To determine the evolutionary pathway and conservation of PLEKHA7 and its domains, Plekha sequences were collected from the National Center for Biotechnology Information (NCBI) reference protein (RefSeq) database (as of December 2021) using protein-BLAST. Since most Plekha proteins exhibit multiple alternatively spliced variants, we used as queries the longest sequence-verified protein isoform (Supplementary Tables S1A,B). The BLAST searches were performed using default parameters and individually for the major taxonomic databases to ensure the collection of all available proteins. Specifically, mammals were searched separately, followed by birds, ray-finned fishes (Actinopterygii), reptiles, amphibia, cartilaginous fish (Chondrichthyes), and all other vertebrate and invertebrate taxa with available genome sequences at NCBI (Supplementary Table S1A). Additionally, we used the ENSEMBL database to search for Hagfish genes (Genome assembly: Eburgeri_3.2; GCA_900186335.2). Because mammals, birds, and ray-finned fishes have disproportionally more species’ genomes sequenced compared to other taxa, we also narrowed our search in a set of representative species, when collecting sequences for phylogenetic tree construction (Supplementary Table S1A).

These BLAST searches retrieved several homologous sequences, including multiple isoforms, verified or putative, across most species (Supplementary Tables S1B, S2–S9). From this original sequence pool in subsequent analyses, we again retained the largest verified (or predicted) Plekha isoform to avoid redundancy. Overall, protein sequences with alignments producing the lowest E-values (denoting nonrandom alignment between query and target sequences), query coverage greater than 60%, and similar domain organization as the query proteins, were collected. Additionally, to identify common ancestors of these genes in other Chordates or invertebrates, sequences with lower coverage, often limited to the PH domain or to the non-PH-containing regions, were also collected and further analyzed.

The selected sequences were then aligned with the Multiple Alignment using Fast Fourier Transform (MAFFT) program version 7 (Katoh et al., 2002), using the E-INS-i strategy with a BLOSUM62 scoring matrix, a gap opening penalty of 1.53, and a 0.0 offset value. Alignments were also in part performed using the SnapGene software (v5.2.1) (from Insightful Science; available at snapgene.com). Domain identification was performed using the conserved domains database (CDD) (Marchler-Bauer et al., 2015) and published work. Multiple sequence alignments were used to generate sequence conservation logos with the WebLogo software (Crooks et al., 2004). Intrinsically Disordered Region (IDR) predictions were performed at the PSIPRED server (McGuffin et al., 2000; Buchan and Jones, 2019) using the DISOPRED3 function (Jones and Cozzetto, 2015).

To determine the phylogenetic relationships of the collected sequences, we generated Neighbor-Joining (NJ) and Maximum likelihood (ML) phylogenetic trees. After analyzing and verifying all sequences’ relationships, different protein sequences were selected in the final dataset as representative sequences (species) based on whether they represented complete sequences and their alignment did not introduce more than 20% of the total gaps.

Maximum-likelihood (ML) was used to find the best model of evolution (MEGA X) (Kumar et al., 2018). Based on the Bayesian Information Criterion (BIC) the substitution pattern was best described by the JTT model with corrections for non-uniform evolutionary rates among sites (+G) and by assuming that a certain fraction of sites is evolutionarily invariable (+I). Phylogenetic trees were generated using the NJ and ML algorithms as implemented in MEGA X. For ML, initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using the JTT model, and then selecting the topology with superior log likelihood value. In both cases, one thousand bootstrap pseudo-replicates were used to test the reliability of the inferred trees. Phylogenetic trees were constructed using a midpoint root and without the use of other outgroups, as previously proposed for families with highly divergent outgroups, which is the case for the PLEKHA family (Wheeler, 1990; Foster and Hickey, 1999; Huelsenbeck et al., 2002; Shavit et al., 2007; Rota-Stabelli and Telford, 2008; Rosenfeld et al., 2012).

To verify orthology and support homology between the selected sequences, intron position and phase conservation were assessed. Both analyses were performed using in-house scripts assembled to a pipeline (www.github.com/bdighera/orthotracer). The intron position and phase for each collected Plekha gene was determined by using Spidey to align genomic sequences and coding sequences collected from GenBank (Wheelan et al., 2001). Raw sequence data were collected via the RefSeq nuccore database API (application programming interface) using biopython version 3.79 Entrez functions. Then, the introns were mapped on a protein multiple sequence alignment (MSA) to illustrate their position along the molecules. The intron phase (0, no codon split by the intron; 1, codon split by intron at the first codon position; 2, codon split by intron at the second codon position) was calculated by the modulo of exon lengths and denoted with different colors.

