A Survey on Tubulin and Arginine Methyltransferase Families Sheds Light on P. lividus Embryo as Model System for Antiproliferative Drug Development

Tubulins and microtubules (MTs) represent targets for taxane-based chemotherapy. To date, several lines of evidence suggest that effectiveness of compounds binding tubulin often relies on different post-translational modifications on tubulins. Among them, methylation was recently associated to drug resistance mechanisms impairing taxanes binding. The sea urchin is recognized as a research model in several fields including fertilization, embryo development and toxicology. To date, some α- and β-tubulin genes have been identified in P. lividus, while no data are available in echinoderms for arginine methyl transferases (PRMT). To evaluate the exploiting of the sea urchin embryo in the field of antiproliferative drug development, we carried out a survey of the expressed α- and β-tubulin gene sets, together with a comprehensive analysis of the PRMT gene family and of the methylable arginine residues in P. lividus tubulins. Because of their specificities, the sea urchin embryo may represent an interesting tool for dissecting mechanisms of tubulin targeting drug action. Therefore, results herein reported provide evidences supporting the P. lividus embryo as animal system for testing antiproliferative drugs.


Introduction
It is well known that different αand β-tubulin isotypes contribute to create an evolutionary conserved network of highly dynamic filaments, with key roles in cellular architecture and physiology [1].
The αand β-tubulin dimers linearly assemble to create polarized proto-filaments with β-tubulin subunits exposed to the solvent at the plus end, while the minus-end is capped by α-tubulin subunits [2,3]. Classically, 13 proto-filaments are known to assemble into an MT, a cylinder of 22 nm in diameter. MTs are responsible for several aspects of cell division and proliferation as well as development and tissue differentiation, signalling pathways and gene expression modulation [4][5][6]. Beyond the role in organizing chromosome movements during cell divisions [7], MTs interacting with kinetochore are responsible for the mechanism of genome surveillance known as spindle assembly checkpoint [8]. Changes in kinetochore MT stability often result in chromosome segregation errors, aneuploidy and tumorigenesis [9,10]. The P. lividus αand β-tubulins are organized in an N-terminal GTPase domain with the canonical tubulin signatures GGGTGSG mapping the amino acid residues 142-148 in αand 140-146 in β-tubulins, respectively, and a C-terminal domain connected by a central helix (Figures 1 and 2). Moreover, tubulins possess an acidic and flexible Carboxy-terminal tail of about 10-15 amino acid residues that in the MTs decorates the external surface and through which MTs interact with microtubule-associated proteins (MAPs) [59].
A computational search for PTMs identified several conserved Ser/Thr phosphorylation recognition sites including S48 and S439 in the α-tubulin chains as well S40, S172, T55, T285 and T290 in β-tubulins (Figures 1 and 2). Even if the functions of these PTMs have remained almost elusive for years, new lines of evidence suggested a control of the dimerization dynamic and a role in the MT behaviour during cell division, especially for β-tubulin phosphorylation at S172 by cyclin-dependent kinase 1 [19,60].
Acetylation of α-tubulin on lysine 40 represents a common PTM especially found on stable MTs [61,62], making microtubules more resistant to mechanical forces [63,64]. Our analyses confirmed the presence of a specific amino acid residue (K40) acetylation/deacetylation site in all α-tubulin ( Figure 1) but not β-tubulin isoforms. Similarly, only the α-tubulins contain a Tyr residue (Y451/453) at the C-terminus which may undergo detyrosination/tyrosination cycles corresponding to assembled or disassembled MTs [20,62]. Interestingly, detyrosination guides chromosome congression during   A MSA analysis confirm the high similarity among members of α-and β-tubulin family and that the main differences located at the C-terminal tail may correspond to the differential isotype class involved in specific heterodimers which in turn may affect MT properties. Additionally, it appears that α isotypes accept more residue variations in the corresponding sequence. An analysis of amino acid substitutions shows that, both in α and β isotypes, about 50% of variability occurs in α helix or β sheet secondary structures. The additional variations showed by α isotypes mainly involve amino acid residues structured in α helix. This is not surprisingly since α helices were already found underrepresented in conserved regions [81,82].

