ABSTRACT
Posterior Hox genes play crucial roles in limb development and specify regions in the proximal-distal (PD) axis of limbs. However, there is no direct genetic evidence that Hox genes are essential for limb regeneration. Moreover, if essential, it is totally unknown which Hox genes have the same or distinct functions between development and regeneration. Here, we mutated hox13 using an efficient CRISPR/Cas9 system in newts (Pleurodeles waltl), which have strong regenerative capacities in various tissues. Triple or double mutants of hox13 paralogs lost their digit and metacarpal/metatarsal bones. Limb regeneration progressed but regenerates lacked the same autopod region. These results showed that hox13 paralogs have the same functions in limb development and regeneration.
INTRODUCTION
The questions to what extent regeneration is similar to development, and what events are specific for regeneration have been topics of discussion for a long time. In limbs, several points have been shown to be regeneration specific. For example, signaling from neurons is necessary for blastema growth in limb regeneration (for review see Stocum, 2017) but not for limb bud growth in development. Although previous studies have suggested that pattern formation of limbs is conducted in a similar manner during development and regeneration (for review see Nacu and Tanaka, 2011), it is still unknown whether the same molecular systems are used.
The posterior Hox genes play critical roles in pattern formation and also limb growth along proximal-distal (PD) and anterior-posterior (AP) axes during development. These Hox functions are involved in determining ZPA and AER activities (for review see Zakany and Duboule, 2007). Analyses of the expression patterns and phenotypes in spontaneous and gene knockout mutations supported a model that Hox9/10, Hox11 and Hox12/13 paralogs specify the identity of the stylopod, zeugopod and autopod, respectively, during limb development (Davis et al., 1995; Fromental-Ramain et al., 1996a; Fromental-Ramain et al., 1996b; Wellik and Capecchi, 2003; Zakany and Duboule, 2007). For example, Hoxa13 and d13 would specify the distal region in the autopod, because double mutant mice lacking Hoxa13 and d13 lost all their digits and metacarpal/tarsal bones (Fromental-Ramain et al., 1996b). In addition, disorders of digit formation, such as hypodactyly, have been linked to human mutations in HOXA13 and HOXD13 (for review, (Lappin et al., 2006). However, there is no direct genetic evidence that Hox genes are essential for limb regeneration as well as limb development. Furthermore, if essential, it is totally unknown which Hox genes have the same or distinct functions between development and regeneration. It is also unknown if the downstream signaling of Hox genes is the same or different between development and regeneration. The reason for this is that there had been no way to perform reversed genetics sufficiently in animals which can regenerate their limbs.
Urodele amphibians have the remarkable ability to regenerate many tissues, including limbs. We developed molecular genetic systems in an urodele animal, Iberian ribbed newt (Pleurodeles waltl) (Hayashi et al., 2013), and recently established a highly efficient gene knockout system using CRISPR/Cas9 (Suzuki et al., 2018). The system enables us to mutate the target gene in the whole body and almost all alleles (more than 99%) of F0 animals (hereinafter referred as to crispants), due to the longer time for the first cleavage in P. waltl (6 h at 25 °C). Therefore, we can disrupt multiple genes in F0 animals by injection of multiple guide RNAs, and analyze the functions without crossing newts. We also performed a comprehensive analysis of a P. waltl transcriptome, and established gene models for almost all protein-coding genes in P. waltl, including all hox genes (Matsunami et al., 2019). In the present paper, using these systems, we analyzed the functions of hox13 paralogs during limb development and regeneration in newts.
RESULTS
Hox13 crispants could not form their digits during limb development
A comprehensive analysis of P. waltl transcriptome revealed that P. waltl has at least three hox13 paralogs (a13, c13 and d13) and transcripts of these three genes were identified in the limb blastemas (Matsunami et al., 2019). Because we also found transcripts showing high homology with hoxb13 of the palmate newt (Lissortriton helveticusi, GenBank:DQ158059) and Axolotl (Ambystoma mexicanum, GenBank:AF298184.1) in P. waltl transcriptome data (Matsunami et al., 2019), we examined mRNA expression of all four hox13 paralogs (a13, b13, c13 and d13) by RT-PCR in limb blastemas (Fig. 1A). The expression of hoxa13, c13 and d13 but not b13 was detected (Fig. 1A). The data is consistent with that in P. waltl transcriptome data (Matsunami et al., 2019). Therefore, we decided to disrupt hox a13, c13 and d13 using a CRISPR/Cas9 system.
