Our lab 26 has previously demonstrated that nucleotide and amino acid phylograms of 10 vertebrate Rbm45 orthologues recapitulate accepted taxonomic relationships between classes Actinopterygii (ray-finned fishes), Amphibia, Reptilia, and Mammalia; additionally, through NCBI database interro gation, we reported Rbm45 orthologues, both empirically confirmed and Gnomon algorithm predicted (https://www.ncbi.nlm.nih.gov/genome/annotation_euk/process/), across metazoan taxa including animals from the non-Bilateria phyla Porifera (sponges) and Cnidaria (e.g., hydra), and within clade Bilateria phyla from the nephrozoan lineages Protostomia (e.g., phyla Mollusca and Arthropoda) and Deuterostomia (e.g., phyla Echinodermata and Chordata). Molecular phylogenetic analysis has been successfully used to reconstruct the evolutionary history of populations, genes, and proteins; furthermore, it has been utilized to understand genome organization and gene conservation 8182. Therefore, to gain a better understanding of the evolutionary history of Rbm45, a gene involved in neuronal development 27 and neuronal pathogenesis 29, we have expanded our phylogenetic analysis to 36 Rbm45 orthologues from 9 phyla: Porifera, Cnidaria, Brachiopoda, Mollusca, Arthropoda, Echinodermata, Priapulid, Hemichordata, and Chordata. When more than one organism’s Rbm45 sequence was available within a phylum, we often chose those species that are on the IUCN Red List of Threatened Species (e.g., Lipotes vexillifer (Yangtze River dolphin) 83, Loxodonta africana (African savanna elephant) 84), are molecular genetic and developmental biology model systems (e.g., Danio rerio (Zebrafish), Xenopus laevis (African clawed frog), Gallus gallus (chicken), and Mus musculus (house mouse)), or give us multiple subgroups in a taxonomic unit (e.g., class Mammalia: order Monotremata (Ornithorhynchus anatinus: platypus), infraclass Marsupialia (Monodelphis domestica:gray short-tailed opossum), and infraclass Placentalia (Homo sapiens: humans)). Additionally, we attempted to have representative members from a variety of crown clades 85: Ecdysozoa (phyla Priapulida and Arthropoda), Spiralia (phyla Brachiopoda and Mollusca), Ambulacraria (phyla Echinodermata and Hemichordata), and Chordata (subphylum Craniata) 8687.

We used MEGA7 software (Materials and Methods) to build rooted phylogenetic trees using 36 Rbm45 cDNA and amino acid orthologous sequences. The bootstrap consensus tree from 1000 iterations is taken to represent the evolutionary history of the gene 51. In our unbiased tree analysis (data not shown), phylum Porifera resolved as the sister group to all other animals; therefore, phylum Porifera served as the outgroup 888990 in the cDNA and amino acid molecular phylogenies (Figure 1 and Figure 2). As we progress “up” the tree from most ancient to most recent lineages, phylum Cnidaria exhibits paraphyly, with the freshwater polyp (Hydra vulgaris) diverging after corals and anemones as sister group to Bilateria, having 87% and 97% bootstrap support for the evolutionary node in the cDNA and amino acid molecular phylogenies, respectively (Figure 1 and Figure 2). These data are an example of incomplete lineage sorting, where a gene tree does not match the history of the taxa 8291, within clade Cnidaria 92. Where we have sequence from more than one organism in a phylum, the phyla are monophyletic in both the cDNA and amino acid molecular phylogeny (e.g., Mollusca, Arthropoda, Echinodermata, and Chordata). We also observed incomplete lineage sorting in the cDNA molecular phylogeny with clades Protostomia and Deuterostomia exhibiting paraphyly (Figure 1). However, since any bootstrap value less than 70 is considered unreliable 9394, this node, at 17% bootstrap support, splitting the cluster taxa Chordata and Hemichordata away from Priapulida, Echinodermata, Arthropoda, Mollusca, and Brachiopoda, is not well supported. However, the node between non-bilaterians and bilaterians has 87% bootstrap support. In contrast, the amino acid molecular phylogeny (Figure 2) shows Protostomia and Deuterostomia as monophyletic clades with 99% bootstrap support of the node at bilaterian diversification. Like the cDNA molecular phylogeny, the node at the diversification of non-bilaterians and bilaterians has 97% bootstrap support in the amino acid molecular phylogeny (Figure 1 and Figure 2).

