Like many other amphibians, Xenopus can regenerate their limbs or tails after they have been severed at larval stages. The Xenopus larval tail contains complex, patterned tissues, including the spinal cord, notochord, and muscles, and is able to completely regenerate within just a week after tail amputation. In contrast, axolotls require several weeks to completely regenerate their tails, which contain cartilage rather than notochord tissue (Sugiura et al., 2004; Echeverri et al., 2001).
Amputation is known to trigger wound healing, followed by the emergence of proliferating cells that eventually form the complete array of tissue types. In the regenerating tail of Xenopus larvae, the notochord and spinal cord regenerate from their pre-existing notochord and spinal cord tissue, while muscle tissue regenerates from satellite cells rather than from pre-existing myofibers (Sugiura et al., 2004; Gargioli and Slack, 2004). In contrast, undifferentiated pluripotent cells comprising the typical blastema are responsible for regeneration of the urodele limb and tail. Several genes have been reported to have similar expression patterns in the embryonic tail bud and regenerating tail (Beck et al., 2003; Christen et al., 2003). On the other hand, some genes showed different expression patterns (Sugiura et al., 2004). However, still little is known about the mechanisms underlying tail regeneration.
Recent progress in DNA macro/microarray technology has allowed us to screen the expression of thousands of genes simultaneously. This technique is well suited for studying differential gene expression patterns between two or more cell populations (Galitski et al., 1999; Mochii et al., 1999; Lockhart and Winzeler, 2000; Young, 2000; Altmann et al., 2001). To elucidate the molecular mechanisms underlying Xenopus larval tail regeneration, we performed a comprehensive analysis using a cDNA macroarray. Here we report the identification of genes involved in tail regeneration, which provides an initial step towards understanding the molecular mechanisms underlying the process of Xenopus larval tail regeneration.
RESULTS AND DISCUSSION
Identification of Genes Involved in Regeneration by Differential Hybridization of Macroarray Filters
To identify genes involved in larval tail regeneration, we performed differential hybridizations of cDNA macroarrays on which 42,240 independent clones were spotted. These cDNA macroarrays represent 19,200 neurula- and 23,040 tailbud-stage Xenopus EST clones, which are considered to represent 17,000 unigenes by 3′ sequence and contig analyses. The sequence information of the cDNA clones is available at the XDB3 website (http://xenopus.nibb.ac.jp). Many genes with unique expression profiles have been identified using this set of 42,240 cDNAs (Chung et al., 2004; Shin et al., 2005; Peiffer et al., 2005). It is well known that gene expression patterns observed during normal development are also observed during regeneration (Stocum, 1995), and Xenopus larval tail regeneration is not an exception, as several genes are reported to have similar expression patterns in the embryonic tail bud and regenerating tail (Beck et al., 2003; Christen et al., 2003; Sugiura et al., 2004). For these reasons, we used this 42,240 cDNA set for differential hybridization studies to elucidate the molecular mechanisms underlying tail regeneration.
Figure 1 illustrates the strategy we used in this study. As in our previous study (Sugiura et al., 2004), we chose two stages of the regeneration process: 1.5 days post-amputation, when the cells proliferate without apparent differentiation, and 3 days post-amputation, when the first morphological evidence of tissue differentiation can be seen. We compared the gene expression profiles of day 0 and day 1.5 or 3 tails by using cDNA macroarrays. To minimize false signals, which may be caused by variability in the amount of DNA spotted onto different arrays, we used two duplicate filters for each probe. Figure 2 shows the changes in the numbers of clones and contigs whose intensities changed by more than twofold in the regeneration process. We found 216 clones (163 contigs) that were up-regulated, and 157 clones (81 contigs) that were down-regulated between day 0 and day 1.5 or 3 post amputation by more than twofold. The amount of DNA in each spot of the macroarray was quite different, as the DNA was directly amplified by PCR from bacterial stocks of the EST library and was not checked by electrophoresis. To obtain more consistent results, clones containing the longest insert for each contig (selected by the first differential hybridization) were PCR amplified using plasmid DNA as a template and checked by electrophoresis. Duplicate spots for each insert were re-arrayed onto the same nylon membrane so that the false signals should be distinguished clearly. To confirm the results obtained by the first differential hybridization, re-arrayed filters were hybridized once again, and our data set was reduced to 98 up-regulated and 39 down-regulated contigs. These clones were sequenced and BLAST searched against the XDB3 database to confirm EST information. Some clones whose sequence information was not present in XDB3 overlapped with other contigs that were isolated in the macroarray screen. We were finally able to narrow down our data set to 56 up-regulated and 28 down-regulated contigs (Fig. 2, Tables 1, 2).
