal functionality.     

Widespread occurrence of spliceosomal introns in the rDNA genes of ascomycetes

Spliceosomal (pre-mRNA) introns have previously been found in eukaryotic protein-coding genes, in the small nuclear RNAs of some fungi, and in the small- and large-subunit ribosomal DNA genes of a limited number of ascomycetes. How the majority of these introns originate remains an open question because few proven cases of recent and pervasive intron origin have been documented. We report here the widespread occurrence of spliceosomal introns (69 introns at 27 different sites) in the small- and large-subunit nuclear-encoded rDNA of lichen-forming and free-living members of the Ascomycota. Our analyses suggest that these spliceosomal introns are of relatively recent origin, i.e., within the Euascomycetes, and have arisen through aberrant reverse-splicing (in trans) of free pre-mRNA introns into rRNAs. The spliceosome itself, and not an external agent (e.g., transposable elements, group II introns), may have given rise to these introns. A nonrandom sequence pattern was found at sites flanking the rRNA spliceosomal introns. This pattern (AG-intron-G) closely resembles the proto-splice site (MAG-intron-R) postulated for intron insertions in pre-mRNA genes. The clustered positions of spliceosomal introns on secondary structures suggest that particular rRNA regions are preferred sites for insertion through reverse-splicing.     

Evidence of splice signal migration from exon to intron during intron evolution

A comparison of the nucleotide sequences around the splice junctions that flank old (shared by two or more major lineages of eukaryotes) and new (lineage-specific) introns in eukaryotic genes reveals substantial differences in the distribution of information between introns and exons. Old introns have a lower information content in the exon regions adjacent to the splice sites than new introns but have a corresponding higher information content in the intron itself. This suggests that introns insert into nonrandom (proto-splice) sites but, during the evolution of an intron after insertion, the splice signal shifts from the flanking exon regions to the ends of the intron itself. Accumulation of information inside the intron during evolution suggests that new introns largely emerge de novo rather than through propagation and migration of old introns.     

Can Codon Usage Bias Explain Intron Phase Distributions and Exon Symmetry?

More introns exist between codons (phase 0) than between the first and the second bases (phase 1) or between the second and the third base (phase 2) within the codon. Many explanations have been suggested for this excess of phase 0. It has, for example, been argued to reflect an ancient utility for introns in separating exons that code for separate protein modules. There may, however, be a simple, alternative explanation. Introns typically require, for correct splicing, particular nucleotides immediately 5 in exons (typically a G) and immediately 3 in the following exon (also often a G). Introns therefore tend to be found between particular nucleotide pairs (e.g., G|G pairs) in the coding sequence. If, owing to bias in usage of different codons, these pairs are especially common at phase 0, then intron phase biases may have a trivial explanation. Here we take codon usage frequencies for a variety of eukaryotes and use these to generate random sequences. We then ask about the phase of putative intron insertion sites. Importantly, in all simulated data sets intron phase distribution is biased in favor of phase 0. In many cases the bias is of the magnitude observed in real data and can be attributed to codon usage bias. It is also known that exons may carry either the same phase (symmetric) or different phases (asymmetric) at the opposite ends. We simulated a distribution of different types of exons using frequencies of introns observed in real genes assuming random combination of intron phases at the opposite sides of exons. Surprisingly the simulated pattern was quite similar to that observed. In the simulants we typically observe a prevalence of symmetric exons carrying phase 0 at both ends, which is common for eukaryotic genes. However, at least in some species, the extent of the bias in favor of symmetric (0,0) exons is not as great in simulants as in real genes. These results emphasize the need to construct a biologically relevant null model of successful intron insertion.Reviewing Editor: Dr. Manyuan Long     

Does Drive Toward Canonic Exonic Splicing Sites Exist in Mammals?

About 2/3 of introns are inserted between G and G/A, which has previously been explained by codon usage frequencies existing during the period of intron insertions. However, less is known about the evolution of exonic splicing sites. Exonic nucleotides that frame introns are involved in both protein coding and splicing. While a compromise between protein coding and splicing constraints is achieved differently in each intron phase, AG|G is the most common site in all phases comprising about one quarter of all such sites. There is also a great variety of other splicing sites. Here we examine evolutionary changes in exonic nucleotides located at positions −2 −1|+1 which occurred after the beginning of eutherian radiation using comparisons of orthologous splicing sites from five mammalian species. AG|G accumulated fewer substitutions and was more conservative than less frequent exonic splicing sites. Such trend could potentially increase frequencies of AG|G during mammalian evolution and cause a decline of less common sites which had higher substitution rates. However, there is a limit to this process determined by the dynamic equilibrium of substitution rates and the frequencies of different splicing sites. It seems that this equilibrium was already achieved at the time of eutherian radiation and a moderate increase in AG|G frequency was observed only in the human genome.     

