Phylogenetic Validation of Mitochondrial Proteins

To confirm the taxonomic identity of the putative mitochondrial-related protein identified in P. canceri and eliminate the possibility of residual contamination, maximum-likelihood phylogenetic trees were constructed (supplementary fig. S5, Supplementary Material online). Except for the mitochondrial ABC transporter gene (atm1), the phylogenetic analysis workflow was performed as follows. All mitochondrial-related proteins identified in P. canceri were queried against the NCBI nr database (August, 2020) with BLAST v.2.1.9 (Altschul et al. 1990 (link)) using the BLASTP algorithm. The top 5,000 hits with an e-value less than 1e-10 (or 1e-5 if few hits were identified) were retrieved and clustered at 90% identity with CD-HIT v.4.8.1 (Edgar 2010 (link)). The predicted proteomes of M. mackini and C. pagurus were searched with BLASTP to retrieve homologous proteins. Lastly, a reciprocal BLASTP in all P. canceri predicted proteoms was performed. The sequences were aligned (Mafft v.7.407 (Katoh and Standley 2013 (link)), mafft-auto). The alignments were trimmed of ambiguous sites with (trimAL v.1.4.1 (Capella-Gutierrez et al. 2009 (link)), -automated1). The amino acid substitution model was determined with IQ-TREE2.1.6.5 using the default settings (Kalyaanamoorthy et al. 2017 (link)). Phylogenies and 1,000 ultrafast bootstrap trees with 1,000 SH-aLRT replicates were constructed with IQ-TREE2 v.1.6.5 (Minh et al. 2013 (link)). These initial phylogenies were visualized in FigTree v.1.4.4 and manually pruned to reduce the number of taxa. The reduced data set was aligned (Mafft v.7.407 (Katoh and Standley 2013 (link)), mafft-linsi). Removal of ambiguous sites, evaluation of amino acid substitution models, and phylogenetic reconstruction proceeded as above. For the putative atm1 transporter, a Hidden Markov Model profile for orthologous group KOG0057 (retrieved from EggNOG 5.0.0 (Huerta-Cepas et al. 2019 (link)) database) was used to retrieve the protein models of P. canceri and M. mackini using the default settings of with hmmsearch. The resulting hits were used as queries against the NCBI nr database (August 2020) as described above. This dataset was supplemented with atm1 sequences reported previously (Freibert et al. 2017 (link)). The proteins were aligned with hmmalign from HMMER v.3.2.1 (http://hmmer.org/) and the Atm1 phylogeny was performed as described above.