Table 1. Summary of the Up-Regulated Genes Identified in the Screena
Temporal expression pattern
Fold increases over day 0
1st screening (macroarray)
2nd screening (rearray)
Clones were categorized based upon functions assigned by sequence homology. Numbers in the column indicates the fold increases over day 0.
Indicates clone whose relative expression level showed less than a twofold increase by quantitative real-time RT-PCR.
Hypothetical protein [Thermotoga maritima] (strain MSB8)
Unnamed protein product [Homo sapiens]
Similar to KIAA0971 protein [Mus musculus]
No significant homology
No significant homology
No significant homology
Muscle and CNS
No significant homology
No significant homology
No significant homology
No significant homology
No significant homology
To assess the reliability of our macroarray screen, we performed quantitative real-time RT-PCR. We chose 14 clones from among the up-regulated genes and analyzed their relative expression during regeneration. The real-time RT-PCR results yielded 1.8- to 34.9-fold increases in expression, while our macroarray data showed 2.0- to 13.8-fold up-regulation. Only one clone had a relative ratio of less than twofold (Table 1), indicating that our macroarray analysis was reliable.
Whole Mount In Situ Hybridization Analysis of Identified Genes
As a secondary screen to verify the expression of the genes that appeared to be up-regulated during tail regeneration, we performed whole mount in situ hybridization analysis on tailbud stage embryos and the regenerating larval tail. In some cases, sections were used to verify expression observed by whole mount. Of the 56 genes analyzed by the secondary screen, 48 were expressed in the regenerating tail and eight were not detected. Following several rounds of hybridization, we considered the expression level of these eight genes to be below detectable levels because these genes exhibited low intensities in the array analyses.
For down-regulated genes, expressions in embryos and normal larval tails were examined by whole mount and sections, respectively. Of the 28 genes examined, 15 were expressed in muscle, seven in the spinal cord, two in the muscle and spinal cord, one in the notochord, and the remainder undetermined (Table 2). Genes associated with the differentiated muscle cell phenotype are down-regulated during myofiber dedifferentiation in newt limb regeneration (Simon et al., 1995). In the case of Xenopus larval tail regeneration, however, dedifferentiation of myofibers does not occur (Gargioli and Slack, 2004). Apparent down-regulation of these muscle genes might depend on the degradation of myofibers during Xenopus tail regeneration. In the regenerating tail, very few myofibers are present in the distal region of the regenerating tail, and the lack of differentiated muscle cells in the collected tissues produces this apparent down-regulation of muscle differentiation genes. Since these genes, such as myosin light chain, appeared to be involved in late differentiation steps, we focused on genes that were up-regulated during regeneration in order to examine early steps in the regeneration process.
Categorization of Up-Regulated Genes
Forty-eight genes identified in the screen were categorized into several groups according to their putative function during regeneration. These categories include genes involved in wound healing, inflammation response, cell proliferation, cell signaling, cell differentiation, and cell structure (Table 1). Individual examples of genes in these categories are highlighted below.
Five genes (10%) from this screen appear to be involved in the wound healing process. For example, annexin II (XL073e04) and plasminogen activator inhibitor 1 (PAI1) (XL074b06), which are part of the fibrinolytic surveillance cascade (Hajjar and Krishnan, 1999), show similar expression patterns in the regenerating tail (Fig. 3B,C,E,F). These genes are expressed on the tip of the notochord at early stages of regeneration (Fig. 3B,E) and in blood cells and prenotochordal cells at later stages (Fig. 3C,F).
Cathepsin B (XL040b13), which was one of the earliest molecules identified in regenerating amphibian limbs and tails (Schmidt, 1966), is first expressed in injured stump tissues (Fig. 3H). At later stages, it is expressed in blood cells and in cells flanking the spinal cord (Fig. 3I).