To determine whether the synteny observed between Plekha and its neighbor genes in humans is conserved in other species, the first four genes on either side, excluding RNA coding genes, and up to 500 kbps from the parent Plekha gene/locus were collected, mapped, and analyzed using the RefSeq genomic information. Other established technology leveraged in the in-house scripts included Clustal 1.2.1 (Sievers et al., 2011) to generate the MSA using default parameters, and the ete3 package for rendering the genomic synteny and intron phase visualizations (Huerta-Cepas et al., 2016; Sievers and Higgins, 2018; Sievers et al., 2020). Final figures were generated using Adobe Illustrator 25.2 and Affinity Designer 1.10.

Protein structural predictions were collected from the AlphaFold Protein Structure Database (Jumper et al., 2021). Figures containing individual three-dimensional models or superimposed structural models were generated using Pymol ™ 1.7.4.5 (Delano Scientific). The models were drawn in cartoon representations and structural alignments were performed using the function “super” in pymol, which executes a sequence independent structure-based dynamic programming alignment between structural elements.

To investigate the evolution and phylogenetic position of Plekha7, we performed BLAST searches using the human PLEKHA sequences as queries against the NCBI protein database for all major metazoan taxonomic groups (Figures 1A,B; Supplementary Table S1A ). In these searches, we considered low E-values (ranging from 0 to 10⁻²⁰) and query coverage >60% as criteria suggesting full-length homology and potential orthology.

Based on these criteria, we consistently identified full-length homologous sequences in all available vertebrate and most invertebrate genomes for the Plekha1, 2, 3, 8 proteins (Figures 1A,B; Supplementary Tables S2–S5). However, the Plekha4, 5, 6, 7 proteins showed consistent full-length homologies only in vertebrates (Figures 1A,B; Supplementary Tables S6–S9). More specifically, searches using the human Plekha4, 5, 6, 7 sequences as queries identified putative orthologs in almost all vertebrate classes examined, including the jawless vertebrate lampreys and hagfish (Figures 1A,B ; Supplementary Tables S6–S9). BLAST searches failed to identify putative Plekha4 orthologs in birds, and in multiple Actinopterygii species (Figure 1B; Supplementary Table S6). Overall, Plekha4 exhibited lower sequence conservation among vertebrates, compared to Plekha5, Plekha6, or Plekha7 (Figures 1A,B ; Supplementary Table S6). Notably, the functionally documented PLEKHA6 region coding for the WW domain, (Sluysmans et al., 2021b), was missing from many Plekha6 hits from different species, possibly due to isoform specificity or incomplete annotation (Figure 1A; Supplementary Table S8).

Contrary to vertebrates, BLAST searches using the Plekha4, 5, 6, 7 sequences as queries against invertebrate genomes, resulted in hits with relatively high E-values and low coverage. Most of these alignments were restricted to the N-terminal region containing the PH domain with a few exceptions (Figures 1A,B ; Supplementary Tables S6–S9–see related taxa-specific tabs). These exceptions included sporadic alignments restricted to the coiled-coiled domains of Plekha7 and Plekha5 or to the extreme C-terminus region of Plekha7, Plekha6 in a few evolutionarily unrelated invertebrates. Similar results were also obtained when we excluded the PH domain and repeated the BLAST searches against invertebrate genomes (Figures 1A,B ; Supplementary Tables S6–S9–see related taxa-specific tabs).

To further investigate the lack of full-length homologs in invertebrates, we performed BLAST searches using a protein sequence (XP_035672282.1) of the cephalochordate Branchiostoma floridae (invertebrate, chordate), as the closest living relative of both vertebrates and non-chordate invertebrates. These searches produced alignments with relatively higher identity scores and from more species in non-chordate invertebrate genomes, compared to the human Plekha4,5,6,7 searches, but were still overall sporadic and limited to the C-terminus (Supplementary Table S10).

