Mechanisms of Development 164 (2020) 103651

Contents lists available at ScienceDirect

Mechanisms of Development

journal homepage: www.elsevier.com/locate/mod

Conservation analysis of core cell cycle regulators and their transcriptional behavior during limb regeneration in Ambystoma mexicanum

a,b ´ a b c,* Annie Espinal-Centeno , Melissa Dipp-Alvarez , Carlos Saldana˜ , Laszlo Bako , Alfredo Cruz-Ramírez a,* a Molecular and Developmental Complexity Group, Unidad de Genomica´ Avanzada (U.G.A.-LANGEBIO) CINVESTAV, Irapuato, Mexico b Facultad de Ciencias Naturales, Campus Juriquilla, Universidad Autonoma´ de Quer´etaro, Av. de las Ciencias S/N, Juriquilla, Quer´etaro, Qro. CP 76230, Mexico c Department of Plant Physiology, Umeå University, Umeå, Sweden

ARTICLE INFO ABSTRACT

Keywords: Ambystoma mexicanum (axolotl) has been one of the major experimental models for the study of regeneration Cell cycle during the past 100 years. Axolotl limb regeneration takes place through a multi-stage and complex develop­ Evolution mental process called epimorphosis that involves diverse events of cell reprogramming. Such events start with Regeneration dedifferentiation of somatic cells and the proliferation of quiescent stem cells to generate a population of pro­ Axolotl liferative cells called blastema. Once the blastema reaches a mature stage, cells undergo progressive differen­ Ambystoma tiation into the diverse cell lineages that will form the new limb. Such pivotal cell reprogramming phenomena depend on the fine-tuned regulation of the cell cycle in each regeneration stage, where cell populations display specific proliferative capacities and differentiation status. The axolotl genome has been fully sequenced and released recently, and diverse RNA-seq approaches have also been generated, enabling the identification and conservatory analysis of core cell cycle regulators in this species. We report here our results from such analyses and present the transcriptional behavior of key regulatory factors during axolotl limb regeneration. We also found conserved protein interactions between axolotl Cyclin Dependent Kinases 2, 4 and 6 and Cyclins type D and E. Canonical CYC-CDK interactions that play major roles in modulating cell cycle progression in eukaryotes.

1. Introduction as legs, gills, tail, retina, spinal cord, regions of heart and brain (Mita­ shov, 1997; Cano-Martínez et al., 2010; Simon and Tanaka, 2013). In multicellular eukaryotes, most developmental processes rely on Axolotl limb regeneration is essentially divided into four major stages: the transition between the proliferative capacity and the differentiated wound healing, dedifferentiation, blastema formation and rediffer­ status of groups of cells. Processes such as embryonic development, entiation (Yocoyama, 2008). The blastema formation is the main char­ postembryonic organ and limb patterning, metamorphosis and regen­ acteristic of epimorphic regeneration and involves the establishment of eration depend on the capacity of cells to divide and to acquire a specific a heterogeneous population of proliferative progenitor cells. Blastema fate. In turn, proliferation and differentiation are influenced by the formation implies that tissue-resident stem cells and differentiated phase of the cell cycle in which cells are residing. Regeneration of tis­ mature cells abandon their G0 status to enter and progress in the cell sues, organs and limbs implies the replacement of populations of lost cycle (Sandoval-Guzman´ et al., 2014; Johnson et al., 2018). cells, which requires that specific remnant cells enter into a highly Pioneer studies in Ambystoma larva limb regeneration characterized proliferative status. Therefore, the understanding of a given regenera­ the cell cycle dynamics after amputation by monitoring the incorpora­ tion process must take into consideration the mechanisms by which the tion of labeled thymidine (Tassava et al., 1987). A more recent study by cell cycle is modulated in cells that are playing an active role in the Johnson et al. (2018) has shown that 3 days post amputation (dpa) of multi-stage and complex phenomenon of regeneration (Lange and axolotl limb, epidermal cells that cover the stump reinitiate the cell cycle ′ Calegari, 2010; Filipczyk et al., 2007; Calegari and Huttner, 2003). as revealed by EdU-positive (5-ethynyl-2 -deoxyuridine) nuclei, indi­ Ambystoma mexicanum is an urodele amphibian that has a great ca­ cating that cells entered the S-phase. Such experiments also demonstrate pacity for regeneration since it can regenerate lost limbs and organs such that EdU signal dramatically increases at blastema stages (7 and 14 dpa).

* Corresponding authors. E-mail addresses: [email protected] (L. Bako), [email protected] (A. Cruz-Ramírez). https://doi.org/10.1016/j.mod.2020.103651 Received 11 July 2020; Received in revised form 1 October 2020; Accepted 19 October 2020 Available online 28 October 2020 0925-4773/© 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). A. Espinal-Centeno et al. Mechanisms of Development 164 (2020) 103651

EdU-positive cells decrease drastically at redifferentiation stages, sug­ axolotl (Fig. 1). gesting that cells exit the cell cycle prior tissue terminal differentiation. The E2F-pRb pathway lies at the heart of cell cycle regulation in most 2. Results multicellular eukaryotes. This pathway not only involves Retinoblas­ toma family proteins and E2F transcription factors, but also regulatory 2.1. Conservation of core cell cycle regulatory genes in axolotl complexes formed by Cyclins (CYCs) and Cyclin Dependent Kinases (CDKs) proteins. The study of this highly conserved pathway in diverse The genomic data for the axolotl has been quite scarce until recent plants and animals has been extensive, and its crucial role in prolifera­ years and the full genome was sequenced and released just recently tion, differentiation, cancer and apoptosis has been described, as (Nowoshilow et al., 2018). Also, a myriad of RNAseq approaches have reviewed in Cross et al. (2011) and Harashima et al. (2013). generated very useful information to search and analyze putative In animals, the activity of the Retinoblastoma protein (pRb1) is functional orthologs of almost any axolotl gene of interest (Stewart modulated by post-translational modifications. One of the well- et al., 2013; Bryant et al., 2017; Caballero-Perez´ et al., 2018). described pRb1 modifications is the phosphorylation of diverse resi­ As a firststep towards the axolotl cell cycle molecular framework, we dues, exerted by distinct CYC-CDK complexes. Phosphorylation status of collected the protein sequence of the most relevant core cell cycle reg­ pRb1 is central in the dynamics of protein-protein interactions between ulators from mouse and human databases since in these two animal pRb1 and E2F factors in cell cycle phases (Knudsen and Knudsen, 2008). species the cell cycle has been thoroughly studied. The obtained se­ When pRb1 is hypophosphorylated, binding between the pRb1 pocket quences were then used to identify putative homologous proteins in domain and the transactivating domain of E2F occurs while in a diverse databases from vertebrate and invertebrate metazoan species. hyperphosphorylated status pRb1 releases E2F. Free E2F1, in turn, Among invertebrates, we included Hydra vulgaris and Schmidtea medi­ positively regulates the transcription of genes involved in cell cycle terranea, two organisms widely used as model systems for the study of progression. In animals, CDK4-CYCD and CDK6-CYCD complexes regeneration. Among the vertebrates, in addition to Mus musculus and phosphorylate pRb1 at different sites in mid G1. Then, in late G1, CDK2- Homo sapiens, we searched in databases available for 4 species of the CYCE complex phosphorylates additional pRb1 sites (Cross et al., 2011). Ambystoma genus (A. mexicanum, A. velasci, A. andersoni and A. mac­ Phosphorylation of pRb1 by the CDK2-CYCA complex occurs from S ulatum), Xenopus laevis, Xenopus tropicalis, Anolis carolinensis, Alligator phase through G2 and by the CDK1(Cdc2)-CYCB complex from G2 to­ mississippiensis and Danio rerio. wards M phase (Knudsen and Knudsen, 2008). By using diverse, in silico, approaches (see methods for details), we Phosphoprotein Phosphatases PP1 and PP2A are responsible for explored the presence of homologs for several key cell cycle regulators pRb1 dephosphorylation in M phase, which induces cell cycle exit to G0, across all the metazoan species investigated (Fig. 2A). Most of the core an event that correlates with the expression of genes that reprograms cell cycle regulators are conserved among vertebrates including the cells to a differentiated status or maintains some stem cells quiescent Ambystoma spp. Pocket proteins (pRb1p105, pRbl1p107 and pRbl2p130), (Rubin et al., 2001; Cohen, 2002; Vietri et al., 2006; Kurimchak and Cyclins (CYCA1, CYCA2, CYCB1, CYCB2, CYCB3, CYCD1, CYCD2, Grana,˜ 2015). Also, de novo synthesis of unphosphorylated pRb1 occurs CYCE1, CYCE2 and CYCH), E2F transcription factors (E2F1 to E2F8), in the resulting daughter cells, which ensures their exit from a prolif­ Cyclin-Dependent Kinases (CDK1Cdc2, CDK2, CDK4, CDK6 and CDK7), erative status, unless mitogens and some growth factors, which promote Protein Phosphatases (PPP1CA and PPP2CA), and two families of CDK the expression/activity of CYC-CDK complexes, drive cells to re-enter inhibitors that include p15, p16, p18, p19, p21, p27 and p57. It is G1. Other important regulators of the cell cycle are the CIP/KIP and important to note that while p16, a CDK inhibitor of the CDKN2/INK4 CDKN2/INK4 protein families that inhibit the activity of CYC-CDKs family is present in mammals and birds, we were not able to identify any leading to a hypo-phosphorylated status of pRb1 thereby preventing homolog for this protein in reptiles, fishes, amphibians, and in­ cell cycle progression (Knudsen and Knudsen, 2008). Cell cycle entry, vertebrates (Fig. 2A). In contrast to vertebrates, we were not able to with consequent local dedifferentiation and active proliferation of cells, identify ortholog sequences for pRb1p105, pRbl2p130, CYCD1, E2F1, is pivotal for the initiation and progress of limb regeneration. However, E2F3, E2F4, E2F6, E2F7, CDKN2/INK4 and CIP/KIP in the datasets studies unraveling the role of cell cycle regulators at the molecular level interrogated for hydra and planaria sequences (see methods). It is also in this process are very scarce. One of the few studies in this context important to note that in planaria, a single E2F transcription factor was reported that in cultured myotubes from newt limb, pRb1 phosphory­ found (more similar in sequence to human E2F5), while in hydra there is lation by CDK4 and CDK6 is required for cell cycle re-entry (Tanaka an expansion of the family with three genes coding for putative homo­ et al., 1997). logs of E2F2, E2F5 and E2F8. In the case of E2F3, Fig. 2A refers to the One of the limitations for the study of the molecular networks that E2F3A isoform, while the E2F3B isoform could not be found in any modulate cell cycle and differentiation during axolotl limb regeneration Ambystoma spp. has been the lack of reliable genetic datasets. With the advent of next- Most of the members of CDKN2/INK4 (p15INK4B, p16INK4A, p18INK4C generation sequencing (NGS), it has recently been possible to obtain and p19INK4D) and CIP/KIP (p21Cip1/Waf1/Sdi1, p27Kip1, and p57Kip2) the full genome sequence and annotation of this species (Nowoshilow protein families are absent in planaria and hydra, with the exception of et al., 2018). Besides the full genome sequencing, various transcriptomic p27 which is absent in planaria, but present already in hydra. The cell approaches have generated very useful information to search and cycle inhibitor p16 is a special case, since this protein is absent in the analyze putative functional orthologs for several genes of interest in genomes of the Amystoma spp. and other amphibians analyzed. p16 axolotl (Stewart et al., 2013; Bryant et al., 2017; Caballero-P´erez et al., homolog is also absent in the genomes of reptiles, fishesand invertebrate 2018). species analyzed (Gilley and Fried, 2001; Kim et al., 2003). With the To establish a molecular framework for the major regulators of cell exception of p16 and E2F3B isoform, which seem to be absent in axolotl cycle in axolotl, we have identifiedseveral putative functional homologs and other Ambystoma spp. analyzed, the rest of the proteins searched of such genes not only in this species, but also in other Ambystoma have homologs with those of human and mouse. species. We have also analyzed their conservation at gene and protein We found that there is a single pRb-like protein encoded in planaria levels and profiled the expression patterns of the respective axolotl and hydra genomes, while the vertebrate genomes analyzed, including mRNAs at 10 different time-points during limb regeneration. Further­ those of Ambystoma spp., encode for three different pRb proteins, which more, we demonstrate in vitro a conserved interaction between specific are homologous to human pRb1p105, pRbl1p107 and pRbl2p130 (Fig. 2A). axolotl CDKs and CYCs. We consider that this molecular framework will Indeed we identifiedall three proteins for X. tropicalis and X. laevis, while be useful and fundamental for researchers aiming to understand the role evolutionary analyses of Cao et al. (2010) reported only one pRb protein of cell cycle modulation throughout the limb regeneration process in in X. tropicalis. Although the sequence of pRb-like in hydra and planaria

