Both Maintenance and Avoidance of RNA-Binding Protein Interactions Constrain Coding Sequence Evolution

Selection on synonymous sites: the unwanted transcript hypothesis

Although translational selection to favour codons that match the most abundant tRNAs is not readily observed in humans, there is nonetheless selection in humans on synonymous mutations. We hypothesize that much of this synonymous site selection can be explained in terms of protection against unwanted RNAs - spurious transcripts, mis-spliced forms or RNAs derived from transposable elements or viruses. We propose not only that selection on synonymous sites functions to reduce the rate of creation of unwanted transcripts (for example, through selection on exonic splice enhancers and cryptic splice sites) but also that high-GC content (but low-CpG content), together with intron presence and position, is both particular to functional native mRNAs and used to recognize transcripts as native. In support of this hypothesis, transcription, nuclear export, liquid phase condensation and RNA degradation have all recently been shown to promote GC-rich transcripts and suppress AU/CpG-rich ones. With such 'traps' being set against AU/CpG-rich transcripts, the codon usage of native genes has, in turn, evolved to avoid such suppression. That parallel filters against AU/CpG-rich transcripts also affect the endosomal import of RNAs further supports the unwanted transcript hypothesis of synonymous site selection and explains the similar design rules that have enabled the successful use of transgenes and RNA vaccines.

Molecular characteristics and clinical implications of serine/arginine-rich splicing factors in human cancer

As critical splicing regulators, serine/arginine-rich splicing factors (SRSFs) play pivotal roles in carcinogenesis. As dysregulation of SRSFs may confer potential cancer risks, targeting SRSFs could provide important insights into cancer therapy. However, a global and comprehensive pattern to elaborate the molecular characteristics, mechanisms, and clinical links of SRSFs in a wide variety of human cancer is still lacking. In this study, a systematic analysis was conducted to reveal the molecular characteristics and clinical implications of SRSFs covering more than 10000 tumour samples of 33 human cancer types. We found that SRSFs experienced prevalent genomic alterations and expression perturbations in multiple cancer types. The DNA methylation, m6A modification, and miRNA regulation of SRSFs were all cancer context-dependent. Importantly, we found that SRSFs were strongly associated with cancer immunity, and were capable of predicting response to immunotherapy. And SRSFs had colossal potential for predicting survival in multiple cancer types, including those that have received immunotherapy. Moreover, we also found that SRSFs could indicate the drug sensitivity of targeted therapy and chemotherapy. Our research highlights the significance of SRSFs in cancer occurrence and development, and provides sufficient resources for understanding the biological characteristics of SRSFs, offering a new and unique perspective for developing cancer therapeutic strategies.

Protein Binding to Cis-Motifs in mRNAs Coding Sequence Is Common and Regulates Transcript Stability and the Rate of Translation

Protein binding to the non-coding regions of mRNAs is relatively well characterized and its functionality has been described in many examples. New results obtained by high-throughput methods indicate that binding to the coding sequence (CDS) by RNA-binding proteins is also quite common, but the functions thereof are more obscure. As described in this review, CDS binding has a role in the regulation of mRNA stability, but it has also a more intriguing role in the regulation of translational efficiency. Global approaches, which suggest the significance of CDS binding along with specific examples of CDS-binding RBPs and their modes of action, are outlined here, pointing to the existence of a relatively less-known regulatory network controlling mRNA stability and translation on yet another level.