These results prompted us to also examine a Drosophila gene named kramer, which has been reported as a homolog of PLEKHA4 (Shami Shah et al., 2019). BLAST searches using the human PLEKHA4 against the insect and Drosophila databases (Supplementary Table S6; see related tabs) showed that the homology between PLEKHA4 and kramer, or other dipteran sequences, was restricted to the PH domain. BLAST searches using kramer as query to search chordate databases retrieved Plekha7 sequences as top hits, as well as several Plekha6, Plekha5, and Plekha4 hits. Reciprocal BLAST searches using the human PLEKHA5,6,7 sequences to query Drosophila databases also retrieved kramer as the top hit (Supplementary Tables S6–S9; Supplementary Table S11). Collectively, these results suggest that kramer is not a Plekha4 ortholog, rather the protein contains a PH domain that is closest to the PH domain of all Plekha4,5,6,7 vertebrate proteins.

Overall, searches using Plekha1 retrieved Plekha2 sequences (and vice versa) and they both retrieved the same single gene in invertebrate species (Figure 1B; Supplementary Tables S2,3). Similarly, Plekha3 searches retrieved Plekha8 sequences (and vice versa) and both retrieved the same single member in each invertebrate species (Figure 1B; Supplementary Tables S4,5). Furthermore, each of the Plekha4, Plekha5, Plekha6, Plekha7 BLAST searches retrieved sequences from the other three members in vertebrates, and all four retrieved only one single sequence per species in invertebrates (Figure 1B; Supplementary Tables S6–S9 ). Additionally, we did not retrieve any alignments between Plekha1,2 and Plekha3,8, or between any of those and Plekha4,5,6,7, either in vertebrates or in invertebrates (Figure 1B; Supplementary Tables S2–S9). Taken together, the BLAST results suggest that there are at least three distinct sets of genes within the PLEKHA family, the Plekha1/2, Plekha3/8, and Plekha4/5/6/7 and that PLEKHA might not be a single family. To further determine the validity of these patterns we performed phylogenetic analysis using the retrieved Plekha proteins.

To investigate the phylogenetic relationships within the PLEKHA family, we generated multiple sequence alignments and constructed phylogenetic trees using Plekha sequences from representative species from all Metazoans retrieved from our BLAST searches (Figure 1B; Supplementary Tables S1–S10). Because Plekha1/2 and Plekha3/8 orthologs were identified throughout metazoa, we included representative species from both vertebrates and invertebrates. In the case of the Plekha4/5/6/7 sequences, we included representative vertebrates and those invertebrate sequences that showed homologies in both the PH and non-PH regions (Figure 1B; Supplementary Tables S6–S10). These phylogenetic analyses clearly separated the PLEKHA family into three distinct clades: the Plekha4/5/6/7, the Plekha1/2, and the Plekha3/8 clades (Figure 2A; Supplementary Figure S1 ). However, confidence was overall low in several internal nodes, since comparisons were based on the PH domain sequence, the only homologous region between all the members of the PLEKHA family.

Within the Plekha1/2 clade, Plekha1 and Plekha2 formed distinct groups within vertebrates, while all invertebrate species studied contained a single member, misannotated in NCBI as either 1 or 2 (Figure 1B, Figure 2A; Supplementary Figure S1). Furthermore, the Plekha1 and Plekha2 sequences follow the species’ phylogeny and were found in almost all species searched (Figure 1B, Figure 2A, Supplementary Figure S1). Together, these findings suggest that the vertebrate Plekha1 and Plekha2 genes emerged after the duplication of a single Plekha1/2 gene that preexisted in invertebrates.

Within the Plekha3/8 clade, Plekha3 and Plekha8 form two distinct groups in vertebrates while invertebrates contain only a single member, misannotated as either Plekha3 or Plekha8 (Figure 1B, Figure 2A; Supplementary Figure S1). The invertebrate Plekha3/8 branching pattern differs from the established species phylogeny although the topology is supported by very low bootstrap scores. Still, results within the Plekha3/8 clade support the idea of a gene duplication event in the last common ancestor of vertebrates that gave rise to Plekha3 and Plekha8 from an ancestral single invertebrate Plekha3/8 gene.

