RNA Biology

The tRNAs are non-coding RNAs (ncRNAs) known for their classical role in decoding mRNAs. Analysis of tRNA sequences from Escherichia coli revealed pairwise similarities for different combinations of tRNAs, as well as complementarities between some tRNA pairs. To explore the physiological significance of the complementarities between the tRNAs, we investigated the pair of tRNAs encoded by lysT (encoding an abundant tRNA, tRNALysT), and argU (encoding a rare tRNA, tRNAArgU). We show that tRNALysT and tRNAArgU anneal to form a heterodimer in vitro. The heterodimerisation is prevented by the presence of DNA oligomers complementary to the interacting sequences, in a dose dependent manner. Consistent with the notion of sequestration of tRNAArgU by tRNALysT, while the overexpression of tRNAArgU did not impact the culture growth, that of tRNALysT did. The tRNALysT mediated inhibition of the culture growth was enhanced at a lower temperature. The AGA minigene (decoded by tRNAArgU) mediated toxicity and hybrid phage ({lambda}imm-P22) growth on ssrA (tmRNA) strains was also consistent with the sequestration of tRNAArgU by tRNALysT. Also, we observed tRNA-derived fragments (tRFs) from tRNALysT and tRNAArgU, which too might facilitate tRNAArgU sequestration. Taken together, these observations support a novel regulatory role of the evolutionary conserved complementarities between tRNAs.

Bacterial small RNAs (sRNAs) regulate gene expression by base pairing with target mRNAs, yet transcriptome-wide interactome mapping has shown that many sRNA-mRNA interactions detected in vivo have modest or no regulatory effect using orthogonal reporter assays. The features that determine functional outcome remain poorly defined. Here, we integrated Hfq-CLASH interactome mapping with matched transcriptomic and proteomic profiling in Escherichia coli and developed an interpretable machine-learning framework to identify the determinants that distinguish functional from non-functional interactions. Using sequence, structural, thermodynamic, duplex and protein-occupancy features, transcriptomic and proteomic responses were predicted with above-chance performance, achieving AUCs of 0.78 and 0.74, respectively. Feature attribution revealed that physical pairing alone is insufficient for regulation; instead, regulatory outcome is shaped by a coordinated interplay between RNA secondary structure, thermodynamic accessibility and local protein-binding context. Target-side Hfq occupancy emerged as a positive predictor of functional regulation, whereas AR2-domain occupancy on the sRNA was associated with non-responsive interactions, suggesting that distinct ribonucleoprotein states may separate productive regulation from non-productive binding. These findings indicate that the regulatory fate of an sRNA-mRNA interaction is an emergent property of its biophysical context and protein-binding environment, rather than a direct consequence of physical pairing alone.

Transfer RNAs (tRNAs) are key molecules that deliver amino acids to the translating ribosome according to their cognate codon encoded in messenger RNA (mRNA). Due to the modified nature of tRNA nucleotides, accurate tRNA quantification can be tedious, especially when dealing with small sample inputs. Here, we took advantage of an RNA-dependent RNA amplification method using T7 polymerase to quantify tRNA abundance in low biological input. Our method, called T7 High-resolution original RNA (Thor)-tRNA-Seq, showed reproducible and quantitative measurement of low tRNA inputs. Thus, our Thor-tRNA-Seq is a robust and reliable approach for the quantification of tRNA in samples with small and precious biological material.

Due to its crucial role in proteobacterial RNA processing and degradation, the synthesis of the endonuclease RNase E is tightly controlled by a post-transcriptional autoregulatory mechanism governed by the 5' untranslated region of rne mRNA, the transcript that encodes it. Previous studies have identified one structural component of the rne 5' untranslated region (hp2) that helps to mediate this process in Escherichia coli by binding to RNase E. Here, we show that another component (hp3), a branched stem-loop, makes an even greater contribution to feedback regulation of RNase E synthesis and that the effects of hp2 and hp3 are additive, suggesting independent function. Despite the structural complexity of hp3, a single constituent branch appears to be the principal source of its regulatory activity. This branch also enables hp3 to bind RNase E with high affinity. The secondary structure of this branch, including the presence of two internal loops essential for its impact, is conserved in diverse Enterobacterales species, whereas the sequence of the internal loops is not. We propose that hp2 and hp3 mediate autoregulation by recruiting RNase E to the rne transcript.