P. lividus α-and β-Tubulin Gene Organization
To characterize the α-and β-tubulin gene structures, efforts were made to isolate genomic loci related to cDNAs.
Seven β-tubulin genomic clones and 11 diverse clones corresponding to α-tubulins were selected. The gene structures of the P. lividus tubulin genes are represented in Figure 3 A,B. Additionally, exclusively the β-tubulin isoforms possess the N-terminal tetrapeptide (MREY) which co-translationally regulates the mRNA degradation by negative β-tubulin autoregulatory binding (Figure 2), thus altering mRNA stability [67].
In the tubulin dimer, the C-terminal tail harboured by αand β-tubulins is exposed to the solvent allowing polyglutamylation [68,69] and it is involved in the regulation of interactions between MTs and different structural and motor MAPs, the regulation of flagellar and ciliary beating and the stability of centrioles/centrosomes. Similarly, polyglycylation also occurs at the C-terminal tail and it has been implicated in the mechanical stabilization of the axoneme [20,70].
Polyglutamylation has been studied also in P. lividus, establishing that this PTM is virtually absent in β-axonemal tubulins and that α-tubulin polyglutamylation (that is not evenly distributed along the flagellar length) is important for flagellar motility [71][72][73].
Computational analyses on sea urchin tubulins mapped the existence of polyglutamylation sites in all the proteins, including E445 and E438 in αand β-tubulins respectively (Figures 1 and 2). This confirms the regulatory role ascribed to these PTMs, connecting tubulins to their associated proteins [20].
Several other PTMs have been identified on tubulins, including ubiquitination and methylation [74]. However, few details on related functions have been provided to date. Interestingly, on the basis of the computational analysis, β-tubulins contain the glycyl-lysine isopeptide (G57K58) that putatively undergoes ubiquitination (Figures 1 and 2).
All these features, including post-transcriptional and post-translation mechanisms, represent a hallmark of the tubulin multigene family and the polymer structural diversity [74][75][76].
A MSA analysis confirm the high similarity among members of αand β-tubulin family and that the main differences located at the C-terminal tail may correspond to the differential isotype class involved in specific heterodimers which in turn may affect MT properties. Additionally, it appears that α isotypes accept more residue variations in the corresponding sequence. An analysis of amino acid substitutions shows that, both in α and β isotypes, about 50% of variability occurs in α helix or β sheet secondary structures. The additional variations showed by α isotypes mainly involve amino acid residues structured in α helix. This is not surprisingly since α helices were already found underrepresented in conserved regions [81,82].

P. lividus αand β-Tubulin Gene Organization
To characterize the αand β-tubulin gene structures, efforts were made to isolate genomic loci related to cDNAs.
Seven β-tubulin genomic clones and 11 diverse clones corresponding to α-tubulins were selected. The gene structures of the P. lividus tubulin genes are represented in Figure 3A,B.
The comparison between cDNA and genomic sequences revealed that the transcription unit of all the β-tubulin genes and α genes including Atub8, Tuba1f, Tuba1e, Tuba1g, Tuba1h and Tuba1a, are composed of three exons interrupted by two introns, whereas in Atub3, Tuba1d, Tuba3 and Tuba1b_1 an additional intron interrupts the coding sequence at the amino acid 125 (phase-0); thus the fourth exon includes sequences encoding the Tub signature, the remaining part of the GTPase domain and the C-terminal domain. In α-tubulin genes, the first intron occurs after the first codon (phase-0) and the second intron is located after the codon number 75 (phase-1). To verify if the third intron occurring in position 125 was acquired before the origin of Echinodermata phylum, an inspection of the α-tubulin expressed genes in a member of the basal echinoid order Cidaroida was performed. Indeed, an α-tubulin gene structured with four exons was found in the slate pencil urchin Eucidaris tribuloides and is reported in Figure 3C. Therefore, it could be reasonably hypothesized that this intron originated before the speciation of Echinodermata and was maintained in the lineage leading to Vertebrata [83].
In β-tubulin genes, the first intron occurs after the codon number 19 (phase-0) and the second intron is located after the codon number 131 (phase-0). All of them possess canonical splicing sites, identified at 5 -end by GT and at 3 -end by AG consensus sequences. Interestingly, the first intron is vertebrates specific, while the second one is shared with invertebrates and is maintained with the alga and fungi [83]. Int    Moreover, computational analyses were performed on the proximal upstream region, revealing the presence of canonical sharp core promoter elements (BRE, TATA box and/or INR) often organized in couples, as reported in Figure 3 and Table 2. Atub8 Both sequence comparison and gene structure analyses provide similar suggestions about the evolution of these gene families, in particular the isotypes that are closely related each other based on tree topology (Figure 4), show the same gene organization and promoter features. This is particularly evident for α-tubulin genes, indeed the divergent ones possess the additional intron and are TATA box dependent (Atub3, Tuba1d, Tuba3 and Tuba1b_1). Additionally, for β-tubulin genes is it possible to observe a stricter relation among genes showing similar first intron length (Btub2, Tubb2a and Btub3).
Both sequence comparison and gene structure analyses provide similar suggestions about the evolution of these gene families, in particular the isotypes that are closely related each other based on tree topology (Figure 4), show the same gene organization and promoter features. This is particularly evident for α-tubulin genes, indeed the divergent ones possess the additional intron and are TATA box dependent (Atub3, Tuba1d, Tuba3 and Tuba1b_1). Additionally, for β-tubulin genes is it possible to observe a stricter relation among genes showing similar first intron length (Btub2, Tubb2a and Btub3). . P. lividus α-and β-tubulin NJ distance trees. The trees were generated using MEGA X. The bootstrap consensus tree inferred from 1000 replicates is taken to represent the reliability of the branches of the analysed sequences. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) greater than 50% are shown in red next to the branches.

