Features generated for computational splice-site prediction correspond to functional elements

Brain cross-protection against SARS-CoV-2 variants by a lentiviral vaccine in new transgenic mice

COVID-19 vaccines already in use or in clinical development may have reduced efficacy against emerging SARS-CoV-2 variants. In addition, although the neurotropism of SARS-CoV-2 is well established, the vaccine strategies currently developed have not taken into account protection of the central nervous system. Here, we generated a transgenic mouse strain expressing the human angiotensin-converting enzyme 2, and displaying unprecedented brain permissiveness to SARS-CoV-2 replication, in addition to high permissiveness levels in the lung. Using this stringent transgenic model, we demonstrated that a non-integrative lentiviral vector, encoding for the spike glycoprotein of the ancestral SARS-CoV-2, used in intramuscular prime and intranasal boost elicits sterilizing protection of lung and brain against both the ancestral virus, and the Gamma (P.1) variant of concern, which carries multiple vaccine escape mutations. Beyond induction of strong neutralizing antibodies, the mechanism underlying this broad protection spectrum involves a robust protective T-cell immunity, unaffected by the recent mutations accumulated in the emerging SARS-CoV-2 variants.

Parallel assembly of actin and tropomyosin, but not myosin II, during de novo actin filament formation in live mice

Many actin filaments in animal cells are co-polymers of actin and tropomyosin. In many cases, non-muscle myosin II associates with these co-polymers to establish a contractile network. However, the temporal relationship of these three proteins in the de novo assembly of actin filaments is not known. Intravital subcellular microscopy of secretory granule exocytosis allows the visualisation and quantification of the formation of an actin scaffold in real time, with the added advantage that it occurs in a living mammal under physiological conditions. We used this model system to investigate the de novo assembly of actin, tropomyosin Tpm3.1 (a short isoform of TPM3) and myosin IIA (the form of non-muscle myosin II with its heavy chain encoded by Myh9) on secretory granules in mouse salivary glands. Blocking actin polymerization with cytochalasin D revealed that Tpm3.1 assembly is dependent on actin assembly. We used time-lapse imaging to determine the timing of the appearance of the actin filament reporter LifeAct-RFP and of Tpm3.1-mNeonGreen on secretory granules in LifeAct-RFP transgenic, Tpm3.1-mNeonGreen and myosin IIA-GFP (GFP-tagged MYH9) knock-in mice. Our findings are consistent with the addition of tropomyosin to actin filaments shortly after the initiation of actin filament nucleation, followed by myosin IIA recruitment.

Effective automated feature construction and selection for classification of biological sequences

BACKGROUND

Many open problems in bioinformatics involve elucidating underlying functional signals in biological sequences. DNA sequences, in particular, are characterized by rich architectures in which functional signals are increasingly found to combine local and distal interactions at the nucleotide level. Problems of interest include detection of regulatory regions, splice sites, exons, hypersensitive sites, and more. These problems naturally lend themselves to formulation as classification problems in machine learning. When classification is based on features extracted from the sequences under investigation, success is critically dependent on the chosen set of features.

METHODOLOGY

We present an algorithmic framework (EFFECT) for automated detection of functional signals in biological sequences. We focus here on classification problems involving DNA sequences which state-of-the-art work in machine learning shows to be challenging and involve complex combinations of local and distal features. EFFECT uses a two-stage process to first construct a set of candidate sequence-based features and then select a most effective subset for the classification task at hand. Both stages make heavy use of evolutionary algorithms to efficiently guide the search towards informative features capable of discriminating between sequences that contain a particular functional signal and those that do not.

RESULTS

To demonstrate its generality, EFFECT is applied to three separate problems of importance in DNA research: the recognition of hypersensitive sites, splice sites, and ALU sites. Comparisons with state-of-the-art algorithms show that the framework is both general and powerful. In addition, a detailed analysis of the constructed features shows that they contain valuable biological information about DNA architecture, allowing biologists and other researchers to directly inspect the features and potentially use the insights obtained to assist wet-laboratory studies on retainment or modification of a specific signal. Code, documentation, and all data for the applications presented here are provided for the community at http://www.cs.gmu.edu/~ashehu/?q=OurTools.

Investigation of dmyc Promoter and Regulatory Regions

Products of the myc gene family integrate extracellular signals by modulating a wide range of their targets involved in cellular biogenesis and metabolism; the purpose of this integration is to regulate cell death, proliferation, and differentiation. However, understanding the regulation of myc at the transcription level remains a challenge. We performed rapid amplification of dmyc cDNA ends (5' RACE) and mapped the transcription start site at P1 promoter, 18 base pairs upstream of the start of the known EST GM01143 and within the 5' UTR. Our data show that the first TATA box, previously computationally predicted, is utilized to generate dmyc full length mRNA. The largest transcript contains all three exons, generated after the removal of the introns by constitutively regulated splicing events. Further investigation of Downstream Promoter Element (DPE) was achieved by studying lacZ reporter activity; investigation revealed that this element and its upstream cluster of binding sites are required for the dmyc intron 2 activity. These findings may provide valuable tools for further analysis of dmyc cis-elements.