We used two sets of guide RNAs; acd (G22) and ad (G24 and G25) (Fig. 1B). G22, G24 and G25 were the names of the CRISPR RNAs (crRNAs). acd and ad were designed to mutate three paralogs (a13, c13 and d13) and two paralogs (a13 and d13), respectively. G22 had complete homologies with one identical sequence in the homeoboxes of hoxa13 and d13 and high homology with a sequence in c13 (19 matched nucleotides among a total of 20 nucleotides of the protospacer sequence). G24 and G25 had complete homology with another sequence in the homeobox of hoxa13 and with a sequence around the splicing acceptor of exon 2 in hoxd13, respectively. Disruption of c13 by G24 was maybe possible, because it had slightly high homology with a sequence in c13 (18 matched nucleotides among a total of 20 nucleotides of the protospacer sequence). All target sequences had a protospacer adjacent motif (PAM) sequence. All three crRNAs showed no high homology (more than 3 nucleotides were unmatched among 20 nucleotides of the protospacer sequences which further had PAM sequences at the 5’ or 3’ end) with all other gene bodies in P. waltl comprehensive transcriptome data, including the sequences of other Homeobox genes (Matsunami et al., 2019).
The ribonucleoprotein complexes (RNPs) including Cas9 and each guide RNA set were injected into fertilized eggs. Uninjected siblings (wild type animals, N=28) showed normal limb development (Fig. 2 and 3). All 9 animals to which acd guide RNA was injected (acd crispants) and 15 among 18 ad crispants lacked all fore-limb and hind-limb digits (Fig. 2A). The remaining three animals in ad crispants had only 1, 2 and 4 digits in their four limbs, respectively, indicating that they lacked 14-17 digits. Bone staining of limbs without digits showed that the stylopod and zeugopod bones formed and appeared to be normal, however, these limbs had neither phalanges nor metacarpal/metatarsal bones (Fig. 2B). The morphology and number of carpal/tarsal bones were abnormal (Fig. 2B). Some carpal/tarsal bones were sharp and the tip stuck out from the skin (Fig. 2B, arrowheads). Observation of limb development in crispants showed the elongation of limb buds, but no digits were formed (Fig. 3). The phenotypes (loss of phalanges and metacarpal/metatarsal bones) are consistent with those in Hoxa13 and d13 double mutant mice (Fromental-Ramain et al., 1996b).
Hox13 genes were mutated in the crispants
Because the target sequences of G22 in acd crispants had a cleavage site for PvuII restriction enzyme, we examined the digestion of amplicons of hoxa13, c13 and d13 in acd crispants to analyze somatic mutations (Fig. S1A). Insufficient or no PCR products were obtained in some cases or the sizes of the main or extra products were smaller than that of a wild type animal, suggesting deletion. Similar results were obtained in the cases of ad crispants (Fig. S1B). For most products from acd crispants in all three genes, which were well amplified, digestion was not detected (Fig. S1A), suggesting mutations in the PvuII site. Some products were digested partially. Since mutations can occur in nucleotides at sites other than a PvuII site, we could not conclude that the digested fragments resulted from wild type alleles.
We also performed amplicon sequencing by next generation sequencing (NGS) to conduct a detailed analysis. Genomic DNAs of hoxa13, c13 and d13 in both acd and ad were investigated. Two crispants of each set, which lost all digits, were examined. Hoxc13 was mutated in acd crispants (mutation rates, 89.2 and 99.8%), but not in ad crispants. Almost all alleles (99.7-100%) in hoxa13 and d13 were mutated in all analyzed animals. One to 4 mutant alleles occupied over 96% of all alleles examined (Table 1, 2, S1 and S2). Mutations in homeoboxes (a13, c13 and d13 in acd and a13 in ad) yielded frameshift mutations or deletions of 2-5 amino acid residues within K10-L16 in helix I (Fig. S2) including conserved residues among homeodomains of posterior Hox proteins (Fig. S2). Helix I is critical for homodimerization, which is required to activate transcription (Zhang et al., 2011). These frameshift mutations and deletions suggested the loss of function of hox13 genes (see also Discussion).
Deletions or insertions around the splicing acceptor of hoxd13 in ad (Table 2 and Table S2) also suggested loss of function of hoxd13 by abnormal splicing. In particular, deletion of the consensus AG sequence in the 3’ splice site makes it impossible to splice at the normal site. Even if new potential splice sites work, the splicing yields frameshift mutations or large deletions in the homeodomain.