USING RDNA GENES TO UNDERSTAND THE ORIGIN AND EVOLUTION OF SPLICEOSOMAL AND GROUP I INTRONS

Ribosomal DNA genes in lichen algae and lichen fungi are astonishingly rich in spliceosomal and group I introns. We use phylogenetic, secondary structure, and biochemical analyses to understand the evolution of these introns. Despite the widespread distribution of spliceosomal introns in nuclear pre-mRNA genes, their general mechanism of origin remains an open question because few proven cases of recent and pervasive intron origin have been documented. The lichen introns are valuable in this respect because they are undoubtedly of a "recent" origin and limited to the Euascomycetes. Our analyses suggest that rDNA spliceosomal introns have arisen through aberrant reverse-splicing (in trans) of free pre-mRNA introns into r RNAs. We propose that the spliceosome itself (and not an external agent; e.g. transposable elements, group II introns) has given rise to the introns. The rDNA introns are found most often between the flanking sequence G (78%) - intron-G (72%), and their clustered positions on secondary structures suggest that particular r RNA regions are preferred sites (i.e., proto-splice sites) for insertion. Mapping of intron positions on the newly available tertiary structures show that they are found most often in exposed regions of the ribosomes. This again is consistent with an intron origin through reverse-splicing. Remarkably, the distribution and phylogenetic relationships of most group I introns in nuclear rDNA genes are also consistent with a reverse-splicing origin. These data underline the value of lichens as a model system for understanding intron origin and stress the importance of RNA-level processes in the spread of these sequences in nuclear coding regions.     

Detailed analysis of base preferences at the cleavage site of a trans-acting HDV ribozyme: a mutation that changes cleavage site specificity.

In our previous attempt at in vitro selection of a trans - acting human hepatitis delta virus (HDV) ribozyme, we found that one of the variants, G10-68-725G, cleaved a 13 nt substrate, HDVS1, at two sites [Nishikawa,F., Kawakami,J., Chiba,A., Shirai,M., Kumar,P.K.R. and Nishikawa,S. (1996) Eur. J. Biochem., 237, 712-718]. One site was the normal cleavage site and the other site was shifted 1 nt toward the 3'-end. To clarify the interactions between nucleotides around the cleavage site of the trans -acting HDV ribozyme, we analyzed the efficiency of the reaction for every possible base pair between the substrate and the ribozyme at positions -1 (-1N:726N) and +1 (+1N:725N) relative to the cleavage site using the genomic HDV ribozyme, TdS4(Xho), and derivatives of the most active variant, G10-68. These mutagenesis analyses revealed that the +1 base of the substrate affects the structure of the catalytic core in the complex with G10-68-725G, substrate and divalent metal ions, and it shifts the cleavage site. In a comparison with other variants of the trans -acting HDV ribozyme, we found that this cleavage site shift occurred only with G10-68-725G.     

Intron mobility in phage T4 is dependent upon a distinctive class of endonucleases and independent of DNA sequences encoding the intron core: mechanistic and evolutionary implications

Although mobility of the phylogenetically widespread group I introns appears to be mechanistically similar, the phage T4 intron-encoded endonucleases that promote mobility of the td and sunY introns are different from their eukaryotic counterparts. Most notably, they cleave at a distance from the intron insertion sites. The td enzyme was shown to cleave 23-26 nt 5' and the sunY endonuclease 13-15 nt 3' to the intron insertion site to generate 3-nt or 2-nt 3'-OH extensions, respectively. The absolute coconversion of exon markers between the distant cleavage and insertion sites is consistent with the double-strand-break repair model for intron mobility. As a further critical test of the model we have demonstrated that the mobility event is independent of DNA sequences that encode the catalytic intron core structure. Thus, in derivatives in which the lacZ or kanR coding sequences replace the intron, these marker genes are efficiently inserted into intron-minus alleles when the cognate endonuclease is provided in trans. The process is therefore endonuclease-dependent, rather than dependent on the intron per se. These findings, which imply that the endonucleases rather than the introns themselves were the primordial mobile elements, are incorporated into a model for the evolution of mobile introns.     

The mechanisms controlling ribosomal protein L1 pre-mRNA splicing are maintained in evolution and rely on conserved intron sequences.