P. lividus α-and β-Tubulin Gene Expression
To analyse the expression profiles of α-and β-tubulin embryonic transcripts, total RNA was isolated from different developmental stages including eggs, early blastula, late blastula and prism and RT-qPCR were performed. The α-and β-tubulin gene families contain members whose transcripts are detected both in unfertilized eggs and during development, and genes whose transcripts showed different expression levels during embryogenesis. . P. lividus αand β-tubulin NJ distance trees. The trees were generated using MEGA X. The bootstrap consensus tree inferred from 1000 replicates is taken to represent the reliability of the branches of the analysed sequences. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) greater than 50% are shown in red next to the branches.

P. lividus αand β-Tubulin Gene Expression
To analyse the expression profiles of αand β-tubulin embryonic transcripts, total RNA was isolated from different developmental stages including eggs, early blastula, late blastula and prism and RT-qPCR were performed. The αand β-tubulin gene families contain members whose transcripts are detected both in unfertilized eggs and during development, and genes whose transcripts showed different expression levels during embryogenesis.
Atub8, Tuba1g, Tuba1h and Tuba1b_1 transcripts resulted maternally inherited since they were detected in unfertilized eggs. Atub8, Tuba1g, Tuba1h resulted expressed also during the development reaching the higher levels at early/late blastula stages and then decreasing at prism, whereas Tuba1b_1 was not expressed during early development and was re-expressed at prism.
The expression of the remaining α-tubulin genes was developmentally regulated rising progressively until early and late blastula as occurred for Tuba1a and Tuba3 or restricted at late blastula stage (Tuba1f) or prism stage (Tuba1e, Atub3 and Tuba1d).
Among β-tubulin genes, only the Btub9 transcript was found in unfertilized eggs and its levels are high; while the mRNA level was reduced during the development. Btub2 and Btub3 transcripts showed similar profiles being expressed in a stage specific manner at early blastula and prism; while Btub6 and Btub4 mRNA expression was restricted at prism. Conversely, Tubb2a and Btub5 showed constitutive expression during development. The results agree with previous reports [47][48][49][50] and are summarized in Table 1.

