Hoang Le

Insertion sequences (IS) are compact and pervasive transposable elements found in bacteria, which encode only the genes necessary for their mobilization and maintenance. IS 200 /IS 605 elements undergo ‘peel-and-paste’ transposition catalyzed by a TnpA transposase, but intriguingly, they also encode diverse, TnpB- and IscB-family proteins that are evolutionarily related to the CRISPR-associated effectors Cas12 and Cas9, respectively. Recent studies demonstrated that TnpB-family enzymes function as RNA-guided DNA endonucleases, but the broader biological role of this activity has remained enigmatic. Here we show that TnpB/IscB are essential to prevent permanent transposon loss as a consequence of the TnpA transposition mechanism. We selected a family of related IS elements from Geobacillus stearothermophilus that encode diverse TnpB/IscB orthologs, and showed that a single TnpA transposase was active for transposon excision. The donor joints formed upon religation of IS-flanking sequences were efficiently targeted for cleavage by RNA-guided TnpB/IscB nucleases, and co-expression of TnpB together with TnpA led to significantly greater transposon retention, relative to conditions in which TnpA was expressed alone. Remarkably, TnpA and TnpB/IscB recognize the same AT-rich transposon-adjacent motif (TAM) during transposon excision and RNA-guided DNA cleavage, respectively, revealing a striking convergence in the evolution of DNA sequence specificity between collaborating transposase and nuclease proteins. Collectively, our study reveals that RNA-guided DNA cleavage is a primal biochemical activity that arose to bias the selfish inheritance and spread of transposable elements, which was later co-opted during the evolution of CRISPR-Cas adaptive immunity for antiviral defense.

ABSTRACT TnpB nucleases represent the evolutionary precursors to CRISPR-Cas12 and are widespread in all domains of life, presumably due to the critical roles they play in transposon proliferation. IS605-family TnpB homologs function in bacteria as programmable homing endonucleases by exploiting transposon-encoded guide RNAs to cleave vacant genomic sites, thereby driving transposon maintenance through DSB-stimulated homologous recombination. Whether this pathway is conserved in other genetic contexts, and in association with other transposases, is unknown. Here we uncover molecular mechanisms of transposition and RNA-guided DNA cleavage by IS607-family elements that, remarkably, also encode catalytic, self-splicing group I introns. After reconstituting and systematically investigating each of these biochemical activities for a candidate ‘IStron’ derived from Clostridium botulinum , we discovered sequence and structural features of the transposon-encoded RNA that satisfy molecular requirements of a group I intron and TnpB guide RNA, while still retaining the ability to be faithfully mobilized at the DNA level by the TnpA transposase. Strikingly, intron splicing was strongly repressed not only by TnpB, but also by the secondary structure of ωRNA alone, allowing the element to carefully control the relative levels of spliced products versus functional guide RNAs. Our results suggest that IStron transcripts have evolved a sensitive equilibrium to balance competing and mutually exclusive activities that promote transposon maintenance while limiting adverse fitness costs on the host. Collectively, this work explains how diverse enzymatic activities emerged during the selfish spread of IS607-family elements and highlights molecular innovation in the multi-functional utility of transposon-encoded noncoding RNAs.

Accurately predicting the molecular subtype of cancer is crucial for personalized diagnosis and treatment. Furthermore, finding reliable biomarkers is essential for achieving personalized medicine and improving patient outcomes overall. This paper presents a novel approach for multi-omics biomarker discovery in Glioblastoma using explainable artificial intelligence. The proposed method integrated data from the TCGA cohort and MOGONET's deep learning model for subtyping problems and then used the Integrated Gradients algorithm to discover important features in the multi-omics data. The identified biomarkers were subsequently validated through two approaches: first, by comparing them with established ground truth data from reference databases, and second, performance evaluation using classical machine learning models.