2 A. Espinal-Centeno et al. Mechanisms of Development 164 (2020) 103651

Fig. 1. Cell cycle regulatory proteins and their roles in the checkpoints of the cycle. After limb amputation in A. mexicanum wound healing occurs and the apical epithelial cap is formed, resident connective tissue cells start to dedifferentiate and stem cells leave its quiescent state. Both types of cells are reprogrammed (Left-up corner) and proliferate to populate the blastema. In metazoans, such reprogramming of cells to a proliferative state depends on the activation of CYCD1 and CDK4/6 by mitogens which inhibit the action of CDKN2/INK4 proteins, which are CYCD-CDK repressors. This molecular dynamics leads cells to enter the cell cycle. CYCD1- CDK4/6 complexes phosphorylate pRb proteins, inactivating its capacity to form a repressive complex with E2F transcription factors. The action of pRb-free E2F TFs in transcribing diverse genes is pivotal for the progression of the cycle to S-Phase. The complete release of E2F to exert such action, depends also on the progressive phosphorylation of pRb, not only by CYCD-CDK complexes, but also by the CYCE-CDK2 complex. The counterpart of CYCE-CDK2 complex is CIP/KIP proteins, which inactivate the complex by protein-protein interactions. pRb hyperphosphorylated status is mandatory for the progression of the cycle in both G1-S and G2-M. At the end of mitosis, pRb is dephosphorylated by PPP1CA and PPP2CA complexes, and de novo synthesis of un-phospho pRB in an environment where CYC-CDKs are not active or present, maintains the daughter cells in G0 and some may differentiate. However, it seems that the cellular environment of the blastema niche leads the daughter cells to enter several rounds of divisions until a gradual differentiation of these cells start to form the diverse tissues of the new limb (Created with BioR ender.com). seems to be more similar to that of AmpRbl1 (p107), the percentage of Whether this is a true functional domain that would affect the nucleo­ identity is low. This is in agreement with previous studies suggesting cytoplasmic distribution of the protein remains to be elucidated in future that¨ modern¨pRb1, pRbl1 and pRbl2 resulted from two events of genome studies. In pRbl1, the red line represents an extra cyclin-like domain at duplications during metazoan evolution from the common ancestral the N-terminus. All the domains reported were predicted by using the retinoblastoma protein or apRb (Liban et al., 2017). InterPro Software (Jones et al., 2014). In the case of E2F transcription factors (E2F TF), our presence and absence analyses indicate that the only E2F TF present in the inverte­ 2.2. Domain conservation analysis in E2F-pRb pathway proteins brate species analyzed is more similar to the E2F5 homologs of the vertebrate species (Fig. 2A). Our domain analysis showed, that although Proteins of the E2F-pRb pathway contain crucial polypeptide do­ the protein varies in length between species, the major functional do­ mains mediating DNA-binding and interaction with other proteins. We mains (DNA-binding and CC-MB domains) in E2F5 homologous proteins therefore performed a protein domain analysis for several members of are conserved, including the predicted homologs from A. mexicanum and the E2F-pRb pathway to further explore functional conservation A. velasci (Supplementary Fig. 1, left). Similarly, E2F1 homologs from A. (Fig. 2B, Supplementary Fig. 1, see methods). mexicanum and A. velasci conserve both DNA-binding and CC-MB do­ For most of the sequences of Cyclins, CDKs and E2Fs families mains (Supplementary Fig. 1A, right). analyzed, we did not observe any obvious differences in the major For Ambystoma spp. CYCD1 and CYCD2 homologs, the canonical protein domains (data not shown). We found that in the case of the Cyclin-like domain is present and its location at the N-terminal region of pocket proteins (pRb1, pRbl1 and pRbl2) the major domains, previously the proteins is conserved when compared to their homologs in all characterized in mouse and humans, are conserved in the respective vertebrate and invertebrate species (Supplementary Fig. 1B). Interest­ homologs from Ambystoma spp. In all three proteins, the small pocket ingly, our analysis could not detect a conserved C-terminal region in the domain (SPD) composed by the A-box and B-box, as well as the C-ter­ single planaria CYCD homolog, sequence identity of which suggests it is minal region which, together with the SPD, comprises the large pocket more similar to CYCD2 from vertebrates (Supplementary Fig. 1B). domain (Fig. 2B). In the case of pRbl1, the only pocket protein present in Collectively, these results show that the key components of the E2F- hydra and planaria, the C-terminal domain is absent in the planaria pRb pathway are highly conserved at the protein domain level. How­ homolog (Fig. 2B). The red line in pRb1 comparison represents a domain ever, deeper analyses of the few putative differences found should be found only in Ambystoma ssp. (A. mexicanum, A. velasci, A. andersoni and carried out and experimentally tested in future studies. Also, a detailed A. maculatum) which is predicted as a putative cytoplasmic domain.

3 A. Espinal-Centeno et al. Mechanisms of Development 164 (2020) 103651

Fig. 2. Conservation of core cell cycle regulatory genes along metazoans. A. Presence and absence resume of core cell cycle regulatory genes in diverse metazoan clades. Where purple filled square means presence, white square means absence, red square means that, although the sequence was not found in the current data available for the species, it may be present since the phylogenetically nearest species have the protein. The percentage inside the square reflects the identity of the protein of each species with respect to the homolog from A. mexicanum. *It means that the sequence is not complete. B. Domain conservation analyses of the pocket proteins (pRb1, pRbl1 y pRbl2) among invertebrates, and vertebrate species. The A-box (red box) and the B-box (green box) comprehend the small pocket domain. In turn, the small pocket domain together with the C-terminal region (pink box), composing the large pocket domain. The red line in pRb1 represents a predicted putative domain found only in Ambystoma ssp. and such feature is predicted as a putative cytoplasmic domain. In pRbl1, the red line represents an extra cyclin-like domain in the N-terminal. The putative domains were predicted by InterProScan tool (Jones et al., 2014). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) analysis of conservation of the residues that have been shown as targets as in keeping a minimal identity to be considered as conserved domain of posttranslational modificationsin the human homologs, remains to be (Fig. 2B). We then performed a phylogenetic reconstruction of the pRb done. pocket protein family including sequences of ancestral pRb (apRb) from five Pre-metazoa, and the three pRb proteins from 304 vertebrate spe­ 2.3. Phylogenetic analysis of the Ambystoma spp. pRb protein family cies, using Pre-metazoan sequences from choanoflagellates and Theca­ monas trahens (Apusozoa) to root collapsed and un-collapsed trees Our results show that the homologs for all the members of the reti­ (Fig. 3, Supplementary Fig. 2, respectively). Both Maximum Likelihood noblastoma pocket protein family, pRb1, pRbl1 and pRbl2, are present phylogenetic trees show that pRb1, pRbl1 and pRbl2 protein sequences in the Ambystoma species analyzed (Fig. 2A). The domain analyses show from Ambystoma spp. grouped with their respective homologs. In the that the major functional domains characteristic for pRb1, pRbl1 and case of Ambystoma pRbl1 homologs they grouped closer to those of pRbl2 are present and their location seems to be conserved too, as well frogs. Interestingly, they also group close to their mammalian homologs,