Read between the lines: Diversity of non-translational selection pressures on local codon usage

Protein coding genes can contain specific motifs within their nucleotide sequence that function as a signal for various biological pathways. The presence of such sequence motifs within a gene can have beneficial or detrimental effects on the phenotype and fitness of an organism, and this can lead to the enrichment or avoidance of this sequence motif. The degeneracy of the genetic code allows for the existence of alternative synonymous sequences that exclude or include these motifs, while keeping the encoded amino acid sequence intact. This implies that locally, there can be a selective pressure for preferentially using a codon over its synonymous alternative in order to avoid or enrich a specific sequence motif. This selective pressure could -in addition to mutation, drift and selection for translation efficiency and accuracy- contribute to shape the codon usage bias. In this review, we discuss patterns of avoidance of (or enrichment for) the various biological signals contained in specific nucleotide sequence motifs: transcription and translation initiation and termination signals, mRNA maturation signals, and antiviral immune system targets. Experimental data on the phenotypic or fitness effects of synonymous mutations in these sequence motifs confirm that they can be targets of local selection pressures on codon usage. We also formulate the hypothesis that transposable elements could have a similar impact on codon usage through their preferred integration sequences. Overall, selection on codon usage appears to be a combination of a global selection pressure imposed by the translation machinery, and a patchwork of local selection pressures related to biological signals contained in specific sequence motifs.

Synonymous variants that disrupt messenger RNA structure are significantly constrained in the human population

BACKGROUND

The role of synonymous single-nucleotide variants in human health and disease is poorly understood, yet evidence suggests that this class of "silent" genetic variation plays multiple regulatory roles in both transcription and translation. One mechanism by which synonymous codons direct and modulate the translational process is through alteration of the elaborate structure formed by single-stranded mRNA molecules. While tools to computationally predict the effect of non-synonymous variants on protein structure are plentiful, analogous tools to systematically assess how synonymous variants might disrupt mRNA structure are lacking.

RESULTS

We developed novel software using a parallel processing framework for large-scale generation of secondary RNA structures and folding statistics for the transcriptome of any species. Focusing our analysis on the human transcriptome, we calculated 5 billion RNA-folding statistics for 469 million single-nucleotide variants in 45,800 transcripts. By considering the impact of all possible synonymous variants globally, we discover that synonymous variants predicted to disrupt mRNA structure have significantly lower rates of incidence in the human population.

CONCLUSIONS

These findings support the hypothesis that synonymous variants may play a role in genetic disorders due to their effects on mRNA structure. To evaluate the potential pathogenic impact of synonymous variants, we provide RNA stability, edge distance, and diversity metrics for every nucleotide in the human transcriptome and introduce a "Structural Predictivity Index" (SPI) to quantify structural constraint operating on any synonymous variant. Because no single RNA-folding metric can capture the diversity of mechanisms by which a variant could alter secondary mRNA structure, we generated a SUmmarized RNA Folding (SURF) metric to provide a single measurement to predict the impact of secondary structure altering variants in human genetic studies.

Evidence for Strong Mutation Bias toward, and Selection against, U Content in SARS-CoV-2: Implications for Vaccine Design

Large-scale re-engineering of synonymous sites is a promising strategy to generate vaccines either through synthesis of attenuated viruses or via codon-optimized genes in DNA vaccines. Attenuation typically relies on deoptimization of codon pairs and maximization of CpG dinucleotide frequencies. So as to formulate evolutionarily informed attenuation strategies that aim to force nucleotide usage against the direction favored by selection, here, we examine available whole-genome sequences of SARS-CoV-2 to infer patterns of mutation and selection on synonymous sites. Analysis of mutational profiles indicates a strong mutation bias toward U. In turn, analysis of observed synonymous site composition implicates selection against U. Accounting for dinucleotide effects reinforces this conclusion, observed UU content being a quarter of that expected under neutrality. Possible mechanisms of selection against U mutations include selection for higher expression, for high mRNA stability or lower immunogenicity of viral genes. Consistent with gene-specific selection against CpG dinucleotides, we observe systematic differences of CpG content between SARS-CoV-2 genes. We propose an evolutionarily informed approach to attenuation that, unusually, seeks to increase usage of the already most common synonymous codons. Comparable analysis of H1N1 and Ebola finds that GC3 deviated from neutral equilibrium is not a universal feature, cautioning against generalization of results.