Both the cDNA (Figure 1) and the amino acid (Figure 2) molecular phylogenies show phylum Hemichordata as the sister group to phylum Chordata. However, in contrast to recent work 59, neither phylogeny places Hemichordata with Echinodermata (i.e., they are paraphyletic) into clade Ambulacraria, the sister group to clade Chordata, albeit these nodes have weak bootstrap support of 52% and 46%, respectively. Conversely, in the amino acid molecular phylogeny (Figure 2), Echinodermata and Hemichordata, though paraphyletic, do form basal groups to Chordata as expected 598687. Interestingly, both the cDNA and amino acid molecular phylogenies place Zebrafish (Danio rerio; group Osteichthyes; class Actinopterygii), with 100% bootstrap support, as ancestral to all other chordates (subphylum Craniata) including cartilaginous fish (class Chondrichthyes). These data are in opposition to accepted cladograms based on morphological and molecular traits 8795, but support a hypothesis that sharks evolved from a common ancestor having a bony skeleton, with generalized bone loss as a synapomorphy for class Chondrichthyes 96. The coelacanth (Latimeria chalumnae; class Sarcopterygii) has 80% (Figure 1) and 71% (Figure 2) bootstrap support as a sister group to clade Tetrapoda in agreement with the current hypothesis on the evolution of the tetrapod lineage 97. Similarly, our phylogenies have 100% (Figure 1) and 92% (Figure 2) bootstrap support for clade Amniota.

Protostomes are separated into clades Spiralia and Ecdysozoa 86. Within the Rbm45 amino acid molecular phylogeny (Figure 2), which delineates clade Protostomia as monophyletic, all nodes, except for the split between octopus and Pacific oyster within phylum Mollusca, have greater than 70% bootstrap support. However, we observed incomplete lineage sorting of Rbm45 within Protostomia as clades Ecdysozoa and Spiralia do not exhibit monophyly. Instead, the taxa that make up these two clades are interspersed within the protostomes. The phylogenetic incongruences (i.e., conflicting branching orders) and lack of coalescence between our Rbm45 molecular phylogenies and currently accepted species phylogeny is not surprising 8298 and is most likely a product of coalescent stochasticity 8299100 and variations in lineage-specific evolutionary rates over time (heterotachy) 101102. In contrast to other published data 103, we found that our amino acid molecular phylogeny recapitulated accepted species tree topology more closely than the cDNA molecular phylogeny 104. Additionally, we have observed that where there is certainty in taxa resolution (e.g., clades Tetrapoda and Amniota), our data also appropriately resolves these nodes; whereas, where there is uncertainty in the placement of taxa (e.g., polytomy of clade Spiralia), our molecular phylogenies also reflect similar difficulties in node placement consistent with previous work by others 105.

Unfortunately, we were unable to include in our analysis two basal groups, phyla Placozoa and Ctenophora, whose positions in the metazoan phylogeny are disputed 106107. The predicted sequence 108 for the placozoan Trichoplax adhaerans Rbm45 orthologue (Gene ID: 6752484; NW_002060945.1, XM_002110678.1, XP_002110714.1; accessed 2024 February 4) is a partial sequence predicted to have at least three exons encoding 274 amino acids roughly corresponding to amino acids 21-317 (including gaps; data not shown) of human RBM45. Therefore, the hypothetical T. adhaerans Rbm45 orthologue sequence lacks a start codon, RNA-binding domain III, homo-oligomer assembly domain, nuclear localization signal, and a stop codon. However, the presence of at least 3 exons supports the assertion that placozoans are more derived than their simple body plan would imply 106108 (Figure 3). Inclusion of the T. adhaerans partial Rbm45 amino acid sequence into our molecular phylogenetic analysis placed it as a sister group to phyla Porifera and Cnidaria as expected 8687 (data not shown). The nuclear genomes of six species from phylum Ctenophora have been sequenced to date (https://www.ncbi.nlm.nih.gov/search/all/?term=Ctenophora; accessed 2024 February 4); however, an Rbm45 orthologue has not yet been identified. This is unfortunate, as we wished to include this gene product in our analysis considering the hypothesis 107109 that ctenophores are the sister group to all animals, and not poriferans, challenging the predominant paradigm that metazoan nervous systems have evolved complexity in a stepwise manner over deep time 110.