Within the Plekha4/5/6/7 clade, the vertebrate sequences grouped separately from the invertebrate ones (Figure 2A; Supplementary Figure S1 ). Within the vertebrate Plekha4/5/6/7 clade, each of the jawed vertebrate (gnathostome) Plekha4, Plekha5, Plekha6 and Plekha7 sequences formed distinct clades (Figure 2A; Supplementary Figure S1 ). However, the three Plekha4/5/6/7 sequences from the jawless vertebrate (agnathan) Petromyzon marinus didn’t follow the same pattern: one of them formed an outgroup to the jawed vertebrate Plekha7 clade and two formed a distinct clade as an outgroup to the jawed vertebrate Plekha5-Plekha6 clade. Notably, in both cases, clustering of the Petromyzon sequences within the jawed vertebrate clades was with relatively low bootstrap values and probably resulted in poor bootstrap values for some of the jawed vertebrate Plekha4,5,6,7 clades (e.g., Plekha4). To resolve the Plekha4/5/6/7 relationships within jawed vertebrates with higher confidence, we excluded the invertebrate and Petromyzon sequences and reconstructed the phylogenetic tree (Figure 2B). Indeed, this approach resolved the jawed vertebrate Plekha4,5,6,7 relationships and provided topologies supported by higher bootstrap values. More specifically, the jawed vertebrate Plekha5 and Plekha6 grouped together, whereas Plekha7 and Plekha4 formed a separate clade. The Plekha5/6 clade exhibits the largest distance from the root, with Plekha5 and Plekha6 clades exhibiting similar distances after their split, suggesting a pattern of divergent evolution (Figure 2B; Supplementary Figure S2). Differently, the Plekha4/7 clade exhibits the shortest branch distance from the root with the Plekha4 and Plekha7 showing different divergence patterns (Figure 2B; Supplementary Figure S2). This phylogenetic pattern is consistent in all jawed vertebrate taxa examined, indicating that putative gene duplications may have occurred in the last common ancestor of jawed vertebrates (Figure 2B; Supplementary Figure S2).

The supposition that the Plekha4/5/6/7 genes are products of duplications in jawed vertebrates is supported by the fact that sequences of the jawless vertebrate Petromyzon marinus, were all consistently placed outside the jawed vertebrate gene-specific clades (Figure 2A; Supplementary Figure S1). This topology predicts that the lamprey sequences might represent lineage specific duplications and that the common vertebrate ancestor contained a single Plekha4/5/6/7 gene. To further evaluate this prediction, we performed phylogenetic analyses including the single Plekha4/5/6/7 sequence (XP_035672282.1; Plekha5-like) from the cephalochordate Branchiostoma floridae, which we considered a proper outgroup for two reasons. First, this sequence was retrieved in all Plekha4,5,6,7 BLAST searches; and second, cephalochordates are considered the closest living relative of the common ancestor of all chordates. Indeed, when we included this sequence, it branched out as an outgroup to all Plekha4,5,6,7 members, including the Petromyzon sequences (Supplementary Figure S3). A potential interpretation of these results is that a single ancestral Plekha4/5/6/7 gene existed in the last common ancestor of all chordates and then multiplied and diversified in the present-day members of the family in vertebrates. Corroborating this notion, all four PLEKHA4,5,6,7 BLAST searches also identified a single Plekha4/5/6/7 sequence in the Tunicates Phallusia mammillata (CAB3264938.1) and Ciona robusta (Ciona intestinalis XP_026694274.1) (Supplementary Tables S6–S9). However, we didn’t include these sequences in subsequent analyses, due to either exhibiting poor similarity or being partially annotated.

Phylogenies of the vertebrate Plekha4,5,6,7 proteins overall follow the established species phylogenies (Figure 2B; Supplementary Figure S2, Supplementary Figures S4–S7). Notably, Plekha7 and Plekha5 Actinopterygii orthologs separate into two subgroups (Figure 2B; Supplementary Figures S4,6), each including two distinct copies of each gene (Plekha7a and Plekha7b; Plekha5 and Plekha5-like; renamed as 5a and 5b). Of the two Plekha7 copies in zebrafish, Plekha7a is the copy previously referred to as Hadp1 (Wythe et al., 2011) (NP_001129715.1; Supplementary Figure S4). Differently, the Actinopterygii Plekha6 and Plekha4 orthologs seem to exist in only one copy (Supplementary Figures S5,7). Plekha4 is also missing from many Actinopterygii species examined, suggesting secondary loss of this gene in these species, similarly to its loss in all birds (Figure 1B; Supplementary Figure S7).