4 A. Espinal-Centeno et al. Mechanisms of Development 164 (2020) 103651 even closer than with those of reptiles (Fig. 3). Although Ambystoma spp. 2.4. Phylogenetic analysis of the Ambystoma spp. CDKN2/INK4 family pRb1 proteins group closer to the Amphibian pRb1s, they are phyloge­ netically closer and appear as a sister clade to the pRb1 homolog of In mammals, members of the CIP/KIP protein family bind CDKs and Easter Newt (Notophthalmus viridescens) than to those of frogs. Pre- inhibit the function of CYC-CDK complexes therefore are inhibitors of metazoan species only have a single apRb protein, which is more cell cycle progression (Besson et al., 2008; Fahmi and Ito, 2019). This similar to the pRb1 homologs of the other species analyzed (Fig. 3). This family is composed of proteins p21Cip1/Waf1/Sdi1, p27Kip1, and p57Kip2 suggests that although pre-metazoan pRb sequences group closer to (Sherr and Roberts, 1999; Fahmi and Ito, 2019). The other protein pRb1 proteins, the three axolotl pRb proteins (AmpRb1, AmpRbl1 and family which has been shown to inhibit the proper function of CYC-CDK AmpRbl2) resulted from gene duplications events, and divergence, from complexes is the CDKN2/INK4 family, composed in several metazoan an ancestral metazoan sequence, as previously reported in other studies, species by p15CDKN2B, p16CDKN2A, p18CDKN2C and p19CDKN2D. Notably, that did not include the Ambystoma spp. sequences (Cao et al., 2010; members of this family bind CYCs instead CDKs, to exert their inhibitory Liban et al., 2017). function (Sherr and Roberts, 1999; Besson et al., 2008; Fahmi and Ito, 2019).

Fig. 3. Maximum Likelihood phylogenetic reconstruction of the retinoblastoma pocket protein family in Metazoa. pRb1p105, pRbl1p107, and pRbl2p130 groups are colored in magenta, orange, and blue, respectively. Only bootstrap values >45 are shown on branches. Pre-metazoan sequences from choanoflagellates and The­ camonas trahens (Apusozoa) were used to root the tree. See Supplementary Fig. 2 for an un-collapsed tree. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Maximum Likelihood phylogenetic reconstruction of the retinoblastoma pocket protein family in Metazoa. pRb1p105, pRbl1p107, and pRbl2p130 groups are colored in magenta, orange, and blue, respectively. Only bootstrap values >45 are shown on branches. Pre-metazoan sequences from choanoflagellatesand Thecamonas trahens (Apusozoa) were used to root the tree. See Supplementary Fig. 2 for an un-collapsed tree. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

5 A. Espinal-Centeno et al. Mechanisms of Development 164 (2020) 103651

(caption on next page) 6 A. Espinal-Centeno et al. Mechanisms of Development 164 (2020) 103651

Fig. 4. Maximum Likelihood phylogenetic tree of four proteins from the CDKN2/INK4 family in Metazoa. p15CDKN2B/p16CDKN2A, p18CDKN2C and p19CDKN2D sub­ families are colored in orange, blue, and magenta, respectively. Only bootstrap values >40 are shown on branches. The NF-kappa-B inhibitor alpha (NKBA in the tree) sequence was used as an external group. In bold we show the Ambystoma ssp. proteins. See Supplementary Fig. 3 for an un-collapsed tree. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Maximum Likelihood phylogenetic tree of four proteins from the CDKN2/INK4 family in Metazoa. p15CDKN2B/p16CDKN2A, p18CDKN2C and p19CDKN2D subfamilies are colored in orange, blue, and magenta, respectively. Only bootstrap values >40 are shown on branches. The NF-kappa-B inhibitor alpha (NKBA in the tree) sequence was used as an external group. In bold we show the Ambystoma ssp. proteins. See Supplementary Fig. 3 for an un-collapsed tree. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Due to their importance as inhibitors of the cell cycle, we decided to CDKN2/INK4 families, we observed that the transcript levels of Amp15 perform a phylogenetic reconstruction for the members of the CDKN2 and Amp21 increased since 12 hpa and along all stages, reaching Amp15 protein family. Our collapsed and uncollapsed Maximum Likelihood a peak at 7 dpa. The transcript levels of Amp18, Amp19, Amp21 and trees (Fig. 4 and Supplementary Fig. 3, respectively) show that p18 Amp57 genes reached peaks of expression at 10 dpa. On the other hand proteins from Ambystoma spp. group together next to their homologs Amp27 expression did not change along regeneration, showing low from amphibians. In contrast, Ambystoma spp. p19 homologs grouped transcript abundances (Fig. 5B). together, but more closely to their homologs from fishes and mammals We also found that AmPPP1CA and AmPPP2CA transcript levels in­ than to those of amphibians. Within the CDKN2 protein family tree, p18 crease at 5, 10 and 28 dpa, with peaks of expression at 10 dpa. The and p19 form a monophyletic clade that could be the result of gene evaluated transcripts correspond to the axolotl homologs of the catalytic duplication in the last common ancestor of chondrichthyes. The fact that subunits of PP1 and PP2A complexes. In monkeys both PP1 and PP2A we do not see a p19 sequence from chondrichthyes could be evidence of complexes have been implicated in pRb1 dephosphorylation. However, gene loss in this group of organisms or could be explained because of the several studies have shown that PP1 is the major pRb1 cell cycle-related absence of a p19 sequence in the interrogated databases. Furthermore, phosphatase (Ludlow et al., 1990). p15 and p16 sequences form a monophyletic clade and within it, the The E2F TFs play a central role in the control of cell cycle, since they putative p15 homologs found in Ambystoma spp. grouped more closely are major transcriptional regulators of gene expression in the diverse with p15/p16 homologs from mammals. An interesting observation phases of the cycle. It has been shown in mammals and plants that some derived from our presence/absence analysis (Fig. 2A) is the absence of E2Fs act as transcriptional activators, while others act as repressors. It available sequences for p16 protein in the invertebrate species analyzed has also been shown that pRb1 inhibits the activity of E2Fs through in this study, as well as absent in Ambystoma spp., birds, and reptiles protein-protein interactions, forming transcriptional repressor com­ (Fig. 4). Our findings are in agreement with previous studies that have plexes. In this study, we identified8 genes coding for E2F TFs in axolotl shown the absence of a functional gene coding for p16 in chicken and and named them according to their similarity with the human homologs newt (Kim et al., 2003; Tanaka et al., 1997). as AmE2F1, AmE2F2, AmE2F3, AmE2F4, AmE2F5, AmE2F6, AmE2F7 and AmE2F8 (Fig. 2A). In humans there are also 8 genes coding for E2F TFs, 2.5. Quantitation of mRNA levels of core cell cycle regulators throughout and the E2F3 locus codes for two isoforms (E2F3A and E2F3B). Previous axolotl limb regeneration studies have shown that E2F1, E2F2 and E2F3A are transcriptional ac­ tivators, while E2F3B, E2F4 and E2F8 act as transcriptional repressors in ˜ Transcription of genes that regulate the cell cycle is dynamic and the humans and mice (Cross et al., 2011; Cuitino et al., 2019). In the RNAseq analysis of their fine-tuned regulation is pivotal to understand the cell datasets and protein databases available for Ambystoma ssp. we found a reprogramming events that take place during limb regeneration. To homolog only for E2F3A, but not for E2F3B. Our RT-qPCR approach explore the transcriptional behavior of the major core cell cycle regu­ shows that, with differences in the levels, all AmE2F genes are tran­ lators in A. mexicanum limb regeneration, we isolated RNA from tissues scriptionally upregulated at 3 dpa (Fig. 5C), being AmE2F6 and AmE2F7 collected at 10 different time points during the process. RNAs were the most upregulated. The expression pattern of AmE2F1 shows a isolated for each time point in three biological replicates and cDNA for maximum of expression at 10 dpa, when blastema is formed and each time point was synthesized. RT-qPCR analyses were carried out for growing and also most AmE2F transcript levels increase when compared each cDNA with a technical triplicate (see Methods for details), to to time 0 dpa. While AmE2F3 and AmE2F8 show similar expression determine the changes in expression at each time point in comparison patterns, their transcript levels increase since 3 dpa and keep expressed with time 0. at the remaining time points tested (Fig. 5C). For a more accurate description of the transcriptional behavior of the Finally, quantitation of the relative expression of mRNA levels for 29 genes included in this analysis, we grouped them in families. In the members of the pRb family shows that AmpRb1, AmpRbl1 and AmpRbl2, case of CYCs-CDKs, considered as cell cycle activators which are albeit with differences in transcript levels, reach their maximum necessary to be expressed in specificcheckpoints of the cycle. Our results expression at 10 dpa. Surprisingly, AmpRb1 is the most upregulated at show that the most upregulated gene corresponds to AmCDK2, which 10 dpa, followed by AmpRbl2 (Fig. 5D). transcript levels increase since 3 dpa, but reach a dramatic change in The data reported above represents the most complete analysis of the expression at 10 dpa, a regeneration stage that corresponds to medium expression pattern for axolotl core cell cycle genes in a developmental blastema (Fig. 5A). Indeed, AmCYCD1, AmCYCE, AmCDK4 and AmCDK6 process and the first such study from the aspect of regeneration. The also reach their top levels of expression at 10 dpa, while transcript levels expression patterns and the observed correlations in expression shown of AmCYCA2, AmCYCD2, AmCDK1 and AmCDK7 are higher at 7 dpa, at in Fig. 5 will be revisited later in the discussion section. early blastema stage. This shows a high correlation of expression among most AmCYCs and AmCDKs, especially at blastema stages. In fact, the 2.6. Protein-protein interaction assays reveal potential AmCYCD1- homologs of these genes, in other species, are well known to interact and AmCDK and AmCYCE1-AmCDK complexes form specific complexes that promote cell cycle progression, several of them, by phosphorylating pRb proteins. In diverse animal models, consecutive phosphorylation of the pRb1 Acting upstream of CYC-CDK complexes are proteins of the CIP/KIP protein by CYCD1-CDK4/6 and CYCE1/CDK2 complexes in G1 phase and CDKN2/INK4 families, which are known to be negative regulators of brings about the release and activation of E2F transcription factors and the cell cycle by binding to either Cyclins or CDKs, inactivating their promotes cell cycle progression. Because of their critical role we thought complexes and in turn, modulating the function of pRb proteins to investigate whether the respective axolotl proteins are capable of (Fig. 5B). Regarding the expression patterns of genes of the CIP/KIP and forming such canonical protein kinase complexes. To test the formation