P. lividus PRMT Orthologues
Due to the growing interest in the protein methylation effects and the recent discover of tubulin methylated sites in neural cells [22], together with the lack of information in the sea urchin methyl transferase, we studied the protein arginine methyltransferase family in P. lividus.
Using human PRMT protein sequences as queries, we detected in the P. lividus EST database partial sequences coding for PRMT1 to PRMT7. To complete coding sequences, 3 RACE PCRs were performed. Clones selection was carried out in the same manner as for tubulin cDNAs. Predicted amino acid sequences, whom features are summarized in Table 3, were in silico characterized. An effort was also carried out to isolate cDNAs coding for PRMT8 and PRMT9, however no matching sequences were retrieved in the dataset. Similarly, a screening of the cDNA and genomic libraries, using degenerate primer sets failed to detect any positive signal (data not shown). Therefore, we hypothesize that in P. lividus no orthologues of such methyltransferases do exist. The PMRT gene family and their related protein features are summarized in Table 3.
Inspection of the predicted domains of P. lividus PRMTs showed that they possess the canonical methyltransferase domain. Overall, a SAM-dependent methyltransferase PRMT-type domain is shared among them. In addition, a Src homology 3 (SH3) domain in PRMT2, a Zinc finger C2H2 superfamily domain in PRMT3, a Pleckstrin homology (PH) like domain similar to CARM1 N-terminal histone-arginine methyltransferase domain in PRMT4, a TIM barrel domain in PRMT5 and a second methyltransferase domain in PRMT7 were found ( Figure 5). On the basis of domain organization, protein comparison with human orthologues (see supplemental material) and phylogenetic analysis ( Figure 6), we likely hypothesize that sea urchin PRMT1, PRMT2, PRMT3, PRMT4 and PRMT6 belong to the type-I enzyme while PRMT5 to type-II and PRMT7 to type III [84].  . Amino acid sequences of P. lividus and Homo sapiens PRMTs were aligned and a distance tree was constructed. The tree was generated using MEGA X. The bootstrap consensus tree inferred from 1000 replicates is taken to represent the reliability of the branches of the analysed sequences. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) greater than 50% are shown in red next to the branches. Pairwise amino acid sequence alignment of human and P. lividus PRMTs are shown in Supplementary Materials Figures S1 and S2.
An MSA of the arginine methyltransferase domains show the conservation between mammal and echinoderm PRMTs (see also Supplementary Figures S1 and S2) and the typical differences into the PRMT family. The active site residues, essential for enzymatic activity, are conserved (i.e., glutamates 144 and 153 in rat, 149 and 158 in P. lividus, Figure 7) as well as the GxGxG motif used to bind the adenosyl part of S-adenosyl-L-methionine (SAM) and the acidic residue at the end of β2 that forms hydrogen bonds to both hydroxyls of the SAM ribose (glutamate 100 in rat, 105 in P. lividus, Figure 7).  . Amino acid sequences of P. lividus and Homo sapiens PRMTs were aligned and a distance tree was constructed. The tree was generated using MEGA X. The bootstrap consensus tree inferred from 1000 replicates is taken to represent the reliability of the branches of the analysed sequences. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) greater than 50% are shown in red next to the branches. Pairwise amino acid sequence alignment of human and P. lividus PRMTs are shown in Supplementary Materials Figures S1 and S2.
An MSA of the arginine methyltransferase domains show the conservation between mammal and echinoderm PRMTs (see also Supplementary Figures S1 and S2) and the typical differences into the PRMT family. The active site residues, essential for enzymatic activity, are conserved (i.e., glutamates 144 and 153 in rat, 149 and 158 in P. lividus, Figure 7) as well as the GxGxG motif used to bind the adenosyl part of S-adenosyl-L-methionine (SAM) and the acidic residue at the end of β2 that forms hydrogen bonds to both hydroxyls of the SAM ribose (glutamate 100 in rat, 105 in P. lividus, Figure 7). Figure 6. Amino acid sequences of P. lividus and Homo sapiens PRMTs were aligned and a distance tree was constructed. The tree was generated using MEGA X. The bootstrap consensus tree inferred from 1000 replicates is taken to represent the reliability of the branches of the analysed sequences. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) greater than 50% are shown in red next to the branches. Pairwise amino acid sequence alignment of human and P. lividus PRMTs are shown in Supplementary Materials Figures S1 and S2.
An MSA of the arginine methyltransferase domains show the conservation between mammal and echinoderm PRMTs (see also Supplementary Figures S1 and S2) and the typical differences into the PRMT family. The active site residues, essential for enzymatic activity, are conserved (i.e., glutamates 144 and 153 in rat, 149 and 158 in P. lividus, Figure 7) as well as the GxGxG motif used to bind the adenosyl part of S-adenosyl-L-methionine (SAM) and the acidic residue at the end of β2 that forms hydrogen bonds to both hydroxyls of the SAM ribose (glutamate 100 in rat, 105 in P. lividus, Figure 7).

P. lividus PRMT Expression
To study the mRNA expression profile of PRMTs during development, we performed RT-qPCR experiments at different developmental stages.
Results (reported in Table 4) show that, with the exception of PRMT3 and 6, PRMTs are expressed in the unfertilized egg, therefore they are maternally inherited. PRMT1 mRNA resulted hugely transcribed until late blastula stage, while it remained expressed at lower levels at prism stage. Conversely, similar RNA expression levels were detected for PRMT2, 4 and 5, showing a basal expression during embryogenesis. Differently, PRMT7 and 3 are developmentally regulated, rising progressively until early and late blastula stage respectively.