Consistent with the idea of a single Plekha4/5/6/7 gene in chordates, the non-chordate sequences used in our analyses formed outgroups to the chordate clade including the Drosophila kramer (Figure 2A). In some taxonomic groups, e.g., Lophotrochozoa or Eleutherozoa, the invertebrate Plekha4/5/6/7 members follow the species phylogeny, and these branches are supported by high bootstrap values. However, in most other cases the branching pattern was supported by very low bootstrap values (Figure 2A). This issue is most probably due to the short homologies, restricted in the PH domains, and due to the fact that these sequences were sporadically found in invertebrates and represent phylogenetically very divergent species (Figure 1B; Supplementary Tables S6–S10).

Taken together, our phylogenetic analyses revealed the presence of three distinct PLEKHA subfamilies in Metazoans, namely PLEKHA1/2, PLEKHA3/8, and PLEKHA4/5/6/7. Additionally, these analyses outline an evolutionary pattern, where the individual members within each of the three subfamilies emerged in vertebrates from a single ancestral invertebrate member. However, this analysis could not fully reconstruct the evolutionary history of the whole Plekha4/5/6/7 protein sequence, which contains an extensive non-PH region that is present in all vertebrates but absent from most invertebrates.

To substantiate the above-described evolutionary patterns and clarify the origins of the Plekha4/5/6/7 vertebrate genes, we sought to investigate their exon-intron organization and syntenic relationships, as performed previously for other gene families (Nikolaidis and Nei, 2004; D'Aniello et al., 2008; Coppola et al., 2019; Nguyen et al., 2020).

Exon-intron organization analysis showed that the three major clades identified in our phylogenetic analyses (Plekha4/5/6/7, Plekha1/2, and Plekha3/8; Figure 2A) have different intron positions and phase (Figure 3). These differences are apparent even within the N-terminus, which includes the PH domain, the only common denominator between all Plekha members (Figure 1A), further underlying the distinct origin and evolution of these three phylogenetic clades.

In contrast, intron position and phase are highly conserved within each phylogenetic clade (Figure 3). In the case of the Plekha1/2 and Plekha3/8 genes, this conservation extends in almost all animal species studied. In the case of the Plekha4/5/6/7 genes, most of the introns were conserved between and within the specific gene clades across vertebrates (Figure 3; Supplementary Figure S8). This conservation (in both the N- and C-termini) is also apparent in the lamprey Petromyzon marinus genes (Figure 3; Supplementary Figure S8). Furthermore, the vertebrate Plekha4/5/6/7 exon-intron organization is also conserved in both the N-terminus and partially in the C-terminus in the single Plekha4/5/6/7 gene from the cephalochordate Branchiostoma floridae. Although many intron positions are also conserved in the other invertebrate sequences analyzed, the phase of these introns is quite different, suggesting changes in the open reading frame of the resulting proteins (Figure 3; Supplementary Figure S8). These partially conserved intron positions in the C-terminus of the invertebrate genes coincide with the coiled-coil domain and the extreme C-terminal domain sequences identified in our BLAST searches (Figure 1A). These findings support the prediction that a single ancestral gene that contained a conserved set of introns, gave rise to the vertebrate Plekha4/5/6/7 clade. In contrast, exon-intron organization of kramer was entirely different from any of the Plekha genes examined, further indicating its high diversification and ambiguous origin (Figure 3; Supplementary Figure S8).

Syntenic analysis showed that the Plekha4/5/6/7 genes share no common genomic neighborhood with the remaining of the PLEKHA family. Instead, the other two subgroups of Plekha paralogs exhibit syntenic relationships: the Plekha1–Plekha2 paralogs with Tacc2–Tacc1 and Htra1–Htra4 paralogs and the Plekha3–Plekha8 with the Fkbp7–Fkbp14 (Supplementary Figure S9). Together with our phylogenetic analyses above, these data support the idea of a separate origin and evolution of the PLEKHA4/5/6/7 subfamily from the PLEKHA1/2 and the PLEKHA3/8 ones. Some of the Plekha4/5/6/7 genes are located on well-studied paralogons (Lagman et al., 2013; Ocampo Daza and Larhammar, 2018), one of which corresponds to paralogon “D” in the reconstruction of the ancestral chordate genome based on the amphioxus genome (Putnam et al., 2008). Similarly, the chromosomal regions around the Plekha3/8 genes correspond to the “C” ancestral chromosome, and the regions around the Plekha1/2 genes correspond to the “E” ancestral chromosome (Putnam et al., 2008).