7 A. Espinal-Centeno et al. Mechanisms of Development 164 (2020) 103651

Fig. 5. Heatmap visualization of RT-qPCR analysis of mRNA levels of core cell cycle regulators during axolotl limb regeneration. RT-qPCR was performed on cDNA using gene-specific primers for diverse axolotl homologous genes to those involved in cell cycle regulation, along 10 time-points after limb amputation. Relative quantification of each gene expression level was normalized according to the AmODC1 gene expression. A. AmCYCs and AmCDKs B. AmCIP/KIP, AmCDKN2/INK4, and AmPhosphatase, families. C. AmE2F Transcription factors. D. AmpRb1 and AmpRb-like genes. The values were validated for three biological replicates by RT- qPCR, and expressed as relative quantification [2^(-ΔΔCт)], data are the mean of three biological replicates, each one with three technical replicates. Purple rep­ resents up-regulation, white represents no change, and yellow represents down-regulation, with respect to time 0. Graduation in colour intensity represents the variations in mRNA expression values, Standard deviations (+/ ) were calculated and presented below the expression value. Limb regeneration stages are described as follows, WH: Wound healing, DD: Dedifferentiation, EB: Early blastema, MB: Medium blastema, LB: Late blastema, ED: Early digit (formation), MD: Medium digit (formation). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

8 A. Espinal-Centeno et al. Mechanisms of Development 164 (2020) 103651

Fig. 6. Protein-Protein interactions between AmCDKs and AmCYCD1-E1. A. Interactions between AmCYCD1-AmCDK2, AmCYCD1-AmCDK4 and AmCYCD1- AmCDK6. B. Interactions between AmCYCE1-AmCDK2, AmCYCE1-AmCDK4 and AmCYCE1-AmCDK6. Proteins were expressed in Arabidopsis protoplasts. Com­ plexes immunoprecipitated with anti-c-Myc antibodies were analyzed on protein gel blots using anti-HA and anti-c-Myc antibodies to detect protein-protein binding (upper panels in A and B) and expression of AmCDKs (lower panels in A and B) respectively. of complexes between AmCYCD1 and AmCDK2, AmCDK4 and AmCDK6, RNA-Seq datasets analyzed nor in the reported protein sequences for any we transiently co-expressed epitope-tagged AmCYCs and AmCDKs in of the Ambystoma spp. (and other amphibians), fishes,reptiles and birds Arabidopsis protoplasts. Co-immunoprecipitation assays (Co-IPs) analyzed. This is an interesting observation, since it has been shown for showed that AmCYCD1 very weakly binds AmCDK2 but strongly binds some species that the absence of p16 is not due to the lack of a gene to AmCDK4. The interaction was also observed between AmCYCD1 and coding for it, but to alterations in genomic arrangements and alternative AmCDK6, although the signal was a lot weaker when compared to that splicing of the resulting transcripts. Regarding this, in the human with AmCDK4 (Fig. 6A). genome p15CDKN2B and p16CDKN2A genes are encoded in the CDKN2B- A similar experimental approach was applied to test the binding ARF-CDKN2A locus, which is composed by the genes that code not only between AmCYCE1 and AmCDK2, AmCDK4 and AmCDK6. We observed for p15 and p16 mRNAs, but also for ARF (ADP ribosylation factor) that AmCYCE1 strongly binds to AmCDK2. Interestingly, we also mRNA. Indeed HsARF mRNA is specified by the second exon of observed that AmCYCE1 blinds to AmCDK4 and AmCDK6, although HsCDKNA via an alternative ORF. A similar genomic arrangement and these interactions appeared to be weaker compared to AmCYCE1- transcription has been reported for mouse p15, p16 and ARF mRNAs AmCDK2 binding (Fig. 6B). (Kim et al., 2003; Szklarczyk et al., 2007). However, Kim et al. (2003), Collectively, our Co-IPs and the transcriptional analyses suggest that demonstrated that, in chicken, a similar locus does not encode for AmCYCD1 may form a functional complex preferentially with AmCDK4, p16CDKN2A. Our results are also in agreement with a previously proposed while AmCYCE1 possibly forms a functional complex with AmCDK2 model in which it was hypothesized that p16 might not be produced in similar to human and mouse. the urodele cells (Tanaka et al., 1997). These observations are inter­ esting since it has been shown in mammals that p16 acts upstream the 3. Discussion CYCD-CDK-pRb1 pathway by inhibiting the action of CDK4 and CDK6 and leading to a hypophosphorylated, therefore repressive, status of The present study represents the most complete catalog of cell cycle- pRb1. Active pRb1 can then promote partial or terminal differentiation related genes in Ambystoma mexicanum reported so far. We identified of proliferative cells. Under such scenario, the predicted absence of p16 their presence in axolotl and other Ambystoma spp., and analyzed their in urodeles, generated in an epoch when almost no genomic data for conservation along diverse animal clades. This was achieved based on urodele amphibians was available, led us to speculate that this might be detailed analyses of both genomic and transcriptomic datasets. More­ a major difference between mammals with a limited capacity to regen­ over, our study describes the transcriptional behavior of most of these erate, and urodele amphibians, with high regeneration capacity (Haas genes during the axolotl limb regeneration process. Such transcriptional and Whited, 2017). Although we cannot exclude that the absence of p16 landscapes include the expression analysis of several genes, which were in axolotl, and other Ambystoma spp., may influence the molecular not explored before in the context of axolotl limb regeneration process. program that potentiates the regeneration capacity in axolotl in com­ Since a finetuned regulation of the transitions between cell proliferation parison to humans or mouse, we speculate that p16 absence alone is not and differentiation governs the regeneration process, our findings are sufficient to explain the high regenerative ability of the axolotl, since pivotal to shed light on the molecular mechanisms that modulate the cell p16 is also absent in birds, reptiles and fishes,species that do not possess cycle in axolotl. the extraordinary regenerative capacity of the axolotl and other urodele Previous studies have focused on the evolutionary aspects of cell amphibians. Also, we can not discard that the regulatory function that cycle regulators across the animal kingdom and the resulting data has p16 protein exerts on its targets could be modulated by other players been extensively discussed (Medina et al., 2016; Liban et al., 2017). Here from the CDKN2/INK4 family. Therefore, future and detailed experi­ we discuss from the perspective of the putative homologs of core cell ments should be done at the mechanistic level to explore these ideas. cycle regulators analyzed from Ambystoma spp., and their expression Another interesting finding, resulted from our presence-absence patterns in A. mexicanum limb regeneration. analysis, is the absence of the E2F3B isoform in axolotl, and in any of Our results showed that p16 transcripts are not present neither in the the Ambystoma spp. datasets analyzed. However, we were able to