Sequences corresponding to the third intron of the X.laevis L1 ribosomal protein gene were isolated from the second copy of the X.laevis gene and from the single copy of X.tropicalis. Sequence comparison revealed that the three introns share an unusual sequence conservation which spans a region of 110 nucleotides. In addition, they have the same suboptimal 5' splice sites. The three introns show similar features upon oocyte microinjection: they have very low splicing efficiency and undergo the same site specific cleavages which lead to the accumulation of truncated molecules. Computer analysis and RNAse digestions have allowed to assign to the conserved region a specific secondary structure. Mutational analysis has shown that this structure is important for conferring the cleavage phenotype to these three introns. Competition experiments show that the cleavage phenotype can be prevented by coinjection of excess amounts of homologous sequences.     

Site selection by Xenopus laevis RNAase P

Investigation of the mechanism of cleavage site selection by Xenopus RNAase P reveals that the acceptor stem, a 7 bp helix common to all tRNA precursors, is required for cleavage. We propose that Xenopus RNAase P recognizes conserved features of the mature tRNA and that the cleavage site is selected by measuring the length of the acceptor stem. In support of this, we demonstrate that insertion of 2 bp in the acceptor stem of yeast pre-tRNA(3Leu) relocates the cleavage site 2 bases 3' to the original one. In addition, insertion of 1 bp in the acceptor stem of the end-matured yeast pre-tRNA(Phe) generates an RNAase P cleavage site: the enzyme produces a mature tRNA with the characteristic 7 bp stem and releases one 5' flanking nucleotide. Since it has previously been shown that cleavage sites of the splicing endonuclease are determined by the length of the anticodon stem, RNAase P and the splicing endonuclease apparently use different stems to determine their cutting sites.     

The genome-wide DNA sequence specificity of the anti-tumour drug bleomycin in human cells

The cancer chemotherapeutic agent, bleomycin, cleaves DNA at specific sites. For the first time, the genome-wide DNA sequence specificity of bleomycin breakage was determined in human cells. Utilising Illumina next-generation DNA sequencing techniques, over 200 million bleomycin cleavage sites were examined to elucidate the bleomycin genome-wide DNA selectivity. The genome-wide bleomycin cleavage data were analysed by four different methods to determine the cellular DNA sequence specificity of bleomycin strand breakage. For the most highly cleaved DNA sequences, the preferred site of bleomycin breakage was at 5′-GT* dinucleotide sequences (where the asterisk indicates the bleomycin cleavage site), with lesser cleavage at 5′-GC* dinucleotides. This investigation also determined longer bleomycin cleavage sequences, with preferred cleavage at 5′-GT*A and 5′- TGT* trinucleotide sequences, and 5′-TGT*A tetranucleotides. For cellular DNA, the hexanucleotide DNA sequence 5′-RTGT*AY (where R is a purine and Y is a pyrimidine) was the most highly cleaved DNA sequence. It was striking that alternating purine–pyrimidine sequences were highly cleaved by bleomycin. The highest intensity cleavage sites in cellular and purified DNA were very similar although there were some minor differences. Statistical nucleotide frequency analysis indicated a G nucleotide was present at the ?3 position (relative to the cleavage site) in cellular DNA but was absent in purified DNA.     

The site-specific DNA endonuclease encoded by a group I intron in the Chlamydomonas pallidostigmatica chloroplast small subunit rRNA gene introduces a single-strand break at low concentrations of Mg2+.

Two group I introns (CpSSU.1 and CpSSU.2) that each potentially encode a protein with two copies of the LAGLI-DADG motif were identified in the Chlamydomonas pallidostigmatica chloroplast small subunit rRNA gene. They both belong to subgroup IA3 and represent novel insertion positions in this gene (sites 508 and 793 in the Escherichia coli 16S rRNA). The proteins encoded by the two introns were synthesized in vitro and tested for their ability to cleave the homing site of their respective introns. Only the CpSSU.1-encoded protein (I-CpaII) was found to display specific DNA endonuclease activity. At 0.1 mM MgCl2, I-CpaII nicks only the bottom (transcribed) DNA strand, but at concentrations ranging from 0.5 to 5.0 mM, it cleaves both DNA strands (leaving a 4 nucleotide single-stranded extension with 3'-OH overhangs) while preferentially nicking the bottom strand. The rate of cleavage of the top strand increases with increasing concentration of MgCl2. The preliminary data derived from these endonuclease assays suggest that the mode of DNA cleavage by I-CpaII is directed by the availability of Mg2+ and the affinity of different binding sites for this cation.