Syntenic analyses and comparisons between the neighboring genomic regions of the four Plekha4/5/6/7 genes further solidified orthology between the different members and revealed important clues on their origin and the gene duplications that generated them (Figure 4; Supplementary Figures S9–S10). More specifically, we observed several conserved syntenic blocks in jawed vertebrates, such as 1) Plekha7-Sox6 and Plekha6-Sox13; 2) Plekha7-Pik3c2a, Plekha6-Pik3c2b, and Plekha5-Pik3c2g; 3) Plekha7-Ppp1r15b, Plekha6-Ppp1r15b, and Plekha4-Ppp1r15a; and 4) Plekha7-Nucb2-Kcnj11 and Plekha4-Nucb1-Kcnj14. The Plekha7-Pppr15b, Plekha7-Alx4, and Plekha6-Alx3 blocks are conserved only in Chondrichthyes and Actinopterygii, but subsequently lost in Coelacanths and Tetrapods (Figure 4; Supplementary Figures S9–S10). Overall, these results show that Plekha7 exhibits the highest number of conserved syntenic blocks compared to its paralogs and that it is the only gene exhibiting conserved syntenic blocks with all the three members of the PLEKHA4/5/6/7 subfamily in jawed vertebrates.

Genomic neighborhood analyses revealed that the lamprey genes also displayed conserved synteny with jawed vertebrate Plekha4,5,6,7 genes. Specifically, a Petromyzon marinus gene (XP_032804545) is in synteny with Sox5, similarly to Sox6-Plekha7 and Sox13-Plekha6, and with Kcna1, similarly to Plekha7-Kcnj11 and to Plekha4-Kcnj14 in jawed vertebrates (Figure 4; Supplementary Figures S9–S10). A second Petromyzon marinus gene (XP_032815647) is in synteny with Insc and with Slco1c1, like the Plekha7-Insc and the Plekha5-Slco1c1, respectively, in teleosts (Figure 4; Supplementary Figures S9–S10). Lastly, the third Petromyzon gene we identified (LOC116956434; XP_032833925.1) does not exhibit any conserved synteny, either with the other two Petromyzon paralogs, or with any of the jawed vertebrate ones (Figure 4; Supplementary Figures S9–S10). These mixed syntenic relationships suggest that neither of these Petromyzon genes are clear Plekha4,5,6,7 orthologs, further supporting the idea that these sequences are the result of lineage-specific duplications of a single gene, after the split of the jawed and jawless vertebrates. For this reason, we renamed these genes as Plekha4/5/6/7a,b,c (Figure 4 and throughout).

Analysis of the lancelet Branchiostoma floridae Plekha4/5/6/7 gene revealed synteny with Liprin-a, which is conserved to the synteny observed between different Plekha4/5/6/7 and Liprin genes in Petromyzon and in many teleosts (Figure 4; Supplementary Figures S9–S10). No syntenic relationships are conserved between non-chordate invertebrate Plekha4/5/6/7 homologs and chordate genes of the subfamily (Supplementary Figure S9). Together, these findings indicate that the Plekha4/5/6/7 genes originated independently in jawed and jawless vertebrates from a single ancestral gene that existed in the last common ancestor of all chordates.

The sequence conservation pattern between the members of the Plekha4/5/6/7 subfamily in vertebrates prompted us to investigate sequence similarities that might infer functionally important regions. The human PLEKHA7 is the best-studied member of the family, with well-identified domains that interact with cell-cell junction components, such as PDZD11, afadin, p120, Nezha, and CGNL1 (Figure 5A) (Meng et al., 2008; Pulimeno et al., 2011; Kurita et al., 2013; Guerrera et al., 2016; Sluysmans et al., 2021b). Amino acid conservation analysis revealed that the binding domains of PLEKHA7 with afadin, p120, Nezha, and CGNL1 are highly conserved in this protein throughout vertebrates (Figure 5B). However, comparisons of the Plekha4, Plekha5, and Plekha6 sequences aligned to p120, Nezha, and CGNL1 domains of Plekha7 yielded overall low amino acid conservation except for specific amino acid sites, which were conserved in all proteins (Figure 5C). Sequence conservation of the PH domain, which overlaps with the afadin-binding domain in PLEKHA7, was to some extent maintained across Plekha4/5/6/7 proteins, but to a far lesser extent across all Plekha proteins, further demonstrating the sequence divergence of the PLEKHA4/5/6/7 subfamily from the rest of the PLEKHA family (Figure 5C; Supplementary Figure S11A).