Considering the strong recapitulation of the metazoan tree of life using Rbm45 amino acid sequences, we also analyzed the conservation of protein domains from sponges to humans across the same 36 taxa used in the phylogenetic analysis (Figure 1 and Figure 2). Multiple sequence alignment using the Clustal Omega algorithm demonstrates that Rbm45 protein domains are linearly conserved from sponges to humans in the order: RBD I, RBD II, HOA, RBD III, and NLS (Figure 3). RBDs I, II, and III show 72%, 80%, and 65 % similarity across taxa, respectively, with the HOA domain displaying 68% similarity across taxa analyzed (data not shown). Furthermore, the monopartite NLS 32111, located just downstream of RBD III (Figure 3), from all 36 orthologues conforms to the canonical consensus sequence K(R/K)(X)(R/K). Example NLSs are sponge KRQK, sea anemone KRPR, mosquito KRMR, and human KRQR (data not shown) exemplifying core basic amino acids.

In contrast to the NLS, the human RBM45 NES does not conform to a classical leucine-rich domain but instead was empirically determined 32 to be made up of a clique of two hydrophobic amino acids: leucine-leucine. Additionally, this clique of two hydrophobic residues L(L/I) are conserved among Rbm45 mammalian orthologues 2732. Using multiple sequence alignment (Materials and Methods), we identified the L(L/I) clique only within clade Tetrapoda of clade Craniata. To identify an NES in non-tetrapods, we interrogated the online NES prediction program LocNES 56 using human RBM45 as the input sequence. The algorithm identified a majority-rule canonical sequence from all 19 Craniata taxa (i.e., clade Tetrapoda, clade Chondrichthyes, and group Osteichthyes) between the HOA and RBD III domains 3132 having the general form: R¹⁵K¹⁶MA(T¹⁴/S²)Q(M¹²/L⁷)VA¹⁶A¹⁸Q¹⁸(L¹¹/M⁵/V³)(A¹¹/M⁴)(S¹⁸/T¹)(M¹⁵/V³) where the superscript indicates the number of taxa with that amino acid, if not at identity, with conserved amino acids (e.g., hydrophobic) grouped in parentheses. In tetrapods, this conserved sequence was downstream and immediately adjacent to the L(L/I) clique. In contrast, using sponge or sea anemone Rbm45 as the query sequence, we were unable to deduce a consensus NES in invertebrates (clades Ambulacraria and Protostomia). These data agree with a pairwise alignment of sponge and human Rbm45 amino acid sequences where the Clustal Omega algorithm inserted a gap in the sponge sequence across from the human NES (data not shown). The inability to identify a putative NES in the invertebrate taxa analyzed is not necessarily surprising given the complexity and variability of NES sequences 56112. Furthermore, the LocNES algorithm searched for CRM1 dependent NESs 56. It is possible that invertebrate Rbm45 proteins use a CRM1-independent pathway (e.g., passive diffusion) like vertebrate TDP-43 and FUS 113. If invertebrate Rbm45 exits the nucleus by passive diffusion, then the evolution of an NES would be a novel trait in the Craniata lineage. Future work on this question would necessitate the empirical determination of whether Rbm45 is able to be trafficked from the nucleus to the cytoplasm in these invertebrate organisms.