P. lividus PRMT Expression
To study the mRNA expression profile of PRMTs during development, we performed RT-qPCR experiments at different developmental stages.
Results (reported in Table 4) show that, with the exception of PRMT3 and 6, PRMTs are expressed in the unfertilized egg, therefore they are maternally inherited. PRMT1 mRNA resulted hugely transcribed until late blastula stage, while it remained expressed at lower levels at prism stage. Conversely, similar RNA expression levels were detected for PRMT2, 4 and 5, showing a basal expression during embryogenesis. Differently, PRMT7 and 3 are developmentally regulated, rising progressively until early and late blastula stage respectively. -
Based on the conservation of type-specific methylation consensus, we deduced that also in P. lividus tubulins the major modifications can be represented by MMA and ADMA. Such modifications are catalysed by type I PRMTs and particularly by PRMT1 which is known as the predominant type I PRMT in vertebrates [86] and also during P. lividus embryo development. Therefore, computational analysis tools were herein used to define the 3D structure of the PRMT1 orthologue in P. lividus.
On the basis of the computational analysis, we determined the key features of P. lividus PRMT1 (Figure 8). Results gave an extremely high accuracy model, with 89% of residues modelled at >90% confidence and only the 40 N-terminal residues, expected disordered, were modelled by ab initio calculations, therefore they are not shown in structure Figures 8 and 9. The results highlight the typical two-domain structure-a SAM binding domain and a barrel-like domain-with the active site pocket located between the two domains.  (8,9). The results highlight the typical two-domain structure-a SAM binding domain and a barrel-like domain-with the active site pocket located between the two domains. P. lividus PRMT1 structure was compared to the known rat structure [85]. There are few structural changes among the protein structures, consisting of a slight difference in the orientation of the loop in the dimerization arm and a loop between β14 and β15 ( Figure 8). Figure 8. Molecular evolution of sea urchin PRMT1. Top: Superposition of three-dimensional models in ribbon representation of rat PRMT1 (1ORH; showed in pale) and P. lividus PRMT1 showed in light blue. Bottom: same superposition in surface (50% transparency) representation with their structures coloured according to the evolutionary conservation of amino acids. The P. lividus 3D structure was created via the Phyre 2 software [87] and rendered by using Chimera package [88]. Variable positions are presented in blue; while conserved amino acids are shown in red as defined in the colour-coding bar. In all the structures only the residues after amino acid 40 are shown, since the amino terminus is disordered.
To recreate the structure of a functional enzyme, a P. lividus PRMT1 dimer was computed by molecular docking avoiding use of the unstructured N-terminal region which could impair the reliability of the model (Figure 9).
The resulting models using the balanced coefficients were compared to the available dimer structure already reported [85]. Interestingly, even if a model recapitulates the already published dimer structure of rat PRMT1 [85], other significant structures were also retrieved. We report in Figure 9 the best ranked model which appears more compact and with additional contacts between monomers, resulting in a loss of the cavity in the ring-like structure (Figure 9 B). Top: Superposition of three-dimensional models in ribbon representation of rat PRMT1 (1ORH; showed in pale) and P. lividus PRMT1 showed in light blue. Bottom: same superposition in surface (50% transparency) representation with their structures coloured according to the evolutionary conservation of amino acids. The P. lividus 3D structure was created via the Phyre 2 software [87] and rendered by using Chimera package [88]. Variable positions are presented in blue; while conserved amino acids are shown in red as defined in the colour-coding bar. In all the structures only the residues after amino acid 40 are shown, since the amino terminus is disordered. P. lividus PRMT1 structure was compared to the known rat structure [85]. There are few structural changes among the protein structures, consisting of a slight difference in the orientation of the loop in the dimerization arm and a loop between β14 and β15 (Figure 8).
To recreate the structure of a functional enzyme, a P. lividus PRMT1 dimer was computed by molecular docking avoiding use of the unstructured N-terminal region which could impair the reliability of the model (Figure 9).
The resulting models using the balanced coefficients were compared to the available dimer structure already reported [85]. Interestingly, even if a model recapitulates the already published dimer structure of rat PRMT1 [85], other significant structures were also retrieved. We report in Figure 9 the best ranked model which appears more compact and with additional contacts between monomers, resulting in a loss of the cavity in the ring-like structure ( Figure 9B). Figure 9. P. lividus PRMT1 dimer structure models obtained by ClusPro [89]. The colour coding is red/orange red for the N-terminal domains, green/light green for the SAM binding domains, orange/yellow for the β barrel structures, and blue/light blue for the dimerization arm. (A) the ring-like model similar to the model described in Zhang and Cheng [85]; (B) the best ranked compact model.

