show Abstracthide AbstractThe microbiome has revealed itself as a key player in health and disease. To better understand its role, in addition to microbial diversity, it is important to understand species-specific activity and gene expression. While metatranscriptomics investigates mRNA abundance2, it does not inform about faster post-transcriptional regulation3. Although prokaryotic translation is a common target for antibiotics, a direct measurement of microbiome ribosome dynamics remains inaccessible. Here we demonstrate that, contrary to expectation, co-translational mRNA degradation is common in prokaryotes, and that in vivo ribosome protection generates widespread 3-nt periodicity in 5´P mRNA decay intermediates. Consequently, 5´P sequencing allows the study of codon and gene specific ribosome stalling in response to stress and drug treatment at single nucleotide resolution. We validate its wide applicability by investigating in vivo species-specific ribosome footprints of clinical and environmental microbiomes and show that amino acid-specific ribosome protection patterns can be used to phenotype microbiome perturbations. Furthermore, we show that multiple RNase activities collaborate to generate in vivo ribosome footprints and that co-translational degradation is phylogenetically conserved across prokaryotes. This strategy opens the way for the study of the metatranslatome, and allows to investigate fast species-specific post-transcriptional responses to environmental and chemical perturbations in unculturable microbial communities. Overall design: We performed 2 to 5 biologically independent experiment (_rX) for individually grown bacteria, mock community or compost microbiomes using either 5PSeq (PMID: 26046441) or HT-5PSeq (doi: https://doi.org/10.1101/2020.06.22.165134). Majority of analyzed samples are from total RNA, with the exception of Bacillus subtilis RNA that was extracted from monosome/polyribosome fractions. Samples analyzed in this study include: Bacillus amyloliquefaciens (bamy), Escherichia coli (ecol), Lactobacillus plantarum (lpla), Lactobacillus reuteri (lreu), Bacillus subtilis (bsub-str168 or bsub), Caulobacter vibrioides (cvib), Synechocystis sp. PCC 6803 (spcc-6803), Alistipes finegoldii (afin), Prevotella copri (pcop), Parabacter merdae (pmer), Prevotella timonesis (ptim), Bacillus subtilis 168 trpC2 rnjA::spc (bsub_del-rnjA), ZymoBIOMICS Microbial Community Standard, cat-no D6300 (zymo), Compost microbiome (comp) and Fecal-cultures (fc). Samples were untreated (ctr), randomly fragmented and re-phosphorylated (frag), and treated with Chloramphenicol (cam), Mupirocin (mup), Erythromycin (ery), Doxycycline (dox), heat shock (hs), NaCL salt stress (saltstress), grown to stationary phase (stat) or exposed to low nutrients (starv). We use 2 different rRNA depletion strategies, either commercial RiBo-zero depletion and PAN-Prokaryote Ribopool (sitoolsbiotech REF dp-K024-000003) using biotinylated probes, or a DSN based rRNA depletion as described in the HT-5PSeq approach. See manuscript for use of specific oligonucleotides and sample specific information regarding treatment condition. Please note that the *mixed sample records were created to include the original mixed fastq files in the records, since the unique molecular identifier information might be lost from the headers of the adapter-trimmed split fastq files (generated for each individual sample).