An evolutionary algorithm approach for feature generation from sequence data and its application to DNA splice site prediction

Associating functional information with biological sequences remains a challenge for machine learning methods. The performance of these methods often depends on deriving predictive features from the sequences sought to be classified. Feature generation is a difficult problem, as the connection between the sequence features and the sought property is not known a priori. It is often the task of domain experts or exhaustive feature enumeration techniques to generate a few features whose predictive power is then tested in the context of classification. This paper proposes an evolutionary algorithm to effectively explore a large feature space and generate predictive features from sequence data. The effectiveness of the algorithm is demonstrated on an important component of the gene-finding problem, DNA splice site prediction. This application is chosen due to the complexity of the features needed to obtain high classification accuracy and precision. Our results test the effectiveness of the obtained features in the context of classification by Support Vector Machines and show significant improvement in accuracy and precision over state-of-the-art approaches.

Natural rules for Arabidopsis thaliana pre-mRNA splicing site selection

The accurate prediction of plant pre-mRNA splicing sites has been studied extensively. The rules for plant pre-mRNA splicing still remain unknown. This study, based on confirmed sequence data, systematically analyzed all expressed genes on Arabidopsis thaliana chromosome IV to quantitatively explore the natural splicing rules. The results indicated that defining Arabidopsis thaliana pre-mRNA splicing sites required a combination of multiple factors including (1) relative conserved consensus sequence at splicing site; (2) individual nucleotide distribution pattern in 50 nucleotides up- and down-stream regions of splicing site; (3) quantitative analysis of individual nucleotide distribution by using the formulations concluded from this study. The combination of all these factors together can bring the accuracy of Arabidopsis thaliana splicing site recognition over 99%. The results provide additional information to the future of plant pre-mRNA splicing research.

A two-stage evolutionary approach for effective classification of hypersensitive DNA sequences

Hypersensitive (HS) sites in genomic sequences are reliable markers of DNA regulatory regions that control gene expression. Annotation of regulatory regions is important in understanding phenotypical differences among cells and diseases linked to pathologies in protein expression. Several computational techniques are devoted to mapping out regulatory regions in DNA by initially identifying HS sequences. Statistical learning techniques like Support Vector Machines (SVM), for instance, are employed to classify DNA sequences as HS or non-HS. This paper proposes a method to automate the basic steps in designing an SVM that improves the accuracy of such classification. The method proceeds in two stages and makes use of evolutionary algorithms. An evolutionary algorithm first designs optimal sequence motifs to associate explicit discriminating feature vectors with input DNA sequences. A second evolutionary algorithm then designs SVM kernel functions and parameters that optimally separate the HS and non-HS classes. Results show that this two-stage method significantly improves SVM classification accuracy. The method promises to be generally useful in automating the analysis of biological sequences, and we post its source code on our website.

Aberrant Single Exon Skipping is not Altered by Age in Exons of NF1, RABAC1, AATF or PCGF2 in Human Blood Cells and Fibroblasts

In human pre-mRNA splicing, infrequent errors occur resulting in erroneous splice products as shown in a genome-wide approach. One characteristic subgroup consists of products lacking one cassette exon. The noise in the splicing process, represented by those misspliced products, can be increased by cold shock treatment or by inhibiting the nonsense mediated decay. Here, we investigated whether the splicing noise frequency increases with age in vivo in peripheral bloods cells or in vitro in cultured and aged fibroblasts from healthy donors. Splicing noise frequency was measured for four erroneously skipped NF1 exons and one exon of RABAC1, AATF and PCGF2 by RT-qPCR. Measurements were validated in cultured fibroblasts treated with cold shock or puromycin. Intragenic but not interpersonal differences were detected in splicing noise frequencies in vivo in peripheral blood cells of 11 healthy donors (15 y–85 y) and in in vitro senescent fibroblasts from three further donors. No correlation to the age of the donors was found in the splicing noise frequencies. Our data demonstrates that splicing error frequencies are not altered by age in peripheral blood cells or in vitro aged fibroblasts in the tested exons of the four investigated genes, indicating a high importance of correct splicing in these proliferating aged cells.

