Referee 1's review:


Review of the original version of the manuscript

Fujishima et al. report computational evidence based on phylogenetic analyses that modern archaeal tRNA originated from the combination of ancestral 5' and 3' tRNA half genes.

The performed network analysis clearly shows that 5' and 3' tRNA halves have a different evolutionary background which is in agreement with previous theories on the origin of the tRNA including the wealth of minihelix work. The authors need to make clear (page 3, line 27) that a minihelix is not exactly the investigated 3' tRNA half sequence which includes part of the anticodon stem. This is important as the complementarity between both acceptor and anticodon stem in the tRNA halves should make them less distinguishable than a minihelix in the network analysis. As both anticodon and CCA end were excluded from the halves, the analysis will be centered on the difference between the well-documented high conservation of the T-loop/stem in the 3' half versus the more variable D-loop in the 5' half.

The 1953 mature tRNA sequences were aligned using CLUSTALW. Such an alignment needs to be manually improved to obtain a structural alignment matching the corresponding tRNA domains based on tRNA structure predictions of programs like tRNAScan-SE, Aragorn and SPLITS. Did the authors see any clustering of tRNAs from hyperthermophilic archaea versus mesophilic organisms as the necessary thermostability of the tRNA structure is usually easily observed by its GC-content?
While I agree that the two tRNA halves have a different evolutionary background, the existence of split tRNAs and permuted tRNA genes could still be a modern development in organism with a fast evolutionary tempo as integration and excision of integrative elements is concentrated at tRNA genes.


Review of the first revised version of the manuscript

The manuscript has been improved significantly and is now acceptable for publication in PLoS ONE.

**********
N.B. These are the comments made by the referee when reviewing an earlier version of this paper. Prior to publication the manuscript has been revised in light of these comments and to address other editorial requirements.