Pervasive Selection against MicroRNA Target Sites in Human Populations

MicroRNA target sites are often conserved during evolution and purifying selection to maintain such sites is expected. On the other hand, comparative analyses identified a paucity of microRNA target sites in coexpressed transcripts, and novel target sites can potentially be deleterious. We proposed that selection against novel target sites pervasive. The analysis of derived allele frequencies revealed that, when the derived allele is a target site, the proportion of nontarget sites is higher than expected, particularly for highly expressed microRNAs. Thus, new alleles generating novel microRNA target sites can be deleterious and selected against. When we analyzed ancestral target sites, the derived (nontarget) allele frequency does not show statistical support for microRNA target allele conservation. We investigated the joint effects of microRNA conservation and expression and found that selection against microRNA target sites depends mostly on the expression level of the microRNA. We identified microRNA target sites with relatively high levels of population differentiation. However, when we analyze separately target sites in which the target allele is ancestral to the population, the proportion of single-nucleotide polymorphisms with high F_st significantly increases. These findings support that population differentiation is more likely in target sites that are lost than in the gain of new target sites. Our results indicate that selection against novel microRNA target sites is prevalent and, although individual sites may have a weak selective pressure, the overall effect across untranslated regions is not negligible and should be accounted when studying the evolution of genomic sequences.

A Depletion of Stop Codons in lincRNA is Owing to Transfer of Selective Constraint from Coding Sequences

Although the constraints on a gene’s sequence are often assumed to reflect the functioning of that gene, here we propose transfer selection, a constraint operating on one class of genes transferred to another, mediated by shared binding factors. We show that such transfer can explain an otherwise paradoxical depletion of stop codons in long intergenic noncoding RNAs (lincRNAs). Serine/arginine-rich proteins direct the splicing machinery by binding exonic splice enhancers (ESEs) in immature mRNA. As coding exons cannot contain stop codons in one reading frame, stop codons should be rare within ESEs. We confirm that the stop codon density (SCD) in ESE motifs is low, even accounting for nucleotide biases. Given that serine/arginine-rich proteins binding ESEs also facilitate lincRNA splicing, a low SCD could transfer to lincRNAs. As predicted, multiexon lincRNA exons are depleted in stop codons, a result not explained by open reading frame (ORF) contamination. Consistent with transfer selection, stop codon depletion in lincRNAs is most acute in exonic regions with the highest ESE density, disappears when ESEs are masked, is consistent with stop codon usage skews in ESEs, and is diminished in both single-exon lincRNAs and introns. Owing to low SCD, the maximum lengths of pseudo-ORFs frequently exceed null expectations. This has implications for ORF annotation and the evolution of de novo protein-coding genes from lincRNAs. We conclude that not all constraints operating on genes need be explained by the functioning of the gene but may instead be transferred owing to shared binding factors.

Integrated structural and evolutionary analysis reveals common mechanisms underlying adaptive evolution in mammals

Understanding the molecular basis of adaptation to the environment is a central question in evolutionary biology, yet linking detected signatures of positive selection to molecular mechanisms remains challenging. Here we demonstrate that combining sequence-based phylogenetic methods with structural information assists in making such mechanistic interpretations on a genomic scale. Our integrative analysis shows that positively selected sites tend to colocalize on protein structures and that positively selected clusters are found in functionally important regions of proteins, indicating that positive selection can contravene the well-known principle of evolutionary conservation of functionally important regions. This unexpected finding, along with our discovery that positive selection acts on structural clusters, opens previously unexplored strategies for the development of better models of protein evolution. Remarkably, proteins where we detect the strongest evidence of clustering belong to just two functional groups: Components of immune response and metabolic enzymes. This gives a coherent picture of pathogens and xenobiotics as important drivers of adaptive evolution of mammals.

