Von Stetina, S. E. | Search bioRxiv

bioRxiv2021

Translation-dependent mRNA localization to Caenorhabditis elegans adherens junctions

mRNA localization is an evolutionarily widespread phenomenon that facilitates sub-cellular protein targeting. Extensive work has focused on mRNA targeting through "zip codes" within untranslated regions (UTRs), while much less is known about translation-dependent cues. Here, we examine mRNA localization in Caenorhabditis elegans embryonic epithelia. From an smFISH-based survey, we identified mRNAs associated with the cell membrane or cortex, and with apical junctions in a stage- and cell type-specific manner. Mutational analyses for one of these transcripts, dlg-1/discs large, revealed that it relied on a translation-dependent process and did not require its 5 or 3 UTR. We suggest a model in which dlg-1 transcripts are co-translationally colocalized with the encoded protein: first the translating complex goes to the cell membrane through sequences of the SH3 domain, and then to the apical junction by the L27 and PDZ sequences. In addition, the Hook and GuK sequences contribute to the second step: they are required for mRNA, but not protein, to accumulate at the apical junctions from locations at or near the membrane. These studies identify a translation-based process for mRNA localization within developing epithelia and determine the necessary cis-acting sequences for dlg-1 mRNA targeting. Summary statementAn smFISH-based survey identified a subset of mRNAs coding for junctional components that localize at or in the proximity of the adherent junction through a translation-dependent mechanism.

developmental biology↗

bioRxiv2020

Attribution of genetic engineering: a practical and accurate machine learning toolkit for biosecurity

The promise of biotechnology is tempered by its potential for accidental or deliberate misuse. Reliably identifying telltale signatures characteristic to different genetic designers, termed genetic engineering attribution, would deter misuse, yet is still considered unsolved. Here, we show that recurrent neural networks trained on DNA motifs and basic phenotype can reach 70% attribution accuracy distinguishing between over 1,300 labs. To make these models usable in practice, we introduce a framework for weighing predictions against other investigative evidence using calibration, and bring our model to within 1.6% of perfect calibration. Additionally, we demonstrate that simple models can accurately predict both the nation-state-of-origin and ancestor labs, forming the foundation of an integrated attribution toolkit which should promote responsible innovation and international security alike.

synthetic biology↗