9 A. Espinal-Centeno et al. Mechanisms of Development 164 (2020) 103651 identify in all Ambystoma spp. analyzed, the homologous sequences future functional experiments. coding for the E2F3A isoform. In mammals, E2F3A acts as a transcrip­ Other interesting observations, which resulted from our transcrip­ tional activator, while E2F3B has been shown to act as a repressor (Asp tional analysis, is the high expression of diverse CYC-CDK repressors. It et al., 2009). This is a quite interesting observation since it has been is noteworthy the transcriptional behavior of Amp21 which reaches its shown that E2F3A and E2F3B exert opposing regulation of SOX2 in highest expression levels 10 dpa. Also, other CDK repressors such as mouse neural precursors (Julian et al., 2013), and axolotl SOX2 is crit­ Amp57 and Amp18 are upregulated mainly at 10 dpa. In contrast, Amp15 ical for promoting the proliferation of neural stem cells upon tail expression is upregulated from 24 hpa until 7 dpa, and its expression is amputation (Fei et al., 2014). Weather AmE2F3A acts regulating down regulated at 10 dpa. This indicates that cell cycle progression is AmSOX2 transcription and the absence of AmE2F3B is of relevance on dynamically modulated, suggesting that CIP/KIP and CDKN2/INK4 AmSOX2-mediated tail regeneration, should be experimentally proteins are being produced to inhibit CDK2, CKD4, CDK6 and preparing addressed in the future. cells to exit the cycle and perhaps, help in the differentiation of certain Developmental processes, such as limb regeneration in multicellular cells in specific stages of the regeneration process. organisms, rely on finely regulated cell proliferation which, in turn, It should also be taken into consideration that transcript levels may critically depends on cell-cycle-dependent mRNA oscillations. It has not coincide with levels of the respective encoded proteins and their been proposed that such oscillatory behavior of cell-cycle-regulated activity. Several cell cycle regulators have been shown to interact with genes are partially related to E2F-dependent transcriptional activation each other, and such interactions are dependent on posttranslational and repression (Giangrande et al., 2004; Li et al., 2008; Cuitino˜ et al., modifications. This is true for pRb proteins, CYCs, CDKs and E2Fs. 2019). Whether critical modifications are conserved in the putative functional Several interesting observations resulted from our expression pattern homologs in axolotl and whether modifications are part of the mecha­ analyses. The transcriptional analyses showed that AmCDK2, AmCDK4 nisms that govern cell reprogramming events during regeneration, need and AmCDK6 reach their maximum peak of expression in blastema to be experimentally explored in future studies. stages, specifically at 10 dpa. Surprisingly, AmCDK2 is the most upre­ As a proof of concept for this, we explored the interaction between gulated gene at 10 dpa, when compared to time 0 dpa. This suggests that axolotl CYCs and CDKs. CYCD1-CDK2/4/6 and CYCE1-CDK2/4/6 several cells in this time point are preparing to enter S-phase (Fig. 1). complexes are two of the most conserved protein-protein interactions AmCDK2 transcription is upregulated since 24 hpa, as well as its putative during plant and animal evolution. The specificity of the formed com­ complex partner AmCYCE. Also, AmCDK4 and AmCYCD2 transcript plexes and their pivotal roles in diverse stages of cell cycle have been levels increase since early times after amputation (Fig. 5A). The early extensively described as major switches for cell proliferation, therefore transcriptional activation of axolotl CYCs and CDKs, since 12 hpa, led us underlying a myriad of developmental processed and morphogenetic speculate on two probable scenarios: i) That existing proteins after events (Cao et al., 2010; Zhu and Pearson, 2013; Medina et al., 2016). amputation may form axolotl CYC-CDK complexes which phosphorylate Our Co-IP assay using Arabidopsis protoplast system has shown very AmpRb and AmE2F may induce the further transcription of AmCYCs and clean results on protein-protein interaction, this system is quite clean AmCDKs in further timepoints or ii) The transcriptional activation of and avoids the use of axolotl cells, however we should consider that the these genes is independent of AmpRb status, for example the tran­ proper interactions in vivo in the axolotl remain to be confirmed.Future scriptional activation of CYCs could be caused by mitogens such as experiments are needed in order to show if the strength and specificityof growth factors that act through the MAP-Kinase signaling pathway, as these interactions is similar in axolotl tissues, where several other previously shown in other models and experimental systems (Roovers axolotl cell cycle players might be influencing these CYCD1-CDK et al., 1999). interactions. Also, our expression analyses showed that all AmE2F tested are Our results showed that AmCYCD1 strongly binds to AmCDK4 transcriptionally induced at 3 dpa (Fig. 5C). By 3 dpa the wound healing (Fig. 6A). Interestingly, the transcriptional patterns of AmCYCD1 and process is ending and de-differentiation may start to take place, and cells AmCDK4 suggest co-expression, since they grouped in the Heatmap start to leave G0 status and enter G1. These expression patterns coincide visualization of our RT-qPCR analyses (Fig. 5A). We also observed that with measurements made in diverse phases of the cell cycle using mouse AmCYCE1 preferentially binds to AmCDK2 (Fig. 6B). A weak interaction embryonic fibroblast(MEF) cell culture, where E2F1, E2F2, E2F3A, E2F7 was also observed between AmCYCE1 and AmCDK4, and AmCDK6. The and E2F8 mRNA levels were higher in G1 phase cells when compared to transcriptional patterns of AmCYCE and AmCDK2 grouped in the Heat­ cells in G0, and their expression peaks at S and G2 phases (Cuitino˜ et al., map visualization of our RT-qPCR analyses (Fig. 5A). Moreover, the 2019). AmE2F1 is the most upregulated gene of this TF family, with relative expression levels of AmCDK6 transcripts are low throughout transcript levels reaching a maximum of expression at 10 dpa. Moreover, regeneration, and less abundant than those of other AmCDK genes, after 3 dpa the expression levels of AmE2F3, AmE2F5, AmE2F6 and suggesting that the most probable AmCYCE partner in vivo is AmCDK2. AmE2F7 remain relatively constant along the process, with exception of These findingsreveal highly conserved interactions, since in human another peak of expression of AmE2F6 and AmE2F7 at 21 dpa. Inter­ and mouse CYCD1 forms complexes with CDK4 and CDK6, to phos­ estingly, mRNA levels of E2F3b, E2F4, E2F5 and E2F6 did not change phorylate pRb at the G1-S checkpoint, while CYCE partner is CDK2 and across all cell-cycle phases in MEF cell cultures (Cuitino˜ et al., 2019). work together to continue pRb phosphorylation in late G1-S, which is a AmE2F1 and AmE2F8 homologs, which in human and mouse act as major event that brakes the interaction between pRb and E2Fs TFs, transcriptional activators and repressors, respectively, are the most allowing E2Fs to exert their role on gene transcription activation for upregulated E2Fs at blastema stages (7 and 10 dpa). These transcrip­ proper cell cycle progression (Cao et al., 2010; Cuitino˜ et al., 2019). tional patterns suggest that in a population of cells that constantly pRb proteins are master regulators of the cell cycle in diverse king­ proliferate, as in blastema stages, both transcriptional activation and doms of life and their role strongly depends on their phosphorylation repression occurs. Indeed it has been hypothesized that transcriptional status which, in turn, relies on the action of CYC-CDK complexes. In this repression mediated by E2F8 in mid-S-G2 allows a temporary and study, genes coding for putative functional homologs of the three pRb repressive state that is necessary for a next wave of cell-cycle-dependent proteins (pRb1, pRbl1 and pRbl2) were identifiedin 4 Ambystoma spp., gene expression to be reactivated during the next round of cell cycle including the axolotl. Phylogenetic and protein domain analyses (Cuitino˜ et al., 2019). It has been shown that E2F8 lacks the ability to revealed conservation with their homologs in the amphibian clade. Our form a repressive complex with pRb1, since mouse and human E2F7 and phylogenetic analyses show that Ambystoma spp. pRb1 sequences group E2F8 lack the DP-dimerization and the retinoblastoma binding domains closer to the pre-metazoan sequences (apRb). Ambystoma pRb1, pRbl1 (Maiti et al., 2005). Whether such mechanisms of cell cycle control are and pRbl2 might follow a similar evolutionary path than that described occurring during axolotl limb regeneration needs to be demonstrated via in previous studies, which indicate that these three proteins resulted

10 A. Espinal-Centeno et al. Mechanisms of Development 164 (2020) 103651 from gene duplications and divergence of an ancestral metazoan spp., we obtained the data from Xenobase (Karimi et al., 2018, http sequence (Cao et al., 2010; Liban et al., 2017). ://www.xenbase.org/, RRID:SCR_003280). All the obtained sequences One of the model organisms with the highest regeneration capacity were used as input to the software MEGA Version 6 (Tamura et al., 2013) in the animal kingdom is Schmidtea mediterranea. The single pRb protein where they were translated with the use of codons for vertebrates, later in fresh-water planaria, Smed-Rb, is more similar to p107 homologs and they were aligned against protein sequences of different animals with is highly expressed in proliferative cells. The characterization of Smed- Clustal Omega. For other species, the sequences were obtained from Rb RNAi lines has shown that it is required for cell-cycle progression, NCBI and Ensembl databases (Cunningham et al., 2019). The sequences while it was not essential for proper differentiation (Zhu and Pearson, curated and annotated for the diverse proteins reported here for 2013). This correlation between the absence of a canonical pRb1 protein Ambystoma spp. were submitted to NCBI and their respective accession in planaria and its high regeneration capacity led us to speculate that numbers are summarized in Supplementary Table I. Other accession there might be a partial loss of the role of AmpRb1 as a canonical numbers for diverse proteins of other species that we submitted to NCBI repressor of the cycle, which together to the absence of p16 and E2F3B are as follows: for p57 of X. tropicalis (MT664955) and X. laevis homologs, may be related with the high regenerative capacity in axolotl. (MT664956), for CYCH from A. mississippiensis (MT653689). The To test such hypotheses more detailed conservation analyses on axolotl different accession numbers for H. vulgaris proteins are: CDK1 pRb proteins, at the level of motifs and residues, should be done to reveal (MT646816), CDK2 (MT653637), CDK7 (MT653650), CYCB1 the existence of subtle differences among pRb proteins from axolotl and (MT653662), CYCH (MT653690), E2F2 (MT664912), E2F5 their homologs in species with a lower regeneration capacity. (MT664924), E2F7 (MT664932) and PPP2CA (MT683295). The acces­ The changes in transcript levels, observed in Fig. 5, are subtle for sion number for S. mediterranea are: CDK4 (MT653642), CDK7 several of the genes tested, and in some cases seem contradictory with (MT653651), CYCB3 (MT653669), CYCH (MT653691), PPP1CA the processes occurring in a regeneration stage. This suggests that some (MT664960) and PPP2CA (MT683296). of these genes might not be regulated at the transcriptional level, but protein modifications may be occurring. We should also consider that 4.2. Domain conservation analyses the change in the expression of a gene might be restricted to a specific subset of cells or cell lineage, and such changes might be masked by the The orthologous sequence for all 35 cell cycle proteins was scanned non-changing transcript levels of the same gene in the rest of the tissue in the InterPro program to find domain old and new putative. InterPro used for RNA isolation. It would be very useful to determine the in situ uses predictive models for various different databases (Jones et al., expression of transcripts, and proteins, encoded in these genes in order 2014). to generate transgenic markers for cell cycle stages. These novel tools could make possible single cell expression analyses of such cell cycle 4.3. Phylogenetic reconstruction of the pocket protein family regulators. Although our study describes the most complete catalog of cell cycle After selecting pRb1, pRbl1 and pRbl2 protein homologs (Supple­ related genes in axolotl reported so far, as well as a detailed analysis of mentary Dataset 1, (TXT)), we performed an amino acid sequence their transcriptional behavior during limb regeneration, the functional alignment with MAFFT (v7.452) using the FFT-NS-i method (309 amino validation for each of their, putatively homologs, genes remains acid sequences, 1244-134 sites) (Katoh et al., 2002; Katoh and Standley, pendant. Therefore, future research must include not only loss and gain 2013). We then edited the alignment and deleted uninformative gaps of function approaches to evaluate each of the genetic networks and using Jalview 2.11 (Waterhouse et al., 2009). We used ModelTest-NG pathways that control cell cycle status, not only during limb regenera­ v0.1.5 (Darriba et al., 2019; https://github.com/ddarriba/modeltest) tion but also during development. For example, the relevance of the on CIPRES to find the best protein evolution model that fit our data absence of p16 and E2F3B, in the context of regeneration, remains to be according to the Akaike Information Criterion (AICc). We posteriorly explored. used RAxML-HPC v.8 (Stamatakis, 2014) on the CIPRES Science Finally, we expect that making public the data described in this study Gateway to make a Maximum Likelihood phylogeny using the GAMMA could be of use and support for future functional studies aimed to reveal model of rate heterogeneity, the JTT amino acid substitution matrix, and and understand the molecular mechanisms, related to cell cycle control, 1000 rapid bootstrap replicates. To root, visualize and edit the phylo­ underlying axolotl regenerative capacity. The differences in such mo­ genetic tree we used FigTree (http://tree.bio.ed.ac.uk/software/figtree lecular mechanisms between axolotl and humans are of key relevance /) and InkScape 0.92. for regenerative medicine. 4.4. Phylogenetic reconstruction of the CDKN2/INK4 proteins family 4. Materials and methods We selected p15 (CDKN2B), p16 (CDKN2A), p18 (CDKN2C) and p19 4.1. Bioinformatic analyses (CDKN2C) protein homologs (Supplementary Dataset 2, (TXT)), and we then made an amino acid sequence alignment with MAFFT (v7.452) The sequences of the genes of interest (AmpRb1, AmpRbl1, AmpRbl2, using the FFT-NS-i method (207 amino acid sequences) (Katoh et al., AmCDK1, AmCDK2, AmCDK4, AmCDK6, AmCDK7, AmCYCA1, 2002; Katoh and Standley, 2013). We then edited the alignment and AmCYCA2, AmCYCB1, AmCYCB2, AmCYCB3, AmCYCD1, AmCYCD2, used Jalview 2.11 (Waterhouse et al., 2009). We used ModelTest-NG AmCYCE1, AmCYCE2 AmCYCH, AmE2F1, AmE2F2, AmE2F3, AmE2F4, v0.1.5 (Darriba et al., 2019; https://github.com/ddarriba/modeltest) AmE2F5, AmE2F6, AmE2F7, AmE2F8, Amp15, Amp18, Amp19, Amp21, on CIPRES to findthe best evolution model that fitour data according to Amp27, Amp57, AmPPP1CA and AmPPP2CA) of A. mexicanum were the Akaike Information Criterion (AICc). We used RAxML-HPC v.8 obtained from transcripts already published in regeneration (Stewart (Stamatakis, 2014) on the CIPRES Science Gateway to make a Maximum et al., 2013) and by tissues generated in the Group of Molecular Likelihood phylogeny using the GAMMA model of rate heterogeneity, Complexity and Development (GCMD, Caballero-Perez´ et al., 2018). For the LG amino acid substitution matrix, and 1000 rapid bootstrap repli­ A. velasci the sequences were obtained from unpublished data sets cates. To edit the resulting phylogenetic tree we used FigTree (http://t generated by our research group, for A. maculatum and A. andersoni, ree.bio.ed.ac.uk/software/figtree/) and InkScape 0.92. sequences were obtained from Dwaraka et al. (2019). For S. mediterranea were from WormBase ParaSite (Howe et al., 2016; Howe et al., 2017, 4.5. Animal limb amputation and tissue processing https://parasite.wormbase.org/index.html), PlanMine (Rozanski et al., 2018, http://planmine.mpi-cbg.de/planmine/begin.do). For Xenopus Juvenile A. mexicanum specimens from Xochimilco, were obtained