Notably, Plekha7 exhibits strong conservation in its C-terminus, specifically within its proline-rich and coiled-coiled domains (Figure 5B). Parts of these C-terminal proline-rich regions are also relatively well conserved across all Plekha4/5/6/7 members in vertebrates (Figure 5C). These regions are of unknown function in all Plekha4/5/6/7 members.

Proline-rich domains are a typical characteristic of intrinsically disordered regions (IDRs) (Jarvelin et al., 2016). Recently, it has been reported that PLEKHA4 contains an IDR (Shami Shah et al., 2019). The significant conservation of the C-terminal regions, including the proline-rich regions, across Plekha4/5/6/7 members prompted us to investigate whether these regions are predicted to be IDRs. Disorder analysis revealed that all four Plekha4/5/6/7 contain two long IDRs, one between the PH domain and the coiled-coiled domain, and a second one after the coiled-coil domain, covering most of the C-terminal region (Figure 6A). The disorder predictions were almost identical in Plekha4/5/6/7 members from other vertebrates, such as mouse, zebrafish, or the jawless lamprey Petromyzon marinus (Figure 6B).

Disorder prediction analysis verified the presence of the first IDR in the Branchiostoma floridae Plekha4/5/6/7 sequence (Figure 6B). However, a sequence immediately adjacent to the C-terminus of the Plekha4/5/6/7 sequence is annotated as partial Plekha7-like sequence (XP_035672283.1; gene LOC118413180) (Supplementary Figure S11B). BLAST searches of this sequence revealed that it aligns considerably well to the vertebrate Plekha4/5/6/7 C-terminal sequences (Supplementary Table S10). Although these findings suggest that this partial sequence is the result of misannotation and could be part of the full-length lancelet Plekha4/5/6/7 protein, we could not reannotate the region to provide an open reading frame that includes both genes. Nevertheless, when we artificially merged the two protein sequences and performed the disorder prediction analysis, we were able to identify the missing C-terminus IDR of the lancelet XP_035672282.1 sequence (Supplementary Figure S11C). These data provide further, indirect support for the lancelet Plekha4/5/6/7 gene being a representative of the last single common ancestor of the vertebrate Plekha4,5,6,7 genes.

In contrast, disorder analysis of the few invertebrate sequences with homologies at their non-PH domain regions (Figure 1B ; Supplementary Tables S6–S9), exhibited different patterns and low IDR predictions except for a region towards the extreme end of their C-terminus (Figure 6C). This region corresponds to the one found using BLAST searches (Figure 1A) and intronic analyses (Figure 3), suggesting that this may be a remnant of the ancestral invertebrate Plekha4/5/6/7 member. Still, disorder analysis of the Drosophila kramer produced entirely distinct patterns, compared to the chordate Plekha4/5/6/7 proteins, further underlining kramer’s sequence diversification (Figure 6C). Finally, the other members of the PLEKHA family (Plekha1, 2, 3, 8) showed lower disorder domain predictions, confirming their evolutionary and functional divergence from the PLEKHA4/5/6/7 subfamily (Supplementary Figure S11D).

Computational predictions of the 3D structures of the human PLEKHA4/5/6/7 proteins confirm the presence of long IDR domains, which seem to engulf the structurally defined protein regions (Figure 7). Furthermore, structural superimposition of the different members reveals that in addition to the PH domain, the only other fully aligned region is the coiled-coiled domain in the C-terminal region of PLEKHA7 (Figure 7). This region is conserved in invertebrates (Figure 1A, Figure 3) and overlaps with the domain that: 1) PLEKHA7 binds to Nezha and the microtubules (Figure 5A) (Meng et al., 2008); 2) is critical for the apical junctional localization of PLEKHA7 and PLEKHA6 (Sluysmans et al., 2021a); and 3) mediates PLEKHA4 oligomerization (Shami Shah et al., 2019). Together, these results further support the idea that this region has been retained from the common invertebrate ancestor of the PLEKHA4/5/6/7 family, due to its functional significance.