This study proposes a novel therapeutic strategy to suppress cancer growth by modulating the MDM2-p53 axis via Alternative Polyadenylation (APA). MDM2 normally promotes tumorigenesis by ubiquitinating and degrading the tumor suppressor p53. In cancer cells, preferential use of proximal polyadenylation signals (PAS) results in shortened 3'UTRs, allowing oncogenic transcripts like MDM2 to evade nuclear sequestration mediated by Inverted Alu (IRAlu) double-stranded RNA structures. We hypothesized that forcing distal PAS usage would elongate the MDM2 mRNA, promoting its nuclear retention and reducing protein translation, thereby restoring p53 activity. Using CRISPR-Cas9, we targeted and deleted the most frequent proximal PAS in the MDM2 3'UTR of A549 cells. Successful genome editing was confirmed via PCR. As expected, Western blot analysis showed a significant reduction in MDM2 expression in PAS-edited cells. However, experimental outcomes contradicted our initial hypothesis: edited cells exhibited higher viability under doxorubicin treatment compared to wild-type cells. Furthermore, despite decreased MDM2 levels, a concurrent reduction in phosphorylated p53 (p-p53) was observed. These unexpected results suggest that MDM2 3'UTR elongation may trigger a non-canonical regulatory mechanism that bypasses the traditional MDM2-p53 interaction. This study highlights the complexity of post-transcriptional regulation and suggests that APA-mediated gene modulation can induce unforeseen compensatory survival pathways in cancer cells, necessitating further investigation into the broader functional landscape of elongated 3'UTRs.

Transfer RNA (tRNA) is highly modified and difficult to profile using reverse-transcription-based sequencing. Here, we established a direct RNA sequencing (tRNA-DRS) workflow to simultaneously characterize tRNA abundance and modification signals in mouse liver. The workflow included small-RNA enrichment, deacylation, splint-adapter ligation, barcoding, reverse transcription, and nanopore sequencing on PromethION with modification-aware basecalling. From 1,678,666 raw reads, 1,421,279 (84.67%) reads were retained after 20-200 nt length filtering, and 17,094 uniquely mapped reads (MAPQ [≥] 10) were obtained. These reads covered 244 tRNA genes and represented all 23 tRNA isotypes. tRNA abundance was strongly skewed, with Asp, Gly, Ser, and Lys accounting for 56.6% of uniquely mapped reads. Read-body coverage indicated that most uniquely mapped reads spanned substantial portions of tRNA molecules (mean 71.1%, median 75.9%; 83.3% reads covering [≥]50% of the reference). Modification analysis identified 41,013 candidate sites, of which 711 passed stringent filtering (modification fraction [≥]50%, coverage [≥]5), spanning eight modification types and dominated by pseudouridine and m5C. Filtered sites showed position-specific clustering and isotype-dependent patterns. Together, these results support tRNA-DRS as a practical single-assay framework for simultaneous tRNA expression and epitranscriptomic profiling in mammalian tissue.