11 A. Espinal-Centeno et al. Mechanisms of Development 164 (2020) 103651 from the Unit of Environmental Management at the Centro de Inves­ org/10.1016/j.mod.2020.103651. tigaciones Acuaticas´ de Cuemanco, Universidad Autonoma´ Metropoli­ tana of Mexico´ City, Campus Xochimilco. Three individuals 7.5–8.5 cm CRediT authorship contribution statement size, in length head-tail for each experiment time of regeneration were used. The animals were anesthetized by immersion in a 0.01% solution Annie Espinal-Centeno: Conceptualization, Investigation, Formal ´ of benzocaine (ethyl 4-aminobenzoate, SIGMA), using a scalpel the analysis, Data curation, Writing - original draft. Melissa Dipp-Alvarez: anterior-right limb was amputated at the mid-forearm level. The limb, or Formal analysis, Data curation, Visualization, Writing - review & edit­ regenerated tissues, of each individual were processed for RNA extrac­ ing. Carlos Saldana:˜ Resources, Writing - review & editing, Funding tion and the obtained RNA was used for cDNA synthesis. acquisition. Laszlo Bako: Conceptualization, Investigation, Resources, The animals experiments were conducted to fully agree with the Writing - review & editing, Funding acquisition, Supervision. Alfredo Mexican OfficialNorm (NOM-062-ZOO-1999) “Technical Specifications Cruz-Ramírez: Conceptualization, Supervision, Resources, Writing - for the Care and Use of Laboratory Animals” based on the Guide for the original draft, Writing - review & editing, Funding acquisition. Care and Use of Laboratory Animals “The Guide”, 2011, NRC, USA, with the Federal Register Number # BOO.02.08.01.01.0095/2014, awarded Declaration of competing interest by the National Health Service, Food Safety and Quality (SENASICA- SADER). The Institutional Animal Care and Use Committee (IACUC) The authors declare no competing interests. from the CINVESTAV approved the project “Management and hus­ bandry of Ambystoma spp and experimental processing of tissue for functional analyses and genetic expression” ID animal use protocol Acknowledgements number: 0209-16. This work was supported by The Swedish International Research 4.6. RT-qPCR quantification links grant 2014-9040-114152-32 granted to A.C.-R. and LB; CONACYT Ciencia Basica´ Grant No. I0017-CB-2015-01-000000000252126 granted ´ Total RNA samples of different times (0 hpa, 12 hpa, 24 hpa, 3 dpa, 5 to A.C.-R. Ciencia Basica Grant No. A1-S-26966 granted to C.S. Authors dpa, 7dpa, 10 dpa, 14 dpa, 21 dpa and 28 dpa) from three individuals of want to thank Dr. Octavio Martinez de la Vega Laboratory and Fernando ´ 7.5–8.5 cm in length, head-tail, for each regeneration time point, were Hernandez-Godinez for technical support. CONACyT provided the PhD used. Each organism was anesthetized by immersion in a 0.01% scholarship to A.E.-C. (No. 420591) and M.D.-A. (No. 487660). benzocaine (ethyl 4-aminobenzoate, SIGMA) solution. Then the anterior right limb was amputated at the mid-forearm level. The tissue of each of Data availability the tree organisms were collected in each time point and processed for RNA extraction using TRIzol Reagent (Invitrogen). RNA was used for the The sequences curated and annotated for the diverse proteins re­ synthesis of cDNA according to the manufacturer protocol SuperScript ported here for Ambystoma spp. were submitted to NCBI and their III RNA to cDNA kit (Invitrogen), we used 3 μg of total RNA per 20 μL respective accession numbers are summarized in Supplementary Table I. reaction. The expression levels of specific mRNAs were determined by Other accession numbers for diverse proteins of other species that we RT-qPCR using SYBR Green PCR master mix in 20 μL reaction. We used submitted to NCBI are as follows: for p57 of X. tropicalis (MT664955) three biological replicates with three independent RT-qPCR reactions and X. laevis (MT664956), for CYCH from A. mississippiensis for each data point and we included the primers efficiency in the cal­ (MT653689). The different accession numbers for H. vulgaris proteins culations. The primer sequences used in these experiments are in Sup­ are: CDK1 (MT646816), CDK2 (MT653637), CDK7 (MT653650), CYCB1 plementary Table II. The resulting data for each time point and (MT653662), CYCH (MT653690), E2F2 (MT664912), E2F5 transcript was normalized with ODC1 expression levels previously re­ (MT664924), E2F7 (MT664932) and PPP2CA (MT683295). The acces­ ported to axolotl limb regeneration (Guelke et al., 2015). The relative sion number for S. mediterranea are: CDK4 (MT653642), CDK7 expression was calculated by 2^( Δ Δ Cт) method, subtracting the (MT653651), CYCB3 (MT653669), CYCH (MT653691), PPP1CA expression levels of time 0 to those each time point analyzed. (MT664960) and PPP2CA (MT683296).