Interplay between coding and exonic splicing regulatory sequences

The inclusion of exons during the splicing process depends on the binding of splicing factors to short low-complexity regulatory sequences. The relationship between exonic splicing regulatory sequences and coding sequences is still poorly understood. We demonstrate that exons that are coregulated by any given splicing factor share a similar nucleotide composition bias and preferentially code for amino acids with similar physicochemical properties because of the nonrandomness of the genetic code. Indeed, amino acids sharing similar physicochemical properties correspond to codons that have the same nucleotide composition bias. In particular, we uncover that the TRA2A and TRA2B splicing factors that bind to adenine-rich motifs promote the inclusion of adenine-rich exons coding preferentially for hydrophilic amino acids that correspond to adenine-rich codons. SRSF2 that binds guanine/cytosine-rich motifs promotes the inclusion of GC-rich exons coding preferentially for small amino acids, whereas SRSF3 that binds cytosine-rich motifs promotes the inclusion of exons coding preferentially for uncharged amino acids, like serine and threonine that can be phosphorylated. Finally, coregulated exons encoding amino acids with similar physicochemical properties correspond to specific protein features. In conclusion, the regulation of an exon by a splicing factor that relies on the affinity of this factor for specific nucleotide(s) is tightly interconnected with the exon-encoded physicochemical properties. We therefore uncover an unanticipated bidirectional interplay between the splicing regulatory process and its biological functional outcome.

Post-transcriptional regulatory patterns revealed by protein-RNA interactions

The coordination of the synthesis of functionally-related proteins can be achieved at the post-transcriptional level by the action of common regulatory molecules, such as RNA–binding proteins (RBPs). Despite advances in the genome-wide identification of RBPs and their binding transcripts, the protein–RNA interaction space is still largely unexplored, thus hindering a broader understanding of the extent of the post-transcriptional regulation of related coding RNAs. Here, we propose a computational approach that combines protein–mRNA interaction networks and statistical analyses to provide an inferred regulatory landscape for more than 800 human RBPs and identify the cellular processes that can be regulated at the post-transcriptional level. We show that 10% of the tested sets of functionally-related mRNAs can be post-transcriptionally regulated. Moreover, we propose a classification of (i) the RBPs and (ii) the functionally-related mRNAs, based on their distinct behaviors in the functional landscape, hinting towards mechanistic regulatory hypotheses. In addition, we demonstrate the usefulness of the inferred functional landscape to investigate the cellular role of both well-characterized and novel RBPs in the context of human diseases.

Refining the Ambush Hypothesis: Evidence That GC- and AT-Rich Bacteria Employ Different Frameshift Defence Strategies

Stop codons are frequently selected for beyond their regular termination function for error control. The “ambush hypothesis” proposes out-of-frame stop codons (OSCs) terminating frameshifted translations are selected for. Although early indirect evidence was partially supportive, recent evidence suggests OSC frequencies are not exceptional when considering underlying nucleotide content. However, prior null tests fail to control amino acid/codon usages or possible local mutational biases. We therefore return to the issue using bacterial genomes, considering several tests defining and testing against a null. We employ simulation approaches preserving amino acid order but shuffling synonymous codons or preserving codons while shuffling amino acid order. Additionally, we compare codon usage in amino acid pairs, where one codon can but the next, otherwise identical codon, cannot encode an OSC. OSC frequencies exceed expectations typically in AT-rich genomes, the +1 frame and for TGA/TAA but not TAG. With this complex evidence, simply rejecting or accepting the ambush hypothesis is not warranted. We propose a refined post hoc model, whereby AT-rich genomes have more accidental frameshifts, handled by RF2–RF3 complexes (associated with TGA/TAA) and are mostly +1 (or −2) slips. Supporting this, excesses positively correlate with in silico predicted frameshift probabilities. Thus, we propose a more viable framework, whereby genomes broadly adopt one of the two strategies to combat frameshifts: preventing frameshifting (GC-rich) or permitting frameshifts but minimizing impacts when most are caught early (AT-rich). Our refined framework holds promise yet some features, such as the bias of out-of-frame sense codons, remain unexplained.