BackgroundMicroRNAs (miRNAs) play key roles in modulating gene expression. Upon transcription, primary miRNA transcripts (pri-miRNAs or MIRNAs) fold into stem-loop structures. Endonucleolytic cleavage releases a miRNA/miRNA* duplex from the stem, which is subsequently matured. In higher plants, the pri-miRNA hairpins vary widely in length and structure. Pri-miRNA processing is extensively regulated by RNA-binding proteins (RBPs) and the range of underlying mechanisms continues to expand. ResultsHere, we performed an unbiased screen to identify Arabidopisis thaliana RBPs that interact with pri-miRNAs, using pri-miRNA159a and pri-miRNA398b as paradigms. Nucleoplasmic proteins were significantly enriched in pulldowns with bead-immobilized in vitro transcripts of the stem-loop regions of these pri-miRNAs compared with pulldowns using empty beads. The enriched proteins included several members of the family of the serine/arginine-rich (SR) splicing factor family, including RS31 and SR34a. Mutant analysis indicated that selected candidate interactors affect pri-miRNA and/or mature miRNA accumulation. To validate interactions with pri-miRNAs in vivo, we performed individual-nucleotide resolution crosslinking and immunoprecipitation (iCLIP). Whereas our conventional plant iCLIP procedure detected only a few crosslink sites on pri-miRNAs, likely because of the low abundance of these transcripts, our improved plant iCLIP2 protocol enabled the identification of binding sites on numerous pri-miRNAs. RS31 and SR34a bound overlapping sets of pri-miRNAs and contacted both stem-loop regions and flanking regions upstream and downstream of the stem-loop. RS31 and SR34 bound both overlapping and distinct regions of pri-miRNAs. ConclusionIn vitro pulldowns of nucleoplasmic proteins identified RBPs associated with pri-miRNA stem-loops, and the improved plant iCLIP2 protocol revealed in vivo binding of RS31 and SR34a to a subset of pri-miRNAs. Together, these findings expand the repertoire of RBPs implicated in plant miRNA biogenesis and support a role for SR proteins as candidate regulators of pri-miRNA processing.

O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=79 SRC="FIGDIR/small/745202v1_ufig1.gif" ALT="Figure 1"> View larger version (21K): org.highwire.dtl.DTLVardef@52e3edorg.highwire.dtl.DTLVardef@1f21637org.highwire.dtl.DTLVardef@909ee8org.highwire.dtl.DTLVardef@b0c907_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOGRAPHICAL ABSTRACTC_FLOATNO C_FIG Crosslinking and immunoprecipitation followed by sequencing (CLIP-seq) is widely used to identify the RNA targets of RNA-binding proteins (RBPs). However, its application to non-canonical RBPs lacking canonical RNA-binding domains and frequently displaying low or transient RNA occupancy, is limited by low signal-to-noise ratios, high input requirements and error-prone ligation steps during library preparation. To overcome these limitations, we developed soniCLIP, a streamlined CLIP-seq workflow that replaces RNase-mediated RNA fragmentation with sonication and uses a ligation-free strategy for library construction. soniCLIP is optimized for reproducible identification of enriched RBP-associated RNA regions from limited starting material. We benchmarked soniCLIP against the widespread eCLIP approach and observed reproducible recovery of known RBP-associated regions and target recovery comparable to ENCODE eCLIP, while requiring only 10% (500 {micro}g) of protein input. We further applied soniCLIP to the glycolytic enzyme and non-canonical RBP pyruvate kinase M2 (PKM2). We identified 197 significantly enriched RNA regions and validated selected targets by RIP-qRT-PCR and in vitro binding assays. By combining reduced input requirements, high reproducibility, a shortened 3.5-day workflow and the elimination of gel-based purification, soniCLIP provides an efficient and robust approach for the identification of RNA targets of canonical and non-canonical RBPs.

