The AlgaeOrtho tool visualizes protein orthologs identified between several algal species using SonicParanoid. This can be helpful for identifying bioengineering targets in silico between evolutionarily distant species, and for proteins that have not yet been annotated for function. The goal of the bioengineering is to differentially express (over-express, knock-out or down-regulate) a gene of interest in a species of choice to obtain the desired genotypic outcome. A common method to achieve this is to use the BLAST algorithm to find sequence similarity of the gene of interest to sequences in the genomes of other species (18). One benefit of focusing on proteins rather than gene regions is to identify orthologs with genetic sequences which are not highly conserved. Some sequences might not appear identical at the gene level but encode proteins that perform the same function. This is because multiple codons can encode for the same or similar amino acids so that multiple genetic sequences can end up producing functionally equivalent proteins.
This method allows for the user to look at several different species simultaneously, as the ortholog groups are produced for the algal species the user selects, including protein sequences from non-publicly available or non-annotated proteomes. Since, over half of algal proteins do not have an annotated function, an annotation-agnostic comparative approach enables the identification of relevant bioengineering target regions in un-sequenced and novel algal species (41). The ability to look through as many species as desired at once is beneficial for exploratory analyses that would be too difficult to undertake, one by one, using BLAST. This is a benefit of multi-sequence alignment algorithms like SonicParanoid, versus single-sequence alignment algorithms such as BLAST.
The ability to identify putative orthologs of a protein species of interest showcases one of the key advantages of the AlgaeOrtho tool which can include protein sequences in a multi-sequence alignment even if the entire proteomic data is not available, in addition to identifying alternative species with publicly available omics resources in which bioengineering of a similar target could be conducted. This was demonstrated in the first case study, where the Chlorella HS2 sp. version of bZIP1 had the highest similarity with a putative ortholog from Chlorella variabilis NC64A (“ChlNC64A”), which is a model species for algal-viral interactions (29), and the next highest with Chlorella sorokiniana UTEX 1602 (“Chloso1602”), which is a feedstock species for lipids (32). Previous studies reported that overexpression of a bZIP gene from Chlorella sp. HS2 (for which a full genome sequence is not yet publicly available to the best of our knowledge) and its ortholog in N. salina, produce an over 70% increase in lipid yield in heterotrophic media (20, 25). Notably, Nannochloropsis-contained bZIP1 that was identified as an ortholog of Chlorella sp. HS2 bZIP1 did not show a high sequence similarity with the others (< 33%), despite being an empirically confirmed ortholog (25). This is logical since the Chlorella sp. HS2 is more closely related to the other Chlorella species than to the Nannochloropsis species and would likely have had less evolutionary divergence. It is worth mentioning that although, Chlorella HS2 bZIP1 protein sequence has been reported in the literature (25), the full proteome of this species is not currently publicly available.
Most importantly, these species may be relevant targets for bioengineering, to test if overexpressing the related gene in the identified species (C. variabilis NC64A, C. sorokiniana UTEX 1602 and Chlorella sp. A99) also results in enhanced lipid content similar to that reported in Chlorella sp. HS2 and N. salina. Both species have been explored as potentially relevant for biofuels, so these sorts of studies could help bioengineer the best traits from one species into the other (31, 42, 43). Indeed, bZIPs in plants and algae that have been implicated in halotolerance mechanisms would constitute other possible targets for bioengineering. These bZIPs could be included in the input files and their orthologs could be identified by AlgaeOrtho from a range of algal species. This would be a good next experiment for a bioengineer looking for putative targets across algal species.
The second case study shows that this analytical pipeline is able to identify putative orthologs that have not been identified, since Ochrophyta sequences with high similarity were included, even when not annotated as LCYBs (Fig. 4, Fig. 5). This is logical, since not all proteins are annotated in all species even in resources like JGI that have robust annotations. There were clusters of species with very high sequence similarity, notably between Nemacystus decipiens species (“Nemde1”) (35), Cladosiphon okamuranus S strain (“Claok1”) (44), and Undaria pinnatifida (“Undpi1”) (45) which are brown algae grown for human consumption, and Ectocarpus siliculosus (“Ectsil”) (46), which is a filamentous brown alga. Phylogenetic analysis for LCYB shows that proteins from closely related species tend to have proteins with more-closely aligned sequences. The results from this study would necessarily need to be verified in the algae before claims could be made about the identification of un-annotated LCYB orthologs.
Chlorella (a freshwater alga in the Chlorophyta) and Nannochloropsis (a marine microalga in the Ochrophyta) are diverged, which is reflected in their lower percentage sequence similarity between their bZIP1 proteins (47). But, since these two proteins have been reported to have the same function, it suggests that bZIPs that are more closely clustered may also have the same function (3, 25). That being said low sequence similarities are still possible between true orthologs, and it is possible to have false positives (sequences with higher similarity that are not actually orthologous) (48, 49). Previous studies have reported that low sequence similarities are possible between true orthologs, and protein sequence similarity may not be able determine the function of a protein (48). There is not necessarily a threshold for which amino acid sequence similarity assures orthology, and the practical cut-off for this may depend on the protein of interest. Nevertheless, sequence similarity is still a common way to identify bioengineering targets (48). Thus, although high sequence similarity between two protein sequences could be evidence that two proteins may have the similar functions, not all orthologs have high sequence similarity. Additionally, ortholog groups from SonicParanoid could also include paralogs (the protein underwent a duplication event, and evolved after a speciation event), which tend to have a lower percent sequence similarity (48). More information about how SonicParanoid identifies orthologs and ortholog groups can be found in the paper in which it is described (17).