Concurrent with protein domain conservation analysis, we also analyzed the gene structure of 25 Rbm45 orthologues (Materials and Methods). Rbm45 from the non-bilaterian phylum Porifera has 2 large exons, and from phylum Cnidaria hydra and star coral have 3 exons (Figure 3) and the pale anemone has 4 exons, including a cryptic exon (data not shown). Interestingly, all protein domains are conserved in the first two exons of cnidarian Rbm45 with a similar spacing to sponge (Figure 3 and data not shown). The bilaterian phyla have between 6 and 13 exons, with the Rbm45 protein domains fragmented across multiple exons (Figure 3 and data not shown). Unsurprisingly, though the localization of protein domains occurs in different exons depending on the organism, within each phylum the domains are localized in very similar exon positions. This phenomenon is especially evident in class Mammalia of phylum Chordata where the domain distribution among exons is almost identical (Figure 3 and data not shown) between taxa. Furthermore, regression analysis of the 25 Rbm45 orthologues revealed a statistically significant strong negative correlation between mean exon length and total number of exons (R² = 0.6169, P < 0.0001; Figure 4) in accord with the work of others 114. Our data are in agreement with earlier work demonstrating that more evolutionarily advanced organisms, as measured by genomic and metabolomic complexity 114115116117, have more short exons and longer introns while less evolutionary advanced organisms have fewer large exons and short introns (Figure 3 and data not shown) consistent with whole genome analysis across the five kingdoms 118 Protozoa, Chromista, Plantae, Fungi, and Animalia 114.

We extended our analysis of the evolution of Rbm45 gene complexity by examining the correlation of a taxon’s approximate lineage age, as determined by a robust analysis of current literature (Materials and Methods), to the number of exons in the Rbm45 orthologue. We demonstrate a statistically significant very strong negative correlation (R² = 0.8057, P < 0.0001; Fig. 5) where the most ancient lineages (e.g., Porifera at 650,000,000 years; Cnidaria at 500-570,000,000 years; see Materials and Methods) have the fewest number of exons, between 2 and 4, while generally more recent taxa lineages have more exons (> 9 exons; e.g., Zebrafish at 150,000,000 years; crown-of-thorns starfish at 4,000,000 years; humans at 300,000 years; see Materials and Methods). The majority of taxa followed the trend of fewer exons correlating to an ancient lineage age and more exons to a more recent lineage age. A notable exception is the acorn worm, Saccoglossus kowalevskii, from phylum Hemichordata (clade Ambulacraria) which has an approximate lineage age of 370,000,000 years (Upper Devonian) 59. The reference genomic sequence (NW_003156738.1) of S. kowalevskii predicts 9 exons which is closer in number to what is observed for other members of clade Bilateria we analyzed. However, all other bilaterian taxa that we used in Figure 5 had lineage ages of less than 250,000,000 years (e.g., horseshoe crab with 6 exons). A rigorous analysis of the gene structure of all 36 Rbm45 orthologues used in this study revealed that the Priapulis caudatus (phylum Priapulida; clade Scalidophora) Rbm45 orthologue reference genomic sequence (NW_014578398.1) is predicted to have 17 exons (data not shown), the most of any organism examined by us. Extant priapulins date from the late Carboniferous (~350,000,000 years ago) 119, while extinct stem- and crown-group priapulins are found in the middle Cambrian (~500,000,000 years ago) 120121. These data suggest that division of the Rbm45 gene into many exons (i.e., > 4) occurred relatively early in the adaptive radiation of the evolutionary complex bilateral body plan during and after the Cambrian period 122123. Taken together, these data indicate an ancient origin for Rbm45 in the metazoan lineage. In accordance with this observation, we were able to identify an Rbm45 orthologue in Monosiga brevicollis (Monbr1│Name: e_gw1.8.135.1; Protein ID: 16822; Location: scaffold_8:72832-74415); https://mycocosm.jgi.doe.gov/cgi-bin/dispGeneModel?db=Monbr1&id=16822) from phylum Choanoflagellata, clade Holozoa. These flagellated protists are hypothesized to be the ancestors of phyla Porifera, and thus all metazoans 124125126, further demonstrating the ancient roots of Rbm45. Consistent with this well-accepted hypothesis 86, an unrooted phylogeny using amino acid sequence data from the 36 Rbm45 orthologues from this study plus the Rbm45 orthologue from Choanoflagellata places choanoflagellates at the root of the tree as the sister to all animals (data not shown). This high level of conservation among crown clades reveals that Rbm45 may play a role in neurogenesis across metazoans.