Transfer RNA (tRNA) modifications are essential for translational accuracy and cellular adaptation to environmental changes. While short-term modification dynamics are well documented, the impact of prolonged stress exposure on the global tRNA landscape remains largely unexplored. This study provides the first systematic profiling of tRNA modifications in Saccharomyces cerevisiae following long-term exposure to distinct stress types: heat, suboptimal pH, oxidative stress (paraquat and diamide), osmotic stress (NaCl and KCl), and genotoxic stress (MMS). To this end, we used a broad-range UPLC-MS protocol to quantify global changes in tRNA modification and identify stress-specific signatures. The results revealed that long-term stress triggers a global reprogramming of the tRNA epitranscriptome in a stress-specific and time-dependent manner. Importantly, while our findings confirm the previously reported temperature-sensitivity of wobble uridine thiolation, we also identified a complete or partial loss of 5-methoxycarbonylmethyl-2-thiouridine (mcm5s2U34) modification upon exposure to paraquat and pH stress. Furthermore, this loss of thiolation is accompanied by an accumulation of the non-thiolated 5-methoxycarbonylmethyl (mcm5U) precursor, suggesting a stress-dependent impairment of the thiolation pathway. Next, we applied MarathonRT-based tRNA sequencing and showed that these modification changes occur independently of tRNA isoacceptor abundance. To further integrate these results, we devised a modification deviation (MDm) index, which indicates that the observed reprogramming is primarily linked to events that are independent from changes in tRNA abundance. Together, this study provides a comprehensive atlas of tRNA modification dynamics under prolonged stress, addressing a critical gap in our understanding of RNA-based translational control and establishes the MDm index as a robust quantitative framework to decouple the influence of tRNA abundance from global modification signals, providing a necessary metric for the field to interpret epitranscriptomic reprogramming. TABLE OF CONTENTS GRAPHIC O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=108 SRC="FIGDIR/small/745200v1_ufig1.gif" ALT="Figure 1"> View larger version (41K): org.highwire.dtl.DTLVardef@2cf571org.highwire.dtl.DTLVardef@1a5f3c2org.highwire.dtl.DTLVardef@266d61org.highwire.dtl.DTLVardef@d7214f_HPS_FORMAT_FIGEXP M_FIG C_FIG

Circular RNAs (circRNAs) are generated by backsplicing of eukaryotic protein-coding transcripts and can regulate microRNAs and RNA binding proteins, or serve as translation templates. Their covalently closed structure confers resistance to exonuclease-mediated degradation, extending their half-life and supporting their development as RNA therapeutics. However, existing overexpression methods often yield substantial contaminating linear RNAs, limiting their utility. Here, we systematically benchmarked plasmid-based circRNA overexpression strategies in human cells, comparing spliceosome- and ribozyme-based mechanisms across constructs incorporating widely used flanking sequences. The ribozyme-based Tornado system produced the highest circRNA yield but introduced extraneous "molecular scars" into the mature product. By contrast, spliceosome-mediated circularization using introns from the Drosophila Laccase2 gene, which contain imperfect complementary repeats, produced scarless circRNA with substantially lower linear RNA contamination. Linear RNA was further reduced by engineering the primary transcript to terminate in a non-polyadenylated end, increasing its susceptibility to exonucleases. Building on this optimized system, we developed a dual-output platform co-expressing a linear fluorescent reporter alongside a circRNA from a single promoter (CIRCUS, circRNA and upstream linear system), enabling efficient screening of circRNA-driven cellular phenotypes, including site-specific A-to-I editing of target mRNAs. Together, this toolkit provides high-purity circRNA production suitable for mechanistic studies and circRNA-based therapeutic development.

In mammalian reproduction, sperm and oocytes fuse to generate embryos. In contrast to maternally inherited small RNAs, sperm-borne small RNAs have been implicated in embryo viability but remain incompletely understood. Unanswered questions about molecular forces shaping small RNA content in sperm and sources of reporting discrepancies hinder a deeper understanding of their post-fertilization roles. Here, we investigate the dynamics and protein association of small RNAs in C57BL/6 mice during spermiogenesis and sperm capacitation. Using unique molecular identifiers and spike-ins to minimize technical biases in low-input sequencing, we demonstrate a selective sperm small RNA repertoire derived from a subset of spermatogenic small RNAs that remains malleable during capacitation. Retained PIWI-interacting RNAs persist despite widespread RNA decay and are insensitive to capacitation, which may be partly attributed to their association with MIWI proteins. Our data reveal RNA species susceptible to PCR duplicate-related abundance overestimation and uncover biological and technical factors shaping experimentally observed sperm small RNA profiles. These findings refine our understanding of sperm-borne RNAs and inform future functional studies and medically assisted reproduction.