Two genes (4%), complement factor B (XL004j13) and complement component C3 (XL028a02), appear to be involved in inflammation response. Complement component C3 is expressed in blood cells and the wound epidermis (Fig. 3K).
Six genes (13%) identified in our screen appear to be involved in cell proliferation. Histone H3.3B (XL092g11), for example, is expressed ubiquitously in the regeneration bud (Fig. 3M). Serotransferrin precursor (XL040d11), which is thought to be required for proliferation of nearly all cells in serum-free medium (Barnes and Sato, 1980), is expressed in the regenerating notochord (Fig. 3O).
Two genes (4%) identified in our screen appear to be involved in cell signaling. Secreted frizzled-related protein 2 (SFRP2) (XL040a14), a Wnt antagonist (Ladher et al., 2000), is expressed in the proximal region of the regenerating tail (Fig. 4B). SFRP2 may interact with XWnt5a, which is expressed in the distal region of the regenerating tail (Sugiura et al., 2004). SFRP2 has also been reported to be expressed in the regenerating caudal fin of the medaka, suggesting that it plays an important role in the regenerating tail of both species (Katogi et al., 2004).
KLG/PTK7 (XL005m14), a receptor tyrosine kinase that lacks endogenous kinase activity (Chou and Hayman, 1991), is expressed in prenotochordal cells and spinal cord in the regenerating tail (Fig. 4D). Recently, PTK7 was reported to be required for convergent extension and neural tube closure (Lu et al., 2004), suggesting that it may function in a similar way in the process of regeneration.
Four genes (8%) identified in our screen appear to be involved in cell differentiation. For example, ATF5 (XL049c20), a member of the activating transcription factor/CREB protein family, which plays a major regulatory role in the progression of neural progenitor cells to neurons (Angelastro et al., 2003), was identified in our screen. Interestingly, ATF5 is expressed in the spinal cord in the regenerating larval tail (Fig. 4F), although no expression was observed in the neural tube at tailbud stage (Fig. 4E).
XL014c20 encodes a gene similar to genes in the cystein-rich protein (CRP) family, which are potent smooth muscle differentiation cofactors (Chang et al., 2004). This gene is expressed in the notochord (Fig. 4G, H), suggesting that it might be involved in differentiation of notochord.
HMG-17 (XL034a15) is a member of the high mobility group family of proteins, and is a chromosomal non-histone protein that marks differentiating tissues and cells during organogenesis (Lehtonen et al., 1998). HMG-17 is expressed in the regenerating tail, particularly in the differentiating cells (Fig. 4J). HMG-X (XL069l01), also a high mobility group protein that is involved in neurogenesis (Kinoshita et al., 1994), is expressed in the regenerating spinal cord and epidermis (Fig. 4L).
Ten genes (21%) identified in our screen appear to be involved in cell structure. For example, endo B keratin (XL024e24) is expressed in the fin, notochord, and spinal cord in the regenerating bud (Fig 4N).
Keratin 8 (XL080o09) has a very similar pattern of expression as endo B keratin, except that it is also expressed in the epidermis at the tailbud stage (Fig 4O) and is not expressed in the fin in the regenerating tail (Fig. 4P).
Eleven genes (23%) identified in our screen could not be easily classified, and their functions during the tail regeneration process were not predictable. For example, clone XL070a09, which resembles 14-KDa transmembrane protein (Morel et al., 1991), is expressed in the regenerating muscle (Fig. 5B).
Glutamine synthetase (XL106n21) is expressed in the terminal vesicle of the regenerating spinal cord (Fig. 5D).
Eight genes (17%) that had no significant homology to known proteins were classified as novel genes. For example, clone XL064e17, XL025c16, XL051a01, and XL063b01 are expressed in the regenerating spinal cord (Fig. 5E–L). XL025c16 is also expressed in the blood cells (Fig. 5H). Interestingly, XL063b01 is not expressed in the tailbud stage embryo (Fig. 5K), suggesting that this gene may be a regeneration-specific gene.