Tubulin Arginine Methylation
Methylation on R339 of some α-tubulin isotypes and R46, R62, R86, R162, R282, R318, R320 and R380 of some β-tubulin isotypes were recently shown by proteomics study of protein methylation (for details see Table 4) [22,[76][77][78]. To provide a comprehensive survey of computed methylable aminoacid residues also in the light of physical constraints, we superpose the P. lividus tubulin 3D structures with human orthologues. As expected, structures appear extremely conserved, maintaining secondary and tertiary structure features with RMSD values <1Å. The predicted methylable arginine residues in P. lividus structures were characterized in terms of evolutionary maintenance and solvent accessibility to provide functional sites for enzymatic modifications. A representative comparison is shown in Figure 10 where methylable ariginine residues are coloured according to their RSA in P. lividus Btub3 and human Tubb3. . P. lividus PRMT1 dimer structure models obtained by ClusPro [89]. The colour coding is red/orange red for the N-terminal domains, green/light green for the SAM binding domains, orange/yellow for the β barrel structures, and blue/light blue for the dimerization arm. (A) the ring-like model similar to the model described in Zhang and Cheng [85]; (B) the best ranked compact model.

Tubulin Arginine Methylation
Methylation on R339 of some α-tubulin isotypes and R46, R62, R86, R162, R282, R318, R320 and R380 of some β-tubulin isotypes were recently shown by proteomics study of protein methylation (for details see Table 4) [22,[76][77][78]. To provide a comprehensive survey of computed methylable aminoacid residues also in the light of physical constraints, we superpose the P. lividus tubulin 3D structures with human orthologues. As expected, structures appear extremely conserved, maintaining secondary and tertiary structure features with RMSD values <1Å. The predicted methylable arginine residues in P. lividus structures were characterized in terms of evolutionary maintenance and solvent accessibility to provide functional sites for enzymatic modifications. A representative comparison is shown in Figure 10 where methylable ariginine residues are coloured according to their RSA in P. lividus Btub3 and human Tubb3.
were similarly computed among the two orthologues, showing a huge conservation in terms of RSA profiles. Therefore, it is reasonable to suppose that MMA and ADMA modifications occurs on sea urchin tubulins in a manner that resembles vertebrate mechanisms. Interestingly, it should be noted that also buried residues which are known to be modified in humans (R62 and R318) are analogously located in sea urchin isotypes, suggesting that changes (including structural modifications after PRMT docking) are likely required for enzyme activity. Figure 10. Comparison between arginine methylation sites in human and P. lividus β-tubulins. Top, ribbon/surface diagrams of the human neural β-tubulin structure Tubb3 (5JCO). Bottom, ribbon/surface diagrams of the neural β-tubulin structure generated by homology modelling. The 3D structures were created via the Phyre 2 software [87] and rendered by using Chimera package [88]. The arginine residues that can be methylated were coloured according to their accessibility on the surface: buried in yellow, intermediate accessibility in cyan, exposed in blue. The last amino acid shown in the structures (Q426 for the human protein and D427 for the P. lividus protein) are coloured in red as reference.

Data Mining
The characterized P. lividus α-and β-tubulin sequences were used initially to retrieve their corresponding sequences from the publicly available EST database at the National Centre for Biotechnology Information (NCBI). To rescue the expressed isoforms, these sequences were used as queries to perform extensive nucleotide BLAST (BLASTN) and protein-nucleotide BLAST (TBLASTN) searches. The P. lividus expressed PRMTs were analogously collected using human PRMT sequences as queries.

Embryo Cultures
Gametes were collected from gonads of the sea urchin P. lividus collected in the South-West coast of Sicily, nearby Capo Granitola. Eggs were fertilized at a concentration of 5000/mL and the embryos were grown under gentle rotation at 18 °C in Millipore filtered seawater (MFSW). Developmental stages were monitored by optical microscopy. Figure 10. Comparison between arginine methylation sites in human and P. lividus β-tubulins. Top, ribbon/surface diagrams of the human neural β-tubulin structure Tubb3 (5JCO). Bottom, ribbon/surface diagrams of the neural β-tubulin structure generated by homology modelling. The 3D structures were created via the Phyre 2 software [87] and rendered by using Chimera package [88]. The arginine residues that can be methylated were coloured according to their accessibility on the surface: buried in yellow, intermediate accessibility in cyan, exposed in blue. The last amino acid shown in the structures (Q426 for the human protein and D427 for the P. lividus protein) are coloured in red as reference.