The maternal-to-zygotic transition (MZT) is a major developmental event during which inherited transcripts are remodeled and zygotic transcription is established. Although parentally inherited long noncoding RNAs (lncRNAs) are present in early embryos, they have been thought to be dispensable. We have identified more than 2000 inherited lncRNAs in zebrafish embryos, but how these RNAs participate in regulatory programs during early development has remained unexplored. Here, the inheritance of selected zebrafish lncRNAs spanning a broad expression range were confirmed at the pre-MZT stage and full-length sequences were captured by Direct RNA nanopore sequencing. We show that 30% inherited intergenic lncRNAs are preferentially associated with active enhancers, annotated as such in DANIO CODE, whereas non-inherited intergenic lncRNAs rarely overlap with enhancers. Perturbation of five inherited intergenic lncRNAs, individually, using antisense oligonucleotides reduced the expression of their respective neighboring genes at 2.5, 4.3, and/or 6 hours post fertilization, indicating that these RNAs act as positive local regulators during MZT. Together, these findings identify inherited intergenic lncRNAs as enhancer-associated regulators with elncRNA-like properties during early embryogenesis.

The majority of eukaryotes encode some intron-containing pre-tRNAs. Splicing of these pre-tRNAs requires a dedicated tRNA splicing machinery. The fungal and trypanosome tRNA ligase, Trl1, and the human RNA ligase, RTCB, catalyze an essential step in tRNA splicing. However, Trl1 and RTCB are nonhomologous and biochemically and structurally distinct from each other. Therefore, Trl1 could serve as a broad-spectrum antifungal and anti-trypanosomal target. While the functions and requirements of the three catalytic Trl1 domains have been extensively characterized in the model yeast Saccharomyces cerevisiae, the roles of Trl1 orthologs in pathogenic fungi remain unexplored. Here, we validate Trl1 as one of the few promising novel drug targets for the development of antifungal therapeutics. Functional analyses of the three Trl1 domains show that only the "sealing" domain is essential for growth and viability in Candida albicans and Aspergillus fumigatus. In contrast, the two "healing" domains are dispensable in these pathogenic fungi, suggesting the presence of redundant healing enzymes, unlike in S. cerevisiae. These findings indicate that only the sealing domain is a good drug target. Our analysis also shows that the Mucor enzyme, which only contains the sealing domain, is essential. Using a Caenorhabditis elegans infection model of C. albicans, we further demonstrated that inhibiting Trl1 expression protects worms during an established infection. In contrast to these fungal pathogens, we show that all three domains of Trl1 are essential in Trypanosoma brucei. Our findings show that the essentiality of the Trl1 sealing is conserved in important human pathogens and provides an impetus for future drug development. SIGNIFICANCEFungal infections are an important cause of human disease and death and difficult to treat and there is an urgent need to develop additional drugs. Based on studies in yeast, one promising target for antifungal drug development is the tRNA splicing pathway. Human tRNA ligase is fundamentally distinct from the fungal one. To investigate the possibility of developing tRNA ligase-targeting drugs, we investigated the function of the catalytic domains of fungal tRNA ligase in different fungal pathogens. Surprisingly, only the first domain is essential in these pathogens and yeast is not a good model fungus. In contrast, all three domains of Trypanosome tRNA ligase are essential. These findings provide an impetus for future drug development.

ABSTRACT: The bacterial small RNA (sRNA) OxyS is expressed in Escherichia coli during oxidative stress. The sRNA OxyS enhances cell survival by controlling genes involved in the regulation of hydrogen peroxide (H2O2) and iron-sulfur (Fe-S) cluster formation. Here, we used the MS2 affinity purification coupled with RNA sequencing (MAPS) technique to identify new target mRNAs of the sRNA OxyS. Our analysis revealed a significant enrichment of mepS mRNA, which encodes a peptidoglycan endopeptidase that promotes cell growth. Our results confirm a previous report on the sRNA OxyS repressing the translation of mepS. We also found that an {Delta}oxyS background facilitates the emergence of mutations, conferring increased resistance to the last-resort antibiotics polymyxin B and E (colistin), but only in the presence of the target mepS gene. This suggests that the translation repression of mepS by OxyS could prevent mutations in bacterial DNA during H2O2-induced oxidative stress. Moreover, we show that adding the antioxidant thiourea or sequestering iron in the {Delta}oxyS background effectively reduces the emergence of resistance against both polymyxin B and colistin. These results suggest that reactive oxygen species (ROS), in conjunction with intracellular iron, play a key role in driving the emergence of antibiotic resistance. Overall, our work underlines a mechanism of antimicrobial emergence implicating oxidative stress, intracellular Fe, and cell wall remodeling in E. coli. IMPORTANCE: This study uncovers an underexplored link between peptidoglycan remodeling and oxidative stress responses during exposure to antibiotics. By elucidating how MepS and the sRNA OxyS interact in the presence of polymyxins and oxidative stress, our study suggests that MepS may exert an anti-mutator function. The repression of mepS translation by OxyS seems to limit the emergence of antibiotic resistance driven by DNA mutations. Together, these findings suggest cell wall remodeling and oxidative stress response pathways as promising targets to enhance antibiotic efficacy and limit the emergence of resistance.