4.7. Co-immunoprecipitacion assays References

For this assay, vector constructs expressing AmCDK2/4/6 in trans­ Asp, A., Acosta-Alvear, D., Tsikitis, M., Oevelen, C.V., Dynlacht, B.D., 2009. E2f3b plays an essential role in myogenic differentiation through isoform-specific gene lational fusion to c-Myc and AmCYCD1/AmCYCE1 to HA epitopes were – – regulation. Genes Dev. 23, 37 53. https://doi.org/10.1101/gad.1727309. generated. Transfected protoplasts were incubated in the dark for 16 24 Besson, A., Dowdy, S., Roberts, J., 2008. CDK inhibitors: cell cycle regulators and h, harvested by centrifugation then proteins were extracted from the cell beyond. Dev. Cell 14, 159–169. https://doi.org/10.1016/j.devcel.2008.01.013. pellet as described previously (Cruz-Ramírez et al., 2012). For the co- Bryant, D.M., Sousounis, K., Payzin-Dogru, D., Bryant, S., Sandoval, A.G.W., Martinez, F. μ J., Mariano, R., Oshiro, R., Wong, A.Y., Leigh, N.D., Johnson, K., Whited, J.L., 2017. immunoprecipitation assays, 100 g total protein extract was incu­ Identificationof regenerative roadblocks via repeat deployment of limb regeneration bated in a total volume of 100 μL extraction buffer containing 150 mM in axolotls. NPJ Regen Med 2, 30. https://doi.org/10.1038/s41536-017-0034-z. ´ NaCl and 1 μg anti-HA or 1.5 μg anti-cMyc antibody (Covance, clone Caballero-Perez, J., Espinal-Centeno, A., Falcon, F., García-Ortega, L.F., Curiel- ´ μ Quesada, E., Cruz-Hernandez, A., Bako, L., Chene, X., Martínez, O., Arteaga- 16B12 for HA and 9E10 for c-Myc, respectively). After 2 h, 10 L Vazquez,´ M.A., Herrera-Estrella, L., Cruz-Ramírez, A., 2018. Transcriptional ProteinG-Sepharose matrix (GE Healthcare) was added which was pre­ landscapes of axolotl (Ambystoma mexicanum). Dev. Biol. 433, 227–239. https:// viously equilibrated in TBS buffer and this mixture was further incu­ doi.org/10.1016/j.ydbio.2017.08.022. ◦ bated for another 2 h on a rotating wheel at 4 C. The matrix was then Calegari, F., Huttner, W.B., 2003. An inhibition of cyclin-dependent kinases that lengthens, but does not arrest, neuroepithelial cell cycle induces premature washed 3 times using 500 μL of washing buffer (1XTBS, 5% glycerol, neurogenesis. J. Cell Sci. 116, 4947–4955. https://doi.org/10.1242/jcs.00825. 0.1% IGEPAL CA-630) and eluted by boiling in 25 μL 1.5× Laemmli Cano-Martínez, A., Vargas-Gonzalez,´ A., Guarner-Lans, V., Prado-Zayago, E., Leon-´ sample buffer. Proteins were then resolved with SDS-PAGE and blotted Oleda, M., Nieto-Lima, B., 2010. Functional and structural regeneration in the axolotl heart (Ambystoma mexicanum) after partial ventricular amputation. Arch to PVDF transfer membrane (Millipore). The presence of the proteins of Cardiol Mex 80 (2), 79–86. interest was tested by immunodetection using rat anti-HA-peroxidase Cao, L., Peng, B., Yao, L., Zhang, X., Sun, K., Yang, X., Yu, L., 2010. The ancient function (3F10, Roche) or chicken anti-c-Myc primary antibody (A2128, of RB-E2F pathway: insights from its evolutionary history. Biol. Direct 5, 55. https:// doi.org/10.1186/1745-6150-5-55. Invitrogen). Cohen, P.T.W., 2002. Protein phosphatase 1 - targeted in many directions. J. Cell Sci. Supplementary data to this article can be found online at https://doi. 115, 241–256.