RNA Purification, Reverse Transcription and DNA Extraction
According to evolutionary maintenance of selected residues, methylation consensus in primary sequence, organization of secondary structural elements and 3D conformation, arginine residues were similarly computed among the two orthologues, showing a huge conservation in terms of RSA profiles. Therefore, it is reasonable to suppose that MMA and ADMA modifications occurs on sea urchin tubulins in a manner that resembles vertebrate mechanisms. Interestingly, it should be noted that also buried residues which are known to be modified in humans (R62 and R318) are analogously located in sea urchin isotypes, suggesting that changes (including structural modifications after PRMT docking) are likely required for enzyme activity.

Data Mining
The characterized P. lividus αand β-tubulin sequences were used initially to retrieve their corresponding sequences from the publicly available EST database at the National Centre for Biotechnology Information (NCBI). To rescue the expressed isoforms, these sequences were used as queries to perform extensive nucleotide BLAST (BLASTN) and protein-nucleotide BLAST (TBLASTN) searches. The P. lividus expressed PRMTs were analogously collected using human PRMT sequences as queries.

Embryo Cultures
Gametes were collected from gonads of the sea urchin P. lividus collected in the South-West coast of Sicily, nearby Capo Granitola. Eggs were fertilized at a concentration of 5000/mL and the embryos were grown under gentle rotation at 18 • C in Millipore filtered seawater (MFSW). Developmental stages were monitored by optical microscopy.

RNA Purification, Reverse Transcription and DNA Extraction
Total RNA was extracted from unfertilized eggs and from embryos at 10 (early blastula), 16 (late blastula) and 36 h (prism stage) of development in MFSW using the RNeasy Mini Kit (Qiagen, Hilden, Germany) following the manufacturer's instructions and performing DNase treatment. RNA concentrations were fluorometrically verified on Qubit 2.0 Fluorometer, while RNA integrity was checked using a 1.5% agarose gel. RNA was stored at −80 • C for future use. An amount of total RNA corresponding to 2 µg was treated with DNAseI, Amplification Grade (Sigma-Aldrich) to remove any residual genomic DNA contamination, and DNAseI was inactivated by adding 50 mM EDTA. First strand cDNA was synthesized from 1 µg DNAseI-treated total RNA samples using SuperScript VILO cDNA Synthesis Kit (Invitrogen), following the manufacturer's instructions. The cDNA mixture was diluted 1:20 prior to use in Real Time qPCR experiments.
Genomic DNA was extracted from sperm cells from a single animal using the GenElute Blood Genomic DNA Kit (Sigma-Aldrich, St. Louis, MO, USA) and further genomic DNA purification steps were performed according to the manufacturer's instructions. DNA concentrations and quality were verified by spectrophotometry (OD at 260 nm), whereas the integrity was checked using a 0.8% agarose gel. The DNA was stored at +4 • C for future use.

cDNA Identification
Based on the partial P. lividus PRMT, αand β-tubulin sequences, the 5 -and 3 -end were obtained by PCR-RACE using the 5 /3 RACE Kit, 2nd Generation application kit (Roche) and the primers ( Table 6) as described in the user manual. Table 6. Oligonucleotides used as primers for the PCR-RACE.
The full-length cDNA, consisting of sequences from original ESTs and additional elements from RACE products, were cloned into the pGEM-T Easy vector (Promega, USA) and transformed into XL1-Blue Escherichia coli cells (Stratagene, San Diego, CA, USA). Plasmid DNAs were purified on Illustra™ plasmidPrep Mini SpinKit (GE Healthcare Life Sciences, Chicago, IL, USA) and sequenced using T7 and SP6 primers.

RS-PCR and αand β-Tubulin Gene Organization
Based on the αand β-tubulin 5 and 3 UTR sequences, the primer sets reported in Table 7 were used to isolate the genomic DNA sequences and define the gene structures. Tuba3 gene amplicons were obtained amplifying two fragments (Up and Down). Moreover, the Restriction-site PCR [90] was carried out to isolate genomic elements adjacent to previously cloned sequences (primer sets are listed in Table 8). Table 8. Oligonucleotides used as primers for the RS-PCRs.
Amplified genomic fragments were cloned in pGEM-T Easy plasmid vector and tubulin clones were fully sequenced by primer walking. Sequences were assembled with Codon Code aligner and were annotated using ab initio or with similarity gene prediction programs (i.e., FGENESH using S. purpuratus specific gene-finding parameters [91]) and then manually curated.