Regeneration-Specific Expression Patterns
Many of the morphological changes and gene expression patterns observed during normal development are also observed during regeneration (Stocum, 1995). In the case of Xenopus tail regeneration, several genes that are expressed in the embryonic tail bud are re-expressed in the regenerating tail (Beck et al., 2003; Christen et al., 2003; Sugiura et al., 2004). Many genes identified in this study had similar expression patterns during tail regeneration and embryonic development. A few genes, however, are known to have different expression patterns during embryonic stages and regeneration. For example, chordin was expressed in the embryonic tail bud but not in the regenerating tail (Sugiura et al., 2004). We show that Keratin 8 (XL080o09) shows a similar loss of expression in the regenerating epidermis (Fig. 4P), though it is highly expressed in the embryonic tail bud (Fig. 4O). At least two genes, ATF5 and XL063b01, were not expressed in the embryonic tail bud but were expressed in the regenerating tail (Figs. 4E,F and 5K,L), suggesting that there exist some regeneration-specific pathways.
Temporal Expression Patterns of Genes Involved in Regeneration
One of the advantages of cDNA array-based analyses is the ability to simultaneously examine a large number of genes at different time points. We used the re-arrayed filter to examine the temporal expression pattern at 0, 1.5, 3, 4.5, and 6 days post-amputation. Genes that showed greater than twofold up-regulation by 1.5 days post-amputation were defined as early responding, while those that were up-regulated at later stages were defined as late responding genes (Fig. 6A,C and Table 1). There were 26 early responding genes, which fell into the following functional categories: inflammation response (2/26), wound healing (4/26), cell signaling (1/26), cell proliferation (2/26), cell differentiation (1/26), cell structure (7/26), unclassified (4/26), and novel (5/26) (Fig. 6B and Table 1). There were 22 late responding genes, which including wound healing (1/22), cell signaling (1/26), cell proliferation (4/22), cell differentiation (3/22), cell structure (3/22), unclassified (7/22), and novel (3/22) genes (Fig. 6D and Table 1). These temporal expression patterns are consistent with the timing of morphological events that occur during regeneration (for example, inflammation response and wound healing are known to be early responses, whereas cell differentiation occurs later). In the case of down-regulated genes, the expression level decreased immediately at 1.5 days post-amputation (Fig. 6E and Table 2).
Analysis of Genes Expressed During Limb Regeneration
To determine whether there is a common set of genes involved in regeneration of both the tail and the limb, we hybridized re-arrayed filters with probes made from normal and regenerating limbs. There were 15 genes, categorized as being involved in wound healing (3/15), cell proliferation (1/15), cell differentiation (1/15), cell structure (4/15), unclassified (4/15), and novel (2/15), whose expression levels were greater than twofold higher in the regenerating limb than in the normal limb (Table 1). It was previously reported that cathepsins, fibronectin, and keratin 8 are expressed in the regenerating limb (Schmidt, 1966; Grinnell, 1984; Ferretti et al., 1989). Interestingly, ATF5 and XL063b01, which were also pulled out in our tail regeneration screen, were among the genes we found to be up-regulated in the regenerating limb.
Here, we present the first identification of candidate genes involved in Xenopus larval tail regeneration using DNA macroarray technology. Many of the genes identified in this screen are thought to be involved in processes associated with regeneration, such as wound healing, inflammation response, cell proliferation, cell signaling, cell differentiation, and cell structure.
One interesting question in the field of regeneration is whether the molecular mechanisms underlying regeneration resemble those of normal development, or whether regeneration-specific gene regulation exists. To answer this question, it is necessary to isolate the genes that are expressed specifically during regeneration. Through a macroarray-based comprehensive screen, we identified two genes with regeneration-specific expression. Interestingly, these two genes were expressed not only in the regenerating tail but also in the regenerating limb, which suggests the possibility that these genes are important for the regeneration process in general. To investigate the function of these genes, manipulation of gene expression will be necessary. Several techniques, such as transgenic methods and electroporation of plasmids or morpholino oligos, have been established (Beck et al., 2003; Echeverri and Tanaka, 2003; Schnapp and Tanaka, 2005). These techniques should prove useful to investigate the function of genes identified in this study and provide a more detailed understanding of the molecular mechanisms underlying the process of Xenopus larval tail regeneration.