12 A. Espinal-Centeno et al. Mechanisms of Development 164 (2020) 103651

Cross, F.R., Buchler, N.E., Skotheim, J.M., 2011. Evolution of networks and sequences in Katoh, K., Standley, D.M., 2013. MAFFT multiple sequence alignment software version 7: eukaryotic cell cycle control. Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. 366 (1584), improvements in performance and usability. Mol. Biol. Evol. 30 (4), 772–780. 3532–3544. https://doi.org/10.1098/rstb.2011.0078. https://doi.org/10.1093/molbev/mst010. Cruz-Ramírez, A., Díaz-Trivino,˜ S., Blilou, I., Grieneisen, V.A., Sozzani, R., Zamioudis, C., Katoh, K., Misawa, K., Kuma, K., Miyata, T., 2002. MAFFT: a novel method for rapid Miskolczi, P., Nieuwland, J., Benjamins, R., Dhonukshe, P., Caballero-Perez,´ J., multiple sequence alignment based on fast Fourier transform. Nucleic Acids Res. 30, Horvath, B., Long, Y., Mah¨ onen,¨ A.P., Zhang, H., Xu, J., Murray, J.A., Benfey, P.N., 3059–3066. https://doi.org/10.1093/nar/gkf436. Bako, L., Mar´ee, A.F., Scheres, B., 2012. A bistable circuit involving SCARECROW- Kim, S.H., Mitchell, M., Fujii, H., Llanos, S., Peters, G., 2003. Absence of p16INK4a and RETINOBLASTOMA integrates cues to inform asymmetric stem cell division. Cell truncation of ARF tumor suppressors in chickens. Proc. Natl. Acad. Sci. U. S. A. 100 150 (5), 1002–1015. https://doi.org/10.1016/j.cell.2012.07.017. (1), 211–216. https://doi.org/10.1073/pnas.0135557100. Cuitino,˜ M.C., Pecot,´ T., Sun, D., Kladney, R., Okano-Uchida, T., Shinde, N., Saeed, R., Knudsen, E., Knudsen, K., 2008. Tailoring to RB: tumor suppressor status and therapeutic Perez-Castro, A.J., Webb, A., Liu, T., Bae, S.I., Clijsters, L., Selner, N., Coppola, V., response. Nat. Rev. Cancer 8 (9), 714–724. https://doi.org/10.1038/nrc2401. Timmers, C., Ostrowsk, M.C., Pagano, M., Leone, G., 2019. Two distinct E2F Kurimchak, A., Grana,˜ X., 2015. PP2A: more than a rest switch to activate pRB proteins transcriptional modules drive cell cycles and differentiation. Cell Rep. 27 (12), during the cell cycle and in response to signaling cues. Cell Cycle 14 (1), 18–30. 3547–3560.e5. https://doi.org/10.1016/j.celrep.2019.05.004. https://doi.org/10.4161/15384101.2014.985069. Cunningham, F., Achuthan, P., Akanni, W., Allen, J., Amode, M.R., Armean, I., Lange, C., Calegari, F., 2010. Cdks and cyclins link G1 length and differentiation of Bennett, R., Bhai, J., Billis, K., Boddu, S., Cummins, C., Davidson, C., Dodiya, K.J., embryonic, neural and hematopoietic stem cells. Cell Cycle 9 (10), 1893–1900. Gall, A., Giron, C.G., Gil, L., Grego, T., Haggerty, L., Haskell, E., Hourlier, T., https://doi.org/10.4161/cc.9.10.11598. Izuogu, O.G., Janacek, S.J., Juettemann, T., Kay, M., Laird, M.R., Lavidas, I., Liu, Z., Li, J., Ran, C., Li, E., Gordon, F., Comstock, G., Siddiqui, H., Cleghorn, W., Chen, H.-Z., Loveland, J.E., Marugan,´ J.C., Maurel, T., MacMahon, A.C., Moore, B., Morales, J., Kornacker, K., Liu, C.-G., Pandit, S.K., Khanizadeh, M., Weinstein, M., Leone, G., de Mudge, J.M., Nuhn, M., Ogeh, D., Parker, A., Parton, A., Patricio, M., Salam, A.I.A., Bruin, A., 2008. Synergistic function of E2F7 and E2F8 is essential for cell survival Schmitt, B.M., Schuilenburg, H., Sheppard, D., Sparrow, H., Stapleton, E., Szuba, M., and embryonic development. Dev. Cell 14, 62–75. https://doi.org/10.1016/j. Taylor, K., Threadgold, G., Thormann, A., Vullo, A., Walts, B., Winterbottom, A., devcel.2007.10.017. Zadissa, A., Chakiachvili, M., Frankish, A., Hunt, S.E., Kostadima, M., Langridge, N., Liban, T.J., Medina, E.M., Tripathi, S., Sengupta, S., Henry, R.W., Buchler, N.E., Rubin, S. Martin, F.J., Muffato, M., Perry, E., Ruffier, M., Staines, D.M., Trevanion, S.J., M., 2017. Conservation and divergence of C-terminal domain structure in the Aken, B.L., Yates, A., Zerbino, D., Flicek, P., 2019. Ensembl 2019. Nucleic Acids Res. retinoblastoma protein family. Proc. Natl. Acad. Sci. U. S. A. 114 (19), 4942–4947. 47 (D1), D745–D751. https://doi.org/10.1093/nar/gky1113. https://doi.org/10.1073/pnas.1619170114. Darriba, D., Posada, D., Kozlov, A.M., Stamatakis, A., Morel, B., Flouri, T., 2019. Ludlow, J.W., Shon, J., Pipas, J.M., Livingston, D.M., DeCaprio, J.A., 1990. The ModelTest-NG: a new and scalable tool for the selection of DNA and protein retinoblastoma susceptibility gene product undergoes cell cycle-dependent evolutionary models. Mol. Biol. Evol. 37 (1), 291–294. https://doi.org/10.1093/ dephosphorylation and binding to and release from SV40 large T. Cell 60, 387–396. molbev/msz189. https://doi.org/10.1016/0092-8674(90)90590-b. Dwaraka, V.B., Smith, J.J., Woodcock, M.R., Voss, S.R., 2019. Comparative Maiti, B., Li, J., de Bruin, A., Gordon, F., Timmers, C., Opavsky, R., Patil, K., Tuttle, J., transcriptomics of limb regeneration: identificationof conserved expression changes Cleghorn, W., Leone, G., 2005. Cloning and characterization of mouse E2F8, a novel among three species of Ambystoma. Genomics 111, 1216–1225. https://doi.org/ mammalian E2F family member capable of blocking cellular proliferation. J. Biol. 10.1016/j.ygeno.2018.07.017. Chem. 280, 18211–18220. https://doi.org/10.1074/jbc.M501410200. Fahmi, M., Ito, M., 2019. Evolutionary approach of intrinsically disordered CIP/KIP Medina, E.M., Turner, J.J., Gordan, R., Skotheim, J.M., Buchler, N.E., 2016. Punctuated proteins. Sci. Rep. 9, 1575. https://doi.org/10.1038/s41598-018-37917-5. evolution and transitional hybrid network in an ancestral cell cycle of fungi. eLife 5, Fei, J.F., Schuez, M., Tazaki, A., Taniguchi, Y., Roensch, K., Tanaka, E.M., 2014. CRISPR- e09492. https://doi.org/10.7554/eLife.09492. mediated genomic deletion of Sox2 in the axolotl shows a requirement in spinal cord Mitashov, V.I., 1997. Retinal regeneration in amphibians. Int J Dev Biol. 41 (6), neural stem cell amplification during tail regeneration. Stem Cell Reports 3 (3), 893–905. 444–459. https://doi.org/10.1016/j.stemcr.2014.06.018. Nowoshilow, S., Schloissnig, S., Fei, J.F., Dahl, A., Pang, A.W.C., Pippel, M., Winkler, S., Filipczyk, A.A., Laslett, A.L., Mummery, C., Pera, M.F., 2007. Differentiation is coupled Hastie, A.R., Young, G., Roscito., J.G., Falcon,´ F., Knapp, D., Powell, S., Cruz, A., to changes in the cell cycle regulatory apparatus of human embryonic stem cells. Cao, H., Habermann, B., Hiller, M., Tanaka, E.M., Myers, E.W., 2018. The axolotl Stem Cell Res. 1, 45–60. https://doi.org/10.1016/j.scr.2007.09.002. genome and the evolution of key tissue formation regulators. Nature 554 (7690), Giangrande, P.H., Zhu, W., Schlisio, S., Sun, X., Mori, S., Gaubatz, S., Nevins, J.R., 2004. 50–55. https://doi.org/10.1038/nature25458. A role for E2F6 in distinguishing G1/S- and G2/M-specifictranscription. Genes Dev. Roovers, K., Davey, G., Zhu, X., Bottazzi, M.E., Assoian, R.K., 1999. Alpha5beta1 integrin 18, 2941–2951. https://doi.org/10.1101/gad.1239304. controls cyclin D1 expression by sustaining mitogen-activated protein kinase activity Gilley, J., Fried, M., 2001. One INK4 gene and no ARF at the Fugu equivalent of the in growth factor-treated cells. Mol. Biol. Cell 10 (10), 3197–3204. https://doi.org/ human INK4A/ARF/INK4B tumour suppressor locus. Oncogene 20 (50), 7447–7452. 10.1091/mbc.10.10.3197. Guelke, E., Vesna, B., Liebsch, C., Lazaridis, A., Radtke, C., Vogt, P.M., Reimers, K., 2015. Rozanski, A., Moon, H., Brandl, H., Martín-Duran,´ J.M., Grohme, M., Hüttner, K., Identificationof reference genes and validation for gene expression studies in diverse Bartscherer, K., Henry, I., Rink, J.C., 2018. PlanMine 3.0—improvements to a axolotl (Ambystoma mexicanum) tissues. Gene 560, 114–123. https://doi.org/ mineable resource of flatwormbiology and biodiversity. Nucleic Acids Res. gky1070. 10.1016/j.gene.2015.01.030. https://doi.org/10.1093/nar/gky1070. Haas, B.J., Whited, J.L., 2017. Advances in decoding axolotl limb regeneration. Trends Rubin, E., Mittnacht, S., Villa-Moruzzi, E., Ludlow, J.W., 2001. Site-specific and Genet. 33 (8), 553–565. https://doi.org/10.1016/j.tig.2017.05.006. temporally-regulated retinoblastoma protein dephosphorylate by protein Harashima, H., Dissmeyer, N., Schnittger, A., 2013. Cell cycle control across the phosphatase type 1. Oncogene 20 (29), 3776–3785. https://doi.org/10.1038/sj. eukaryotic kingdom. Trends Cell Biol. 23 (7), 345–356. https://doi.org/10.1016/j. onc.1204518. tcb.2013.03.002. Sandoval-Guzman,´ T., Wang, H., Khattak, S., Schuez, M., Roensch, K., Nacu, E., Howe, K.L., Bolt, B.J., Cain, S., Chan, J.C., Chen, W.J., Davis, P., Done, J., Down, T., Tazaki, A., Joven, A., Tanaka, E.M., Simon, A., 2014. Fundamental differences in Gao, S., Grove, C., Harris, T.W., Kishore, R., Lee, R., Lomax, J., Li, Y., Muller, H.M., dedifferentiation and stem cell recruitment during skeletal muscle regeneration in Nakamura, C., Nuin, P., Paulini, M., Raciti, D., Schindelman, G., Stanley, E., Tuli, M. two salamander species. Cell Stem Cell 14 (2), 174–187. https://doi.org/10.1016/j. A., Auken, K.V., Wang, D., Wang, X., Williams, G., Wright, A., Yook, K., stem.2013.11.007. Berriman, M., Kersey, P., Schedl, T., Stein, L., Sternberg, P.W., 2016. WormBase Sherr, C.J., Roberts, J.M., 1999. CDK inhibitors: positive and negative regulators of G1- 2016: expanding to enable helminth genomic research. Nucleic Acids Res. 44 (D1), phase progression. Genes Dev. 14, 1501–1512. https://doi.org/10.1016/j. D774–D780. https://doi.org/10.1093/nar/gkv1217. devcel.2008.01.013. Howe, K.L., Bolt, B.J., Shafie,M., Kersey, P., Berriman, M., 2017. WormBase ParaSite a Simon, A., Tanaka, E.M., 2013. Limb regeneration. Wiley Interdiscip. Rev. Dev. Biol. 2, comprehensive resource for helminth genomics. Mol. Biochem. Parasitol. 215, 2–10. 291–300. https://doi.org/10.1002/wdev.73. https://doi.org/10.1016/j.molbiopara.2016.11.005. Stamatakis, A., 2014. RAxML version 8: a tool for phylogenetic analysis and post-analysis Johnson, K., Bateman, J., DiTommaso, T., Wong, A.Y., Whited, J.L., 2018. Systemic cell of large phylogenies. Bioinformatics 30 (9), 1312–1313. https://doi.org/10.1093/ cycle activation is induced following complex tissue injury in axolotl. Dev. Biol. 433 bioinformatics/btu033. (2), 461–472. https://doi.org/10.1016/j.ydbio.2017.07.010. Stewart, R., Rascon,´ C.A., Tian, S., Nie, J., Barry, C., Chu, L.F., Ardalani, H., Wagner, R.J., Jones, P., Binns, D., Chang, H., Fraser, M., Li, W., McAnulla, C., McWilliam, H., Probasco, M.D., Bolin, J.M., Leng, N., Sengupta, S., Volkmer, M., Habermann, B., Maslen, J., Mitchell, A., Nuka, G., Pesseat, S., Quinn, A.F., Sangrador-Vegas, A., Tanaka, E.M., Thomson, J.A., 2013. Comparative RNA-seq analysis in the Scheremetjew, M., Yong, S., Lopez,´ R., Hunter, S., 2014. InterProScan 5: genome- unsequenced axolotl: the oncogene burst highlights early gene expression in the scale protein function classification. Bioinformatics 30 (9), 1236–1240. https://doi. blastema. PLoS Comput. Biol. 9 (3), e1002936 https://doi.org/10.1371/journal. org/10.1093/bioinformatics/btu031. pcbi.1002936. Julian, L.M., Vandenbosch, R., Pakenham, C.A., Andrusiak, M.G., Nguyen, A.P., Szklarczyk, R., Heringa, J., Pond, S.K., Nekrutenko, A., 2007. Rapid asymmetric McClellan, K.A., Svoboda, D.S., Lagace, D.C., Park, D.S., Leone, G., Blais, A., Slack, R. evolution of a dual-coding tumor suppressor INK4a/ARF locus contradicts its S., 2013. Opposing regulation of Sox2 by cell-cycle effectors E2f3a and E2f3b in function. Proc. Natl. Acad. Sci. U. S. A. 104 (31), 12807–12812. https://doi.org/ neural stem cells. Cell Stem Cell 12 (4), 440–452. https://doi.org/10.1016/j. 10.1073/pnas.0703238104. stem.2013.02.001. Tamura, K., Stecher, G., Peterson, D., Filipski, A., Kumar, S., 2013. MEGA6: molecular Karimi, K., Fortriede, J.D., Lotay, V.S., Burns, K.A., Wang, D.Z., Fisher, M.E., Pells, T.J., evolutionary genetics analysis version 6.0. Mol. Biol. Evol. 30 (12), 2725–2729. James-Zorn, C., Wang, Y., Ponferrada, V.G., Chu, S., Chaturvedi, P., Zorn, A.M., https://doi.org/10.1093/molbev/mst197. Vize, P.D., 2018. Xenbase: a genomic, epigenomic and transcriptomic model Tanaka, E.M., Gann, A.A., Gates, P.B., Brockes, J.P., 1997. Newt myotubes reenter the organism database. Nucleic Acids Res. 46 (D1), D861–D868. https://doi.org/ cell cycle by phosphorylation of the retinoblastoma protein. J. Cell Biol. 136 (1), 10.1093/nar/gkx936. 155–165. https://doi.org/10.1083/jcb.136.1.155.

13 A. Espinal-Centeno et al. Mechanisms of Development 164 (2020) 103651

Tassava, R.A., Goldhamer, D.J., Tomlinson, B.L., 1987. Cell cycle controls and the role of Waterhouse, A.M., Procter, J.B., Martin, D.M.A., Clamp, M., Barton, G.J., 2009. Jalview nerves and the regenerate epithelium in urodele forelimb regeneration: possible version 2 - a multiple sequence alignment editor and analysis workbench. modifications of basic concepts. Biochem. Cell Biol. 65 (8), 739–749. https://doi. Bioinformatics 25 (9), 1189–1191. https://doi.org/10.1093/bioinformatics/btp033. org/10.1139/o87-097. Yocoyama, H., 2008. Initiation of limb regeneration: the critical steps for regenerative Vietri, M., Bianchi, M., Ludlow, J.W., Mittnacht, S., Villa-Moruzzi, E., 2006. Direct capacity. Develop Growth Differ Development, Growth y Differentiation 50 (1), interaction between the catalytic subunit of protein phosphatase l and pRb. Cancer 13–22. https://doi.org/10.1111/j.1440-169x.2007.00973.x. Cell Int. 6 (3) https://doi.org/10.1186/1475-2867-6-3. Zhu, S.J., Pearson, B.J., 2013. The retinoblastoma pathway regulates stem cell proliferation in freshwater planarians. Dev. Biol. 373 (2), 442–452. https://doi.org/ 10.1016/j.ydbio.2012.10.025.

14