RT-qPCR
Real-time PCRs were performed in 96 well plates in a 20 µL mixture containing 1 µL of a 1:20 dilution of the cDNA preparations, in the BIO-RAD CFX96 system using the following PCR parameters: 95 • C for 5 min, followed by 40 cycles of 95 • C for 10 s, 60 • C for 30 s. The sequences of the specific primer pairs used for qPCR are shown in Table 9. Samples were run in triplicate, with positive controls (100 pg plasmid clone DNA) and negative controls (no cDNA). The absence of nonspecific products was confirmed by both the analysis of the melt curves and electrophoresis in 2 % agarose gels. To obtain sample quantification, the E −∆∆Ct method was used, and the relative changes in gene expression were analysed as described in the CFX Manager software manual. 18S rRNA was used as internal reference, in order to compensate for variations in input RNA amounts. Table 9. Oligonucleotides used as primers for the qPCR.

Sequence and Structural Analyses
Functional sites and domains in the predicted amino acid sequences were predicted using the InterProScan software [92].
MSA were constructed using T-Coffee (Tree-based Consistency Objective Function For alignment Evaluation) [93]. Phylogenetic and molecular evolutionary analyses were conducted using a Neighbour Joining (NJ) method, implemented in MEGA version X [94] in which Poisson correction, pairwise deletion and bootstrapping (1000 replicates) were considered as parameters, to define the diversification of different protein families herein analysed. Moreover, the 3D structures were reconstructed by homology modelling via the Protein Homology/analogY Recognition Engine 2.0 (Phyre 2) software [87] using the intensive modelling model. Candidate structures for homology modelling were selected according to pairwise alignment. At least two different structures were used as a template for each generated structure as reported elsewhere [82,[95][96][97] and extremely high accurate homology models were built for all the sets of proteins. Secondary structure assignments and relative solvent accessibility (RSA) were calculated by the DSSP program as implemented in ENDscript [98]. Additionally, the evolutionary variability of amino acids onto the reconstructed structures were rendered using the UCSF Chimera package [88].
The Multimer Mode Docking option of ClusPro tool [89] was used to compute the dimer structure of P. lividus PRMT1, with the number of subunits set to 2 and selecting the option "Remove Unstructured Terminal Residues".

Conclusions
In the present work we provide a comprehensive analysis of the expressed αand β-tubulin gene families in the P. lividus model system. Tubulins are recognized targets for antiproliferative drugs used as chemotherapeutic agents. To date, several lines of evidence suggest their use for drug assessing and toxicity evaluation. Moreover, effectiveness of MT binding compounds has been shown to rely often on different PTM degree on tubulins, thus providing different or controversial issues. Among PTMs, methylation carried out by PRMTs were recently associated to drug resistance mechanisms. In particular, some αand β-tubulin mutations that impair taxanes and epothilones binding to tubulin include arginine residues that undergo methylation (β282 Arg→Gln ) [99]. Moreover, PRMTs are known to be expressed with different subcellular locations in different cancer types and on the other hand, it has been reported that methylation of R282 reduces Taxol binding [22]. Therefore, it has been suggested [23] to use PRMT inhibitors in combination with classic chemotherapy drugs to overcame Taxol resistance. In cancer-targeted therapies, cell division could be inhibited impeding arginine methylation of tubulins, providing full access to Taxol. Alternatively, drug development may also require the production of Taxol derivatives that bind also to methylated tubulin.
Herein, we show that the tubulin and PRMT system in P. lividus extremely resembles the mammal counterpart, in terms of sequences, phylogeny and structures. Additionally, chemical-physical constraints on modifiable arginine residues resulted conserved among the systems.
Herein, we provide additional elements to extend the reliability of the sea urchin embryo also to MT targeting drugs that are affected by arginine methylation. Transfer of nucleic acid, proteins and drugs performed by microinjection in egg or in one of the blastomere of the early stage embryos are established procedures [112,113] that can allow the modulation of drug actions and responses in order to comprehend the activity of targeted elements and to choose a specific compound. Therefore, experimental set up based on modulation of selected PRMTs and tubulins expression level (by means of miRNA or mRNA transfer experiments) could result in the identification of arginine methylation non-sensitive MT targeting drugs. Moreover, local injection of drugs and their derivatives in different blastomeres allow the observation of local effects on the mitotic apparatus [42].
Additionally, in times of the complex identification of the proper animal system due to ethical issues, P. lividus embryos solve many problems also in accordance with international and local laws. This offers the opportunity to test novel anti-proliferative drugs in the nearest non-chordate deuterostome relative, thus providing an animal system supporting cancer research programs including those "from the bench to the bedside".