One animal in ad (ad#16) showed frameshift mutations and deletion of AG sequence in the 3’ splice sites in all alleles of hoxa13 and hoxd13, respectively (Table 2). Therefore, both the hoxa13 and d13 genes were almost certainly disrupted in at least this animal. All newts in acd and ad including #16 showed the same phenotype in the autopod (loss of phalanges and metacarpal/metatarsals), which was also observed in Hoxa13 and d13 double knockout mice, suggesting that at least hoxa13 and d13 were disrupted in hox13 crispants.
Limb regeneration progressed but regenerates lacked digits in hox13 crispants
Hox 13 crispants grew and metamorphosed, although 5 among 9 animals in acd and 7 among 18 animals in ad died mainly around the time of metamorphosis. One possible reason was difficulty in swimming up to the surface of the water to breath without digits. However, the fact that almost half of the animals survived after metamorphosis was very important for the analysis of limb regeneration, and was different from Hoxa13 and d13 double mutant mice which showed embryonic lethality. The fore-limbs of young crispants after metamorphosis (after 16 weeks post fertilization), which had lost all digits, in acd (4 animals) and ad (10 animals) and uninjected siblings (wild type, 11 animals) were amputated at the proximal region in the stylopod. All four animals in which hox13 mutations were identified by amplicon sequencing (Table 1, 2, S1 and S2) were included. Regenerates of all wild type animals showed normal morphology (Fig. 4, 5 and movie 1). During limb regeneration, no apparent phenotypes were observed before the notch stage, however, no crispants showed notch structure or digit formation (Fig. 4 and movie 2). Bone staining of limbs showed that all regenerates lacked all digits and metacarpal/tarsal bones in the limb regenerates. The bone structure of regenerates was very similar to that of developed limbs before amputation (Fig. 5). These results first showed that hox13 has the same function (patterning of the autopod) in limb regeneration as well as limb development. We could not find any regeneration specific phenotypes.
DISCUSSION
Our results showed that crispants, in which hox13 genes were mutated, lacked the distal autopod region of the limbs during development and regeneration, suggesting that hox13 was essential for morphogenesis of the autopod region in newts as well as mammals, and in both development and regeneration.
The results presented in this paper suggest that hox13 genes were disrupted. Frameshift mutations cause the substitution of downstream amino acid residues and/or truncation by abnormal stop codons, resulting in the loss of over 50% of homeodomains. The lost regions include the recognition helix (helix III, Fig. S2), which is essential for Hox functions. Deletions in helix I would also affect Hox function. Two residues in helix I of HOXA13 (corresponding residues are R18 and E19 in Fig. S2) are critical for dimerization, which is required to activate transcription (Zhang et al., 2011). Deletions of 2-5 amino acid residues within K10-L16 of helix I (Fig. S2), close to the critical residues for dimerization (R18 and E19), would affect the homodimerization by truncation or structural abnormalities in helix I. Therefore, mutations observed in homeoboxes suggest the loss of function of hox13 genes.
acd and ad animals exhibited the same phenotypes in limbs (loss of digits) (Fig. 2-5). These data showed that mutations in at least hoxa13 and d13 caused the phenotypes. We could not find any hoxc13 specific functions in the limb development and regeneration. However, it is still possible that disruption of hoxc13 showed specific or similar phenotypes and that these were masked by a deficiency of hoxa13 and d13. Analysis of single mutants of hoxc13 in newts should be able to examine this possibility.
We believe that the phenotypes shown here did not result from off-target effects which mutated genes other than the hox13 gene, for the following reasons. (1) Target sequences were carefully determined, and the sequences used did not show high homology with any other gene bodies in P. waltl comprehensive transcriptome data (more than 3 mismatches). (2) Sequences with more than two mismatches would not be mutated, because hoxc13 was not mutated by G24. (3) All crispants in acd and ad showed the same phenotypes (loss of digits), although different crRNAs were used. It is almost impossible that another gene was disrupted both in acd and ad, or that different genes which showed the same phenotypes were disrupted in each Set.