RNA-targeting CRISPR systems are commonly evaluated by RT-qPCR, but guide RNA binding can confound these measurements. We show that crRNA alone produces apparent knockdown without reducing target RNA abundance, whereas RNA sequencing remains unbiased and reveals guide-associated transcriptomic perturbations. A simple RNA denaturation step before reverse transcription restores accurate RT-qPCR quantification, providing practical guidance for RNA-targeting CRISPR analysis and guide design.

RNAs regulate gene expression and cellular processes, often relying on specific conformations for function. RNA folding is hierarchical and sequence-dependent, with nearest-neighbor thermodynamic models commonly used to predict secondary structure. Current models were developed using optical melting experiments in 1 M NaCl, which does not represent the cellular environment. To address this, we developed a new model in Advanced Dulbecco's Modified Eagle Medium (Adv. DMEM), which mimics mammalian extracellular ionic composition. This in vivo-like model provides RNA folding parameters for helical base stacks and loop motifs. Optical melting experiments revealed helical stacks, particularly tandem G-U pairs, are less stabilizing in Adv. DMEM. Loop parameters were generally destabilizing but highly dependent on both sequence and loop type, with internal loops displaying idiosyncratic behavior. Structure prediction benchmarking revealed minimal differences overall, except for tRNAs, which showed improved prediction reliability and enhanced cloverleaf stability. Notably, tRNAs lack internal loops, suggesting further studies in Adv. DMEM could refine secondary structure predictions. This in vivo-like parameter set is included in the RNAstructure software package. Grounding these parameters in a physiologically relevant environment, we improve the biological relevance of RNA secondary structure predictions and establish a foundation for studying RNA folding under in vivo conditions.

In recent years, the catalog of mammalian box C/D small nucleolar RNAs (SNORDs) has expanded considerably, with several hundred members now annotated in humans. Despite being a well-characterized family of antisense small noncoding RNAs, the full diversity of mammalian box C/D SNORDs remains incompletely understood. To address this, we combined deep sequencing of Fibrillarin- and Snu13-associated RNAs with systematic mining of a large collection of publicly available small RNA-seq datasets spanning diverse mouse tissues and cell types. This integrative approach uncovered an unexpectedly large repertoire of box C/D-like RNAs, comprising more than 20 000 distinct species. These RNAs display the defining hallmarks of intronic SNORDs but are expressed at low levels and exhibit poor evolutionary conservation. Strikingly, approximately 70% are only 20-50 nucleotides long, making them substantially shorter than canonical SNORDs. We designate this previously unrecognized class of box C/D-like RNAs as tiny SNORDs (tiSNORDs). Functional characterization of a representative 36-nucleotide tiSNORD demonstrates that, despite its reduced size, it retains nucleolar 2'-O-ribose methylation guide activity. The discovery of this previously unrecognized class of box C/D-like RNAs substantially expands the known repertoire of SNORDs and raises fundamental questions regarding their biogenesis, functional potential and evolutionary significance.