Exonic splice regulation imposes strong selection at synonymous sites

What proportion of coding sequence nucleotides have roles in splicing, and how strong is the selection that maintains them? Despite a large body of research into exonic splice regulatory signals, these questions have not been answered. This is because, to our knowledge, previous investigations have not explicitly disentangled the frequency of splice regulatory elements from the strength of the evolutionary constraint under which they evolve. Current data are consistent both with a scenario of weak and diffuse constraint, enveloping large swaths of sequence, as well as with well-defined pockets of strong purifying selection. In the former case, natural selection on exonic splice enhancers (ESEs) might primarily act as a slight modifier of codon usage bias. In the latter, mutations that disrupt ESEs are likely to have large fitness and, potentially, clinical effects. To distinguish between these scenarios, we used several different methods to determine the distribution of selection coefficients for new mutations within ESEs. The analyses converged to suggest that ∼15%-20% of fourfold degenerate sites are part of functional ESEs. Most of these sites are under strong evolutionary constraint. Therefore, exonic splice regulation does not simply impose a weak bias that gently nudges coding sequence evolution in a particular direction. Rather, the selection to preserve these motifs is a strong force that severely constrains the evolution of a substantial proportion of coding nucleotides. Thus synonymous mutations that disrupt ESEs should be considered as a potentially common cause of single-locus genetic disorders.

Adenine Enrichment at the Fourth CDS Residue in Bacterial Genes Is Consistent with Error Proofing for +1 Frameshifts

Beyond selection for optimal protein functioning, coding sequences (CDSs) are under selection at the RNA and DNA levels. Here, we identify a possible signature of “dual-coding,” namely extensive adenine (A) enrichment at bacterial CDS fourth sites. In 99.07% of studied bacterial genomes, fourth site A use is greater than expected given genomic A-starting codon use. Arguing for nucleotide level selection, A-starting serine and arginine second codons are heavily utilized when compared with their non-A starting synonyms. Several models have the ability to explain some of this trend. In part, A-enrichment likely reduces 5′ mRNA stability, promoting translation initiation. However T/U, which may also reduce stability, is avoided. Further, +1 frameshifts on the initiating ATG encode a stop codon (TGA) provided A is the fourth residue, acting either as a frameshift “catch and destroy” or a frameshift stop and adjust mechanism and hence implicated in translation initiation. Consistent with both, genomes lacking TGA stop codons exhibit weaker fourth site A-enrichment. Sequences lacking a Shine–Dalgarno sequence and those without upstream leader genes, that may be more error prone during initiation, have greater utilization of A, again suggesting a role in initiation. The frameshift correction model is consistent with the notion that many genomic features are error-mitigation factors and provides the first evidence for site-specific out of frame stop codon selection. We conjecture that the NTG universal start codon may have evolved as a consequence of TGA being a stop codon and the ability of NTGA to rapidly terminate or adjust a ribosome.

Depletion of somatic mutations in splicing-associated sequences in cancer genomes

Background

An important goal of cancer genomics is to identify systematically cancer-causing mutations. A common approach is to identify sites with high ratios of non-synonymous to synonymous mutations; however, if synonymous mutations are under purifying selection, this methodology leads to identification of false-positive mutations. Here, using synonymous somatic mutations (SSMs) identified in over 4000 tumours across 15 different cancer types, we sought to test this assumption by focusing on coding regions required for splicing.

Results

Exon flanks, which are enriched for sequences required for splicing fidelity, have ~ 17% lower SSM density compared to exonic cores, even after excluding canonical splice sites. While it is impossible to eliminate a mutation bias of unknown cause, multiple lines of evidence support a purifying selection model above a mutational bias explanation. The flank/core difference is not explained by skewed nucleotide content, replication timing, nucleosome occupancy or deficiency in mismatch repair. The depletion is not seen in tumour suppressors, consistent with their role in positive tumour selection, but is otherwise observed in cancer-associated and non-cancer genes, both essential and non-essential. Consistent with a role in splicing modulation, exonic splice enhancers have a lower SSM density before and after controlling for nucleotide composition; moreover, flanks at the 5’ end of the exons have significantly lower SSM density than at the 3’ end.

