Thousands of primer-free, high-quality, full-length SSU rRNA sequences from all domains of life

1 Ribosomal RNA (rRNA) genes are the consensus marker for determination of microbial diversity on 2 the planet, invaluable in studies of evolution and, for the past decade, high-throughput sequencing 3 of variable regions of ribosomal RNA genes has become the backbone of most microbial ecology 4 studies. However, the underlying reference databases of full-length rRNA gene sequences are 5 underpopulated, ecosystem skewed, and subject to primer bias, which hamper our ability to study 6 the true diversity of ecosystems. Here we present an approach that combines reverse transcription 7 of full-length small subunit (SSU) rRNA genes and synthetic long read sequencing by molecular 8 tagging, to generate primer-free, full-length SSU rRNA gene sequences from all domains of life, with 9 a median raw error rate of 0.17%. We generated thousands of full-length SSU rRNA sequences from 10 five well-studied ecosystems (soil, human gut, fresh water, anaerobic digestion, and activated 11 sludge) and obtained sequences covering all domains of life and the majority of all described phyla. 12 Interestingly, 30% of all bacterial operational taxonomic units were novel, compared to the SILVA 13 database (less than 97% similarity). For the Eukaryotes, the novelty was even larger with 63% of all 14 OTUs representing novel taxa. In addition, 15% of the 18S rRNA OTUs were highly novel sequences 15 with less than 80% similarity to the databases. The generation of primer-free full-length SSU rRNA 16 sequences enabled eco-system specific estimation of primer-bias and, especially for eukaryotes, 17 showed a dramatic discrepancy between the in-silico evaluation and primer-free data generated in 18 this study. The large amount of novel sequences obtained here reaffirms that there is still vast, 19 untapped microbial diversity lacking representatives in the SSU rRNA databases and that there might 20 be more than millions after all. With our new approach, it is possible to readily expand the rRNA 21 databases by orders of magnitude within a short timeframe. This will, for the first time, enable a 22 broad census of the tree of life. 23 To obtain primer-free and full-length SSU rRNA sequences, we combined and optimized methods for 24 producing full-length SSU rRNA cDNA from total RNA with synthetic long read sequencing enabled by 25 molecular tagging. Full-length SSU rRNA molecules were enriched from extracted total RNA and 26 converted to double-stranded cDNA, enabled by poly(A) tailing and single-stranded ligation, thereby 27 avoiding the use of conventional SSU rRNA PCR primers and the resulting taxonomic bias (Fig. 1A). 28 During first and second strand cDNA synthesis, the individual SSU rRNA molecules are uniquely tagged 29 . CC-BY 4.0 International license peer-reviewed) is the author/funder. It is made available under a The copyright holder for this preprint (which was not . http://dx.doi.org/10.1101/070771 doi: bioRxiv preprint first posted online Aug. 22, 2016;