Evolutionally, amphibians acquired limbs with the autopods found in present-day tetrapods. How did amphibians acquire the autopods? Hox13 genes have been considered to play important roles in acquisition of the autopod structure, and changes in patterns and activities of Hox13 genes may be involved in acquisition (for reviews see (Leite-Castro et al., 2016; Paço and Freitas, 2018; Schneider and Shubin, 2013; Tanaka, 2016; Woltering and Duboule, 2010). Loss of function experiments, using the CRISPR/Cas9 system, on hoxa13 and d13 in zebrafish showed a marked reduction of fin ray, suggesting that digits originated via the transition of the distal structure of the fin (Freitas et al., 2012; Nakamura et al., 2016). As described above, Hoxa13 and d13 would specify the distal region in the autopod of mice (Fromental-Ramain et al., 1996b). Our studies shown here connect the mouse to zebrafish in terms of Hox13 functions, and indicate that Hox13 genes are essential for formation of the distal region of the appendages (fins and limbs) in mammals, amphibians and fishes. Studies on expression patterns and cis elements of hox13 genes in newts, and comparisons with those of mice and fish will be important. Another fundamental question is what is the biological significance of the acquisition of digits? If we can use animals without digits for physical and/or behavioral experiments, we should be able to investigate the functions of digits and their significance in evolution. It is difficult to use Hox13 mutant mice due to embryonic lethality. However, the newt mutants shown here can be used for these experiments. These studies will provide important insight for understanding the molecular mechanisms that enabled the fin to limb transition, and its biological significance.
MATERIALS AND METHODS
Newts
In this study, we used Iberian ribbed newts (Pleurodeles waltl) which were raised in our laboratory. The animals were reared as described previously (Hayashi et al., 2013). All procedures were carried out in accordance with the Institutional Animal Care and Use Committee of Tottori University (Tottori, Japan) and national guidelines of the Ministry of Education, Culture, Sports, Science & Technology of Japan.
RT-PCR
Four blastemas at late bud (Koriyama et al., 2018) were used for sampling of mRNA. RT-PCR was performed as described in a previous study (Toyoda et al., 2003). Primers are shown in Table S3. GAPDH was used as the endogenous control.
Preparation of RNPs and microinjection
crRNAs were designed by using CRISPR-direct (Naito et al., 2015). Positions of the targets are shown in Fig. 3 and the sequences are G22: CTACACCAAGGTGCAGCTGAAGG, G24: TCATCACCAAAGACAAGCGGAGG and G25: CCCTTTTGTCTTTGCCGCAGGGG (PAM sequences are undelined). The target sequence of G22 has PvuII cleavage site (CAGCTG). The synthetic tracrRNA, crRNA and Cas9 protein were obtained from Integrated DNA Technologies (IDT; Skokie, IA, USA). The tracrRNA and crRNA were annealed, and then RNPs including Cas 9 were produced in accordance with the manufacturer’s instructions just before injection. Microinjection of RNPs was performed based on our previous reports (Hayashi et al., 2019; Suzuki et al., 2018).
Limb amputation and bone staining
Limb amputation and bone staining were performed as described previously (Koriyama et al., 2018). The fore- and hind-limbs were amputated at the most proximal region in the stylopod. The developed and regenerated limbs were fixed in 10% formalin-Holtfreter’s solution at 4°C for more than 2 days and stained with Alizarin Red and Alcian Blue.
Genotyping
Genomic DNA was extracted from tail fins. The target region was amplified from the lysates using EX-Taq enzyme (for conventional PCR analysis, Takara Bio, Japan) or KAPA HiFi enzyme (for NGS analysis, Roche Diagnostics, Basel, Switzerland) with primer sets (Table S3). Digestion of amplicons of acd crispants was examined using PvuII (New England BioLabs Japan, Japan). Amplicons for NGS analysis were obtained by primers containing barcoded overhang adaptor sequences according to a 16S Metagenomic Sequencing Library Preparation Kit, and purified using AMPure XP (Beckman Coulter, Pasadena, CA, USA). All amplicons were mixed and subjected to an Illumina MiSeq run (paired-end 300; Macrogen). Sequencing data was analyzed according to our previous study (Suzuki et al., 2018).
Observation of developing and regenerating limbs
The developing and regenerating limbs were observed with a stereoscopic microscope (Leica, MZFL III) and images including limbs before and after bone staining were acquired with a microscopy camera (Nikon, DS-Ri2).
Competing interests
The authors declare no competing or financial interests.
Funding
This work was partially supported by JSPS KAKENHI Grant no. 26650082 and 16H04794 to T.T., Grant no. 18K06257 to K.T.S. Grant no. 16K08467 to T.H.
Competing interests
The authors declare no competing or financial interests.
Supplemental Figures, tables and movies
Movie 1. Regeneration of the right forelimb of a wild-type animal
Movie 2. Regeneration of the right forelimb of ad animal
Acknowledgements
This work was supported by NIBB Collaborative Research Program (18–204) to T.H.