Very short RNAs play critical roles in modern biology, and are thought to have been crucial for genome replication during the origin of life. Next-generation sequencing is an essential tool for characterizing pools of small RNAs, but current library preparation methods suffer from strong size and sequence biases. Here we present tinyRNA-seq, an optimized library preparation method designed to minimize length- and sequence-dependent capture bias enabling the sequencing of RNA fragments as short as 2 nucleotides. We use degenerate adaptor regions to reduce ligation sequence bias and facilitate unique molecular identifier (UMI) installation. We benchmarked tinyRNA-seq against commercial kits using a model primordial RNA genome consisting of hundreds of defined oligonucleotides ranging from 2 to 12 nucleotides. tinyRNA-seq reproduced the input RNA distribution without the size and sequence bias of the commercial kits. tinyRNA-seq also enables the detection of de novo oligonucleotide generation, an important process for the origins of life. Applied to biologically derived small RNAs including miRNAs, piRNAs, and cityRNAs, tinyRNA-seq showed significantly lower capture bias and recovered a wider range of sequences than commercial kits. tinyRNA-seq may thus provide a more complete and quantitatively accurate representation of small RNAs from both biological and chemical sources. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=94 SRC="FIGDIR/small/743385v1_ufig1.gif" ALT="Figure 1"> View larger version (29K): org.highwire.dtl.DTLVardef@95ee64org.highwire.dtl.DTLVardef@155fb06org.highwire.dtl.DTLVardef@1d3665forg.highwire.dtl.DTLVardef@1e61404_HPS_FORMAT_FIGEXP M_FIG C_FIG

Double-stranded RNA (dsRNA) biopesticides that silence essential fungal genes by spray-induced gene silencing (SIGS) are advancing towards registration, underpinned by the claim that silencing is confined to sequence-matched target organisms, a claim that has never been tested systematically against beneficial fungi. We screened twelve dsRNA constructs (the whole transcripts of eleven canonical antifungal target genes of Fusarium graminearum: three CYP51 sterol 14-demethylases, six chitin synthases and two {beta}-tubulins, plus a concatenation of the three CYP51 transcripts modelling the flagship CYP3RNA design) for perfect 21-nucleotide (21-nt) identity, the primary match criterion of recent regulatory bioinformatics frameworks, against the transcriptomes of seven non-target fungi, including commercial biocontrol agents (Trichoderma harzianum, T. virens, Beauveria bassiana, Metarhizium anisopliae, M. brunneum), the arbuscular mycorrhizal fungus Rhizophagus irregularis and Saccharomyces cerevisiae. Eleven of the twelve constructs carried off-target hazard; only CYP51C was completely clean. TUB2b carried 580 perfect 21-nt matches and contained no hazard-free window of 250 consecutive 21-mer start positions, whereas 26.6-60.0% of screened windows were hazard-free in the designable genes. Hits landed overwhelmingly on orthologues of the targeted gene and concentrated in the Hypocreales; R. irregularis and S. cerevisiae were nearly clean. These are hazard findings, not risk findings: perfect siRNA-length matches to beneficial fungi are common in canonical antifungal dsRNA targets, and current regulatory screening panels contain no fungus that would detect them.

MicroRNAs (miRNAs) are processed from structured precursors and subsequently loaded into Argonaute proteins to repress target mRNAs. Yet generalized computational frameworks capable of decoding mature miRNAs from precursor context remain limited, constraining both cross-species annotation and rational artificial miRNA design. Here, we present miRstring, a biogenesis-aware RNA language framework that decodes the four boundaries defining the miRNA/miRNA duplex. Trained on 77,708 miRNA precursors spanning 414 species, miRstring outperforms existing methods under family- and species-held-out evaluations and accurately identifies the first nucleotide of mature miRNAs. Importantly, its attention mechanism highlights the miRNA/ miRNA* boundary sites cleaved by endonucleases, indicating that the model captures biologically meaningful features. Furthermore, we employed miRstring to design optimal pre-miRNA scaffolds for artificial miRNAs and validated its efficacy in repressing target mRNAs. Taken together, miRstring establishes a scalable route from cross-species mature-miRNA annotation to predictive design, enabling artificial intelligence-driven miRNA engineering and extending computational miRNA analysis toward broadly applicable small-RNA biotechnology.