Wagh, A. | bioRxiv Science

bioRxiv2026

Organyl 5'-Phosphates in siRNA Guide Strands: Structure-Function Relationships Governing Anchoring in Argonaute 2 and Metabolic Stability

Efficient siRNA loading into Argonaute2 (AGO2) requires a 5'-phosphate (5'-P) on the guide strand, yet this group is vulnerable to metabolic degradation in vivo. Although numerous chemical mimics of 5'-P have been reported, structural principles governing AGO2 interactions with organyl substituents on the 5'-P remain unclear. Moreover, structural determinants of 5'-P mimic recognition by known degradative enzymes (principally phosphatases and 5'-exonucleases) are also poorly understood. The 5'-P binding site of the AGO2 MID domain contains a stack of aromatic residues (Y527/F811/Y815), presenting a structural basis for augmenting canonical anchoring interactions. Herein, we systematically synthesized and characterized a diverse panel of organyl 5'-phosphates (5'-POR; R = 35 variable substituents) as guide strand 5'-P mimics designed to engage this unique hydrophobic pocket. Among the compounds evaluated, 5'-POR guide strands bearing methyl (Me) or phenylpropargyl (PhPrp) substituents are well-tolerated by AGO2 in cells. Previously uncharacterized 5'-P mimics, including 5'-phosphorothioate (5'-PS), phenylpropargyl 5'-phosphorothioate (5'-PS-PhPrp), and 5'-mesylphosphoramidate (5'-MsPA), maintain comparable AGO2 compatibility. All examined 5'-P mimics are markedly resistant to phosphatase, while 5'-POR variants and 5'-PS-PhPrp are also resistant to 5'-exonuclease degradation due to masking a negative charge of 5'-P. A crystal structure of a 5'-PO-PhPrp guide strand loaded into AGO2 reveals an unexpected network of {pi}-{pi} interactions between the rigid phenylpropargyl group and the targeted hydrophobic pocket of the MID domain. Collectively, these findings expand the functional chemical space of 5'-P mimics and define new modes for metabolically stabilizing the guide strand 5'-end while augmenting AGO2 MID anchoring. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=85 SRC="FIGDIR/small/705631v1_ufig1.gif" ALT="Figure 1"> View larger version (35K): org.highwire.dtl.DTLVardef@e6f88forg.highwire.dtl.DTLVardef@1c87dfaorg.highwire.dtl.DTLVardef@1c6e68corg.highwire.dtl.DTLVardef@14a1f60_HPS_FORMAT_FIGEXP M_FIG C_FIG

molecular biology↗

bioRxiv2025

CRYPTID-exon: streamlined detection of cryptic exons from RNA-seq data

Cryptic splicing has emerged as a pervasive feature of mammalian gene expression, with recent studies uncovering thousands of previously unannotated splice sites. Despite its prevalence, the functional consequences of this hidden layer of splicing remain largely unknown due to challenges in identifying the exact exonic regions introduced into mRNA transcripts. Here, we introduce a novel computational approach, CRYPTID-exon, that accurately predicts exon boundaries by modeling RNA-seq read coverage around empirically derived splice sites. We use CRYPTID-exon to identify and characterize thousands of cryptic exons in nascent and mature RNA from human cells. Additionally, we demonstrate that CRYPTID-exon is well powered to identify exons that are sensitive to translation-mediated degradation processes. Finally, given the growing interest in leveraging cryptic exons to modulate gene expression levels, we use our approach to identify cryptic exons in disease-relevant genes. We see that targeting these cryptic exons with splice-switching antisense oligonucleotides (ASOs) can alter gene expression and splicing patterns of the parent genes. Our study provides a framework to systematically identify and characterize cryptic exons, which will enable downstream insights into their impact on mRNA stability and translation.

genomics↗