Animals and Surgical Operation
Xenopus laevis tadpoles obtained after ovulation and copulation induced by gonadotropic hormones (Teikoku Zouki) were subjected to 50% tail removal at stage 48 (according to Nieuwkoop and Faber, 1956). Operated tadpoles were cultured in autoclaved tap water with 10,000 IU/ml penicillin (Meiji) and 10 mg/ml streptomycin (Meiji) at 22°C.
The bacterial stocks of 42,240 independent cDNA clones, ranging from XL001a01 to XL110p24 from the NIBB Mochii normalized gastrula and tailbud cDNA libraries, generated by the NIBB/NIG Xenopus laevis EST project (Kitayama, unpublished data), were as used templates to PCR amplify the inserts using vector-specific primers. Inserts were arrayed onto a nylon membrane filter (Hybond-N; Amersham Pharmacia) using a gridding robot (Bio Robotic) at a high-density (384 × 5 × 5 cDNA grids per 8 × 12 cm filter). The sequence information of the cDNA clones is available at the XDB3 website (http://xenopus.nibb.ac.jp). The filters were denatured with 0.5M NaOH and 1.5M NaCl, neutralized with 0.5M Tris-HCl (pH 7.0), 1 mM EDTA, and 1.5M NaCl, and baked at 80°C for 2 h.
The regenerating tails were cut again at about 0.5 mm from the point of amputation at each stage (0, 1.5, 3, 4.5, 6 days after operation). Total RNA was extracted using Trizol reagent (Invitrogen, La Jolla, CA), according to the manufacturer's instructions.
For probe labeling, 10 μg of total RNA (days 0, 1.5, 3) was incubated with oligo(dT) primer, Superscript RTase (Invitrogen) and 32P-dCTP at 37°C for 2.5 h. Two duplicate filters were hybridized as previously described (Mochii et al. 1999). Hybridized signals were detected using the Fuji BAS system and quantified with Array Gauge software (Fuji film). After quantification, the intensities of corresponding spots from the two duplicate filters were averaged. The averaged intensities for day-1.5 and -3 samples were normalized to the total intensity of the day-0 spots.
cDNA Clone Recovery and Rearray
Clones that were up-regulated or down-regulated during tail regeneration were isolated from the library and their plasmid DNA was purified using the Automatic DNA Isolation System PI-50 (Kurabo). Inserts that were PCR amplified using plasmid DNA as templates were then re-arrayed at high density (384 × 4 × 4 cDNA grids per 8 × 12 cm filter), as described above. EF1-alpha was used to normalize the data.
Real-Time Quantitative RT-PCR
Complementary DNA was synthesized from total RNA (1 μg) using Ready-to-Go You-Prime First-Strand Beads (Amersham Biosciences) and random hexamers, according to the manufacturer's instructions. The quantification of each gene was done using real-time RT-PCR. The specific primers used to quantitate were as described in Table 3. The amplification profile for these primer pairs was as follows: 95°C for 10 min, 40 cycles of 95°C for 15 s, and 65°C for 1 min, performed in an ABI PRISM 7000 Sequence Detection System (PE Applied Byosystems). Standard curves were generated for each gene and unknown quantities relative to the standard curve for each gene were calculated, yielding transcriptional quantitation of each gene relative to the endogenous standard (EF1-alpha).
Table 3. Primer Pairs Used for Quantitative Real-Time RT-PCR
In Situ Hybridization
The insert encoding each selected gene was PCR amplified using T3/T7 primers of the Bluescript vector that flanked the insert. Antisense digoxigenin (DIG)-labeled RNA probes were synthesized with T7 RNA polymerase (Invitrogen) and a DIG-labeled nucleotide mixture (Roche). Whole mount and section in situ hybridization were performed as previously described (Sugiura et al., 2004) using BM purple (Roche) for color detection.
This work was supported by the “Research for the Future Program” of the Japanese Society for the Promotion of Science to N.U. and a Grant-in-Aid from the Ministry of Education, Science, Sports, and Culture, Japan, to M.M. We thank Dr. Orii, Takuji Sugiura, Yuka Taniguchi, and other members of our laboratory for helpful discussion and maintenance of animals.