Transposable elements in disease-associated cryptic exons

Transposable elements (TEs) make up a half of the human genome, but the extent of their contribution to cryptic exon activation that results in genetic disease is unknown. Here, a comprehensive survey of 78 mutation-induced cryptic exons previously identified in 51 disease genes revealed the presence of TEs in 40 cases (51%). Most TE-containing exons were derived from short interspersed nuclear elements (SINEs), with Alus and mammalian interspersed repeats (MIRs) covering >18 and >16% of the exonized sequences, respectively. The majority of SINE-derived cryptic exons had splice sites at the same positions of the Alu/MIR consensus as existing SINE exons and their inclusion in the mRNA was facilitated by phylogenetically conserved changes that improved both traditional and auxiliary splicing signals, thus marking intronic TEs amenable for pathogenic exonization. The overrepresentation of MIRs among TE exons is likely to result from their high average exon inclusion levels, which reflect their strong splice sites, a lack of splicing silencers and a high density of enhancers, particularly (G)AA(G) motifs. These elements were markedly depleted in antisense Alu exons, had the most prominent position on the exon-intron gradient scale and are proposed to promote exon definition through enhanced tertiary RNA interactions involving unpaired (di)adenosines. The identification of common mechanisms by which the most dynamic parts of the genome contribute both to new exon creation and genetic disease will facilitate detection of intronic mutations and the development of computational tools that predict TE hot-spots of cryptic exon activation.

SpliceIT: a hybrid method for splice signal identification based on probabilistic and biological inference

Splice sites define the boundaries of exonic regions and dictate protein synthesis and function. The splicing mechanism involves complex interactions among positional and compositional features of different lengths. Computational modeling of the underlying constructive information is especially challenging, in order to decipher splicing-inducing elements and alternative splicing factors. SpliceIT (Splice Identification Technique) introduces a hybrid method for splice site prediction that couples probabilistic modeling with discriminative computational or experimental features inferred from published studies in two subsequent classification steps. The first step is undertaken by a Gaussian support vector machine (SVM) trained on the probabilistic profile that is extracted using two alternative position-dependent feature selection methods. In the second step, the extracted predictions are combined with known species-specific regulatory elements, in order to induce a tree-based modeling. The performance evaluation on human and Arabidopsis thaliana splice site datasets shows that SpliceIT is highly accurate compared to current state-of-the-art predictors in terms of the maximum sensitivity, specificity tradeoff without compromising space complexity and in a time-effective way. The source code and supplementary material are available at: http://www.med.auth.gr/research/spliceit/.

Sim4cc: a cross-species spliced alignment program

Advances in sequencing technologies have accelerated the sequencing of new genomes, far outpacing the generation of gene and protein resources needed to annotate them. Direct comparison and alignment of existing cDNA sequences from a related species is an effective and readily available means to determine genes in the new genomes. Current spliced alignment programs are inadequate for comparing sequences between different species, owing to their low sensitivity and splice junction accuracy. A new spliced alignment tool, sim4cc, overcomes problems in the earlier tools by incorporating three new features: universal spaced seeds, to increase sensitivity and allow comparisons between species at various evolutionary distances, and powerful splice signal models and evolutionarily-aware alignment techniques, to improve the accuracy of gene models. When tested on vertebrate comparisons at diverse evolutionary distances, sim4cc had significantly higher sensitivity compared to existing alignment programs, more than 10% higher than the closest competitor for some comparisons, while being comparable in speed to its predecessor, sim4. Sim4cc can be used in one-to-one or one-to-many comparisons of genomic and cDNA sequences, and can also be effectively incorporated into a high-throughput annotation engine, as demonstrated by the mapping of 64 000 Fagus grandifolia 454 ESTs and unigenes to the poplar genome.

Global control of aberrant splice-site activation by auxiliary splicing sequences: evidence for a gradient in exon and intron definition

Auxiliary splicing signals play a major role in the regulation of constitutive and alternative pre-mRNA splicing, but their relative importance in selection of mutation-induced cryptic or de novo splice sites is poorly understood. Here, we show that exonic sequences between authentic and aberrant splice sites that were activated by splice-site mutations in human disease genes have lower frequencies of splicing enhancers and higher frequencies of splicing silencers than average exons. Conversely, sequences between authentic and intronic aberrant splice sites have more enhancers and less silencers than average introns. Exons that were skipped as a result of splice-site mutations were smaller, had lower SF2/ASF motif scores, a decreased availability of decoy splice sites and a higher density of silencers than exons in which splice-site mutation activated cryptic splice sites. These four variables were the strongest predictors of the two aberrant splicing events in a logistic regression model. Elimination or weakening of predicted silencers in two reporters consistently promoted use of intron-proximal splice sites if these elements were maintained at their original positions, with their modular combinations producing expected modification of splicing. Together, these results show the existence of a gradient in exon and intron definition at the level of pre-mRNA splicing and provide a basis for the development of computational tools that predict aberrant splicing outcomes.