Conclusions

These results suggest that the observable mutational spectrum of cancer genomes is not simply a product of various mutational processes and positive selection, but might also be shaped by negative selection.

Electronic supplementary material

The online version of this article (doi:10.1186/s13059-017-1337-5) contains supplementary material, which is available to authorized users.

ASXL gain-of-function truncation mutants: defective and dysregulated forms of a natural ribosomal frameshifting product?

BACKGROUND

Programmed ribosomal frameshifting (PRF) is a gene expression mechanism which enables the translation of two N-terminally coincident, C-terminally distinct protein products from a single mRNA. Many viruses utilize PRF to control or regulate gene expression, but very few phylogenetically conserved examples are known in vertebrate genes. Additional sex combs-like (ASXL) genes 1 and 2 encode important epigenetic and transcriptional regulatory proteins that control the expression of homeotic genes during key developmental stages. Here we describe an ~150-codon overlapping ORF (termed TF) in ASXL1 and ASXL2 that, with few exceptions, is conserved throughout vertebrates.

RESULTS

Conservation of the TF ORF, strong suppression of synonymous site variation in the overlap region, and the completely conserved presence of an EH[N/S]Y motif (a known binding site for Host Cell Factor-1, HCF-1, an epigenetic regulatory factor), all indicate that TF is a protein-coding sequence. A highly conserved UCC_UUU_CGU sequence (identical to the known site of +1 ribosomal frameshifting for influenza virus PA-X expression) occurs at the 5' end of the region of enhanced synonymous site conservation in ASXL1. Similarly, a highly conserved RG_GUC_UCU sequence (identical to a known site of -2 ribosomal frameshifting for arterivirus nsp2TF expression) occurs at the 5' end of the region of enhanced synonymous site conservation in ASXL2.

CONCLUSIONS

Due to a lack of appropriate splice forms, or initiation sites, the most plausible mechanism for translation of the ASXL1 and 2 TF regions is ribosomal frameshifting, resulting in a transframe fusion of the N-terminal half of ASXL1 or 2 to the TF product, termed ASXL-TF. Truncation or frameshift mutants of ASXL are linked to myeloid malignancies and genetic diseases, such as Bohring-Opitz syndrome, likely at least in part as a result of gain-of-function or dominant-negative effects. Our hypothesis now indicates that these disease-associated mutant forms represent overexpressed defective versions of ASXL-TF.

REVIEWERS

This article was reviewed by Laurence Hurst and Eugene Koonin.

Estimating the prevalence of functional exonic splice regulatory information

In addition to coding information, human exons contain sequences necessary for correct splicing. These elements are known to be under purifying selection and their disruption can cause disease. However, the density of functional exonic splicing information remains profoundly uncertain. Several groups have experimentally investigated how mutations at different exonic positions affect splicing. They have found splice information to be distributed widely in exons, with one estimate putting the proportion of splicing-relevant nucleotides at >90%. These results suggest that splicing could place a major pressure on exon evolution. However, analyses of sequence conservation have concluded that the need to preserve splice regulatory signals only slightly constrains exon evolution, with a resulting decrease in the average human rate of synonymous evolution of only 1–4%. Why do these two lines of research come to such different conclusions? Among other reasons, we suggest that the methods are measuring different things: one assays the density of sites that affect splicing, the other the density of sites whose effects on splicing are visible to selection. In addition, the experimental methods typically consider short exons, thereby enriching for nucleotides close to the splice junction, such sites being enriched for splice-control elements. By contrast, in part owing to correction for nucleotide composition biases and to the assumption that constraint only operates on exon ends, the conservation-based methods can be overly conservative.